Original Article

Filicide (child homicide by parents) in Hong Kong

Hong Kong Med J 2025;31:Epub 1 Apr 2025

ORIGINAL ARTICLE

Filicide (child homicide by parents) in Hong Kong

Yuen Dorothy Yee Tang, MB, BS, FHKAM (Psychiatry)¹; Jessica PY Lam, MB, BS, FHKAM (Psychiatry)²; Amy CY Liu, MB, ChB, FHKAM (Psychiatry)¹; Bonnie WM Siu, MB, ChB, FHKAM (Psychiatry)¹

¹ Department of Forensic Psychiatry, Castle Peak Hospital, Hong Kong SAR, China

² Department of Psychiatry, Queen Mary Hospital, Hong Kong SAR, China

Corresponding author: Dr Yuen Dorothy Yee Tang (tyy551@ha.org.hk)

Full paper in PDF

Abstract

Introduction: Filicide refers to an act in which a parent or stepparent kills a child. This retrospective study provides the first comprehensive analysis of filicides in Hong Kong over a 15-year period.

Methods: The study explored the local epidemiology, differences between maternal and paternal filicides, associated mental illnesses, and the criminal responsibility of the perpetrators.

Results: Among 81 filicide cases (43 female victims, 37 male victims, and 1 victim of unknown gender), the incidence rate was 0.7 per 100 000 population. Mothers were responsible for two-thirds (66.7%) of the cases, fathers for 19.8%, and the remainder involved both parents. Victims aged <1 year (n=44) were nearly equal in number to those aged between 1 and 17 years (n=41). Mental illness was diagnosed in 31.0% of the perpetrators, predominantly depression and psychotic disorders. Paternal perpetrators exhibited a higher prevalence of mental illness and were more frequently involved in filicide-suicides. One-third (33%) of perpetrators with mental illness invoked the psychiatric defence of diminished responsibility, resulting in Hospital Order sentencing. Reduced culpability due to mental illness and the application of infanticide provisions provided legal protections for mothers who killed their children aged <1 year.

Conclusion: Understanding the local epidemiology of filicide and the mental health conditions of perpetrators may help identify at-risk populations and develop effective intervention strategies.

New knowledge added by this study

The epidemiology, differences between maternal and paternal filicides, associated mental illnesses, and the criminal responsibility of the perpetrators in Hong Kong from 2003 to 2017 were explored.
Maternal perpetrators were disproportionately responsible for infanticides, highlighting the protective legal provisions applied to mothers who kill their children aged <1 year.

Implications for clinical practice or policy

Understanding the local epidemiology of filicide and the mental health conditions of perpetrators may help identify at-risk populations and develop effective intervention strategies.
Enhanced mental health screening and support for parents, particularly mothers of infants, could potentially prevent cases of filicide.

[Abstract in Chinese]

Introduction

Child homicide represents a rare but important global issue with devastating consequences for families and communities. The global homicide rate among children aged 0 to 17 years was 1.6 per 100 000 population in 2016,1 and approximately 95 000 children are murdered annually.2 A 2017 review by Stöckl et al3 found that the majority of child homicides were committed by a family member; parents were responsible for over half of the cases involving child victims.3

Filicide

Filicide refers to the act of killing one’s own child. Subcategories of filicide include neonaticide, a term introduced by Resnick4 to describe the murder of a child within the first 24 hours after birth, and infanticide, which applies when the victim is aged <1 year. Resnick4 identified various motives for filicide. In altruistic filicide, the parent believes that the act is in the child’s best interests. An acutely psychotic parent may kill a child under the influence of severe mental illness. In unwanted child filicide, a parent kills a child who is perceived as a hindrance. Accidental/fatal maltreatment describes the unintentional death of a child due to parental abuse or neglect. Spouse revenge filicide occurs when a child is killed as a means of exacting revenge upon the spouse or the other parent. Bourget and Bradford5 later emphasised the importance of the perpetrator’s gender by introducing paternal filicide as a distinct category.

Victim and perpetrator characteristics vary in cases of filicide. The first year of life is a critical period, and the highest risk of filicide occurs within the first 24 hours. Neonaticides are predominantly committed by mothers,6 and mothers are overrepresented across the entire spectrum of filicide.4 5 However, contradictory results have been reported.5 7 8 The gender distribution of victims also varies. Male children aged <1 year are at greater risk in high-income Western countries, such as the US9 and the UK10; the opposite trend has been observed in India and China.11 Some studies have shown that boys are overrepresented among victims,7 12 whereas others have identified comparable numbers of male and female filicide victims.13

Maternal and paternal perpetrators of filicide exhibit distinct characteristics.14 15 Maternal perpetrators tend to be younger and have younger victims compared with fathers.15 Younger maternal perpetrators are often poor, experience psychosocial stress, and lack family and community support, whereas older maternal perpetrators frequently have mental illnesses and lack criminal histories.13 14 16 In contrast, paternal perpetrators are more commonly driven by anger, jealousy, or marital and life discord.15 Fatal abuse and acts of retaliation are more prevalent among paternal perpetrators than among maternal perpetrators.17 Fathers are also more likely to attempt or die by suicide12 17 18 when committing filicide.14 18 Additionally, fathers typically use more violent methods to cause death.19

Filicide and mental illness

Pathological filicide, characterised by altruistic or actively psychotic motives, constitutes one of the most common categories of filicide.17 Psychiatric factors are involved in 36% to 85% of all filicide cases.5 16 20 21 22 Maternal perpetrators are more likely to have a history of mental illness and to exhibit symptoms at the time of the offence.22 The most frequent diagnosis among maternal perpetrators is major depressive disorder, followed by schizophrenia.5 16 20 23 Personality disorders and substance use are more often associated with paternal filicides.8

The criminal justice system and infanticide laws

Filicide presents unique challenges for the criminal justice system. Societal attitudes regarding parents who kill their children are often ambivalent, balancing the need for justice due to loss of innocent life against calls for mercy towards offenders who may require care rather than punishment.

Legal systems worldwide acknowledge that filicide should be treated differently from other forms of homicide. The UK enacted the Infanticide Act in 1922 (amended in 1938)24 to recognise the biological vulnerability of women to psychiatric illnesses during the perinatal period. The Act mandated sentences of probation and psychiatric treatment for offenders.24 By the late 20th century, 29 countries had revised penalties for infanticide to consider unique biological and psychological changes associated with childbirth.25

In Hong Kong, perpetrators with mental illnesses can invoke psychiatric defences, including insanity or diminished responsibility. The insanity defence is based on the M’Naghten principles, which hold that it is unjust to punish an individual for an action performed without the mental capacity to control it. The defence of diminished responsibility applies when the offender demonstrates abnormal mental function arising from a recognised medical condition, which has substantially impaired their ability to either understand the nature of their conduct, form a rational judgement, or exercise self-control (or any combination of these impairments). Perpetrators with mental illnesses who are found not guilty by reason of insanity, or who successfully raise the partial defence of diminished responsibility—thereby reducing the charge from murder to manslaughter—may be sentenced to a Hospital Order at the Correctional Services Department Psychiatric Centre (Siu Lam Psychiatric Centre [SLPC]), under Section 75 of the Criminal Procedure Ordinance26 or Section 45 of the Mental Health Ordinance,27 respectively, for psychiatric observation and management.

A separate legal provision exists for mothers who kill their children aged <1 year. Hong Kong has adopted the UK concept of infanticide, in which mothers experiencing vulnerability after childbirth are charged with infanticide rather than murder, under Section 47C of the Offences against the Person Ordinance.28

A study has shown that the local homicide rate in Hong Kong is lower than global averages (0.32 vs 6.1 victims per 100 000 population in 2017),29 but no filicide-specific data are available. The underlying hypothesis in this study was that the incidence of filicide would be lower in Hong Kong than in Western countries, consistent with the lower local homicide rate and the protective effects of cultural factors. The objectives of this study were to describe the epidemiology of filicide in Hong Kong, examine the characteristics of victims and perpetrators (including associated mental illnesses), and evaluate the local criminal justice system’s response to infanticide and other forms of filicide.

Methods

Data were obtained from the Hong Kong Police Force regarding child homicide cases that occurred from 2003 to 2017. These data included the age and gender of the victim, relationship of the perpetrator to the victim, mode of death, year of offence, and charges against the defendant along with corresponding outcomes and sentences. Medical records from the Hospital Authority and the SLPC of the Correctional Services Department were reviewed to determine any history of mental illness. Psychiatric diagnoses of the perpetrators, based on the International Classification of Diseases, Tenth Revision, were documented during forensic psychiatric assessments conducted by two psychiatrists, at least one of whom was a specialist. For the minority of defendants who were not sent to psychiatric hospitals or SLPC after the offences, the presence or absence of mental illness was cross-referenced using newspaper articles. Charges and sentences were verified through judgements available on the Judiciary’s official website.

All statistical analyses were performed using SPSS software (Windows version 21.0; IBM Corp, Armonk [NY], US). Data were analysed with descriptive statistics, including the mean, median, standard deviation, 95% confidence interval, and percentages for categorical variables. Differences between groups in demographic characteristics were assessed using t tests and univariate analysis of variance for continuous data. For nominal data, the Kruskal–Wallis and Chi squared tests were utilised.

Results

Epidemiology of child homicide

From 2003 to 2017, 107 child homicide victims were recorded in Hong Kong, equating to approximately 0.70 death per 100 000 population, based on a population of 1 024 000 children aged <18 years in 2010.30 Among these victims, 81 (75.7%) were killed by their parents (Fig).

Figure. Child homicide cases in Hong Kong from 2003 to 2017

Characteristics of victims and perpetrators

Among the filicide victims (n=81), 53.1% were female, 45.7% were male, and the gender of the remaining victim was unknown. There was no significant correlation between the gender of the victim and the gender of the perpetrator (χ²=0.13; P=0.82). The median age of the victims was 6 years (interquartile range [IQR]=0-8) [Table 1].

Table 1. Demographics of victims in infanticide and other filicide cases

Of the 81 filicide victims, 54 (66.7%) were killed by their mothers, 16 (19.8%) by their fathers, and 11 (13.6%) by both parents. The median age of victims varied across perpetrator groups; the paternal group victims had a median age of 7.5 years (IQR=5-10.25), compared with 0 year (IQR=0-3.5) for the maternal group and 2 years (IQR=0-5) for the parental couple group (H²=14.31; P<0.001).

Characteristics of infanticide and other filicide cases

Forty victims aged <1 year were killed by 44 perpetrators, and 41 victims aged ≥1 year were killed by 40 perpetrators. No significant gender differences were observed among the victims (Table 1).

The median age of paternal perpetrators, 43.5 years, was significantly older than the median ages of maternal and parental couple perpetrators (H2=16.50; P<0.001). The median age of offenders in the infanticide group was younger than that of offenders in the other filicide group. In the infanticide group, nine mothers (26.5%) were <20 years, and all pregnancies had been concealed. These infants were killed immediately after birth. Single offenders were more prevalent in the infanticide group, whereas married offenders were more common in the other filicide group. Biological mothers were the main perpetrators in both groups; similar to paternal and couple perpetrators, maternal perpetrators were younger in the infanticide group (Table 2). The maternal group was responsible for 40% of victims aged <4 years, compared with 7.1% in the paternal group. A higher prevalence of mental illness was identified among perpetrators, particularly mothers, in the other filicide group. Among perpetrators in the infanticide group, depression (40%) was the most common diagnosis, followed by a psychotic disorder (20%), mental and behavioural disorders due to psychoactive substance use (20%), and mental retardation (20%). The only biological father in the infanticide group was diagnosed with harmful use of alcohol. In the other filicide group, among maternal perpetrators, 25.0% had a psychotic disorder, 18.8% had depression, 6.4% had bipolar affective disorder, and the remainder had unknown diagnoses. Among paternal perpetrators, 18.0% had depression, 9.1% had a psychotic disorder, and the remainder had undocumented diagnoses.

Table 2. Demographics of offenders in infanticide and other filicide cases

Suffocation or strangulation was the most common mode of death in infanticides, occurring in 95.7% of cases with maternal perpetrators. In contrast, paternal perpetrators (100%) and couples (50%) caused death mainly by bashing, throwing, or shaking the infants. The two most common modes of death across all filicides were drug overdose or poisoning (including charcoal burning) and stabbing. Drug overdose or poisoning was most frequently performed by maternal perpetrators (36.8%) and couples (57.1%), whereas paternal perpetrators most often engaged in stabbing (57.1%).

Excluding the four perpetrators who died by suicide, 80.0% of perpetrators in the infanticide group faced criminal charges and were convicted. The most common convictions were concealing the birth of a child, manslaughter, and infanticide (Table 2). In the other filicide group, excluding the 18 perpetrators who died by suicide, 95.5% of perpetrators were charged and convicted; manslaughter was the most common conviction, followed by murder. Sentences significantly differed between the infanticide and other filicide groups. Noncustodial sentences were more frequent in the infanticide group than in the other filicide group. Given the higher prevalence of mental illness in the other filicide group, 33.3% (5/15) of the perpetrators were convicted of manslaughter under diminished responsibility and sentenced to a Hospital Order, compared with 6.3% in the infanticide group (Table 2). Among paternal and couple perpetrators, 80% in the infanticide group and 92.3% in the other filicide group received prison sentences, ranging from 3 to 10 years and 18 months to life imprisonment, respectively. Similar proportions of maternal perpetrators in both groups—41.0% in the infanticide group and 42.9% in the other filicide group—were imprisoned. Among maternal perpetrators in the infanticide group, all but one received prison sentences of <1 year; the exception received an 8-year sentence. In the other filicide group, maternal perpetrators received sentences of 4 to 7 years.

Filicide-suicide is defined as the perpetrator dying by suicide within 24 hours of committing filicide. A significantly greater proportion of filicide-suicides occurred in the other filicide group. In the infanticide group, all perpetrators were biological mothers. In contrast, within the other filicide group, half of maternal perpetrators and 66.7% of paternal perpetrators had a diagnosed mental illness. The difference in mental illness prevalence between the two groups was not statistically significant (Table 3).

Table 3. Characteristics of perpetrators in filicide-suicide cases (n=22)

Mental illness of filicide offenders

Of the 84 filicide perpetrators, 26 (31.0%) were diagnosed with mental illness. No mental illness was reported in the parental couple group. A higher prevalence of mental illness was observed among paternal perpetrators (58.3%) than among maternal perpetrators (38.0%), although the difference was not statistically significant. Depression was the most common diagnosis, followed by psychotic disorder. In cases of filicide-suicide, mental illness prevalence was higher among paternal perpetrators; this difference was not statistically significant (Table 4).

Excluding perpetrators who died by suicide, 41.7% of maternal perpetrators with mental illness received a Hospital Order for an unspecified period. Among the three paternal perpetrators with mental illness who did not die by suicide, only one (33.3%) was sentenced to a Hospital Order for an unspecified period.

Table 4. Mental illness in filicide perpetrators (n=26)

Discussion

The incidence of child homicide in Hong Kong, at 0.7 per 100 000 population, is lower than the global average (1.6 per 100 000 population)1 and lower than that of Asian countries with similar socio-economic status, such as South Korea (1.03 per 100 000 population).31 The protective influence of traditional Confucian cultural values may play a prominent role in Hong Kong.32 An idiom from the Sung dynasty, ‘even a vicious tiger would not eat its cubs’, continues to be taught in modern primary schools. This cultural ethos could explain why the incidence of child maltreatment in Hong Kong, at <0.14%,30 remains lower than the global rate of 0.3% to 0.4%.33 Consistent with studies worldwide,3 most child homicides in Hong Kong were perpetrated by parents. Mothers were the predominant perpetrators in filicides. The typical profile of an infanticidal perpetrator was a young, single mother who suffocated or strangled the infant. Some cases may represent neonaticides, as suggested by charges of concealing the birth of a child. Among cases involving the filicide of older children, perpetrator characteristics were more heterogeneous. Perpetrators tended to be older and married; they used methods such as overdosing, poisoning, or stabbing. The profiles of perpetrators and victims in this group also differed. The median age of maternal perpetrators was younger and their victims tended to be younger. Mothers most often caused death through overdosing or poisoning, whereas fathers were more likely to kill by stabbing.

Mental illness in filicides

In the present study, 31.0% of filicidal perpetrators had a diagnosed mental illness, a lower rate compared with other population studies.8 20 22 23 This discrepancy could be attributed to the lower prevalence of mental illness in Hong Kong. The Hong Kong Mental Morbidity Survey (2010-2013) revealed a 13.3% prevalence of mental disorders among Chinese adults,34 compared with 18.5% among adults in the US in 2013.35 It is also plausible that some perpetrators, especially those involved in filicide-suicide cases, had no prior contact with mental health services and may have had undiagnosed psychiatric illnesses. Mental illness prevalence was higher among paternal perpetrators than among maternal perpetrators in our filicide sample. This finding may be related to the small sample size or could reflect societal changes, such as fathers assuming greater childcare responsibilities.17 Consistent with some studies,20 22 depression was the most common diagnosis, followed by psychotic disorder.

Filicide-suicides

Substantial proportions of filicide perpetrators (23.0% of maternal and 34.8% of paternal) died by suicide during or after committing the act. Charcoal burning was the most common method, comparable to the frequency of jumping from height. Charcoal burning is a relatively recent suicidal method,36 which has spread as a contagious phenomenon in other Asian countries; it is often portrayed as a ‘peaceful way of dying’ and has been used during >10% of suicides in the region.37 The proportion of filicide-suicides observed in this study was lower than that reported in other studies.17 23 This difference may be related to the lower prevalence of mental illness in our sample, the relatively lower lethality of charcoal burning in Hong Kong compared with firearm use in Western countries, or the possibility that attempted suicides not resulting in death were not captured in our data. Filicide-suicide events were more frequent in cases involving older children than in infanticides, potentially due to differences in underlying motives. Half of the filicide-suicide perpetrators in the present study had a history of mental illness, suggesting that altruistic motives were involved. Depression was the most frequently diagnosed condition in these cases.18 20

The local law and filicides

The majority of perpetrators with mental illness were convicted of manslaughter under diminished responsibility and sentenced to a Hospital Order at SLPC for an unspecified period under Section 45 of the Mental Health Ordinance.27 No insanity pleas were recorded in our sample. Consistent with international studies,38 maternal perpetrators in Hong Kong received more lenient outcomes relative to paternal perpetrators. Some young mothers who killed their children aged <1 year were released without charge; among those convicted, a few received noncustodial sentences. In contrast, all fathers who killed their children were imprisoned, with the exception of one who was sentenced to a Hospital Order at SLPC.

Hong Kong developed its legislation based on the UK law, including the British Infanticide Act of 1922.21 24 Section 47C of the Offences against the Person Ordinance28 defines the offence of infanticide as follows: “Where a woman by any wilful act or omission causes the death of her child being a child under the age of 12 months but at the time of the act or omission the balance of her mind was disturbed by reason of her not having fully recovered from the effect of giving birth to the child or by reason of the effect of lactation consequent upon the birth of the child, then, notwithstanding that the circumstances were such that but for the provisions of this section the offence would have amounted to murder, she shall be guilty of infanticide, and shall be liable to be punished as if she were guilty of manslaughter.” In the present study, eight mothers who killed their children aged <1 year were convicted under the infanticide provision. There appears to be considerable application of this provision in Hong Kong; lenient noncustodial sentences are issued to mothers in such cases.

Limitations

First, information provided by the Police was restricted to arrest cases; thus, the study may underreport the true incidence of filicides in Hong Kong. Second, although multiple sources of information were utilised, details regarding the perpetrators’ and victims’ abuse or victimisation histories, involvement with social services, or autopsy reports were unavailable. Third, the classification of neonaticides was challenging, although charges of concealing the birth of a child may indicate the death of a victim within 24 hours of birth. Fourth, although most diagnoses of offenders with mental illnesses were accessible, the availability of psychiatric records was limited. Information for a small number of cases (<5) was obtained from newspaper reports. Sixth, the absence of critical details, such as the onset of mental illness, symptomatology, and medication adherence, impeded a thorough exploration of the relationship between mental illness and filicides. A more comprehensive approach, such as conducting psychological autopsies—particularly in filicide-suicide cases—would provide deeper insights. Finally, the sample size was insufficient to allow for robust comparisons among perpetrators in maternal, paternal, parental couple, and stepparent filicide groups.

Conclusion

In this study, most child homicides were perpetrated by parents; mothers committed filicide more frequently than fathers. Maternal perpetrators and their victims were younger than their counterparts in the paternal perpetrator group. Mental illness was prevalent among filicidal perpetrators of both genders, with a higher prevalence in paternal perpetrators. Filicide-suicide is a substantial problem. Psychiatrists should remain vigilant in identifying depressed or psychotic parents and in eliciting self-harm or filicidal ideations among both mothers and fathers. Social support and child protection services should be actively offered to young single mothers. In Hong Kong, a comprehensive child development service has been established since 2005,39 with the aim of identifying and intervening early in cases that involve children and mothers in need; this service seeks to improve health outcomes for children and families. However, no local policies specifically address the needs of fathers. A multidisciplinary approach involving mental health professionals and social workers is recommended to screen fathers experiencing mental illness or distress and to identify early warning signs of risk. Finally, given the high prevalence of mental illness among filicidal perpetrators, forensic psychiatrists and related professionals should maintain a high index of suspicion for the presence of mental illness when evaluating filicidal offenders.

Author contributions

Concept or design: All authors.
Acquisition of data: YDY Tang.
Analysis or interpretation of data: YDY Tang, JPY Lam.
Drafting of the manuscript: YDY Tang.
Critical revision of the manuscript for important intellectual content: YDY Tang.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This research was approved by the New Territories West Cluster Research Ethics Committee of the Hospital Authority, Hong Kong (Ref No.: NTWC/REC/19021). A waiver for informed patient consent was granted by the Committee due to the retrospective nature of the research.

References

1. United Nations Office on Drugs and Crime. Global Study on Homicide. Killing of Children and Young Adults. 2019. Available from: https://www.unodc.org/documents/data-and-analysis/gsh/Booklet_6new.pdf. Accessed 6 Apr 2023.

2. UNICEF. Hidden in plain sight: a statistical analysis of violence against children. New York: United Nations International Children’s Emergency Fund. 2014 Sep 4. Available from: https://data.unicef.org/resources/hidden-in-plain-sight-a-statistical-analysis-of-violence-against-children/. Accessed 6 Apr 2023.

3. Stöckl H, Dekel B, Morris-Gehring A, Watts C, Abrahams N. Child homicide perpetrators worldwide: a systematic review. BMJ Paediatr Open 2017;1:e000112. Crossref

4. Resnick PJ. Child murder by parents: a psychiatric review of filicide. Am J Psychiatry 1969;126:325-34. Crossref

5. Bourget D, Bradford JM. Homicidal parents. Can J Psychiatry 1990;35:233-8. Crossref

6. Wilson RF, Klevens J, Fortson B, Williams D, Xu L, Yuan K. Neonaticides in the United States—2008-2017. Acad Forensic Pathol 2022;12:3-14. Crossref

7. Vanamo T, Kauppi A, Karkola K, Merikanto J, Räsänen E. Intra-familial child homicide in Finland 1970-1994: incidence, causes of death and demographic characteristics. Forensic Sci Int 2001;117:199-204. Crossref

8. Bourget D, Gagné P. Paternal filicide in Québec. J Am Acad Psychiatry Law 2005;33:354-60.

9. Mariano TY, Chan HC, Myers WC. Toward a more holistic understanding of filicide: a multidisciplinary analysis of 32 years of U.S. arrest data. Forensic Sci Int 2014;236:46-53. Crossref

10. Brookman F, Nolan J. The dark figure of infanticide in England and Wales: complexities of diagnosis. J Interpers Violence 2006;21:869-89. Crossref

11. Sahni M, Verma N, Narula D, Varghese RM, Sreenivas V, Puliyel JM. Missing girls in India: infanticide, feticide and made-to-order pregnancies? Insights from hospital-based sex-ratio-at-birth over the last century. PLoS One 2008;3:e2224. Crossref

12. Dawson M. Canadian trends in filicide by gender of the accused, 1961–2011. Child Abuse Negl 2015;47:162-74. Crossref

13. Camperio Ciani AS, Fontanesi L. Mothers who kill their offspring: testing evolutionary hypothesis in a 110-case Italian sample. Child Abuse Negl 2012;36:519-27. Crossref

14. Harris GT, Hilton NZ, Rice ME, Eke AW. Children killed by genetic parents versus stepparents. Evol Hum Behav 2007;28:85-95. Crossref

15. West SG, Friedman SH, Resnick PJ. Fathers who kill their children: an analysis of the literature. J Forensic Sci 2009;54:463-8. Crossref

16. Friedman SH, Horwitz SM, Resnick PJ. Child murder by mothers: a critical analysis of the current state of knowledge and a research agenda. Am J Psychiatry 2005;162:1578-87. Crossref

17. Bourget D, Grace J, Whitehurst L. A review of maternal and paternal filicide. J Am Acad Psychiatry Law 2007;35:74-82.

18. Hatters Friedman S, Hrouda DR, Holden CE, Noffsinger SG, Resnick PJ. Filicide-suicide: common factors in parents who kill their children and themselves. J Am Acad Psychiatry Law 2005;33:496-504.

19. West SG, Hatters Friedman S. Filicide: a research update. In: Browne RC, editor. Forensic Psychiatry Research Trends. New York: Nova Science Publishers; 2007: 29-62.

20. Bourget D, Gagné P. Maternal filicide in Québec. J Am Acad Psychiatry Law 2002;30:345-51.

21. Hatters Friedman S, Resnick PJ. Child murder by mothers: patterns and prevention. World Psychiatry 2007;6:137-41.

22. Flynn SM, Shaw JJ, Abel KM. Filicide: mental illness in those who kill their children. PLoS One 2013;8:e58981. Crossref

23. Kauppi A, Kumpulainen K, Karkola K, Vanamo T, Merikanto J. Maternal and paternal filicides: a retrospective review of filicides in Finland. J Am Acad Psychiatry Law 2010;38:229-38.

24. Legislation.gov.uk. Infanticide Act 1938. Available from: https://www.legislation.gov.uk/ukpga/Geo6/1-2/36. Accessed 25 Mar 2025.

25. Oberman M. Mothers who kill: coming to terms with modern American infanticide. Am Crim L Rev 1996;34:1-110.

26. Hong Kong SAR Government. Criminal Procedure Ordinance (Cap 221). Available from: https://www.elegislation.gov.hk/hk/cap221. Accessed 17 Mar 2025.

27. Hong Kong SAR Government. Mental Health Ordinance (Cap 136). Available from: https://www.elegislation.gov.hk/hk/cap136. Accessed 6 Apr 2023.

28. Hong Kong SAR Government. Offences against the Person Ordinance (Cap 212). Available from: https://www.elegislation.gov.hk/hk/cap212. Accessed 6 Apr 2023.

29. Hong Kong Police Force. Crime statistics comparison. 2017. Available from: https://www.police.gov.hk/ppp_en/09_statistics/csc.html. Accessed 6 Apr 2023.

30. Child Fatality Review Panel, Social Welfare Department, Hong Kong SAR Government. Second report for child death cases in 2010-2011. July 2015. Available from: https://www.swd.gov.hk/storage/asset/section/655/en/fcw/CFRP2R-Eng.pdf. Accessed 6 Apr 2023.

31. Jung K, Kim H, Lee E, et al. Cluster analysis of child homicide in South Korea. Child Abuse Negl 2020;101:104322. Crossref

32. Lassi N. A Confucian theory of crime [dissertation]. University of North Dakota; 2018.

33. Stoltenborgh M, Bakermans-Kranenburg MJ, Alink LR, van IJzendoorn MH. The prevalence of child maltreatment across the globe: review of a series of meta-analyses. Child Abuse Rev 2015;24:37-50. Crossref

34. Lam LC, Wong CS, Wang MJ, et al. Prevalence, psychosocial correlates and service utilization of depressive and anxiety disorders in Hong Kong: the Hong Kong Mental Morbidity Survey (HKMMS). Soc Psychiatry Psychiatr Epidemiol 2015;50:1379-88. Crossref

35. Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration, US Department of Health and Human Services. Results from the 2013 National Survey on Drug Use and Health: Mental Health Findings. November 2014. Available from: https://www.samhsa.gov/data/sites/default/files/NSDUHmhfr2013/NSDUHmhfr2013.pdf. Accessed 6 Apr 2023.

36. The Hong Kong Jockey Club Centre for Suicide Research and Prevention, The University of Hong Kong. Statistics of suicide data in Hong Kong (by year). Distribution of method of suicide by age group in Hong Kong. 2020. Available from: https://www.csrp.hku.hk/statistics/. Accessed 6 Apr 2023.

37. Chang SS, Chen YY, Yip PS, Lee WJ, Hagihara A, Gunnell D. Regional changes in charcoal-burning suicide rates in East/ Southeast Asia from 1995 to 2011: a time trend analysis. PLoS Med 2014;11:e1001622. Crossref

38. Porter T, Gavin H. Infanticide and neonaticide: a review of 40 years of research literature on incidence and causes. Trauma Violence Abuse 2010;11:99-112. Crossref

39. Education Bureau, Hong Kong SAR Government. Comprehensive Child Development Service. 2021. Available from: https://www.edb.gov.hk/en/edu-system/preprimary-kindergarten/comprehensive-child-development-service/index.html. Accessed 18 Mar 2025.

Use of pronase in screening for early cancers of the upper gastrointestinal tract

Hong Kong Med J 2025 Feb;31(1):41–7

https://doi.org/10.12809/hkmj2411578

ORIGINAL ARTICLE (HEALTHCARE IN CHINA)

Use of pronase in screening for early cancers of the upper gastrointestinal tract

Zhengqi Wu, BSc¹; Shihua Li, BSc²; Linzhi Lu, BSc²; Zhiyi Zhang, BSc²; Guiqi Wang, BSc, MD³; Tianyan Qin, MSc²; Guangyuan Zhao, MSc²; Jindian Liu, MSc²

¹ Department of Gastroenterology, Wuwei Liangzhou Hospital, Wuwei, China

² Department of Gastroenterology, Wuwei Tumor Hospital, Wuwei, China

³ Department of Endoscopy, Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, China

Corresponding author: Prof Zhengqi Wu (wzqwwzl@163.com)

Full paper in PDF

Abstract

Introduction: This study aimed to investigate the effectiveness of pronase in improving the detection rate of early cancer and enhancing visual field clarity during gastroscopy in China.

Methods: In total, 1450 patients who participated in an early diagnosis and treatment programme of upper gastrointestinal cancer in Wuwei, Gansu Province between 2020 and 2021 were enrolled. Cluster randomisation was utilised at the community level. All patients underwent endoscopy and biopsy. The experimental group (n=725) received pronase granules and dimethicone prior to gastroscopy; the control group (n=725) received dimethicone alone. Endoscopic visibility scores, examination durations, and lesion detection rates were recorded for both groups.

Results: Visibility scores for all regions of the stomach were significantly lower in the experimental group than in the control group (P<0.001). This finding remained consistent after adjustment for confounding factors in multiple linear regression analysis. The detection rate of precancerous lesions and early cancer was significantly higher in the experimental group than in the control group (77.5% vs 62.5%; P<0.001). Binary logistic regression analysis indicated that the likelihood of detecting early cancer was greater in the experimental group, with an odds ratio of 3.840 (95% confidence interval=1.204-12.241; P=0.023). Also, average gastroscopy time was significantly shorter in the experimental group than in the control group (6.52±2.51 min vs 10.03±1.23 min, t=33.81; P=0.001).

Conclusion: The administration of pronase prior to gastroscopy enhances visual field clarity, reduces examination time, and increases the detection rates of precancerous lesions and early cancer.

New knowledge added by this study

Pronase enhances visual field clarity during gastroscopy and reduces examination time.
Pronase can enhance diagnostic precision by minimising misdiagnoses and missed lesions.

Implications for clinical practice or policy

Pronase improves the detection rates of precancerous lesions and early cancer. The results provide a strong scientific foundation for using pronase in endoscopic screening during clinical diagnostic examinations.
The findings support adoption of pronase as a standard adjunct in gastroscopy to improve diagnostic accuracy and procedural efficiency.

[Abstract in Chinese]

Introduction

The implementation of early gastric cancer screening in community populations and performance of endoscopic examinations in high-risk groups represents a feasible, cost-effective, and efficient strategy to address the challenges of gastric cancer diagnosis and treatment in China.1 More than 80% of early-stage gastric cancer cases are identified in asymptomatic community populations aged ≥40 years. Thus, community-based screening programmes are important for increased detection of early-stage cancer. Gastroscopy remains the gold standard for diagnosing upper gastrointestinal diseases. High-quality intragastric visibility is essential for ensuring diagnostic accuracy, minimising the risks of misdiagnosis and missed diagnosis, and improving the detection of minimal-change gastric lesions. However, air bubbles and mucus in the stomach often reduce gastroscopic field visibility, leading to missed diagnoses and prolonged examination times. Pretreatment with defoaming agents and mucolytic agents enhances gastroscopic field visibility.2 Pronase, a proteolytic enzyme isolated from the culture filtrate of Streptomyces griseus, effectively cleaves the peptide bonds of glycoproteins, thereby dissolving and eliminating gastric mucus.3 This study aimed to evaluate the impact of pronase on the detection rate of precancerous lesions and early cancer, clarifying its utility in early gastric cancer screening. The findings will provide foundational evidence for the incorporation of pronase in endoscopic screening for upper gastrointestinal tract cancers and clinical diagnostic examinations.

Methods

Participants

This study enrolled 1450 individuals aged 40 to 70 years from a community population who participated in the 2020-2021 Upper Gastrointestinal Cancer Screening Programme in Wuwei, Gansu Province, China. The inclusion criteria were: (1) ability to cooperate with the gastroscopic procedure; (2) ability to discontinue anticoagulant medications 1 week prior to endoscopy; and (3) voluntary participation and provision of written informed consent. The exclusion criteria were: (1) contraindications to gastroscopy; (2) severe heart disease or heart failure; (3) severe respiratory disease; (4) posterior pharyngeal abscess or severe spinal deformity; (5) other serious illnesses or physical conditions that precluded tolerance of endoscopy; and (6) bleeding tendency.

Gastroscopy examinations

Using a random number table, all 1450 participants from the community population were randomly assigned to either an experimental group (n=725) or a control group (n=725). All participants underwent gastroscopy and tissue biopsy. In the experimental group, 1 sachet (20 000 U) of pronase (Beijing Tide Pharmaceutical, Beijing, China) and 1 g of sodium bicarbonate were dissolved in 50 to 80 mL of drinking water (20-40°C) by shaking. The solution was orally administered 15 to 30 minutes before gastroscopy (GIF-H290; Olympus, Tokyo, Japan). Dimethicone was also given orally to lubricate the cavity and remove gastric bubbles. To ensure that pronase reached all areas of the stomach, participants laid flat on a bed under a nurse’s guidance, then turned sideways three to five times. Subsequently, routine gastroscopy was performed. In the control group, participants received oral dimethicone 15 to 30 minutes before routine gastroscopy (GIF-H290).

The gastroscopy examinations were performed by two physicians holding the title of associate chief physician or higher, each having >10 years of experience in gastroscopy. The visibility of each part of the visual field was evaluated during the procedure; pathological examinations were conducted on tissue biopsies collected from minimal-change lesions.

Observation indicators

Endoscopic visibility scores were compared between the two groups. Scoring criteria were as follows4: 1 point, no mucus; 2 points, a small amount of mucus but no blurring of the visual field; 3 points, a large amount of mucus with a blurred visual field, requiring <30 mL of water for rinsing; and 4 points, very thick mucus with a blurred visual field, requiring ≥30 mL of water for rinsing. Lower scores indicated better endoscopic visibility. To minimise errors during the scoring process, each visibility score was recorded as the average of scores assigned by the two physicians who performed gastroscopy. The lesion detection rate was defined as the percentage of subjects within a group in whom lesions were identified. Gastroscopy time was measured from entry of the gastroscope into the oesophagus until its removal. Adverse reactions included nausea, vomiting, difficulty breathing, facial flushing, and other symptoms.

Statistical analyses

R software (version 4.0.5) was used for statistical analysis. Quantitative data were expressed as mean±standard deviation; intergroup differences were analysed using independent sample t tests. Qualitative data were expressed as frequency and percentage; intergroup differences were assessed using the Chi squared test or Fisher’s exact test. Multivariable linear regression analysis was performed to evaluate the effect of group assignment on visibility scores after adjustment for confounding factors. Differences in early cancer detection rates between the two groups were analysed using multivariable binary logistic regression analysis. All statistical tests were two-sided, and P values <0.05 were considered statistically significant.

Results

A summary of the baseline characteristics of the experimental and control groups is provided in Table 1. Among the 1450 patients in the cohort, 416 (28.7%) had a family history of gastrointestinal disease, 172 (11.9%) had a history of smoking, 91 (6.3%) had a history of alcohol consumption, and 335 (23.1%) had a history of gastrointestinal disease. Significant differences between the two groups were observed in the proportions of patients with a history of smoking, alcohol consumption, and gastrointestinal disease.

Table 1. Baseline characteristics of the study groups

Average visibility scores for the oesophagus, cardia, gastric fundus, gastric body, gastric antrum, gastric angle, and duodenum were significantly lower in the experimental group than in the control group (P<0.001 for all comparisons) [Table 2]. The visibility of different regions of the stomach under gastroscopy substantially differed between the two groups (Fig).

Table 2. Gastroscopy visibility scores of the study groups

Figure. Images of each part of the stomach under gastroscopy: (a) oesophagus, (b) cardia, (c) fundus, (d) corpus, and (e) duodenum. Upper and lower images show experimental and control groups, respectively

Effect of pronase on visibility score

Multiple linear regression analysis was performed with the visibility score for each site as the dependent variable and group assignment as the independent variable; adjustments were conducted for sex, age, marital status, education level, smoking status, alcohol consumption, history of gastrointestinal disease, and family history of gastrointestinal disease. After adjustment for these confounding factors, the visibility scores for all regions of the stomach remained significantly higher in the control group than in the experimental group (P<0.001 for all visibility scores) [Table 3].

Table 3. Effect of pronase on visibility score

Lesion and early cancer detection rates

Chi squared test analyses revealed that the detection rates of precancerous lesions (including atrophic gastritis, intestinal metaplasia, and low-grade intraepithelial neoplasia5) and early cancer were significantly higher in the experimental group than in the control group (77.5% vs 62.5%; P<0.001) [Table 4].

Table 4. Rates of lesion detection in the study groups

Multivariable binary logistic regression analysis was performed with early cancer detection as the dependent variable and group assignment as the independent variable; adjustments were conducted for sex, age, marital status, education level, smoking status, alcohol consumption, history of gastrointestinal disease, and family history of gastrointestinal disease. The likelihood of early cancer detection was significantly higher in the experimental group compared with the control group, with an odds ratio of 3.840 (95% confidence interval=1.204-12.241; P=0.023) [Table 5].

Table 5. Comparison of early cancer detection rates between the study groups

Examination time

Average gastroscopy times were 6.52±2.51 minutes in the experimental group and 10.03±1.23 minutes in the control group. Gastroscopy time significantly differed between the two groups (t=33.81; P=0.001).

Adverse reactions

No adverse reactions, such as nausea, vomiting, dyspnoea, or facial flushing, were reported in either group.

Discussion

Currently, approximately 90% of primary gastric cancers in China are diagnosed at an advanced stage.6 The prognosis of affected patients is closely related to the timing of diagnosis and treatment. Despite surgical intervention, the 5-year survival rate for patients with advanced gastric cancer remains <30%.7 After treatment, the 5-year survival rate for patients with early gastric cancer exceeds 90%, and cure may be achieved.8 However, the rates of early diagnosis and treatment of gastric cancer in China are <10%, substantially lower than rates reported in Japan (70%) and South Korea (50%).9 In Wuwei, the incidence and mortality rates of gastric cancer remain among the highest in the country; gastric cancer ranks first among malignant tumours in the city.10 Screening for upper gastrointestinal cancer is one of the most effective methods for population-level detection of early-stage cancer. Since 2010, Wuwei Tumour Hospital has implemented an upper gastrointestinal cancer screening programme (endoscopy combined with tissue biopsy) in Wuwei. Improvements in the detection rates of precancerous lesions and upper gastrointestinal cancer are key objectives of this screening initiative.

Gastroscopy is currently a widely used method for the clinical diagnosis and treatment of gastrointestinal diseases. A clear endoscopic field of vision is essential for accurate diagnosis and effective treatment by endoscopists. To optimise gastroscopy outcomes and enhance visibility within the stomach, bubbles and mucus must be removed. The use of pronase in combination with defoaming agents is recommended by the Consensus on Early Gastric Cancer Screening and Endoscopic Diagnosis and Treatment in China11 and the Guidelines for Endoscopic Diagnosis of Early Gastric Cancer (2019 edition) developed by the Japan Gastroenterological Endoscopy Society.12

Lee et al13 demonstrated that administering pronase 10 to 20 minutes before gastroscopy significantly improved the visibility of the endoscopic visual field and reduced the number of water washes required. Similarly, a multicentre randomised controlled study by Liu et al14 indicated that the combination of pronase and dimethicone significantly enhanced the visibility of the upper gastrointestinal mucosa. Pronase has also been utilised in narrow-band imaging endoscopy. A randomised controlled study by Cha et al15 compared the effects of orally administering pronase and simethicone 10 minutes before narrow-band imaging endoscopy on mucosal visibility and diagnostic performance. The results showed that mucosal visibility within the proximal stomach was significantly better in the pronase group than in the simethicone group.15 In the present study, the visibility scores for all sites in patients who received pronase were approximately 1 point, indicating minimal mucus adhesion. After adjustment for confounding factors, multiple linear regression analysis confirmed that visibility scores remained significantly lower in the pronase group than in the control group at all sites; this finding further validated the effectiveness of pronase. The present study also revealed that the average endoscopic examination time was significantly shorter (approximately 5 minutes) in the pronase group than in the control group. This reduced examination time was attributed to the near-complete absence of mucus adhesion after pronase administration, which decreased the number of rinses needed during the procedure. The shorter examination also enhanced patient comfort and increased compliance for subsequent screenings.

Zhang et al16 and Gao et al17 conducted retrospective analyses of 25 314 patients who underwent gastroscopy at Nanfang Hospital of Southern Medical University and 166 260 patients at Bazhong Central Hospital, revealing early cancer detection rates of 0.2% and 0.62%, respectively. Zhang et al1 performed a follow-up analysis of individuals in Liangzhou District in Wuwei who underwent upper gastrointestinal cancer screening in 2017; they observed an early cancer detection rate of 2.8%.1 In the present study, lesion detection rates for the experimental and control groups were 77.5% and 62.5%, respectively; corresponding early cancer detection rates were 3.0% and 2.1%. These percentages align with findings from the previous study in Wuwei1 and are substantially higher than those reported for other regions.16 17 The present results suggest that in Wuwei, a region displaying one of the highest incidences of upper gastrointestinal cancer in China, early cancer screening should be actively promoted. Furthermore, the detection rates of precancerous lesions and early cancer can be improved by using endoscopy combined with tissue biopsy.

The efficacy of pronase in improving the endoscopic visual field is well established, but studies investigating its impacts on the detection rates of precancerous lesions and early cancer have yielded inconsistent results.14 18 19 Chen et al18 conducted a randomised controlled trial that enrolled older patients undergoing gastroscopy; they found that the detection rate of minimal-change lesions was higher in the pronase group than in the control group (45.2% vs 27.5%; P<0.05).18 Lee et al19 demonstrated that the use of pronase when rinsing a lesion during endoscopy significantly increased the tissue depth of endoscopic biopsies and improved the anatomical localisation of biopsy sites, thereby enhancing the accuracy of disease diagnosis. In the present study, the detection rates of precancerous lesions and early cancer were significantly higher in the experimental group than in the control group (P<0.001). After adjustment for confounding factors, multivariable logistic regression showed that the likelihood of detecting early cancer was significantly greater in the experimental group than in the control group (odds ratio=3.840; P=0.023) [Table 5]. This finding indicates that pronase pretreatment before gastroscopy can enhance the detection rates of precancerous lesions and early cancer. The enhancement may be attributed to the clear visual field provided by pronase, which facilitates accurate selection of biopsy sites and improves recognition of minimal-change lesions. Gastroscopy physicians have substantial daily workloads and manage large numbers of patients requiring treatment. The use of pronase reduced the time required for endoscopy, potentially improving patient compliance with clinical microscopy.

Limitations

As an early cancer screening study, this investigation had a relatively small sample size; therefore, the findings require further validation in large-scale clinical studies. Cluster randomisation was used in this study, leading to baseline differences between groups; however, adjustments for these factors were included in the statistical analyses. The gastroscopy procedures were performed by highly skilled endoscopists. The generalisability of the findings to all endoscopists warrants additional investigation.

Conclusion

Pronase pretreatment before gastroscopy improves visual field clarity, reduces examination time, increases the detection rates of precancerous lesions and early cancer, and demonstrates good safety. This approach is beneficial for early cancer screening in regions with a high incidence of upper gastrointestinal cancer. The practical value of this method requires confirmation in large-scale clinical studies.

Author contributions

Concept or design: Z Wu, S Li, G Wang.
Acquisition of data: L Lu, G Zhao, J Liu, S Li.
Analysis or interpretation of data: T Qin.
Drafting of the manuscript: Z Zhang.
Critical revision of the manuscript for important intellectual content: Z Wu.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Funding/support

This research was supported by the National Key Research and Development Program of China (Ref No.: 2017YFC0908302). The funder had no role in study design, data collection, analysis, interpretation, or manuscript preparation.

Ethics approval

This research was approved by the Medical Ethics Committee of Wuwei Cancer Hospital, Wuwei, Gansu, China (Ref No.: 2019-Ethical review-11). The trial was registered with the Chinese Clinical Trial Registry (Ref No.: ChiCTR2200064855). Informed consent was obtained from all study participants, including consent for the publication of their anonymised data and clinical photos.

References

1. Zhang Z, Wu Z, Lu L, et al. Analysis of the upper gastrointestinal cancer screening and follow-up results in Liangzhou District of Wuwei City from 2009 to 2017 [in Chinese]. Chin J Cancer Prev Treat 2019;23:1750-5.

2. Choi IJ. Gastric preparation for upper endoscopy. Clin Endosc 2012;45:113-4. Crossref

3. Kim GH, Cho YK, Cha JM, Lee SY, Chung IK. Effect of pronase as mucolytic agent on imaging quality of magnifying endoscopy. World J Gastroenterol 2015;21:2483-9. Crossref

4. Beg S, Ragunath K, Wyman A, et al. Quality standards in upper gastrointestinal endoscopy: a position statement of the British Society of Gastroenterology (BSG) and Association of Upper Gastrointestinal Surgeons of Great Britain and Ireland (AUGIS). Gut 2017;66:1886-99. Crossref

5. Gomceli I, Demiriz B, Tez M. Gastric carcinogenesis. World J Gastroenterol 2012;18:5164-70. Crossref

6. Committee of Laboratory Medicine of Chinese Association of Integrative Medicine. Chinese Expert Consensus on Detection Technologies for Early-stage Gastric Cancer Screening [in Chinese]. Chin J Lab Med 2023;46:347-59.

7. Katai H, Ishikawa T, Akazawa K, et al. Five-year survival analysis of surgically resected gastric cancer cases in Japan: a retrospective analysis of more than 100,000 patients from the nationwide registry of the Japanese Gastric Cancer Association (2001-2007). Gastric Cancer 2018;21:144-54. Crossref

8. Sumiyama K. Past and current trends in endoscopic diagnosis for early-stage gastric cancer in Japan. Gastric Cancer 2017;20(Suppl 1):20-7. Crossref

9. Ren W, Yu J, Zhang Z, Song Y, Li Y, Wang L. Missed diagnosis of early gastric cancer or high-grade intraepithelial neoplasia. World J Gastroenterol 2013;19:2092-6. Crossref

10. Lu L, Nie P, Zhang Z. Analysis of incidence and mortality of stomach cancer from 2011 to 2015 in Wuwei City, Gansu Province [in Chinese]. China Cancer 2020;29:677-81.

11. Chinese Society of Digestive Endoscopy; Chinese Anti-Cancer Association The Society of Tumor Endoscopy. Chinese Consensus on Screening and Endoscopic Diagnosis and Management of Early Gastric Cancer (Changsha, April 2014) [in Chinese]. Chin J Gastroenterol 2014;19:408-27.

12. Yao K, Uedo N, Kamada T, et al. Guidelines for endoscopic diagnosis of early gastric cancer. Dig Endosc 2020;32:663-98. Crossref

13. Lee GJ, Park SJ, Kim SJ, Kim HH, Park MI, Moon W. Effectiveness of premedication with pronase for visualization of the mucosa during endoscopy: a randomized, controlled trial. Clin Endosc 2012;45:161-4. Crossref

14. Liu X, Guan CT, Xue LY, et al. Effect of premedication on lesion detection rate and visualization of the mucosa during upper gastrointestinal endoscopy: a multicenter large sample randomized controlled double-blind study. Surg Endosc 2018;32:3548-56. Crossref

15. Cha JM, Won KY, Chung IK, Kim GH, Lee SY, Cho YK. Effect of pronase premedication on narrow-band imaging endoscopy in patients with precancerous conditions of stomach. Dig Dis Sci 2014;59:2735-41. Crossref

16. Zhang Q, Chen Z, Chen C, et al. Training in early gastric cancer diagnosis improves the detection rate of early gastric cancer: an observational study in China. Medicine (Baltimore) 2015;94:e384. Crossref

17. Gao Z, Liang S, Li M, et al. Clinicopathological features and trends of 1025 cases of early gastric cancer, 2006-2020 [in Chinese]. J Cancer Control Treat 2021;34:649-54.

18. Chen L, Feng Y, Wang W, Zheng P. Clinical value of pronase combined with sodium bicarbonate in gastroscopy of elderly patients [in Chinese]. Zhejiang JITCWM 2018;28:225-7.

19. Lee SY, Han HS, Cha JM, Cho YK, Kim GH, Chung IK. Endoscopic flushing with pronase improves the quantity and quality of gastric biopsy: a prospective study. Endoscopy 2014;46:747-53. Crossref

Liver- and tumour-specific indicators predicting suboptimal survival following repeat transarterial chemoembolisation in patients with hepatocellular carcinoma

Hong Kong Med J 2025 Feb;31(1):32–40

https://doi.org/10.12809/hkmj2311208

ORIGINAL ARTICLE

Liver- and tumour-specific indicators predicting suboptimal survival following repeat transarterial chemoembolisation in patients with hepatocellular carcinoma

LM Chen, PhD^1,2,3,4; Simon CH Yu, MB, BS, FHKAM (Radiology)^1,2; Leung Li, MB, ChB, FRCP⁵; Edwin P Hui, MB, ChB, FHKAM (Medicine)⁵; Winnie Yeo, MB, BS, FHKAM (Medicine)^5,6; Stephen L Chan, MB, BS, FHKAM (Medicine)^5,6

¹ Department of Imaging and Interventional Radiology, Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong SAR, China

² Vascular and Interventional Radiology Foundation Clinical Science Centre, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong SAR, China

³ Department of Medical Ultrasonics, The Sixth Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China

⁴ Biomedical Innovation Center, The Sixth Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China

⁵ Department of Clinical Oncology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong SAR, China

⁶ State Key Laboratory of Translational Oncology, China

Corresponding author: Prof Simon CH Yu (simonyu@cuhk.edu.hk)

Full paper in PDF

Abstract

Introduction: This study explored liver- and tumour-specific indicators predictive of suboptimal survival outcomes following repeat transarterial chemoembolisation (TACE) in intermediate-stage hepatocellular carcinoma (HCC) patients after an initial TACE.

Methods: This study included 300 HCC patients who underwent TACE treatment. Based on whether persistent albumin–bilirubin (ALBI) grade deterioration (PABD) occurred after the initial TACE, defining as a shift in ALBI grade to a higher grade from baseline without recovery within 90 days, patients were divided into PABD and non-PABD groups. Overall survival of non-PABD and PABD groups according to subgroups stratified by baseline ALBI grade and tumour burden was compared with that of patients receiving only sorafenib or supportive care during the same period.

Results: Repeat TACE provided a survival benefit over systemic therapy or supportive care for patients in all post-TACE non-PABD or most PABD subgroups, regardless of baseline liver condition (ie, modified albumin–bilirubin [mALBI] grade and tumour burden). This benefit was absent in two subgroups among patients who developed PABD after the initial TACE, namely, (1) those with a baseline liver condition of mALBI grade 1 or 2a and tumour burden exceeding the up-to-11 criteria, and (2) those with a baseline liver condition of mALBI grade 2b, regardless of tumour burden.

Conclusion: Repeat TACE is not recommended for patients with persistent liver function deterioration after the initial TACE, particularly those exhibiting suboptimal baseline liver function or excessive tumour burden. Understanding the liver condition and tumour burden in HCC patients may assist clinicians in planning optimal treatment strategies, leading to better prognosis.

New knowledge added by this study

The identification of objective and specific indicators predictive of suboptimal survival outcomes following repeat transarterial chemoembolisation (TACE) would be clinically valuable.
The survival benefit of repeat TACE was not significant in two subgroups of patients who developed persistent albumin–bilirubin (ALBI) grade deterioration after the initial TACE, namely, (1) those with a baseline liver condition of modified albumin–bilirubin (mALBI) grade 1 or 2a and tumour burden exceeding the up-to-11 criteria, and (2) those with a baseline liver condition of mALBI grade 2b, regardless of tumour burden.

Implications for clinical practice or policy

Liver function changes after initial TACE combined with tumour burden could serve as indicators to select patients suitable for repeat TACE.
Repeat TACE is not recommended for patients with persistent liver function deterioration and a baseline liver condition of mALBI grade 1 or 2a and tumour burden exceeding the up-to-11 criteria, or for those with a baseline liver condition of mALBI grade 2b, regardless of tumour burden.

[Abstract in Chinese]

Introduction

Hepatocellular carcinoma (HCC) imposes a substantial cancer burden worldwide; its incidence rate in 2020 was ranked seventh, whereas its mortality rate was ranked second.1 Transarterial chemoembolisation (TACE) is commonly used as a first-line treatment for patients with intermediate-stage HCC, preserved liver function, and good performance status.2 3

Liver function deterioration occurs in 15.1% to 52% of patients after TACE4 5 6 7 8 9; among these patients, 3% to 31% experience chronic or irreversible liver function deterioration.5 6 7 9 Patients with post-TACE liver function deterioration may have a suboptimal long-term prognosis.5 8 10 Repeat TACE is indicated when residual tumour remains or when a new tumour is detected after the initial TACE.2 Patients with tumours refractory to TACE are preferably treated with systemic therapy; switching to such therapy has demonstrated a survival benefit and better liver function preservation relative to continued TACE.11 12

Liver condition is crucial to the clinical outcome of repeat TACE. Patients with suboptimal liver function are more likely to experience irreversible liver function deterioration after repeat TACE, leading to suboptimal survival outcomes. Such patients also exhibit risks of reduced treatment efficacy and compromised safety during subsequent treatment with systemic therapy. In patients with HCC, liver condition is inevitably linked to tumour burden; liver function deterioration occurs more frequently in those with a high tumour burden.5 13

The identification of objective and specific indicators predictive of suboptimal survival outcomes following repeat TACE would be clinically valuable because such indicators could guide decisions regarding whether to pursue repeat TACE or switch to systemic therapy. We hypothesised that specific indicators based on liver condition and tumour burden, predictive of suboptimal survival outcomes following repeat TACE, could be identified. In this study, we sought to identify liver- and tumour-specific indicators predictive of suboptimal survival outcomes with repeat TACE relative to sorafenib or supportive care (SC) in patients who had received an initial TACE.

Methods

All patients presenting to our institution with unresectable HCC between January 2005 and December 2019 who met the eligibility criteria were recruited. Inclusion criteria consisted of treatment-naïve unresectable HCC confirmed by biopsy or contrast-enhanced imaging demonstrating typical enhancement features, Barcelona Clinic Liver Cancer stage B disease, and treatment with one of three options: TACE, sorafenib, or SC. Exclusion criteria were age <18 years, intrahepatic tumours with vascular invasion, extrahepatic metastases, liver function classified as albumin–bilirubin (ALBI) grade 3, or incomplete post-TACE liver function data. According to standard practice at our institution during the study period, patients with unresectable intermediate-stage HCC and no contraindication to TACE were prioritised for TACE. Patients who refused TACE were treated with sorafenib; those who declined both treatments received SC.

Liver condition indicator

Liver condition was assessed using the modified albumin–bilirubin (mALBI) grade.14 The grade was defined by the ALBI score, which was calculated using the following equation: log₁₀ (bilirubin [in μmol/L])×0.66+albumin [in g/L]×(-0.085). Patients were categorised into four grades: 1 (ALBI score ≤-2.60), 2a (ALBI score >-2.60 and ≤-2.27), 2b (ALBI score >-2.27 and ≤-1.39), and 3 (ALBI score >-1.39). Post–transarterial chemoembolisation liver condition was classified into three categories based on post-TACE ALBI grade deterioration, defined as a shift to a higher grade from baseline following TACE, such as from grade 1 to grade 2-3, grade 2a to 2b-3, or grade 2b to 3. No ALBI grade deterioration (NABD) was regarded as the lack of a shift to a higher ALBI grade after TACE. Temporary ALBI grade deterioration (TABD) constituted ALBI grade deterioration that resolved within 90 days after TACE. Persistent ALBI grade deterioration (PABD) was defined as ALBI grade deterioration that did not resolve within 90 days after TACE. Patients in NABD and TABD groups were categorised as non- PABD group.

Tumour burden indicators

Tumour burden was assessed using the up-to-7 and up-to-11 criteria, defined as the sum of the tumour number and the largest tumour diameter in centimetres, with thresholds set at 7 and 11, respectively. Tumour burden was subclassified into four categories: within or beyond the up-to-7 or up-to-11 criteria.

Study design

At our institution, it was standard practice for patients initially treated with TACE to receive repeat TACE if residual or recurrent intrahepatic tumours were present, until a contraindication to TACE occurred. Contraindications included an Eastern Cooperative Oncology Group performance status score >2 or a Child-Pugh score >7, regardless of liver condition changes following the initial TACE. Assuming that patients with PABD after the initial TACE have a higher risk of further liver damage and worse survival outcomes if subjected to repeat TACE, such patients were targeted in this study. The overall survival (OS) of patients with or without PABD after the initial TACE was compared with the OS of patients receiving only sorafenib or SC during the same period. Among patients with or without post-TACE PABD, we identified subgroups with baseline mALBI grade and tumour burden who showed no survival benefit over sorafenib or SC; these patients were considered unsuitable for repeat TACE. Overall survival was calculated from the date of TACE or sorafenib initiation to the date of death from any cause. For patients who received SC, OS was calculated from the date of HCC diagnosis to the date of death from any cause. Censoring was applied to patients who were lost to follow-up, underwent subsequent liver resection, or were last known to be alive.

Transarterial chemoembolisation

The TACE procedure was performed under local anaesthesia and guided by digital subtraction angiography. An emulsion consisting of aqueous cisplatin (Platosin; Pharmachemie BV, Haarlem, the Netherlands) and ethiodised oil in a 1:1 volume ratio was delivered transarterially into the tumour vasculature until flow stagnation occurred or a maximum dose of 40 mL emulsion was reached. Tumour-feeding arteries were subsequently embolised using 5 to 10 mL of gelatin sponge. The completeness of the procedure was verified using digital subtraction angiography, with or without non-contrast multiplanar computed tomography (CT).

Systemic therapy

Oral sorafenib was administered twice daily at a standard dose of 400 mg. Dose adjustments or drug discontinuation were performed at the discretion of the oncologist based on patient tolerance.

Statistical analysis

Categorical variables are presented as numbers (percentages) and continuous variables are presented as medians (interquartile ranges). The Chi squared test was used to compare categorical data. The Mann-Whitney U test or Kruskal–Wallis test was performed for comparisons of continuous data. Differences in OS between subgroups were analysed using the log-rank test and hazard ratios (HRs) with 95% confidence intervals (CIs). Interaction terms were included to evaluate whether the survival benefit of the post-TACE PABD or non-PABD group over the sorafenib or SC group varied across subgroups. P values <0.05 were considered statistically significant. Data analysis was performed using SPSS (Windows version 25.0; IBM Corp, Armonk [NY], United States).

Results

Study participants

In total, 300 treatment-naïve patients with HCC received TACE. The median age was 65 years (interquartile range, 56-72); the cohort included 255 men and 45 women. After the first TACE, 235 of 300 patients experienced ALBI deterioration: 154 exhibited TABD and 81 displayed PABD. The demographics of patients with NABD, TABD, and PABD are listed in Table 1. The OS was similar for patients with NABD and TABD (22.40 vs 23.83 months), indicating that TABD did not adversely affect treatment outcomes. Therefore, patients with NABD and TABD were combined into the non-PABD group. The demographics of patients in non-PABD group and PABD group were compared to sorafenib group and SC group, as listed in Table 2. Patients in non-PABD group and PABD group had significantly better OS than those in sorafenib and SC group (23.13, 8.03, 5.11, and 2.57 months, respectively).

Table 1. Demographics of patients with different albumin–bilirubin deterioration statuses after the initial transarterial chemoembolisation

Table 2. Demographics of patients with different albumin–bilirubin deterioration statuses after the initial transarterial chemoembolisation relative to those receiving sorafenib or supportive care

Overall survival

Patients with post–transarterial chemoembolisation persistent albumin–bilirubin grade deterioration versus sorafenib in subgroups

Online supplementary Figure 1 illustrates the median OS of patients with post-TACE PABD relative to patients treated with sorafenib. Patients receiving TACE who developed post-TACE PABD had significantly longer median OS than those receiving sorafenib in subgroups within and beyond the up-to-7 criteria (19.63 vs 5.17 months; P=0.019 and 7.63 vs 5.11 months; P=0.030, respectively).

A significantly longer median OS was observed in patients receiving TACE who developed post-TACE PABD relative to those receiving sorafenib in the subgroup within the up-to-11 criteria (10.20 vs 5.37 months; P=0.016). However, this difference was not significant in the subgroup beyond the up-to-11 criteria (8.00 vs 4.94 months; P=0.083). Similarly, OS was significantly improved in the post-TACE PABD group relative to the sorafenib group within the mALBI grade 1 or 2a subgroup (11.50 vs 6.60 months; P=0.001). However, no significant difference was observed in the mALBI grade 2b subgroup (3.47 vs 4.39 months; P=0.517) [online supplementary Fig 1].

Based on stratification according to mALBI grade and the up-to-7 criteria, patients receiving TACE who developed post-TACE PABD had significantly longer median OS relative to those receiving sorafenib in the subgroup with mALBI grade 1 or 2a and within the up-to-7 criteria (29.57 vs 5.17 months; P=0.003) and the subgroup with mALBI grade 1 or 2a and beyond the up-to-7 criteria (10.57 vs 6.60 months; P=0.020). However, OS was not significantly improved in the subgroup with mALBI grade 2b and within the up-to-7 criteria (6.40 vs 4.39 months; P=0.071) or in the subgroup with mALBI grade 2b and beyond the up-to-7 criteria (3.07 vs 4.39 months; P=0.891). The interaction between treatment effects in subgroups stratified according to mALBI grade and the up-to-7 criteria had a 5% level of significance, with a tendency of a significant interaction that warrants further studies (P=0.058) [online supplementary Fig 1].

Based on stratification according to mALBI grade and the up-to-11 criteria, patients receiving TACE who developed post-TACE PABD had significantly longer median OS relative to those receiving sorafenib in the subgroup with mALBI grade 1 or 2a and within the up-to-11 criteria (13.37 vs 5.76 months; P=0.004). However, OS was not significantly improved in the subgroup with mALBI grade 1 or 2a and beyond the up-to-11 criteria (11.50 vs 6.60 months; P=0.061), the subgroup with mALBI grade 2b and within the up-to-11 criteria (5.07 vs 4.52 months; P=0.313), or the subgroup with mALBI grade 2b and beyond the up-to-11 criteria (3.07 vs 4.10 months; P=0.316). The interaction between treatment effects in subgroups stratified according to mALBI grade and the up-to-11 criteria had a 5% level of significance, with a tendency of a significant interaction that warrants further studies (P=0.071) [online supplementary Fig 1].

Patients with post–transarterial chemoembolisation persistent albumin–bilirubin grade deterioration versus sorafenib in subgroups

The median OS of patients who developed post-TACE PABD relative to those receiving SC is shown in online supplementary Figure 2. Patients receiving TACE who developed post-TACE PABD had significantly longer median OS compared with those receiving SC in the subgroup with mALBI grade 1 or 2a and within the up-to-7 criteria (29.57 vs 15.38 months; P=0.036) and the subgroup with mALBI grade 1 or 2a and beyond the up-to-7 criteria (10.57 vs 3.32 months; P<0.001). However, no significant improvement in OS was observed in the subgroup with mALBI grade 2b and within the up-to-7 criteria (6.40 vs 5.40 months; P=0.266) or in the subgroup with mALBI grade 2b and beyond the up-to-7 criteria (3.07 vs 2.18 months; P=0.051).

Patients receiving TACE who developed post- TACE PABD also had significantly longer median OS relative to those receiving SC in the subgroup with mALBI grade 1 or 2a and within the up-to-11 criteria (13.37 vs 4.29 months; P=0.035) and the subgroup with mALBI grade 1 or 2a and beyond the up-to-11 criteria (11.50 vs 3.32 months; P=0.001). However, no significant improvement in OS was observed in the subgroup with mALBI grade 2b and within the up-to-11 criteria (5.07 vs 2.57 months; P=0.084) or in the subgroup with mALBI grade 2b and beyond the up-to-11 criteria (3.07 vs 2.08 months; P=0.269) [online supplementary Fig 2].

Patients with post–transarterial chemoembolisation non-persistent albumin–bilirubin grade deterioration versus sorafenib or supportive care in subgroups

Significantly longer median OS was observed among patients in the non-PABD group after TACE relative to those receiving sorafenib (all P<0.001) [online supplementary Fig 3] or SC in all subgroups (all P<0.001, except for the subgroup with mALBI grade 1 or 2a and within the up-to-7 criteria, which displayed a P value of 0.012) [online supplementary Fig 4] stratified according to various criteria.

Discussion

Principal findings

This study demonstrated that repeat TACE provided a survival benefit over systemic therapy or SC for patients who developed TABD or PABD after the first TACE, regardless of baseline liver condition (according to ALBI grade, tumour burden, or liver function). However, this benefit was absent in the following two subgroups among patients who developed PABD after the first TACE: (1) those with a baseline liver condition of mALBI grade 1 or 2a and tumour burden exceeding the up-to-11 criteria, and (2) those with a baseline liver condition of mALBI grade 2b, regardless of tumour burden. These two subgroups could serve as specific indicators to guide the decision against prescribing repeat TACE for individual patients, based on their baseline liver condition, tumour burden, and occurrence of PABD after the initial TACE. In such cases, the treatment outcomes of repeat TACE are unlikely to differ from those of sorafenib or SC. Notably, there was a 5% level of significance, with a tendency of a significant interaction that warrants further studies.

Current knowledge of previous studies

Liver function deterioration after TACE is associated with worsened long-term survival.5 8 10 Patients with no increase in Child-Pugh score 1 month after TACE had significantly better survival rates than those with an increased Child-Pugh score at the same time point (84.5% vs 44.4%, 43.75% vs 18.5%, and 8.3% vs 0% for 1-year, 2-year, and 3-year survivals, respectively).8 The extent of liver function deterioration after TACE also impacts survival outcomes. The median OS was significantly longer in patients with ALBI grade migration to grade 2 than in patients with migration to grade 3 during both the acute phase (30.9 months vs 8.9 months; P<0.001) and the chronic phase (30.9 months vs 5.7 months; P<0.001).5 Higher tumour burden is linked to liver function deterioration and worse survival outcomes after TACE.15 16 17 Based on the 7-11 criteria, patients with high tumour burden experienced significantly higher rates of liver function deterioration (24.4% vs 14.9% or 14.4%) and shorter median survival (11.9 vs 22.3 or 33.1 months) relative to those with low or intermediate tumour burden.17 Currently, there are no reports in the literature concerning studies that identified liver- and tumour-specific indicators to predict survival benefits of repeat TACE.

Implications for clinical practice

Repeat TACE can damage liver function and worsen long-term survival. If a patient’s liver function is irreversibly and severely impaired by repeat TACE, the opportunity to switch to systemic therapy may be missed. To maximise survival benefits, the decision to repeat TACE, discontinue TACE, or transition to systemic therapy should be carefully considered and individualised. Two scoring systems have been developed to guide retreatment strategies,18 19 but universal validation of their predictive value is needed. Studies have shown that these systems are ineffective in terms of supporting decision-making for sequential treatment.20 21

Most patients who develop TABD are able to spontaneously recover their baseline liver function. In this study, similar median OS was observed among patients with TABD and NABD (23.83 vs 22.40 months). Transarterial chemoembolisation provided a statistically significant survival benefit for patients within the non-PABD group, regardless of tumour burden, relative to those receiving sorafenib or SC. This finding suggests that TABD has minimal impact on survival benefit or long-term prognosis after TACE, and repeat TACE remains feasible in these patients with reversed or reversible liver function. Based on the present findings, repeat TACE is not recommended for patients with PABD and a baseline liver condition of mALBI grade 2b, regardless of tumour burden, because survival outcomes in this subgroup are unlikely to be superior to those achieved with sorafenib or SC. For the same reason, repeat TACE is not recommended for patients with PABD, a baseline liver condition of mALBI grade 1 or 2a, and tumour burden beyond the up-to-11 criteria. Systemic therapy is preferred for this subgroup, considering that its effectiveness is likely maximised in patients with better liver function (eg, those with ALBI grade 1 or mALBI grade 2a, as stated in an expert consensus).22

Limitations

We acknowledge that sorafenib is no longer first-line systemic therapy for HCC. Regimens such as lenvatinib23 or atezolizumab-bevacizumab24 have been associated with significantly better OS relative to sorafenib. We recognise that the use of sorafenib as a control was a limitation of this study. However, no alternative was available because a sufficiently large database with long-term clinical outcomes for newer systemic therapies was not accessible for the local population. The primary objective of this study was not to evaluate the role of sorafenib compared with TACE, but to use sorafenib as a control to identify specific liver and tumour indicators predictive of suboptimal survival outcomes after repeat chemoembolisation. These indicators are intended to serve as contraindications for repeat TACE in patients with the corresponding liver and tumour conditions. The use of a systemic drug with lower OS benefit, such as sorafenib, as a control might lead to overestimation of the value of repeat TACE and, consequently, to the identification of indicators under worse liver and tumour conditions. However, this observation does not compromise the validity of these indicators as criteria for contraindicating repeat TACE.

Other limitations of the study include the relatively small sample size in patient groups receiving sorafenib or SC. Patient numbers were further reduced in some subgroups after stratification according to liver function and tumour burden, which could introduce bias in survival comparisons. Serum alpha-fetoprotein (AFP) levels and tumour response after TACE were not analysed in this study. Considering that elevated AFP levels have been associated with ALBI deterioration, AFP may be partially represented in the baseline ALBI grade. The median time to Child-Pugh deterioration was significantly longer in patients who responded to the initial TACE than in those who were refractory to the initial TACE (55.9 vs 19.6 months).25 Most patients (22/27, 81.5%) ineligible for repeat TACE due to hepatic decompensation exhibited tumour progression at the time of TACE discontinuation.26 Target lesion progression has been associated with no survival improvement and an increased risk of liver dysfunction after repeat TACE.27 Based on findings in the above studies, poor tumour response may eventually lead to liver function deterioration. Although tumour response was not analysed in this study, it is reasonable to assume that tumour response varies according to treatment effectiveness. Given that treatment effectiveness is assumed to remain consistent under the same treatment protocol within a single centre, it may be argued that the overall effect of tumour response in individual patients was reflected in liver function deterioration.

Conclusion

This study found that repeat TACE is not recommended for patients with persistent liver function deterioration after the initial TACE, particularly those exhibiting suboptimal baseline liver function or excessive tumour burden. Understanding the liver condition and tumour burden in HCC patients may assist clinicians in planning optimal treatment strategies and improving patient prognosis.

Author contributions

Concept or design: SCH Yu.
Acquisition of data: LM Chen, L Li, EP Hui, W Yeo, SL Chan.
Analysis or interpretation of data: LM Chen, SCH Yu.
Drafting of the manuscript: LM Chen, SCH Yu.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Funding/support

This research was funded by the Vascular and Interventional Radiology Foundation. The funding body was not involved in the study design, data collection, analysis, interpretation, or manuscript preparation.

Ethics approval

This research was approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: 2020.672). The research was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonisation, Good Clinical Practice. The requirement for written informed patient consent was waived by the Committee due to the retrospective nature of the research.

Supplementary material

The supplementary material was provided by the authors, and some information may not have been peer reviewed. Accepted supplementary material will be published as submitted by the authors, without any editing or formatting. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by the Hong Kong Academy of Medicine and the Hong Kong Medical Association. The Hong Kong Academy of Medicine and the Hong Kong Medical Association disclaim all liability and responsibility arising from any reliance placed on the content.

References

1. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209-49. Crossref

2. Cheung TT, Kwok PC, Chan S, et al. Hong Kong consensus statements for the management of unresectable hepatocellular carcinoma. Liver Cancer 2018;7:40-54. Crossref

3. European Association for the Study of the Liver. EASL Clinical Practice Guidelines: management of hepatocellular carcinoma. J Hepatol 2018;69:182-236. Crossref

4. Min YW, Kim J, Kim S, et al. Risk factors and a predictive model for acute hepatic failure after transcatheter arterial chemoembolization in patients with hepatocellular carcinoma. Liver Int 2013;33:197-202. Crossref

5. Chi CT, Lee IC, Lee RC, et al. Effect of transarterial chemoembolization on ALBI grade in intermediate-stage hepatocellular carcinoma: criteria for unsuitable cases selection. Cancers (Basel) 2021;13:4325. Crossref

6. Hiraoka A, Kumada T, Kudo M, et al. Hepatic function during repeated TACE procedures and prognosis after introducing sorafenib in patients with unresectable hepatocellular carcinoma: multicenter analysis. Dig Dis 2017;35:602-10. Crossref

7. Miksad RA, Ogasawara S, Xia F, Fellous M, Piscaglia F. Liver function changes after transarterial chemoembolization in US hepatocellular carcinoma patients: the LiverT study. BMC Cancer 2019;19:795. Crossref

8. Kohla MA, Abu Zeid MI, Al-Warraky M, Taha H, Gish RG. Predictors of hepatic decompensation after TACE for hepatocellular carcinoma. BMJ Open Gastroenterol 2015;2:e000032. Crossref

9. Park KH, Kim JH, Choe WH, et al. Risk factors for liver function deterioration after transarterial chemoembolization refractoriness in Child-Pugh class A hepatocellular carcinoma patients. Korean J Gastroenterol 2020;75:147-56. Crossref

10. Sun Z, Li G, Ai X, et al. Hepatic and biliary damage after transarterial chemoembolization for malignant hepatic tumors: incidence, diagnosis, treatment, outcome and mechanism. Crit Rev Oncol Hematol 2011;79:164-74. Crossref

11. Ogasawara S, Ooka Y, Koroki K, et al. Switching to systemic therapy after locoregional treatment failure: definition and best timing. Clin Mol Hepatol 2020;26:155-62. Crossref

12. Piscaglia F, Ogasawara S. Patient selection for transarterial chemoembolization in hepatocellular carcinoma: importance of benefit/risk assessment. Liver Cancer 2018;7:104-19. Crossref

13. Yasui Y, Tsuchiya K, Kurosaki M, et al. Up-to-seven criteria as a useful predictor for tumor downstaging to within Milan criteria and Child-Pugh grade deterioration after initial conventional transarterial chemoembolization. Hepatol Res 2018;48:442-50. Crossref

14. Hiraoka A, Michitaka K, Kumada T, et al. Validation and potential of albumin–bilirubin grade and prognostication in a nationwide survey of 46,681 hepatocellular carcinoma patients in Japan: the need for a more detailed evaluation of hepatic function. Liver Cancer 2017;6:325-36. Crossref

15. Khisti R, Patidar Y, Garg L, Mukund A, Thomas SS, Sarin SK. Correlation of baseline portal pressure (hepatic venous pressure gradient) and indocyanine green clearance test with post–transarterial chemoembolization acute hepatic failure. J Clin Exp Hepatol 2019;9:447-52. Crossref

16. Siriwardana RC, Niriella MA, Dassanayake AS, et al. Factors affecting post-embolization fever and liver failure after trans-arterial chemo-embolization in a cohort without background infective hepatitis–a prospective analysis. BMC Gastroenterol 2015;15:96. Crossref

17. Hung YW, Lee IC, Chi CT, et al. Redefining tumor burden in patients with intermediate-stage hepatocellular carcinoma: the seven-eleven criteria. Liver Cancer 2021;10:629-40. Crossref

18. Sieghart W, Hucke F, Pinter M, et al. The ART of decision making: retreatment with transarterial chemoembolization in patients with hepatocellular carcinoma. Hepatology 2013;57:2261-73. Crossref

19. Adhoute X, Penaranda G, Naude S, et al. Retreatment with TACE: the ABCR SCORE, an aid to the decision-making process. J Hepatol 2015;62:855-62. Crossref

20. Arizumi T, Ueshima K, Iwanishi M, et al. Evaluation of ART scores for repeated transarterial chemoembolization in Japanese patients with hepatocellular carcinoma. Oncology 2015;89 Suppl 2:4-10. Crossref

21. Kloeckner R, Pitton MB, Dueber C, et al. Validation of clinical scoring systems ART and ABCR after transarterial chemoembolization of hepatocellular carcinoma. J Vasc Interv Radiol 2017;28:94-102. Crossref

22. Kudo M, Han KH, Ye SL, et al. A changing paradigm for the treatment of intermediate-stage hepatocellular carcinoma: Asia-Pacific primary liver cancer expert consensus statements. Liver Cancer 2020;9:245-60. Crossref

23. Kudo M, Finn RS, Qin S, et al. Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet 2018;391:1163-73. Crossref

24. Cheng AL, Qin S, Ikeda M, et al. Updated efficacy and safety data from IMbrave150: atezolizumab plus bevacizumab vs. sorafenib for unresectable hepatocellular carcinoma. J Hepatol 2022;76:862-73. Crossref

25. Maesaka K, Sakamori R, Yamada R, et al. Initial treatment response to transarterial chemoembolization as a predictive factor for Child-Pugh class deterioration prior to refractoriness in hepatocellular carcinoma. Hepatol Res 2020;50:1275-83. Crossref

26. Labeur TA, Takkenberg RB, Klümpen HJ, van Delden OM. Reason of discontinuation after transarterial chemoembolization influences survival in patients with hepatocellular carcinoma. Cardiovasc Intervent Radiol 2019;42:230-8. Crossref

27. Zhang YF, Guo RP, Ouyang HY, et al. Target lesion response predicts survival of patients with hepatocellular carcinoma retreated with transarterial chemoembolization. Liver Int 2016;36:1516-24. Crossref

Success rate of induction of labour in twin pregnancies relative to singleton pregnancies in a predominantly Chinese population

Hong Kong Med J 2025 Feb;31(1):24–31 | Epub 12 Feb 2025

https://doi.org/10.12809/hkmj2311333

ORIGINAL ARTICLE CME

Success rate of induction of labour in twin pregnancies relative to singleton pregnancies in a predominantly Chinese population

CK Wong, MB, ChB, FHKAM (Obstetrics and Gynaecology); Catherine MW Hung, MB, ChB, FHKAM (Obstetrics and Gynaecology); Vivian KS Ng, MB, ChB, FHKAM (Obstetrics and Gynaecology); WK Yung, MB, BS, FHKAM (Obstetrics and Gynaecology); WC Leung, MD, FHKAM (Obstetrics and Gynaecology); WL Lau, MB, BS, FHKAM (Obstetrics and Gynaecology)

Department of Obstetrics and Gynaecology, Kwong Wah Hospital, Hong Kong SAR, China

Corresponding author: Dr CK Wong (wck936@ha.org.hk)

Full paper in PDF

Abstract

Introduction: This study assessed the efficacy of induction of labour in twin pregnancies relative to singleton pregnancies within a predominantly Chinese patient population.

Methods: This retrospective case-matched cohort study included patients with twin pregnancies who underwent induction of labour at our institution in Hong Kong between 2012 and 2020. Patients with twin pregnancies were matched one-to-one with singleton pregnancies based on parity, maternal age, and the indication for induction of labour. The primary outcome was the mode of delivery. Secondary outcomes included the time from oxytocin infusion to delivery, indications for caesarean or instrumental delivery, and maternal and neonatal outcomes.

Results: In total, 160 women with twin pregnancies met the inclusion criteria and were matched with 160 singleton pregnancies. Caesarean section was performed in 42 patients (26.3%) with twin pregnancies and 27 patients (16.9%) with singleton pregnancies undergoing induction of labour. Patients with twin pregnancies had a significantly higher risk of caesarean section relative to those with singleton pregnancies (odds ratio=2.14, 95% confidence interval=1.14-4.04; P=0.024). Internal podalic version was required in 13.6% of cases for the vaginal delivery of the second twin. There was no significant difference between the groups in the time from oxytocin administration to vaginal delivery (P=0.143).

Conclusion: Despite a higher induction failure rate, about three quarters of twin pregnancy patients achieved successful vaginal deliveries. Our findings inform decision making for patients and obstetricians, emphasising the importance of training for internal podalic version to aid second twin delivery and reduce caesarean rates in twin pregnancies.

New knowledge added by this study

Approximately three-quarters of patients with twin pregnancies who underwent induction of labour achieved successful vaginal deliveries.
The failure rate of induction of labour was higher in twin pregnancies than in singleton pregnancies.
The probability of requiring a caesarean section for the second twin when the first twin is delivered vaginally is only 0.8% if experts in twin vaginal delivery are available. Internal podalic version was necessary in 13.6% of cases for the vaginal delivery of the second twin.

Implications for clinical practice or policy

Patients with twin pregnancies undergoing induction of labour should be counselled regarding the increased risk of unsuccessful labour induction relative to singleton pregnancies.
Proficient obstetricians skilled in internal podalic version should be readily available during the delivery of the second twin to improve the success rate of vaginal delivery for the second twin.
Greater emphasis should be placed on implementing training opportunities, simulation models, and practices for junior obstetricians to enhance proficiency in internal podalic version. These measures can facilitate second twin delivery and reduce the need for caesarean sections in second twin births.

[Abstract in Chinese]

Introduction

The global twin birth rate has increased by one-third since the 1980s, rising from 9.1 to 12 per 1000 deliveries, resulting in approximately 1.6 million twin pairs born annually.1 One major factor contributing to this trend is the growing use of assisted reproductive techniques in recent decades.1 Relative to singleton pregnancies, twin pregnancies are associated with higher incidences of maternal and fetal complications, which may require earlier delivery.2 3 Even in uncomplicated cases, the National Institute for Health and Care Excellence (NICE) guidelines recommend delivery at 37 weeks for dichorionic twin pregnancies and 36 weeks for monochorionic twin pregnancies.4 Consequently, earlier delivery is frequently required in twin pregnancies. The Twin Birth Study,5 a large multicentre randomised controlled trial published in 2013, demonstrated the safety of both vaginal and caesarean birth in twin pregnancies where the first twin presented in cephalic position at 32 weeks of gestation or later. These findings have supported an increase in vaginal deliveries for twin pregnancies through induction of labour.

The success rate, benefits, and complications associated with induction of labour in singleton pregnancies have been extensively studied.6 7 However, only a limited number of studies have compared the success rate of induction of labour in twin pregnancies relative to singleton pregnancies.8 9 10 11 According to Loscul et al8 and Okby et al,9 induction of labour in twin pregnancies increases the likelihood of caesarean section. In contrast, Fausett et al10 and Taylor et al11 reported that the risk of caesarean delivery in twin pregnancies is comparable to the risk in singleton pregnancies undergoing induction of labour. These conflicting findings may arise from variations in induction methods, ethnicity-related factors, selection biases, and differences in study designs.

To provide appropriate counselling to patients, obstetricians must understand the likelihood of vaginal delivery after induction of labour. Reliance on data from singleton deliveries to estimate this likelihood for twin pregnancies may be inappropriate due to inherent differences between twin and singleton pregnancies.12 Considering the current lack of robust evidence regarding induction of labour in twin pregnancies, this study aimed to evaluate the rate of caesarean section (including classical or lower segment caesarean sections) and associated outcomes in twin pregnancies undergoing induction of labour, compared with singleton pregnancies.

Methods

Study design

Our institution, a regional public hospital in Hong Kong, provides obstetric services for 3000 to 5000 deliveries annually. All cases of twin pregnancies are recorded in a specialised twin pregnancy clinic registry and managed by a dedicated team of obstetricians and midwives in the Twin Pregnancy Clinic. The medical professionals overseeing this clinic have specialised expertise in maternal fetal medicine.13 In accordance with departmental protocol, patients attend regular follow-up appointments and undergo ultrasound examinations. When a patient approaches term or requires earlier delivery, the attending obstetrician discusses the mode of delivery with the patient. For uncomplicated dichorionic-diamniotic and monochorionic-diamniotic twin pregnancies, vaginal delivery is encouraged if the first twin presents in cephalic position.

The same induction of labour protocol is applied to both twin and singleton pregnancies. Patients are admitted to the hospital and a cervical examination is conducted to assess the modified Bishop score. If the cervix is unfavourable with a modified Bishop score (Calder score) <6, cervical priming is performed using either dinoprostone tablets or a cervical ripening balloon (Cook Medical, Bloomington [IN], United States). If the cervix is favourable with a modified Bishop score ≥6, the patient is transferred to the labour ward for artificial rupture of membranes and administration of synthetic oxytocin. The induction of labour protocol used in this study aligns with NICE recommendations, except that the modified Bishop score was used instead of the Bishop score to assess cervical readiness for induction.14 15

Medical records of patients with twin pregnancies who underwent induction of labour and delivered at our institution between January 2012 and December 2020 were retrospectively identified using the International Classification of Diseases codes through the Clinical Data Analysis and Reporting System of Hospital Authority. The identified medical records were individually reviewed. Study participants were required to meet all of the following inclusion criteria: gestational age ≥24 weeks, intact membranes, and planned induction of labour. Exclusion criteria included premature rupture of membranes, labour in the latent or active phase, threatened preterm labour resulting in spontaneous labour, and intrauterine fetal death.

Each twin pregnancy patient who underwent induction of labour was matched with a singleton pregnancy patient at a 1:1 ratio in the same hospital during the same study period. Matching was based on specific criteria, including parity (nulliparous or multiparous),16 maternal age (advanced maternal age of ≥35 years, or not),16 17 and the indication for induction of labour.18 These criteria were selected to minimise confounding factors that could affect the success rate of induction of labour. To further reduce the impact of variations in medical practice during the study period, the singleton pregnancy patient with the delivery date closest to that of the twin pregnancy patient was selected.

For both twin and singleton pregnancies, demographic data, past obstetric history, parity, modified Bishop score, method of cervical priming, indication for induction of labour, and use of epidural analgesia were recorded. The primary outcome was the mode of delivery. Secondary outcomes included the time from oxytocin infusion to vaginal delivery or caesarean section, indications for caesarean section or instrumental delivery, and maternal and neonatal outcomes. Postpartum haemorrhage was defined as blood loss of ≥500 mL, regardless of the mode of delivery. Patients who underwent caesarean section for the second twin after vaginal delivery of the first twin were considered to have undergone caesarean section.

Statistical analyses

To calculate the required sample size, it was assumed that the incidences of caesarean section were 25% in the control group and 40% in the study group. The proportion of discordant pairs was assumed to be 0.45. A two-sided significance level of 0.05 was selected, and the study aimed to achieve a 1:1 comparison between the groups. Calculations in G*Power software (version 3.1.9.6; Erdfelder, Faul, & Buchner, Germany) indicated that a total sample size of 160 pairs would provide 80% power for the analysis.

Categorical variables are reported as numerator and denominator values (%), whereas continuous variables are presented as mean±standard deviation. McNemar’s test was used to analyse the primary outcome for twin pregnancies and their matched controls. Data for the matched pairs are presented in a 2×2 table, showing concordant and discordant study pairs. Odds ratios (ORs) were calculated as the ratio of discordant pairs, and the test statistic was derived from McNemar’s test. For the remaining outcomes, paired t tests, Wilcoxon signed rank test and analyses of variance were used to analyse continuous variables, whereas McNemar’s tests or Fisher’s exact tests were utilised for categorical variables when comparing the case and control groups. For non-matched data, the Chi squared test was applied for categorical data, and unpaired t tests were used for normally distributed continuous data.

Statistical analyses of data using McNemar’s test were performed with Epi Info (version 7.2.5.0; Centers for Disease Control and Prevention, Atlanta [GA], US). All other analyses were conducted with SPSS (Windows version 27.0; IBM Corp, Armonk [NY], US). Two-sided P values <0.05 were considered statistically significant.

Results

During the study period, 760 women with twin pregnancies were recorded out of 42 280 maternity cases. Of these, 160 women met the inclusion criteria for this study. The incidence of twin pregnancies was 1.8% and the rate of induction of labour in twin pregnancies was 21.1%.

The study group consisted of women with twin pregnancies who underwent induction of labour, whereas the control group comprised women with singleton pregnancies who delivered at the same hospital during the same period. The two groups were matched in terms of age, parity and indication for induction of labour, and were well balanced with respect to these matching factors. Patients with twin pregnancies had a significantly lower body mass index (21.1 kg/m² vs 22.2 kg/m²; P=0.008) and a significantly higher modified Bishop score (6.2 vs 5.4; P<0.001) relative to the control group. The mean gestational age at delivery was significantly earlier in twin pregnancies than in the control group (37.1 weeks vs 40.2 weeks; P<0.001). Other baseline characteristics, including ethnicity, prior caesarean section, and use of epidural anaesthesia, did not significantly differ between the two groups (Table 1).

Table 1. Baseline maternal characteristics

Success rate of induction of labour

Out of 160 pairs, 44 were discordant (ie, one member of the pair had a caesarean section and the other had a vaginal delivery), and 116 were concordant (ie, both members of the pair had either a caesarean section or a vaginal delivery) [Fig]. Patients with twin pregnancies who underwent induction of labour had a significantly higher risk of caesarean section relative to those with singleton pregnancies (OR=2.14, 95% confidence interval [CI]=1.14-4.04; P=0.024).

Figure. Mode of delivery results in matched twin and singleton pregnancies

Among the patients with twin pregnancies, there were 118 vaginal deliveries, 41 caesarean sections, and one case in which the first twin was delivered vaginally and the second twin was delivered by caesarean section. Among the patients with singleton pregnancies, there were 133 vaginal deliveries and 27 caesarean sections. Instrumental deliveries were performed in 32 patients with twin pregnancies and eight patients with singleton pregnancies (Table 2).

Table 2. Maternal outcomes

There was no significant difference between the groups in the time from oxytocin administration to vaginal delivery (P=0.143) or the time from oxytocin administration to caesarean section (P=0.054). In total, eight patients with twin pregnancies and three patients with singleton pregnancies delivered after dinoprostone insertion without requiring artificial rupture of membranes or oxytocin infusion. Three patients with twin pregnancies and one patient with a singleton pregnancy had an unfavourable cervix after repeated doses of dinoprostone; thus, caesarean section was performed (Table 2).

Obstetric outcomes

Twin pregnancies were associated with significantly greater blood loss relative to singleton pregnancies (median: 400 mL vs 250 mL; P<0.001) and a higher incidence of postpartum haemorrhage (35.0% vs 10.6%; P<0.001). However, there was no significant difference between the groups in blood transfusion rates (8.1% vs 3.8%; P=0.115). The aetiology of postpartum haemorrhage and need for second-line treatments were also comparable between the two groups (Table 2).

Mode of delivery in twin pregnancies

Among those 118 vaginal deliveries, 45 had vaginal cephalic deliveries of both twins, whereas 42 had a vaginal cephalic delivery of the first twin followed by a vaginal breech delivery of the second twin. Additionally, 32 patients required instrumental delivery with vacuum or forceps for at least one twin (Table 3).

Table 3. Mode of delivery in twin pregnancies (n=160)

In cases where vaginal breech delivery was required for the second twin, most babies were in breech presentation. Internal podalic version was performed in 16 patients (13.6%) to facilitate delivery of the second twin (Tables 3 and 4). Notably, even in five cases where the second twin was in cephalic presentation, internal podalic version was performed by manual upward displacement of the fetal head to expedite delivery due to fetal bradycardia or cord presentation.

Table 4. Vaginal delivery of the second twin (n=118)

Neonatal outcomes

When neonatal outcomes were compared between the two groups, no significant differences were observed in Apgar scores at 1 and 5 minutes or in rates of admission to neonatal intensive care units. However, both the first and second twins were significantly lighter in weight relative to neonates in singleton pregnancies (Table 5).

Table 5. Neonatal outcomes

Discussion

Primary outcomes

This case-control study, utilising matched controls, demonstrated that the rate of failed induction of labour was significantly higher in twin pregnancies than in singleton pregnancies. Nevertheless, 73.8% of patients with twin pregnancies achieved successful vaginal deliveries. This study represents the largest cohort investigation of its kind in a predominantly Chinese population and differs from previous studies conducted in Western countries.8 9 10 11 A previous study19 revealed that ethnic variation can influence the success of induction of labour; thus, our findings provide valuable insights for counselling and managing Chinese patients with twin pregnancies.

Comparison with previous studies

The literature on induction of labour in twin pregnancies compared with singleton pregnancies remains limited. The present findings are consistent with those reported by Loscul et al8 and Okby et al,9 both of which identified an increased risk of caesarean delivery after induction of labour in twin pregnancies. Loscul et al8 reported an adjusted OR of 1.8 (95% CI=1.4-2.2), whereas Okby et al9 reported an adjusted OR of 2.2 (95% CI=1.7-2.7). Similarly, the current study demonstrated an OR of 2.14, reinforcing the notion that induction of labour in twin pregnancies is associated with a higher rate of caesarean section relative to singleton pregnancies. However, limitations existed in these studies. For instance, the large cohort study by Loscul et al,8 which included 1995 twin deliveries and 2771 singleton deliveries, did not consider chorionicity, and the methods of induction were not described; considering the multicentre retrospective design of that study, interhospital variations may have existed in terms of induction methods, intrapartum assessment, and decisions regarding caesarean section. Furthermore, Okby et al9 included 191 twin deliveries and 25 913 singleton deliveries, but did not provide details regarding induction methods, cervical status prior to induction, Bishop score, or the chorionicity of twin pregnancies. Conversely, Fausett et al10 and Taylor et al11 included smaller cohorts of twin pregnancies (62 and 100 patients, respectively), and their findings may have been influenced by the small sample sizes. The method of random patient selection used in the control group of the study by Taylor et al11 may have introduced potential bias. Another factor potentially contributing to differences in findings among these studies is ethnic variation.

Physiological explanations

Physiological differences in the myometrium between twin and singleton pregnancies may explain the higher incidence of failed labour induction in twin pregnancies. Research has shown that myometrial activity in twin pregnancies is characterised by shorter and more frequent contractions compared with singleton pregnancies, particularly at term.20 Shortened contraction duration may result in ineffective and dysfunctional contractions, increasing the likelihood of failed labour induction. Additionally, the uterus undergoes greater distension and stretching in twin pregnancies. Physiological studies have indicated that increased myometrial stretching is associated with reduced uterine contraction in response to oxytocin stimulation.21 One potential mechanism for this phenomenon is that prolonged stretching enhances the expression or activity of TWIK-related K+ channels, which subsequently diminish myometrial contraction in response to oxytocin.21 Further physiological and molecular investigations are warranted to explore the differences between singleton and twin pregnancies in greater detail.

Clinical implications of secondary outcomes

In conjunction with the primary outcomes, the secondary outcomes of this study provide important clinical insights and have substantial implications. Notably, 26.3% of twin pregnancies delivering vaginally required a vaginal breech delivery for the second twin (Tables 3 and 4). Therefore, we recommend that senior obstetricians with expertise in internal podalic version and breech extraction be present during such deliveries.

This study identified one patient who required a caesarean section for the second twin after the first twin had been delivered vaginally. When the first twin was delivered vaginally in our cohort, the probability of caesarean section for the second twin was 0.8%. Patients should be informed of this potential risk prior to induction of labour. Previous studies have shown that the risk of caesarean section for the second twin after vaginal delivery of the first twin ranges from 4.3% to 10.7%.5 22 23 24 In our cohort, the percentage of caesarean deliveries for second twins was much lower than that in other series, including another retrospective study conducted in Hong Kong with the same ethnic population.22 23 24 This discrepancy may be attributed to selection bias because the present study included only patients undergoing induction of labour, whereas other studies included patients with both induction of labour and spontaneous onset of labour. Furthermore, elective induction of labour for twin pregnancies in our unit is typically scheduled during daytime hours. This practice ensures the availability of experienced staff proficient in internal podalic version, potentially improving the likelihood of successful vaginal delivery for the second twin. Our findings showed that 13.6% of cases involving second twin deliveries required internal podalic version, primarily due to transverse or oblique lie (Table 4). Even when the second twin presented in cephalic position (as observed in five cases), internal podalic version was required to expedite delivery because of complications such as cord presentation or fetal bradycardia. The presence of experienced staff skilled in performing internal podalic version can significantly increase the likelihood of achieving successful vaginal delivery for the second twin. At our hospital, vaginal twin deliveries during daytime hours are typically supervised by experienced obstetric consultants or associate consultants.

Strengths and limitations

To minimise the impact of variations in medical practice during the study period, we utilised a rigorous matching approach in which the singleton pregnancy patient with the delivery date closest to that of the twin pregnancy patient was selected. This approach effectively reduced the potential for confounding factors, including variations in medical practices, and ensured that the same induction of labour protocol was applied to both patient groups. Also, we recorded detailed information concerning chorionicity, indications for induction of labour, and the modified Bishop score. Finally, the induction of labour protocol used in this study aligns with NICE recommendations; thus, the results are applicable to other centres that use a similar protocol.

However, this study had some limitations. The earlier gestational age at delivery in the twin pregnancy group, as recommended by international guidelines4 even for uncomplicated twin pregnancies, may have affected the efficacy of induction of labour and influenced the overall outcomes. Additionally, patients with twin pregnancies had a higher initial modified Bishop score relative to those with singleton pregnancies. This difference may be due to selection bias because obstetricians often discourage vaginal delivery in patients with low initial modified Bishop scores; instead, they recommend caesarean section. Despite the higher initial modified Bishop score in twin pregnancies, the success rate of vaginal delivery remained lower in this group than in singleton pregnancies, suggesting that this factor did not significantly influence the study’s results.

Although the induction of labour protocol was consistent for both twin and singleton pregnancies, variations in obstetricians’ assessments of cervical dilatation, labour progression, and confidence in managing vaginal twin deliveries may have influenced study outcomes. Obstetricians with less experience or confidence in vaginal twin delivery may have been more likely to diagnose failed induction of labour and proceed with caesarean section. However, no statistically significant difference was observed in the time from initiation of oxytocin to selection of caesarean section between the two groups. The retrospective nature of the study introduced potential biases, and the limited incidence of twin pregnancies in a single regional hospital restricted the sample size. A larger sample size and multicentre design would enhance the generalisability of the findings. Furthermore, because the study primarily included Chinese patients, the applicability of these conclusions to other ethnic groups is limited; there is a need for further research in this area.

Conclusion

The failure rate of induction of labour was higher in twin pregnancies than in singleton pregnancies. Nevertheless, 73.8% of patients with twin pregnancies achieved successful vaginal deliveries; approximately 20% required instrumental delivery for at least one twin. Furthermore, twin pregnancies were associated with a higher incidence of postpartum haemorrhage. These findings can help facilitate informed decision making for patients and obstetricians when considering induction of labour and selecting the most appropriate mode of delivery for patients with twin pregnancies.

Author contributions

Concept or design: CK Wong, WL Lau.
Acquisition of data: CK Wong.
Analysis or interpretation of data: CK Wong.
Drafting of the manuscript: CK Wong, CMW Hung.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Acknowledgement

The authors thank the Kwong Wah Hospital Clinical Research Centre and Mr Steven Lau, Biostatistician from the Centre for Clinical Research and Biostatistics of The Chinese University of Hong Kong, for their statistical advice.

Declaration

This work was posted on Authorea as a registered online preprint (https://doi.org/10.22541/au.169320065.53300017/v1).

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This research was approved by the Kowloon Central Cluster/Kowloon East Cluster Research Ethics Committee of Hospital Authority, Hong Kong (Ref No.: KC/KE-22-0113/ER-3). The requirement for informed patient consent was waived by the Committee due to the retrospective nature of the research and the use of anonymised data in the research.

References

1. Monden C, Pison G, Smits J. Twin peaks: more twinning in humans than ever before. Hum Reprod 2021;36:1666-73. Crossref

2. Santana DS, Cecatti JG, Surita FG, et al. Twin pregnancy and severe maternal outcomes: the World Health Organization multicountry survey on maternal and newborn health. Obstet Gynecol 2016;127:631-41. Crossref

3. Cheong-See F, Schuit E, Arroyo-Manzano D, et al. Prospective risk of stillbirth and neonatal complications in twin pregnancies: systematic review and meta-analysis. BMJ 2016;354:i4353. Crossref

4. National Institute for Health and Care Excellence. NICE guideline [NG137]. Twin and triplet pregnancy. London: National Institute for Health and Care Excellence; 2019.

5. Barrett JF, Hannah ME, Hutton EK, et al. A randomized trial of planned cesarean or vaginal delivery for twin pregnancy. N Engl J Med 2013;369:1295-305. Crossref

6. Sotiriadis A, Petousis S, Thilaganathan B, et al. Maternal and perinatal outcomes after elective induction of labor at 39 weeks in uncomplicated singleton pregnancy: a meta-analysis. Ultrasound Obstet Gynecol 2019;53:26-35. Crossref

7. Stock SJ, Ferguson E, Duffy A, Ford I, Chalmers J, Norman JE. Outcomes of elective induction of labour compared with expectant management: population-based study. BMJ 2012;344:e2838. Crossref

8. Loscul C, Schmitz T, Blanc-Petitjean P, Goffinet F, Le Ray C; JUMODA and MEDIP study groups. Risk of cesarean after induction of labor in twin compared to singleton pregnancies. Eur J Obstet Gynecol Reprod Biol 2019;237:68-73. Crossref

9. Okby R, Shoham-Vardi I, Ruslan S, Sheiner E. Is induction of labor risky for twins compare to singleton pregnancies? J Matern Fetal Neonatal Med 2013;26:1804-6. Crossref

10. Fausett MB, Barth WH Jr, Yoder BA, Satin AJ. Oxytocin labor stimulation of twin gestations: effective and efficient. Obstet Gynecol 1997;90:202-4. Crossref

11. Taylor M, Rebarber A, Saltzman DH, Klauser CK, Roman AS, Fox NS. Induction of labor in twin compared with singleton pregnancies. Obstet Gynecol 2012;120:297-301. Crossref

12. Amikam U, Hiersch L, Barrett J, Melamed N. Labour induction in twin pregnancies. Best Pract Res Clin Obstet Gynaecol 2022;79:55-69. Crossref

13. Yung WK, Liu AL, Lai SF, et al. A specialised twin pregnancy clinic in a public hospital. Hong Kong J Gynaecol Obstet Midwifery 2012;12:21-32.

14. National Institute for Health and Care Excellence. NICE guideline [NG207]. Inducing labour. London: National Institute for Health and Care Excellence; 2021.

15. Thomas J, Kavanagh J, Kelly A, editors. RCOG Evidence-based Clinical Guidelines Induction of labour. RCOG Press; 2001.

16. Batinelli L, Serafini A, Nante N, Petraglia F, Severi FM, Messina G. Induction of labour: clinical predictive factors for success and failure. J Obstet Gynaecol 2018;38:352-8. Crossref

17. Jeong Y, Choo SP, Yun J, Kim EH. Effect of maternal age on maternal and perinatal outcomes including cesarean delivery following induction of labor in uncomplicated elderly primigravidae. Medicine (Baltimore) 2021;100:e27063. Crossref

18. Chan YY, Lo TK, Yu EL, Ho LF. Indications for induction of labour and mode of delivery in nulliparous term women with an unfavourable cervix. Hong Kong J Gynaecol Obstet Midwifery 2021;21:69-75. Crossref

19. Papoutsis D, Antonakou A, Tzavara C. The effect of ethnic variation on the success of induced labour in nulliparous women with postdates pregnancies. Scientifica (Cairo) 2016;2016:9569725. Crossref

20. Turton P, Arrowsmith S, Prescott J, et al. A comparison of the contractile properties of myometrium from singleton and twin pregnancies. PLoS One 2013;8:e63800. Crossref

21. Yin Z, He W, Li Y, et al. Adaptive reduction of human myometrium contractile activity in response to prolonged uterine stretch during term and twin pregnancy. Role of TREK-1 channel. Biochem Pharmacol 2018;152:252-63. Crossref

22. Mok SL, Lo TK. Vaginal delivery of second twins: factors predictive of failure and adverse perinatal outcomes. Hong Kong Med J 2022;28:376-82. Crossref

23. Tang HT, Liu AL, Chan SY, et al. Twin pregnancy outcomes after increasing rate of vaginal twin delivery: retrospective cohort study in a Hong Kong regional obstetric unit. J Matern Fetal Neonatal Med 2016;29:1094-100. Crossref

24. Kong CW, To WW. The predicting factors and outcomes of caesarean section of the second twin. J Obstet Gynaecol 2017;37:709-13. Crossref

Prevalence, risk factors, and outcomes of systemic sclerosis–associated interstitial lung disease in a Chinese population

Hong Kong Med J 2025 Feb;31(1):16–23 | Epub 12 Feb 2025

https://doi.org/10.12809/hkmj2411807

ORIGINAL ARTICLE CME

Prevalence, risk factors, and outcomes of systemic sclerosis–associated interstitial lung disease in a Chinese population

Dennis TH Chan, MRCP (UK), FHKAM (Medicine)¹; Lydia HP Tam, MRCP (UK), FHKAM (Medicine)²; Tommy TO Lam, MRCP (UK), FHKAM (Medicine)²; Jacqueline So, MRCP (UK), FHKAM (Medicine)²; LY Ho, MRCP (UK), FHKAM (Medicine)²; LS Tam, MRCP (UK), FRCP (Lond)²; Ho So, FHKAM (Medicine), FRCP (Lond)²

¹ Department of Medicine, Alice Ho Miu Ling Nethersole Hospital, Hong Kong SAR, China

² Department of Medicine and Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong SAR, China

³ Department of Medicine and Geriatrics, Tai Po Hospital, Hong Kong

Corresponding author: Dr Dennis TH Chan (cdt978@ha.org.hk)

Full paper in PDF

Abstract

Introduction: Systemic sclerosis–associated interstitial lung disease (SSc-ILD) is a leading cause of mortality among systemic sclerosis (SSc) patients. This multicentre cohort study sought to determine the prevalence of SSc-ILD, identify risk factors for ILD development in SSc patients, and explore poor prognostic factors in SSc-ILD patients.

Methods: Medical records were retrospectively reviewed for Chinese patients who met the 2013 American College of Rheumatology/European League Against Rheumatism classification criteria for SSc. Univariable and multivariable analyses were performed to compare SSc patients with and without ILD, as well as SSc-ILD patients with and without disease progression. Survival analysis was also conducted.

Results: The study cohort comprised 223 SSc patients with a median follow-up duration of 8.1 years. The prevalence of ILD was 49.8%. A history of bibasal crackles (hazard ratio [HR]=2.813; P=0.001) was independently associated with ILD development. Among ILD patients, 64.1% exhibited progressive disease. An elevated C-reactive protein (CRP) level at ILD diagnosis (HR=1.064; P=0.002) constituted an independent predictor of ILD progression. The overall mortality rate was 24.2% and pneumonia was the most common cause of death. Predictors of mortality included age at SSc diagnosis (HR=1.101; P=0.002), history of smoking (HR=5.173; P=0.028), and CRP level at SSc diagnosis (HR=1.103; P=0.009).

Conclusion: Interstitial lung disease was prevalent among SSc patients in this cohort and the majority exhibited disease progression. Comprehensive clinical assessment, supported by investigations such as CRP level measurement, is essential to identify predictors of poor prognosis.

New knowledge added by this study

Interstitial lung disease (ILD) is common and often progressive among systemic sclerosis (SSc) patients in the Hong Kong Chinese population.
Baseline C-reactive protein level is independently associated with ILD progression and mortality in SSc patients.

Implications for clinical practice or policy

Interstitial lung disease screening is recommended for all SSc patients.
C-reactive protein level may serve as a predictor of ILD progression and mortality in SSc patients.
Prospective studies are necessary to develop personalised monitoring and treatment strategies.

[Abstract in Chinese]

Introduction

Systemic sclerosis (SSc) is a heterogeneous connective tissue disorder involving multiple organ systems. Its subtypes comprise limited cutaneous SSc (lcSSc) and diffuse cutaneous SSc (dcSSc).1 Common features include Raynaud’s phenomenon, skin sclerosis, and musculoskeletal inflammation. Organ-based manifestations, such as interstitial lung disease (ILD), pulmonary hypertension (PH), and scleroderma renal crisis, are particularly important because they substantially affect patient quality of life and survival. Systemic sclerosis–associated interstitial lung disease (SSc-ILD) is the leading cause of mortality in SSc, contributing to 35% of disease-related deaths.2 In Hong Kong, SSc has one of the highest standardised mortality ratios among rheumatic diseases.3

Systemic sclerosis–associated interstitial lung disease arises from chronic microinjuries to lung endothelial and epithelial cells, which activate the immune system and lead to the recruitment and transformation of fibroblasts into myofibroblasts that secrete excessive collagen-rich extracellular matrix.4 5 This pathological process causes pathological lung stiffness and architectural disruption, producing restrictive lung disease through reductions of lung compliance and volume.6

It is well established that there is an ethnic disparity in SSc-ILD; prevalence rates considerably vary among ethnic groups, ranging from 25% to 90%.7 The prevalence of SSc-ILD is reportedly higher in Asian populations than in Western populations.8 However, data concerning the prevalence and predictive factors of SSc-ILD in Southern Chinese individuals remain limited. A prospective case-control study investigating functioning and health-related quality of life in Hong Kong showed that among 78 SSc patients recruited, 24 (30.8%) had ILD.9

The clinical course of SSc-ILD ranges from asymptomatic presentation to rapidly progressive disease, which can lead to mortality. Severe disease develops in approximately 25% to 33% of SSc-ILD patients.4 Thus, it is essential to identify patients with early-stage SSc who are asymptomatic but exhibit a risk of ILD development and progression. This approach enables closer monitoring and facilitates timely treatment. Numerous risk factors for ILD development and progression in SSc patients have been reported.8 10 11 According to the 2020 European consensus statements on the identification and management of ILD in SSc,10 predictive factors include respiratory symptoms, smoking history, ethnicity (eg, native American or African heritage), dcSSc, presence of anti-topoisomerase antibody (ATA), and male sex. However, most of these findings were based on studies conducted in Western populations.10

To improve the identification and management of SSc patients at risk of ILD development or progression, we conducted a multicentre study that aimed to assess the prevalence of SSc-ILD in the Hong Kong Chinese population, investigate associated risk factors, and identify potential indicators of poor prognosis. The findings of this study are expected to enhance early detection and monitoring of ILD in SSc patients, enabling timely and effective interventions.

Methods

Study design and patients

This retrospective longitudinal study included SSc patients who attended Alice Ho Miu Ling Nethersole Hospital, Prince of Wales Hospital, and North District Hospital. These patients were identified via the Clinical Data Analysis and Reporting System, a database established for record keeping and research purposes in Hong Kong, which has been utilised in epidemiological studies.12 The International Classification of Diseases, Ninth Revision, Clinical Modification code 710.1 (Systemic sclerosis) was used to identify SSc patients within the Clinical Data Analysis and Reporting System. The search period spanned from January 2008 to December 2022. Clinical information for each patient was reviewed in the electronic health record. Patients were included if they had attended more than one follow-up appointment and met the 2013 American College of Rheumatology/European League Against Rheumatism classification criteria for SSc.13 Exclusion criteria were age at onset <18 years, overlap syndrome, and non-Chinese ethnicity. Patients with ILD were identified based on radiologists’ reports of high-resolution computed tomography (HRCT) of the thorax. For patients without HRCT records, chest radiographs were reviewed to identify evidence of ILD. Investigations, treatments, and the frequency of follow-ups were determined by the treating physicians.

Clinical variable collection

Demographic variables, including sex, smoking history, age at symptom onset, age at SSc diagnosis, and age at ILD diagnosis, were recorded. The first clinical symptoms attributed to SSc, as judged by the treating physicians, and symptoms observed during the follow-up period were documented. These symptoms included Raynaud’s phenomenon, puffy fingers, sclerodactyly, digital ulcers, oesophageal dysmotility, arthralgia, dyspnoea, and cough.14 The presence of bibasal crackles on physical examination by the treating physicians was also documented. The status of PH was recorded based on findings from echocardiography or right heart catheterisation. Disease duration was defined as the time from onset of the first symptom to the last visit. The SSc subtype was categorised as dcSSc or lcSSc based on the extent of skin involvement, using criteria established by LeRoy and Medsger.1

Laboratory data, including autoantibodies, C-reactive protein (CRP), and erythrocyte sedimentation rate (ESR) levels, were recorded. C-reactive protein and ESR levels at baseline and at ILD diagnosis were documented. Pulmonary function test (PFT) results at baseline and at the latest available assessment were retrieved. Forced expiratory volume in 1 second, forced vital capacity (FVC), and diffusing capacity of the lungs for carbon monoxide (DL_CO) were recorded. In ILD cases, the radiological pattern on HRCT, including non-specific interstitial pneumonitis, usual interstitial pneumonia, or other patterns, was noted.

Systemic sclerosis–associated interstitial lung disease outcomes were assessed based on ILD progression and mortality. Disease progression was defined as an increase in ILD extent on serial HRCT, as reported by radiologists, or a decline in FVC of ≥10% from baseline. Alternatively, progression was defined as an FVC decline of 5% to 9% accompanied by a DL_CO decline of ≥15%.15 Causes of death were categorised as SSc-related or SSc-unrelated, based on assessment by the treating physicians (when available) or the authors. Clinical variables with >20% missing data were excluded from statistical analyses.

Statistical analyses

Descriptive data for continuous variables were presented as mean±standard deviation or median (interquartile range [IQR]), as appropriate. Categorical variables were presented as numbers with percentages. Student’s t test or the Mann-Whitney U test was used for comparisons of continuous variables, depending on the data distribution. Categorical variables were compared using Fisher’s exact test or the Chi squared test. Patients with and without ILD were compared using univariable and multivariable Cox regression analyses to identify risk factors associated with the development of SSc-ILD. Among SSc-ILD patients, those displaying progressive ILD were compared with those lacking progression via univariable and multivariable analyses to identify risk factors for disease progression. The univariate effects of covariates on survival were evaluated using Kaplan–Meier curves; the log-rank test was utilised to assess differences in survival. Multivariable Cox regression analyses were conducted to identify independent predictors of adverse outcomes. Variables with P value <0.2 in univariable analyses were included in the multivariable Cox regression analysis. All statistical analyses were performed using SPSS (Windows version 27.0; IBM Corp, Armonk [NY], United States). P values <0.05 were considered statistically significant.

Results

Demographics and clinical characteristics

In total, 223 SSc patients were included in this study (Fig). Table 1 summarises the patients’ baseline characteristics. The median follow-up duration was 8.1 years (IQR=4.0-10.2) and the total cumulative follow-up period was 1951 person-years. The majority of patients were female (86.1%). The median age at SSc diagnosis was 55 years (IQR=48-64). A majority of patients (86.5%) underwent HRCT scans during the follow-up period. Among those without HRCT, none had chest radiographs suggestive of ILD. Limited cutaneous SSc was the most common subtype, displayed by 71.3% of the cohort. Anti-topoisomerase antibody was the most frequently detected autoantibody, present in 39.0% of patients.

Figure. Patient recruitment

Table 1. Baseline characteristics of systemic sclerosis patients in this study

The overall prevalence of ILD among SSc patients was 49.8%. The age at ILD diagnosis ranged from 20 to 85 years, with a median of 57 years. Most patients in the SSc-ILD subgroup were female (86.5%) and non-smokers (86.5%); these characteristics did not significantly differ relative to patients without ILD (Table 1). The median interval from onset of the first SSc symptom to ILD diagnosis was 2.4 years (IQR=1.3-5.4). Among ILD cases, 51.3% were diagnosed within the first 3 years after SSc symptom onset, and 64.0% were diagnosed within 5 years. Of the ILD patients, 18.9% were asymptomatic, whereas symptomatic patients experienced a median interval of 2.4 years (IQR=1.2-6.3) from respiratory symptom onset to ILD diagnosis.

The frequency of dcSSc was significantly greater in patients with ILD than in patients without ILD (39.6% vs 16.1%; P<0.001). Conversely, lcSSc was more common in patients without ILD than in patients with ILD (83.0% vs 59.5%; P<0.001). In the ILD group, ATA was the most frequently detected autoantibody (57.7%), whereas anti-centromere antibody (ACA) was more common in the non-ILD group (49.1%) [Table 1].

The frequencies of non-respiratory clinical features were comparable between the ILD and non-ILD groups. However, respiratory features, including dyspnoea, cough and bibasal crackles significantly differed between the two groups, both at presentation and during follow-up (P<0.001 for all comparisons). Pulmonary hypertension was significantly more frequent in the ILD group throughout the follow-up period (19.8% vs 2.7%; P<0.001). The ILD group also exhibited a numerically higher baseline ESR, with a median of 21.5 mm/hr (IQR=14-40.5), whereas the non-ILD group displayed a median of 18 mm/hr (IQR=11-30; P=0.074) [online supplementary Table 1].

Associative factors of interstitial lung disease development

Univariable analysis showed that several factors were associated with the presence of ILD (online supplementary Table 2). These included dcSSc, ATA, history of dyspnoea, history of cough, history of bibasal crackles, history of PH, and baseline ESR level. Conversely, ACA and lcSSc were negatively associated with ILD development. According to multivariable Cox regression analysis, a history of bibasal crackles was independently associated with the presence of ILD, and a history of dyspnoea showed a trend towards significance.

Predictors of interstitial lung disease progression

Among patients with ILD, 64.1% exhibited progression during follow-up. Patients with progressive ILD were younger at ILD diagnosis, displaying a mean age of 54 years (range, 20-85) compared with 60 years (range, 31-81; P=0.051) in patients with non-progressive ILD. The proportions of dcSSc and lcSSc were similar between the progressive and non-progressive ILD groups. Anti-topoisomerase antibody was the predominant autoantibody in both groups, with proportions of 62.7% and 54.5%, respectively (P=0.444) [online supplementary Table 3]. Regarding clinical characteristics, only a history of digital ulcers showed a significant difference; its prevalence was higher in the progressive ILD group (42.4% vs 15.2%; P=0.008) [online supplementary Table 4].

Table 2 compares the results of laboratory and PFT between the progressive and non-progressive ILD groups. C-reactive protein levels at both SSc diagnosis and ILD diagnosis were higher in the progressive ILD group; however, only CRP level at ILD diagnosis showed a trend towards significance (P=0.130). The latest values for the predicted percentages of forced expiratory volume in 1 second, FVC and DL_CO were significantly lower in the progressive ILD group (all P≤0.001), but baseline values did not differ between the groups. Regarding HRCT patterns, no significant differences were observed between the two groups.

Table 2. Laboratory and pulmonary function test results of systemic sclerosis patients with progressive and non-progressive interstitial lung disease

The results of the Cox regression analysis for ILD progression are presented in Table 3. In the univariable analysis, factors associated with ILD progression included CRP level at ILD diagnosis (hazard ratio [HR]=1.504; P=0.005) and the latest predicted percentage of DL_CO (HR=0.962; P<0.001). Multivariable analysis identified CRP level at ILD diagnosis (HR=1.064; P=0.002) as an independent factor associated with ILD progression, whereas a history of digital ulcers (HR=1.874; P=0.076) showed a trend towards significance.

Table 3. Univariable and multivariable Cox regression for predictors of interstitial lung disease progression

Mortality

The overall mortality rate in the cohort during the follow-up period was 24.2%; a higher rate was observed in the SSc-ILD group relative to the non-ILD group (29.7% vs 18.8%, P=0.056) [online supplementary Fig]. Among the causes of death, infections were most common, followed by malignancy (online supplementary Table 5). In patients with ILD, 63.6% of deaths resulted from pneumonia; this proportion was 42.9% among patients without ILD. The univariable analysis indicated that factors associated with mortality were older age at SSc diagnosis, male sex, history of smoking, presence of PH, baseline CRP level, and baseline ESR level. Multivariable analysis revealed the following independent predictors of mortality: older age at SSc diagnosis (HR=1.101; P=0.002), history of smoking (HR=5.173; P=0.028), and higher baseline CRP level (HR=1.103; P=0.009) [Table 4].

Table 4. Univariable and multivariable Cox regression for predictors of mortality

Discussion

We observed an SSc-ILD prevalence of 49.8% in a multicentre cohort of Southern Chinese SSc patients. Among Asian countries, the reported prevalence of SSc-ILD varies. In Korea, prevalence rates range from 40% to 58%,16 17 whereas in Japan, they range from 42% to 51%.18 19 However, considerably higher prevalence estimates of 63% to 85% have been reported in centres from Northern China.20 21 These findings suggest ethnic or geographic variations in the prevalence of SSc-ILD within the Asian population. Given the high prevalence in our cohort and the observed delay between respiratory symptom onset and ILD diagnosis (median=2.4 years), early universal screening for ILD is necessary among SSc patients. This is particularly important because a substantial proportion of patients (18.9%) were asymptomatic.

Consistent with previous studies, our findings confirmed that in Chinese SSc patients, the dcSSc subtype and presence of ATA were associated with a higher likelihood of ILD development, whereas the lcSSc subtype and presence of ACA were inversely related to ILD risk.11 22 23 Also, our study showed that a history of bibasal crackles was independently associated with ILD development, similar to the findings of a retrospective cohort study in South Africa.24 However, it is important to recognise that the presence of crackles often reflects established disease, and the new onset of respiratory symptoms may indicate ILD development. Irrespective of the presence of respiratory symptoms, all SSc patients are recommended to undergo screening for ILD via HRCT and PFTs, as specified by expert consensuses.10 25 Regular auscultation for bibasal crackles during follow-up is equally important because it facilitates the identification of individuals who may require repeat investigations.

C-reactive protein has been proposed as a biomarker for predicting SSc-ILD progression.26 Similar to our findings, a retrospective cohort study in France27 revealed a significant difference in CRP levels between SSc patients with and without ILD (P=0.003). The multivariate analysis in that study also demonstrated a negative correlation between CRP levels and FVC.27 C-reactive protein production is driven by interleukin 6, and interleukin 6 inhibitors have shown efficacy in preserving lung function among SSc-ILD patients during a phase three randomised controlled trial.28 These findings provide a mechanistic rationale for using CRP levels to identify SSc-ILD patients who may benefit from early investigation and treatment.

The cumulative survival rates reported in our study align with those observed in Western populations.11 29 However, a European Scleroderma Trials and Research Group cohort study conducted in China20 demonstrated a higher cumulative survival rate of 87.8% at 10 years, a lower overall mortality rate of 8.9%, and fewer SSc-ILD–related deaths (2.5%). This disparity may be attributed to the higher frequency of infection-related deaths and the greater proportion of patients with progressive disease in our cohort. Although assessments of treatment regimen and response were beyond the scope of our study, due to the confounding by indication involved in its retrospective design, immunosuppressive agents commonly used in the past may have predisposed patients to infections. Indeed, infection has previously been identified as the leading cause of death in local SSc patients.3 Considering the high rate of infection-related mortality, recently available antifibrotic treatments may be preferable to immunosuppressive therapy in selected patients who exhibit increased infection risk. Furthermore, consistent with well-established evidence, we identified increased age15 30 31 32 and elevated CRP levels at SSc diagnosis33 34 as predictors of mortality. It remains unclear whether more aggressive early treatment in patients with elevated baseline CRP levels would improve survival; further investigation is warranted.

Limitations

Some limitations should be acknowledged in our study. The data were extracted from the electronic health record, making undercoding of diagnoses unavoidable. Due to the retrospective study design, some clinical data essential to this study might not have been fully documented, and disease progression monitoring was not systematic, which could introduce bias. The presence of symptoms and ILD was assessed by the treating physicians and radiologists, respectively; these assessments potentially lacked specificity or sensitivity. Follow-up investigations were primarily ordered based on clinical judgement, leading to potential selection bias. Our analyses did not adjust for patients with progressive disease who may have received treatment leading to ILD stabilisation, which could have resulted in classification of their ILD as non-progressive. Furthermore, no standardised criteria currently exist for defining SSc-ILD progression. Quantitative assessments of ILD involvement on HRCT, such as percentage involvement or the Warrick score,35 and the extensiveness of skin disease using the modified Rodnan skin score,36 were also unavailable.

Conclusion

This is the first multicentre cohort study to investigate SSc-ILD in Hong Kong. Our findings demonstrated a high prevalence of ILD among Chinese SSc patients, with a significant proportion of these patients exhibiting disease progression. Universal ILD screening is recommended for SSc patients, with particular attention to those who develop respiratory symptoms and signs. In addition to imaging and PFTs, CRP levels could serve as a biomarker for ILD progression and poor prognosis.

Author contributions

Concept or design: DTH Chan, H So.
Acquisition of data: DTH Chan, LHP Tam.
Analysis or interpretation of data: DTH Chan, H So.
Drafting of the manuscript: DTH Chan, H So.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Declaration

The results of this study were presented as poster presentation at 26th Asia-Pacific League of Associations for Rheumatology Congress 2024 in Singapore, 21-25 August 2024.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This research was approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: CREC-2023-393). The requirement for informed patient consent was waived by the Committee due to the retrospective nature of the research.

Supplementary material

References

1. LeRoy EC, Medsger TA Jr. Criteria for the classification of early systemic sclerosis. J Rheumatol 2001;28:1573-6.

2. Tyndall AJ, Bannert B, Vonk M, et al. Causes and risk factors for death in systemic sclerosis: a study from the EULAR Scleroderma Trials and Research (EUSTAR) database. Ann Rheum Dis 2010;69:1809-15. Crossref

3. Mok CC, Kwok CL, Ho LY, Chan PT, Yip SF. Life expectancy, standardized mortality ratios, and causes of death in six rheumatic diseases in Hong Kong, China. Arthritis Rheum 2011;63:1182-9. Crossref

4. Khanna D, Tashkin DP, Denton CP, Renzoni EA, Desai SR, Varga J. Etiology, risk factors, and biomarkers in systemic sclerosis with interstitial lung disease. Am J Respir Crit Care Med 2020;201:650-60. Crossref

5. Mostmans Y, Cutolo M, Giddelo C, et al. The role of endothelial cells in the vasculopathy of systemic sclerosis: a systematic review. Autoimmun Rev 2017;16:774-86. Crossref

6. Khanna D, Lescoat A, Roofeh D, et al. Systemic sclerosis–associated interstitial lung disease: how to incorporate two Food and Drug Administration–approved therapies in clinical practice. Arthritis Rheumatol 2022;74:13-27. Crossref

7. Qiu M, Nian X, Pang L, Yu P, Zou S. Prevalence and risk factors of systemic sclerosis–associated interstitial lung disease in East Asia: a systematic review and meta-analysis. Int J Rheum Dis 2021;24:1449-59. Crossref

8. Chan DT, So H. Systemic sclerosis–associated interstitial lung disease: prevalence and risk factors. J Clin Rheumatol Immunol 2023;23:15-24. Crossref

9. Chan PT, Mok CC, Chan KL, Ho LY. Functioning and health-related quality of life in Chinese patients with systemic sclerosis: a case-control study. Clin Rheumatol 2014;33:659-66. Crossref

10. Hoffmann-Vold AM, Maher TM, Philpot EE, et al. The identification and management of interstitial lung disease in systemic sclerosis: evidence-based European consensus statements. Lancet Rheumatol 2020;2:e71-83. Crossref

11. Nihtyanova SI, Schreiber BE, Ong VH, et al. Prediction of pulmonary complications and long-term survival in systemic sclerosis. Arthritis Rheumatol 2014;66:1625-35. Crossref

12. So H, So J, Lam TT, et al. Performance of the 2017 European Alliance of Associations for Rheumatology/American College of Rheumatology classification criteria in patients with idiopathic inflammatory myopathy and anti–melanoma differentiation–associated protein 5 positivity. Arthritis Rheumatol 2022;74:1588-92. Crossref

13. van den Hoogen F, Khanna D, Fransen J, et al. 2013 classification criteria for systemic sclerosis: an American College of Rheumatology/European League against Rheumatism Collaborative Initiative. Arthritis Rheum 2013;65:2737-47. Crossref

14. van den Hombergh WM, Carreira PE, Knaapen-Hans HK, van den Hoogen FH, Fransen J, Vonk MC. An easy prediction rule for diffuse cutaneous systemic sclerosis using only the timing and type of first symptoms and auto-antibodies: derivation and validation. Rheumatology (Oxford) 2016;55:2023-32. Crossref

15. Goh NS, Hoyles RK, Denton CP, et al. Short-term pulmonary function trends are predictive of mortality in interstitial lung disease associated with systemic sclerosis. Arthritis Rheumatol 2017;69:1670-8. Crossref

16. Jung E, Suh CH, Kim HA, Jung JY. Clinical characteristics of systemic sclerosis with interstitial lung disease. Arch Rheumatol 2018;33:322-7. Crossref

17. Kim J, Park SK, Moon KW, et al. The prognostic factors of systemic sclerosis for survival among Koreans. Clin Rheumatol 2010;29:297-302. Crossref

18. Sekiguchi A, Inoue Y, Yamazaki S, et al. Prevalence and clinical characteristics of earlobe crease in systemic sclerosis: possible association with vascular dysfunction. J Dermatol 2020;47:870-5. Crossref

19. Aozasa N, Hatano M, Saigusa R, et al. Clinical significance of endothelial vasodilatory function evaluated by EndoPAT in patients with systemic sclerosis. J Dermatol 2020;47:609-14. Crossref

20. Hu S, Hou Y, Wang Q, Li M, Xu D, Zeng X. Prognostic profile of systemic sclerosis: analysis of the clinical EUSTAR cohort in China. Arthritis Res Ther 2018;20:235. Crossref

21. Wang J, Assassi S, Guo G, et al. Clinical and serological features of systemic sclerosis in a Chinese cohort. Clin Rheumatol 2013;32:617-21. Crossref

22. Sánchez-Cano D, Ortego-Centeno N, Callejas JL, et al. Interstitial lung disease in systemic sclerosis: data from the Spanish scleroderma study group. Rheumatol Int 2018;38:363-74. Crossref

23. Gelber AC, Manno RL, Shah AA, et al. Race and association with disease manifestations and mortality in scleroderma: a 20-year experience at the Johns Hopkins Scleroderma Center and review of the literature. Medicine (Baltimore) 2013;92:191-205. Crossref

24. Ashmore P, Tikly M, Wong M, Ickinger C. Interstitial lung disease in South Africans with systemic sclerosis. Rheumatol Int 2018;38:657-62. Crossref

25. Rahaghi FF, Hsu VM, Kaner RJ, et al. Expert consensus on the management of systemic sclerosis–associated interstitial lung disease. Respir Res 2023;24:6. Crossref

26. Distler O, Assassi S, Cottin V, et al. Predictors of progression in systemic sclerosis patients with interstitial lung disease. Eur Respir J 2020;55:1902026. Crossref

27. Chikhoune L, Brousseau T, Morell-Dubois S, et al. Association between routine laboratory parameters and the severity and progression of systemic sclerosis. J Clin Med 2022;11:5087. Crossref

28. Khanna D, Lin CJ, Furst DE, et al. Long-term safety and efficacy of tocilizumab in early systemic sclerosis–interstitial lung disease: open-label extension of a phase 3 randomized controlled trial. Am J Respir Crit Care Med 2022;205:674-84. Crossref

29. Pokeerbux MR, Giovannelli J, Dauchet L, et al. Survival and prognosis factors in systemic sclerosis: data of a French multicenter cohort, systematic review, and meta-analysis of the literature. Arthritis Res Ther 2019;21:86. Crossref

30. Volkmann ER, Tashkin DP, Sim M, et al. Short-term progression of interstitial lung disease in systemic sclerosis predicts long-term survival in two independent clinical trial cohorts. Ann Rheum Dis 2019;78:122-30. Crossref

31. Volkmann ER, Saggar R, Khanna D, et al. Improved transplant-free survival in patients with systemic sclerosis–associated pulmonary hypertension and interstitial lung disease. Arthritis Rheumatol 2014;66:1900-8. Crossref

32. Takei R, Arita M, Kumagai S, et al. Radiographic fibrosis score predicts survival in systemic sclerosis–associated interstitial lung disease. Respirology 2018;23:385-91. Crossref

33. Liu X, Mayes MD, Pedroza C, et al. Does C-reactive protein predict the long-term progression of interstitial lung disease and survival in patients with early systemic sclerosis? Arthritis Care Res (Hoboken) 2013;65:1375-80. Crossref

34. Le Gouellec N, Duhamel A, Perez T, et al. Predictors of lung function test severity and outcome in systemic sclerosis–associated interstitial lung disease. PLoS One 2017;12:e0181692. Crossref

35. Warrick JH, Bhalla M, Schabel SI, Silver RM. High resolution computed tomography in early scleroderma lung disease. J Rheumatol 1991;18:1520-8.

36. Khanna D, Furst DE, Clements PJ, et al. Standardization of the modified Rodnan skin score for use in clinical trials of systemic sclerosis. J Scleroderma Relat Disord 2017;2:11-8. Crossref

Artificial intelligence–based computer-aided diagnosis for breast cancer detection on digital mammography in Hong Kong

Hong Kong Med J 2024 Dec;30(6):468–77 | Epub 19 Dec 2024

https://doi.org/10.12809/hkmj2310920

ORIGINAL ARTICLE

Artificial intelligence–based computer-aided diagnosis for breast cancer detection on digital mammography in Hong Kong

SM Yu, MB, BS, FHKAM (Radiology)¹; Catherine YM Young, MB, BS, FRCR¹; YH Chan, MB, ChB, FRCR¹; YS Chan, MB, ChB, FHKAM (Radiology)¹; Carita Tsoi, MB, ChB, FHKAM (Radiology)¹; Melinda NY Choi, MHSc, GCB¹; TH Chan, BSc, MSc¹; Jason Leung, MSc²; Winnie CW Chu, MB, ChB, FHKAM (Radiology)¹; Esther HY Hung, MB, ChB, FHKAM (Radiology)¹; Helen HL Chau, MB, ChB, FHKAM (Radiology)

¹ Department of Imaging and Interventional Radiology, Prince of Wales Hospital, Hong Kong SAR, China

² The Jockey Club Centre for Osteoporosis Care and Control, The Chinese University of Hong Kong, Hong Kong SAR, China

Corresponding author: Dr SM Yu (ysm687@ha.org.hk)

Full paper in PDF

Abstract

Introduction: Research concerning artificial intelligence in breast cancer detection has primarily focused on population screening. However, Hong Kong lacks a population-based screening programme. This study aimed to evaluate the potential of artificial intelligence–based computer-assisted diagnosis (AI-CAD) program in symptomatic clinics in Hong Kong and analyse the impact of radio-pathological breast cancer phenotype on AI-CAD performance.

Methods: In total, 398 consecutive patients with 414 breast cancers were retrospectively identified from a local, prospectively maintained database managed by two tertiary referral centres between January 2020 and September 2022. The full-field digital mammography images were processed using a commercial AI-CAD algorithm. An abnormality score <30 was considered a false negative, whereas a score of ≥90 indicated a high-score tumour. Abnormality scores were analysed with respect to the clinical and radio-pathological characteristics of breast cancer, tumour-to–breast area ratio (TBAR), and tumour distance from the chest wall for cancers presenting as a mass.

Results: The median abnormality score across the 414 breast cancers was 95.6; sensitivity was 91.5% and specificity was 96.3%. High-score cancers were more often palpable, invasive, and presented as masses or architectural distortion (P<0.001). False-negative cancers were smaller, more common in dense breast tissue, and presented as asymmetrical densities (P<0.001). Large tumours with extreme TBARs and locations near the chest wall were associated with lower abnormality scores (P<0.001). Several strengths and limitations of AI-CAD were observed and discussed in detail.

Conclusion: Artificial intelligence–based computer-assisted diagnosis shows potential value as a tool for breast cancer detection in symptomatic setting, which could provide substantial benefits to patients.

New knowledge added by this study

With a threshold score of 30, a commercially available artificial intelligence–based computer-assisted diagnosis (AI-CAD) program showed high sensitivity and specificity for breast cancer detection on digital mammography in symptomatic settings, offering a valuable diagnostic adjunct.
The performance of AI-CAD varied according to the radio-pathological characteristics of breast cancer. Notably, the program demonstrated promising accuracy in detecting breast cancers that exhibit architectural distortion, which remains a diagnostic challenge.
Observed limitations of AI-CAD, such as underscoring cancers that present as large masses or exhibit nipple retraction as well as its inability to compare with previous studies, highlight concerns regarding standalone use of AI for triage in symptomatic clinics.

Implications for clinical practice or policy

Artificial intelligence–based computer-assisted diagnosis exhibits substantial potential for detecting breast cancers in symptomatic settings.
To make study findings clinically viable, larger validation studies are needed.

[Abstract in Chinese]

Introduction

Mammography is the principal modality used for breast cancer screening and detection in women worldwide.1 However, 10% to 30% of breast cancers may be undetected during mammography due to factors such as dense breast tissue, poor imaging technique, perceptual error, and subtle mammographic abnormalities.2

Conventional computer-aided diagnosis systems have been developed for more than two decades; however, large-scale studies have shown no significant benefit of such systems in enhancing radiologists’ diagnostic performance.3 4 Such systems do not facilitate differentiation between benign and malignant breast lesions, resulting in numerous false-positive results that require radiologist review, which may lead to reader fatigue and unnecessary additional investigations.

Currently, artificial intelligence–based computer-assisted diagnosis (AI-CAD) is widely implemented in mammography to improve diagnostic accuracy and reduce radiologist workload.5 6 The AI-CAD systems developed using deep-learning algorithms make independent decisions and self-learn without the need for feature engineering and computation.7 Artificial intelligence algorithms have been applied to multiple aspects of breast cancer screening, including risk stratification, triage, lesion interpretation, and patient recall.8 As of 2022, the US Food and Drug Administration has approved >15 AI tools for mammography applications, including density assessment, triage, lesion detection, and classification.9 Most commercial AI-CAD programs provide heatmaps with abnormality scores. Generally, higher abnormality scores indicate more suspicious radiological features and a greater likelihood of cancer.

Most existing evidence in the literature is derived from population-based screening studies.5 10 11 However, unlike other developed Asian countries such as Singapore and Korea, Hong Kong lacks a large-scale population screening programme.12 13 Our patient population primarily consists of symptomatic individuals. Evidence concerning the application of AI-CAD in symptomatic breast imaging is limited. This study aimed to evaluate the potential of AI-CAD in Hong Kong, focusing on the impact of radio-pathological phenotypes of breast cancer on AI-CAD performance. We analysed the distinctive characteristics of high-score versus low-score breast cancers. We also discuss observed strengths and limitations of AI-CAD in identifying breast cancer.

Methods

Study population

In total, 488 consecutive patients with histology-confirmed breast cancers were identified from a prospectively maintained database managed by two tertiary referral centres in Hong Kong during the period between January 2020 and September 2022. In our centres, all patients referred for diagnostic mammography were symptomatic, presenting with various clinical symptoms. We included patients with breast cancers confirmed by core needle biopsy under ultrasound guidance or stereotactic-guided vacuum-assisted breast biopsy performed at our centres. We excluded patients with diagnostic mammography performed at outside facilities (n=6), chest wall recurrence after mastectomy (n=14), cancers identified only in axillary nodes (n=3), tumour locations not feasible for mammography (n=3), and mammographically occult breast cancers undetectable by both reporting radiologists and AI-CAD (n=64) [Fig 1]. Finally, 398 patients with 414 breast cancers and 347 unaffected breasts were included in the study. Sixteen patients were diagnosed with bilateral breast cancer. Among the 382 patients with unilateral breast cancer, 35 had previously undergone contralateral mastectomy.

Figure 1. Inclusion and exclusion criteria

Image acquisition and analysis

Full-field digital mammography (MAMMOMAT Inspiration; Siemens, Erlangen, Germany or Selenia Dimensions; Hologic, Newark [DE], US) was performed prior to each biopsy. The included mammograms were exported and processed by a commercial AI-CAD program (INSIGHT MMG, version 1.10.2; Lunit, Seoul, South Korea), which is approved by the US Food and Drug Administration for lesion detection and classification in breast imaging.9

The AI-CAD algorithm used in the current study was developed and validated through multinational studies.14 15 This algorithm provides a heatmap that highlights mammographic abnormalities and generates a score ranging from 0 to 100 for each view (craniocaudal and mediolateral oblique views). The abnormality score is the maximum value for each breast, reflecting the likelihood of malignancy.

All mammograms were interpreted by radiologists subspecialising in breast radiology (with 4 to 20 years of experience in breast imaging). Mammography reports from the time of breast cancer diagnosis were retrieved from the radiology information system and retrospectively reviewed for breast density, dominant mammographic features of breast cancer, and any axillary lymphadenopathy. The clinical findings, pathological results, and molecular profiles of breast cancers were also recorded. Breast density was categorised from 1 to 4 using the BI-RADS (Breast Imaging Reporting and Data System) classification.16 The cancers were classified according to their dominant mammographic features as asymmetrical density, mass (with or without calcifications), calcifications alone, or architectural distortion.

For breast cancers presenting as a mass without calcifications, the tumour distances from the chest wall and the tumour-to–breast area ratio (TBAR) were measured in mammograms using the picture archiving and communication system by a radiologist with 2 years of experience in breast imaging. Tumour distance from the chest wall was defined as the shortest distance between the tumour and the pectoralis major in the mediolateral oblique view (Fig 2a). Tumours partially visible within the lower breast in the mediolateral oblique view, where the pectoralis muscle is not discernible, were assigned a chest wall distance of 0 cm. The TBAR was calculated via division of the tumour area by the breast area, as measured using the freehand region-of-interest tool (Fig 2b).

Figure 2. (a) Index cancer (white arrows) and measurement of tumour distance from the chest wall on the picture archiving and communication system (PACS) [double arrow]. (b) Measurement of tumour-to–breast area ratio by freehand region-of-interest on the PACS, indicated by curved arrow (tumour area) and open arrows (breast area)

The radiologists matched the index lesion to the AI-CAD heatmap to determine whether the AI-CAD correctly localised the known cancer. When the cancer was correctly localised by the AI-CAD, an abnormality score of ≥30 was regarded as a true positive, whereas a score &LT;30 was considered a false negative. When the cancer was undetected or incorrectly localised by the AI-CAD, this result also was regarded as a false negative. Breast cancers with abnormality scores of ≥90 and <30 were designated as ‘high-score tumour’ and ‘low-score tumour’, respectively.

Statistical analysis

Abnormality scores are presented as medians with interquartile ranges. The scores were analysed according to patient symptoms, breast density, mammographic findings, cancer histology, and molecular profile using the Mann-Whitney U test or Kruskal–Wallis H test. The AI-CAD abnormality scores were divided into three intervals: 0 to <30, 30-90, and >90 to 100. The Chi squared test and Mantel-Haenszel test for trend were used to analyse associations with different factors. For cancers presenting as a mass, mean abnormality scores across various TBARs and distances to the chest wall were evaluated using analysis of variance with pairwise comparisons. Statistical analyses were performed using SPSS (Windows version 26; IBM Corp, Armonk [NY], US). P values <0.05 were considered statistically significant.

Results

In total, 398 patients (mean age, 62.4 years; range, 35-100) with 414 breast cancers and 347 unaffected breasts were included in the study. The cohort consisted of two men and 396 women. Among the 414 breast cancer cases, 284 (68.6%) were palpable (Table 1).

Table 1. Median abnormality scores assigned by artificial intelligence–based computer-assisted diagnosis according to clinical, radiological, and pathological phenotypes of breast cancers (n=414)

Distribution of abnormality scores

The median and mean abnormality scores for the 414 breast cancers were 95.6 and 80.6, respectively (range, 0.4-99.9). The distribution of breast cancers according to abnormality score interval is presented in Figure 3. The sensitivity of the AI-CAD algorithm in detecting breast cancers was 91.5%, based on breast cancer identification using an abnormality score of ≥30. Overall, 65.7% of breast cancers were classified as high-score tumours, whereas 8.5% were classified as low-score tumours with abnormality scores <30; these low-score tumours were regarded as false-negative cases. Table 1 presents the medians and interquartile ranges of abnormality scores according to clinical, radiological, and pathological phenotypes.

Figure 3. Distribution of breast cancers according to abnormality score interval (n=414)

Palpable lesions, cancers in entirely fatty or scattered fibroglandular breasts, cancers presenting as masses with or without calcifications and architectural distortion, and larger cancers were associated with higher abnormality scores (all P<0.001) [Table 1]. Invasive cancers had higher abnormality scores compared with ductal carcinoma in situ (P=0.010). Axillary nodal status (P=0.078) and cancer molecular subtype (P=0.820) were not associated with abnormality scores (Table 2).

Table 2. Comparison of clinical, radiological, and pathological phenotypes of breast cancers between false-negative and truepositive results of artificial intelligence–based computer-assisted diagnosis (n=414)

Phenotypic features of high-score breast cancer

High-score breast cancers had higher prevalences of palpable disease, cancers presenting as masses with or without calcifications, invasive cancers, and larger cancers (>1 cm) [Table 2].

Phenotypic features of low-score, false-negative breast cancer

The false-negative rate for AI-CAD was 8.5% (35/414). These cancers had higher prevalences of non-palpable disease, cancers presenting as asymmetrical densities, small cancers (<1 cm), and locations in heterogeneously dense or extremely dense breast tissue.

Impact of tumour-to–breast area ratio and tumour distance from chest wall on abnormality score

Overall, 158 cancers presenting as masses without calcifications were included in this analysis. The mean abnormality score for cancers with a TBAR of ≥30% was significantly lower than for those with a TBAR of <30% (86.7 vs 54.4; P<0.001). Tumours bordering the chest wall (ie, distance of 0 cm from chest wall) demonstrated significantly lower abnormality scores compared with those located 1 cm and ≥2 cm away from the chest wall (mean, 65.5 vs 89.2 vs 87.2; P<0.001).

Distribution of abnormality scores for unaffected breasts

In the analysis of 347 unaffected breasts (regarded as negative findings by reporting radiologists), the median abnormality score was 0 (mean, 3.5; range, 0-81). Using a threshold score of 30, the false-positive rate was 3.7% (13/347), indicating 96.3% specificity. Most of these false positives (11/13) scored between 30 and 50; none scored >90. One case with known postoperative changes from breast conservative surgery showed stable mammographic finding for 10 years, scored 81 by AI-CAD. One case with a breast cyst scored 73, which was confirmed via fine needle aspiration cytology.

Discussion

Performance and potentials

Most AI-CAD algorithms provide heatmaps with abnormality scores ranging from 0 to 100; a higher score generally implies a greater likelihood of cancer. Previous AI-CAD studies have used various threshold scores; some set a threshold of 10 for population screening,18 19 20 whereas Weigel et al21 set a threshold of 28 for detecting malignant calcifications. However, the clinical implications of the abnormality score itself have not been clarified; a score range from 10 to 100 may be too broad for distinguishing malignancies in clinical practice. These aspects highlight the need for further validation of the appropriate reference score provided by AI-CAD algorithms. In this study, we set the threshold at 30 because, unlike population screening approaches, our patients were symptomatic individuals. A higher threshold score appears more practical in the clinical setting of symptomatic patients.

In our study, the AI-CAD algorithm detected 91.5% (379/414) of breast cancers with an abnormality score of >30; of these 379 cancers, 71.7% exhibited a high abnormality score of >90. The false-negative rate of 8.5% is comparable to previously reported rate for this AI-CAD algorithm.5

All cancers presenting as architectural distortion in our study were correctly localised by the AI-CAD, with abnormality scores >30; 87.5% of them were assigned high abnormality scores of >90 (Fig 4a and b). Unlike cancers presenting as masses or calcifications, cancers presenting as architectural distortion remain challenging for radiologists to detect and interpret.22 23 24 Wan et al25 showed that a standalone AI algorithm did not outperform radiologists; however, with AI assistance, junior radiologists demonstrated significant improvements in diagnostic accuracy for architectural distortion.

Figure 4. Cases illustrating the strengths of artificial intelligence–based computer-assisted diagnosis (AI-CAD). (a) A 52-year-old woman presenting with a right breast mass. The mediolateral-oblique mammographic view shows an architectural distortion (white arrow) in the upper right breast. (b) The AI-CAD program successfully detected this asymmetrical distortion within heterogeneously dense breast tissue, assigning a high abnormality score of 97. (c) A 55-year-old woman with a subtle asymmetrical density, identified as ductal carcinoma in situ on biopsy. The mediolateral-oblique mammographic view shows a subtle asymmetrical density (white arrow) in the upper left breast. The reporting radiologist did not detect the lesion on mammography but detected it via concurrent diagnostic ultrasound. (d) The AI-CAD program detected the subtle asymmetrical density, assigning an abnormality score of 68. (e) A 50-year-old woman with bilateral polyacrylamide gel implants presenting with a small lump in the left breast. The mediolateral-oblique mammographic view shows that the gel had been injected into various layers of the anterior chest wall (behind and within breast tissue, subcutaneous layer, and muscle). A subtle group of amorphous calcifications is visible in the upper left breast (white arrow). (f) The AI-CAD program detected these grouped calcifications in the context of breast augmentation, assigning a high abnormality score of 91

One case of breast cancer presenting as asymmetric density in heterogeneously dense breast tissue was missed by the reporting radiologist but detected by AI-CAD, which assigned an abnormality score of 68. The cancer was later identified by the radiologist via ultrasound, which is part of routine workup for symptomatic patients in our centre. Retrospective review indicated that the asymmetric density was visible on mammography (Fig 4c and d). In a study by Kim et al,26 40 of 128 mammographically occult breast cancers were correctly identified by the AI algorithm, demonstrating its added value in detecting such cancers.

The 64 cases of mammographically occult breast cancer not detected by either the AI-CAD or the radiologists were excluded from the study. Of these cases, 84.3% were found in heterogeneously dense and extremely dense breast tissue (BI-RADS 3 and 4).16 Dense breast tissue is recognised as a significant feature associated with mammographically occult and missed cancers.27 28 29 30 We suspect that mammographic signs of cancer are masked or obscured by dense breast parenchyma, thus evading detection by the AI-CAD. Conversely, both radiologists and the AI-CAD tended to more effectively detect cancers in fatty breasts.18

In our study, the AI-CAD correctly localised a small breast cancer with a high abnormality score (>90) in a patient with polyacrylamide hydrogel (PAAG)–injected augmentation mammoplasty (Fig 4e and f). The diagnosis of breast cancer after PAAG-injected augmentation mammoplasty is challenging. Lesion visualisation may be masked by the presence of polyacrylamide gel, and extravasated polyacrylamide gel may mimic a lesion on mammography, potentially delaying early cancer detection. In such cases, assessments of suspicious calcifications and parenchymal distortion within visible breast parenchyma are considered the main goals of screening mammography.31 32 The effectiveness of AI-CAD in detecting breast cancer among patients with augmentation mammaplasty remains uncertain, warranting further studies.

Detection challenges and future directions

Isolated cases of large, clearly visible lesions that evaded AI detection have been described by Lång et al33 and Choi et al.18 To our knowledge, our study is the first to investigate factors contributing to such evasion. In this study, the AI algorithm tended to underscore cancers presenting as large masses (Fig 5a and b). Cancers with a TBAR of ≥30% had significantly lower mean abnormality scores relative to those with a ratio of <30%. Tumours bordering the chest wall (0 cm distance) also showed significantly lower abnormality scores than those located away from the chest wall. The underlying cause remains unclear; however, these findings highlight concerns regarding the use of AI-CAD as a standalone tool for triaging cases in symptomatic populations. We also noted that the AI-CAD missed certain cancers with obvious findings, such as nipple retraction and diffuse dermal thickening (Fig 5c to f).

Figure 5. Cases illustrating the limitations of artificial intelligence–based computer-assisted diagnosis (AI-CAD). (a) A 48-year-old woman presenting with a left breast mass. The craniocaudal mammographic view shows a retracted left breast mostly replaced by a large, irregular, high-density mass with dermal infiltration and suspected pectoralis involvement. (b) The AI-CAD program detected the tumour but assigned it a low abnormality score of 30. (c) A 48-year-old woman presenting with a right breast mass. The mediolateral-oblique mammographic view shows an irregular mass with indistinct margins in the periareolar region of the right breast with nipple retraction (white arrow). (d) The AI-CAD program correctly localised the right breast mass but assigned a low abnormality score of 19, despite the presence of nipple retraction. (e) A 57-year-old woman presenting with a right breast mass. The mediolateral-oblique mammographic view shows a large right breast mass with diffuse skin thickening (white arrows). (f) The AI-CAD program detected the breast mass but assigned a low abnormality score of 32, despite the presence of diffuse skin thickening. (g) A 62-year-old woman—with a history of breast-conserving surgery for breast cancer— exhibited local recurrence on surveillance mammography. The previous mediolateral-oblique mammographic view shows postoperative changes and macrocalcification in the upper right breast; no suspicious lesion was identified. (h) The follow-up mediolateral-oblique mammographic view shows a newly developed small, irregular mass (white arrow) in the upper right breast adjacent to the macrocalcification; biopsy confirmed invasive carcinoma. (i) The AI-CAD program did not detect this lesion, assigning a low abnormality score of 8

Moreover, the inability of AI-CAD to compare mammograms with previous studies may hinder its effectiveness in specific scenarios, such as the detection of subtle developing symmetries and identification of early recurrence in postoperative cases (Fig 5g to i). In contrast, radiologists can compare mammograms with previous studies, improving mammogram interpretation accuracy.

Studies have shown that the diagnostic performances of AI algorithms are comparable to those of radiologists in terms of assessing screening mammograms; the use of AI to triage screening mammograms could potentially reduce radiologists’ workload.5 34 35 We identified potential limitations and weaknesses of AI-CAD in diagnosing breast cancers under certain conditions, highlighting the need for further large-scale studies to investigate clinical applications of AI-CAD in symptomatic patients.

Strengths and limitations

This study had several key strengths. To our knowledge, it is the first to evaluate AI-CAD for breast cancer detection in Hong Kong, using an AI-CAD system that had not previously been exposed to images from our centres during their product development. Additionally, all digital mammograms were obtained before biopsies, avoiding any biopsy-related changes which could potentially affect AI-CAD performance. Limitations of the study include its retrospective design and inclusion of cancer-enriched datasets, which may lead to overestimation of AI-CAD performance; the use of a single AI vendor, hindering applicability to other AI algorithms; and the lack of BI-RADS correlation. Furthermore, there was a lack of information concerning progression in unaffected breasts over an extended follow-up interval (≥2 years), which could impact the false-positive rate of the AI-CAD. An extended observation period is needed to identify potential malignancies that may have been initially missed by radiologists.

Conclusion

Unlike other developed cities or countries, Hong Kong does not have population-based screening programmes. The adoption and implementation of AI programs in Hong Kong for breast imaging remains in early stages, mainly due to ongoing debates about efficacy and a lack of sufficient local data to support widespread application. Current literature is almost entirely based on population screening data, which may not be applicable to cities without screening programmes. In our study, AI-CAD demonstrated promising accuracy in detecting breast cancers within symptomatic settings; its performance varied according to radio-pathological characteristics. To translate these research findings into practical clinical applications, further validation studies with larger sample sizes are required; these would confirm the reliability of AI-CAD systems. The development of protocols for integrating AI-CAD into existing clinical workflows, formulation of usage guidelines, and initiation of training programmes for radiologists to effectively utilise AI as a second reader are essential elements of this process. Collaborations with information technology departments and hospital management are necessary to ensure successful integration. Although further investigation is needed, this study provides encouraging evidence to support the use of AI-CAD as a breast cancer detection tool in symptomatic settings, ultimately benefitting patients.

Author contributions

Concept or design: SM Yu, MNY Choi, EHY Hung, HHL Chau.
Acquisition of data: SM Yu, MNY Choi, TH Chan, CYM Young, YH Chan, YS Chan, C Tsoi.
Analysis or interpretation of data: SM Yu, TH Chan, J Leung.
Drafting of the manuscript: SM Yu, CYM Young.
Critical revision of the manuscript for important intellectual content: SM Yu, CYM Young, WCW Chu, EHY Hung, HHL Chau.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This research was approved by the New Territories East Cluster Research Ethics Committee/Institutional Review Board of Hospital Authority, Hong Kong (Ref No.: NTEC-2023-074). The requirement for informed patient consent was waived by the Committee due to the retrospective nature of the research.

References

1. Tabár L, Vitak B, Chen HH, Yen MF, Duffy SW, Smith RA. Beyond randomized controlled trials: organized mammographic screening substantially reduces breast carcinoma mortality. Cancer 2001;91:1724-31. Crossref

2. Majid AS, de Paredes ES, Doherty RD, Sharma NR, Salvador X. Missed breast carcinoma: pitfalls and pearls. Radiographics 2003;23:881-95. Crossref

3. Fenton JJ, Taplin SH, Carney PA, et al. Influence of computer-aided detection on performance of screening mammography. N Engl J Med 2007;356:1399-409. Crossref

4. Lehman CD, Wellman RD, Buist DS, et al. Diagnostic accuracy of digital screening mammography with and without computer-aided detection. JAMA Intern Med 2015;175:1828-37. Crossref

5. Dembrower K, Wåhlin E, Liu Y, et al. Effect of artificial intelligence–based triaging of breast cancer screening mammograms on cancer detection and radiologist workload: a retrospective simulation study. Lancet Digit Health 2020;2:e468-74. Crossref

6. Raya-Povedano JL, Romero-Martín S, Elías-Cabot E, Gubern-Mérida A, Rodríguez-Ruiz A, Álvarez-Benito M. AI-based strategies to reduce workload in breast cancer screening with mammography and tomosynthesis: a retrospective evaluation. Radiology 2021;300:57-65. Crossref

7. Erickson BJ, Korfiatis P, Kline TL, Akkus Z, Philbrick K, Weston AD. Deep learning in radiology: does one size fit all? J Am Coll Radiol 2018;15(3 Pt B):521-6. Crossref

8. Schünemann HJ, Lerda D, Quinn C, et al. Breast cancer screening and diagnosis: a synopsis of the European breast guidelines. Ann Intern Med 2020;172:46-56. Crossref

9. Bahl M. Artificial intelligence: a primer for breast imaging radiologists. J Breast Imaging 2020;2:304-14. Crossref

10. Hickman SE, Woitek R, Le EP, et al. Machine learning for workflow applications in screening mammography: systematic review and meta-analysis. Radiology 2022;302:88-104. Crossref

11. Leibig C, Brehmer M, Bunk S, Byng D, Pinker K, Umutlu L. Combining the strengths of radiologists and AI for breast cancer screening: a retrospective analysis. Lancet Digit Health 2022;4:e507-19. Crossref

12. Intelligence Unit, The Economist. Breast cancer in Asia—the challenge and response. A report from the Economist Intelligence Unit. 2016. Available from: https://www.eiuperspectives.economist.com/sites/default/files/EIU Breast Cancer in Asia_Final.pdf. Accessed 19 Nov 2017.

13. Lim YX, Lim ZL, Ho PJ, Li J. Breast cancer in Asia: incidence, mortality, early detection, mammography programs, and risk-based screening initiatives. Cancers (Basel) 2022;14:4218. Crossref

14. Salim M, Wåhlin E, Dembrower K, et al. External evaluation of 3 commercial artificial intelligence algorithms for independent assessment of screening mammograms. JAMA Oncol 2020;6:1581-8. Crossref

15. Kim HE, Kim HH, Han BK, et al. Changes in cancer detection and false-positive recall in mammography using artificial intelligence: a retrospective, multireader study. Lancet Digit Health 2020;2:e138-48. Crossref

16. D’Orsi CJ, Sickles EA, Mendelson EB, et al. ACR BIRADS® Atlas, Breast Imaging Reporting and Data System. Reston, VA: American College of Radiology; 2013.

17. McKinney SM, Sieniek M, Godbole V, et al. International evaluation of an AI system for breast cancer screening. Nature 2020;577:89-94. Crossref

18. Choi WJ, An JK, Woo JJ, Kwak HY. Comparison of diagnostic performance in mammography assessment: radiologist with reference to clinical information versus standalone artificial intelligence detection. Diagnostics (Basel) 2022;13:117. Crossref

19. Lee SE, Han K, Yoon JH, Youk JH, Kim EK. Depiction of breast cancers on digital mammograms by artificial intelligence–based computer-assisted diagnosis according to cancer characteristics. Eur Radiol 2022;32:7400-8. Crossref

20. Koch HW, Larsen M, Bartsch H, Kurz KD, Hofvind S. Artificial intelligence in BreastScreen Norway: a retrospective analysis of a cancer-enriched sample including 1254 breast cancer cases. Eur Radiol 2023;33:3735-43. Crossref

21. Weigel S, Brehl AK, Heindel W, Kerschke L. Artificial intelligence for indication of invasive assessment of calcifications in mammography screening [in English, German]. Rofo 2023;195:38-46. Crossref

22. Suleiman WI, McEntee MF, Lewis SJ, et al. In the digital era, architectural distortion remains a challenging radiological task. Clin Radiol 2016;71:e35-40. Crossref

23. Babkina TM, Gurando AV, Kozarenko TM, Gurando VR, Telniy VV, Pominchuk DV. Detection of breast cancers represented as architectural distortion: a comparison of full-field digital mammography and digital breast tomosynthesis. Wiad Lek 2021;74:1674-9. Crossref

24. Alshafeiy TI, Nguyen JV, Rochman CM, Nicholson BT, Patrie JT, Harvey JA. Outcome of architectural distortion detected only at breast tomosynthesis versus 2D mammography. Radiology 2018;288:38-46. Crossref

25. Wan Y, Tong Y, Liu Y, et al. Evaluation of the combination of artificial intelligence and radiologist assessments to interpret malignant architectural distortion on mammography. Front Oncol 2022;12:880150. Crossref

26. Kim HJ, Kim HH, Kim KH, et al. Mammographically occult breast cancers detected with AI-based diagnosis supporting software: clinical and histopathologic characteristics. Insights Imaging. 2022;13:57. Crossref

27. Lian J, Li K. A review of breast density implications and breast cancer screening. Clin Breast Cancer 2020;20:283-90. Crossref

28. Freer PE. Mammographic breast density: impact on breast cancer risk and implications for screening. Radiographics 2015;35:302-15. Crossref

29. Arora N, King TA, Jacks LM, et al. Impact of breast density on the presenting features of malignancy. Ann Surg Oncol 2010;17 Suppl 3:211-8. Crossref

30. Ma L, Fishell E, Wright B, Hanna W, Allan S, Boyd NF. Case-control study of factors associated with failure to detect breast cancer by mammography. J Natl Cancer Inst 1992;84:781-5. Crossref

31. Cheng NX, Liu LG, Hui L, Chen YL, Xu SL. Breast cancer following augmentation mammaplasty with polyacrylamide hydrogel (PAAG) injection. Aesthetic Plast Surg 2009;33:563-69. Crossref

32. Teo SY, Wang SC. Radiologic features of polyacrylamide gel mammoplasty. AJR Am J Roentgenol 2008;191:W89-95. Crossref

33. Lång K, Dustler M, Dahlblom V, Åkesson A, Andersson I, Zackrisson S. Identifying normal mammograms in a large screening population using artificial intelligence. Eur Radiol 2021;31:1687-92. Crossref

34. Rodriguez-Ruiz A, Lång K, Gubern-Merida A, et al. Stand-alone artificial intelligence for breast cancer detection in mammography: comparison with 101 radiologists. J Natl Cancer Inst 2019;111:916-22. Crossref

35. Rodríguez-Ruiz A, Krupinski E, Mordang JJ, et al. Detection of breast cancer with mammography: effect of an artificial intelligence support system. Radiology 2019;290:305-14. Crossref

Incidence of 30-day readmission after total knee arthroplasty and its associated factors in Hong Kong

Hong Kong Med J 2024 Dec;30(6):461–7 | Epub 5 Dec 2024

https://doi.org/10.12809/hkmj2310733

ORIGINAL ARTICLE

Incidence of 30-day readmission after total knee arthroplasty and its associated factors in Hong Kong

Omar WK Tsui¹; PK Chan, FHKAM (Orthopaedic Surgery), FHKCOS²; Jeffery HY Leung, BSc²; Amy Cheung, FHKAM (Orthopaedic Surgery), FHKCOS³; Vincent WK Chan, FHKAM (Orthopaedic Surgery), FHKCOS³; Michelle Hilda Luk, FHKAM (Orthopaedic Surgery), FHKCOS³; MH Cheung, FHKAM (Orthopaedic Surgery), FHKCOS²; Henry Fu, FHKAM (Orthopaedic Surgery), FHKCOS²; KY Chiu, FHKAM (Orthopaedic Surgery), FHKCOS²

¹ Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China

² Department of Orthopaedics and Traumatology, School of Clinical Medicine, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China

³ Department of Orthopaedics and Traumatology, Queen Mary Hospital, Hong Kong SAR, China

Corresponding author: Dr PK Chan (lewis@ortho.hku.hk)

Full paper in PDF

Abstract

Introduction: Total knee arthroplasty (TKA) is one of the most commonly performed orthopaedic procedures worldwide, due to the increased prevalence of osteoarthritis associated with an ageing global population. Although many studies have focused on the causes of readmission among TKA patients within 30 days post-surgery, none have been conducted in Hong Kong. This study investigated the 30-day readmission rate, causes, and risk factors among TKA patients in Hong Kong.

Methods: This retrospective review included patients who underwent TKA at a local university-affiliated hospital between 2001 and 2020. Eligible patients were identified using the Clinical Data Analysis and Reporting System and electronic patient records. Their data were analysed to determine the 30-day readmission rate, risk factors, and underlying causes.

Results: Among the 3827 TKA patients included, the male-to-female ratio was 1:2.78 (1012:2815) and the mean age (±standard deviation) was 71.11±8.82 years. Of these patients, 3.4% underwent unplanned readmission to hospitals through the Accident and Emergency Department within 30 days of TKA. The most common causes of readmission were knee pain (33.1%), knee swelling (26.2%), and gastrointestinal-related conditions (8.5%). Age ≥80 years (odds ratio [OR]=1.63; P=0.01) and hypertension (OR=2.08; P<0.001) were risk factors for readmission. Bilateral simultaneous TKA (OR=0.42; P=0.005) was associated with lower risk of readmission.

Conclusion: The readmission rate in this study was 3.4%, comparable to rates in previous reports. Enhanced patient education and optimised perioperative pain management are needed to minimise hospital readmissions. Fall prevention, cautious painkiller prescribing, and improved nursing care are recommended to prevent readmission.

New knowledge added by this study

Pain (33.1%) and swelling (26.2%) are the most common causes of readmission after total knee arthroplasty (TKA) in Hong Kong.
Age ≥80 years and hypertension are major risk factors for readmission, whereas simultaneous bilateral TKA is associated with a lower risk of readmission.
The male-to-female ratio is 1:2.78 in Hong Kong, which is lower than the ratio in other countries.

Implications for clinical practice or policy

Pain management and education should be enhanced.
Fall prevention, cautious painkiller prescribing, and improved nursing care are recommended.

[Abstract in Chinese]

Introduction

Due to the increasing incidence of osteoarthritis associated with the ageing global population, total knee arthroplasty (TKA) has become one of the most commonly performed orthopaedic procedures worldwide. The most common approach to determine causes and risk factors involves analysing readmission episodes among TKA patients within 30 days post-surgery.1 2 The 30-day readmission rate provides insight into the prevalence of postoperative complications, whereas the length of stay after readmission reflects the severity of those complications. A review of readmission causes is needed to assess the quality of hospital care and determine the adequacy of patient education (eg, wound management).3 An understanding of the 30-day readmission rate, causes, and risk factors can help hospitals improve clinical guidelines, reduce medical and surgical complications,4 and reduce the financial burden of treatment for these complications.5

Although multiple studies worldwide have adequately explored the 30-day readmission causes, rate, and length of stay among TKA patients,6 7 revealing important clinical insights, no such studies have been conducted in Hong Kong. The current study aimed to investigate the 30-day readmission rate, causes, and risk factors among TKA patients in the city.

Methods

This retrospective study included all patients who underwent TKA at our local university-affiliated hospital and were readmitted through an Accident and Emergency Department (AED) between 2001 and 2020. It evaluated the epidemiological characteristics, readmission causes, and preoperative co-morbidities of TKA patients.

We utilised data from the Clinical Data Analysis and Reporting System (CDARS), a well-established platform developed by the Hospital Authority (HA). The CDARS contains patient data, such as laboratory reports and radiological images; it covers all outpatients and inpatients at 43 public hospitals and institutions across seven service clusters in Hong Kong. Records in the CDARS include the details of patients with unplanned 30-day readmission to the AED of an HA hospital from either their homes or rehabilitation facilities, along with their discharge information. This platform is extensively used by research teams across Hong Kong.8 We obtained a list of TKA patients who underwent surgery at the study hospital and were readmitted to an HA hospital within 30 days. We matched these patient names with their corresponding electronic patient records to determine the reasons for readmission.

For patients who experienced 30-day readmission, both the records in the CDARS and electronic patient records were reviewed. For patients who did not require 30-day readmission, only CDARS records were reviewed. Medication records (ie, dispensing dates, dosages, and durations) were extracted from CDARS records to identify co-morbidities (online supplementary Appendix). All patient data were de-identified.

Based on factors described by Roger et al,6 we classified reasons for readmission into the following categories: orthopaedics-related, surgery-related, gastrointestinal-related, urological-related, neurological-related, cardiac-related, respiratory-related, renal-related, medication-related, and others. Orthopaedic specialists performed the classification to determine the cause of readmission.

The inclusion criteria were a recent history of TKA at our institution, readmission through the AED of an HA hospital, and inpatient admission. The exclusion criteria were a history of knee surgery, incomplete clinical assessment data, and/or orthopaedic tumours in the knee (for paediatric patients only).

Analyses of readmission cause, number, and rate, as well as organ dysfunction episodes, were episode-based. The analysis of risk factors for readmission was patient-based. Risk factors/co-morbidities were identified based on medications prescribed to the patients. If a patient received antihypertensive medication, that patient was assumed to have hypertension.

Data analysis was performed using R (R Foundation for Statistical Computing, Vienna, Austria) and R Studio software. All statistical tests were two-sided, and a 5% significance threshold was applied. The investigators and their research assistants were responsible for data collection and had access to the source data and study records. To evaluate categorical variables, Chi squared tests and/or Fisher’s exact tests were conducted, depending on the observed frequencies. To evaluate continuous variables, the Kruskal–Wallis test was used.

Results

In total, 3878 records were initially reviewed; of these, 43 were excluded due to the presence of tumours (ie, osteosarcoma in the distal tibia), three were excluded due to incorrect data entry for revision surgery, and five were excluded because they constituted duplicate entries for the same readmission episode (online supplementary Fig 1).

Basic demographic data

Of the 3827 valid patient records, 2855 were included in the initial analysis after removal of duplicate records for 972 patients who underwent two unilateral TKAs during different admission episodes. Of the 3827 patients, 2815 (73.6%) were women and 1012 (26.4%) were men. The mean ages were 71.11 years for TKA patients who did not experience readmission and 73.10 years for TKA patients who experienced readmission (Table 1 6 9 10 11 12). The mean postoperative length of stay (±standard deviation) was 6.85±6.19 days (Table 2). Thus, the readmission rate at our institution was 3.4%, similar to rates reported worldwide (Table 1).6 7 13 There was an increase in the 30-day readmission rate between 2001 and 2020 (Table 2). The number of TKAs performed in our institution increased from 2001 to 2014 and remained consistently high (>200 TKAs annually except in 2020) [online supplementary Fig 2], in line with published literature.14

Table 1. Comparison of results with reports worldwide

Table 2. Trends in patient readmission (n=3827)

In total, 130 patients with valid readmission records were analysed to identify causes and risk factors (online supplementary Fig 1). Of these patients, 90 (69.2%) were women and 40 (30.8%) were men. The median length of stay after readmission was 2 days (interquartile range=1.25-5) and the mean time between surgery and readmission was 22 days (Table 3).

Table 3. General epidemiology of readmitted patients (n=130)*

Causes of readmission and associated risk factors

Unilateral knee pain (33.1%), unilateral knee swelling (26.2%), and gastrointestinal-related conditions (8.5%) were the most common causes of readmission (Table 4). Hypertension (67.7%) and diabetes mellitus (22.7% before March 2017 from which our institution modified the preoperative management pathway) were the most common co-morbidities among readmitted patients. Additionally, hypertension (odds ratio [OR]=2.08; P<0.001) and age ≥80 years (OR=1.63; P=0.01) were identified as significant risk factors for readmission (Table 5 and online supplementary Table 1). Patients who underwent bilateral TKA had a 58% lower risk of readmission (Table 6), possibly because they had better health condition before surgery and received more rigorous preoperative screening for high-risk co-morbidities.

Table 4. Causes of readmission among total knee arthroplasty patients (n=130)*

Table 5. Risk factors for readmission

Table 6. Numbers of unilateral and bilateral patients according to episode-based data

Discussion

Our results suggest that there is a substantial rate of readmission due to pain and swelling among TKA patients in Hong Kong, which is higher than the rates in previous studies (Table 1).6 9 12 Hospital resources should be reviewed to determine whether these patients require admission because most readmitted patients have non-severe conditions. To reduce the unnecessary allocation of clinical resources to non-severe cases, alternatives such as designated nurse clinics15 and patient consultation hotlines can provide medical advice for managing minor conditions (eg, pain and swelling) at home. These measures can reduce the workload of orthopaedic surgeons and improve postoperative follow-up care.

Hypertension and age ≥80 years were significant risk factors for readmission. The mean age of TKA patients in Hong Kong was higher than the mean ages of TKA patients in similar studies worldwide (Table 1)6 9 10 11 12; this difference aligns with the fact that Hong Kong has the longest life expectancy globally (mean age of 85.16 years in 2022).16

The increased risk of readmission with old age, consistent with findings in a previous study,17 may be related to the greater likelihood for older individuals to visit the AED for non-orthopaedic issues. In contrast, patients who underwent simultaneous bilateral TKA had a lower risk of readmission (OR=0.42; P=0.005) [Table 6].

At our institution, patients who underwent simultaneous bilateral TKA were aged <75 years and had no clinically significant cardiovascular co-morbidities (eg, stroke). Furthermore, in March 2017, our institution introduced routine glycated haemoglobin screening to identify diabetic and prediabetic patients, with the goal of minimising postoperative complications. Diabetes mellitus is known to increase the risk of periprosthetic joint infection after surgery.18 Patients with elevated glycated haemoglobin levels were referred to endocrinologists for better management of diabetes mellitus prior to TKA, thereby decreasing the OR for readmission from 1.24 to 0.74 (Table 5 and online supplementary Table 1). Considering that a substantial number of patients with the aforementioned co-morbidities exhibit a higher risk of readmission, the perioperative protocol could be improved. Suggested changes could include better coordination with each patient’s family medicine specialists or general practitioners, who usually have a better understanding of the patient’s underlying medical conditions, to develop effective preoperative and postoperative care plans.

Overall, 4.6% (n=6) of the patients were readmitted primarily due to falls (Table 4). This finding highlights the need for enhanced nursing support and education to prevent post-surgical falls among TKA patients. Furthermore, stronger occupational therapy and household aid programmes can help prevent falls at home and improve patient rehabilitation. Another 4.6% of patients were readmitted due to the adverse drug effects (Table 4 and online supplementary Table 2), particularly from tramadol/codeine/morphine and related medications; symptoms included vomiting and constipation. These results indicate a need for cautious painkiller prescribing to prevent future medication-related readmissions.

Strengths and limitations

To the best of our knowledge, this is the first study on the readmission rate, causes, and risk factors among TKA patients within 30 days post-surgery. It also compared data collected in Hong Kong with results from other studies, yielding insights for local orthopaedic surgeons who seek to improve suboptimal surgical outcomes.

Notably, there were some limitations. Patient data in the CDARS may be incomplete because some doctors might have omitted the International Classification of Diseases, Ninth Revision codes for certain co-morbidities. To mitigate this issue, prescribed drugs were used to identify patients’ co-morbidities. However, this approach may have missed some patients with co-morbidities and no associated medication records.

Conclusion

This study showed that older TKA patients with hypertension were more likely to be readmitted through the AED within 30 days post-surgery. The most common reasons for readmission were pain, swelling, and gastrointestinal-related symptoms. To reduce readmissions, hospitals should place greater emphasis on pain and wound management for TKA patients. Furthermore, patient education efforts should be strengthened to increase awareness of pain and wound management.

Author contributions

Concept or design: OWK Tsui, PK Chan.
Acquisition of data: OWK Tsui, PK Chan, JHY Leung.
Analysis or interpretation of data: OWK Tsui, PK Chan, JHY Leung.
Drafting of the manuscript: OWK Tsui.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Declaration

The research was presented at 42nd Annual Congress of the Hong Kong Orthopaedic Association, 5 November 2022, Hong Kong.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This research was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster, Hong Kong (Ref No.: UW-22-313). The requirement for patient consent was waived by the Board due to the retrospective nature of the research.

Supplementary material

The supplementary material was provided by the authors and some information may not have been peer reviewed. Accepted supplementary material will be published as submitted by the authors, without any editing or formatting. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by the Hong Kong Academy of Medicine and the Hong Kong Medical Association. The Hong Kong Academy of Medicine and the Hong Kong Medical Association disclaim all liability and responsibility arising from any reliance placed on the content.

References

1. Howie CM, Mears SC, Barnes CL, Stambough JB. Readmission, complication, and disposition calculators in total joint arthroplasty: a systemic review. J Arthroplasty 2021;36:1823-31. Crossref

2. D’Apuzzo M, Westrich G, Hidaka C, Jung Pan T, Lyman S. All-cause versus complication-specific readmission following total knee arthroplasty. J Bone Joint Surg Am 2017;99:1093-103. Crossref

3. Chambers MC, El-Othmani MM, Anoushiravani AA, Sayeed Z, Saleh KJ. Reducing 30-day readmission after joint replacement. Orthop Clin North Am 2016;47:673-80. Crossref

4. Bosco JA 3rd, Karkenny AJ, Hutzler LH, Slover JD, Iorio R. Cost burden of 30-day readmissions following Medicare total hip and knee arthroplasty. J Arthroplasty 2014;29:903-5. Crossref

5. Gould D, Dowsey MM, Spelman T, et al. Patient-related risk factors for unplanned 30-day hospital readmission following primary and revision total knee arthroplasty: a systematic review and meta-analysis. J Clin Med 2021;10:134. Crossref

6. Roger C, Debuyzer E, Dehl M, et al. Factors associated with hospital stay length, discharge destination, and 30-day readmission rate after primary hip or knee arthroplasty: retrospective cohort study. Orthop Traumatol Surg Res 2019;105:949-55. Crossref

7. Bovonratwet P, Shen TS, Ast MP, Mayman DJ, Haas SB, Su EP. Reasons and risk factors for 30-day readmission after outpatient total knee arthroplasty: a review of 3015 cases. J Arthroplasty 2020;35:2451-7. Crossref

8. Sing CW, Woo YC, Lee AC, et al. Validity of major osteoporotic fracture diagnosis codes in the Clinical Data Analysis and Reporting System in Hong Kong. Pharmacoepidemiol Drug Saf 2017;26:973-6. Crossref

9. Phruetthiphat OA, Otero JE, Zampogna B, Vasta S, Gao Y, Callaghan JJ. Predictors for readmission following primary total hip and total knee arthroplasty. J Orthop Surg (Hong Kong) 2020;28:2309499020959160. Crossref

10. Ross TD, Dvorani E, Saskin R, Khoshbin A, Atrey A, Ward SE. Temporal trends and predictors of thirty-day readmissions and emergency department visits following total knee arthroplasty in Ontario between 2003 and 2016. J Arthroplasty 2020;35:364-70. Crossref

11. Kurtz SM, Lau EC, Ong KL, Adler EM, Kolisek FR, Manley MT. Which hospital and clinical factors drive 30- and 90-day readmission after TKA? J Arthroplasty 2016;31:2099-107. Crossref

12. Ali AM, Loeffler MD, Aylin P, Bottle A. Predictors of 30-day readmission after total knee arthroplasty: analysis of 566,323 procedures in the United Kingdom. J Arthroplasty 2019;34:242-8.e1. Crossref

13. Urish KL, Qin Y, Li BY, et al. Predictors and cost of readmission in total knee arthroplasty. J Arthroplasty 2018;33:2759-63. Crossref

14. Yan CH, Chiu KY, Ng FY. Total knee arthroplasty for primary knee osteoarthritis: changing pattern over the past 10 years. Hong Kong Med J 2011;17:20-5.

15. Fan JC, Lo CK, Kwok CK, Fung KY. Nurse-led orthopaedic clinic in total joint replacement. Hong Kong Med J 2014;20:511-8. Crossref

16. MacroTrends. Hong Kong life expectancy 1950-2024. Available from: https://www.macrotrends.net/countries/HKG/hong-kong/life-expectancy. Accessed 28 Nov 2024.

17. Cheung A, Fu H, Cheung MH, et al. How well do elderly patients do after total knee arthroplasty in the era of fasttrack surgery? Arthroplasty 2020;2:16. Crossref

18. Chan VW, Chan PK, Woo YC, et al. Universal haemoglobin A1c screening reveals high prevalence of dysglycaemia in patients undergoing total knee arthroplasty. Hong Kong Med J 2020;26:304-10. Crossref

Public fertility preservation programme for cancer patients in Hong Kong

Hong Kong Med J 2024 Dec;30(6):452–60 | Epub 17 Dec 2024

https://doi.org/10.12809/hkmj2311060

ORIGINAL ARTICLE

Public fertility preservation programme for cancer patients in Hong Kong

Dorothy TY Chan, MB, BS; Jennifer KY Ko, MB, BS, MRCOG; Kevin KW Lam, BSc, PhD; YW Tong, MB, BS, MRCOG; Evelyn Wong, MB, BS, MRCOG; Heidi HY Cheng, MB, BS, MRCOG; Sofie SF Yung, MB, BS, MRCOG; Raymond HW Li, MD, FRCOG; Ernest HY Ng, MD, FRCOG

Department of Obstetrics and Gynaecology, School of Clinical Medicine, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China

Corresponding author: Dr Raymond HW Li (raymondli@hku.hk)

Full paper in PDF

Abstract

Introduction: Fertility preservation (FP) offers cancer patients the opportunity to have biological children after completing treatment. This study was performed to review the experience and changes in service demand since the implementation of a public FP programme for cancer patients in Hong Kong.

Methods: This retrospective study included men and women who attended an assisted reproduction unit for public FP services before cancer treatment from August 2020 to February 2023. Their medical records were reviewed and the results were compared with findings from our previous study to evaluate trends in service demand.

Results: During the study period, there were 48 consultations for female FP, compared with 72 women who presented for FP from 2010 to 2020 prior to establishment of the public FP programme. The median time from referral to consultation was 3 days (interquartile range [IQR]=2-5). Eighteen women (37.5%) underwent 19 cycles of ovarian stimulation for oocyte or embryo cryopreservation. Thirty women (62.5%) received gonadotropin-releasing hormone agonists during cancer treatment. There were 58 consultations for male FP during the study period, compared with 265 men who presented for sperm cryopreservation from 2005 to 2020. The median time from referral to consultation was 4 days (IQR=2-7). Fifty-five men (94.8%) attempted sperm cryopreservation, and 49 (84.5%) successfully preserved sperm.

Conclusion: Since the establishment of a public FP programme for cancer patients, there has been an increase in the demand for FP services at our centre. Regular review of FP services is warranted to assess changes in demand and identify areas for improvement.

New knowledge added by this study

Since the establishment of a public fertility preservation (FP) programme, there has been an increase in the number of patients seeking FP services at our centre.
Reproductive-age men seeking FP were more likely than reproductive-age women to undergo gamete cryopreservation.
Only 62.5% of women received gonadotropin-releasing hormone agonists during cancer treatment; the reasons for not receiving the agonists were not recorded.

Implications for clinical practice or policy

The cost of FP may be a barrier to patients considering this option.
Public funding for medications and gamete storage can support reproductive-age patients in pursuing FP before cancer treatment.
Further research is needed to improve FP, especially for reproductive-age women.

[Abstract in Chinese]

Introduction

Many individuals are diagnosed with cancer during childhood, adolescence, and young adulthood. Worldwide, there were approximately 1 335 100 new cancer cases among adolescents and young adults in 20191; the incidence rate was 44.99 per 100 000 people.1 In 2020, the incidence rate for cancer among Hong Kong children and adolescents (aged 0-19 years) was 160 cases per 1 000 000 people.2 There were 177 newly diagnosed cancer cases in this age-group (92 in male patients and 85 in female patients).2 The survival rates for childhood and adolescent cancers are encouraging. In a retrospective cohort study from a research hospital in the United States,3 the 5-year overall survival rate exceeded 83%. Similarly, in Hong Kong, the 5-year survival rate among women diagnosed with breast cancer, the most common cancer in reproductive-age women, was 84% between 2010 and 2017.2

Chemotherapy or pelvic radiotherapy may affect fertility, either temporarily or permanently. Considering advances in cancer treatment and improved post-treatment survival rates, fertility should be discussed at the time of cancer diagnosis, especially for younger patients who have not yet completed their families. International guidelines regarding fertility preservation (FP) recommend that clinicians inform cancer patients about the potential effects of cancer and its treatment on reproductive function, as well as FP options.4 5 In a semi-structured phone interview study of female cancer survivors who were diagnosed with invasive cervical cancer, breast cancer, Hodgkin lymphoma, or non-Hodgkin lymphoma at age ≤40 years, participants were interviewed an average of 10 years after diagnosis.6 Those who had wanted children at the time of diagnosis but were unable to conceive subsequently reported distress related to their interrupted fertility.6 Additionally, patients who do not receive accurate and timely information regarding FP are at risk for psychological distress.7 In our recently published cross-sectional questionnaire study of reproductive-age women in Hong Kong who had been diagnosed with breast cancer,8 only 44% of those women were aware of FP; however, 46% of the women felt that fertility concerns affected their cancer treatment decisions.8

The most common FP options include sperm cryopreservation for men and embryo or oocyte cryopreservation for women. Other options for women include pharmacological ovarian protection using gonadotropin-releasing hormone (GnRH) agonists, ovarian tissue cryopreservation, and ovarian transposition. In Hong Kong, FP was previously self-funded and only available through private services. Sperm cryopreservation costs approximately HK$4400 to HK$6600 for 2 years, whereas oocyte and embryo cryopreservation costs are approximately HK$15 000 to HK$20 000.9 Our centre launched the first public FP programme for cancer patients in Hong Kong, beginning in August 2020. Here, we review the two-and-a-half-year experience of providing public FP services to cancer patients in Hong Kong.

Methods

This retrospective study included men and women who attended the Centre of Assisted Reproduction and Embryology at The University of Hong Kong—Queen Mary Hospital for FP services before cancer treatment, from the establishment of our public FP programme in August 2020 until the end of February 2023.

Criteria for public fertility preservation services

During the study period, we provided public FP for cancer patients <35 years old, expressed a desire for future fertility, had a survival rate exceeding 50% after cancer treatment, had no living children, and had not undergone prior chemotherapy or pelvic radiotherapy. In women, an antral follicle count of >7 on pelvic ultrasound was required. These criteria were adapted from The Edinburgh Selection Criteria for ovarian tissue cryopreservation.10

There was no minimum age requirement for FP. Male adolescents could undergo sperm freezing if they were able to provide sperm samples for cryopreservation. For patients aged <18 years, we included their parents in discussions prior to proceeding with FP treatment.

During the study period, the public FP programme offered up to 40 cycles of sperm freezing and 20 cycles of oocyte/embryo freezing per year.

Referral process

Patients diagnosed with cancer who were expected to undergo gonadotoxic treatments were referred to our FP service by surgeons, oncologists, paediatricians, haematologists, private practitioners, and cancer support groups. Clinicians completed a referral letter, which can be downloaded from our centre’s website.11 Patients or their doctors can also contact us via email. Additionally, a chat group was established between Hong Kong Children’s Hospital and our centre to facilitate rapid referrals.

After we received a referral, the patient was scheduled for an appointment in the public FP clinic within 1 week. Our centre maintained a flexible clinic schedule, which allowed urgent cases to be accommodated within the existing clinic framework, 5 days per week.

Fertility preservation counselling

The details of our FP programme were previously published.12 Information sheets and videos about the FP services offered by our centre were readily accessible to the general population and patients through our website13 and YouTube channel.14 Patients were encouraged to review these materials before attending the public FP clinic. For men, sperm banking was arranged on the same day as counselling. For women, the options of oocyte and embryo preservation were discussed if feasible. Embryo preservation was only offered to women who were legally married. In Hong Kong, assisted reproductive technology is regulated by the Human Reproductive Technology Ordinance.15 This ordinance limits the storage duration for frozen gametes in cancer patients to 10 years or until the patient reaches the age of 55 years, whichever is longer.15 The storage duration for frozen embryos is limited to 10 years.15 Cryopreserved gametes and embryos can only be used after a patient recovers from their illness and is legally married.15 Posthumous use of cryopreserved gametes and embryos is prohibited.15

The use of GnRH agonists for pharmacological ovarian protection was discussed either after cryopreservation or if cryopreservation was not feasible. Such agonists were usually administered monthly or every 3 months during chemotherapy.

The characteristics of men who underwent sperm cryopreservation and women who underwent ovarian stimulation for oocyte or embryo cryopreservation were prospectively entered into our database. Medical records (both in paper and electronic formats), including data from the assisted reproductive technology database at our centre and the Hospital Authority’s electronic clinical management system, were retrieved and reviewed. These records encompassed demographic data, cancer type, cancer treatment, FP method chosen, ovarian stimulation cycle characteristics, semen analysis, reproductive outcomes, and follow-up information, if available.

All women who attended our centre for FP were asked to return to our late-effects clinic for gonadal function monitoring after the completion of cancer treatment. All men were asked to undergo semen analysis when they wished to conceive after the completion of cancer treatment.

Statistical analysis

Data were analysed using SPSS (Windows version 26; IBM Corp, Armonk [NY], United States) and are presented as median (interquartile range [IQR]) or as number (percentage). P value was calculated by Chi squared test.

Methods

Women

Fifty-two women were referred to our public FP clinic between August 2020 and February 2023. Three women were excluded from the analysis because they had non-malignant conditions, including rheumatological disease (systemic lupus erythematosus) and neurological disease (multiple sclerosis). Additionally, one woman missed her clinic appointment. Therefore, the final analysis included 48 women (Fig a). The median age of these women was 30 years (IQR=25-33). The cancer outcomes of these women are shown in Table 1. Regarding marital status, 36 women (75.0%) were single, 11 (22.9%) were married, and one (2.1%) was divorced. All were nulliparous, except for one married woman (2.1%) with a livebirth was ineligible for publicly funded FP due to the programme’s criteria. She then selected GnRH agonist treatment after counselling. The median time from referral to consultation was 3 days (IQR=2-5).

Figure. Patients referred for fertility preservation. (a) Female patients. (b) Male patients

Table 1. Diagnoses and outcomes of cancer among women seeking fertility preservation (n=48)

Eighteen women underwent 19 cycles of ovarian stimulation for oocyte or embryo cryopreservation (Table 2). One woman underwent an additional self-financed stimulation cycle because she only achieved two frozen oocytes in the first cycle. She achieved two additional frozen oocytes in the second attempt. Thirteen women cryopreserved oocytes, whereas five women cryopreserved embryos (three at the cleavage stage and two at the blastocyst stage). The median time between consultation and ovarian stimulation was 5 days (IQR=2-12) [Table 2]. All women with breast cancer received letrozole co-treatment during ovarian stimulation.

Table 2. Cycle characteristics of women who underwent ovarian stimulation (n=19)

One woman developed moderate ovarian hyperstimulation syndrome requiring hospital admission. Oocyte retrieval was uneventful, and 45 oocytes were retrieved. However, 3 days after oocyte retrieval, she was admitted with abdominal distension, shortness of breath, and vomiting. She was diagnosed with moderate ovarian hyperstimulation syndrome, which resolved with conservative management.

There was no significant age difference between women who proceeded with oocyte/embryo cryopreservation and those who did not. The median age of women who proceeded with oocyte/embryo cryopreservation was 28 years (IQR=24.0-32.8), whereas the median age of women who did not proceed with oocyte/embryo cryopreservation was 31 years (IQR=26.8-33.0) [Table 3].

Table 3. Characteristics of women with or without oocyte/embryo cryopreservation (n=48)

Among patients with breast and gynaecological cancers, six of 10 (60.0%) and six of 13 (46.2%) underwent oocyte/embryo cryopreservation, respectively, compared with five of 19 (26.3%) women with haematological cancers and one of six (16.7%) women with other solid tumours (Table 3).

Among the 48 women who attended the clinic, nine (18.8%) proceeded with oocyte or embryo cryopreservation alone, nine (18.8%) underwent cryopreservation followed by the use of GnRH agonists, 21 (43.8%) received GnRH agonists alone, five (10.4%) decided against FP after counselling, and four (8.3%) were lost to follow-up. Those who chose GnRH agonists received this treatment from their primary oncology team.

At the end of February 2023, among the 48 women, 22 exhibited disease remission, 21 were continuing treatment, four were deceased, and one had been lost to follow-up (Table 1). None of the women have returned to use their frozen oocytes or embryos, nor have any reported natural conception since their cancer diagnosis.

Men

Sixty-six men were referred to our public FP clinic during the study period (Fig b). Five men were excluded: four had exceeded the age limit and one had already begun chemotherapy. Fertility preservation counselling at a private clinic was offered to those who were not eligible for the public FP service. One man, who exceeded the age limit, underwent self-funded sperm cryopreservation. Three men missed their clinic appointments. Therefore, the final analysis included 58 men (Fig b). The median age of the men was 26 years (IQR=18.3-32.8). The cancer outcomes of these men are shown in Table 4. Regarding marital status, 51 men (87.9%) were single and seven men (12.1%) were married. One man (1.7%) had a child but was unmarried. The remaining 57 men (98.3%) had no offspring. The median time from referral to consultation was 4 days (IQR=2-7).

Table 4. Diagnoses and outcomes of cancer in men seeking fertility preservation (n=58)

Among the 58 men who attended the clinic, 55 attempted sperm freezing and three chose not to undergo cryopreservation after counselling. Six men were unable to cryopreserve sperm (Fig b). One, aged 14 years, was unable to provide a semen sample; four men submitted semen samples containing no sperm. One man had previously attempted sperm cryopreservation at a private hospital, but no sperm were found in his ejaculate. He subsequently underwent testicular sperm extraction at our hospital; no sperm were retrieved. The ages of the men with no sperm in their semen ranged from 15 to 34 years.

The median number of vials of cryopreserved sperm was 5 (IQR=5-5) and the median sperm concentration was 18.8 million/mL (IQR=4.3-52.8).

At the end of February 2023, among the 58 men, 29 exhibited disease remission, 18 were continuing treatment, six were deceased, and five had been lost to follow-up (Table 4). None of the men have returned to use their frozen sperm.

As of this writing, six men and four woman who attended the FP clinic have died.

Discussion

This is the first review of a public FP programme for cancer patients in Hong Kong. Our study showed that among the 48 women who attended during the study period, 37.5% (n=18) underwent oocyte/embryo cryopreservation and 62.5% (n=30) chose GnRH agonists for FP. In contrast, among the 58 men who attended for FP before cancer treatment, >90% attempted sperm cryopreservation.

We previously published a review of our self-funded FP service from 2010 to 2020.12 During that period, 72 women attended consultations for FP, and 20 of them underwent 22 cycles of ovarian stimulation for oocyte or embryo cryopreservation.12 Additionally, from 1995 to 2020, 265 men underwent sperm cryopreservation.12 Over the years, there were increases in the numbers of men and women seeking FP; the increase was more prominent among women.

For comparison, we selected the period from 2018 to 2020 (ie, the 2.5 years immediately preceding the launch of the public FP programme). During that period, 19 women were referred for self-funded FP prior to cancer treatment, and 10 (52.6%) underwent oocyte or embryo cryopreservation. Fifty-eight men were referred for FP and underwent sperm cryopreservation. In the years prior to the launch of the publicly funded FP programme, we had already begun networking with various specialties, which likely contributed to the gradual increase in awareness and demand for FP services.

Public fertility preservation programme

A successful FP programme requires good networking, flexibility, and a patient-friendly clinic environment. During the establishment of the public FP programme, we have networked with other specialties to enhance collaboration. Our centre aimed to simplify logistics so that consultations could be arranged as quickly as possible, allowing FP counselling and procedures to be completed within the short window of opportunity before cancer treatment. In our public FP clinic, the median waiting times from referral to consultation were 3 days for women and 4 days for men. Among women who chose oocyte or embryo cryopreservation, the median time from consultation to the start of ovarian stimulation was 5 days (IQR=2-12). In our previous study, the time from consultation to oocyte retrieval was 17 days (IQR=13-30).12 Notably, our previous study did not investigate the waiting time from referral to consultation; therefore, direct comparisons cannot be performed. Compared with our previous study regarding FP for cancer patients at our centre,12 the proportion of women who ultimately underwent oocyte or embryo cryopreservation increased from 28% to 38% in the public FP programme. However, further monitoring is needed to determine whether this difference represents a true upward trend due to increased awareness and easier access to the service. Additionally, patient characteristics and cancer types may vary across time periods.

For reproductive-age women with cancer, the receipt of specialised counselling regarding fertility issues, followed by FP, has been linked to less regret and improved quality of life among survivors.16 Providing our patients with accessible FP counselling and affordable treatments is an essential aspect of comprehensive oncology care. A clinical practice guideline from the American Society of Clinical Oncology indicates that FP should be initiated as early as possible in the treatment process to allow for the widest range of options.17 Referral to FP services enables patients to receive counselling from reproductive medicine specialists, empowering them to make informed decisions about fertility treatment.

At our FP clinic, patients were able to consult reproductive medicine specialists who discussed the potential effects of gonadotoxic cancer treatments on future fertility and described FP options. Local regulations concerning gamete storage and assisted reproduction were also explained. Patients were informed that they must be legally married to use frozen gametes in the future, and that gametes cannot be used posthumously.

Gonadotropin-releasing hormone agonists

In our cohort, only 62.5% of women received GnRH agonists during cancer treatment; the reasons for not receiving GnRH agonists were not recorded. Gonadotropin-releasing hormone agonists are usually administered monthly or every 3 months during cancer treatment, although their effectiveness depends on the type of cancer treatment. Some studies of breast cancer patients have shown that GnRH agonists can reduce the risk of premature ovarian insufficiency, but the fertility benefit remains uncertain.18 19 20 Most studies have focused on outcomes such as the maintenance or resumption of menstruation, prevention of treatment-related premature ovarian failure, and ovulation. In a Cochrane review20 which discussed randomised controlled trials that examined the effect of GnRH analogues for chemotherapy-induced ovarian failure in premenopausal women, 12 randomised controlled trials were included. Eleven studies reported rates of menstruation recovery or maintenance, four studies measured treatment-related premature ovarian failure, and seven studies reported the rates of pregnancy.20 However, there are limited data regarding live birth rates.20 A meta-analysis of randomised studies concerning ovarian suppression using GnRH agonists during chemotherapy in breast cancer patients found that temporary ovarian suppression with a GnRH agonist in young breast cancer patients was associated with a reduced risk of chemotherapy-induced premature ovarian insufficiency; it also appeared to increase the pregnancy rate without negatively influencing prognosis.21 Thus far, the benefit of GnRH agonists in other malignancies is unclear. A long-term analysis of young female lymphoma patients showed that GnRH agonists were not effective in preventing chemotherapy-induced premature ovarian insufficiency and did not improve future pregnancy rates.22 According to the European Society of Human Reproduction and Embryology guideline on female FP,4 GnRH agonists should be offered as an option for protecting ovarian function in premenopausal breast cancer patients undergoing chemotherapy; importantly, limited evidence exists regarding their protective effects on ovarian reserve and potential future pregnancies.4 In malignancies other than breast cancer, GnRH agonists should not be routinely offered as an option for protecting ovarian function protection and FP without discussing the uncertainty of their benefit.4 Gonadotropin-releasing hormone agonists during chemotherapy should not be considered as a substitute for established FP techniques, such as cryopreservation. They can be offered in addition to cryopreservation or when such techniques are not feasible.4 Despite the use of GnRH agonists, patients may experience premature ovarian insufficiency. Gonadotropin-releasing hormone agonists are currently provided as a self-financed option; women are often referred back to their oncology team, who prescribes and administers these agonists after FP counselling. The proportion of patients who underwent oocyte/embryo cryopreservation was higher among those with gynaecological or breast cancers than among those with haematological malignancies. This difference is likely due to the nature of their diseases and the urgency of initiating cancer treatment.

Oocyte or embryo cryopreservation

For women who chose to proceed with ovarian stimulation for oocyte or embryo cryopreservation, oocytes were retrieved during a stimulated cycle. Recombinant follicle-stimulating hormone could be initiated on any day of the menstrual cycle for ovarian stimulation (ie, ‘random-start’), using either a GnRH antagonist or progestin-primed protocol. This random-start approach allowed ovarian stimulation without substantial delays and did not affect the number or quality of retrieved oocytes.4 For women with hormone-sensitive cancers (eg, breast cancer), letrozole was routinely used during ovarian stimulation. The concomitant use of letrozole reduced circulating oestrogen levels and did not impair the efficacy of ovarian stimulation.23 A systematic review and meta-analysis regarding the safety of hormonal stimulation in young women with breast cancer before starting cancer treatment, as well as survivors who underwent assisted reproduction after cancer treatment, showed no increased risk of breast cancer recurrence in women who underwent ovarian stimulation with concomitant letrozole treatment.24 Despite using the ‘random-start’ approach, one cycle of ovarian stimulation required approximately 2 weeks.

Limitations

This study had some limitations. It was a retrospective, single-centre study conducted over a short period of time; thus, it may not reflect situations in other regions. Due to resource constraints, we only included cancer patients who had not begun cancer treatment. Patients who did not meet the criteria for the public FP programme but still wished to pursue FP were referred to private clinics or other private centres upon receipt of their referral and therefore were excluded from this review. Patients who had already begun cancer treatment were also excluded from the public FP service. However, they could still be referred to our centre after stabilisation to assess fertility and explore self-funded FP options before undergoing more toxic chemotherapy, non–fertility-sparing radiotherapy, or surgeries. At the time of writing, our centre has not yet offered ovarian or testicular tissue cryopreservation. A 2018 survey of several Asian countries (eg, Australia, China, and India) revealed that ovarian tissue cryopreservation was available for prepubertal girls and postpubertal women who were unable to delay the initiation of chemotherapy.25 Testicular tissue cryopreservation also was provided to prepubertal boys in Australia, China, India, Indonesia, Japan, and Taiwan.25 A recently published pilot study from Hong Kong demonstrated the feasibility of ovarian tissue cryopreservation and transplantation using xenografts in nude mice26; ovarian tissue cryopreservation has recently become available in Hong Kong.

The patients included in this study were counselled for FP, and many are still undergoing cancer treatment and monitoring; none have returned to use the stored material. They were advised to return after cancer treatment for follow-up regarding their gonadal function. Patient satisfaction should also be evaluated. However, at the time of cryopreservation—typically close to the time of cancer diagnosis—patients may feel overwhelmed by the diagnosis and planned cancer treatments. Thus, patient satisfaction may be more accurately evaluated when the cancer is controlled or in remission.

Future outlook

Despite the presence of the public FP programme, patients were required to pay for the medications used in ovarian stimulation, as well as the fees involved in oocyte handling, freezing, and storage of frozen gametes or embryos; these costs were considerably reduced compared with expenses in private clinics. Cost remains a major barrier to accessing FP services. Although a public healthcare system has been established in Hong Kong, cancer patients are often financially overwhelmed due to the loss of income after a cancer diagnosis, along with additional expenditures for various self-funded investigations or treatments. We recently performed a survey of the knowledge, attitudes, and intentions regarding FP among breast cancer patients; most participants thought that FP should be subsidised by the government or provided at no cost.8

Conclusion

Since the establishment of a public FP programme for cancer patients, there has been an increase in the number of patients seeking FP services. More than 90% of men attempted sperm cryopreservation, whereas 37.5% of women underwent oocyte/embryo cryopreservation and 62.5% of women received GnRH agonists during cancer treatment. With further promotion, changes in funding policies, and a more accessible FP programme, the demand for FP services is expected to increase. Fertility preservation services should be regularly reviewed to assess changes in demand and identify areas for improvement.

Author contributions

Concept or design: JKY Ko, EHY Ng.
Acquisition of data: DTY Chan.
Analysis or interpretation of data: DTY Chan, JKY Ko, EHY Ng.
Drafting of the manuscript: DTY Chan.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

The study protocol was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster, Hong Kong (Ref No.: UW 23-334). The requirement for informed consent was waived by the Board due to the retrospective nature of the research.

References

1. You L, Lv Z, Li C, et al. Worldwide cancer statistics of adolescents and young adults in 2019: a systematic analysis of the Global Burden of Disease Study 2019. ESMO Open 2021;6:100255. Crossref

2. Hong Kong Cancer Registry, Hospital Authority. Overview of Hong Kong Cancer Statistics of 2020. Available from: https://www3.ha.org.hk/cancereg/pdf/overview/Overview%20of%20HK%20Cancer%20Stat%202020.pdf. Accessed 7 May 2023.

3. Bhakta N, Liu Q, Ness KK, et al. The cumulative burden of surviving childhood cancer: an initial report from the St Jude Lifetime Cohort Study (SJLIFE). Lancet 2017;390:2569-82. Crossref

4. ESHRE Guideline Group on Female Fertility Preservation; Anderson RA, Amant F, et al. ESHRE guideline: female fertility preservation. Hum Reprod Open 2020;2020:hoaa052. Crossref

5. Practice Committee of the American Society for Reproductive Medicine. Fertility preservation in patients undergoing gonadotoxic therapy or gonadectomy: a committee opinion. Fertil Steril 2019;112:1022-33. Crossref

6. Canada AL, Schover LR. The psychosocial impact of interrupted childbearing in long-term female cancer survivors. Psychooncology 2012;21:134-43. Crossref

7. Rosen A, Rodriguez-Wallberg KA, Rosenzweig L. Psychosocial distress in young cancer survivors. Semin Oncol Nurs 2009;25:268-77. Crossref

8. Ko JK, Cheung CS, Cheng HH, et al. Knowledge, attitudes and intention on fertility preservation among breast cancer patients. Sci Rep 2023;13:9645. Crossref

9. Yeung SY, Ng EY, Lao TT, Li TC, Chung JP. Fertility preservation in Hong Kong Chinese society: awareness, knowledge and acceptance. BMC Womens Health 2020;20:86. Crossref

10. Wallace WH, Smith AG, Kelsey TW, Edgar AE, Anderson RA. Fertility preservation for girls and young women with cancer: population-based validation of criteria for ovarian tissue cryopreservation. Lancet Oncol 2014;15:1129-36. Crossref

11. Centre of Assisted Reproduction and Embryology, The University of Hong Kong–Queen Mary Hospital. How to make an appointment. Available from: https://hkuivf.hku.hk/en/services/fertility-preservation/how-to-make-an-appointment/. Accessed 27 Nov 2024.

12. Ko JK, Lam KK, Cheng HH, et al. Fertility preservation programme in a tertiary-assisted reproduction unit in Hong Kong. Fertil Reprod 2021;3:94-100. Crossref

13. Centre of Assisted Reproduction and Embryology, The University of Hong Kong–Queen Mary Hospital. Fertility preservation. Available from: https://hkuivf.hku.hk/en/services/fertility-preservation/. Accessed 27 Nov 2024.

14. Centre of Assisted Reproduction and Embryology, The University of Hong Kong–Queen Mary Hospital. YouTube Channel. Available from: https://www.youtube.com/@HKUIVF. Accessed 27 Nov 2024.

15. Hong Kong e-Legislation, Hong Kong SAR Government. Cap 561 Human Reproductive Technology Ordinance. Available from: https://www.elegislation.gov.hk/hk/cap561!en-zh-Hant-HK. Accessed 27 Nov 2024.

16. Letourneau JM, Ebbel EE, Katz PP, et al. Pretreatment fertility counseling and fertility preservation improve quality of life in reproductive age women with cancer. Cancer 2012;118:1710-7. Crossref

17. Oktay K, Harvey BE, Partridge AH, et al. Fertility preservation in patients with cancer: ASCO clinical practice guideline update. J Clin Oncol 2018;36:1994-2001. Crossref

18. Blumenfeld Z. Fertility preservation using GnRH agonists: rationale, possible mechanisms, and explanation of controversy. Clin Med Insights Reprod Health 2019;13:1179558119870163. Crossref

19. Zong X, Yu Y, Yang H, et al. Effects of gonadotropin-releasing hormone analogs on ovarian function against chemotherapy-induced gonadotoxic effects in premenopausal women with breast cancer in China: a randomized clinical trial. JAMA Oncol 2022;8:252-8. Crossref

20. Chen H, Xiao L, Li J, Cui L, Huang W. Adjuvant gonadotropin-releasing hormone analogues for the prevention of chemotherapy-induced premature ovarian failure in premenopausal women. Cochrane Database Syst Rev 2019;3:CD008018. Crossref

21. Lambertini M, Ceppi M, Poggio F, et al. Ovarian suppression using luteinizing hormone-releasing hormone agonists during chemotherapy to preserve ovarian function and fertility of breast cancer patients: a meta-analysis of randomized studies. Ann Oncol 2015;26:2408-19. Crossref

22. Demeestere I, Brice P, Peccatori FA, et al. No evidence for the benefit of gonadotropin-releasing hormone agonist in preserving ovarian function and fertility in lymphoma survivors treated with chemotherapy: final long-term report of a prospective randomized trial. J Clin Oncol 2016;34:2568-74. Crossref

23. Moravek MB, Confino R, Lawson AK, et al. Predictors and outcomes in breast cancer patients who did or did not pursue fertility preservation. Breast Cancer Res Treat 2021;186:429-37. Crossref

24. Arecco L, Blondeaux E, Bruzzone M, et al. Safety of fertility preservation techniques before and after anticancer treatments in young women with breast cancer: a systematic review and meta-analysis. Hum Reprod 2022;37:954-68. Crossref

25. Takae S, Lee JR, Mahajan N, et al. Fertility preservation for child and adolescent cancer patients in Asian countries. Front Endocrinol (Lausanne) 2019;10:655. Crossref

26. Chung JP, Chan DY, Song Y, et al. Implementation of ovarian tissue cryopreservation in Hong Kong. Hong Kong Med J 2023;29:121-31. Crossref

Consolidated and updated ultrasonographic fetal biometry and estimated fetal weight references for the Hong Kong Chinese population

Hong Kong Med J 2024 Dec;30(6):444–51 | Epub 16 Dec 2024

https://doi.org/10.12809/hkmj2310910

ORIGINAL ARTICLE

Consolidated and updated ultrasonographic fetal biometry and estimated fetal weight references for the Hong Kong Chinese population

Fangzi Liu, MB, ChB, MRCOG¹; Jing Lu, MD²; Angel HW Kwan, MB, ChB, FHKAM (Obstetrics and Gynaecology)¹; YK Yeung, MB, BS¹; Lo Wong, MB, BS, FHKAM (Obstetrics and Gynaecology)¹; Christopher PH Chiu, MB, BS, FHKAM (Obstetrics and Gynaecology)¹; Liona CY Poon, MB, BS, MD³; Daljit Singh Sahota, BEng, PhD³

¹ Department of Obstetrics and Gynaecology, Prince of Wales Hospital, Hong Kong SAR, China

² Department of Obstetrics and Gynaecology, The First Affiliated Hospital of Xiamen University, Xiamen, China

³ Department of Obstetrics and Gynaecology, The Chinese University of Hong Kong, Hong Kong SAR, China

Corresponding author: Prof Daljit Singh Sahota (daljit@cuhk.edu.hk)

Full paper in PDF

Abstract

Introduction: This study aimed to construct consolidated and updated ultrasonographic fetal biometry and estimated fetal weight (EFW) references for the Hong Kong Chinese population and evaluate the extent of under- and overdiagnosis of small-for-gestational-age (SGA) and large-for-gestational-age (LGA) using these new references.

Methods: Fetal biometry and EFW references were constructed using the Generalised Additive Model for Location, Scale, and Shape, based on data from 1679 singleton pregnancies in non-smoking Chinese women. Ultrasound scans were performed at 12 to 40 weeks of gestation to measure biparietal diameter, head circumference, abdominal circumference (AC), and femur length, following standardised protocols. The rates of SGA and LGA diagnoses using the existing and updated Hong Kong fetal biometry references were compared in an independent cohort of 10 229 pregnancies.

Results: The median number of scans per gestational week between 20 and 39 weeks was 75 (interquartile range=67-83). Compared with existing references, the new AC reference would significantly (P<0.001) increase the proportions of SGA fetuses with AC measurements at <3rd and <10th percentiles from 1.7% and 6.1% to 3.4% and 10.0%, respectively. Conversely, it would significantly decrease (P<0.001) the proportions of LGA fetuses with AC at >90th and >97th percentiles from 15.0% and 4.9% to 11.5% and 3.5%, respectively.

Conclusion: Adoption of the new references, particularly for AC, may lead to increased identification of SGA cases and decreased identification of LGA cases. The proportions of these cases will be more consistent with their intended diagnostic thresholds. Further studies are needed to determine how these references impact pregnancy outcomes.

New knowledge added by this study

Updated biometry and estimated fetal weight (EFW) references were constructed for antenatal assessment of fetal size.
Improved detection of small-for-gestational-age (SGA) fetuses was achieved.
Reduced identification of fetuses classified as large-for-gestational-age was noted.

Implications for clinical practice or policy

The updated biometry and EFW references were implemented in clinical practice by hospitals managed by the Hospital Authority in the second quarter of 2023.
There is a need for clinicians to prepare for an increase in the number of cases requiring closer monitoring and potentially earlier interventions for SGA fetuses and a need for clear guidelines to manage the increased number of potential SGA pregnancies without overtreatment.

[Abstract in Chinese]

Introduction

Fetal biometry and estimated fetal weight (EFW) are routinely documented by sonographers and ultrasound providers during the antenatal period as early indicators of suspected or actual abnormal fetal growth. At a given gestational age (GA), small or large fetal size is often suspected when biometry measurements are below or above the reference extremes. Small for gestational age (SGA), typically defined as a fetus with an abdominal circumference (AC) or EFW <10th percentile, is associated with increased risks of stillbirth, preterm delivery, and neonatal morbidity and mortality1 2; this diagnosis requires more frequent ultrasound monitoring. In contrast, large for gestational age (LGA) refers to a fetus with AC or EFW >90th percentile and is associated with increased risks of macrosomia, shoulder dystocia, neonatal hypoglycaemia, caesarean delivery, and postpartum haemorrhage.3 4 Management of an LGA fetus may include strict maternal glycaemic control in cases of gestational diabetes, early induction of labour, or scheduled caesarean delivery. Therefore, reliable reference charts for fetal biometry and size are essential in obstetric practice to optimise the use of antenatal surveillance resources, especially in public medical institutions.

The current fetal biometry references adopted by obstetricians and ultrasound providers in Hong Kong were constructed using a cohort of Hong Kong Chinese women from 1999 to 2000, based on best practices available at that time, and were published in 2008.5 However, the clinical utility of these 2008 biometry references for identifying SGA and LGA was not evaluated until 2016 by Cheng et al,6 who found that the percentile thresholds used to classify fetuses as SGA and LGA led to underdiagnosis of SGA and overdiagnosis of LGA. Specifically, only 4.6% of fetuses had an AC <10th percentile, whereas 13.3% had an AC >90th percentile,6 raising concerns about the validity of the measurements in 20085 and whether they still reflect current fetal size, considering changes in population and socio-demographic characteristics.

The aims of the current study were to construct revised ultrasonographic fetal biometry and EFW references for the Hong Kong Chinese population, using statistical methods recommended by the World Health Organization (WHO), and to compare the rates of SGA and LGA diagnoses based on the new and existing references.

Methods

This study utilised fetal biometry data from three population cohort studies previously conducted at Prince of Wales Hospital, The Chinese University of Hong Kong.5 6 7 Fetal biometry data from two of the cohorts5 7 were used to construct the revised biometry and EFW references, while the remaining cohort6 was used to assess the clinical utility of specific percentiles from the updated biometry references. This study followed the TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) reporting guideline.8

Derivation of biometry and estimated fetal weight references

The new fetal biometry references were developed using data collected from non-smoking Chinese women with viable, spontaneously conceived singleton pregnancies, recruited at 11 to 13 weeks of gestation from the general obstetric population in the years 1999-20005 and 2015-2016.7 Women who consented to participate in either cohort were randomly selected to undergo a study-specific ultrasound examination of fetal size by a maternal-fetal medicine specialist at GAs ranging from 12 to 40 weeks. Gestational age at recruitment was calculated based on the first date of the last menstrual period if it corresponded to the crown-rump length measurement within a 4-day margin; otherwise, the GA was adjusted using a crown-rump length formula specific to the Chinese population.9 Pregnancies with fetal anomalies were excluded from both cohorts.

Transabdominal ultrasounds were performed using standard commercially available transducers and machines present in the hospital, as described in the original studies.5 7 Fetal biometric measurements, including head circumference (HC), biparietal diameter (BPD) measured in an outer-inner manner, AC, and femur length (FL) were obtained using identical standardised protocols, as previously described.5 7 Estimated fetal weight was derived from biometric data using the formula EFW=10^{(1.326+0.0107×HC+0.0438×AC+0.158×FL−0.00326×AC×FL)}, as previously published by Hadlock et al10 and adopted by the WHO.11

Biometry reference models for HC, BPD, AC, FL and EFW according to GA were constructed using the Generalised Additive Model for Location, Scale, and Shape (GAMLSS) package (version 5.0) in R statistical software (version 3.3.2). Best-fit models were developed in a stepwise manner, beginning with models based on the normal distribution and considering alternatives such as the Box—Cox power exponential, as appropriate. Gestational age was included as a polynomial term, and all measurements were transformed to their natural logarithm equivalent before model construction. Goodness of fit was assessed by inspecting residuals using quantile—quantile plots and worm plots to determine whether kurtosis adjustments were necessary.12

Biometry models were constructed for 12 to 40 weeks of gestation, whereas EFW models were constructed for 20 to 40 weeks. Final smoothing models were chosen by balancing smoothness of percentiles, goodness of fit, and model simplicity. These final models were used to calculate smoothed values for the 50th, 10th, and 90th percentiles (Z_α= ±1.281), as well as the 3rd and 97th percentiles (Z_α= ±1.881). Percentiles were determined using the expression μ × (1+υσZ_α)^1/υ, where Z_α represents the percentile of interest and μ, υ, and σ are dependent on the time covariate (ie, GA).

Standard errors (SEs) of the 50th percentile were estimated using the expression

, assuming that the SE of the percentile of interest can be expressed as a multiple of the standard deviation (SD).13 14

Clinical utility of the revised biometry references

The expected clinical performance of the revised references was evaluated based on the same cohort of second- and third-trimester fetal ultrasound scans previously used to assess the INTERGROWTH-21st standards.6 This cohort consisted of biometry measurements from 10 229 fetuses, with respective median birthweight and GA at delivery of 3140 g (interquartile range [IQR]=2850-3412) and 275 days (IQR=268-281); of these fetuses, 5419 (53.0%) were male.6 All fetal scans were performed transabdominally by either maternal-fetal medicine specialists or midwives who had passed the American Registry for Diagnostic Medical Sonography certification, using standard commercially available transducers and ultrasound machines.

To compare the relative performances of the revised and existing biometry references, Z-scores were calculated as recommended by Salomon et al.15 Expected median and SD values were determined for each gestational week. Z-scores for each fetal parameter were then calculated using the formula: (observed value − expected median) / expected SD. These fetal parameter Z-scores were used to determine the proportion of biometry measurements in the cohort that were <10th or >90th percentiles and <3rd or >97th percentiles, with ±1.282 and ±1.881 as respective thresholds.

Results

Updated biometry references were constructed from a combined cohort of 1679 pregnancies. The median maternal age at expected date of delivery, as well as weight and height at recruitment, were 32 years (IQR=28-34), 53 kg (IQR=38.5-58.1), and 157 cm (IQR=154-161), respectively. Of the pregnancies, 892 (53.1%) were nulliparous women. Birth details were unavailable for 115 (6.8%) pregnancies, all from the cohort recruited by Leung et al,5 which was used to construct the existing biometry reference. In the 1564 (93.2%) pregnancies with documented birth details, the median birthweight, GA at delivery, and male sex proportion were 3160 g (IQR=2900-3405), 277 days (IQR=270-283), and 830 (53.1%), respectively. The median number of scans per gestational week between 20 and 39 weeks was 75 (IQR=67-83).

The best-fitting GAMLSS for fetal biometry and EFW are reported in online supplementary Tables 1 and 2, respectively. The distribution of residuals from the fitted models approximated that of a normal standard distribution, with means of 0, variances of 1, skewness ranging from 0 to 0.1, and kurtosis ranging from 3.22 to 3.69. The Figure shows the fitted 50th, 3rd/97th, and 10th/90th smoothed percentiles.

Figure. Fetal size references for the Hong Kong Chinese population, showing raw data and fitted 50th, 3rd/97th, and 10th/90th smoothed percentiles versus gestational age for (a) abdominal circumference, (b) head circumference, (c) biparietal diameter (outer to inner), (d) femur length, and (e) natural logarithm of estimated fetal weight

The Table summarises the comparison of the proportions of fetuses whose biometry was assessed for fetal size above and below specific percentiles across the 10 229 pregnancies. The proportions of fetuses identified <3rd and >97th percentiles, as well as <10th and >90th percentiles, by the revised biometry references were approximately 3% and 10%, respectively, except for the FL reference.

Table. Comparison of the proportion of fetal biometry measurements among the 10 229 fetuses above and below specific percentiles for the updated local biometry reference and the existing reference5

The analysis showed that, compared with the existing AC biometry reference,5 the revised AC biometry reference would significantly increase the proportions of fetuses with AC measurements at <3rd and <10th percentiles from 1.7% and 6.1% to 3.4% and 10.0%, respectively (both P<0.001). It would also significantly decrease the proportions of fetuses with AC measurements at >90th and >97th percentiles from 15.0% and 4.9% to 11.5% and 3.5%, respectively (both P<0.001). Compared with the existing biometry references,5 the revised biometry references would identify greater numbers of fetuses with short FL (<3rd percentile P=0.002; <10th percentile P<0.001) and smaller HC (<3rd percentile P=0.23; <10th percentile P=0.003) at the extreme lower percentile limits.

Discussion

Principal findings

In this study, we developed updated biometry and EFW references, then assessed how they compare with existing references created over 20 years ago.5 These new references serve as a guide for local obstetricians and ultrasound providers, both in public institutions and private practice, to assess relative and absolute fetal sizes.

Results in the context of current knowledge

In recent years, both the INTERGROWTH-21st project16 and the WHO11 have published biometry and EFW charts according to GA. The INTERGROWTH-21st reference was proposed as a universal standard, based on the premise that fetuses of well-nourished mothers, irrespective of ethnicity or parental characteristics, grow at similar rates.16 Thus, a single INTERGROWTH-21st standard was recommended for assessing fetal size and growth worldwide. In contrast, the WHO suggested that its references could be customised to accommodate local populations, adjusting diagnostic thresholds for SGA and LGA to reflect population-specific characteristics.11 Local studies assessing the suitability and impact of adopting the INTERGROWTH-21st and WHO charts have indicated that these approaches would lead to substantial misclassification of fetuses as small.6 7 17 Similar concerns about the potential for inaccurate classification have been reported by other research groups that assessed either or both the INTERGROWTH-21st and WHO biometry charts.18 19 20 Customisation of the WHO charts to fit the Hong Kong population would be comparable to developing a locally tailored biometry reference, the approach we have taken in this study.

Implications for clinical practice

The revised references had minimal impact on measurements of bony structures, such as HC, BPD, and FL. However, AC, which reflects fetal subcutaneous fat mass and nutritional status,21 plays a greater role in calculating EFW, particularly in the third trimester.10 The revised references should reduce the misdiagnosis of SGA and LGA, given that they are mainly based on AC and EFW. However, this change might increase the workload for obstetricians because additional scans will be needed to distinguish constitutional smallness from growth restriction.

The revised biometry and newly developed EFW references replaced the existing Leung et al’s biometry references5 previously used for antenatal management in hospitals managed by the Hospital Authority starting from the second quarter of 2023. The major clinical impact of the revised biometry references was expected to be an increase in the proportion of fetuses classified as SGA and a decrease in those classified as LGA, such that the proportions become more consistent with their intended diagnostic thresholds at the 3rd and 10th percentiles. By definition, the smallest 10% of fetuses are regarded as SGA,1 2 and the largest 10% are considered LGA.3 4 Although not all of these fetuses exhibit restricted growth, these classifications carry prognostic importance because they predict risks of perinatal morbidity and mortality, especially for SGA. Furthermore, fetuses classified as LGA are more likely to require induction of labour or caesarean delivery. Fetal biometry and EFW references can serve as screening tools to detect fetuses at both extremes of the growth spectrum. Further evaluation, such as assessments of growth velocity, performance of Doppler studies, and use of biophysical profiles, can help differentiate between those at high risk and those who are constitutionally small or large.1

One key purpose of biometry references is to reduce obstetric complications such as shoulder dystocia, stillbirth, and neonatal morbidity and mortality by improving the identification of SGA and LGA fetuses. Further studies will be needed to determine whether revision of the percentiles, particularly the AC reference, and development of a local EFW reference will show significant correlations with perinatal outcomes. However, such studies will need to be conducted over several years and require support from a funding body, considering the generally low incidence of adverse perinatal outcomes in Hong Kong pregnancies.22 In a review of stillbirth rates from 2000 to 2020, Wong et al23 concluded that although stillbirth rates had declined from approximately 3.3 to 2.9 per 1000 births between the first and second decades, further improvements remained necessary regarding early identification of early fetal growth restriction. This analysis indicated that 16% of all stillbirths were related to fetal growth restriction of unknown cause.23 Whether the revised references, by classifying an increased number of fetuses as SGA, lead to improved early detection of fetal growth restriction requires prospective investigation. One approach could involve using information obtained during first-trimester Down syndrome screening to identify fetuses at increased risk of being considered SGA, followed by either longitudinal or cross-sectional assessments later in pregnancy. Leung et al24 previously reported that low serum levels of pregnancy associated plasma protein-A and smaller fetal crown-rump length at 11 to 13 weeks of gestation were independent predictors of SGA status. More recently, Papastefanou et al25 proposed a model for predicting SGA classification using a combination of maternal factors and the same biomarkers included in preeclampsia screening to identify potential fetuses at risk of SGA status.

Strengths and limitations

The revised biometry and newly developed EFW references were derived from a larger cohort, improving the precision of the estimated percentiles, specifically those used for clinical decision-making. By combining two cohorts with similar inclusion and exclusion criteria and using standardised ultrasound measurement protocols,5 7 the precision of the estimated percentiles has been enhanced. The existing biometry references were based on 706 pregnancies, yielding SEs of 0.05 SD for the 10th and 90th percentiles and 0.06 SD for the 3rd and 97th percentiles. By developing the revised references from 1679 cases, we have improved the precision; the abovementioned SEs are now 0.03 SD and 0.04 SD, respectively. Additionally, consistent with biometry references reported by other groups, we used the semi-parametric GAMLSS method to concurrently model the mean, variance, skew, and kurtosis; conversely, the approach by Leung et al5 utilised a simpler mean±k×SD model and assumed no kurtosis or skewness. The GAMLSS method is recommended by the WHO,11 26 27 which adopted this approach during the development of its biometry and EFW references because the GAMLSS enabled more accurate prediction and smoother curves compared with earlier modelling approaches.26 Finally, we avoided a common limitation, identified in a previous review,28 by not retrospectively using routinely collected fetal measurements to derive biometry references—this could lead to skewed charts and inaccurate percentile limits.

A limitation of the newly revised references is that they are monoethnic because they were derived from pregnancies in Chinese women at a single hospital, which provides medical care to approximately 18% of the territory’s population.29 Hong Kong is a largely homogenous society in which approximately 92% of individuals are Han Chinese.30 However, considering possible ethnic differences, especially when comparing East and Southeast Asians with other groups, caution may be needed when interpreting biometry and EFW measurements in other ethnic populations.31 32

Conclusion

We have constructed and updated ultrasonographic fetal biometry and EFW reference percentiles for the antenatal assessment of fetal size in Hong Kong Chinese singleton pregnancies. The adoption of these updated biometry percentile references, particularly regarding AC, is expected to result in an increased proportion of fetuses classified as SGA and a decreased proportion of fetuses considered LGA. The proportions of SGA and LGA cases will be more consistent with the intended diagnostic thresholds. Further prospective studies are needed to determine whether the introduction of these revised biometry and EFW reference percentiles by the hospitals of the Hospital Authority will lead to improved perinatal outcomes.

Author contributions

Concept or design: F Liu, DS Sahota.
Acquisition of data: F Liu, J Lu, AHW Kwan, L Wong.
Analysis or interpretation of data: F Liu, YK Yeung, CPH Chiu, DS Sahota.
Drafting of the manuscript: F Liu, DS Sahota.
Critical revision of the manuscript for important intellectual content: LC Poon, DS Sahota.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

All authors have disclosed no conflicts of interest.

Acknowledgement

The authors thank the pregnant women in this study, as well as the Fetal Medicine team, midwives, and research assistants at the Prince of Wales Hospital who recruited participants and performed fetal scans in the primary study cohorts used to construct the updated biometry references.

Funding/support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

This is a retrospective analysis of data that were collected as part of approved studies conducted by the Joint Chinese University of Hong Kong–New Territories Cluster Clinical Research Ethics Committee, Hong Kong, for the same use and purpose (Ref Nos.: CRE-9019, CRE-2012.538, and CRE 2014.507). Informed consent was obtained from patients when the data was originally collected.

Supplementary material

References

1. Lees CC, Stampalija T, Baschat A, et al. ISUOG Practice Guidelines: diagnosis and management of small-for-gestational-age fetus and fetal growth restriction. Ultrasound Obstet Gynecol 2020;56:298-312. Crossref

2. ACOG Practice Bulletin No. 227: Fetal Growth Restriction: correction [editorial]. Obstet Gynecol 2021;137:754. Crossref

3. Evidence review for large-for-gestational age baby: intrapartum care for women with existing medical conditions or obstetric complications and their babies. Evidence review Q. NICE Guideline No. 121. National Institute for Health and Care Excellence: London; 2019.

4. Macrosomia: ACOG Practice Bulletin, Number 216. Obstet Gynecol 2020;135:e18-35. Crossref

5. Leung TN, Pang MW, Daljit SS, et al. Fetal biometry in ethnic Chinese: biparietal diameter, head circumference, abdominal circumference and femur length. Ultrasound Obstet Gynecol 2008;31:321-7. Crossref

6. Cheng YK, Leung TY, Lao TT, Chan YM, Sahota DS. Impact of replacing Chinese ethnicity-specific fetal biometry charts with the INTERGROWTH-21(st) standard. BJOG 2016;123 Suppl 3:48-55. Crossref

7. Cheng YK, Lu J, Leung TY, Chan YM, Sahota DS. Prospective assessment of INTERGROWTH-21st and World Health Organization estimated fetal weight reference curves. Ultrasound Obstet Gynecol 2018;51:792-8. Crossref

8. Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD statement. BMC Med 2015;13:1. Crossref

9. Sahota DS, Leung TY, Leung TN, Chan OK, Lau TK. Fetal crown-rump length and estimation of gestational age in an ethnic Chinese population. Ultrasound Obstet Gynecol 2009;33:157-60. Crossref

10. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the use of head, body, and femur measurements—a prospective study. Am J Obstet Gynecol 1985;151:333-7. Crossref

11. Kiserud T, Piaggio G, Carroli G, et al. The World Health Organization fetal growth charts: a multinational longitudinal study of ultrasound biometric measurements and estimated fetal weight. PLoS Med 2017;14:e1002220. Crossref

12. van Buuren S, Fredriks M. Worm plot: a simple diagnostic device for modelling growth reference curves. Stat Med 2001;20:1259-77. Crossref

13. Healy MJ. Notes on the statistics of growth standards. Ann Hum Biol 1974;1:41-6. Crossref

14. Royston P. Constructing time-specific reference ranges. Stat Med 1991;10:675-90. Crossref

15. Salomon LJ, Bernard JP, Duyme M, Buvat I, Ville Y. The impact of choice of reference charts and equations on the assessment of fetal biometry. Ultrasound Obstet Gynecol 2005;25:559-65. Crossref

16. Papageorghiou AT, Ohuma EO, Altman DG, et al. International standards for fetal growth based on serial ultrasound measurements: the Fetal Growth Longitudinal Study of the INTERGROWTH-21st Project. Lancet 2014;384:869-79. Crossref

17. Lok IW, Kong MC, To WW. Updated gestational age specific birthweight reference of Hong Kong Chinese newborns and comparison with local and international growth charts. Open J Obstet Gynecol 2021;11:940-54. Crossref

18. Liu S, Metcalfe A, León JA, et al. Evaluation of the INTERGROWTH-21st project newborn standard for use in Canada. PLoS One 2017;12:e0172910. Crossref

19. Jakubowski D, Salloum D, Maciejewski M, et al. Comparison of application of Fenton, Intergrowth-21st and WHO growth charts in a population of Polish newborns. Clin Exp Obstet Gynecol 2021;48:949-54. Crossref

20. Huang TM, Tsai CH, Hung FY, Huang MC. A novel reference chart and growth standard of fetal biometry in the Taiwanese population. Taiwan J Obstet Gynecol 2022;61:794-9. Crossref

21. Gardeil F, Greene R, Stuart B, Turner MJ. Subcutaneous fat in the fetal abdomen as a predictor of growth restriction. Obstet Gynecol 1999;94:209-12. Crossref

22. Hong Kong College of Obstetricians and Gynaecologists. Territory-Wide Audit in Obstetrics and Gynaecology. 2014. Available from: https://www.hkcog.org.hk/hkcog/Download/Territory-wide_Audit_in_Obstetrics_Gynaecology_2014.pdf. Accessed 1 Apr 2023.

23. Wong ST, Tse WT, Lau SL, Sahota DS, Leung TY. Stillbirth rate in singleton pregnancies: a 20-year retrospective study from a public obstetric unit in Hong Kong. Hong Kong Med J 2022;28:285-93. Crossref

24. Leung TY, Sahota DS, Chan LW, et al. Prediction of birth weight by fetal crown-rump length and maternal serum levels of pregnancy-associated plasma protein-A in the first trimester. Ultrasound Obstet Gynecol 2008;31:10-4. Crossref

25. Papastefanou I, Wright D, Nicolaides KH. Competing-risks model for prediction of small-for-gestational-age neonate from maternal characteristics and medical history. Ultrasound Obstet Gynecol 2020;56:196-205. Crossref

26. Stasinopoulos DM, Rigby RA. Generalized Additive Models for Location, Scale and Shape (GAMLSS) in R. J Stat Softw 2007;23:1-46. Crossref

27. Borghi E, de Onis M, Garza C, et al. Construction of the World Health Organization child growth standards: selection of methods for attained growth curves. Stat Med 2006;30:247-65. Crossref

28. Ioannou C, Talbot K, Ohuma E, et al. Systematic review of methodology used in ultrasound studies aimed at creating charts of fetal size. BJOG 2012;119:1425-39. Crossref

29. Census and Statistics Department, Hong Kong SAR Government. District profiles (population and households). 2023. Available from: https://www.censtatd.gov.hk/en/map_ghs.html. Accessed 1 Apr 2023.

30. Race Relations Unit, Home Affairs Department, Hong Kong SAR Government. The demographics: ethnic groups. Available from: https://www.had.gov.hk/rru/english/info/demographics.htm. Accessed 1 Apr 2023.

31. Shiono PH, Klebanoff MA, Graubard BI, Berendes HW, Rhoads GG. Birth weight among women of different ethnic groups. JAMA 1986;255:48-52. Crossref

32. Kierans WJ, Joseph KS, Luo ZC, Platt R, Wilkins R, Kramer MS. Does one size fit all? The case for ethnicspecific standards of fetal growth. BMC Pregnancy Childbirth 2008;8:1. Crossref

Factors affecting human papillomavirus vaccine acceptance among parents of Primary 4 to 6 boys and girls in Hong Kong

Hong Kong Med J 2024 Oct;30(5):386–99

https://doi.org/10.12809/hkmj2311144

ORIGINAL ARTICLE

Factors affecting human papillomavirus vaccine acceptance among parents of Primary 4 to 6 boys and girls in Hong Kong

Jody KP Chu, MClinPharm¹; CW Sing, PhD¹; Y Li, BPharm¹; Patrick H Wong, BSc²; Eric YT So, MPH²; Ian CK Wong, PhD^1,3

¹ Department of Pharmacology and Pharmacy, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China

² Merck Sharp and Dohme (Asia) Ltd, Hong Kong SAR, China

³ Research Department of Practice and Policy, School of Pharmacy, University College London, London, United Kingdom

Corresponding author: Ms Jody KP Chu (chukpj@hku.hk)

Full paper in PDF

Abstract

Introduction: Human papillomavirus (HPV) poses a substantial but underestimated healthcare burden in Hong Kong. This study investigated factors affecting parental acceptance of HPV vaccination after the introduction of an immunisation programme for primary school girls. We assessed parental perceptions and related factors concerning HPV vaccination for both boys and girls.

Methods: We conducted a cross-sectional survey between December 2021 and February 2022 among parents of Primary 4 to 6 students in Hong Kong. Our self-administered online survey collected data regarding socio-demographic characteristics, awareness and knowledge of HPV vaccination, attitudes towards HPV vaccination, and acceptance of HPV vaccination. Characteristics were compared between boys’ parents and girls’ parents. Factors associated with vaccine acceptance were analysed by multivariate logistic regression.

Results: We observed high awareness of HPV vaccination among boys’ parents and girls’ parents; however, they demonstrated relatively poor knowledge of HPV and the HPV vaccine. An alarming low HPV vaccination uptake rate was also observed. Attitudes towards the HPV vaccine were similar between parent groups. A majority of parents believed that the HPV vaccine was safe and effective in preventing infection. Parents of boys showed lower HPV vaccine acceptance. Factors associated with acceptance differed between parent groups.

Conclusion: High awareness of HPV and HPV vaccine is predictive of vaccine acceptance. Boys’ parents are less likely to accept HPV vaccination and emphasis should be placed on addressing potential HPV vaccine hesitancy in this group. Public education should also aim to raise awareness of government vaccination programme, and implementation of catch-up vaccination programme to school children beyond primary school should be considered.

New knowledge added by this study

Awareness of human papillomavirus (HPV) was similar between parents of boys and parents of girls (P=0.346); 81.4% of boys’ parents and 78.5% of girls’ parents had heard of HPV.
Overall, attitudes towards HPV and the HPV vaccine were similar between parents of boys and parents of girls.
High acceptance of their child receiving the HPV vaccine in both parents of boys and girls was observed; parents of girls were more likely to accept the vaccine, compared with parents of boys (89.7% vs 73.8%; P<0.001).

Implications for clinical practice or policy

The awareness of the HPV vaccination programme among girls’ parents is low, echoing the problem of insufficient information provision concerning HPV vaccination, especially during the coronavirus disease 2019 pandemic.
To prevent future healthcare burdens caused by immunisation gaps, catch-up vaccination services for affected children should be considered and implemented as soon as possible.

[Abstract in Chinese]

Introduction

Human papillomavirus (HPV) is a common sexually transmitted infection that constitutes a substantial global healthcare burden. It is associated with genital warts and various cancers (eg, cervical, penile, anal, oropharyngeal, and head and neck cancers). Human papillomavirus causes 4.5% (630 000) of all new cancer cases worldwide.1

In Hong Kong, cervical cancer is the ninth most common cancer, with a crude incidence of 12.9 per 100 000 women and girls.2 3 There are limited data regarding HPV infection or HPV-associated cancers in men and boys. A local study estimated that the incidence of genital warts in Hong Kong was 203.7 per 100 000 person-years. Men and boys had a higher incidence compared with women and girls (292.2 per 100 000 person-years vs 124.9 per 100 000 person-years, respectively), suggesting a similar or possibly higher prevalence of HPV infection in men and boys.4

Human papillomavirus vaccination is a highly effective preventive measure against HPV infection and its complications. National HPV vaccination programmes targeting adolescent girls have significantly reduced the incidences of HPV-associated diseases.5 The World Health Organization recommends including HPV vaccination in routine programmes for girls aged 9 to 14 years, with possible extension to boys if feasible.6

Universal HPV vaccination programmes, covering both adolescent boys and adolescent girls, have become increasingly common in recent years, particularly in developed countries such as the United States, Canada, Australia, and 20 European nations. Female-only vaccination programmes with high vaccine coverage rates have demonstrated substantial public health impact concerning several HPV-related diseases and cancers.7 Gender-neutral vaccination programmes, targeting both boys and girls, have shown greater resilience8 and faster elimination of cervical cancer9; they also provide direct protection to reduce disease burden in all men and in subpopulations of men (eg, men who have sex with men and men who have sex abroad).10 11 12 The achievement of an 80% vaccination rate in both sexes is expected to enable the elimination of HPV subtypes 6, 11, 16, and 18.13

Despite the benefits of a high vaccination rate, the current rate of HPV vaccine is much lower than desired. In Hong Kong, the rate of vaccine uptake reportedly ranged from 2.2% to 7.2% in adolescent girls and 0.6% in adolescent boys before the HPV vaccine was incorporated into the Hong Kong Childhood Immunisation Programme (HKCIP).14 15 16 The 9-valent HPV vaccine was introduced into the Programme for Primary 5 and 6 girls, with a reported first-dose uptake rate of 85% among Primary 5 girls in 2020.17 However, vaccination rates are expected to remain low among adolescent boys.

Prior to the inclusion of HPV vaccination in the HKCIP, few local studies explored parental decision-making14 15 16 18; those that did primarily focused on girls, with limited examination of factors influencing vaccine acceptance or uptake. One survey did include parents of adolescent boys but was hindered by its small sample size (162 boys’ parents).14 Considering the recent implementation of the HPV vaccination programme for primary school girls and the lack of sufficient data concerning HPV vaccination in Hong Kong, further local research is warranted. An understanding of parental acceptance, particularly for boys, can inform strategies to improve vaccine uptake.

This study aimed to identify factors affecting HPV vaccination acceptance among Hong Kong parents of Primary 4 to 6 students. It compared the knowledge, attitudes, and acceptance between girls’ parents and boys’ parents, then explored the underlying reasons for their vaccination decisions.

Methods

Study design

This cross-sectional survey was conducted from December 2021 to February 2022. Invitation letters were sent to 554 primary schools in Hong Kong, including public local schools, private local schools, and Direct Subsidy Scheme schools. In total, 65 schools agreed to participate. After consent had been obtained from participating schools, parents of Primary 4 to 6 students at those schools received a self-administered online survey through the Qualtrics platform.

Measures

The survey was divided into four sections, namely, (1) socio-demographic characteristics, (2) awareness and knowledge regarding HPV and HPV vaccination, (3) attitudes towards HPV vaccination, and (4) acceptance of HPV vaccination. The questionnaires were available in both English and Chinese (online supplementary Appendices 1 and 2, respectively). A detailed description of the survey sections is provided in online supplementary Appendix 3. Upon completion of the online survey, the results were stored in the Qualtrics platform for further analysis.

Data analysis

Human papillomavirus vaccination attitudes were measured using a five-point Likert scale. Two statements (Questions 41 and 42) with strong internal consistency (Cronbach’s alpha=0.81) were combined to form the variable ‘Worried about HPV infection’, and the mean score of the two statements was used for analysis. Similarly, two other statements (Questions 46 and 47) with strong internal consistency (Cronbach’s alpha=0.91) were merged into the variable ‘Worried that HPV vaccine might negatively impact child’s sexual activity’.

Descriptive statistics were used to characterise the participants and study variables. The first analysis compared the knowledge, attitudes, and acceptance of the HPV vaccine between boys’ parents and girls’ parents. Significant differences between groups were identified using the Chi squared test for nominal variables, the t test for continuous variables, and the Mann-Whitney U test for ordinal variables (age-group, household income, and education level). In the second analysis, we investigated factors associated with the acceptance of HPV vaccination for children. Participants whose children had already received HPV vaccination were excluded from the analysis due to missing values in some variables (Questions 49 to 51). Univariate logistic regression was used to estimate the crude odds ratio (OR) and 95% confidence interval (CI). The study variables were included as independent predictors; the acceptance of HPV vaccination for children was regarded as a binary dependent variable (‘Yes’ or ‘No’). Variables with P values <0.1 were entered into multivariate logistic regression. P values <0.05 were considered statistically significant.

Because free HPV vaccination was only provided for primary school girls in Hong Kong, we assumed that factors affecting the acceptance of HPV vaccination for their child varied between boys’ parents and girls’ parents. Consequently, the second analysis was conducted separately for each parent group.

All analyses were conducted using R software (version 4.1.1).

Results

In total, 844 participants completed the survey. Of these, 43.8% were parents of boys and 56.2% were parents of girls. The socio-demographic characteristics of the parents are presented in Table 1.

Table 1. Socio-demographic characteristics of parents of boys and girls (n=844)

Vaccine uptake rate

The HPV vaccine uptake rate is low with boys’ parents reported 6.8% and girls’ parents reported 4.9%. Among children who have received HPV vaccine, >90% of parents in both groups reported that their children received the HPV vaccine through the HKCIP (Table 1).

Awareness and knowledge of human papillomavirus and the vaccine among parents

Awareness of HPV was similar between boys’ parents and girls’ parents (81.4% vs 78.5%; P=0.346). Knowledge scores regarding HPV and the HPV vaccine were low in both parent groups; parents of boys had higher mean scores compared with parents of girls (6.48 vs 6.03; P=0.012). More boys’ parents discussed sexually transmitted disease (STDs) with their children, relative to girls’ parents (33.0% vs 15.2%; P<0.001) [Table 2].

Table 2. Knowledge, attitudes, and acceptance of the human papillomavirus vaccine among parents of boys and girls

Attitudes towards the vaccine in parents

Two questions addressed the timing of vaccination, namely: ‘At what age should a child receive the HPV vaccine?’, and the yes/no statement ‘I believe it’s better for my child to receive the HPV vaccine before they become sexually active’. A majority of parents, both of boys (83.6%) and of girls (85.9%), believed that their children should receive the HPV vaccine at age ≥13 years. Additionally, more parents of boys (55.7%) believed that their children should receive the HPV vaccine before becoming sexually active; more parents of girls (51.1%) reported a neutral perspective on this statement. Regarding HPV infection and HPV vaccine effectiveness, parents in both groups were worried about HPV infection (mean±standard deviation [SD] out of 5: 3.56±0.74 in boys’ parents; 3.48±0.76 in girls’ parents). Over 70% in parents of both groups believe that their children cannot be protected from HPV without HPV vaccination, furthermore a majority of parents in both groups also believe in the vaccine’s effectiveness (90.2% in boys’ parents and 84.6% in girls’ parents) [Table 2].

Concerning vaccine safety, impacts, and cost, most parents of boys (90.8%) and parents of girls (84.8%) agreed that the HPV vaccine is safe. They had a neutral perspective or were less worried about the vaccine’s short-term (62.5% and 67.6%, respectively) and long-term side-effects (80.5% and 82.3%, respectively) [Table 2].

Additionally, parents had a neutral perspective or were less worried about the vaccine’s negative impacts or influence on adolescent development. However, most parents agreed that the HPV vaccine is too expensive (88.1% and 79.8%, respectively) [Table 2].

Vaccine acceptance in parents of boys and girls

We observed high acceptance of the HPV vaccine for their children in boys’ parents (73.8%) and girls’ parents (89.7%). If the HPV vaccine were subsidised under the HKCIP, acceptance in parents would slightly increase because the government would cover the cost (78.9% and 92.6%, respectively) [Table 2].

The reasons for accepting the HPV vaccine for their children were similar between parent groups (Fig), with a majority citing concerns about HPV infection (91.0% in boys’ parents and 78.6% in girls’ parents). Acceptance was least influenced by religions and culture (<3%) or advertisements (<5%) in parents of both sexes. The reasons for declining the HPV vaccine for their children were somewhat different between boys’ parents and girls’ parents. ‘The HPV vaccine is too expensive’ was the top reason chosen by both boys’ parents (46.4%) and girls’ parents (42.9%). The other two reasons most often selected by boys’ parents were ‘Not enough information about the HPV vaccine provided to me’ (32.0%) and ‘My child doesn’t like vaccinations’ (22.7%). For girls’ parents, the other two reasons were ‘My child doesn’t like vaccinations’ (38.8%) and ‘The HPV vaccine can cause adverse effects/is not safe’ (36.2%) [online supplementary Fig].

Figure. Reasons for allowing their child to receive the human papillomavirus vaccine among parents of boys (n=273) and parents of girls (n=425)

Factors associated with vaccine acceptance for children

The association analysis excluded 25 parents of boys and 23 parents of girls whose children had already received the HPV vaccine. The acceptance rates of the HPV vaccine for children of boys’ parents and girls’ parents, stratified according to the study variables, are listed in online supplementary Tables 1 and 2, respectively.

Regarding boys’ parents, 24 study variables with P values <0.1 in univariate logistic regression were entered into multivariate logistic regression (Table 3). Factors associated with higher acceptance of the HPV vaccine for children included parental receipt of the HPV vaccine (OR=9.36, 95% CI=1.5-63.82; P=0.018), knowledge of the HPV vaccine (OR=10.16, 95% CI=3.02-39.07; P<0.001), and stronger beliefs that ‘it’s better for my child to receive the HPV vaccine before they become sexually active’ (OR=3.27, 95% CI=1.66-7.09; P=0.001) and ‘I am worried that the HPV vaccine might affect adolescent development’ (OR=2.56, 95% CI=1.39-5.03; P=0.004). Conversely, factors associated with lower acceptance of the HPV vaccine were the presence (in the respondents’ families) of more children in Primary 4 to Primary 6 (OR=0.28, 95% CI=0.12-0.63; P=0.002), a history of discussing STD prevention with their children (OR=0.23, 95% CI=0.08-0.64; P=0.005), receipt of regular seasonal influenza vaccines (OR=0.15, 95% CI=0.04-0.48; P=0.002), child’s receipt of regular seasonal influenza vaccines (OR=0.25, 95% CI=0.08-0.80; P=0.021), and stronger beliefs that ‘my child can be protected from HPV without HPV vaccination’ (OR=0.28, 95% CI=0.11-0.66; P=0.005) [Table 3].

Table 3. Associations of variables and acceptance of the human papillomavirus vaccine for children among parents of boys

Regarding girls’ parents, 19 study variables were entered into multivariate logistic regression. Higher acceptance of the HPV vaccine for their children was associated with higher monthly household income (OR=4.3, 95% CI=1.95-10.47; P=0.001) and the combined variable ‘worried about HPV infection’ (OR=2.39, 95% CI=1.08-5.73; P=0.038). Older age-group (OR=0.38, 95% CI=0.17-0.82; P=0.018) was the only variable associated with lower acceptance of the HPV vaccine (Table 4).

Table 4. Associations of variables and acceptance of the human papillomavirus vaccine for children among parents of girls

Discussion

This survey of 844 Hong Kong parents (370 boys’ parents and 474 girls’ parents) revealed high HPV vaccine awareness but relatively low knowledge of HPV and the HPV vaccine. Parents believed the vaccine was safe and effective in preventing HPV infection. Acceptance of the HPV vaccine was lower among boys’ parents than among girls’ parents, and factors associated with acceptance differed between the two parent groups. Differences in socio-demographic characteristics were observed, such that more boys’ parents discussed STDs with their children and had experience with regular seasonal influenza vaccines, the HPV vaccine, and Pap smears.

Understanding of human papillomavirus and the vaccine

Although a majority of parents of both sexes had knowledge of the HPV vaccine, their average scores indicated a low overall understanding of HPV and HPV vaccination. This finding is consistent with the results of previous studies, which showed that general knowledge and awareness of HPV among parents in Hong Kong remain low despite some improvement over time.14 15 16 18 19 20 Considering the substantial healthcare burden associated with HPV-related diseases in Hong Kong, there is an urgent need for educational or promotional programmes to enhance vaccine acceptance and uptake.

In our study, parents expressed concern about HPV infection and strongly favoured HPV vaccination for their children before the children became sexually active. These beliefs support educational and promotional campaigns targeting the early adolescent age-group.

The reported HPV vaccine uptake rate is low in both groups (6.8% in boys and 4.9% in girls). The low vaccine uptake rate reported in girls is particularly alarming considering the recent inclusion of the HPV vaccine in the HKCIP and the high vaccination rate of 85% reported in the 2019/2020 school year.17 Among those parents who reported their children of receiving the HPV vaccine, >90% of them, including boys’ parents, indicated that their children received the vaccine through the HKCIP. This finding provides evidence suggesting insufficient public health campaigns, resulting in a lack of knowledge among parents on the HPV vaccination programme and the HKCIP, subsequently leading to potential confusion among parents.

Notably, girls’ parents in our study reported a belief that the HPV vaccine is too expensive, despite the availability of free HPV vaccination through the HKCIP. This finding again reinforces a potential lack of awareness regarding the Programme, possibly due to inadequate dissemination of information during the coronavirus disease 2019 pandemic. Similar trends have been observed in other Western countries, where routine vaccinations (including HPV vaccination) were disrupted by the pandemic.21 22 Catch-up vaccination services for affected children should be implemented promptly to prevent future healthcare burdens.23 24 25 26

Concern for cost and vaccine safety

This study examined the factors influencing parental acceptance of HPV vaccination for boys and girls. Parents who had more children in Primary 4 to 6 were less likely to accept the vaccine, possibly due to cost concerns. Discussions with children about STD prevention and previous receipt of seasonal flu vaccines did not lead to higher acceptance rates. These findings imply that vaccination is not a common topic in STD prevention campaigns, a point that warrants attention in future educational efforts focused on STD prevention. Intriguingly, parents with greater concern that the HPV vaccine affects adolescent development were more likely to accept it; they also had higher knowledge and awareness of HPV (online supplementary Table 3). This result highlights the need to increase parental understanding of HPV and the HPV vaccine, including efforts to clarify potential misconceptions and mitigate safety concerns.

Our data indicate that parental concerns about HPV infection strongly influence vaccine acceptance, whereas concerns about genital warts and HPV-related cancers are less impactful. This discrepancy may be attributed to an optimistic bias, where parents associate HPV complications with promiscuity and believe that their children have low STD risk.18

Notably, parents ranked HPV vaccine recommendations from healthcare professionals, relatives and friends, and schools as more important reasons to accept the vaccine, compared with recommendations by health authorities. This result may suggest that government initiatives provide suboptimal education concerning HPV and the HPV vaccine.

Barriers to HPV vaccine acceptance include costs and children’s preferences, which may explain the discrepancies between uptake and acceptance. Cost is a well-established barrier to vaccination uptake. However, we note that the vaccine is free for girls in our study population, which highlights the importance of awareness. Health messages to boys’ parents should emphasise the value of HPV vaccination as a long-term investment in their sons’ health.14 Concerns about vaccine safety and adverse effects, as well as a lack of recommendations from healthcare professionals or a lack of general knowledge, may also hinder vaccine acceptance.

We found that parental knowledge of HPV and the HPV vaccine significantly influenced decision-making in boys’ parents, indicating that educational campaigns targeting HPV acceptance may be more effective for these parents than for girls’ parents. This difference might be partly related to the feminisation of HPV, especially in Hong Kong. This phenomenon has been observed in a regional qualitative study focusing on men’s perceptions of HPV and HPV vaccination.27 Because the Chinese translation of the HPV vaccine is ‘cervical cancer vaccine’, many boys and men in Hong Kong perceive a low risk of HPV infection.27 28 29 In this context, campaigns or strategies using a fear-based approach to increase the perceived risk of HPV infection may be more effective for boys’ parents.

Limitations

This study had several limitations. First, it was a cross-sectional study and thus provided less robust evidence than would be obtained in a longitudinal study. Vaccine acceptance is merely an indicator of potential uptake, and it is unclear whether this acceptance will be translated into action. Second, this study relied on parents to self-report their outcomes, and it lacked the ability to verify information provided by participants. Third, the results may have been influenced by volunteer bias or other selection biases. Because the survey was self-administered, random sampling of the general study population could not be achieved due to intrinsic differences between those who did and did not choose to participate. Volunteer bias may explain the variation in baseline characteristics between boys’ parents and girls’ parents. This bias limits the generalisability of the study results to the broader population. Fourth, the use of previously validated scales or items was limited. Previous studies were used as a reference to construct the survey questionnaire, but questions were not directly adapted. Although such validated measures exist, due to the lack of research regarding HPV and HPV vaccination, no measures have been validated in Hong Kong.30 31

One possible future research direction involves conducting longitudinal studies to examine the factors affecting vaccine uptake. These studies can produce stronger evidence and more effectively inform strategies for improved vaccine uptake. Furthermore, because this study only screened for variables involved in parental decision-making, a more thorough investigation could be done to better understand this process. Qualitative studies (eg, involving focus groups or interviews) can provide a more in-depth understanding of parents’ attitudes, perceptions, and decision-making processes regarding HPV vaccination acceptance.

Conclusion

This study represents the most extensive local investigation into factors affecting parental acceptance of HPV vaccination in Hong Kong after the implementation of a school-based outreach programme. We found that high awareness of HPV and the HPV vaccine is predictive of vaccine acceptance. To increase vaccination rates among adolescents, we recommend targeted interventions based on the identified factors, including public education for parents and children to raise awareness of HPV risks, the benefits of vaccination for boys, and STD prevention. We also suggest including HPV vaccination for boys in the HKCIP and implementing catch-up vaccination for affected children. Extension of the catch-up programme to school children beyond Primary 6 should be considered to maintain high vaccination rates.

Author contributions

All authors (except for PH Wong and EYT So) contributed to the concept or design of the study, acquisition of the data, analysis or interpretation of the data, drafting of the manuscript, and critical revision of the manuscript for important intellectual content. PH Wong and EYT So contributed to the concept and design of the study questionnaire. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Conflicts of interest

PH Wong and EYT So are employees of Merck Sharp and Dohme (Asia) Ltd. Other authors have disclosed no conflicts of interest.

Acknowledgement

The authors thank Dr Ka-yu Tse from the Division of Gynaecology Oncology of the Department of Obstetrics and Gynaecology of The University of Hong Kong for review of survey questions.

Funding/support

This research was sponsored by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co, Inc (Rahway [NJ], United States) [Ref No.: NIS009837]. The sponsor had no role in collection, analysis, or interpretation of the data, nor did it participate in manuscript preparation.

Ethics approval

This study was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster, Hong Kong (Ref No.: UW21-574). Participants provided informed consent via the online survey platform before survey completion.

Supplementary material

References

1. de Martel C, Plummer M, Vignat J, Franceschi S. Worldwide burden of cancer attributable to HPV by site, country and HPV type. Int J Cancer 2017;141:664-70. Crossref

2. Hong Kong Cancer Registry, Hospital Authority, Hong Kong SAR Government. Cervical Cancer in 2017. October 2019. Available from: https://www3.ha.org.hk/cancereg/pdf/factsheet/2017/cx_2017.pdf. Accessed 10 Feb 2022.

3. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Cervical Cancer. 2024 January 12. Available from: https://www.chp.gov.hk/en/healthtopics/content/25/56.html. Accessed 9 Sep 2024.

4. Lin C, Lau JT, Ho KM, Lau MC, Tsui HY, Lo KK. Incidence of genital warts among the Hong Kong general adult population. BMC Infect Dis 2010;10,272. Crossref

5. Garland SM, Kjaer SK, Muñoz N, et al. Impact and effectiveness of the quadrivalent human papillomavirus vaccine: a systematic review of 10 years of real-world experience. Clin Infect Dis 2016;63:519-27. Crossref

6. World Health Organization. Human papillomavirus vaccines: WHO position paper, May 2017-Recommendations. Vaccine 2017;35:5753-5. Crossref

7. Markowitz LE, Hariri S, Lin C, et al. Reduction in human papillomavirus (HPV) prevalence among young women following HPV vaccine introduction in the United States, National Health and Nutrition Examination Surveys, 2003-2010. J Infect Dis 2013;208:385-93. Crossref

8. Elfström KM, Lazzarato F, Franceschi S, Dillner J, Baussano I. Human papillomavirus vaccination of boys and extended catch-up vaccination: effects on the resilience of programs. J Infect Dis 2016;213:199-205. Crossref

9. Lehtinen M, Gray P, Louvanto K, Vänskä S. In 30 years, gender-neutral vaccination eradicates oncogenic human papillomavirus (HPV) types while screening eliminates HPV-associated cancers. Expert Rev Vaccines 2022;21:735-8.Crossref

10. Division of Cancer Epidemiology & Genetics, National Cancer Institute. HPV vaccine may provide men with herd immunity against oral HPV infections. 2019 October 10. Available from: https://dceg.cancer.gov/news-events/news/2019/hpv-vaccine-herd-immunity. Accessed 10 Feb 2022.

11. Kahn JA, Brown DR, Ding L, et al. Vaccine-type human papillomavirus and evidence of herd protection after vaccine introduction. Pediatrics 2012;130:e249-56. Crossref

12. Merriel SW, Nadarzynski T, Kesten JM, Flannagan C, Prue G. ‘Jabs for the boys’: time to deliver on HPV vaccination recommendations. Br J Gen Pract 2018;68:406-7. Crossref

13. Brisson M, Bénard É, Drolet M, et al. Population-level impact, herd immunity, and elimination after human papillomavirus vaccination: a systematic review and meta-analysis of predictions from transmission-dynamic models. Lancet Public Health 2016;1:e8-17. Crossref

14. Wang Z, Wang J, Fang Y, et al. Parental acceptability of HPV vaccination for boys and girls aged 9-13 years in China—a population-based study. Vaccine 2018;36:2657-65. Crossref

15. Li SL, Lau YL, Lam TH, Yip PS, Fan SY, Ip P. HPV vaccination in Hong Kong: uptake and reasons for non-vaccination amongst Chinese adolescent girls. Vaccine 2013;31:5785-8. Crossref

16. Choi HC, Leung GM, Woo PP, Jit M, Wu JT. Acceptability and uptake of female adolescent HPV vaccination in Hong Kong: a survey of mothers and adolescents. Vaccine 2013;32:78-84. Crossref

17. Hong Kong SAR Government. LCQ10: human papillomavirus vaccination programme. 2021 January 20. Available from: https://www.info.gov.hk/gia/general/202101/20/P2021012000507.htm. Accessed 13 Feb 2022.

18. Wang LD, Lam WW, Fielding R. Determinants of human papillomavirus vaccination uptake among adolescent girls: a theory-based longitudinal study among Hong Kong Chinese parents. Prev Med 2017;102:24-30. Crossref

19. Jones CL, Jensen JD, Scherr CL, Brown NR, Christy K, Weaver J. The Health Belief Model as an explanatory framework in communication research: exploring parallel, serial, and moderated mediation. Health Comm 2015;30:566-76. Crossref

20. Chen JM, Leung DY. Factors associated with human papillomavirus vaccination among Chinese female university students in Hong Kong. Am Int J Soc Sci 2014;3:56-62.

21. Silva TM, Nogueira de Sá AC, Beinner MA, et al. Impact of the COVID-19 pandemic on human papillomavirus vaccination in Brazil. Int J Public Health 2022;67:1604224. Crossref

22. Damgacioglu H, Sonawane K, Chhatwal J, et al. Long-term impact of HPV vaccination and COVID-19 pandemic on oropharyngeal cancer incidence and burden among men in the USA: a modeling study. Lancet Reg Health Am 2022;8:100143. Crossref

23. Shet A, Carr K, Danovaro-Holliday MC, et al. Impact of the SARS-CoV-2 pandemic on routine immunisation services: evidence of disruption and recovery from 170 countries and territories. Lancet Glob Health 2022;10:e186-94. Crossref

24. Ryan G, Gilbert PA, Ashida S, Charlton ME, Scherer A, Askelson NM. Challenges to adolescent HPV vaccination and implementation of evidence-based interventions to promote vaccine uptake during the COVID-19 pandemic: “HPV is probably not at the top of our list”. Prev Chronic Dis 2022;19:E15. Crossref

25. Ogilvie GS, Remple VP, Marra F, et al. Intention of parents to have male children vaccinated with the human papillomavirus vaccine. Sex Transm Infect 2008;84:318-23. Crossref

26. Karafillakis E, Simas C, Jarrett C, et al. HPV vaccination in a context of public mistrust and uncertainty: a systematic literature review of determinants of HPV vaccine hesitancy in Europe. Hum Vaccin Immunother 2019;15:1615-27. Crossref

27. Siu JY, Fung TK, Leung LH. Barriers to receiving HPV vaccination among men in a Chinese community: a qualitative study in Hong Kong. Am J Mens Health 2019;13:1557988319831912. Crossref

28. Holman DM, Benard V, Roland KB, Watson M, Liddon N, Stokley S. Barriers to human papillomavirus vaccination among US adolescents: a systematic review of the literature. JAMA Pediatr 2014;168:76-82. Crossref

29. Trim K, Nagji N, Elit L, Roy K. Parental knowledge, attitudes, and behaviours towards human papillomavirus vaccination for their children: a systematic review from 2001 to 2011. Obstet Gynecol Int 2012;2012:921236. Crossref

30. Waller J, Ostini R, Marlow LA, McCaffery K, Zimet G. Validation of a measure of knowledge about human papillomavirus (HPV) using item response theory and classical test theory. Prev Med 2013;56:35-40. Crossref

31. Perez S, Tatar O, Ostini R, et al. Extending and validating a human papillomavirus (HPV) knowledge measure in a national sample of Canadian parents of boys. Prev Med 2016;91:43-9. Crossref

Pages