Advancements in diagnostic and treatment strategies have led to substantial improvements in treatment prognoses for children with cancer. The 5-year survival rate of childhood cancers has increased dramatically in high-resource settings, from <50% in the 1970s to >80% in the past decade.1 Consequently, there has been a global surge in the population of survivors of childhood cancer, especially in developed regions such as Hong Kong. According to the Hospital Authority Cancer Registry, from 2001 to 2017 in Hong Kong, approximately 180 paediatric patients <19 years of age were diagnosed with cancer each year.2 It is unquestionably necessary to further improve survival rates, and recent efforts and resources have been dedicated to improving the quality of life and health outcomes of survivors of childhood cancer in Hong Kong.3 4 5

Cancer survivors are susceptible to developing a spectrum of late effects because of their previous treatment exposures.6 7 Studies have shown that histories of specific treatment exposures, coupled with continued engagement in health-damaging behaviours during survivorship, may accelerate or exacerbate the development of late effects.8 9 The Children’s Oncology Group (COG)10 provide details of common health-damaging behaviours and their potential impacts on various treatment-related chronic conditions.

Adult and paediatric oncology research has suggested that cancer survivors and their families are often highly receptive to education regarding optimal lifestyles during the early survivorship period.11 Thus, the COG and other international oncology groups have specified that an ideal cancer survivorship programme should comprise recommended screening/surveillance protocols to detect recurrence and late effects, health promotion activities, specialty referrals, and psychosocial interventions.7 12 One systematic review reported that a comprehensive cancer survivorship care programme is associated with positive behavioural change and better health outcomes in survivors.3

In Hong Kong, a recent study by Chan et al13 showed that, although the rates of smoking and alcohol consumption were low among local survivors of childhood cancer, survivors were less likely than their healthy siblings to participate in cancer screening. However, the study did not examine frequencies of engagement in other health-protective behaviours, such as participation in physical activity, undergoing immunisation, using sunscreen, and consuming a balanced diet. Furthermore, survivors’ expectations of a comprehensive survivorship programme have not been investigated. The identification of predictors of poor health-behaviour practices and elucidation of survivors’ needs will presumably assist clinicians in developing targeted interventions to address the needs of this special population.

The primary aim of this study was to identify factors associated with engagement in health-protective and health-damaging behaviours among local survivors of childhood cancer. The secondary aim was to examine cancer survivors’ expectations of a comprehensive survivorship programme in Hong Kong.

This prospective, observational study was conducted at the paediatric oncology/haematology long-term follow-up clinic of the Prince of Wales Hospital, Hong Kong. Eligible participants were recruited through convenience sampling. Between June 2019 and March 2020, the study investigators obtained the list of patients who were scheduled to attend follow-up consultations at the long-term follow-up clinic; this clinic was typically held once per week. Patients were then screened for eligibility using the in-house electronic patient record system (Clinical Management System). All eligible patients who subsequently attended the long-term follow-up clinic were invited to participate in the study.

The inclusion criteria were as follows: diagnosis with primary cancer before 18 years of age; treatment in any medical institutions in Hong Kong; survival for at least 2 years since the completion of cancer treatment or 5 years since diagnosis; and ability to communicate in Cantonese. A parent was recruited if the survivor was aged ≤16 years, or if the survivor was cognitively impaired. Patients were excluded if they were diagnosed with non-cancer conditions (eg, aplastic anaemia, thalassemia), did not understand Cantonese, were still on active treatment, or had incomplete treatment data.

Clinical data regarding cancer diagnosis, treatment history, commodities, and relapse status were retrieved from survivors’ electronic health records. A 20-minute structured questionnaire was interviewer-administered. Participants self-reported their socioeconomic information (ie, highest education attainment, medical insurance, and monthly family income).

Health behaviours were measured using a version of the 2013 National Youth Risk Behaviour Survey14 that had been modified and translated into Traditional Chinese. To adapt the survey for use within the study population, questions pertaining to the healthy behaviour practices of young adult cancer survivors were added. These additional questions were developed based on the health behaviours and practices most frequently reported in studies of survivors of childhood cancer in other countries.9 15 16 Health-protective behaviours refer to engagement in physical activity, balanced diet, sun protection, and immunisation programmes. Health-damaging practices refer to alcohol consumption, smoking, and exposure to second-hand smoke. Alcohol consumption and smoking practices were evaluated in adult survivors only, as the legal age for purchasing tobacco and alcoholic products is 18 years in Hong Kong.

Participants were asked to rate the perceived values of recommended components of a comprehensive survivorship programme12 in the categories of health, psychosocial, parenting, and financial issues. Ratings were conducted using a 5-point Likert scale (1=least important, 5=most important). Participants were also asked to report their preferred time of enrolment into a survivorship programme and modes of services.

The current analysis is part of a broader study17 that aimed to evaluate the effect of an educational intervention on improving awareness of personal health risks among survivors (primary outcome), as well as general health literacy and health behaviours among survivors (auxiliary outcomes). The tailored educational intervention included a review of the survivor’s cancer treatment summary and teaching materials that contained simplified health promotion messages derived from the COG Health Links.10 Sample size was determined based on the primary outcome (awareness of personal health risks). A similar study by Landier et al18 showed that the proportion of survivors of childhood cancer who adequately understood their health risks (defined as awareness of >75% of treatment-related late effects for which they were at risk) was approximately 55% after two sessions of the tailored intervention (ie, θ=0.55). At α=0.05, the required sample size for achieving 80% power to detect a difference in proportion (target θ₀=0.45) between pre- and post-intervention assessments was 195. The current analysis reported the health behavioural practices of participants who provided baseline, pre-intervention assessments.

The SAS University Edition (version 2015; SAS Institute Inc, Cary [NC], US) software was used for all statistical analyses. Descriptive statistics were used to summarise participants’ demographics, clinical characteristics, and frequencies of health behaviours. Multivariable logistic regression analysis was conducted to identify factors that were associated with health behaviours. Associations were presented using odds ratios (ORs) and 95% confidence intervals (95% CIs). Based on a literature review,9 15 19 20 the hypothesised predictors comprised demographics and clinical characteristics (sex, age, cancer diagnosis, and time since diagnosis), as well as socioeconomic status (medical insurance status, monthly household income, and highest education attainment [in adult survivors only]). Finally, descriptive statistics were used to summarise participants’ preferences for the components of a comprehensive survivorship programme.

In total, 252 survivors were screened for eligibility; 39 were excluded for <2 years since treatment or <5 years since diagnosis, the presence of non-cancer diagnoses (eg, benign ovarian teratoma), inability to understand Chinese, or treatment performed outside of Hong Kong. Subsequently, 213 eligible participants were approached. Eight survivors declined to participate, while the remaining 205 eligible survivors provided informed consent and completed the study. Five participants were subsequently excluded because of incomplete treatment records. Finally, data from 200 participants were analysed (response rate 93.9%) [Fig 1].

The mean (± standard deviation) ages at interview were 26.9 ± 6.4 years and 11.1 ± 3.6 years among adult and paediatric survivors, respectively (Table 1). The mean age at cancer diagnosis was 7.3 ± 5.2 years. The mean time since treatment completion was 13.4 ± 7.6 years; 41.0% (n=82) of survivors were within 10 years after treatment. The most common diagnoses were leukaemia (n=78, 39.0%), lymphoma (n=28, 14.0%), and bone tumour (n=18, 9.0%). In total, 185 survivors (92.5%) had undergone chemotherapy, 130 survivors (65.0%) had received radiation, and 85 survivors (42.5%) had undergone surgery. Only 30 survivors (15.0%) had received hematopoietic stem cell transplantation.

All paediatric survivors were students. Among adult survivors, 107 (69.0%) had completed post-secondary education. Only 32 survivors (16.0%) reported a monthly household income of less than HKD$15 000, and 86 survivors (43.0%) had private medical insurance.

Interviews for paediatric survivors (n=45) were completed by parents (Table 1). The mean age of parents was 43.4 ± 7.7 years, and 43 (95.6%) parents had completed secondary school or higher education.

The health-protective and health-damaging behaviours of survivors are summarised in Table 2. The least frequently practised health-protective health-protective behaviour was physical activity. Only 16 survivors (8%) met the World Health Organisation recommendation of engagement in 20 minutes of aerobic physical activity for ≥4 days per week. Of the survivors, 104 (52.0%) reported that they exercised rarely (≤1 day per week) and 135 (67.5%) applied sunscreen rarely. Fewer than 40% of survivors “always” and “frequently” had a balanced diet. In terms of immunisation practice, 49 (24.5%) survivors reported that they did not undergo any immunisation and 37 (18.5%) were unsure about their immunisation history.

Among 155 adult survivors, 110 (71.0%) were never-drinkers, whereas 45 (29.0%) identified themselves as social drinkers. These social drinkers met the “moderate” and “low-risk” drinker definitions established by the National Institute on Alcohol Abuse and Alcoholism.21 Of the 155 adult survivors, 144 (92.9%) were never-smokers and three (1.9%) were ever-smokers. Only eight survivors (5.2%) were current smokers; they smoked a median of 7.5 cigarettes per day (interquartile range=5-10). Of the survivors, 85 (42.5%) were exposed to second-hand smoke; 41 (20.5%) from family members and 29 (14.5%) from colleagues.

Compared with survivors of central nervous system (CNS) tumours, survivors who had been diagnosed with haematological malignancies were more likely to adopt a balanced diet (OR=2.45; 95% CI=1.29-4.68). Younger age at interview was also a significant predictor of adoption of a balanced diet (OR=0.95; 95% CI=0.91-0.99) [Table 3].

Female survivors had more than fivefold greater odds of regular sunscreen use, compared with male survivors (OR=5.66; 95% CI=2.40-13.34). Lower education level in adult survivors (OR=0.20; 95% CI=0.05-0.83) and lower monthly household income (OR=0.35; 95% CI=0.15-0.84) were associated with inadequate sun protection (Table 3).

Older survivors were less likely than younger survivors to participate in immunisation programmes (OR=0.76; 95% CI=0.53-0.97). Although the difference was not statistically significant (P=0.051), immunisation practices tended to be less common in survivors who did not have private medical insurance, compared with survivors who did (OR=0.78; 95% CI=0.45-1.06).

In terms of health-damaging behaviours (Table 4), compared with survivors who had completed education to a higher level than secondary school, adult survivors with a lower education level had greater odds of being current or ever-smokers (OR=5.13; 95% CI=1.48-17.75) and of being exposed to second-hand smoke (OR=3.52; 95% CI=1.42-8.69).

Nearly all participants stated that the provision of survivorship education (n=168, 84%) and late-effects screening services (n=157, 78.5%) would be the most important components of a survivorship programme (Fig 2). Moreover, helping survivors to understand and confront the fear of relapse (n=161, 80.5%) and addressing psychological distress (n=154, 77%) were the most popular psychosocial services. Among parents (n=45), learning how to parent a child with cancer (n=36, 80%) and psychosocial support for parents (n=33, 73.3%) were regarded as essential components.

Most participants (n=81, 40.5%) stated that their preferred enrolment time into a survivorship programme would be at the initiation of cancer treatment; some participants (n=58, 29.0%) stated that their preferred enrolment time would be during the transition from active treatment to follow-up (Table 5). In terms of modes of services, most respondents preferred dialogue sessions with healthcare professionals (n=156, 78.0%), interactive platforms (n=100, 50.0%), and support groups (n=101, 50.5%) [Table 5].

There is growing evidence that physical activity is a therapeutic strategy that may reduce the risks of systemic recurrence and mortality in cancer survivors.22 Similar to findings from other countries, we found that physical inactivity was highly prevalent among survivors of childhood cancer.9 23 This observation was not surprising, considering that the rate of physical activity is low among the general population in Hong Kong; only 40% to 46% of children and youth met physical activity guidelines for a mean duration of 60 minutes of moderate-to-vigorous physical activity per day.24 Although we did not identify any significant predictors of physical inactivity, this large proportion of inactive survivors indicates the need to further explore the reasons for this phenomenon and devise interventions to address them. For example, interventions targeting the survivor–parent dyad may promote common lifestyle behaviours within the families of survivors.25 Moreover, local adventure-based training and experiential learning programmes may enhance self-efficacy in survivors, thereby empowering them to initiate and maintain a physically active lifestyle.26 27

Our study found that only 38% of survivors reported frequently consuming a balanced diet. For example, survivors of CNS tumours were more likely to have poor dietary habits, compared with survivors of non-CNS malignancies. This is concerning because patients with CNS tumours are more vulnerable to developing metabolic syndromes related to complications associated with cranial radiation and neurosurgery. Poor dietary habits may further exacerbate the disease course of these late effects.28 This finding suggests that dietitians should give advice regarding stricter dietary control to optimise the health of CNS cancer survivors in Hong Kong.

Despite the extensive promotion of the seasonal influenza vaccination programme by the Hong Kong Special Administrative Region Government,29 only 30% of survivors indicated that they had received the influenza vaccine in the past year. Younger survivors were more likely to have participated in vaccination programmes, probably because school-age children are generally enrolled into the government immunisation programme that provides the hepatitis B, pneumococcal, and annual influenza vaccines, as well as the recently added human papillomavirus (HPV) vaccine.30 We acknowledge that our findings must be interpreted with caution because survivors might inaccurately recall or report their vaccination histories. However, these results have two important implications that warrant attention from the medical community. First, there is a need to educate survivors regarding the role of vaccination in preventing severe complications from infection (eg, influenza and pneumococcal vaccines for preventing seasonal flu and pneumonia, respectively) and other malignancies (eg, HPV vaccine for preventing cervical cancer). In particular, collaborations among schools and community physicians may help promote the uptake of HPV vaccines among adolescent female survivors.31 32 Second, children treated with chemotherapy for childhood malignancies reportedly may develop acquired immunological defects in both cell-mediated and humoral immunity, resulting in the loss of protection conferred by prior vaccinations.33 Future work should involve the development of clinical consensus guidelines regarding vaccination administration schedules for non-transplant survivors of childhood cancer, particularly survivors who have received intensive chemotherapy treatment.

Similar to the findings of Chan et al,13 we found that health-damaging practices are uncommon among local survivors. Both drinking and smoking rates were lower in this study than in studies from other developed countries.9 19 34 However, the reported rate (20.5%) of exposure to second-hand smoke in the home was surprisingly high. We speculate that this high rate is because the smoking rates of individuals above the age of 40 years remain relatively high in the general population (16.9% to 26.2% in men and 1.3% to 5.1% in women).35 Older family members, particularly men, might remain the main source of second-hand smoke for survivors. This observation underscores the need for continual efforts to encourage survivors to abstain from harmful health practices (particularly during the early survivorship phase) and the need for smoking cessation interventions to be provided for the comparatively few survivors and family members who are current smokers.

In this study, lower socioeconomic status was significantly associated with poor health practices. The association between possession of private medical insurance and immunisation practice was close to statistically significant, further suggesting that socioeconomic disparities hinder access to preventive care among cancer survivors. Underserved survivors may require special navigation services to support their adherence to surveillance, preventive care, and health-protective behaviours.36 Considering that only half of the survivors had private health insurance, collaborations between clinicians and policymakers could enable the establishment of a universal vaccination and late-effects screening programme for cancer survivors. Local research is needed to identify barriers to—and facilitators of—quality care and effective methods of outreach to underserved survivors.

Most survivors indicated that they would prefer to be enrolled into a survivorship programme early in the cancer care continuum. This is a promising prospect because survivors who had early access to structured survivorship programmes reportedly were more aware of their late effects, visited emergency departments less frequently, had higher cancer-specific health literacy, and tended to experience less emotional stress.3 Therefore, a structured survivorship programme is recommended to include cancer and late-effects screening, a specialist referral network, and psychosocial services for survivors and caregivers (Fig 3).4 12 In Hong Kong, the five major institutions that provide paediatric oncology care typically include these core services in their long-term follow-up programmes, although the specific services offered may differ among institutions. Overall, the centralisation of paediatric oncology services in the new Hong Kong Children’s Hospital has provided an unprecedented opportunity for oncologists and allied health professionals to initiate a formal paediatric cancer survivorship programme in Hong Kong. This will facilitate the development of a survivorship care model that is tailored to the healthcare system in Hong Kong.

Our participants regarded services concerning health issues (eg, education and screening for late effects) as the most important aspects of survivorship care. The COG has developed a set of “risk-based” guidelines, which refer to a personalised systematic plan of regular screening, surveillance, and prevention strategies based on a patient’s treatment, cancer experience, and personal factors.5 10 In an effort to improve the awareness of health issues in Chinese cancer survivors, we collaborated with the COG and launched a Chinese version of the Health Links patient education materials in May 2020.10 To our knowledge, this is the first set of publicly available authoritative resources regarding late effects that is written in a native Chinese language. Such initiatives are anticipated to assist survivors in taking age-appropriate ownership of their health and engaging as active partners with their health providers during the survivorship phase.

Our findings should be considered in the context of the following limitations. First, this single-centre study comprised a moderately small sample of survivors who were recruited through a convenience sampling approach. Moreover, eligible participants were identified from a long-term follow-up clinic that had a mean loss to follow-up rate of 15% to 20%. This is a recognised challenge in survivorship research because this population is often lost to follow-up from primary paediatric clinics as a result of their growing independence and mobility during advancement into adulthood.37 These study limitations may have introduced sampling bias because our participants may have been more likely to be health conscious than non-participants and survivors who had been lost to follow-up. Hence, the true uptake of health-protective behaviours among local survivors may be lower than the rates reported in this study, and our findings might not be generalisable to other survivors of childhood cancer in Hong Kong. Second, social desirability and recall bias may have affected the accuracy of the self-reported results. Future studies should adopt validated and more sensitive instruments to achieve a more objective evaluation of health behaviour. For example, physical activity and sleep can be better measured with actigraphy studies. Finally, the multiple predictors and covariates analysed in this study may have increased the risk of a Type I error. However, lifestyle itself is a complex phenotype that is likely to be influenced by intrinsic and extrinsic factors. Our findings should be validated using a larger-scale study that involves the prospective collection of outcome data to better reflect the trajectory of health behaviour changes and correlate these findings with the results in local cancer survivors.

Despite the low prevalences of health-damaging behaviours, the frequencies with which Hong Kong survivors of childhood cancer engaged in health-protective behaviours were unsatisfactory. These findings highlight the need to empower survivors to adopt health-protective behaviours. A potential intervention opportunity may involve engaging survivors and families in a structured comprehensive survivorship programme during their transition to survivorship. A multidisciplinary and interactive programme addressing late effects and psychosocial aspects may address the multifaceted needs of Hong Kong survivors of childhood cancer. Future work should aim to improve preventive care for underserved groups through advocacy and care coordination.

Concept or design: All authors.
Acquisition of data: YT Cheung, LS Yang, JCT Ma, PHK Woo, TCH Chan, SMS Luk.
Analysis or interpretation of data: YT Cheung, LS Yang, JCT Ma, PHK Woo, TCH Chan, SMS Luk.
Drafting of the manuscript: YT Cheung, TCH Chan, SMS Luk.
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.

All authors have disclosed no conflicts of interest.

We thank Dr Smita Bhatia and Dr Wendy Landier from the Institute for Cancer Outcomes and Survivorship, The University of Alabama in Birmingham, for serving as consultants on this project.

A portion of this work was presented at the 52nd Congress of the International Society of Paediatric Oncology (SIOP)–Virtual conference (14-17 October 2020), as well as the HKPS/HKCOP/HKPNA/HKCPN Joint Annual Scientific Meeting on 7 November 2020.

This study was supported by the Health and Medical Research Fund Research Fellowship, the Food and Health Bureau, The Government of the Hong Kong Special Administrative Region (Ref 03170047).

The study protocol was approved by The Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref: 2018.338). Written informed consent was obtained from all participants.

1. Howlader N, Noone AM, Krapcho M, et al. SEER Cancer Statistics Review (CSR) 1975-2014. Available from: https://seer.cancer.gov/archive/csr/1975_2014/. Accessed 3 Sep 2020.

2. Hong Kong Cancer Registry, Hospital Authority, Hong Kong SAR Government. Cancer in children and adolescents (0-18 years). Available from: https://www3.ha.org.hk/cancereg/children.asp. Accessed 3 Sep 2020.

3. Signorelli C, Wakefield CE, Fardell JE, et al. The impact of long-term follow-up care for childhood cancer survivors: a systematic review. Crit Rev Oncol Hematol 2017;114:131-8. Crossref

4. Children’s Oncology Group Nursing Discipline Clinical Practice Subcommittee/Survivorship in collaboration with the Late Effects Committee. Establishing and enhancing services for childhood cancer survivors. Long-term follow-up program resource guide. 2007. Available from: http://www.survivorshipguidelines.org/pdf/LTFUResourceGuide.pdf. Accessed 3 Sep 2020.

5. Landier W, Bhatia S, Eshelman DA, et al. Development of risk-based guidelines for paediatric cancer survivors: the Children’s Oncology Group long-term follow-up guidelines from the Children’s Oncology Group Late Effects Committee and Nursing Discipline. J Clin Oncol 2004;22:4979-90. Crossref

6. Hudson MM, Ness KK, Gurney JG, et al. Clinical ascertainment of health outcomes among adults treated for childhood cancer. JAMA 2013;309:2371-81. Crossref

7. Poon LH, Yu CP, Peng L, et al. Clinical ascertainment of health outcomes in Asian survivors of childhood cancer: a systematic review. J Cancer Surviv 2019;13:374-96. Crossref

8. Dixon SB, Bjornard KL, Alberts NM, et al. Factors influencing risk-based care of the childhood cancer survivor in the 21st century. CA Cancer J Clin 2018;68:133-52. Crossref

9. Ford JS, Barnett M, Werk R. Health behaviours of childhood cancer survivors. Children (Basel) 2014;1:355-73. Crossref

10. Children’s Oncology Group. Long-term follow-up guidelines for survivors of childhood, adolescent, and young adult cancers, version 5.0. October 2018. Available from: http://www.survivorshipguidelines.org/pdf/2018/COG_LTFU_Guidelines_v5.pdf. Accessed 3 Sep 2020.

11. George SM, Smith AW. Commentary: understanding risk behaviour among adolescent cancer survivors–are they more like healthy adolescents or is cancer a teachable moment? A commentary on Klosky and colleagues’ article on health behaviours in survivors of childhood cancer and their siblings. J Pediatr Psychol 2012;37:647-9. Crossref

12. Nekhlyudov L, Ganz PA, Arora NK, Rowland JH. Going beyond being lost in transition: a decade of progress in cancer survivorship. J Clin Oncol 2017;35:1978-81. Crossref

13. Chan CW, Choi KC, Chien WT, et al. Health behaviours of Chinese childhood cancer survivors: a comparison study with their siblings. Int J Environ Res Public Health 2020;17:6136. Crossref

14. Centres for Disease Control and Prevention, US Department of Health & Human Services. Youth risk behaviour surveillance–United States, 2013. Available from: https://www.cdc.gov/nchhstp/dear_colleague/2014/dcl-061314-hiv-prep.html. Accessed 4 Sep 2020.

15. Lown EA, Hijiya N, Zhang N, et al. Patterns and predictors of clustered risky health behaviours among adult survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. Cancer 2016;122:2747-56. Crossref

16. Klosky JL, Howell CR, Li Z, et al. Risky health behaviour among adolescents in the childhood cancer survivor study cohort. J Pediatr Psychol 2012;37:634-46. Crossref

17. Health and Medical Research Fund, Food and Health Bureau, Hong Kong SAR Government. Research Fellowship Scheme Approved Projects since 2016-17. Ref 03170047: Personalized risk-based care and education for early survivors of childhood cancer in Hong Kong. Available from: https://rfs2.fhb.gov.hk/images/fundedproject/Approved_Projects_Fellowship.pdf. Accessed 5 Jan 2021.

18. Landier W, Chen Y, Namdar G, et al. Impact of tailored education on awareness of personal risk for therapy-related complications among childhood cancer survivors. J Clin Oncol 2015;33:3887-93. Crossref

19. Lown EA, Goldsby R, Mertens AC, et al. Alcohol consumption patterns and risk factors among childhood cancer survivors compared to siblings and general population peers. Addiction 2008;103:1139-48. Crossref

20. Devine KA, Mertens AC, Whitton JA, et al. Factors associated with physical activity among adolescent and young adult survivors of early childhood cancer: a report from the childhood cancer survivor study (CCSS). Psychooncology 2018;27:613-9. Crossref

21. National Institute on Alcohol Abuse and Alcoholism, National Institute of Health, US Government. Drinking levels defined. Available from: https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking. Accessed 3 Sep 2020.

22. Cormie P, Zopf EM, Zhang X, Schmitz KH. The impact of exercise on cancer mortality, recurrence, and treatment-related adverse effects. Epidemiol Rev 2017;39:71-92. Crossref

23. Ness KK, Leisenring WM, Huang S, et al. Predictors of inactive lifestyle among adult survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. Cancer 2009;115:1984-94. Crossref

24. Huang WY, Wong SH, Sit CH, et al. Results from the Hong Kong’s 2018 report card on physical activity for children and youth. J Exerc Sci Fit 2019;17:14-9. Crossref

25. Badr H, Paxton RJ, Ater JL, Urbauer D, Demark-Wahnefried W. Health behaviours and weight status of childhood cancer survivors and their parents: similarities and opportunities for joint interventions. J Am Diet Assoc 2011;111:1917-23. Crossref

26. Li HC, Chung OK, Ho KY, Chiu SY, Lopez V. Effectiveness of an integrated adventure-based training and health education program in promoting regular physical activity among childhood cancer survivors. Psychooncology 2013;22:2601-10. Crossref

27. Ruble K, Scarvalone S, Gallicchio L, Davis C, Wells D. Group physical activity intervention for childhood cancer survivors: a pilot study. J Phys Act Health 2016;13:352-9. Crossref

28. Smith WA, Li C, Nottage KA, et al. Lifestyle and metabolic syndrome in adult survivors of childhood cancer: a report from the St. Jude Lifetime Cohort Study. Cancer 2014;120:2742-50. Crossref

29. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Vaccination subsidy scheme–general public. Available from: https://www.chp.gov.hk/en/features/46107.html. Accessed 14 Sep 2020.

30. Family Health Service, Department of Health, Hong Kong SAR Government. Child health. Schedule of Hong Kong Childhood Immunisation Programme. Available from: https://www.fhs.gov.hk/english/main_ser/child_health/child_health_recommend.html. Accessed 14 Sep 2020.

31. Chau JP, Lo SH, Choi KC, et al. Effects of a multidisciplinary team-led school-based human papillomavirus vaccination health-promotion programme on improving vaccine acceptance and uptake among female adolescents: a cluster randomized controlled trial. Medicine (Baltimore) 2020;99:e22072. Crossref

32. Yuen WW, Lee A, Chan PK, Tran L, Sayko E. Uptake of human papillomavirus (HPV) vaccination in Hong Kong: facilitators and barriers among adolescent girls and their parents. PLoS One 2018;13:e0194159. Crossref

33. Han JH, Harmoney KM, Dokmeci E, et al. Dynamic re-immunization of off-treatment childhood cancer survivors: an implementation feasibility study. PLoS One 2018;13:e0191804. Crossref

34. Gibson TM, Liu W, Armstrong GT, et al. Longitudinal smoking patterns in survivors of childhood cancer: an update from the Childhood Cancer Survivor Study. Cancer 2015;121:4035-43. Crossref

35. Census and Statistics Department, Hong Kong SAR Government. Pattern of smoking. In: Census and Statistics Department, The Government of the Hong Kong Special Administrative Region. Hong Kong Monthly Digest of Statistics January 2019. Available from: https://www.statistics.gov.hk/pub/B10100022019MM01B0100.pdf. Accessed 14 Sep 2020.

36. Lee Smith J, Hall IJ. Advancing health equity in cancer survivorship: opportunities for public health. Am J Prev Med 2015;49:S477-82. Crossref

37. Rokitka DA, Curtin C, Heffler JE, Zevon MA, Attwood K, Mahoney MC. Patterns of loss to follow-up care among childhood cancer survivors. J Adolesc Young Adult Oncol 2017;6:67-73. Crossref

Intracranial aneurysms constitute approximately 80% of all nontraumatic subarachnoid haemorrhages.1 Aneurysmal subarachnoid haemorrhage could be associated with 30-day mortality of 45%, and approximately half of the survivors will have irreversible brain damage.2 Knowledge of the intracranial aneurysm prevalence in a population allows health authorities to assess the severity of the problem and formulate appropriate healthcare policies. The prevalence of unruptured intracranial aneurysm has considerably varied among studies according to the detection method used.3 4 5 6 7 8 9 10 11 12 13 14 15 Notably, the prevalence of unruptured intracranial aneurysm generally increases with imaging sensitivity. Studies that used three-dimensional time-of-flight magnetic resonance angiography (MRA) with 3-T magnetic resonance imaging (MRI) showed a prevalence of 7.0% in a general population of Chinese adults age 35 to 75 years.14 Time-of-flight MRA is an ideal imaging technique for the analysis of intracranial aneurysm prevalence because of its non-invasive nature and its diagnostic accuracy, which is comparable to the accuracy of digital subtraction angiography.16 17 Information concerning the prevalence of intracranial aneurysms in the general population is important but has been unavailable in Hong Kong. We aimed to determine the prevalences of unruptured intracranial aneurysms in the Hong Kong population, individuals with symptoms related to intracranial aneurysms, and individuals with cerebrovascular disease (CVD).

Data on cerebral MRA examinations of Hong Kong residents conducted at St Teresa’s Hospital, a private community hospital that serves patients throughout Hong Kong, were extracted from hospital electronic records. Data were excluded if they were repeat MRA examinations or if the individual had a known history of ruptured/unruptured aneurysm or a family history of cerebral aneurysm.

Magnetic resonance angiography was the first-line imaging method used for assessment of intracranial vessels at St Teresa’s Hospital during the study period. Information retrieved from clinical records included the indication for MRA, presenting symptoms, medical history, family history, and MRA findings. Three groups were formed for analysis: a Symptom group that consisted of individuals who presented with any single symptom or combination of symptoms such as headache, any neurological symptoms related to intracranial aneurysms such as localised pain above or behind the eye, nausea, vomiting, or any visual symptom related to intracranial aneurysms (eg, diplopia, vision blurring, proptosis, or ptosis); a General group that consisted of individuals without any symptoms that were criteria for inclusion in the Symptom group and without a known history of CVD; and a CVD group that consisted of individuals with a known history of ischaemic stroke, transient ischaemic attack, intracranial stenosis, a history of intracerebral haemorrhage, arteriovenous malformation, or any other CVD other than cerebral aneurysm. The prevalences of cerebral aneurysm within these groups were recorded and compared among groups. The characteristics of aneurysms in the General group were extensively characterised.

The usual contra-indications for MRI were adopted.18 Magnetic resonance angiography was performed with a Siemens MAGNETOM MRI scanner (1.5T Vision, 1.5T Sonata, 1.5T Avanto, 3T Trio) using non-contrast time-of-flight sequence (repetition time 21-40 ms, echo time 3.8-7.2 ms, flip angle 18°-25°, matrix 195×512 to 250×512, field of view 200-230, slab 3-6, slices per slab 36-48, slice thickness 0.5-0.6 mm, acquisition time 4.32-8.45 minutes, with or without magnetic transfer). All MRI examinations were performed and reported by the same team of radiographers and radiologists; each member had at least 5 years of experience. The diagnosis of aneurysms in each participant was based on definite findings in the radiology report. Cases involving an inconclusive diagnosis of aneurysm without confirmatory computed tomography angiography or digital subtraction angiography findings were not included. Aneurysm types were saccular or fusiform. Aneurysm sizes were recorded as ≤5 mm, >5 mm to <10 mm, 10 to 20 mm, or >20 mm. Aneurysm locations were internal carotid artery, middle cerebral artery, anterior cerebral artery, posterior cerebral artery, or vertebrobasilar artery. Anterior communicating artery aneurysms were included in the anterior cerebral artery location category. Posterior communicating artery aneurysms were included in the posterior cerebral artery location category.

Prevalence analysis was based on participants, rather than aneurysms. In the General group, sex- and age-specific prevalences were analysed. The age threshold that demarcated the greatest change in prevalence was identified using odds ratio (OR) for the General group overall, as well as for all male participants and for all female participants within the General group. The prevalences in the General group overall, as well as its sex- and age-specific subgroups, were compared with the prevalences in the Symptom and CVD groups. The sizes and locations of unruptured intracranial aneurysms were analysed in the General group. For participants with multiple aneurysms, the largest aneurysm was selected for size and location analysis. The prevalence in the Hong Kong population was estimated from the sex- and age-specific prevalence in the General group in this study, with reference to the sex and age composition of the Hong Kong population. Because this cross-sectional analysis involved a long study period (ie, 15 years), a separate analysis of the sex- and age-specific prevalences of aneurysms among participants examined in the final 5 years was performed to determine whether any significant variations in sex- and age-specific prevalences occurred over time.

Statistical analyses were conducted using SPSS for Windows (version 20.0; IBM Corp, Armonk [NY], United States). All categorical variables are presented as number and percentage. All continuous variables are presented as median and interquartile range. Comparisons of prevalences among groups and subgroups were carried out using the Chi squared test. The Mann-Whitney U test was used for comparisons of continuous variables. P<0.05 was considered to indicate statistical significance. Comparisons of aneurysm prevalences between participant groups according to an age threshold were conducted using OR and 95% confidence interval (95% CI).

During the study period between July 1994 and December 2009, 8959 cerebral MRA examinations were conducted; 97.9% of the examined individuals were Chinese. Among them, 2252 repeat MRA examinations and 70 individuals with known history of ruptured/unruptured aneurysm or a family history of cerebral aneurysm were excluded (Fig 1). Finally, first cerebral MRA examinations of 6637 individuals were included for analysis (Table 1).

Table 1 shows the prevalences of unruptured intracranial aneurysm among the study groups. In the General group, the overall prevalence was 4.9%; the prevalences in women and men were 5.9% and 3.9%, respectively (P=0.004) [Table 2]. The prevalence was significantly higher in women than in men for almost all age ranges; it generally increased with age (Fig 2). The age threshold with the greatest change in unruptured intracranial aneurysm prevalence was 45 years in the General group overall (OR=3.1; 95% CI, 1.9-4.9), among men in the General group (OR=3.6, 95% CI, 1.6-8.3), and among women in the General group (OR=2.9, 95% CI, 1.6-5.2). In the General group overall, as well among men and women in the General group, the prevalences significantly differed between participants age <45 years and participants age ≥45 years (Table 2).

In the General group, the prevalences of unruptured intracranial aneurysm in men age <35, 35 to 44, 45 to 54, 55 to 64, 65 to 74 and >74 years were 1.2%, 1.8%, 2%, 3.4%, 7% and 6.6%, respectively; the corresponding prevalences in women were 1.9%, 1.7%, 4.8%, 7.3%, 7%, and 10.5%, respectively. Census data for Hong Kong from 2019 indicated that the numbers of men in the above age ranges were 1.25 million, 0.4652 million, 0.4896 million, 0.5952 million, 0.3804 million and 0.2514 million, respectively; the numbers of women in those age ranges were 1.3542 million, 0.7141 million, 0.6601 million, 0.6404 million, 0.394 million and 0.3262 million, respectively.19 Thus, we estimated the overall prevalence of unruptured intracranial aneurysm in the Hong Kong population to be 3.6%. There were no statistically significant differences in sex- and age-specific prevalences among participants examined in the final 5 years of the study, compared with participants examined throughout the 15.5-year study period (Table 3).

The unruptured intracranial aneurysm prevalence was significantly higher in the Symptom group than in the General group (6.3% vs 4.9%, P=0.018) [Table 2]. The prevalence was also significantly higher in the Symptom group than in the General group among all women participants, participants age <45 years, participants age ≥45 years, women age <45 years, and women age ≥45 years. Otherwise, the prevalences of unruptured intracranial aneurysm did not significantly differ between the General group (overall or subgroups) and the other main groups (Table 2).

Using the prevalence at age 35 to 44 years as the reference value, the prevalences for the subgroups of age 45 to 54, 55 to 64 and 65 to 74 years were 2.8-fold, 4.3-fold and 4.1-fold greater than the reference value, respectively, for women in the General group; these prevalences were 1.1-fold, 1.9-fold and 3.9-fold greater than the reference value, respectively, for men in the General group. For the subgroups of age 35 to 44, 45 to 54, 55 to 64 and 65 to 74 years, the prevalence ratios of women to men were 0.9, 2.4, 2.1 and 1, respectively, in the General group. Additionally, the prevalence ratios of the Symptom group to the General group in the four age subgroups varied from 4.2 to 1.5 for women and from 1.5 to 0.8 for men (Table 4, Fig 2).

Among 176 participants with one or more aneurysms in the General group, the largest aneurysm in each participant was selected for size and location analysis. Eleven participants (6.3%) had fusiform aneurysms, while 165 participants (93.8%) had saccular aneurysms. Furthermore, 157 participants (89.2%) had a single aneurysm, 16 participants (9.1%) had two aneurysms, and two participants (1.1%) had three aneurysms. Mirror aneurysm occurred in six participants (3.4%): five men and one woman. Mirror aneurysms were located in the cavernous internal carotid artery in two participants and in the non-cavernous internal carotid artery in four participants. The mirror aneurysms were bilaterally symmetrical; they measured ≤5 mm (four participants), 5-10 mm (one participant), and 10-20 mm (one participant). The size and location distributions of aneurysms in the General and Symptom groups are shown in Table 4. The proportion of posterior cerebral artery aneurysms was significantly greater in the Symptom group than in the General group (10.5% vs 3.4%, P=0.01) [Table 5].

It is resource-intensive to determine the unruptured intracranial aneurysm prevalence in any population because of the need for large-scale studies with prospective and random selection of participants, as well as the utilisation of accurate diagnostic means for aneurysm detection. To achieve this objective within the constraints of a hospital-based retrospective study, our study model used a hospital that served the whole population of Hong Kong; study participants comprised individuals without a family history, past history, or symptoms related to intracranial aneurysms, all of whom were referred to the hospital for body check with MRA. These individuals were assumed to be closely representative of a random sample of the Hong Kong population because there was no identifiable medical reason when they presented themselves to a doctor, therefore the sample was not biased. Because the prevalence of unruptured intracranial aneurysm is sex- and age-dependent, 11 the overall prevalence is dependent on the sex and age composition of participants in the study sample. Sex- and age-specific prevalences are more meaningful than overall prevalences because they allow comparisons among distinct target groups in the same study and in other studies. The analysis of unruptured intracranial aneurysm prevalence in individuals with symptoms and individuals with a known history of CVD provides useful information for physicians who must counsel such patients.

Studies of intracranial aneurysm prevalence have multiple limitations. Retrospective studies tend to show lower prevalence rates than do prospective studies.4 13 20 21 22 Prevalences vary according to the modality of aneurysm detection, such that prevalence rates tend to be much lower in autopsy studies than in arteriogram studies or MRA studies. Published prevalences have been 0.4% to 0.5% in retrospective autopsy studies,4 3.1% to 4.1% in prospective autopsy studies,4 0.65% to 4.4% in retrospective arteriogram studies,4 3% to 6.8% in prospective arteriogram studies,4 3.2% to 4.3% in retrospective MRA studies,21 22 and 5% to 7% in prospective MRA studies.13 14 The overall prevalences are also limited by the sex and age compositions of participants in each study. Autopsy and hospital-based retrospective studies usually have biased samples comprising individuals who presented to the hospital with clinical indications for investigation. Furthermore, the identification of an intracranial aneurysm at autopsy is greatly affected by the interest and enthusiasm of the examiner23; intracranial aneurysms are commonly overlooked at autopsy.24 These observations may explain the generally low prevalence rates of 0.2% to 2.2% reported in autopsy studies.6 25 26 27

Unruptured intracranial aneurysms are more common in women and relatively older individuals.4 11 14 21 Comparisons of unruptured intracranial aneurysm prevalence between published studies may not be meaningful without considering the sex ratio and age distribution of the study samples; such comparisons are not feasible for studies in which sex ratio and age distribution data are unavailable, and such studies are not uncommon. We compared our results with findings from a prospective study of Chinese individuals in Shanghai,14 in which the sex- and age-specific prevalences had been analysed; both studies shared some common trends. In the previous study, the overall prevalence was 7%; moreover, prevalence increased with age and the female-to-male prevalence ratio decreased with increasing age. These prevalence patterns were similar to patterns observed in the current study. Notably, the age- and sex-specific prevalences were consistently higher in the Shanghai study14 than in the current study, which suggests differences in intracranial aneurysm prevalence between the two populations.

There were two main limitations in this study. First, its duration (15 years) was relatively long. Although there were prevalence studies which involved cerebral arteriogram or autopsy lasting for up to 11 years or 30 years respectively,6 20 studies involving MRA typically lasted for 2 to 4 years.14 21 However, the results obtained from the long study period can be regarded as similar to the aggregate results of a series of consecutive short retrospective studies. The overall age- and sex-specific prevalences obtained during the long study period can be regarded as the average value of age- and sex-specific prevalences in the individual shorter studies. A major concern related to a long study period is the potential for variations in population demographics, particularly with respect to age and sex. Because analyses of age- and sex-specific prevalence were conducted in the present study, variations in age and sex composition throughout the population in various time periods were presumed not to affect the analysis outcome. The overall prevalence of a particular disease at any time point can be calculated from the sex and age composition of the population at that particular time point, together with the age- and sex-specific prevalences of that disease. Many published studies report an overall prevalence, which is meaningful only for a specific time period because it is dependent on the age and sex composition of the population during the study period. Age- and sex-specific prevalences are more meaningful than an overall prevalence, they could be regarded as part of the natural process of the disease specific to the population and are expected to exhibit fewer changes over time, therefore, they are more directly applicable to clinical practice. Consistent with that expectation, we found no statistically significant differences in age- and sex-specific prevalences in the final 5 years of the study period, compared with the study period overall.

Second, the assumption that the General group is a close representative of the general population in Hong Kong requires validation because certain factors associated with high aneurysm risk may be more common among individuals referred for examination via MRA. Based on the selection criteria for the General group, selection bias may have caused this group to be poorly representative of the general population in Hong Kong. To minimise the potential for such bias, factors associated with a high risk of cerebral aneurysm were addressed by the exclusion of individuals with a family history, past history, or symptoms related to cerebral aneurysm. Additionally, the proportions of individuals with hypertension and a smoking habit in the General group did not significantly differ from those proportions in the wider Hong Kong population. Finally, potential biases involving advanced age and female sex were addressed with age- and sex-specific analyses.

In conclusion, the estimated overall prevalence of unruptured intracranial aneurysm in the Hong Kong population was 3.6%. The prevalence was significantly higher in the Symptom group than in the General group. However, the unruptured intracranial aneurysm prevalences did not differ between the General and CVD groups, or between the Symptom and CVD groups.

Concept or design: SCH Yu.
Acquisition of data: PW Cheng, GE Antonio, HTG Ma, SCC Chan.
Analysis or interpretation of data: SCH Yu, TWW Lau, SCC Chan.
Drafting of the manuscript: SCH Yu, PW Cheng.
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.

The authors declare that they have no conflict of interest.

The authors thank the following individuals for substantial contributions to data interpretation and manuscript revision: HKM Cheng, BMH Lai, GKC Wong, VKY Pang, ACO Tsang, DPH Wong, KM Leung, R Lee, TKT Chan, and CB Tan.

This study was funded by the Vascular and Interventional Radiology Foundation. The funding body had no involvement in the design of the study; in the collection, analysis, and interpretation of data; or in writing the manuscript.

This study was approved by the Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (CRE-2010-381). The requirement for informed consent was waived owing to the retrospective nature of the study.

1. Brown RD. Unruptured intracranial aneurysms. Semin Neurol 2010;30:537-44. Crossref

2. Graves EJ. Detailed diagnoses and procedures, national hospital discharge survey, 1990. Vital Health Stat 13 1992;113:1-225.

3. Ujiie H, Sato K, Onda H, et al. Clinical analysis of incidentally discovered unruptured aneurysms. Stroke 1993;24:1850-6. Crossref

4. Rinkel GJ, Djibuti M, Algra A, van Gijn J. Prevalence and risk of rupture of intracranial aneurysms: a systematic review. Stroke 1998;29:251-6. Crossref

5. International Study of Unruptured Intracranial Aneurysms Investigators. Unruptured intracranial aneurysms—risk of rupture and risks of surgical intervention. N Engl J Med 1998;339:1725-33. Crossref

6. Iwamoto H, Kiyohara Y, Fujishima M, et al. Prevalence of intracranial saccular aneurysms in a Japanese community based on a consecutive autopsy series during a 30-year observation period. The Hisayama study. Stroke 1999;30:1390-5. Crossref

7. Katzman GL, Dagher AP, Patronas NJ. Incidental findings on brain magnetic resonance imaging from 1000 asymptomatic volunteers. JAMA 1999;282:36-9. Crossref

8. Wardlaw JM, White PM. The detection and management of unruptured intracranial aneurysms. Brain 2000;123:205-21. Crossref

9. Horikoshi T, Akiyama I, Yamagata Z, Nukui H. Retrospective analysis of the prevalence of asymptomatic cerebral aneurysm in 4518 patients undergoing magnetic resonance angiography—when does cerebral aneurysm develop? Neurol Med Chir (Tokyo) 2002;42:105-12. Crossref

10. Vernooij MW, Ikram MA, Tanghe HL, et al. Incidental findings on brain MRI in the general population. N Engl J Med 2007;357:1821-88. Crossref

11. Vlak MH, Algra A, Brandenburg R, Rinkel GJ. Prevalence of unruptured intracranial aneurysms, with emphasis on sex, age, comorbidity, country, and time period: a systematic review and meta-analysis. Lancet Neurol 2011;10:626-36. Crossref

12. Agarwal N, Gala NB, Choudhry OJ, et al. Prevalence of asymptomatic incidental aneurysms: a review of 2685 computed tomographic angiograms. World Neurosurg 2014;82:1086-90. Crossref

13. Chan DY, Abrigo JM, Cheung TC, et al. Screening for intracranial aneurysms? Prevalence of unruptured intracranial aneurysms in Hong Kong Chinese. J Neurosurg 2016;124:1245-9. Crossref

14. Li MH, Chen SW, Li YD, et al. Prevalence of unruptured cerebral aneurysms in Chinese adults aged 35 to 75 years: a cross-sectional study. Ann Intern Med 2013; 159:514-21. Crossref

15. Igase K, Matsubara I, Igase M, Miyazaki H, Sadamoto K. Initial experience in evaluating the prevalence of unruptured intracranial aneurysms detected on 3-Tesla MRI. Cerebrovasc Dis 2012;33:348-53. Crossref

16. Li MH, Cheng YS, Li YD, et al. Large-cohort comparison between three-dimensional time-of-flight magnetic resonance and rotational digital subtraction angiographies in intracranial aneurysm detection. Stroke 2009;40:3127-9. Crossref

17. Li MH, Li YD, Tan HQ, Gu BX, et al. Contrast-free MRA at 3.0 T for the detection of intracranial aneurysms. Neurology 2011;77:667-76. Crossref

18. T Dill. Contraindications to magnetic resonance imaging. Heart 2008;94:943-8. Crossref

19. Census and Statistics Department, Hong Kong SAR Government. Available from: www.censtatd.gov.hk/en/web_table.html?id=1A. Accessed 18 Sep 2020.

20. Winn HR, Jane JA Sr, Taylor J, Kaiser D, Britz GW. Prevalence of asymptomatic incidental aneurysms: review of 4568 arteriograms. J Neurosurg 2002;96:43-9. Crossref

21. Harada K, Fukuyama K, Shirouzu T, et al. Prevalence of unruptured intracranial aneurysms in healthy asymptomatic Japanese adults: differences in gender and age. Acta Neurochir (Wien) 2013;155:2037-43. Crossref

22. Imaizumi Y, Mizutani T, Shimizu K, Sato Y, Taguchi J. Detection rates and sites of unruptured intracranial aneurysms according to sex and age: an analysis of MR angiography–based brain examinations of 4070 healthy Japanese adults. J Neurosurg 2018;130:573-8. Crossref

23. Chason JL, Hindman WM. Berry aneurysms of the circle of Willis; results of a planned autopsy study. Neurology 1958,8:41-4. Crossref

24. Stehbens WE. Aneurysms and anatomical variation of cerebral arteries. Arch Pathol 1963;75:45-64.

25. Housepian EM, Pool JL. A systematic analysis of intracranial aneurysms from the autopsy file of the Presbyterian Hospital, 1914 to 1956. J Neuropathol Exp Neurol 1958,17:409-23. Crossref

26. McCormick WF, Nofzinger JD. Saccular intracranial aneurysms: an autopsy study. J Neurosurg 1965;22:155-9. Crossref

27. Inagawa T, Hirano A. Autopsy study of unruptured incidental intracranial aneurysms. Surg Neurol 1990;34:361-5. Crossref

Enhanced recovery after surgery (ERAS) practices were developed in the 1990s whereby multiple modalities of intervention1 were introduced perioperatively to improve postoperative recovery,2 reduce length of stay (LOS),3 and lower the incidence of perioperative morbidity.4 These practices have been widely adopted in many surgical fields,5 6 7 including orthopaedics.8 Further enhancements of postoperative pain management, venous thromboembolism prophylaxis, and early mobilisation have led to encouraging results in primary total hip arthroplasty (THA) and total knee arthroplasty (TKA); such results have included earlier recovery,9 LOS reduction,10 improved function,11 and lower venous thromboembolism incidence12 without declines in patient satisfaction, postoperative complication rate,13 or cost.14 The development of ERAS practices has matured with the progressive introduction of standardised clinical pathways for all patients receiving THA or TKA,15 also referred as fast-track hip and knee arthroplasty.16 These ERAS practices have become the standard of care in most joint replacement centres.17 18

Because of the ageing population and increasing incidence of degenerative joint disease,19 the wait time for elective TKA in a public joint replacement service can reach 5 years in Hong Kong20; thus, many patients visit private orthopaedic surgeons for earlier surgery. Despite the presence of robust joint replacement options in the Hong Kong orthopaedic community, there remain differences between public and private hospital settings in terms of the environment, perioperative medical care, and service availability. To our knowledge, no studies have been published regarding the effects of ERAS practices on LOS after lower limb total joint arthroplasty, or whether ERAS practices could be implemented as a standardised protocol by private hospitals in Hong Kong.

Therefore, this study investigated whether the promising results of ERAS practices could be reproduced in private hospitals by comparison of LOS among patients who underwent unilateral primary THA or TKA by a single surgeon at Hong Kong Sanatorium & Hospital, a private hospital in Hong Kong, before and after the implementation of ERAS.

Patients who had undergone unilateral primary THA or TKA by the senior author (KYC) at Hong Kong Sanatorium & Hospital, Hong Kong, were included in the study cohort. Patients with revision arthroplasty, one-stage bilateral arthroplasty, and unicompartmental knee arthroplasty were excluded. Because ERAS practices were progressively implemented from 2015 to 2016, we allocated patients who were treated from 2012 to 2014 into the conventional group and patients who were treated from 2017 to 2018 into the ERAS group.

Our ERAS practices were generally similar to conventional practices. Most patients underwent surgery on the morning after an evening admission. Most patients received spinal anaesthesia unless contra-indicated (eg, ankylosing spondylitis, severe spinal deformity, coagulopathy, or fixed cardiac output state); routine sedation (using intravenous midazolam and propofol) was also conducted to improve patient comfort. Cementless THA systems, either a Pinnacle acetabular cup with a Summit femoral stem (DePuy, Warsaw [IN], US) or an R3 acetabular cup with a Synergy femoral stem (Smith & Nephew, Auckland, New Zealand), were implemented by means of a posterolateral approach. Total knee arthroplasty was performed using a medial parapatellar approach with thigh tourniquet and conventional instruments. Cemented rotating platform TKA systems were used: either a Legacy Posterior Stabilised Flex Mobile prosthesis (Zimmer, Warsaw [IN], US) or an Attune prosthesis (DePuy). A Foley urinary catheter was inserted only on urinary retention with bladder volume of >800 mL.21 Prophylactic antibiotics were administered on the induction of anaesthesia, then continued for 2 days after surgery. Prophylaxis against venous thromboembolism, both pharmacological (with subcutaneous enoxaparin) and mechanical (with a sequential compression device), was routinely implemented. Patients were discharged home when they could safely exit their beds without assistance and stably walk using an assistive device without any sign of complications.

When using ERAS practices, a higher dose of intravenous steroid is administered on the induction of anaesthesia for both THA and TKA. In the conventional group, 4 to 8 mg of dexamethasone was administered; in the ERAS group, 125 mg of methylprednisolone (equivalent to 25 mg of dexamethasone) was administered instead.22 Notably, high-dose glucocorticoids before arthroplasty are reportedly safe and recommended for routine use.23

In the conventional group, no standard pain control regimen was established. Pain medications were prescribed at the discretion of anaesthetists or surgeons, including the use of femoral nerve block and postoperative patient-controlled analgesia pump. Pain management was standardised and optimised in the ERAS group, particularly for patients undergoing TKA. Pre-emptive analgesia was implemented, such that patients routinely began oral pregabalin and transdermal buprenorphine patch treatments before surgery. Preventive analgesia was also employed both intra- and post-operatively. A periarticular “cocktail” injection of local infiltrative analgesia24—consisting of ropivacaine, ketorolac, and 1:1000 adrenaline—was injected into the posterior joint capsule before implantation of the prosthesis; it was also injected into the subcutaneous layer anteriorly and intra-articularly during wound closure. After surgery, patients received multimodal oral analgesia including regular cyclooxygenase-2 inhibitors or non-steroidal anti-inflammatory drugs, pregabalin, and paracetamol. Buprenorphine patch treatment was maintained for 5 to 7 days. Patient-controlled analgesia was omitted when using ERAS practices. In contrast, the pain control requirement was lower for patients undergoing THA. In both conventional and ERAS groups, local anaesthetic (bupivacaine) was injected into the subcutaneous plane before skin closure, while oral paracetamol was prescribed after surgery. After discharge from the hospital, patients who underwent TKA were administered non-steroidal anti-inflammatory drugs, pregabalin, and paracetamol for up to 5 weeks after surgery; most patients undergoing THA were prescribed paracetamol alone.

Prophylactic intravenous palonosetron was routinely administered to prevent postoperative nausea and vomiting; intravenous metoclopramide was used to manage breakthrough symptoms.

Tranexamic acid was routinely used in the ERAS group to minimise bleeding and the need for transfusion. For patients undergoing TKA, 1 g of tranexamic acid was injected intra-articularly after deep layer closure. No routine use of tranexamic acid was adopted in conventional practices. A deep drain was used when adhering to conventional practices but not when adhering to ERAS practices. For patients undergoing THA, intravenous tranexamic acid was administered at the same time as induction of anaesthesia; a deep drain was also used and removed the next morning.

While hypnotics were only administered on request when in the conventional group, patients in the ERAS group were routinely prescribed hypnotics the night before surgery and the first 2 to 3 days after surgery. This helped patients in the ERAS group to comply with the rehabilitation programme after surgery.

Same-day or day-zero rehabilitation was implemented in the ERAS group. Patients in the conventional group had bed rest on the day of surgery, then began mobilisation on postoperative day 1. Conversely, patients in the ERAS group who underwent morning surgery were encouraged to mobilise in the afternoon or evening on the same day, under physiotherapist supervision.

The outcome was postoperative LOS, which was denoted by the number of days after surgery when the patient was discharged from the hospital. The day of surgery was regarded as postoperative day 0. Discharge criteria remained consistent throughout the study period (ie, safe exit from bed without assistance and stable walking using an assistive device), as described above. The proportion of patients discharged on or before postoperative day 3 in each group was compared. We also investigated the effects of age, sex, nature of surgery (THA versus TKA), and class of hospital bed (general versus private ward) on LOS.

Patient data were anonymously entered into an encrypted file to ensure privacy. Data analysis was performed using SPSS (Window version 26.0; IBM Corp, Armonk [NY], US). The Chi squared test, independent samples t test with two-tailed significance, and one-way analysis of variance were used for comparisons. A P value of <0.05 was considered statistically significant.

In total, 228 patients were included: 117 in the conventional group and 111 in the ERAS group. The mean and median ages did not significantly differ between the conventional and ERAS groups (Table 1). Most patients were aged between 50 and 89 years (89.7% in the conventional group and 97.2% in the ERAS group); however, the distribution of ages did not significantly differ between groups. The distributions of sex, nature of surgery, and class of hospital bed also did not significantly differ between the conventional and ERAS groups (Table 1).

The mean LOS significantly improved from 5.16 ± 2.06 days to 3.28 ± 1.04 days (P<0.001) after ERAS implementation (Table 2). Patients discharged on or before postoperative day 3 comprised 13.7% of the conventional group and 77.5% of the ERAS group (P<0.001).

Subgroup analysis was performed to examine the effects of sex, nature of surgery, class of hospital bed (Table 3), and age-group (Fig) on LOS.

In the conventional group, there were no significant differences in mean LOS between female and male patients, patients receiving TKA and patients receiving THA, or general ward and private ward patients (Table 3). One-way analysis of variance showed a significant difference in mean LOS among age-groups (F [7, 109]=2.58, P=0.017) [Fig]. The mean LOS generally increased as age increased from the third decade (3 days) to the ninth decade (7 days); however, two patients in the 20-29 age-group had exceptionally long hospital stays.

In the ERAS group, there were no significant differences in mean LOS between female and male patients or between patients receiving TKA and patients receiving THA (Table 3). One-way analysis of variance showed that age did not have a significant effect on the mean LOS (F [5, 105]=1.13, P=0.348) [Fig]. However, the mean LOS significantly differed between general ward and private ward patients (3.06 ± 0.59 vs 3.66 ± 1.46 days, P=0.003) [Table 3].

No postoperative complications or instances of 30-day re-admission were observed among patients who underwent TKA. Among patients who underwent THA, three (two from the conventional group, one from the ERAS group) had complications. In the conventional group, one patient with spondyloepiphyseal dysplasia experienced dislocation during in-patient stay, which required closed reduction; one patient experienced dislocation during postoperative week 3, which required re-admission and revision to offset the liner and a longer neck hip ball to improve soft tissue tension. In the ERAS group, one patient had periprosthetic femoral fracture after an accidental fall on postoperative day 13, which required re-admission with revision to the long cementless stem, as well as cable fixation. No patients in either group experienced postoperative wounds or periprosthetic infections.

Despite more efficient service provision, postoperative LOS in private hospitals might be limited by confounders that surgeons cannot control (eg, patient preference and financial factors).25 Nevertheless, it was unsurprising that our results were consistent with previous literature: ERAS practices are effective for reducing the LOS after unilateral primary arthroplasty.

Regarding factors that affect postoperative LOS, a significant difference in the mean LOS was observed between general ward and private ward patients in the ERAS group. In public hospitals, the LOS among patients with worse socio-economic backgrounds is often limited by inadequate social support from family after discharge26 or a suboptimal home environment (eg, non-lift landing flats in older urban buildings).27 While placement issues are rarely problematic for patients in private hospitals,28 a possible explanation for the difference in LOS between general ward and private ward patients, where the cost difference is on average 5 times higher, is that patients with better socio-economic backgrounds may have higher expectations for surgical outcomes29; thus, they may tolerate longer hospital stays for rehabilitation, despite the higher costs of such stays. Private insurance is also reportedly an independent predictor of discharge delay despite objective readiness for discharge30; however, we presumed that the effect of insurance was not applicable in the present study because fewer than 10% of patients in our cohort had no insurance coverage. Furthermore, no significant differences in the mean LOS were noted with regard to the nature of surgery, sex, or age in the ERAS group. These findings may be related to the use of standardised ERAS practices and perioperative protocols, which have minimised variation in patient management.31

The implementation of ERAS practices in private hospitals is potentially beneficial to all stakeholders (including hospital administrators) because it facilitates hospital bed availability, while reducing costs via shorter convalescence duration and reduced morbidity.32 However, there are some important challenges for surgeons who wish to initiate ERAS practices in private centres. These challenges include occasional requirements for minor alterations in ward environments, changes in anaesthesia technique, rapid turnover of in-house surgical staff, and noncompliance with ERAS practices.33 Furthermore, a large caseload might be necessary to attract a dedicated multidisciplinary team for the sustainable development of ERAS practices in private centres. While it may be challenging to achieve full ERAS implementation in a short period of time, the stepwise addition of ERAS components might improve patient outcomes in private hospitals.34

There were some limitations in this study. First, this study used a retrospective design without randomisation, which may have led to imbalance and bias in the results. Second, this study only involved patients from a single surgeon; thus, the sample size was small. Third, differences in functional status and co-morbidities were not considered in the analysis, as the electronic health record sharing system between public and private hospitals was only established in 2016 so complete acquisition of patient’s parameters was not possible. Finally, other clinical outcome parameters and patient satisfaction were not investigated; these will be examined in a future study, where a thorough documentation in patient reported outcome measure and clinician-based outcome measure will improve the validity of results.

In conclusion, ERAS practices produced significant improvement in mean postoperative LOS, compared to conventional practices, for patients who underwent unilateral primary THA or TKA in a private hospital. Specifically, a significantly greater proportion of patients in the ERAS group were able to return home on or before postoperative day 3. The findings indicate that the good outcomes of ERAS practices in public joint replacement centres can be reproduced in private hospitals with sufficient caseloads and consistent implementation of ERAS practices.

Concept or design: KY Chiu.
Acquisition of data: MMT Chung.
Analysis or interpretation of data: MMT Chung.
Drafting of the manuscript: MMT Chung, JKF Ng, FY Ng, PK Chan.
Critical revision of the manuscript for important intellectual content: KY Chiu.

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.

All authors have disclosed no conflicts of interest.

The results of this study were presented in the Hong Kong Orthopaedic Association 40th Annual Congress in Hong Kong (31 October to 1 November 2020).

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

This study was approved by the Hong Kong Sanatorium & Hospital Medical Group Research Committee (Ref RC-2019- 25). The requirement for patient consent was waived for this retrospective study.

1. Kehlet H. Multimodal approach to postoperative recovery. Curr Opin Crit Care 2009;15:355-8. Crossref

2. Kehlet H, Dahl JB. Anaesthesia, surgery, and challenges in postoperative recovery. Lancet 2003;362:1921-8. Crossref

3. Kehlet H. Fast-track hip and knee arthroplasty. Lancet 2013;381:1600-2. Crossref

4. Rogers LJ, Bleetman D, Messenger DE, et al. The impact of enhanced recovery after surgery (ERAS) protocol compliance on morbidity from resection for primary lung cancer. J Thoracic Cardiovasc Surg 2018;155:1843-52. Crossref

5. Geltzeiler CB, Rotramel A, Wilson C, Deng L, Whiteford MH, Frankhouse J. Prospective study of colorectal enhanced recovery after surgery in a community hospital. JAMA Surg 2014;149:955-61. Crossref

6. Melnyk M, Casey RG, Black P, Koupparis AJ. Enhanced recovery after surgery (ERAS) protocols: time to change practice? Can Urol Assoc J 2011;5:342-8. Crossref

7. Scheib SA, Thomassee M, Kenner JL. Enhanced recovery after surgery in gynecology: a review of the literature. J Minim Invasive Gynecol 2019;26:327-43. Crossref

8. Andersen LØ, Gaarn-Larsen L, Kristensen BB, Husted H, Otte KS, Kehlet H. Subacute pain and function after fast-track hip and knee arthroplasty. Anaesthesia 2009;64:508-13. Crossref

9. Wu CL, Raja SN. Treatment of acute postoperative pain. Lancet 2011;377:2215-25. Crossref

10. Auyong DB, Allen CJ, Pahang JA, Clabeaux JJ, MacDonald KM, Hanson NA. Reduced length of hospitalization in primary total knee arthroplasty patients using an updated enhanced recovery after surgery (ERAS) pathway. J Arthroplasty 2015;30:1705-9. Crossref

11. Pua YH, Ong PH. Association of early ambulation with length of stay and costs in total knee arthroplasty: retrospective cohort study. Am J Phys Med Rehabil 2014;93:962-70. Crossref

12. Pearse EO, Caldwell BF, Lockwood RJ, Hollard J. Early mobilisation after conventional knee replacement may reduce the risk of postoperative venous thromboembolism. J Bone Joint Surg Br 2007;89:316-22. Crossref

13. Zhu S, Qian W, Jiang C, Ye C, Chen X. Enhanced recovery after surgery for hip mand knee arthroplasty: a systematic review and meta-analysis. Postgrad Med J 2017;93:736-42. Crossref

14. Duncan CM, Hall Long K, Warner DO, Hebl JR. The economic implications of a multimodal analgesic regimen for patients undergoing major orthopaedic surgery: a comparative study of direct costs. Reg Anesth Pain Med 2009;34:301-7. Crossref

15. Duggal S, Flics S, Cornell CN. Introduction of clinical pathways in orthopedic surgical care: the experience of the hospital for special surgery. In: Ronald MacKenzie C, Cornell CN, Memtsoudis SG, editors. Perioperative Care of the Orthopedic Patient. New York: Springer; 2014: 365-71. Crossref

16. Husted H. Fast-track hip and knee arthroplasty: clinical and organizational aspects. Acta Orthop Suppl 2012;83:1-39. Crossref

17. Christelis N, Wallace S, Sage CE, et al. An enhanced recovery after surgery program for hip and knee arthroplasty. Med J Aust 2015;202:363-8. Crossref

18. Soffin EM, YaDeau JT. Enhanced recovery after surgery for primary hip and knee arthroplasty: a review of the evidence. Br J Anaesth 2016;117(Suppl 3):iii62-72. Crossref

19. 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.

20. Hospital Authority, Hong Kong SAR Government. Elective total joint replacement surgery. 2019. Available from: https://www.ha.org.hk/visitor/ha_visitor_index.asp?Content_ID=221223&Lang=EN%20G&Dimension=100&Parent_ID=214172&Ver=HTML. Accessed 25 Feb 2020. Crossref

21. Bjerregaard LS, Hornum U, Troldborg C, Bogoe S, Bagi P, Kehlet H. Postoperative urinary catheterization thresholds of 500 versus 800 ml after fast-track total hip and knee arthroplasty: a randomized, open-label, controlled trial. Anesthesiology 2016;124:1256-64. Crossref

22. Lunn TH, Kristensen BB, Andersen LØ, et al. Effect of high-dose preoperative methylprednisolone on pain and recovery after total knee arthroplasty: a randomized, placebo-controlled trial. Br J Anaesth 2011;106:230-8. Crossref

23. Kehlet H, Lindberg-Larsen V. High-dose glucocorticoid before hip and knee arthroplasty: to use or not to use-that's the question. Acta Orthop 2018;89:477-9. Crossref

24. Ng FY, Ng JK, Chiu KY, Yan CH, Chan CW. Multimodal periarticular injection vs continuous femoral nerve block after total knee arthroplasty: a prospective, crossover, randomized clinical trial. J Arthroplasty 2012;27:1234-8. Crossref

25. Badham J, Brandrup J. Length of stay comparisons for private and public hospitals. Aust Health Rev 2000;23:162-70. Crossref

26. Freitas A, Silva-Costa T, Lopes F, et al. Factors influencing hospital high length of stay outliers. BMC Health Serv Res 2012;12:265. Crossref

27. Waring J, Marshall F, Bishop S, et al. An ethnographic study of knowledge sharing across the boundaries between care processes, services and organisations: the contributions to ‘safe’ hospital discharge. Southampton (UK): NIHR Journals Library; Sep 2014. Crossref

28. Perelman J, Closon MC. Impact of socioeconomic factors on in-patient length of stay and their consequences in per case hospital payment systems. J Health Serv Res Policy 2011;16:197-202. Crossref

29. Li X, Galvin JW, Li C, Agrawal R, Curry EJ. The impact of socioeconomic status on outcomes in orthopaedic surgery. J Bone Joint Surg Am 2020;102:428-44. Crossref

30. Celio DA, Poggi R, Schmalzbauer M, Rosso R, Majno P, Christoforidis D. ERAS, length of stay and private insurance: a retrospective study. Int J Colorectal Dis 2019;34:1865-70. Crossref

31. Barbieri A, Vanhaecht K, Van Herck P, et al. Effects of clinical pathways in the joint replacement: a meta-analysis. BMC Med 2009;7:32. Crossref

32. Stowers MD, Lemanu DP, Hill AG. Health economics in enhanced recovery after surgery programs. Can J Anaesth 2015;62:219-30. Crossref

33. Kahokehr A, Sammour T, Zargar-Shoshtari K, Thompson L, Hill AG. Implementation of ERAS and how to overcome the barriers. Int J Surg 2009;7:16-9. Crossref

34. Tan NL, Hunt JL, Gwini SM. Does implementation of an enhanced recovery after surgery program for hip replacement improve quality of recovery in an Australian private hospital: a quality improvement study. BMC Anesthesiol 2018;18:64. Crossref

Labour is regarded as a time of suffering in a woman’s life, during which she may experience intensive pain that lasts for many hours. Ineffective labour pain management could create a negative life experience for a woman, which may negatively impact postpartum sexual and marital satisfaction.1 2 Labour pain involves both physical and psychological elements such as uterine contractions, tension, fear, anxiety, and the sensations of powerlessness and a loss of control.3 Current remedies for labour pain include pharmacological and non-pharmacological interventions. The most common pharmacological interventions include nitrous oxide inhalation, the injection of narcotic analgesics (eg, pethidine), and epidural analgesia. However, these methods are associated with adverse effects such as nausea and vomiting, longer first and second stages of labour, hypotension, motor blockade, fever, and urinary retention; they can also lead to neonatal respiratory depression and newborn sleepiness that affects breastfeeding.4 5 6 7 8 9 Hence, women prefer safer and simpler non-pharmacological pain relief methods.10 11

A notable non-pharmacological remedy is massage, which may provide pain relief to the site of application, along with overall psychological relaxation.12 The pressure applied during massage is presumed to block the transmission of pain impulses to the brain, while stimulating local release of endorphins.13 Randomised controlled trials concerning intrapartum massage have been conducted in various countries over the past two decades.12 14 15 16 17 18 19 20 21 22 23 However, there have been conflicting findings concerning beneficial effects (ie, reductions in pain score or the use of pharmacological analgesia)12 14 15 16 17 21 because of small sample sizes which ranged from 28 women12 to 176 women.21 Furthermore, the duration of intrapartum massage was either unspecified15 21 or lasted for only 30 to 40 minutes.12 14 18 19 22 23 In addition, intrapartum massage was performed by various types of people: student midwives,22 therapists17 19 or partners who had received training by therapists immediately before labour.12 14 These factors probably influenced the effectiveness, consistency, and duration of the application of massage. A recent Cochrane review concluded that the current quality of evidence regarding intrapartum massage is low to very low.24 Therefore, this randomised controlled study investigated the efficacy of a comprehensive massage programme, combined with controlled breathing and visualisation—all initiated during the antenatal period—as a non-pharmacological pain relief method during labour, with the goal of reducing pethidine or epidural analgesia use.

This randomised controlled study was conducted in two public hospitals in Hong Kong, where the midwives were responsible for intrapartum management and natural vaginal delivery of low-risk pregnancies. The respective annual childbirth rates were approximately 5000 and 7000; the caesarean section rates were 21% and 23%.25 Recruitment commenced in September 2016 and completed in December 2017. The recruitment of women was conducted at 32 to 36 weeks of gestation during their routine antenatal visit by a team of research midwives. The inclusion criteria were low-risk nulliparous Chinese women aged ≥18 years, who could communicate in Cantonese, and who carried a singleton pregnancy without known contraindications for vaginal delivery. Exclusion criteria were the use of massage among women in the control group, the absence of a partner to learn the massage technique, planned delivery in hospitals other than the study sites, and planned caesarean delivery. There was no exclusion of recruited women who attempted vaginal delivery or induction of labour but eventually required intrapartum caesarean delivery.

Randomisation was conducted via two-by-two blocking with a block size of 4; a computer-generated number indicating either the study or control group was sealed in an opaque envelope. After a woman had provided written informed consent to participate, the midwife revealed the group allocation by opening the envelope. Because there were multiple midwifery staff responsible for participant recruitment at different occasions, none of the staff were aware of the allocation of previous participants; hence, they were unable to guess the group allocation.

Couples (ie, participating women and their partners) randomised to the massage group were invited to attend a 2-hour childbirth massage programme class at 36 weeks of gestation. This programme was based on the United Kingdom’s Royal College of Midwives accredited course ‘Towards Natural Childbirth and Beyond’.26 It included a 30-minute theoretical explanation of the evidence underpinning the childbirth massage programme, followed by a 90-minute practicum. During the 90-minute practicum, the couples received training by accredited midwifery trainers with respect to the massage technique, controlled breathing, and visualisation, in accordance with the methods used in previous studies.16 20 The massage areas included the lower back and four limbs. Couples were taught how to synchronise the massage strokes with slow rhythmic breathing. Visualisation (ie, a mind mapping component) was also taught.26 In this process, the woman was asked to imagine something comfortable, which could bring her to a relaxed state. Subsequently, the couples were encouraged to practise the massage technique regularly at home in the evening, in a dimly lit and quiet environment, with the aim of encouraging relaxation and improving the quality and duration of sleep.27 The control group received standard antenatal education without instruction concerning massage, controlled breathing, or visualisation techniques.

When a woman in the massage group was admitted to the study hospital at onset of labour or for planned labour induction, her partner was first asked to demonstrate massage technique to the research team midwives to ensure that the partner could perform the procedure properly. If labour was not yet established, each woman was encouraged to relax through self-massage on her abdomen and legs. When labour commenced, the partner stayed to provide arm and shoulder massage for relaxation or lateral sacral massage for pain relief, according to the woman’s preference. There was no time limit for massage as long as the couple was happy and felt comfortable to continue the procedure throughout the labour. The partner could take a break in times of fatigue, or when the woman fell asleep. The partners of women in the control group were also encouraged to accompany the women during labour and delivery. Women in both groups otherwise received the same intrapartum care. They received explanations concerning the effectiveness of various analgesic methods according to the ranking of reported efficacy4 7: epidural was ranked highest, followed by pethidine, then nitrous oxide and other non-pharmacological analgesia methods (including transcutaneous nerve stimulation, birthing ball, and warm pads). Women could choose various methods or a combination of methods according to their pain tolerance and acceptance, using a step-up approach or direct implementation of the most effective methods. The degree of labour pain was assessed using the visual analogue scale for pain (ranging from 0 [no pain] to 10 [most painful]) at different stages of labour: latent phase (cervical dilatation of 1-3 cm), active phase (cervical dilatation of 4-7 cm), late active phase (cervical dilatation of 8-9 cm), and second stage (cervical dilatation of 10 cm); it was also assessed when the women first requested pethidine or epidural analgesia.

The primary outcome of this study was the use of the two most effective pharmacological methods (as described above): intramuscular pethidine injection or epidural analgesia. Women were also categorised according to the type of analgesia that they eventually received: none of the analgesic methods; non-pharmacological methods only; nitrous oxide ± non-pharmacological methods; pethidine ± other pain relief except epidural; or epidural ± above methods. The proportions of women that received each type of analgesia were also compared as one of the secondary outcomes. Other secondary outcomes included intrapartum caesarean rate, duration of labour, the pain score at the point when the women first requested pethidine or epidural analgesia, the interval between the onset of labour to the time of making such a request, and the cervical dilatation at which such a request was made.

A previous study reported a reduction of 60% in the epidural rate with the use of intrapartum massage when compared with the control group.2 Therefore, our study sample size was calculated based on the assumption that the requirement for pethidine injection or epidural analgesia could be reduced by 60% (ie, from the current 15% according to Hospital Authority data to 6%) in the study group. Using an 80% power (beta) threshold and a two-tailed alpha value of 5%, we calculated that 181 participants were required in each arm. The method of calculation was obtained from the website of Department of Obstetrics and Gynaecology, the Chinese University of Hong Kong (http://www.obg.cuhk.edu.hk/ResearchSupport/StatTools/index.php). Because we anticipated that 40% of the recruited participants would be excluded (eg, because of a shift to a private hospital for delivery, change to elective caesarean section, or withdrawal from the study), we planned to recruit 300 participants for each arm.

The Chi squared test and t test were used to assess differences in baseline characteristics, obstetric outcomes, neonatal outcomes, and the proportions of women using specific pharmacological pain relief methods between the massage and control groups. Linear-by-linear association was used to assess trends regarding the use of different types of analgesia. The t test was used to compare between-group differences in the stage of labour, cervical dilatation, and pain score among participants who used pethidine/epidural, as well as the mean pain scores in different phases of labour among participants who did not use any pain relief modalities. A P value of <0.05 was considered statistically significant. All analyses used a per-protocol approach. Intention-to-treat analysis (including all participants recruited at baseline) could not be conducted because information collected during labour (eg, the use of pain relief modalities) was not available for participants who delivered in other hospitals, required caesarean section before pain labour commenced, or withdrew from the study. Statistical analyses were performed using SPSS software (Windows version 22.0; IBM Corp, Armonk [NY], United States).

Of the 1130 women eligible for this study, 528 were excluded for reasons shown in the Figure; thus, 602 women were randomised to the massage group (n=302) and control group (n=300). Furthermore, 69 (22.8%) and 54 (18.0%) women were subsequently excluded from the massage and control groups, respectively, for reasons such as delivery in private hospitals, planned caesarean section, development of complications before labour, or withdrawal from the study. Finally, 479 pregnant women (233 in the massage group and 246 in the control group) were included in the per-protocol analysis.

There were no significant differences between groups in terms of maternal age, height, or demographic characteristics nor in the proportions of women who underwent induction of labour, augmentation of labour, or delivered by caesarean section (Table 1). The mean duration of labour did not differ between the massage and control groups. No significant differences were found in gestational age at delivery, birthweight, or the proportion of babies with Apgar score ≥8 at 5 minutes (Table 1). All women in the massage group practised massage during labour (n=233). The duration of massage ranged from 35 minutes to 195 minutes (median, 100 minutes).

The proportion of women who used pethidine or epidural did not significantly differ between the massage and control groups (12.0% vs 15.9%; P=0.226). However, linear-by-linear association analysis showed a significant shift in the massage group, from using stronger analgesics (eg, epidural analgesia: 2.1% in the massage group vs 5.7% in the control group) to weaker analgesics. Thus, more women in the massage group required none of the analgesics, compared with women in the control group (29.2% vs 21.5%; P=0.041) [Table 2].

Among women who needed pethidine or epidural for pain control, there was no difference between the two groups in terms of the pain score at the point when they requested these pain relief modalities, or the interval between the onset of labour to the time of requesting these modalities (Table 3). However, the cervical dilatation at which pethidine or epidural was first requested was significantly greater in the massage group (3.8 ± 1.7 cm) than in the control group (2.3 ± 1.0 cm; P<0.001) [Table 3].

Among women who needed none of the pain relief modalities, the pain scores progressively increased with cervical dilatation, although there were no differences between the two groups (Table 4).

Although our study did not show a statistically significant reduction in the number of women who used either pethidine or epidural analgesia with the practice of massage (12.0% vs 15.9%), linear-by-linear analysis revealed that there was a statistically significant overall shift in the pattern of analgesics use in the massage group: a smaller proportion of women requested epidural analgesia (2.1% vs 5.7%) and a larger proportion of women requested none of the pain relief methods (29.2% vs 21.5%). Our results suggest that the pain perceptions of labouring women were improved by the training and practice of massage, controlled breathing, and visualisation. Thus, some women who initially requested the stronger methods (eg, epidural analgesia) might have achieved satisfactory pain control with weaker analgesic methods (eg, pethidine or nitrous oxide). Similarly, women who initially requested pethidine or nitrous oxide might have shifted to non-pharmacological methods only; this led to a greater proportion of women in the massage group who requested no analgesia.

Although pain scores are commonly used to compare analgesic effectiveness, such comparisons are often difficult on the basis of a single pain score during labour. This is because the labour process is generally long and its characteristics are variable; labouring women might use more than one method of pain relief at different stages of labour. Nonetheless, if the pain is intolerable, labouring women require a stronger analgesic method.28 Hence, we used a pattern of analgesic utility (rather than a single pain score) as an indicator for the effectiveness of the massage programme; our linear-by-linear association findings indicated that the massage programme may reduce pain perception among labouring women. Furthermore, the mean cervical dilatation at the time of pethidine or epidural analgesia request was higher in the massage group than in the control group (3.8 ± 1.7 cm vs 2.3 ± 1.0 cm). Notably, among women who requested pethidine or epidural analgesia, the pain score at the point of first pethidine or epidural analgesia request was very similar between the massage and control groups (7.6 ± 2.2 vs 7.6 ± 2.8). Although the midwives were not blinded to the allocation in this study, women in both groups received the same intrapartum care and could choose pain relief methods according to their pain tolerance and acceptance. These results further support the notion that the practice of massage might have modulated the pain perception among labouring women, such that they only requested stronger pharmacological pain relief during later phases of labour; additional studies are required to confirm the underlying biological mechanism.

Janssen et al17 also reported a delay in epidural insertion by 1 cm of cervical dilatation (5.9 cm in the massage group vs 4.9 cm in the control group) in their randomised controlled trial. However, they failed to show a significant reduction in the rate of epidural analgesia use (81.1% in the massage group vs 65.0% in the control group). Importantly, their participants only learned and practised massage at the time of labour, while our participants began learning the massage programme during the antenatal period.

In another randomised controlled trial, Levett et al21 showed that the incidence of epidural analgesia was significantly reduced (from 68.2% to 23.9%) in a cohort of 176 Australian patients. However, their control group had a baseline epidural analgesia rate of 68.2%, which was much higher than the rate in our control group (ie, 5.7%, which is similar to the 6.6% reported previously in Hong Kong29) Possible reasons for the large difference in epidural rates between Australia and Hong Kong include variations in midwifery practices, pain tolerance among labouring women, and limited resources in Hong Kong public hospitals. Other obstetric practice differences include an overall (massage and control groups combined) higher normal vaginal delivery rate in our cohort than in the cohort reported by Levett et al21 (76.4% vs 57.9%); our overall cohort also exhibited a lower instrumental delivery rate (10.9% vs 17.0%) and a lower caesarean section rate (12.7% vs 25.1%). Regardless of our low baseline epidural rate, we found a 60% reduction in epidural use in the massage group (2.1%), compared with the control group (5.7%). Finally, Levett et al21 only reported the incidence of simultaneous use of pethidine and nitrous oxide, whereas we stratified analgesic methods according to the strength of pain relief; this allowed us to detect an overall shift towards weaker analgesics among women in the massage group.

Importantly, neither study (this study or the study by Levett et al21) demonstrated any reduction in the overall duration of labour in nulliparous women, although the cohort reported by Levett et al21 exhibited a marked reduction in the rate of epidural use. In contrast, Bolbol-Haghighi et al22 showed that massage practice was associated with significantly shorter durations in both the first stage (9.0 hours vs 11.5 hours) and the second stage of labour (49 minutes vs 64 minutes) among a cohort of Iranian women. However, their study included multiparous women with an overall vaginal delivery rate over 95%; they did not describe the availability of epidural analgesia for their participants. In summary, it remains unclear whether the practice of massage has consistent effects on labour progression; the underlying mechanisms of such effects are unknown.

This study had several strengths. First, it involved a large number of nulliparous labouring women. To our knowledge, this is the largest number of such women among similar published studies; it allowed us to identify any changes in the utilisation of different levels of analgesic methods, without any confounding effects related to multiparity.20 22 Second, this study involved a team of accredited and dedicated midwife trainers, which enabled us to ensure that a consistent high-quality massage technique was applied to women in the study. Third, training at 36 weeks of gestation allowed each couple (ie, a participating woman and her partner) to have sufficient time to practise massage at home and refine their technique before the woman began labour. Fourth, on admission prior to delivery, an accredited midwife trainer was available to verify each couple’s massage technique and ensure quality. Finally, there was no limit to the duration of intrapartum massage; women could receive their preferred amount of massage to achieve optimal results.

There were some notable limitations in this study. First, because of the pain relief methods used, we were unable to incorporate blinding in the trial design. However, the midwives providing intrapartum care were not involved in data collection. Second, approximately one-fifth of the participants in each group had changes to their childbirth plan, including shift to a private hospital or to planned caesarean delivery; thus, they were excluded from the final analysis. Third, we could only assess the intrapartum massage provided by the participating women’s partners; we could not assess the breathing and visualisation practice at home, which are also essential components of the overall massage programme. Finally, because continuous foetal heart monitoring was the standard method of intrapartum foetal surveillance in Hong Kong during the study period, women were unable to move freely during labour; this restriction might have limited the ability to perform certain massage techniques. Nevertheless, the shifts towards less epidural analgesia use and higher rates of analgesic-free labour, in the absence of adverse labour outcomes, support the efficacy of our massage programme.

This study demonstrated an overall shift towards using weaker pain relief modalities among women participating in an intrapartum massage programme. The findings imply that massage, in combination with controlled breathing and visualisation, may modulate pain perception among labouring women, leading to higher rates of analgesic-free labour.

Concept or design: CY Lai.
Acquisition of data: MKW Wong, WH Tong, SY Chu, KY Lau, AML Tam.
Analysis or interpretation of data: CY Lai, LL Hui.
Drafting of the manuscript: CY Lai, TTH Lao, TY Leung.
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.

All authors have disclosed no conflicts of interest.

The authors thank Ms Linda Kimber, Midwife/Director, and Ms Mary McNabb, Scientific Advisor and Academic Midwife, both of Childbirth Essentials, Banbury, United Kingdom, for training the midwives involved in the present study and for advising the authors on the study design.

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

The study was approved by The Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (CREC Ref: 2016.332). The study has been registered at the Centre for Clinical Research and Biostatistics, The Chinese University of Hong Kong, (Unique Trial Number: CUHK_CCRB00525; https://www2.ccrb.cuhk.edu.hk/registry/public/393). All participants were informed about the nature of the study and provided written consent to participate before randomisation into study groups.

1. Kabeyama K, Miyoshi M. Longitudinal study of the intensity of memorized labour pain. Int J Nurs Pract 2001;7:46-53. Crossref

2. Rijnders M, Baston H, Schönbeck Y, et al. Perinatal factors related to negative or positive recall of birth experience in women 3 years postpartum in the Netherlands. Birth 2008;35:107-16. Crossref

3. Rachmawati IN. Maternal reflection on labour pain management and influencing factors. Br J Midwifery 2012;20:263-70. Crossref

4. Yerby M. Managing pain in labour. Part 3: pharmacological methods of pain relief. Mod Midwife 1996;6:22-5.

5. Cawthra AM. The use of pethidine in labour. Midwives Chron 1986;99:178-81.

6. Rajan L. The impact of obstetric procedure and analgesia/anaesthesia during labour and delivery on breastfeeding. Midwifery 1994;10:87-103. Crossref

7. Jones L, Othman M, Dowswell T, et al. Pain management for women in labour: an overview of systematic reviews. Cochrane Database Syst Rev 2012;(3):CD009234. Crossref

8. Anim-Somuah M, Smyth RM, Cyna AM, Cuthbert A. Epidural versus non-epidural or no analgesia for pain management in labour. Cochrane Database Syst Rev 2018;(5):CD000331. Crossref

9. Klomp T, van Poppel M, Jones L, Lazet J, Di Nisio M, Lagro-Janssen AL. Inhaled analgesia for pain management in labour. Cochrane Database Syst Rev 2012;(9):CD009351. Crossref

10. Thomson G, Feeley C, Moran VH, Downe S, Oladapo OT. 2019 Women’s experiences of pharmacological and non-pharmacological pain relief methods for labour and childbirth: a qualitative systematic review. Reprod Health 2019;16:71. Crossref

11. Chaillet N, Belaid L, Crochetière C, et al. Nonpharmacologic approaches for pain management during labor compared with usual care: a meta-analysis. Birth 2014;41:122-37. Crossref

12. Field T, Hemandez-Reif M, Taylor S, Quintino O, Burman I. Labor pain is reduced by massage therapy. J Psychosom Obstet Gynecol 1997;18:286-91. Crossref

13. McCaffery M, Beebe A. Pain: clinical manual for nursing practice. J Pain Symptom Manage 1990;5:338-9. Crossref

14. Chang MY, Wang SY, Chen CH. Effects of massage on pain and anxiety during labour: a randomized controlled trial in Taiwan. J Adv Nurs 2002;38:68-73. Crossref

15. Karami NK, Safarzadeh A, Fathizadeh N. Effect of massage therapy on severity of pain and outcome of labour in primipara. Iran J Nurs Midwifery Res 2007;12:6-9.

16. Kimber L, McNabb M, Mc Court C, Haines A, Brocklehurst P. Massage or music for pain relief in labour: a pilot randomised placebo controlled trial. Eur J Pain 2008;12:961-9. Crossref

17. Janssen P, Shroff F, Jaspar P. Massage therapy and labor outcomes: a randomized controlled trial. Int J Ther Massage Bodywork 2012;5:15-20. Crossref

18. Mortazavi SH, Khaki S, Moradi R, Heidari K, Vasegh Rahimparvar SF. Effects of massage therapy and presence of attendant on pain, anxiety and satisfaction during labour. Arch Gynecol Obstet 2012;286:19-23. Crossref

19. Silva Gallo RB, Santana LS, Jorge Ferreira CH, et al. Massage reduced severity of pain during labour: a randomised trial. J Physiother 2013;59:109-16. Crossref

20. Mc Nabb MT, Kimber L, Haines A, McCourt C. Does regular massage from late pregnancy to birth decrease maternal pain perception during labour and birth? A feasibility study to investigate a programme of massage, controlled breathing and visualization, from 36 weeks of pregnancy until birth. Complement Ther Clin Pract 2006;12:222-31. Crossref

21. Levett KM, Smith CA, Bensoussan A, Dahlen HG. Complementary therapies for labour and birth study: a randomised controlled trial of antenatal integrative medicine for pain management in labour. BMJ Open 2016;6:e010691. Crossref

22. Bolbol-Haghighi N, Masoumi SZ, Kazemi F. Effect of massage therapy on duration of labour: a randomized controlled trial. J Clin Diagn Res 2016;10:12-5. Crossref

23. Taghinejad H, Delphisheh A, Suhrabi Z. Comparison between massage and music therapies to relieve the severity of labour pain. Womens Health (Lond) 2010;6:377-81. Crossref

24. Smith CA, Levett KM, Collins CT, Dahlen HG, Ee CC, Suganuma M. Massage, reflexology and other manual methods for pain management in labour. Cochrane Database Syst Rev 2018;(3):CD009290. Crossref

25. Hui AS, Lao TT, Leung TY, Schaaf JM, Sahota DS. Trends in preterm birth in singleton deliveries in a Hong Kong population. Int J Gynaecol Obstet 2014;127:248-53. Crossref

26. The Royal College of Midwives. Childbirth essentials towards natural childbirth and beyond. Available from: https://www.rcm.org.uk/promoting/learning-careers/ accredited-learning/childbirth-essentials-towards-natural-childbirth-and-beyond/. Accessed 19 Feb 2020.

27. Fuchs AR, Behrens O, Liu HC. Correlation of nocturnal increase in plasma oxytocin with a decrease in plasma estradiol/progesterone ratio in late pregnancy. Am J Obstet Gynecol 1992;167:1559-63. Crossref

28. Tsui MH, Ngan Kee WD, Ng FF, Lau TK. A double blinded randomised placebo-controlled study of intramuscular pethidine for pain relief in the first stage of labour. BJOG 2004;111:648-55. Crossref

29. Hong Kong College of Obstetricians and Gynaecologists. Report of the territory-wide audit in obstetrics & gynaecology. 2014. Available from: https://www.hkcog.org.hk/hkcog/pages_4_307.html. Accessed 19 Feb 2020.

There has been a gradual decline in the global tuberculosis (TB) incidence and mortality rates over time, by approximately 2% and 3% per year, respectively.1 Despite this trend, TB remains a leading cause of death, and Mycobacterium tuberculosis drug resistance continues to be a major public health issue. Globally, isoniazid (INH)-resistant, rifampicin (RIF)-susceptible TB (Hr-TB) is the most common drug-resistant form of disease.2 In 2017, the rates of Hr-TB were 7.1% among patients with new TB diagnoses and 7.9% in patients who had previously undergone treatment for TB.1 In Hong Kong, the respective rates were 5.3% and 9.5%, with a combined rate of 5.7% in 2016.3 The Hr-TB is associated with worse clinical outcome and development of multidrug resistance (MDR),4 5 although the findings have been inconsistent.6 7

Isoniazid constitutes a key component in the treatment of drug-susceptible TB, through its inhibition of mycolic acid biosynthesis in the mycobacterial cell wall. Over 85% of INH resistance is conferred by mutations residing in the katG gene and inhA promoter region,8 leading to high and low levels of resistance, respectively. These mutations can readily be detected by commercial molecular line-probe assays. The level of resistance can also vary according to the co-occurrences of less common mutations in other regions.

For the treatment of Hr-TB, current World Health Organization (WHO) guidelines recommend 6 months of RIF, pyrazinamide (PZA), ethambutol, and levofloxacin (LVX). The guidelines also recommend examination of resistances towards fluoroquinolones and PZA, prior to treatment.9

In our locality, routine testing of LVX and PZA susceptibility has not been implemented for the treatment of Hr-TB. To determine the most cost-effective testing strategy, this study reviewed an annual collection of Hr-TB isolates, with the aim of determining the prevalences of LVX resistance and pncA mutations (as a molecular marker of PZA resistance in Hr-TB) and identifying factors associated with these resistances.

Data were reviewed from the TB Reference Laboratory of the Department of Health. This laboratory processes local M tuberculosis strains from specimens collected in out-patient clinics and in-patient hospitals in both public and private sectors. All phenotypic Hr-TB isolates in 2018 were identified and de-duplicated using unique patient identifiers. The following data were included in this analysis: basic patient demographics, phenotypic susceptibility results of first-line drugs (INH at critical concentrations 0.1 μg/mL and 0.4 μg/mL as recommended by the WHO, RIF, streptomycin, and ethambutol), and genotypic susceptibility results (inhA promoter region, katG codon 315, and rpoB hotspot region) from the GenoType MTBDRplus assay, version 2.0 (Hain Lifescience). Any discrepant or unsuccessful test results were resolved by the agar proportion method and/or DNA sequencing; when applicable, these additional data were reported as the final results. All available data from the Laboratory Information System were reviewed for each patient to determine any history of TB, fluorescence microscopy results (according to Global Laboratory Initiative grading10), site of isolation (pulmonary versus extrapulmonary), any documented sputum culture conversion and the duration (ie, time between the date of first positive culture for current infection and the date of consistently negative culture), and microbiological outcome.

Selected M tuberculosis strains were retrieved from the laboratory archive and sub-cultured, then subjected to LVX susceptibility testing using the BD BACTEC MGIT 960 system (Becton, Dickson and Company), in accordance with the manufacturer’s instructions. The LVX critical concentration at 1.0 μg/mL was used as recommended by the WHO. DNA extraction was performed using GenoLyse (Hain Lifescience). The pncA gene was amplified with the primers pncA-1F (5′-CGCTCAGCTGGTCATGTTC-3′) and pncA-1R (5′-CCCACCGGGTCTTCGAC-3′) to produce an amplicon of 798 bp, using the GeneAmp PCR System 9700 (Applied Biosystems). Each 25-μL reaction mixture contained 12.5 μL of GoTaq G2 Hot Start Colorless Master Mix (Promega) [1×], 0.25 μL of each primer (1.0 pM/μL), 9.5 μL PCR-grade water, and 2.5 μL of template DNA. Reaction mixtures were amplified using the following protocol: 2 minutes at 95°C; 35 cycles of 1 min at 95°C, 1 minute at 65°C, and 1 minute at 72°C; and 10 minutes at 72°C. Sequencing was performed using the 3730 × l DNA Analyzer (Applied Biosystems) with the same primers. Resulting sequences were compared with the sequence of wild-type M tuberculosis H37Rv for detection of pncA mutations. Strains with detected pncA mutations were subjected to further analysis of PZA susceptibility via the MGIT 960 system in accordance with the manufacturer’s instructions, with a critical concentration of 100 μg/mL.

Univariate analysis comprised odds ratio calculations and Fisher’s exact test. Firth logistic regression was employed for multivariable analysis because of the low outcome frequency. IBM SPSS Statistics Subscription (Windows version, IBM Corp, Armonk [NY], United States) was used for data analysis. A P value of <0.05 was considered statistically significant. This article complies with the STROBE Statement reporting guidelines.

In total, 8865 M tuberculosis isolates from 3411 patients were processed in 2018. Susceptibility profiles were available for all except repeated strains collected from the same patient within 3 months. De-duplication of 393 isolates yielded 160 patients with phenotypic Hr-TB, amounting to a case rate of 4.7%. rpoB hotspot mutations were identified in five cases by GenoType MDRTBplus, three of which were confirmed to confer RIF resistance according to rpoB sequencing results. These three cases were considered to be RIF-resistant and excluded from further analysis. The mean age of the patients in the remaining 157 cases was 61.3 years (range, 15-95 years); the male to female ratio was 1.8. Pulmonary TB was present in 90.4% of affected patients (n=142), while 9.6% of affected patients (n=15) had extrapulmonary involvement. There was a documented history of TB by positive culture in 4.5% of affected patients (n=7). Microscopy results were available for cases in which direct specimens were received by our laboratory at the time of initial diagnosis (n=45). The majority of specimens was acid-fast bacillus smear-negative (n=33), followed by 1-4 acid-fast bacilli per length (n=4), scanty (n=4), 1+ (n=2), and 2+ and 3+ (n=1 each). In terms of microbiological outcome, 76.4% of affected patients (n=120) had documented sputum culture conversion without recurrence after follow-up of at least 9 months, among which 106 attained conversion within 5 months after diagnosis. The median interval required for culture conversion was 72.5 days (range, 9-622 days). No clearance was documented for patients in the remaining 37 cases, for whom no follow-up specimens were received for >1 year.

Table 1 shows the patterns of resistance among first-line drugs and distributions of cases with LVX resistance and pncA mutations. Table 2 shows the patterns of mutations detected. With respect to INH, 45.2% (n=71) of the isolates exhibited low-level resistance (minimum inhibitory concentration 0.1-0.4 μg/mL) and 54.8% (n=86) of the isolates exhibited high-level resistance (minimum inhibitory concentration >0.4 μg/mL).

For LVX, only one strain was resistant, while 155 were sensitive (one isolate failed to be recovered). The number of resistant cases was insufficient for correlation analysis. pncA mutations were detected in 4.5% (7/157 cases), and all detected mutations were previously identified as PZA resistance–related. All isolates with pncA mutations were phenotypically resistant to PZA. Univariate analysis showed that pncA mutations were associated with documented history of TB in the affected patient (odds ratio [OR]=11.60, 95% confidence interval [CI]=1.79-75.00; P=0.03) and INH poly-resistance (OR=19.06, 95% CI=1.07-339.78; P=0.04) but not mono-resistance, as shown in Table 3. In multivariable logistic regression, pncA mutations were associated with documented history of TB in the affected patient (OR=18.03, 95% CI=2.25-153.85; P=0.008), INH triple resistance (OR=409.11, 95% CI=6.52 to >1000; P<0.001), and INH double resistance (OR=13.50, 95% CI=1.43 to >1000; P=0.019).

In this study, the frequency of LVX resistance was very low in Hr-TB. Levofloxacin is an important drug in the management of drug-resistant TB. In a 2009 study in Shanghai, Xu et al11 estimated the rates of LVX resistance to be 1.9% in strains pan-susceptible to first-line drugs, 6.7% in INH mono-resistant strains, and ≤25% in MDR-TB. Independent risk factors associated with LVX resistance included MDR, RIF mono-resistance, poly-resistance (resistance to ≥2 first-line drugs, but not MDR), age ≥46 years, and TB re-treatment, but not INH mono-resistance. These findings were consistent with the results of a 2017 study in Ningbo (a city near Shanghai), whereby Che et al12 revealed no LVX resistance in strains pan-susceptible to first-line drugs, 3% in strains with any resistance to first-line drugs except MDR, and ≤30% in MDR-TB. Most fluoroquinolone resistance was present in the MDR group. Che et al12 also found that prevalence of LVX resistance in MDR-TB increased with duration of inappropriate treatment before LVX; this relationship was stronger than any relationships with previous fluoroquinolone exposures. Further studies are needed to determine whether this finding also applies to Hr-TB. Levofloxacin resistance is uncommon in non-MDR strains, especially in the present study. Since 2015, our laboratory has performed LVX susceptibility testing on 640 Hr-TB strains, identifying only two resistant cases (0.31%; unpublished data), including the one in this study. Clinicians could treat affected patients empirically, in accordance with WHO recommendations; further testing could be arranged based on clinical, radiological, and microbiological responses during early follow-up. In patients with Hr-TB that exhibits LVX resistance or poly-resistance to other primary agents, individualised adjustment of the treatment regimen is recommended in accordance with WHO guidelines9 (eg, exclusion of LVX or inclusion of second-line TB medicines). In our single case with LVX resistance, the patient had no history of TB. His sputum acid-fast bacillus smear result was 2+. The Hr-TB strain isolated was initially LVX-sensitive without mutations in rpoB, gyrA, or gyrB on diagnosis and follow-up at 8 months. The patient continued to exhibit sputum culture-positive results after 15 months of treatment; further analysis revealed that the isolate had acquired gyrA mutation (detected by line-probe assays) and LVX phenotypic resistance. It then developed into an MDR strain; the patient eventually achieved culture conversion at approximately 18 months after diagnosis. Underlying hetero-resistance was a possible contributing factor.

Phenotypic PZA susceptibility testing has often been problematic. pncA mutations constitute the most common and primary determinant of PZA resistance, with reported sensitivity of approximately 80% to 95% and specificity of approximately 85% to 100%.13 14 Resistance-conferring mutations are known to be diverse and scattered over the gene’s full length. Our seven isolates all had unique mutations (Gln10Arg, His51Tyr, Trp68Stop, Thr47Pro, Ala102Thr, Asp63Gly, and Val139Ala), which were associated with PZA resistance at a high probability of 0.985 according to available literature15 (except Val139Ala, probability of resistance=0.783). These seven isolates were also confirmed to be phenotypically resistant to PZA in our study.

Thus far, investigations of phenotypic and genotypic PZA resistance have generally focused on MDR-TB. Concerning non-MDR-resistant strains, there is considerable geographical variation in the reported rates, from 2.92% to 4.8% in the United States,16 17 to 24% in the Western Pacific and 75% in South East Asia.18 For Hr-TB in particular, phenotypic PZA resistance was reported in 2% of INH mono-resistant strains in South Africa,19 no PZA resistance was detected (phenotypic or pncA mutation) in Georgia (Eastern Europe) among Hr-TB strains,20 and ≤14.9% of Hr-TB strains exhibited pncA mutations in Vietnam.21 In our study, we observed that 4.5% of Hr-TB isolates had pncA mutations. Such variation may be related to multiple factors, including differences in inclusion criteria, testing method, treatment, infection control practice, and disease burden. In terms of risk factors, we found that history of TB in the affected patient and poly-resistance (but not mono-resistance) were significantly associated with pncA mutations. These results are consistent with findings from previous studies, which also showed an association with MDR.21 22 23 Clinicians are advised to consider the possibility of PZA resistance in patients with Hr-TB and these risk factors. Routine testing for LVX or PZA susceptibility in patients with INH mono-resistant TB alone (ie, no other risk factors) is not considered cost-effective, based on our findings.

In our study, most patients with Hr-TB had a satisfactory microbiological outcome without recurrence. However, the outcome was unknown in 23.6% of cases. Most of these involved patients were diagnosed and managed in hospitals, where only positive isolates were subjected to reference testing at our laboratory. Based on the recommended management regimen for TB in Hong Kong, these patients would have been followed up until sputum culture conversion. Considering the time elapsed since the latest positive laboratory result, the patients in most cases with unknown outcomes likely already had culture conversion.

In general, Hr-TB is presumably associated with higher rates of treatment failure and MDR acquisition, compared with drug-susceptible strains.4 5 Important considerations for disease control include recognition of risk factors associated with Hr-TB (eg, history of TB treatment, incarceration in prisons, and homelessness7 24), early detection of INH resistance, exclusion of RIF resistance by appropriate molecular methods (eg, line-probe assays),2 and compliance with current treatment guidelines. With the maturation of whole-genome sequencing, the technology is becoming integrated into the local routine drug susceptibility testing protocol for all M tuberculosis isolates; multiple and uncommon molecular markers of drug resistance might thus be identified earlier in a single set of assays for proper treatment guidance.

Because our laboratory serves as a local TB reference laboratory, this study was able to use a representative M tuberculosis collection that reflected the status of Hr-TB in Hong Kong. Furthermore, because our laboratory serves as a supranational reference laboratory, we were able to perform an array of confirmatory tests in accordance with WHO recommendations, thus ensuring reliable results. However, this study had some limitations. First, it lacked data regarding microscopy results, drug treatment, and clinical outcome, thus preventing a more comprehensive assessment. Second, the small sample size and low outcome frequencies of LVX and PZA non-susceptibilities led to limited correlation analysis. Third, less common mutations mediating INH resistance were not examined to determine their prevalences; Hr-TB cases with these mutations would only be identified on the basis of phenotypic results. A substantial proportion of cases (22.3%; n=35) had negative GenoType MDRTBplus results. Because this test only detects the most common mutations at the inhA promoter region and katG codon 315, other relevant mutations might have been missed. Furthermore, in 3.2% (n=5) of cases with high-level INH resistance, mutations were detected in the inhA promoter region alone. The availability of preliminary information before receiving the INH phenotypic susceptibility results might lead to mismanagement of patient treatment (eg, INH inclusion in the therapeutic regimen). Fourth, low-level RIF resistance mediated by mutations outside the rpoB hotspot region and undetected by phenotypic testing was not ruled out in this study. Inclusion of such isolates might have led to overestimation of true resistance in Hr-TB. Fifth, PZA resistance could be mediated by other less common mutations, which were not examined by performing phenotypic PZA susceptibility testing on all Hr-TB isolates; thus, we may have underestimated its prevalence. Our laboratory has found that all PZA-resistant M tuberculosis isolates thus far could be confirmed through pncA mutation analysis. Nevertheless, we presume that these limitations do not greatly affect the reliability and overall interpretation of our findings.

In Hong Kong, the rate of Hr-TB was 4.7% among active TB cases in 2018. Levofloxacin and PZA resistances among Hr-TB were uncommon (0.6% and 4.5%, respectively). History of TB in the affected patient and INH poly-resistance, including both double and triple resistances, were independent risk factors for pncA mutations. The findings from this study do not support routine susceptibility testing for these two drugs in patients with INH mono-resistant TB who lack additional risk factors.

Concept or design: KKM Ng.
Acquisition of data: All authors.
Analysis or interpretation of data: All authors.
Drafting of the manuscript: KKM Ng.
Critical revision of the manuscript for important intellectual content: KKM Ng.

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.

All authors have disclosed no conflicts of interest.

The authors acknowledge the excellent work and contribution by staff, especially Mr Steven CW Lui and Ms Angela WL Lau, at the Tuberculosis Laboratory and Special Investigation Laboratory of Public Health Laboratory Services Branch, Centre for Health Protection, Department of Health, Hong Kong SAR Government.

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

Ethics approval for this study was obtained from the Ethics Committee of the Department of Health, Hong Kong SAR Government (Ref: LM 425/2019). All included patients consented to testing for tuberculosis.

1. World Health Organization. Global tuberculosis report 2018. Available from: https://www.who.int/tb/publications/global_report/en/. Accessed 19 Sep 2019.

2. Romanowski K, Campbell JR, Oxlade O, Fregonese F, Menzies D, Johnston JC. The impact of improved detection and treatment of isoniazid resistant tuberculosis on prevalence of multi-drug resistant tuberculosis: a modelling study. PLoS One 2019;14:e0211355. Crossref

3. Tuberculosis & Chest Service of Department of Health, Hong Kong SAR Government. Annual report 2016. Available from: https://www.info.gov.hk/tb_chest/doc/Annual_Report_2016.pdf. Accessed 19 Sep 2019.

4. Gegia M, Winters N, Benedetti A, van Soolingen D, Menzies D. Treatment of isoniazid-resistant tuberculosis with first-line drugs: a systematic review and meta-analysis. Lancet Infect Dis 2017;17:223-34. Crossref

5. Karo B, Kohlenberg A, Hollo V, et al. Isoniazid (INH) mono-resistance and tuberculosis (TB) treatment success: analysis of European surveillance data, 2002 to 2014. Euro Surveill 2019;24:1800392. Crossref

6. Salindri AD, Sales RF, DiMiceli L, Schechter MC, Kempker RR, Magee MJ. Isoniazid monoresistance and rate of culture conversion among patients in the State of Georgia with confirmed tuberculosis, 2009-2014. Ann Am Thorac Soc 2018;15:331-40. Crossref

7. Cattamanchi A, Dantes RB, Metcalfe JZ, et al. Clinical characteristics and treatment outcomes of isoniazid-monoresistant tuberculosis. Clin Infect Dis 2009;48:179-85. Crossref

8. Seifert M, Catanzaro D, Catanzaro A, Rodwell TC. Genetic mutations associated with isoniazid resistance in Mycobacterium tuberculosis: a systematic review. PLoS One 2015;10:e0119628. Crossref

9. World Health Organization. WHO consolidated guidelines on drug-resistant tuberculosis treatment. Available from: https://www.who.int/tb/publications/2019/consolidated-guidelines-drug-resistant-TB-treatment/en/. Accessed 19 Sep 2019. Crossref

10. Global Laboratory Initiative. Mycobacteriology Laboratory Manual. 1st ed. Global Laboratory Initiative; 2014

11. Xu P, Li X, Zhao M, et al. Prevalence of fluoroquinolone resistance among tuberculosis patients in Shanghai, China. Antimicrob Agents Chemother 2009;53:3170-2. Crossref

12. Che Y, Song Q, Yang T, Ping G, Yu M. Fluoroquinolone resistance in multidrug-resistant Mycobacterium tuberculosis independent of fluoroquinolone use. Eur Respir J 2017;50:1701633. Crossref

13. Streicher EM, Maharaj K, York T, et al. Rapid sequencing of the Mycobacterium tuberculosis pncA gene for detection of pyrazinamide susceptibility. J Clin Microbiol 2014;52:4056-7. Crossref

14. Khan MT, Malik SI, Ali S, et al. Pyrazinamide resistance and mutations in pncA among isolates of Mycobacterium tuberculosis from Khyber Pakhtunkhwa, Pakistan. BMC Infect Dis 2019;19:116. Crossref

15. Miotto P, Cabibbe AM, Feuerriegel S, et al. Mycobacterium tuberculosis pyrazinamide resistance determinants: a multicenter study. mBio 2014;5:e01819-4.Crossref

16. Kurbatova EV, Cavanaugh JS, Dalton T, Click ES, Cegielski JP. Epidemiology of pyrazinamide-resistant tuberculosis in the United States, 1999-2009. Clin Infect Dis 2013;57:1081-93. Crossref

17. Budzik JM, Jarlsberg LG, Higashi J, et al. Pyrazinamide resistance, Mycobacterium tuberculosis lineage and treatment outcomes in San Francisco, California. PLoS One 2014;9:e95645. Crossref

18. Whitfield MG, Soeters HM, Warren RM, et al. A global perspective on pyrazinamide resistance: systematic review and meta-analysis. PLoS One 2015;10:e0133869. Crossref

19. Whitfield MG, Streicher EM, Dolby T, et al. Prevalence of pyrazinamide resistance across the spectrum of drug resistant phenotypes of Mycobacterium tuberculosis. Tuberculosis (Edinb) 2016;99:128-30. Crossref

20. Sengstake S, Bergval IL, Schuitema AR, et al. Pyrazinamide resistance-conferring mutations in pncA and the transmission of multidrug resistant TB in Georgia. BMC Infect Dis 2017;17:491. Crossref

21. Huy NQ, Lucie C, Hoa T, et al. Molecular analysis of pyrazinamide resistance in Mycobacterium tuberculosis in Vietnam highlights the rate of pyrazinamide resistance-associated mutations in clinical isolates. Emerg Microbes Infect 2017;6:e86. Crossref

22. Li D, Hu Y, Werngren J, et al. Multicenter study of the emergence and genetic characteristics of pyrazinamide-resistant tuberculosis in China. Antimicrob Agents Chemother 2016;60:5159-66. Crossref

23. Chiu YC, Huang SF, Yu KW, Lee YC, Feng JY, Su JY. Characteristics of pncA mutations in multidrug-resistant tuberculosis in Taiwan. BMC Infect Dis 2011;11:240. Crossref

24. Smith CM, Trienekens SC, Anderson C, et al. Twenty years and counting: epidemiology of an outbreak of isoniazid-resistant tuberculosis in England and Wales, 1995 to 2014. Euro Surveill 2017;22:30467. Crossref