Miscarriage is the most common early pregnancy complication, which constitutes a large burden for patients and the overall healthcare system. Approximately one in four women experiences early pregnancy loss during her lifetime; such losses have significant negative psychological and social impacts on affected women.1 2 3 4 Miscarriage has been ranked the second most common diagnosis for admissions in the past 10 years in Hong Kong.5 In-patient hospital admissions for miscarriage were considerably reduced following establishment of out-patient early pregnancy assessment clinics (EPACs) in various hospitals. Women visit EPACs to seek reassurance that their pregnancy remains viable, which may be difficult because gestational sac size and embryonic growth are not uniform6; moreover, estimates of gestational age are misleading in women with irregular menstrual periods.7 8 A score that signifies the likelihood of a viable pregnancy may be helpful for clinicians working in EPACs.9 10 This score would enable clinicians to target appropriate early psychological or clinical support, thereby minimising psychological morbidity, especially for patients with finite resources.

Probability models to predict pregnancy viability have been reported; these combine clinical history, sonographic assessment of gestational sac and heart pulsation, and biochemical measurements.3 9 11 12 13 However, some models cannot be utilised in initial counselling, because they require calculators or biochemical measurements which may not yet be available.9 12 13 In a recent systematic review regarding prediction of miscarriage in women with viable intrauterine pregnancy, it was found that many combinations of markers have been tested with varying diagnostic accuracy; however, no meta-analysis could be performed on combination models.14 Furthermore, the meta-analysis did not involve assessment of the proportions of women with intrauterine pregnancy of uncertain viability, and no scoring system or cut-off value could be derived for counselling.

The Bottomley score11 is a scoring system independent of biochemical measurements, which utilises clinical and ultrasound parameters, including maternal age, severity of bleeding, and ultrasound features (eg, mean sizes of gestational sac and yolk sac, as well as presence of fetal heart pulsation). It has been validated in comparison with more complicated probability-based models in women with intrauterine pregnancy of uncertain viability; however, the validation study was limited by small sample size and the exclusion of one third of the eligible women due to missing variables and absence of data regarding pregnancy outcomes.12 The scoring system was also limited by the absence of data regarding body mass index (BMI) and smoking status, which might affect the likelihood of viable pregnancy.11 Furthermore, ethnicity influences miscarriage risk—black women are reportedly more likely to miscarry than white women,15 whereas rates among South and East Asian women are reportedly similar to those of white women after adjustment for confounders.16 The objective of the present study was to assess and validate the Bottomley score for prediction pregnancy viability until 16 weeks of gestation in a cohort of Chinese pregnant women who presented to our hospital with threatened miscarriage or abdominal pain before 12 weeks of gestation.

This non-interventional prospective observational study was performed at the Prince of Wales Hospital, Hong Kong, between July 2013 and June 2015. An out-patient EPAC is available in this hospital to receive referrals of first trimester pregnant women with vaginal bleeding, abdominal pain, or both, and suspected threatened miscarriage, threatened miscarriage with uncertain viability, and/or abdominal pain complicating intrauterine pregnancy; referrals were made by general practitioners or accident and emergency medical officers. All gynaecologists at the EPAC had ≥3 years of experience in ultrasound scans, as well as in diagnosis and treatment of miscarriage.

For this study, Chinese women aged ≥18 years with a singleton intrauterine pregnancy, referred before 12 weeks of gestation based on last menstrual period, were invited to participate. Participants provided demographic data for both standard clinical treatment and determination of the pregnancy viability score. Women were excluded if they underwent pregnancy termination, had an ectopic or multiple pregnancy, had pregnancy at an unknown location, or were diagnosed with miscarriage at the time of initial presentation. Women with intrauterine pregnancy of uncertain viability underwent a second ultrasound examination after 7 to 14 days to determine fetal viability, in accordance with published guidelines.17 18 19

Detailed information regarding obstetrics history and history of the current pregnancy were obtained, including abdominal pain (graded by pain score) and vaginal bleeding (assessed using a pictorial blood loss chart with number of pads used; no bleeding was regarded as a score of 0, while clots or flooding was regarded as a score of 4). Information regarding smoking status (smoker or non-smoker), alcohol intake, and BMI were collected. Smokers included women who continued to smoke, as well as those who had discontinued smoking ≤2 weeks before presentation to our hospital.20 Alcohol users included women who consumed >2 units per day, once or twice per week, in the month before and during their pregnancy.21 Body mass index was classified in accordance with international classification as underweight, normal, overweight, or obese.22

All women underwent a structured ultrasound assessment. All transvaginal ultrasound scans were performed using a GE Voluson 730 ultrasound machine (GE Healthcare, Zipf, Austria) to ascertain the location and viability of the pregnancy. Mean gestational sac diameter, mean yolk sac diameter, size of fetal pole, and presence of fetal heart pulsation were documented. The Royal College of Obstetricians and Gynaecologists guidelines17 18 and National Institute for Health and Care Excellence guidelines19 were used for diagnosis of miscarriage, intrauterine pregnancy of uncertain viability, viable pregnancy, ectopic pregnancy, or pregnancy of unknown location. In women with a history of vaginal bleeding, a hypoechoic or anechoic crescent-shaped area on ultrasound images was regarded as a subchorionic haematoma; its three-dimensional size was classified as small, medium, or large when its size ratio (relative to gestational sac size) was <0.2, 0.2-0.5, or >0.5, respectively.

Each pregnancy was assigned a Bottomley score based on clinical history and ultrasound parameters.11 All clinicians were blinded to the score and all women in this study were treated in accordance with our established standard clinical protocols. Pregnancies were categorised as viable or miscarriage at the repeat ultrasound scan conducted between 13 and 16 weeks of gestation, according to the presence or absence of fetal heart pulsation. Women with miscarriage were treated in accordance with current guidelines.17 19 Watchful waiting, or medical or surgical evacuation of the uterus, were offered according to each patient’s clinical condition. For medical evacuation, misoprostol 800 μg vaginally was used as first-line treatment, with follow-up assessment to check for complete evacuation. Patient treatment was not affected by participation in the study.

Ethics approval was obtained from the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CREC-2013.348). Written informed consent was obtained from all participants.

Bottomley and colleagues11 reported that the areas under the receiver operating characteristic (ROC) curves of their score were 0.90 and 0.72 using a combination of history and ultrasound parameters and history alone, respectively. To determine whether the Bottomley score in our local population exceeded the discriminatory power of the history-alone model and achieved discriminatory power similar to that of the combined model, sample size analysis, using MedCalc Statistical Software version 18.5 (MedCalc Software bv, Ostend, Belgium), showed that a minimum sample size of 750 was required for a type 1 error of 1% and power of 90%, with the assumption that miscarriage occurred in one of every 10 pregnancies. The planned sample size was increased to 1500 to allow for the worst-case scenario of 50% non-participation rate and 50% loss to follow-up rate.

Women were divided into viable or miscarriage groups according to pregnancy outcome. Comparisons of socio-demographic and pregnancy characteristics between the two outcome groups were performed using the Chi squared test or Fisher’s exact test, where appropriate, for categorical variables; the Mann-Whitney U test with post hoc Bonferroni correction was used for comparisons of continuous variables. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) of significant predictors of miscarriage were determined by multivariate logistic regression analysis. Receiver operating characteristic curves were constructed to determine the discriminatory performance of the Bottomley score, as well as to determine the Bottomley score which predicted 90% viable pregnancies at the repeat ultrasound scan. Probit regression, with Bottomley score as the only independent predictor, was performed to estimate the probability of an ongoing viable pregnancy beyond the first trimester. Bottomley scores were truncated to a minimum value of -12 and a maximum value of 18, prior to Probit regression and ROC analyses. Patients lost to follow-up were excluded from the study. All analyses were performed using SPSS Statistics, version 20.0 (IBM Corp., Armonk [NY], United States) and MedCalc. P<0.05 was considered statistically significant.

Of the 1508 women invited to participate, 54 (3.6%) declined and two (0.1%) were lost to follow-up at 13 to 16 weeks of gestation (Fig 1). Table 1 summarises the socio-demographic characteristics, clinical signs and symptoms, and ultrasound measurements in the two outcome groups. The Bottomley score ranged from -41 to 24; 36 women (2.8%) had a score of ≤-12, and 64 (5.0%) had a score ≥18, while the score ranged from -12 to +1 in women with a pregnancy of uncertain viability at first assessment.

Unadjusted and adjusted ORs for risk of miscarriage after adjusting for covariates are summarised in Table 2. Subchorionic haematoma was noted in 167 pregnancies (13.1%), including 138 (82.6%) with bleeding. Haematoma size relative to gestational sac size was significantly associated with miscarriage (χ²=70.5, P<0.05) [Tables 1 and 2]. A pregnancy with a large haematoma was nearly two-fold more likely to miscarry, compared with other pregnancies (17/160 [10.6%] vs 66/1111 [5.9%]).

The area under the ROC curve of the discriminatory performance of the Bottomley score in all women was 0.91 (95% CI: 0.89-0.93, P<0.001). A score of ≥1 had a sensitivity of 91% (95% CI: 85.8-95.1%) and a false positive rate of 26.7% (95% CI: 24.1-29.4%). Figure 2 and Table 3 show the observed and estimated viability of a pregnancy after the first trimester, compared with the viability estimates by Bottomley et al11 at each viability score. The estimated probability of viability for a particular pregnancy, based on the Bottomley score, was determined: Probit(p) = 1.15109 + 0.17188 × score.

The estimated probability of viability reported by Bottomley and colleagues11 was within the 95% CI of the estimated probability determined by our Probit(p) function if the viability score was ≥0. The area under the ROC curve of the Bottomley score in women with a pregnancy of uncertain viability at initial presentation was only 0.74 (95% CI: 0.67-0.80).

To the best of our knowledge, this is the first report of independent external validation of the scoring system proposed by Bottomley et al.11 to predict early second trimester pregnancy viability in women with intrauterine pregnancy before 12 weeks of gestation. Moreover, it is the first study in a homogenous Chinese population. Our findings indicated that the Bottomley score could be used to predict the likely outcome of pregnancy, thus potentially alleviating maternal anxiety; this is particularly useful for women with symptoms of threatened miscarriage. The scoring system is simple to use and does not require a calculator (in contrast to previous models)8 11; moreover, it relies solely on information that can be readily obtained by any gynaecologist, without the need for blood tests.3 12 Women with a Bottomley score of ≥1 had a >90% probability of pregnancy viability beyond the first trimester. These women could be reassured, and further ultrasounds could be avoided. A low Bottomley score was associated with increased likelihood of miscarriage, such that half of the women with a score of -7 or -6 were expected to miscarry; this proportion reached 80% if the score was ≤-12. Proper counselling could be offered to prepare these women psychologically, thereby reducing the impact of pregnancy loss.

This scoring system incorporates several different variables that interact with each other. For example, a larger but empty gestational sac increases the likelihood of miscarriage, thus resulting in a lower (ie, more negative) score; however, this lower score would be counterbalanced by the presence of fetal heart pulsation and an appropriately sized yolk sac. In addition, although a negative score was unexpectedly determined for pregnancies with the presence of fetal heart pulsation, this negative score could be counterbalanced by a positive score for a larger gestational sac size.

Notable strengths in this study include its use of a priori determination of sample size to assess discriminatory performance. Moreover, the participation rate was high and few pregnancies were lost to follow-up. The resulting large sample size enabled assessment of discriminatory performance of the scoring system; it also allowed identification of predictors of miscarriage in our cohort of Chinese women and provided estimated sensitivities (±5%) at specific viability scores. Whereas only 10% of included patients were of Asian ethnicity in the study by Bottomley et al,11 our study was performed in a homogenous Chinese population; therefore, our findings suggest that the score is likely to be valid for various Asian populations, although further studies are necessary to validate its use in different Asian subgroups.

This scoring system was designed for use in all pregnant women with an intrauterine pregnancy before 12 weeks of gestation. Our analysis of the performance of this scoring system in pregnancies with uncertain viability alone also showed reasonable performance: pregnancy failure could be predicted in women with pregnancies of uncertain viability, with an area under the ROC curve of >0.5. However, this finding should be interpreted with caution, because there were only 223 women in this subgroup; our sample size was only sufficient to detect an area under the ROC curve of 0.65, assuming that the ratio of miscarriage to viability was 1:1. Of note, while the scoring system was reliable for estimation of pregnancy viability until 16 weeks of gestation, the implication of each score differed, compared with the previous study.11

Lastly, the miscarriage rate in this study was 13%, approximately 50% lower than the 20% to 30% rates reported by Kong et al1 and Bottomley et al.11 However, the observed rate of miscarriage among women with intrauterine pregnancy of uncertain viability at presentation was consistent with the rates observed in other studies.11 12 The relatively low overall miscarriage rate could have been due to differences in local referral practices, because some women attending our EPAC were asymptomatic, whereas women with heavy vaginal bleeding or severe abdominal pain might have been admitted directly for treatment; we were unable to ascertain how many women with early pregnancy loss were directly admitted without referral to the EPAC. We excluded women with ectopic pregnancy or pregnancy of unknown location because these women had empty uteri. By focusing on women with intrauterine pregnancy irrespective of viability, we presumed that our study would be more likely to generate useful clinical information for counselling if the model were validated. Notably, excluded women comprised only 0.9% of all women invited to participate in the study.

Other potential explanations for the low miscarriage rate could be differences in lifestyle factors, such as smoking and alcohol consumption, as well as incidence of obesity; in contrast to the findings in published literature,23 24 these factors were not associated with pregnancy outcome. Smoking during pregnancy is rare in Chinese women.25 Obesity is also uncommon; in the present study only 35 women (2.7%) had a BMI ≥30 kg/m², while the median weight of 53.3 kg and median BMI of approximately 21-21.5 kg/m² in this study were similar to the characteristics of women attending a first trimester Down syndrome screening clinic and of pregnant women enrolled in another prospective study (regarding the pelvic floor) in our centre.26 27 Lastly, the median duration of gestation at presentation to EPAC was 55 days in our study, whereas it was 50 days in the study by Bottomley et al.11 This could have contributed to our lower miscarriage rate, which decreases with gestational age.

Consistent with the findings of other studies, our multivariate analysis indicated that the following factors were associated with increased likelihood of miscarriage: increasing age, absence of fetal heart pulsation, heavier bleeding, and a large subchorionic haematoma. In addition to miscarriage, subchorionic haematoma increases the risk of placental abruption and preterm premature rupture of membranes.28 We previously proposed classification of the three sizes of the subchorionic haematoma, in relation to gestational sac size.29 We suggested assessment of the size of subchorionic haematoma relative to the gestational sac, rather than the mere presence or absolute size of subchorionic haematoma, in accordance with the approach used by clinicians in the United States.30 This parameter was expected to enhance prediction of miscarriage, but this hypothesis was not supported by multivariate analysis results; further studies are needed to more thoroughly investigate the usefulness of this parameter.

The results of our external validation study suggested that the scoring system would reliably predict probable pregnancy viability, despite slight differences in score implication. Application of this score could potentially enhance the treatment of women who initially present with early pregnancy complications. The cut-off value obtained in this study may be useful when counselling pregnant women. Further studies will be performed in our study centre to determine whether this externally validated scoring system could be utilised to reduce psychological morbidity by reassuring the women likely to maintain a viable pregnancy, while psychologically preparing other women for expected miscarriage.

Concept or design: OYK Wan, SSC Chan, JPW Chung.
Acquisition of data: OYK Wan, JWK Kwok
Analysis or interpretation of data: OYK Wan, JWK Kwok, DS Sahota.
Drafting of the manuscript: OYK Wan, DS Sahota, TTH Lao.
Critical revision of the manuscript for important intellectual content: OYK Wan, SSC Chan, JPW Chung, TTH Lao, DS Sahota.

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

As an editor of the journal, JPW Chung was not involved in the peer review process. The remaining authors have no conflicts of interest to disclose.

The results from this research have been presented, in part, at the following conferences:
1. Wan O, Chan SS, Kong G. A prospective observational study to validate the reliability of the early pregnancy viability scoring system. Ultrasound Obstet Gynecol 2015;46(Suppl 1):42. (Oral presentation)
2. Kwok J, Wan O, Chan SS. Subchorionic hematoma: its size and association with early pregnancy outcome. Ultrasound Obstet Gynecol 2015;46(Suppl 1):172. (Poster presentation)
3. Wan OY, Chan SS, Kong GW. A prospective observational study to validate the reliability of the early pregnancy viability scoring system. FOCUS in O&G 2015 Congress; 2015. May 30-31; Shatin, Hong Kong. The Chinese University of Hong Kong; 2015. (Oral presentation)
4. Wan OY, Chan SS, Kong GW. External validation of an early pregnancy viability prediction model via a prospective observational study. RCOG World Congress; 2016. Jun 20-22; Birmingham: United Kingdom; 2016. (Poster presentation)

This research was supported by a research grant from the Health and Medical Research Fund of Food and Health Bureau, Hong Kong SAR (HMRF Reference: 12131091). The study sponsor was not involved in the collection, analysis, or interpretation of data, or in the writing of the manuscript.

Ethics approval was obtained from the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CREC-2013.348). Written informed consent was obtained from all the participants.

1. Kong GW, Lok IH, Yiu AK, Hui AS, Lai BP, Chung TK. Clinical and psychological impact after surgical, medical or expectant management of first-trimester miscarriage—a randomised controlled trial. Aust N Z J Obstet Gynaecol 2013;53:170-7. Crossref

2. Lok IH, Yip AS, Lee DT, Sahota D, Chung TK. A 1-year longitudinal study of psychological morbidity after miscarriage. Fertil Steril 2010;93:1966-75. Crossref

3. Elson J, Salim R, Tailor A, Banerjee S, Zosmer N, Jurkovic D. Prediction of early pregnancy viability in the absence of an ultrasonically detectable embryo. Ultrasound Obstet Gynecol 2003;21:57-61. Crossref

4. Volgsten H, Jansson C, Darj E, Stavreus-Evers A. Women’s experiences of miscarriage related to diagnosis, duration, and type of treatment. Acta Obstet Gynecol Scand 2018;97:1491-8. Crossref

5. Hong Kong College of Obstetricians and Gynaecologists. Territory-wide Obstetrics and Gynaecology Audit Report 2009. Hong Kong College of Obstetricians and Gynaecologists; 2014.

6. Abdallah Y, Daemen A, Guha S, et al. Gestational sac and embryonic growth are not useful as criteria to define miscarriage: a multicenter observational study. Ultrasound Obstet Gynecol 2011;38:503-9. Crossref

7. Bottomley C, Bourne T. Dating and growth in the first trimester. Best Pract Res Clin Obstet Gynaecol 2009;23:439-52. Crossref

8. Doubilet PM. Should a first trimester dating scan be routine for all pregnancies? Semin Perinatol 2013;37:307-9. Crossref

9. Oates J, Casikar I, Campain A, et al. A prediction model for viability at the end of the first trimester after a single early pregnancy evaluation. Aust N Z J Obstet Gynaecol 2013;53:51-7. Crossref

10. Stamatopoulos N, Lu C, Casikar I, et al. Prediction of subsequent miscarriage risk in women who present with a viable pregnancy at the first early pregnancy scan. Aust N Z J Obstet Gynaecol 2015;55:464-72. Crossref

11. Bottomley C, Van Belle V, Kirk E, Van Huffel S, Timmerman D, Bourne T. Accurate prediction of pregnancy viability by means of a simple scoring system. Hum Reprod 2013;28:68-76. Crossref

12. Guha S, Van Belle V, Bottomley C, et al. External validation of models and simple scoring systems to predict miscarriage in intrauterine pregnancies of uncertain viability. Hum Reprod 2013;28:2905-11. Crossref

13. Ku CW, Allen JC Jr, Malhotra R, et al. How can we better predict the risk of spontaneous miscarriage among women experiencing threatened miscarriage? Gynecol Endocrinol 2015;31:647-51. Crossref

14. Pillai RN, Konje JC, Richardson M, Tincello DG, Potdar N. Prediction of miscarriage in women with viable intrauterine pregnancy—a systematic review and diagnostic accuracy meta-analysis. Eur J Obstet Gynecol Reprod Biol 2018;220:122-31. Crossref

15. Mukherjee S, Velez Edwards DR, Baird DD, Savitz DA, Hartmann KE. Risk of miscarriage among black women and white women in a U.S. prospective cohort study. Am J Epidemiol 2013;177:1271-8. Crossref

16. Khalil A, Rezende J, Akolekar R, Syngelaki A, Nicolaides KH. Maternal racial origin and adverse pregnancy outcome: a cohort study. Ultrasound Obstet Gynaecol 2013;41:278-85. Crossref

17. Royal College of Obstetricians and Gynaecologists. The Management of Early Pregnancy Loss. Green-top Guideline No 25, 2006. Available from: http://www.jsog. org/GuideLines/The_management_of_early_pregnancy_ loss.pdf. Accessed 7 Aug 2019.

18. Royal College of Obstetricians and Gynaecologists. Addendum to GTG No 25 (Oct 2006). Available from: https://www.rcog.org.uk/globalassets/documents/news/ addendum-to-gtg-no-25.pdf. Accessed 7 Aug 2019.

19. National Institute for Health and Care Excellence. Ectopic pregnancy and miscarriage: diagnosis and initial management. Clinical guideline [CG154]. Available from: https://www.nice.org.uk/Guidance/CG154. Accessed 7 Aug 2019.

20. National Institute for Health and Care Excellence. Smoking: stopping in pregnancy and after childbirth. Public Health guideline [PH26]. Available from: https://www.nice.org. uk/guidance/ph26. Accessed 7 Aug 2019.

21. National Institute for Health and Care Excellence. Antenatal care for uncomplicated pregnancies. Clinical guideline [CG62]. Available from: https://www.nice.org. uk/Guidance/CG62. Accessed 7 Aug 2019.

22. WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363:157-63. Crossref

23. Pan Y, Zhang S, Wang Q, et al. Investigating the association between prepregnancy body mass index and adverse pregnancy outcomes: a large cohort study of 536 098 Chinese pregnant women in rural China. BMJ Open 2016;6:e011227. Crossref

24. Pineles BL, Park E, Samet JM. Systematic review and metaanalysis of miscarriage and maternal exposure to tobacco smoke during pregnancy. Am J Epidemiol 2014;179:807- 23. Crossref

25. Kong GW, Tam WH, Sahota DS, Nelson EA. Smoking pattern during pregnancy in Hong Kong Chinese. Aust N Z J Obstet Gynaecol 2008;48:280-5. Crossref

26. Sahota DS, Leung WC, Chan WP, To WW, Lau ET, Leung TY. Prospective assessment of the Hong Kong Hospital Authority universal Down syndrome screening programme. Hong Kong Med J 2013;19:101-8.

27. Chan SS, Cheung RY, Yiu AK, et al. Prevalence of levator ani muscle injury in Chinese women after first delivery. Ultrasound Obstet Gynecol 2012;39:704-9. Crossref

28. Tuuli MG, Norman SM, Odibo AO, Macones GA, Cahill AG. Perinatal outcomes in women with subchorionic hematoma: a systematic review and meta-analysis. Obstet Gynecol 2011;117:1205-12. Crossref

29. Kwok J, Wan O, Chan SS. P16.02: Subchorionic hematoma: its size and association with early pregnancy outcome. Ultrasound Obstet Gynecol 2015;46(Suppl 1):172. Crossref

30. Heller HT, Asch EA, Durfee SM, et al. Subchorionic hematoma: correlation of grading techniques with first-trimester pregnancy outcome. J Ultrasound Med 2018;37:1725-32. Crossref

Half of all children with vision problems worldwide reside in China,1 and many children in rural China are visually impaired. In particular, a recent study of vision problems among primary school children in rural northwestern China showed that 24% of 9- to 11-year-old students in grades 4 and 5 of primary school had myopia.2 Among 13- to 15-year-old students in lower secondary school, half (50.4%) of all students reportedly exhibit visual impairment.3 Other studies have shown similarly high rates of myopia among children in rural China.4 5

Children with poor vision experience academic difficulty, such that poor vision is significantly negatively correlated with academic performance.2 Appropriate spectacles wear has been shown to increase the capacity and motivation of students to learn. In two recent randomised controlled trials in China, students who were provided free spectacles demonstrated statistically significant improvement in academic performance.3 5 Although spectacles enable non-invasive, inexpensive treatment of myopia, a significant proportion of rural Chinese children who need spectacles do not own them.6 In a sample of 18 915 students in grades 4 to 6 in rural Gansu Province, >97% of near-sighted students did not own spectacles.7 Among 9- to 11-year-old students in primary school, as few as one in six students with poor uncorrected vision owned spectacles.2 In lower secondary school, only 37% of students who needed spectacles were observed to own spectacles.

In addition to the outright failure to own spectacles, under-correction or the use of poorly fitted spectacles (including spectacles with refractive inaccuracy and spectacles that were not upgraded in a timely manner) may increase the burden of children with poor vision due to refractive error.6 Researchers have discovered high rates of inaccuracy in spectacles prescriptions in rural China. In a random sample of 3226 children in 15 middle schools (grades 7-10) in Guangdong Province, nearly 67% of student-owned spectacles were inaccurate. A spectacles quality-check study conducted in 33 primary schools in Shaanxi and 36 primary schools in Gansu showed that nearly half (46%) of students with spectacles had a Snellen visual acuity (VA) that was not corrected to ≥6/12.8 9

An understanding of vision problems among rural students in elementary and lower secondary schools is beginning to emerge in the literature; however, to the best of our knowledge, few studies have examined visual impairment and spectacles ownership among the approximately 13.5 million upper secondary school (grades 10-12) students in rural China. Empirical studies regarding spectacles quality among these students are similarly scarce. As a result, factors determining spectacles ownership and quality at this age are poorly understood. Vision problems among upper secondary school students are of particular concern in rural China because visual impairment has been shown to increase as students age through the school system; notably lower secondary school students experience higher rates of impairment and uncorrected poor vision, compared with primary school students.10 Therefore, there may be a greater need for quality vision care among upper secondary school students.

Furthermore, these vision problems among students in rural China are of particular concern because of China’s two-track upper secondary school system. Specifically, there are two types of upper secondary schools, academic upper secondary schools (AUSSs) and vocational upper secondary schools (VUSSs). Empirical studies comparing AUSS and VUSS students in China have shown that VUSS attendance does not lead to gains in specific skills and that it reduces general skills (relative to AUSS attendance). Moreover, VUSS students have a higher tendency to drop out.11 The primary determinant of whether a student attends an AUSS or a VUSS is that student’s performance on an entrance exam. Given that uncorrected vision in primary and middle school can lead to reduced academic performance,3 7 students without spectacles or with incorrect prescriptions may be attending VUSSs at higher rates than are consistent with their abilities. There may also exist a feedback mechanism between the skills VUSS students are likely to learn and the rates of spectacles ownership, as well as the quality of spectacles, among these students. In particular, high rates of uncorrected refractive error could mean that VUSS students experience smaller benefits from school than AUSS students; this might cause students (or their parents) to have less interest in spectacles. However, without empirical studies to assess potential links between spectacles ownership and quality in both AUSS and VUSS, there is no basis to study causal mechanisms. Understanding differences in vision and access to quality vision care in China’s upper secondary school environment can help to support future studies of this issue.

The overall objective of this study was to assess the prevalence of visual impairment, rate of spectacles ownership, and quality of spectacles among upper secondary school students in northwestern rural China. To fulfil this objective, there were three specific aims. First, the prevalence of refractive error, rate of spectacles ownership, and quality of spectacles were assessed in both AUSS and VUSS students. Second, the factors associated with spectacles ownership were assessed among students who had refractive errors. Finally, the association between spectacles ownership and type of upper secondary school (AUSS or VUSS) was investigated.

The protocol for this study was approved in full by Institutional Review Boards at Stanford University, Palo Alto, US, and the Zhongshan Ophthalmic Center, Guangzhou, China. Permission was received from the local Board of Education in each county, as well as the principals of all schools involved. The principles of the Declaration of Helsinki were followed throughout.

This study was carried out in Mei and Qianyang counties, both of which are located within Baoji prefecture in the southern part of Shaanxi Province. The Shaanxi Province per capita gross domestic product (GDP) was $7728 in 2015.12 Baoji prefecture ranked approximately in the middle of all prefectures within Shaanxi Province (ranked 4/10 at $7651 per year).13 The per capita GDP in Baoji prefecture is similar to the per capita GDP in Shaanxi Province. In terms of per capita GDP, Mei county and Qianyang county ranked approximately in the middle of all counties in Shaanxi Province (Mei county ranked 44/107 at $5667 per year, while Qianyang county ranked 55/107 at $5272 per year)13; therefore, the per capita GDPs of these two counties are relatively similar to the per capita GDP of Shaanxi Province ($7728). The per capita GDPs in these two counties are similar to the per capita GDPs in Ningxia Autonomous Region ($6210) and Qinghai Province ($5845).14 In addition to its average GDP, our sample area is broadly representative of poorer rural parts of northwestern China due to the scarcity of agricultural resources (cultivated land per capita of only 0.11 hectares per household, similar to the regional average).12

Data for this study were collected in Mei county during March 2016, and in Qianyang county during April 2016. The study was conducted at all six upper secondary schools in Mei and Qianyang counties, including four AUSSs (3 in Mei county and 1 in Qianyang county) and two VUSSs (1 in each county). Total enrolment for the six schools was 5583 students; 4549 attended AUSSs and 1034 attended VUSSs. Access to these schools for research purposes was granted by the Board of Education of each county. All students in all grades 10 and 11 in each sample county and sample school participated in the data collection process.

Data were collected in two parts by enumerators, graduate students from Shaanxi Normal University who had been trained by optometrist from the Zhongshan Ophthalmic Center. In the first part of the data collection effort, enumerators administered a questionnaire to all students, regarding a number of student and household characteristics. The variables that were created from the data included spectacles ownership (defined as the ability to produce spectacles at school on the day the questionnaire was administered, following prior instructions to bring them that day), grade level, sex, urban or rural residence, school type, boarding status at school, parental migration status (both parents worked away from home this semester), and parental educational level. In the second part of the data collection effort, the survey team conducted a VA test to measure the sharpness of each student’s vision. Visual acuity was measured without refractive correction for all children, as well as with habitually worn correction for children who owned spectacles. The data collection procedure is shown in the Figure.

The VA assessment was conducted using Early Treatment in Diabetic Retinopathy Study Tumbling-E charts (Precision Vision, La Salle [IL], US). The chart has 14 rows of optotypes (represented by a capital letter E) with five optotypes pointing randomly in different directions in each row. The sizes of the optotypes become smaller with vertical movement downward on the chart. Visual acuity is recorded as 6/X, such that X varies between 60 (at the top of the chart) and 3 (at the bottom of the chart) when tested at a distance of 4 m.

The VA was measured without refractive correction for all children, as well as with habitually worn correction for children who owned spectacles. Each eye was tested separately in a well-lit, indoor area of each school. Students stood at a distance of 4 m from a Tumbling-E chart. Each student began testing from the top row (6/60). If the orientation of at least four of the five optotypes was correctly identified, the student was examined on row 4 (6/30). Failure was defined as an inability to correctly identify the orientation of at least four of the five optotypes in a given row. If one or fewer optotypes was missed on row 4, the testing resumed at row 7 (6/15) and continued to row 11 (6/6). If a student failed, the row immediately above the failed row was tested until the student identified at least four of the five optotypes in a row. The lowest row read successfully was recorded as the VA for the eye undergoing testing. If the top row was missed at 4 m, the student was progressively advanced to 1 m with progression down the chart as described above, and the VA was divided by 4. In this study, we defined normal vision as VA >6/12 in both eyes, in accordance with published definitions.4 Thus, a failed visual screening was defined as uncorrected VA ≤6/12 in either eye.

We also performed a subgroup analysis of students with poor vision, which was defined as uncorrected VA ≤6/12 in the better eye. The definition of visual impairment in the better eye included only binocular visual impairment. Using the VA of the better eye as a cut-off for poor vision is a common practice in clinical vision research, because the VA of the better-seeing eye is the best indication of an individual’s level of vision.4 6 15 We stratified poor vision into mild poor vision (presenting VA ≤6/12 to 6/18), moderate poor vision (presenting VA <6/18 to 6/60), and severe poor vision (presenting VA <6/60), in accordance with categories proposed in the World Health Organization International Statistical Classification of Disease 10th Revision.2

To calculate and compare different VA levels in this study, we used a linear scale with constant increments, logMAR (logarithm of the minimum angle of resolution).2 logMAR is one of the most commonly used continuous scales to describe VA, and uses the logarithm transformation: logMAR = log₁₀ (1/VA). logMAR offers a relatively intuitive interpretation of VA measurement. Its scale uses constant increments of 0.1, where each increment indicates approximately one line of VA loss in the Early Treatment in Diabetic Retinopathy Study chart. Higher logMAR values indicate worse VA.2

We first conducted a general descriptive analysis of visual impairment, spectacles ownership, and spectacles quality. We also investigated the prevalence of various baseline variables among students attending AUSSs and those attending VUSSs. Variables included logMAR VA, level of poor vision (ie, mild, moderate, or severe), school type (AUSS=1), age, grade (grade 10=1), sex (male=1), boarding status (living at school=1), residence (urban=1), parental migration status (both migrated=1), and parental education for both mother and father (completed ≥12 years of education=1).

Next, to explore potential factors influencing spectacles ownership, we investigated all baseline variables as predictors of ownership among students who needed spectacles, using simple and multiple logistic regression analyses. We included only students with uncorrected VA ≤6/12 in either eye in our regression analyses. To explore the relationship between spectacles ownership and the type of upper secondary school (AUSS or VUSS), we employed a second logistic regression analysis method, which controlled for VA, age, sex, urban or rural residence, parental migration status, and parental education levels; this analysis included only students with uncorrected VA ≤6/12 in either eye. All analyses were performed using Stata version 14.0 statistical software (StataCorp, College Station [TX], US). All tests were two-sided and P<0.05 was considered statistically significant.

Among the 5583 students (54% boys, mean age 16.4±1.0 years) in grades 10 and grade 11 at four AUSSs (n=4549) and two VUSSs (n=1034), 100% completed both the questionnaire and the vision screening. Uncorrected VA of ≤6/12 was present in 4026 students (72%); of these, 3425 were in AUSSs (75% of AUSS students) and 601 were in VUSSs (58% of VUSS students). A greater proportion of AUSS students had poor vision, compared with VUSS students (P=0.004, Table 1).

Among students with poor uncorrected vision, 2763 (69%) demonstrated owning spectacles. Of the students who owned spectacles, 2551 were in AUSSs (75% of students with poor vision attending AUSSs) and 212 were in VUSSs (35% of students with poor vision attending VUSSs). Among students who needed spectacles, a greater proportion of students in VUSSs did not own spectacles, compared with students in AUSSs (Table 2).

When spectacles ownership was stratified by severity of visual impairment in the better eye, 72% of students with mild poor vision (726/1009) in AUSSs and 32% of students with mild poor vision (61/190) in VUSSs had spectacles. Of students with moderate poor vision, 95% (1148/1207) in AUSSs and 63% (95/152) in VUSSs had spectacles. Finally, of students with severe poor vision, 99% (461/468) in AUSSs and 90% (35/39) in VUSSs had spectacles (Table 2). The spectacles owned by the students were often under-corrected or exhibited refractive inaccuracy; 36% of students who had spectacles failed the VA screening while wearing their spectacles. Both AUSS and VUSS students who had spectacles were equally likely to have under-corrected vision (Table 2).

Simple logistic regression models for predicting spectacles ownership revealed that the following characteristics were significant predictors among students with poor vision (Table 3): worse uncorrected VA (P<0.001); AUSS attendance (P<0.001); a lower grade level (P=0.006); male sex (P<0.001); boarding at school (P=0.001); residence in an urban area (P=0.001); and both parents out-migrated for work (P=0.015). Students whose mothers had completed at least 12 years of education were also more likely to own spectacles than students whose mothers did not finish upper secondary school (P=0.006); similarly, students whose fathers completed at least 12 years of education were more likely to own spectacles (P=0.020).

Multiple logistic regression analysis revealed that the following characteristics were associated with spectacles ownership: worse uncorrected VA (P<0.001); AUSS attendance (P<0.001); male sex (P<0.001); and residence in an urban area (P=0.034). In this model, age, boarding status, parental out-migration, and education status were not associated with spectacles ownership (Table 3). In the multiple logistic regression model investigating the relationship between spectacles ownership and type of upper secondary school, spectacles wear was strongly associated with AUSS attendance (P<0.001); the same result was obtained when controlling for student characteristics and parental characteristics (Table 4).

Our study found a high prevalence of poor vision among upper secondary school students in rural northwestern China: nearly 72% of these students had refractive errors, which was consistent with the results of a prior analysis in which a high prevalence of poor vision was found among upper secondary school students in rural areas.15 Furthermore, rates of correction were well below poor vision prevalence rates: overall, 31% of students who needed spectacles were not wearing them in this rural northwestern Chinese cohort.

Furthermore, nearly 40% of the students who wore spectacles had prescriptions with under-correction or refractive inaccuracy. Thus, nearly half of students with poor vision and more than one third of all students are attending school without clear vision. This demonstrates a high rate of unmet need in visual care among upper secondary school students, which is consistent with the findings of studies from primary and lower secondary schools in rural northwestern China.2 6 16

Based on our results, we suspect that several factors may contribute to the low rates of spectacles ownership among rural upper secondary school students. In particular, rates of spectacles ownership were higher in students with worse uncorrected VA, which suggests that students with mild poor vision do not experience a need to correct their vision; these students or their parents may mistakenly presume that spectacles are unnecessary or associated with deteriorations in eyesight.17

Another factor that may affect spectacles ownership is a family’s level of income/wealth. Spectacles ownership rates were higher among students residing in urban areas, as well as among students with better-educated fathers. These factors may act as proxies for family wealth18 19; thus, our results suggest that children in a wealthier family are more likely to wear spectacles when needed. Rural students tend to be much poorer than their urban counterparts, which may explain why many do not own spectacles. Our findings are consistent with those found in the literature.2 9

Our results also showed that spectacles ownership was strongly associated with AUSS attendance. This finding implies involvement of several factors in spectacles ownership. One of these factors may be the expectations of academic performance for VUSS students, who have lower rates of spectacles ownership than their AUSS peers; VUSS students may have a lighter academic burden,20 which allows them to tolerate poorer vision. Their parents may also presume that they do not need spectacles. Moreover, the number of VUSS students with poor vision may also contribute to lowered academic achievements and expectations. Our results are only correlative, but future studies of causal mechanisms are needed. Furthermore, given that VUSS graduates have poorer general skills compared with AUSS graduates,20 VUSS students comprise an at-risk population that urgently requires additional investigation.

The overall implication of the study, similar to the findings in studies of primary school and lower secondary school,2 3 5 is that systematic implementation of vision care programmes is needed in rural Chinese schools, including upper secondary schools. There is a need to more clearly identify which students exhibit poor vision, as well as a need for programmes that provide spectacles or encourage spectacles wear among affected students. Finally, there is a need for programmes that provide high-quality spectacles.

This study was limited in three ways. First, the results were based on cross-sectional data, such that it was difficult to perform a causal analysis between spectacles ownership and upper secondary school educational choice. Second, data regarding school performance and spectacles ownership were not collected when students were in lower secondary school (prior to AUSS or VUSS attendance); these data might have been useful in a causal analysis. Therefore, we cannot exclude the possibility that factors specific to AUSSs (for instance, more homework leading to less time spent outdoors) might place AUSS students at greater risk of visual impairment. Furthermore, we cannot exclude the possibility that inability to afford spectacles and the lack of spectacles use prior to taking the entrance exam would result in lower overall learning, which would partially contribute to the type of upper secondary school attended; this contribution may lead to a positive correlation between VUSS attendance and poor vision. Third, the participants in this study were recruited from two counties in rural northwestern China, which limits the external validity of the findings.

Despite these limitations, we believe the results from this study provide compelling evidence that a number of students in rural northwestern China are not receiving adequate vision care: poorer students whose parents have less education are less likely to have spectacles when needed; VUSS students are also less likely to have spectacles when needed, compared with AUSS students. We presume that these findings will aid programme planners in targeting vulnerable populations when formulating strategies to reduce the burden of uncorrected myopia in rural China. China’s health and education policymakers should consider incorporating programmes that specifically address spectacles ownership and quality in rural upper secondary schools, especially among VUSS students and poorer students.

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

Concept or design: All authors.
Acquisition of data: J Zhao, H Guan, K Du, Y Shi.
Analysis or interpretation of data: J Zhao, H Guan, K Du, Y Shi.
Drafting of the article: J Zhao, H Guan, H Wang, M Boswell, AOsborn.
Critical revision for important intellectual content: H Wang, M Boswell, S Rozelle, N Congdon.

All authors have disclosed no conflicts of interest.

This study was supported by the 111 Project (Grant No. B16031). H Guan is funded by the National Natural Science Foundation of China, grant number 7180310.

This study was approved by the institutional review board at Stanford University, Palo Alto, United States (Ref 52514). Permission was received from local boards of education in each region and the principals of all schools. A letter was given to children and the parents informing them of the vision screening of their child, and a receipt returned to the school to show the parent consent. The presented data are anonymised, and the risk of identification is low. This study was performed in accordance with the principles of the Declaration of Helsinki.

Eczema is a prevailing childhood dermatological disease associated with skin dryness and impaired quality of life.1 2 3 4 Eczema, particularly during disease flare-ups, is typically characterised by unbearable itchiness, sleep disturbance, erythematous rash, dryness, and inflammation.1 5 6 In the clinical setting, subjective and objective assessments are used in combination to evaluate eczema disease severity.7 8

SCORing Atopic Dermatitis (SCORAD) is a clinical tool often used to evaluate disease severity.7 9 Eczematous areas and intensity of eczema are recorded; these are considered the objective components in SCORAD (ie, objective SCORAD). Subjective symptoms including the degree of skin itchiness and sleep disturbance due to skin itchiness are also recorded to determine the overall SCORAD assessment outcome.10 Nottingham Eczema Severity Score (NESS) is a clinical assessment tool that adopts subjective and objective approaches for evaluation of the severity of eczema.4 7 11 12 Using an assessment form, patients record their symptoms of sleep impairment due to skin itchiness in terms of the number of nights and length of the presence of eczematous symptoms within the previous 12 months; they also report the areas of eczematous skin.11 In Hong Kong, a validated Chinese version of the NESS is used.12

In addition to clinical scores, skin measurements including skin hydration (SH) and transepidermal water loss (TEWL) are important indicators of the barrier function of the skin.8 13 Skin hydration and TEWL measurements have been shown to correlate with eczema severity and quality of life impairment.8 13 Measurements of these non-invasive parameters may be useful in prognostic prediction of the development of eczema in infants14; these parameters have been use in Hong Kong as auxiliary tools in eczema therapeutic trials.7 8 15 16 To the best of our knowledge, there have been no studies in Hong Kong regarding the influences of age and sex on these parameters. This study aimed to compare the age and sex of Chinese children with and without eczema in terms of SH and TEWL, and to evaluate the correlations of SH and TEWL with clinical scores.

Chinese patients with eczema, who were aged 4 weeks to 18 years and were attending the paediatric dermatology clinic at Prince of Wales Hospital, were recruited from July 2018 to October 2018. Eczema was diagnosed in accordance with the United Kingdom working diagnosis.9 In addition, Chinese patients without eczema (eg, naevi or warts), aged 4 weeks to 18 years, were also recruited from the same paediatric dermatology clinic at Prince of Wales Hospital during the study period. Patients without eczema were defined as those who had not been clinically diagnosed with eczema in their medical history. Patients and their caregivers received an explanation of the details of the study including its aim and procedures, as well as each patient’s rights and possible risks related to involvement in the study. Written informed consent for participating in the overall study was obtained from patients or caregivers before enrolment in the study. Patients were then separated into two groups; those with a current or prior clinical diagnosis of eczema were regarded as children with eczema, while those without clinically diagnosed eczema were regarded as children without eczema. Both children with and without eczema underwent the same study measurements and received the same treatment procedures throughout the study.

Eczema was routinely assessed by clinicians using the SCORAD tool.9 10 The overall SCORAD assessment outcome was determined by two objective components and one subjective component. The extent of eczematous areas involved were assessed by observing patient’s eczematous body surface area covering head and neck (9%), bilateral upper limbs (9% each), bilateral lower limbs (18% each), anterior trunk (18%), back (18%) and genitals (1%), which were aggregated to 100% of whole body surface areas. Intensity of dermatological lesions was assessed by indicating none (0), mild (1), moderate (2) and severe (3) onto the eczematous areas in terms of six categories: redness, swelling, oozing/crust, excoriation, lichenification, and dryness on non-eczematous areas. The score of intensity of dermatological lesions was obtained by adding up the scores of these six categories. Subjective components of SCORAD were assessed by evaluation of subjective symptoms (ie, pruritus and sleep loss) over the course of the previous three nights, which were rated on a scale from 0 to 10.9 10 Objective SCORAD scores <15, 15 to 40, and >40 were classified as mild, moderate, and severe disease, respectively; overall SCORAD scores <25, 25 to 50, and >50 were classified as mild, moderate, and severe disease, respectively.9 10 17 Both overall SCORAD and objective SCORAD scores were used in this study. The NESS tool was also used to determine the severity of eczema.11 18 With respect to the past 12 months, the duration of eczema and number of nights impacted by skin itchiness weekly were rated from 1 to 5. A higher score indicated greater eczema severity. In addition, areas of skin with eczematous lesions (eg, rash, lichenified skin, and/or bleeding) were recorded. Scores ranging from 3 to 8, 9 to 11, and 12 to 15 were categorised as mild, moderate, and severe disease.11 The validated Chinese version of the NESS was used.12

Patients were first taken to an air-conditioned treatment room and rested for 20 minutes. All children included in this study (regardless of the presence or absence of eczema) reported no use of emollient in any form within 24 hours prior to the skin measurements. Skin hydration was measured topically using the MoistureMeterSC (Delfin Technologies Ltd., Kuopio, Finland) at a standard site 2 cm below the antecubital fossa of each patient’s right arm.8 13 The SH is measured in arbitrary units (a.u.), which range from 0 to 300 a.u. The MoistureMeterSC serves as a capacitance meter, measuring the skin capacitance detected from the probe head of equipment onto the skin surface layer, based on the relationship between skin capacitance and the water content of the surface layer of skin19 20; a higher SH value indicates greater skin moisture. Transepidermal water loss was topically measured using the Vapometer SWL5 (Delfin Technologies Ltd.) at the same standard site 2 cm below the antecubital fossa of each patient’s right arm.8 13 The TWEL is measured in g/m²h, which range from 0 to 300. The Vapometer is equipped with a closed cylindrical chamber containing sensors which detect changes in relative humidity and temperature after placement on the skin surface. The TEWL is then measured based on the change in relative humidity19 20; an increasing TEWL value indicates greater skin dryness. Skin measurements were conducted by trained research personnel, and fewer than 5 minutes were required to complete both procedures. During the measurements, no psychological or physiological discomforts occurred.

Clinical data were de-identified and analysed using SPSS Statistics for Windows, version 25.0 (IBM Corp, Armonk [NY], US). Frequency distributions were used to describe patients’ demographic data. Independent-samples t tests were used to compare SH and TEWL between patients in terms of sex, age, and presence or absence of disease. Pearson correlation was used to investigate associations between skin measurements and SCORAD subcategories. Trend analysis was performed regarding the effects of age on SH. Multiple regression model of the interactions of SH and TEWL with age and sex in children with and without eczema was done. P values <0.05 were considered statistically significant and 95% confidence intervals were adopted for all statistical comparisons in this study.

In total, datasets of clinical scores (SCORAD) and equipment measurements were reviewed for 252 patients during the period from August 2018 to October 2018. Among the 252 patients (mean age, 5.84±5.97 years; 52.0% boys), 132 were children with eczema (mean age, 7.05±6.60 years; 52.3% boys), while 120 were children without eczema (mean age, 4.52±4.91 years; 51.7% boys). Furthermore, 121 patients were aged ≤2 years, while 131 patients were aged >2 years.

Table 1 shows that, among children with eczema (n=132), the mean SH of boys aged ≤2 years was significantly higher than the mean SH of boys aged >2 years (P<0.001). Similarly, the mean SH of girls aged ≤2 years was significantly higher than the mean SH of girls aged >2 years (P<0.001). Among children without eczema (n=120), the mean SH of boys aged ≤2 years was significantly higher than the SH of boys aged >2 years (P=0.046). However, no statistically significant difference in SH (P=0.980) was observed in comparisons between girls aged ≤2 years and girls aged >2 years. These results demonstrated that SH significantly differed in an age-related manner among both boys and girls with eczema.

Table 1 shows that, among children with eczema (n=132), the mean SH of girls aged >2 years was significantly higher than the mean SH of boys aged >2 years (P=0.006). Similarly, among children without eczema, the mean SH of girls aged >2 years was significantly higher than the mean SH of boys aged >2 years (P=0.009). However, in both children with eczema and children without eczema, there were no sex-related differences in SH among patients aged ≤2 years. These results demonstrated that SH significantly differed in a sex-related manner among patients aged >2 years, regardless of the presence or absence of eczema.

Table 1 shows that, among boys in this study (n=131), the mean SH of children with eczema aged >2 years was significantly lower than the mean SH of children without eczema aged >2 years (P<0.001). Furthermore, the mean SH of boys with eczema aged ≤2 years was significantly lower than the mean SH of boys without eczema aged ≤2 years (P=0.005). Similarly, among girls in this study (n=121), the mean SH of children with eczema aged >2 years was significantly lower than the mean SH of children without eczema aged >2 years (P<0.001). In addition, the mean SH of girls with eczema aged ≤2 years was significantly lower than the mean SH of girls without eczema aged ≤2 years (P=0.009). The results showed that SH was consistently higher in children without eczema than in children with eczema, regardless of age and sex.

Among children with eczema (Table 2; n=132), Pearson correlation demonstrated that SH was significantly negatively correlated with objective SCORAD (P<0.001), overall SCORAD (P<0.001), oedema/papulation (P=0.041), lichenification (P=0.002), dryness (P<0.001), and intensity (P=0.025). In addition, Pearson correlation demonstrated that TEWL was significantly positively correlated with objective SCORAD (P=0.018), overall SCORAD (P=0.015), and lichenification (P=0.043). No significant correlation was observed between TEWL and SH (P=0.565) in children with eczema, whereas a correlation was present in children without eczema (P<0.001).

Trend analysis revealed a marked decline in SH during the first 50 months of life among children with eczema, relative to children without eczema (Fig). Beginning at 51 months of age, the magnitude of SH began to gradually decline. Conversely, for children without eczema, a slight, steady decline in SH was observed in the first 50 months of life.

Table 1 shows that, among children with eczema (n=132), the mean TEWL of girls aged ≤2 years was significantly higher than the mean TEWL of girls aged >2 years (P=0.046). However, no statistically significant difference in TEWL (P=0.877) was observed in comparisons between boys aged ≤2 years and boys aged >2 years. Similarly, among both boys and girls without eczema in this study, no statistically significant difference in TEWL was found between children aged ≤2 years and children aged >2 years. The findings demonstrated that TEWL was affected by age only among girls with eczema.

Regardless of age and disease status, no statistically significant difference in TEWL was found between boys and girls (Table 1). Likewise, regardless of age and sex, no statistically significant difference in TEWL was found between children with eczema and children without eczema. The results showed that TEWL did not significantly differ among patients based on sex, age, or disease status.

Multiple regression models were constructed with TEWL and SH as the outcome variables, using age and sex as independent predictors, to examine the effects of the age and sex interaction on TEWL and SH in children with and without eczema. Table 3 shows that, among children without eczema, the age×sex interaction term could explain 40.9% of the variance in SH (P=0.031); however, the age×sex interaction term could not explain the variance in SH (P=0.290) among children with eczema. In addition, when used as a predictive variable, age could not explain the variance in SH among children with eczema (P=0.364) or among children without eczema (P=0.143). Similarly, when used as a predictive variable, sex could not explain the variance in SH among children with eczema (P=0.377) or among children without eczema (P=0.132). These results showed that the interaction between age and sex significantly affected SH among children without eczema, but not among children with eczema; when considered alone, neither age nor sex significantly affected SH among any of the children in the study.

In contrast, Table 3 shows that, when used as a predictive variable, age could explain 67.8% of the variance in TEWL (P=0.009) among children without eczema, but could not explain the variance in TEWL (P=0.156) among children with eczema. In addition, when used as a predictive variable, sex could not explain the variance in TEWL among children with eczema (P=0.379) or among children without eczema (P=0.952). Similarly, the age×sex interaction term could not explain the variance in TEWL among children with eczema (P=0.194) or among children without eczema (P=0.088). These results showed that age alone significantly affected TEWL among children without eczema, but not among children with eczema; sex alone and the age×sex interaction term did not significantly affect TEWL among any of the children in the study.

Skin hydration and TEWL are useful auxiliary tools in assessment of adult eczema for research purposes.8 13 21 The findings in this study confirmed our previous observations that SH and TEWL were correlated with disease severity (ie, objective SCORAD and overall SCORAD) in paediatric patients with eczema.13 As SH declines (ie, skin dryness increases), the clinical scores of objective SCORAD and overall SCORAD increase, indicating more severe eczema. Furthermore, as TEWL increases (ie, skin loses water through the epidermis), the clinical scores of objective SCORAD and overall SCORAD increase, indicating more severe eczema. In our study, SH and TEWL were also correlated with lichenification, which is an indicator of the chronicity of eczema.13 21 This study also demonstrated that age and sex could influence SH, but not TEWL. First, SH was significantly lower in children with eczema than in children without eczema, regardless of age or sex; no such difference was demonstrated with respect to TEWL. Second, SH was generally higher in young children, whereas TEWL did not differ on the basis of age. Hence, researchers should consider these relationships when interpreting SH measurements in infants.

Skin hydration was lower in children with eczema than in children without eczema (Fig). Trend analysis showed that SH declined more markedly in children with eczema than in children without eczema during the first 2 years of life. This phenomenon might reflect the onset and progression of eczema in early life. Skin hydration was correlated with skin dryness and lichenification (Table 2). With disease progression, skin dryness increases and lichenification develops as a sign of chronicity. Therefore, age is a key factor that affects SH and eczema severity. Among children aged >2 years, SH was higher in girls than in boys, regardless of the presence or absence of eczema. This universal phenomenon might be related to sex differences in cutaneous physiology. Furthermore, the observation of a sex difference in SH suggests that skin is less dry and more manageable with emollients in girls, compared with boys. We previously studied quality of life in children with eczema and found that not all aspects of quality of life were affected equally in children with eczema.22 Age and sex were important in that clothes/shoes caused more problems for girls, whereas itching and sleep disturbance primarily affected younger children. In addition, eczema damaged the epidermal layer of skin and caused greater self-image difficulties in girls than in boys, especially during childhood and adolescence. These findings indicated that psychosocial influences should be considered among patients with eczema. Notably, age influenced SH and TEWL among children without eczema. Although regression analysis showed that age was not a significant predictor of SH or TEWL among children with eczema, the partial eta squared (η_p²) for age was high, which suggested that it may explain some variance in both SH and TEWL. Thus, age should be considered in further studies of SH and/or TEWL.

The design and findings of this study were quite different from those in studies by Walters et al and Kong et al.23 24 The participants in the study by Walters et al were mainly Caucasian, while those in the study by Kong et al were mainland Chinese.23 In addition, the study by Kong et al was conducted in winter, such that the temperature and relative humidity were extremely low.24 Hong Kong is a subtropical climate and the skin measurements in the present study were performed using an established standard protocol. In addition, both prior studies focused on comparisons between children without eczema and their mothers. In contrast, our study compared SH and TEWL between children with eczema and children without eczema. Furthermore, our study showed the following: SH was significantly higher among both boys and girls aged ≤2 years than among boys and girls aged >2 years, regardless of the presence or absence of eczema; this age-related difference was not observed with respect to TEWL. Among children aged >2 years, SH was higher among girls with and without eczema than among boys with and without eczema; this sex-related difference was not observed with respect to TEWL. Regardless of age and sex, SH was lower among children with eczema than among children without eczema; this disease status-related difference was not observed with respect to TEWL. Hence, we concluded that age and sex affect SH. Kong et al24 also demonstrated a more permeable skin barrier in younger children, compared with both older children and adults. Potential underlying reasons for the findings in our study require further investigation because of the cross-sectional study design and limited sample size.

There were some limitations in this study. First, the sample size was small and did not allow in-depth subgroup analysis of children with eczema or children without eczema; thus, our findings cannot be used to establish age-based population norms for the assessment parameters used in this study. Second, the recruitment of patients was restricted to a single site; thus the sample may not have been representative of the population, and the generalisability of the results may be limited. Finally, the efficacy of skin barrier treatment was not evaluated. Future studies should address these limitations and confirm our findings.

This study investigated the effects of age and sex on SH and TEWL in children with and without eczema. Skin hydration was lower in children with eczema than in children without eczema and generally decreased with age, especially during infancy and in patients with more severe disease. Among children aged >2 years, SH was generally higher in girls than in boys. Thus, we presume that SH can be used to distinguish clinical differences in eczema based on age, sex, and disease status.

Concept or design: KL Hon.
Acquisition of data: PH Lam, WG Ng, JS Kung.
Analysis or interpretation of data: PH Lam.
Drafting of the article: KL Hon, PH Lam.
Critical revision for important intellectual content: KL Hon, WG Ng, JS Kung, NS Cheng, ZX Lin, CM Chow, TF Leung.

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.

As an editor of the journal, KL Hon was not involved in the peer review process. Other authors have no conflicts of interest to disclose.

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 Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CRE.2016.118). Written informed consent to participate was obtained from patients or caregivers before enrolment in the study.

Radical radiotherapy is a standard treatment option for patients with early-stage and locally advanced non-metastatic prostate cancer. Advances in radiotherapy in the past 20 years include the use of androgen deprivation therapy for patients with this type of cancer, as well as the application of more precise radiotherapy techniques.1 2 Intensity-modulated radiation therapy (IMRT) has emerged as the standard radiotherapy technique.3 Its benefits have been explored in terms of the effects of dose escalation or hypofractionation on survival outcomes.4 5 For patients undergoing this type of treatment, toxicities are the primary concern. Long-term side-effects (ie, complications occurring ≥3 months after radiotherapy) have a major impact on the quality of life for affected patients; this is particularly important for patients with genitourinary or rectal toxicities. Late rectal toxicities, including per-rectal bleeding, faecal incontinence, and proctitis, have been reported to occur at rates of 5% to 21%.1 34 5 6 7 8 9

Associations have been reported between late rectal toxicities and various clinical and dosimetric parameters; however, most data were collected using the conventional three-dimensional conformal technique.8 1011 12 In addition, there have been limited reports of such associations among patients in Hong Kong. In particular, Poon et al8 reported that 8% of patients exhibited grade ≥2 late rectal toxicities following IMRT in a retrospective cohort study. Although several clinical parameters were assessed, most failed to show statistically significant associations, with the exception of the presence of acute rectal toxicities.8 To the best of our knowledge, pharmacological parameters following IMRT for prostate cancer have not yet been studied in local populations. Some previous reports showed a significant association between anticoagulant use and late rectal toxicities, whereas an association between antiplatelet use and androgen deprivation was inconsistent among studies.12 13 14

Multiple strategies have been used for the treatment of late rectal toxicities. The use of hyperbaric oxygen has shown promising results in some retrospective studies, but it has not been available in Hong Kong until recently.12 15 16 Treatments with sucralfate, prednisolone enaema, short-chain fatty acids, and antifibrinolytics have been evaluated in small trials.17 18 19 Thus far, no standard approach has been established, and there are no published data regarding local management practices.

Late rectal toxicities may represent clinically significant complications because of their non-negligible incidences. Insights regarding any factors predictive of their occurrence could aid in improved treatment planning and early identification of toxicity. This study was performed to assess the incidence of late rectal toxicities and to identify factors predictive for late proctitis in patients treated with prostate-specific IMRT in Hong Kong.

This retrospective longitudinal observational study included patients with prostate cancer who received IMRT with radical intent in a tertiary referral institution in Hong Kong from January 2007 to December 2011. Patients were excluded if they were followed up for fewer than 12 months from the start of radiotherapy, if they did not complete the course of radiotherapy, if they were not at risk of proctitis (eg, those with post-abdominoperineal resection), or if they did not have a retrievable radiotherapy plan due to technical difficulties. The cut-off date for data collection was 31 December 2018.

Patients underwent treatment with a comfortably full bladder and an empty rectum, with laxatives administered 1 day prior to simulation computed tomography. Patients were asked to empty the bladder prior to attending the radiotherapy suite, and then drink a comfortable volume of water. A pelvic thermoplastic mould was used for immobilisation. Intravenous contrast was administered prior to computed tomography. Re-simulation was performed automatically if bladder volume was below 150 cc, if prominent rectal gas was present, or upon request by the attending oncologist. Contouring was performed by designated oncologists with confirmation by at least one specialist. Tumour and whole prostate were contoured as a single volume; the clinical target volume (CTV) was the volume of the tumour, whole prostate, and base of the seminal vesicle (defined as 1 cm of the central seminal vesicle proximal to the base of the prostate). Whole seminal vesicle was included in the CTV if seminal vesicle involvement was observed. Planning target volume (PTV) was determined by expanding the CTV by a radial margin of 1.5 cm, except posteriorly where a smaller margin was used (0.7 cm). Pelvic lymph node irradiation was not performed. Patients received 70 Gy in 35 daily fractions over 7 weeks at 100% of the isodose level. Rectal volume was contoured in accordance with the Radiation Therapy Oncology Group Consensus Contouring Guidelines for normal male pelvic tissue. Dose constraints for organs at risk followed our departmental protocol: for the rectum, we classified the plan as fulfilling the first, second, or third criteria. First criteria were satisfied if V40 (% of organ volume receiving 40 Gy) <35% or V65 (% of organ volume receiving 65 Gy) <17%; second criteria were satisfied if V53 (% of organ volume receiving 53 Gy) <45% or V68 (% of organ volume receiving 68 Gy) <20%; and third criteria were satisfied if V60 (% of organ volume receiving 60 Gy) <50%, V65 (% of organ volume receiving 65 Gy) <35%, or V70 (% of organ volume receiving 70 Gy) <25%. Hormonal treatment was administered based on the risk stratification used in the United Kingdom National Institute for Health and Care Excellence guidelines. Patients were followed up at 3–6-month intervals until the patient died or defaulted, and data were censored at the last recorded follow-up. Dose distributions, doses administered to organs at risk, and dose volume histograms were evaluated by the Eclipse and Planning System (Varian Medical Systems; Palo Alto [CA], United States).

For each patient, basic demographic data were documented, including age; Eastern Cooperative Oncology Group performance score; smoking habit; pretreatment albumin level; co-morbidities such as hypertension, diabetes, lipid disorder, history of cerebrovascular disease, ischaemic heart disease, and/or chronic renal impairment; medical history of abdominal surgery; drug history including antihypertensives, oral glycaemic agents, antiplatelets, anticoagulants, lipid-lowering agents, and antipurine agents; androgen deprivation therapies, including medical or surgical castration; and use of immunosuppressants. Tumour characteristics were also recorded, including pretreatment prostate-specific antigen level, clinical T-staging determined by clinical and radiological findings (based on AJCC 7th edition20), and Gleason score.

Acute and late rectal toxicities, including proctitis, incontinence, and per-rectal bleeding, were recorded and classified in accordance with Common Terminology Criteria for Adverse Events version 4.21 Late rectal toxicities were defined as those that occurred at least 3 months after the completion of radiotherapy. Late proctitis was defined as either the presence of rectal symptoms listed in Common Terminology Criteria for Adverse Events version 4, or colonoscopy findings of proctitis (eg, telangiectasia, ulcers, or inflammation). If a patient presented with per-rectal bleeding, colonoscopy findings were referenced whenever present to differentiate proctitis or other causes of bleeding, such as diverticulosis or haemorrhoids. Per-rectal bleeding only was recorded if no endoscopic proctitis features were present; otherwise, both per-rectal bleeding and proctitis were recorded. Additional parameters recorded included time of onset of late rectal toxicities, as well as treatment modalities used.

Dosimetric parameters (eg, V40, V50, V60, V70, Dmax [maximum dose], mean dose to rectum, and contoured rectal volume) were evaluated with the radiotherapy planning system. The use of static beam or volumetric arc technique was recorded, as was the compliance with rectal dose constraints.

Incidences of grade ≥1 late rectal toxicities with 95% confidence interval (CI) were calculated at 1, 2, and 5 years after treatment. The Kaplan-Meier curve method was used to illustrate the time to onset of late rectal toxicities. The Chi squared test, Fisher’s exact test, independent t test, or Mann-Whitney U test were used to compare baseline patient characteristics, pharmacological and dosimetric parameters between patients in grades 0 and ≥1 late toxicities, as well as in patients with late proctitis. The association of each parameter with late proctitis was examined using a multivariable binary logistic regression model with a backward stepwise selection method, including variables with P<0.1 in univariable regression analyses. The presence of multicollinearity was determined by using variance inflation factors. Statistical analyses were performed using SPSS (Windows version 22.0; IBM Corp, Armonk [NY], United States). The threshold of statistical significance was set at P<0.05. The STROBE checklist was followed to ensure standardised reporting.

From January 2007 to December 2011, a total of 238 patients with prostatic cancer received radical radiotherapy in our institution. As shown in the Figure, 232 patients were included in the analysis. The mean age of patients was 72.3 ± 4.8 years at time of radiotherapy (Table 1). The mean follow-up period was 7.3 ± 2.1 years, and there were 157 (67.7%) surviving patients at the cut-off date for data collection. Forty-two (18.1%) patients had been diagnosed with biochemical recurrence during the study period, based on the Phoenix definition.22 In total, 229 patients received a PTV dose of ≤70 Gy. Owing to genuine bowel invasion, or as a component of individualised dose escalation, four patients received a PTV dose of 66 to 76 Gy, of which three were >70 Gy. Colonoscopy was performed in 103 (44.4%) patients during follow-up. Among patients with per-rectal bleeding, 93 (88.6%) had undergone colonoscopy.

Occurrences of acute and late rectal toxicities throughout the study period are shown in Table 2. The rates of all-grade acute and late rectal toxicities were 36.2% and 46.5%, respectively; the rates of grade ≥2 late rectal toxicities and proctitis were 28.4% and 25.0%, respectively. Nineteen (8.2%) patients had grade 3 per-rectal bleeding, with 15 (78.9%) requiring blood transfusion and eight (42.1%) requiring endoscopic coagulation. The cumulative incidences of rectal toxicities at 1, 2, and 5 years after treatment are shown in Table 3. The median times of onset of late proctitis, late faecal incontinence, and late per-rectal bleeding were 15, 21.8, and 18.4 months, respectively.

Patients’ detailed demographic, pharmacological, and dosimetric parameters are listed in Table 1. Factors including history of haemorrhoid, PTV dose, and V70 were significantly different between patients with and without late rectal toxicities. In addition, age was the sole demographic factor significantly associated with late proctitis. There was no significant association between antiplatelet use and late rectal toxicities (P=0.066). No associations were found between late proctitis and other demographic or pharmacological characteristics (eg, PTV dose and history of haemorrhoid) in this study.

Univariable and stepwise multivariable analyses were performed to identify factors predictive of

late proctitis (Table 4). In univariable analysis, the presence of acute rectal toxicities, antiplatelet use, age at radiotherapy, Dmax, and dose/volume histogram parameters (ie V50, V60, V70, and rectal constraints) were identified as potential risk factors. In the regression model with all potential risk factors included, multicollinearity was detected among the dose/volume histogram parameters (variance inflation factors of 7.21, 8.69, 3.05, and 4.97 for V50, V60, V70, and rectal constraints, respectively). Compared to V50 and V60, V70 (ie, the high-dose region) showed a stronger association with late proctitis in univariable analysis. Multicollinearity was resolved by exclusion of V50 and V60 from the multivariable regression model. The final multivariable regression model revealed increased odds of late proctitis in patients with older age, higher V70, and the presence of acute rectal toxicities. Antiplatelet use tended to show higher odds, but this finding was not statistically significant (odds ratio=1.98, 95% CI=0.95-4.14). Dmax and satisfaction of the 3rd criteria alone were associated with late proctitis in univariable analysis, but the associations were not significant in multivariable analysis.

Common treatment modalities among patients with grade ≥2 late proctitis were also recorded. Topical agents such as Ultraproct® (commercial preparation of fluocortolone pivalate, fluocortolone hexanoate, and cinchocaine hydrochloride), bismuth ointment, or an antifibrinolytic agent (eg, tranexamic acid) are commonly used as first-line treatment.23 More than half (53.4%) of the patients had been administered an antifibrinolytic agent, while 77.6% and 19% of the patients were prescribed Ultraproct® and bismuth, respectively. Prednisolone enaema was also administered in 22 (37.9%) patients; the median duration of enaema use was 3.5 months (interquartile range, 1-7.25 months). Subjective improvement was reported by eight (36.4%) patients who received enaema treatment.

Radiation proctitis and other long-term rectal toxicities are clinically significant complications of radiotherapy to the prostate, due to their detrimental effects on patients’ quality of life, as well as the expected long duration of post-treatment survival. In our cohort, the incidences of late proctitis (30.2%) and overall rectal toxicities (46.5%) were slightly higher than those in previous reports (5%-21%).1 3 4 5 6 7 8 9 Comparison of baseline characteristics showed that more patients had ≥T3 disease in our cohort, although we found no statistically significant association between T-staging and a higher incidence of proctitis; similarly, no association between these parameters were reported in other studies.8 12 Other variables with possible interactions were similar between our study and prior studies; these included age, dosimetric parameters (eg, V70, which was 14% in our study and 10% to 23% in previous studies), and the use of antiplatelets.8 11 12

There are two possible explanations for the higher incidences of late proctitis and overall rectal toxicities. First, our study involved frequent utilisation of colonoscopy for any rectal symptoms, which may lead to a higher rate of recognition; notably, the rate of utilisation was not reported in previous studies. Second, our study had a relatively long follow-up period. Previous studies described the incidence of toxicity throughout the study period. The mean follow-up period in our study was 7.3 years, whereas that of most previous studies was 38.9 to 66 months; in one notable exception, the follow-up period was 8.4 years (the incidence was 21% in that study).3 The longer study period may also have contributed to a higher number of late rectal toxicities.

Previous reports suggested that a variety of parameters are associated with late proctitis; knowledge of these parameters could help clinicians to predict the risk of proctitis in each patient. In our study, age, and dosimetric parameters including V50, V60, and V70 were associated with late proctitis; history of haemorrhoid and V70 were associated with overall late rectal toxicities. These findings are consistent with the results of previous studies.10 11 12 13 14 24 Some factors identified in prior studies, including diabetes, previous abdominal surgery, and the use of antiandrogen or anticoagulant medication,11 13 25 failed to demonstrate any associations in the present study. Of note, <10% of the patients in our study had a history of abdominal surgery or inflammatory bowel disease; this could have influenced our ability to identify a statistically significant association. Recall bias, incomplete documentation of coexisting medical conditions and pharmacological histories, and the relatively small sample size in our cohort may have influenced our conclusions regarding factors associated with overall late rectal toxicities and/or late proctitis.

Several dosimetric parameters and dose/volume histogram data (including V50, V60, and V70) were also associated with late proctitis, as in previous studies.8 Our in-house rectal constraints did not demonstrate significant associations with the occurrence of proctitis (P=0.092). Notably, in the present study, the PTV dose was associated with overall late toxicities, but not with late proctitis specifically. Most patients received 70 Gy in this study; therefore, the effects of PTV dose on complications were difficult to establish.

Regression analysis was used to predict the odds of late proctitis among patients in our study. As shown in Table 4, higher V70, older age, and the presence of acute rectal toxicities were found to increase the odds of late proctitis. Poon et al8 also reported similar findings concerning acute rectal toxicities; however, they did not find associations with V70 or age. The increased incidence of late proctitis in our study may have enhanced our ability to identify significantly associated factors. Nevertheless, both our present study and the study of Poon et al8 demonstrated that patients with acute rectal toxicities during radiotherapy had higher incidences of late proctitis than patients without acute rectal toxicities. Similar results were reported by Fellin et al.11 Taken together, the present and prior results indicate that the presence of acute toxicities is predictive for late proctitis. Clinicians should be vigilant and perform prompt investigations when patients with acute toxicities report any rectal symptoms during subsequent follow-up.

Theoretically, dosimetric parameters are expected to be associated with late proctitis. In our study, the dosimetric parameters exhibited modest associations with late proctitis. Notably, we did not find a significant association between our in-house rectal constraints and the occurrence of late proctitis. Fellin et al11 demonstrated similar associations between late proctitis and V70, as well as other dosimetric parameters, in their cohort. This suggests that the presence of confounding factors may reduce the strength of associations with late proctitis. A notable factor is the inter-fractional variation of rectal and bladder filling; specifically, Miralbell et al26 found that rectal filling was significantly associated with late rectal toxicities. Imaging-guided radiotherapy with inter-fractional bowel and bladder control has been suggested in accordance with the nomogram designed by Delobel et al9; this type of therapy could reduce the risks of acute and late rectal toxicities. In our study, there was no strict inter-fractional bowel or imaging control for bladder and rectal volumes during the course of IMRT. Although we found no statistically significant difference in the mean rectal volume during simulation computed tomography between patients with and without late proctitis, we could not retrieve the inter-fractional variation in rectal volumes for analysis in this study; this factor was also excluded from analysis in the study by Fellin et al.11 Although identical instructions were provided to patients during simulation and treatment, inter-fractional variations may have been statistically significant. To further confirm whether dosimetric parameters are predictive of late proctitis, a prospective study is needed in which strict interfractional rectal and bladder control are performed, in combination with improved treatment verification strategies (eg, the use of cone beam computed tomography).

There were a few weaknesses in this study. First, this was a retrospective study in which incomplete reporting may have occurred and data might have been missing. Second, the small sample size and the low prevalences of some clinical factors and events may have affected the statistical power to determine associations between rates of complications and potential predictive factors (eg, use of anticoagulants and presence of inflammatory bowel disease). Third, confounding factors might have been present as mentioned earlier in the Discussion, and could not be controlled because of the retrospective nature of this study. However, this study did identify factors that clinicians could use to predict the occurrence of late proctitis. The significant association of V70 with late proctitis should be applied to radiotherapy planning, in that high doses to the rectal volume should be limited where possible.

In summary, late rectal toxicities were frequent among patients in this study in Hong Kong. The occurrence of late proctitis was associated with age, V50, V60, and V70; the occurrence overall late rectal toxicities was associated with a history of haemorrhoid, PTV dose, and V70. Multivariable regression analysis suggested that age, V70, and the presence of acute rectal toxicities could predict the occurrence of late proctitis. Clinicians should closely monitor patients for the occurrence of late proctitis if they exhibit these high-risk factors.

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.

Concept or design: BYH Ng, ACK Cheng.
Acquisition of data: BYH Ng.
Analysis or interpretation of data: BYH Ng, ELM Yu, TTS Lau.
Drafting of the article: BYH Ng, ELM Yu, TTS Lau, KS Law.
Critical revision for important intellectual content: BYH Ng, ELM Yu, KS Law, ACK Cheng.

All authors have disclosed no conflict of interest.

The initial abstract was presented at the “ESTRO meets Asia” Conference 2019, Singapore, 6-8 December 2019.

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

This study was approved by the Kowloon West Cluster research ethics committee (Ref KW/EX-19-020(131-08)) and the requirement for patient consent was waived by the committee.

1. Bolla M, Collette L, Blank L, et al. Long-term results with immediate androgen suppression and external irradiation in patients with locally advanced prostate cancer (an EORTC study): a phase III randomised trial. Lancet 2002;360:103-6. Crossref

2. Pilepich MV, Winter K, Lawton C, et al. Androgen suppression adjuvant to definitive radiotherapy in prostate carcinoma—long-term results of phase III RTOG 85-31. Int J Radiat Oncol Biol Phys 2005;61:1285-90. Crossref

3. Michalski J, Moughan J, Purdy J, et al. Effect of standard vs dose-escalated radiation therapy for patients with intermediate-risk prostate cancer: The NRG Oncology RTOG 0126 randomized clinical trial. JAMA Oncol 2018;4:e180039. Crossref

4. Dearnaley D, Syndikus I, Mossop H, et al. Conventional versus hypofractionated high-dose intensity-modulated radiotherapy for prostate cancer: 5-year outcomes of the randomized, non inferiority, phase 3 CHHiP trial. Lancet Oncol 2016;17:1047-60. Crossref

5. Aluwini S, Pos F, Schimmel E, et al. Hypofractionated versus conventionally fractionated radiotherapy for patients with prostate cancer (HYPRO): acute toxicity results from a randomised non-inferiority phase 3 trial. Lancet Oncol 2015;16:274-83. Crossref

6. Zelefsky MJ, Levin EJ, Hunt M, et al. Incidence of late rectal and urinary toxicities after three-dimensional conformal radiotherapy and intensity-modulated radiotherapy for localized prostate cancer. Int J Radiat Oncol Biol Phys 2008;70:1124-9. Crossref

7. Heemsbergen WD, Peeters ST, Koper PC, Hoogeman MS, Lebesque JV. Acute and late gastrointestinal toxicity after radiotherapy in prostate cancer patients: consequential late damage. Int J Radiat Oncol Biol Phys 2006;66:3-10. Crossref

8. Poon DM, Chan SL, Leung CM, et al. Efficacy and toxicity of intensity-modulated radiation therapy for prostate cancer in Chinese patient. Hong Kong Med J 2013;19:407-15. Crossref

9. Delobel J, Gnep K, Ospina JD, et al. Nomogram to predict rectal toxicity following prostate cancer radiotherapy. PLoS One 2017;12:e0179845. Crossref

10. Feng M, Hanlon AL, Pisansky T, et al. Predictive factors for late genitourinary and gastrointestinal toxicity in patients with prostate cancer treated with adjuvant or salvage radiotherapy. Int J Radiat Oncol Biol Phys 2007;68:1417-23. Crossref

11. Fellin G, Fiorino C, Rancati T, et al. Clinical and dosimetric predictors of late rectal toxicity after conformal radiation for localized prostate cancer: results of a large multicenter observational study. Radiother Oncol 2009;93:197-202. Crossref

12. Fuentes-Raspall R, Inoriza J, Rosello-Serrano A, Auñón- Sanz C, Garcia-Martin P, Oliu-Isern G. Late rectal and bladder toxicity following radiation therapy for prostate cancer: predictive factors and treatment results. Rep Pract Oncol Radiother 2013;18:298-303. Crossref

13. Takeda K, Ogawa Y, Ariga H, et al. Clinical correlations between treatment with anticoagulants/antiaggregants and late toxicity after radiotherapy for prostate cancer. Anticancer Res 2009;29:1831-4.

14. Sanguineti G, Agostinelli S, Foppiano S, et al. Adjuvant androgen deprivation impacts late rectal toxicity after conformal radiotherapy of prostate carcinoma. Br J Cancer 2002;86:1743-7. Crossref

15. Fuentes-Raspall R, Inoriza J, Martí-Utzet MJ, Auñón-Sanz C, Garcia-Martin P, Oliu-Isern G. Hyperbaric oxygen therapy for late rectal and bladder toxicity after radiation in prostate cancer patients. A symptom control and qualityof- life study. Clin Oncol (R Coll Radiol) 2012;24:e126. Crossref

16. Jones K, Evans A, Bristow R, Levin W. Treatment of radiation proctitis with hyperbaric oxygen. Radiother Oncol 2006;78:91-4. Crossref

17. Kochhar R, Sriram PV, Sharma SC, Goel RC, Patel F. Natural history of late radiation proctosigmoiditis treated with topical sucralfate suspension. Dig Dis Sci 1999;44:973-8. Crossref

18. Talley NA, Chen F, King D, Jones M, Talley NJ. Shortchain fatty acids in the treatment of radiation proctitis: a randomized double-blind, placebo-controlled, cross-over pilot trial. Dis Colon Rectum 1997;40:1046-50. Crossref

19. Denton A, Andreyev H, Forbes A, Maher EJ. Systematic review for non-surgical interventions for the management of late radiation proctitis. Br J Cancer 2002;87:134-43. Crossref

20. American Joint Committee on Cancer, Prostate Cancer Staging, 7th edition. Available from: https://cancerstaging.org/references-tools/quickreferences/Documents/ProstateSmall.pdf. Accessed 1 Apr 2019.

21. US Department of Health and Human Services, National Institutes of Health, National Cancer Institute, US Government. Common Terminology Criteria for Adverse Events (CTCAE) version 4.03. 2010. Available from: https://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm#ctc_40. Accessed 1 Apr 2019.

22. Abramowitz M, Li T, Buyyounouski MK, et al. The Phoenix definition of biochemical failure predicts for overall survival in patients with prostate cancer. Cancer 2008;112:55-60. Crossref

23. Bansai N, Soni A, Kaur P, Chauhan AK. Exploring the management of radiation proctitis in current clinical practice. J Clin Diagn Res 2016;10:XE01-XE06. Crossref

24. Skwarchuk MW, Jackson A, Zelefsky MJ, et al. Late rectal toxicity after conformal radiotherapy of prostate cancer (I): multivariate analysis and dose-response. Int J Radiat Oncol Biol Phys 2000;47:103-13. Crossref

25. Herold DM, Hanlon AL, Hanks GE. Diabetes mellitus: a predictor for late radiation morbidity. Int J Radiat Oncol Biol Phys 1999;43:475-9. Crossref

26. Miralbell R, Taussky D, Rinaldi O, et al. Influence of rectal volume changes during radiotherapy for prostate cancer: a predictive model for mild-to-moderate late rectal toxicity. Int J Radiat Oncol Biol Phys 2003;57:1280-4. Crossref

With the ageing of the Hong Kong population, clinicians must monitor increasing numbers of patients with neurodegenerative disease. Progressive supranuclear palsy (PSP) is a common form of atypical parkinsonism,1 with a prevalence of up to 18 cases per 100 000 people.1 Pathologically, PSP is characterised by the presence of neurofibrillary tangles, neuropil threads, or both in the basal ganglia and brainstem.1 In patients with classical Richardson’s syndrome, the disease is characterised by early postural instability, falls, vertical gaze palsy, parkinsonism with poor response to levodopa, pseudobulbar palsy, and frontal release signs.2 Increasingly, patients with PSP have been reported to exhibit the following manifestations: parkinsonism, progressive gait freezing, corticobasal syndrome, apraxia of speech, frontal presentation, or cerebellar ataxia.2 Despite advancements in understanding the disease, there remains no approved treatment for PSP.

In addition to the need for accurate diagnosis of PSP, the natural clinical course of the disease is a concern for caregivers.3 4 Important problems for patients with PSP include increased risks of falls, dysphagia, aspiration pneumonia, pressure injuries, caregiver stress leading to institutionalisation, and long-term mortality.1 Given the lack of definitive treatment, it remains prudent for clinicians to educate caregivers regarding the natural course of PSP, which will ensure that caregivers are better prepared to care for their relatives; it will also allow implementation of different methods to avoid long-term complications, and may facilitate discussions of advanced care planning (ACP) at earlier stages of disease.3 In particular, clinicians need to identify the appropriate timing to discuss ACP. Previous studies have shown that the mean age at onset of PSP is 61 to 67.2 years, and that the disease affects both sexes equally; moreover, the median survival ranges from 5.3 to 10.2 years.5 Factors predictive of mortality included age at onset, early clinical milestones (eg, falls, vertical gaze palsy, neck or limb stiffness, dysphagia, and incontinence), cognitive impairment, language impairment, autonomic dysfunction, male sex, and certain subtypes of PSP, such as classical Richardson’s syndrome and pneumonia.6 7 8 9 10 11 12 13 14 15 16

To the best of our knowledge, there have been no studies of the natural clinical course of disease in Chinese patients with PSP. This information is important for clinical treatment and the design of future intervention studies (eg, for the purposes of sample size estimation). In the present study, we hypothesised that bulbar symptoms and pneumonia could predict mortality in patients with PSP. The aims of this study were (1) to calculate the prevalences of motor symptoms, cognitive symptoms, bulbar symptoms, other systemic symptoms, and long-term outcomes (eg, falls, tube feeding, pressure injuries, and institutionalisation) during the clinical course of PSP, and (2) to identify factors predictive of mortality among patients with PSP.

This retrospective study protocol was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB; Approval No. UW 17-483) and New Territories West Cluster Clinical and Research Ethics Committee (NTWC CREC; Approval No. NTWC/CREC/17127); the requirement for informed consent was waived by the review board. This study comprised a retrospective review of the clinical records of all patients who presented to the geriatrics clinics of Queen Mary Hospital and Tuen Mun Hospital between 1 January 2008 and 30 December 2017. All patients had at least 1 year of clinical follow-up and fulfilled the latest Movement Disorder Society Criteria for clinical diagnosis of PSP.2 In addition, magnetic resonance imaging scans showed radiological evidence of midbrain atrophy (Hummingbird sign or Morning Glory sign) in all patients, according to radiological reports prepared by licensed radiologists.1 Twenty-five patients with clinically probable PSP and radiological evidence of midbrain atrophy were considered for inclusion in this study. Four patients were excluded because their clinical history and radiology findings were not suggestive of PSP. Finally, 21 patients were included: 19 had Richardson syndrome variant, one had PSP-corticobasal syndrome, and one had PSP with language impairment. The patient who had PSP with language impairment was previously described.17 Seven and 14 patients were recruited from Queen Mary Hospital and Tuen Mun Hospital, respectively.

Clinical information was retrospectively retrieved from clinical records, including age at onset, age at presentation, age at death, duration of symptoms, level of education, sex, presenting scores on Cantonese version of Mini-Mental State Examination (C-MMSE),18 clinical symptoms, and history of levodopa or dopamine agonists intake or response. Clinical symptoms were clustered into the following categories: motor symptoms (including limb or neck stiffness, slowness of movement, balance impairment, gait impairment, falls, tremor, and vertical gaze palsy), bulbar symptoms (including dysarthria, dysphagia, and drooling), cognitive symptoms (including memory impairment, apathy, apraxia, dysexecutive syndrome, behavioural disinhibition, repetitive motor behaviour, hyperorality, and visual hallucination), and others (including faecal or urinary incontinence, constipation, insomnia, depression, and caregiver stress).6 The dates of development of the above symptom clusters were retrospectively determined.

‘Age at presentation’ was defined as the age at which the patient first presented to the geriatrics clinic. ‘Duration of symptoms’ was defined as the time between the first appearance of clinical symptoms of neurodegenerative disease and the first presentation. ‘Age at onset’ was defined as the difference between the ‘age at presentation’ and the ‘duration of symptoms’. ‘Disease duration’ was defined as the difference between ‘age at onset’ and ‘age at death’ or the last date of follow-up. ‘Time to diagnosis’ was defined as the time between the date of disease onset and the date of diagnosis with PSP. The times to development of the above categories of symptoms were calculated in relation to both disease onset and presentation.

Long-term outcomes were recorded, including falls, dysphagia, pneumonia, pressure sore development, and mortality. ‘Time to event’ was defined by the difference between the onset of clinical symptoms and first appearance of these long-term events. The time to each event from the time of first presentation was also calculated.

These were defined as events that resulted in the patient’s body or body part inadvertently coming to rest on the ground or other surface lower than the body. The dates and numbers of falls were recorded. Geriatric day hospital training was recorded, including the pre-/post-training elderly mobility scale.19 Parkinsonism medications were often titrated in the geriatric day hospital.

A diagnosis of pneumonia was made based on the following criteria: clinical signs and symptoms, white cell count of ≥10×10⁹/L or proportion of neutrophils of ≥80%, fever (body temperature of ≥37.6℃), and new infiltrates or consolidations on chest radiography (X-ray or computed tomography).20 The dates and total numbers of pneumonia diagnoses were recorded.

Any documentation of dysphagia by a speech therapist was recorded; alternatively (if available), reports of video fluoroscopic swallowing studies were obtained. Penetration was defined as the entry of barium material into the airway without passage below the vocal cords; aspiration was defined as the passage of barium material below the level of the vocal cords.21 The dates of diagnosis of dysphagia and tube feeding were recorded.

The locations of pressure injuries and their stages, according to National Pressure Ulcer Advisory Panel guidelines, were recorded.22 The dates of discovery of pressure injuries were recorded.

This was defined as institutionalisation of the patient, regardless of the level of care (eg, personal care facility or health care facility). The dates of institutionalisation were recorded where possible.

The date and cause of death were recorded.

For descriptive statistics, continuous variables with normal distributions were expressed as means ± standard deviations; variables that did not exhibit a normal distribution were expressed as medians (interquartile ranges). Symptom prevalences (cumulative incidences) were estimated using Kaplan-Meier method, with the first presentation defined as time zero. Patients who did not exhibit a particular symptom by the most recent assessment were censored at that point. Median times to clinical events were used as cut-offs to define ‘early’ or ‘late’ development of those events (binary classification). Time zero was consistently defined as the first clinical presentation or onset of disease, while the event time was defined as the number of months from first presentation or disease onset to occurrence of the event. Cox proportional hazards modelling was used to identify factors predictive of mortality based on the above binary classification or clinical events as time-dependent covariates. Pearson correlation coefficients were used to study the correlations between age at presentation and time to mortality (from the date of presentation). A two-tailed P value of <0.05 was considered statistically significant. All statistical analyses were performed using SPSS for Windows (version 24; IBM Corp, Armonk [NY], United States).

Twenty-one patients were included in this analysis, with a total of 1671.4 months of follow-up from onset (1428.4 months from presentation). The baseline demographics are summarised in Table 1. The mean age of the patients at presentation was 67.6 ± 9.4 years, and most patients were men (76.2%). Fifty-seven percent of the patients received another diagnosis at the time of presentation: nine patients were diagnosed with Parkinson’s disease, one patient was diagnosed with Lewy body dementia, one patient was diagnosed with cervical myelopathy, and one patient was diagnosed with myasthenia gravis. None of the patients showed improvement when treated with levodopa or a dopamine agonist. Six patients (28.6%) exhibited dementia at the time of presentation. Seventeen patients (81%) had been referred to the geriatric day hospital for rehabilitation after a median of 20 months from presentation; they showed improvement in elderly mobility scale score (pre-geriatric day hospital elderly mobility scale score vs post-geriatric day hospital elderly mobility scale score: 14 ± 4.6 vs 16 ± 4.3, respectively, P=0.02). Fifteen patients (71.4%) died during follow-up with mean survival of 6.1 ± 2.6 years from onset (5.2 ± 3.2 years from presentation); 12 of these 15 patients (80%) died of pneumonia, while two (13.3%) died of sudden cardiac arrest.

Among the categories of potential symptoms, motor symptoms were most prevalent during initial presentation (specific symptoms affected up to 33.3% of patients) [Fig 1]. Motor manifestations were among the earliest clinical features observed in patients with PSP (online supplementary Appendix). The most frequent motor symptoms at the time of presentation were limb stiffness (33.3%) and gait impairment (28.6%). Vertical gaze palsy was present in 19% of patients at the time of presentation, but eventually affected all patients (100%). The prevalence of gait impairment increased rapidly, such that ≥80% of the patients were affected within 3 years after presentation. All motor features showed increased in prevalence over time, with final prevalences ranging from 47.6% to 100%. Regarding bulbar symptoms, dysarthria was the most frequent presenting symptom (9.5%). Both dysarthria and dysphagia reached 100% prevalence over time (Fig 1).

Regarding cognitive symptoms, memory impairment and apathy were the two most frequent presenting symptoms, with prevalences of 33.3% and 23.8%, respectively (Fig 2). The respective prevalences of memory impairment and apathy increased to >65% and >45% over time. The prevalence of dysexecutive syndrome reached 28.6% during the clinical course of disease. Other cognitive symptoms relevant to patients with PSP showed lower prevalence, including apraxia, behavioural disinhibition, repetitive motor behaviour, hyperorality, and visual hallucination; the highest prevalence for any of these symptoms was 9.5% throughout the clinical course of disease. The degree of caregiver stress also increased with progression of disease, such that it reached approximately 40% within 5 years after initial presentation.

Regarding systemic symptoms, faecal and urinary incontinence showed the highest prevalences (both reached approximately 40%), particularly in the later stages of disease (Fig 2). Regarding long-term outcomes, the prevalences of aspiration pneumonia and dysphagia requiring Ryle’s tube insertion both reached 100% over time (Fig 2).

Using median time to clinical events as cut-off (binary classification) [online supplementary Appendix] and with analysis in a Cox proportional hazards model, our results showed that earlier development of vertical gaze palsy (hazard ratio [HR]=4.4, 95% confidence interval [CI]=1.4-13.9, P=0.01) and earlier development of pneumonia (HR=10.9, 95% CI=2.2-53.3, P=0.003) were predictive of mortality from disease onset (Table 2). Multivariate analysis showed that earlier development of vertical gaze palsy (HR=3.8, 95% CI=1.1-13.0, P=0.04) and earlier development of pneumonia (HR=10.4, 95% CI=1.9-55.7, P=0.006) were predictive of mortality from disease onset (Table 2). Earlier development of vertical gaze palsy (HR=3.3, 95% CI=1.1-10.5, P=0.01), earlier development of dysphagia (HR=3.7, 95% CI=1.0-12.9, P=0.04), and earlier development of pneumonia (HR=10.6, 95% CI=2.2-52.0, P=0.004) were predictive of mortality from disease presentation (Table 2). Multivariate analysis showed that only earlier development of pneumonia (HR=9.9, 95% CI=1.9-51.8, P=0.007) was predictive of mortality from presentation (Table 2). The number of episodes of pneumonia was also predictive of mortality in patients with PSP, indicating that pneumonia is a major cause of mortality (Table 2). There was a significant negative correlation between the age at presentation and time to mortality from presentation (Pearson correlation=-0.54, P=0.04).

Using clinical events as time-dependent covariates in Cox modelling for prediction of mortality, we found that apathy, dysphagia, Ryle’s tube feeding, pneumonia, and pressure injuries were predictive of mortality from both disease onset and presentation (Table 3). Caregiver stress was only predictive of mortality from presentation (Table 3).

An accurate diagnosis of PSP is important for management of the disease in affected patients. However, only 43% of the patients in this study received a correct diagnosis at the time of initial presentation. This is potentially because vertical gaze palsy was not present initially and only developed during clinical follow-up (median time to develop, 19.6 months; online supplementary Appendix). In addition, 43% of patients were initially misdiagnosed with Parkinson’s disease; this group of patients may have had PSP with parkinsonism.2 Clinicians should regularly assess patients with parkinsonism for the presence of any vertical gaze palsy or poor response to levodopa, in order to correctly identify patients with PSP. Our reported mean time to diagnosis of 3 years was similar to the duration reported in previous studies (mean, 3.1-4 years).1 2 5 6 7 8 9 10 11 12 13 14 15 16 17 23 24 25 26 27 28

With regard to clinical features, our cohort of patients exhibited early development of motor symptoms, which was consistent with previous studies.1 2 5 6 7 8 9 10 11 12 13 14 15 16 17 23 24 25 26 27 28 Our cohort of patients showed evidence of improved mobility, as reflected by changes in elderly mobility scale score after training in the geriatric day hospital, with a median time to referral of 20 months from the initial clinical presentation. Patients with PSP should be referred to a physiotherapist and an occupational therapist for fall assessment, as well as guidance regarding the potential need for walking aids. Relatives should be educated to ensure close monitoring of environmental risks for falls. The patients in our cohort showed relatively early development of memory problems at a median duration of 9 months, which contrasted with the mean duration of 12 months observed in another study.6 There may have been bias in the current cohort, which recruited patients with PSP from a geriatrics clinic whereas the patients in the previous study were recruited from a neurology clinic.6

A previous meta-analysis showed mixed results with regard to whether the age at onset of PSP was predictive of mortality.5 Pooled results from six studies showed no prognostic effect of yearly increases in age at disease onset in either univariate or multivariate analyses.5 However, pooled results from nine studies using median age at onset as cut-off showed a pooled HR of 1.75 (95% CI=1.32-2.32) in multivariate analysis.5 Other predictors of mortality found in the present study, including vertical gaze palsy, dysphagia, pneumonia, or pressure injuries, were previously reported in other studies.6 7 8 9 10 11 12 13 14 15 16 It remains unknown whether resolution of dysphagia in patients with PSP can prevent pneumonia and reduce mortality. It is important for clinicians to refer patients with PSP involving dysphagia to a speech therapist for advice regarding food texture and appropriate swallowing posture (ie, chin-tuck) [Table 4].29 During the clinical course of disease, approximately 40% of primary caregivers complained of caregiver stress, which was also determined to be a significant predictor of mortality. Patient aggression and depression have been reported as sources of stress.4 When these positive predictors of mortality appear, clinicians should consider discussing ACP with patients or caregivers (Table 4).

The finding of apathy as a time-dependent covariate for prediction of mortality is notable. Recently, apathy was found to predict survival in a cohort of 124 patients with syndromes associated with frontotemporal lobar degeneration (including 35 patients with PSP, mean age of 72.2 ± 8.5 years).30 The development of apathy in patients with PSP was related to brainstem, midbrain, and frontal atrophy.30 It remains unknown whether apathy accelerates the decline to death or indirectly signifies the degree of brainstem degeneration, including the development of dysphagia, which is also related to greater mortality. Future clinical trials may consider the use of therapeutic measures to address apathy, in order to assess their impacts on the survival of patients with PSP.

Because there is currently no disease-modifying treatment for patients with PSP, ACP is an important component of clinical care, for which patients and their caregivers can reach a consensus during the clinical course of disease.4 In addition, symptomatic care plays an important role. Symptoms relevant to patients with PSP include dystonia, drooling, gaze palsy (also known as reduced blinking), constipation, and apathy.4 Drooling could be managed by the administration of sublingual atropine drops (Table 4).4 Reduced blinking could be managed by frequent application of lubricating eyedrops. Gaze palsy could be managed by the use of prisms or audiobooks (Table 4).4 Apathy could be minimised by addressing sensory deficits, such as through the use of eyeglasses or hearing aids (Table 4).4 Up to 75% of patients with PSP could be discharged home after stabilisation of symptoms.4

There are multiple limitations in the current study. First, it was a retrospective study involving reviews of clinical charts, and may be biased due to inconsistent documentation of symptoms. Second, there was a limited number of patients included, none of whom had autopsy and pathological confirmation of their diagnosis; however, we had radiological evidence of PSP. Third, our descriptive statistical results should be regarded as preliminary findings; only limited variables could be included in our Cox proportional hazards modelling for survival analysis. Notably, no specific scales were used to assess the severity of parkinsonism or other symptoms, including response to levodopa; most of our evaluations were subjective. Sequential C-MMSE scores were not recorded; thus, we were unable to examine the development of dementia over time. Fourth, we only included patients attending the geriatrics clinic; therefore, our cohort may not be fully representative of patients with PSP who present to most neurology clinics. Finally, the limited numbers of patients precluded stratified analyses based on subtypes of PSP.

In conclusion, important clinical milestones, including the development of dysphagia, vertical gaze palsy, significant caregiver stress, pressure injuries, and pneumonia, may be used to guide the ideal timing for discussions of ACP for patients with PSP, in order to facilitate long-term care.

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.

Concept or design: YF Shea, ACK Shum.
Acquisition of data: YF Shea, ACK Shum.
Analysis or interpretation of data: All authors.
Drafting of the article: YF Shea, ACK Shum.
Critical revision for important intellectual content: All authors.

The authors have no conflicts of interest to disclose.

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

This retrospective study protocol was approved by the Institutional Review Board of the University of Hong Kong/ Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB; Ref UW 17-483) and New Territories West Cluster Clinical and Research Ethics Committee (NTWC CREC; Ref NTWC/CREC/17127). The requirement for informed consent was waived by the review board.

1. Boxer AL, Yu JT, Golbe LI, Litvan I, Lang AE, Höglinger GU. Advances in progressive supranuclear palsy: new diagnostic criteria, biomarkers, and therapeutic approaches. Lancet Neurol 2017;16:552-63. Crossref

2. Höglinger GU, Respondek G, Stamelou M, et al. Clinical diagnosis of progressive supranuclear palsy: The movement disorder society criteria. Mov Disord 2017;32:853-64. Crossref

3. Luk JK, Chan FH. End-of-life care for advanced dementia patients in residential care home—a Hong Kong perspective. Ann Palliat Med 2018;7:359-64. Crossref

4. Wiblin L, Lee M, Burn D. Palliative care and its emerging role in multiple system atrophy and progressive supranuclear palsy. Parkinsonism Relat Disord 2017;34:7-14. Crossref

5. Glasmacher SA, Leigh PN, Saha RA. Predictors of survival in progressive supranuclear palsy and multiple system atrophy: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry 2017;88:402-11. Crossref

6. Arena JE, Weigand SD, Whitwell JL, et al. Progressive supranuclear palsy: progression and survival. J Neurol 2016;263:380-9. Crossref

7. Golbe LI, Davis PH, Schoenberg BS, Duvoisin RC. Prevalence and natural history of progressive supranuclear palsy. Neurology 1988;38:1031-4. Crossref

8. Nath U, Ben-Shlomo Y, Thomson RG, Lees AJ, Burn DJ. Clinical features and natural history of progressive supranuclear palsy: a clinical cohort study. Neurology 2003;60:910-6. Crossref

9. Oliveira MC, Ling H, Lees AJ, Holton JL, De Pablo-Fernandez E, Warner TT. Association of autonomic symptoms with disease progression and survival in progressive supranuclear palsy. J Neurol Neurosurg Psychiatry 2019;90:555-61. Crossref

10. Cosseddu M, Benussi A, Gazzina S, et al. Natural history and predictors of survival in progressive supranuclear palsy. J Neurol Sci 2017;382:105-7. Crossref

11. Chiu WZ, Kaat LD, Seelaar H, et al. Survival in progressive supranuclear palsy and frontotemporal dementia. J Neurol Neurosurg Psychiatry 2010;81:441-5. Crossref

12. Dell’Aquila C, Zoccolella S, Cardinali V, et al. Predictors of survival in a series of clinically diagnosed progressive supranuclear palsy patients. Parkinsonism Relat Disord 2013;19:980-5. Crossref

13. Santacruz P, Uttl B, Litvan I, Grafman J. Progressive supranuclear palsy: a survey of the disease course. Neurology 1998;50:1637-47. Crossref

Litvan I, Mangone CA, McKee A, et al. Natural history of progressive supranuclear palsy (Steele-Richardson-Olszewski syndrome) and clinical predictors of survival: a clinicopathological study. J Neurol Neurosurg Psychiatry 1996;60:615-20. Crossref

15. O’Sullivan SS, Massey LA, Williams DR, et al. Clinical outcomes of progressive supranuclear palsy and multiple system atrophy. Brain 2008;131:1362-72. Crossref

16. Papapetropoulos S, Gonzalez J, Mash DC. Natural history of progressive supranuclear palsy: a clinicopathologic study from a population of brain donors. Eur Neurol 2005;54:1-9. Crossref

17. Shea YF, Ha J, Chu LW. Progressive supranuclear palsy presenting initially as semantic dementia. J Am Geriatr Soc 2014;62:2459-60. Crossref

18. Chiu HF, Lee HC, Chung WS, Kwong PK. Reliability and validity of Cantonese version of the Mini-Mental State Examination: a preliminary study. J Hong Kong Col Psych 1994;4:25-8.

19. Prosser L, Canby A. Further validation of the elderly mobility scale for measurement of mobility of hospitalized elderly people. Clin Rehabil 1997;11:338-43. Crossref

20. Tomita S, Oeda T, Umemura A, et al. Impact of aspiration pneumonia on the clinical course of progressive supranuclear palsy: a retrospective cohort study. PLoS One 2015;10:e0135823. Crossref

21. Rosenbek JC, Robbins JA, Roecker EB, Coyle JL, Wood JL. A penetration-aspiration scale. Dysphagia 1996;11:93-8. Crossref

22. Edsberg LE, Black JM, Goldberg M, McNichol L, Moore L, Sieggreen M. Revised national pressure ulcer advisory panel pressure injury staging system: revised pressure injury staging system. J Wound Ostomy Continence Nurs 2016;43:585-97. Crossref

23. Testa D, Monza D, Ferrarini M, Soliveri P, Girotti F, Filippini G. Comparison of natural histories of progressive supranuclear palsy and multiple system atrophy. Neurol Sci 2001;22:247-51. Crossref

24. Diroma C, Dell’Aquila C, Fraddosio A, et al. Natural history and clinical features of progressive supranuclear palsy: a clinical study. Neurol Sci 2003;24:176-7. Crossref

25. Goetz CG, Leurgans S, Lang AE, Litvan I. Progression of gait, speech and swallowing deficits in progressive supranuclear palsy. Neurology 2003;60:917-22. Crossref

26. Jecmenica-Lukic M, Petrovic IN, Pekmezovic T, Kostic VS. Clinical outcomes of two main variants of progressive supranuclear palsy and multiple system atrophy: a prospective natural history study. J Neurol 2014;261:1575-83. Crossref

27. Litvan I, Kong M. Rate of decline in progressive supranuclear palsy. Mov Disord 2014;29:463-8. Crossref

28. Ou R, Liu H, Hou Y, et al. Executive dysfunction, behavioral changes and quality of life in Chinese patients with progressive supranuclear palsy. J Neurol Sci 2017;380:182-6. Crossref

29. Luk JK, Chan DK. Preventing aspiration pneumonia in older people: do we have the ‘know-how’? Hong Kong Med J 2014;20:421-7. Crossref

30. Lansdall CJ, Coyle-Gilchrist IT, Vázquez Rodríguez P, et al. Prognostic importance of apathy in syndromes associated with frontotemporal lobar degeneration. Neurology 2019;92:e1547-57. Crossref