For rural residents, who account for approximately 41% of China’s population, primary healthcare is the main source of medical care.1 To meet their healthcare needs, China promoted a tiered medical system in 2015 to encourage people to fully utilise primary healthcare.2 3 Nevertheless, while the government invests various resources, primary healthcare in rural areas remains underused.4 5 6 From 2015 to 2018, the total number of out-patient visits in rural primary healthcare institutions decreased from 2.90 billion to 2.70 billion, while the number of hospital visits increased from 3.10 billion to 3.60 billion.7

Disparity exists in the utilisation of primary healthcare across different regions of China. Compared with rural residents in eastern China, the utilisation of primary healthcare among those in western China is relatively low.8 9 The lack of medical resources and sparseness of the land in western China contributes to the inconvenience of accessing primary healthcare among rural residents.10 11 12 In addition to the difference between eastern and western China, differences in the utilisation of primary healthcare exists among the provinces in western China.11 Although many studies have reported on the utilisation of primary healthcare in Southwest China, most have focused on the perspective of rural residents while neglecting the importance of providers, who play important roles in primary healthcare.3 4 Therefore, research concerning village healthcare providers in Southwest China could be conducive to understanding the utilisation of primary healthcare.

Southwest China is home to more ethnic minority village healthcare providers than other areas in China.11 To investigate village healthcare providers in Southwest China, the ethnicity of providers, which may be linked to the utilisation of healthcare, is a factor that cannot be ignored. Previous studies have illustrated that ethnic minority providers are more likely to attract patients from the same ethnicity, and this finding has been attributed to the patients’ preference instead of the capability of minority providers.13 14 Existing studies were mainly conducted outside China or were related to providers who performed traditional Chinese medicine.13 14 15 Therefore, knowledge regarding whether differences exist in the utilisation of primary healthcare among Chinese ethnic minority providers and Han Chinese majority providers is limited.

Clinical competence may influence the utilisation of healthcare and serves as a practical way to measure a doctor’s working performance and quality.16 17 Some studies in China have shown that the ethnicity of medical students might be related to their future clinical competence.18 19 20 21 22 Studies conducted in Southwest China confirmed the underdeveloped clinical competence of village providers, but most studies failed to distinguish the ethnicity of the healthcare providers.23 24 25 26 Whether ethnic minority healthcare providers in rural areas of China have underdeveloped clinical competence remains unclear.

Considering the above, the purpose of the present study was to investigate the utilisation of primary healthcare in western China. In particular, the aims were to clarify whether the ethnicity of healthcare providers affects utilisation of primary healthcare; whether there are differences in clinical competence among different ethnic groups; and whether clinical competence of healthcare providers affects utilisation of primary healthcare.

This study was a cross-sectional survey conducted in three prefectures (ie, prefecture-level cities) in Yunnan Province, an economically developing area in Southwest China. In 2017, the per capita gross domestic product in Yunnan Province was US$5068, which is lower than the national average (US$8777).27 The total population in Yunnan Province was 47.71 million, and the proportion of the rural population in Yunnan Province was 53.31%, which is much higher than the overall proportion of rural population in China (41.48%).27 The three prefectures included in our study have a total rural population of 6.50 million, accounting for 20.00% of the total rural population in Yunnan Province.

To investigate the utilisation of primary healthcare in rural China, we conducted a cross-sectional study in 330 village clinics (VCs), representing the first tiers of China’s three-tiered rural health system.4 In summary, we selected a random sample of 330 healthcare providers from three prefectures in three steps (Fig). First, we selected 10 counties (in three prefectures) at random, after excluding three urban counties and 13 counties with a minority population greater than 20%. Second, we used probability proportional to size sampling to randomly select 330 VCs proportional to the number of VCs in each county. Finally, we asked each clinic to list all staff serving in the clinic and describe their responsibilities. Considering the measurement of different types of medical practitioners, we excluded healthcare providers other than Western medicine practitioners (ie, traditional Chinese medicine practitioners or those responsible for public health services only). Then, we randomly selected one of the remaining village healthcare providers as our sample.

The data collection was carried out in July 2017 by trained investigators. The survey consisted of a clinic form administered to the head of the VC and a clinician form administered to providers in each clinic verbally.

The clinic form (Supplementary Table 1) was used to collect basic information regarding the VCs, including the number of equipment per clinic, whether the drugs sold by the clinics met the zero price difference of medicine (a policy requires no mark-ups above the cost of drugs),6 the number of clinics within 5 km, the number of out-patient visits per clinic, and the number of providers per clinic.

The clinician form (Supplementary Table 2) consisted of two parts. The first part was used to obtain information regarding the provider’s demographics, working time allocation and income. This part included age, sex, ethnicity, basic salary, local residence (whether he/she was born and raised in the sample village), and time spent performing public health services. The second part was used to obtain detailed information on the providers’ in-service training participation in 2016 (the year before the survey year) and their clinical competence, including education level, certificate in rural medicine or higher, length of experience, and medical study.

The primary healthcare system provides generalist clinical care and basic public health services.1 China has promoted the three-tiered healthcare system to improve the use of primary healthcare. In rural China, the three-tiered healthcare system consists of VCs, township health centres, and county hospitals. Township health centres and VCs play a role in primary healthcare, and VCs mainly provide out-patient services under common clinical conditions. To assess the utilisation of VCs, the investigators asked the heads of the VCs to estimate (on average) the total number of out-patient visits during the previous month. The utilisation of the VCs was calculated using data related to the total number of out-patient visits, intramuscular injection visits, intravenous infusion visits, and number of healthcare providers in the clinics. Specifically, the utilisation of VCs in our research is measured on a per-provider basis. Thus, the utilisation of VCs equals the total number of out-patient visits divided by the number of healthcare providers.

Providers’ clinical competence is the quality of healthcare providers pertaining to medical knowledge, treatment quality, medical experience, medical study background, and medical training.28 In our study, we evaluated the providers’ clinical competence in the following dimensions: education above college level (measured medical knowledge); attainment of certificates in rural medicine or higher (measured treatment quality); length of experience (measured medical experience); participation in medical studies (measured medical study background); and completion of in-service training (measured medical training).

In clinical competence, multiple characteristics are often correlated; multicollinearity is a limitation of applying a multiple logistic regression analysis to measure clinical competence. A principal component analysis is often used to address this issue by transforming the data into one or two dimensions that serve as a summary of the characteristics, such as constructing an index.29 Therefore, we constructed the Clinical Competence Index to assess the overall clinical competence of the providers using a principal component analysis approach.

To explore the relationships between the healthcare providers and the utilisation of VCs, we used a multiple logistic regression. We conducted three types of regressions, and the outcome variable was out-patient visits per doctor. In the first regression, we measured the relationship between the village providers’ ethnicity and the utilisation of VCs. In the second regression, we included only the variables measuring the providers’ clinical competence. This relationship might vary across different ethnicities, and a heterogeneity analysis is needed. In the third regression, we performed a heterogeneity analysis by including the interaction term ‘Clinical Competence Index ethnic minority providers’ to measure the providers’ clinical competence and its relationship with the utilisation of VCs across different ethnicities.

In each regression mentioned above, we assessed the correlations with a fixed set of facility-level and provider-level characteristics. These characteristics included the number of equipment per clinic, whether the drugs sold by the clinics met the zero price difference of medicine, the number of clinics within 5 km, the percentage of minority residents in the town, the providers’ sex, salary, local residency, and time spent providing public health services. Notably, in all regressions, we also controlled for county-fixed effects.

All statistical analyses were performed using Stata version 15.0 statistical software (Stata Corp, College Station [TX], United States). The results with a P value <0.05 were considered statistically significant.

The questionnaire surveys were sent to 330 VCs and 330 healthcare providers, and valid results were obtained from all (participation rate 100%). There were no significant differences in terms of the amount of clinic equipment (P=0.395) and the implementation of the zero price difference of medicine (P=0.396) among the VCs (Table 1). However, the proportion of minority residents in the towns significantly differed with type of VCs (P<0.001). We also found that VCs where Han Chinese providers worked faced more competition, as the number of clinics within 5 km of these clinics was significantly higher than that for clinics where ethnic minority providers worked (P=0.002). The average number of healthcare providers in the VCs was 2.79. There was a significant difference in the number of providers across clinics (P=0.016), and this difference was mainly due to the number of providers who treat patients (P=0.044). Among the 330 VCs, mean number of out-patient visits per doctor was 146. The results indicate that the Han Chinese providers conducted a significantly higher average number of out-patient visits than the ethnic minority providers (151 vs 101; P=0.008) [Table 1]. In addition, we documented intramuscular injection visits and intravenous infusion visits. There was also a significant difference between the ethnic minority and Han Chinese majority providers in intravenous infusion visits (P=0.023), but no significant difference was found in intramuscular injection visits (P=0.834).

Most healthcare providers were male (64.5%), and 81.8% of providers were local residents (Table 2). The mean ± standard deviation annual salary of all providers was US$926±399, with no significant difference between the ethnic minority and Han Chinese majority providers (P=0.327). More than half of the healthcare providers devoted close to 60% of their work time to providing public health services (P=0.577).

Among the 330 providers, 26.4% had a college degree or higher, and the proportion among the Han Chinese providers was similar to that (26.4%) among the ethnic minority providers (25.7%, P=0.927) [Table 2]. In total, 91.9% of the Han Chinese providers were confirmed to have at least certificates in rural medicine; however, the proportion was 74.3% among the ethnic minority providers (P=0.001). In addition, the length of experience of the Han Chinese providers was significantly higher than that of the ethnic minority providers (P=0.013). The proportion of full-time medical studies conducted by all providers was 52.7%, and more ethnic minority providers than Han Chinese providers received in-service medical training (P=0.085).

The ethnicity of the providers was negatively associated with the utilisation of VCs (odds ratio [OR]=0.53, 95% confidence interval [CI]=0.29-0.96; P=0.037) [Table 3]; the providers’ clinical competence was positively associated with the utilisation of VCs (OR=1.49, 95% CI=1.12-2.00; P=0.007) [Table 3]. To better examine this relationship, we performed a heterogeneity analysis (Table 4). The results suggest that ethnic minority providers were likely to have underdeveloped clinical competence, which could further limit the utilisation of VCs (OR=0.45, 95% CI=0.23-0.89; P=0.022) [Table 4]. We also measured the determinants of intravenous infusion visits, and the results suggest that the providers’ clinical competence (OR=1.43, 95% CI=1.16-1.77; P=0.001) [Supplementary Table 4] was associated with the utilisation of VCs.

The data enabled an analysis of the utilisation of primary healthcare in rural areas in Southwest China. In general, there are three key findings in this study. First, we found significant differences between Han Chinese and ethnic minority providers in the utilisation of VCs. Second, compared with Han Chinese providers, ethnic minority providers in our sample had poorer clinical competence in two dimensions (possession of rural physician certificate or length of experience). Finally, our results indicate that underdeveloped clinical competence is a factor responsible for the lower utilisation of VCs among ethnic minority providers.

Our survey data show that the average number of out-patient visits in the sampling VCs is 146 per month per doctor, which is lower than the number of out-patient visits in eastern China (188 per month per doctor, 2017).9 Consistent with previous studies, we believe that the inconvenience of accessing primary healthcare among rural residents contributes to this difference.23 Compared with Yunnan Province (116 per month per doctor, 2017), the utilisation of VCs in the sampling area was higher.9 Based on our investigation, the high utilisation of VCs in the sampling area could be explained by both the population density and local economic development. According to the China Statistical Yearbook, the rural residents in the three prefectures we studied accounted for 20.00% of the total rural residents in Yunnan Province, which is quite large compared with the rate in other regions.30 The gross domestic product of the sampling area is also relatively high in Yunnan Province.27

However, the number of out-patient visits to VCs where the main providers are ethnic minorities is significantly lower than that of VCs where the main providers are Han Chinese individuals. Previous studies have shown that patients visiting providers of their own race were more satisfied and deliberately chose providers of their own race because of personal preference and language issues.13 14 In contrast to these studies, we excluded the preference and language issues of patients. First, their studies were generally conducted in large hospitals with many providers, but the average number of providers in our sampled VCs was approximately two, limiting the patients’ choices. Second, regarding language, most (81.8%) providers in our study were local residents who were fluent in the local dialect. These providers rarely have communication problems with their patients. After the above exclusions, an investigation of the characteristics of different ethnic providers was performed. Consistent with previous studies, there was no significant difference in provider characteristics and income, and the amount of clinic equipment and implementation of zero price difference of the medicine did not differ.19 20

Furthermore, we find significant differences between ethnic minority and Han Chinese village providers in their clinical competence. As a crucial aspect of primary healthcare services in rural China, village providers are obligated to provide qualified medical services, which require abundant clinical competence.31 32 However, based on our evidence, even if rural patients in Southwest China were asked to follow the instructions of policymakers and seek care primarily in VCs, they would visit higher-level facilities with a relatively high cost considering the local providers’ poor clinical competence. In fact, this type of situation is already very common in rural China.3 16 25

Under such circumstances, our findings imply that the difference in the utilisation of VCs might be related to ethnic minority providers’ underdeveloped clinical competence. Despite the small number of previous studies, the current evidence is consistent with the main findings.1 24 On the one hand, early studies suggested that the quality of health providers limited the utilisation of primary healthcare.3 4 16 On the other hand, ethnic minority providers usually experience less supportive learning environments during their medical studies, which might help explain the lag in their clinical competence.33 34 35 Therefore, we believe that uniformly improving the clinical competence of village providers, especially that of ethnic minority village providers, is conducive to improving the utilisation of primary healthcare.1 36

The Government of the People’s Republic of China understands the benefits of improving the clinical competence of village providers. A series of human resource policies for health launched in 2010 had a positive impact in rural China.37 Among these policies, the encouragement of external training for healthcare providers has been proven effective and necessary.36 38 39 In-service training for village providers could help these providers be informed of the latest knowledge and skills to manage diseases, which could significantly improve the quality of their medical services.36 The government has implemented actions to encourage the training of ethnic minorities since 2009.40 Thus, the appropriate introduction of re-education and more medical training should be adopted among ethnic minority providers.

Considering previous studies with analyses based mainly on the utilisation of healthcare from the patient perspective, our study might be the first investigation to examine healthcare providers’ ethnicity and clinical competence.10 11 32 In addition, our quantitative data include a rich set of detailed information of VCs regarding out-patient visits and providers’ clinical competence. We hope that our findings offer key insights into the utilisation of primary healthcare in rural China.

Our study has two main limitations. First, the participants in this study were recruited from rural Southwest China, so unavoidably, the nationwide validity of our findings is limited. Second, the results were limited by the cross-sectional nature of the study, and no causal effect between the providers’ clinical competence and the utilisation of primary healthcare was detected. Future research could investigate how different ethnic providers influence the utilisation of VCs and seek to adopt multiple measures to reduce bias in investigations.

In conclusion, this study reveals differences in the utilisation of primary healthcare between ethnic minority providers and Han Chinese providers at VCs in rural areas in Southwest China. Notably, the results indicate that higher clinical competence is more likely to drive the utilisation of VCs. We believe that the results of this study provide compelling evidence that ethnic minority healthcare providers in Southwest China require further enhancement with respect to their clinical competence.

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

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

As an adviser of the journal, Y Shi was not involved in the peer review process. Other authors have disclosed no conflicts of interest.

H Yi received funding for this study from the Health and Hope Fund of the Business Development Center of the Red Cross Society of China and UCB of Belgium. Y Shi received funding from 111 Project (Grant number B16031), and Q Gao received funding from Innovation Capability Support Program of Shaanxi (Grant number 2022KRM007). The funders had no role in study design, data collection/analysis/interpretation, or manuscript preparation.

The study was approved by the Peking University Institutional Review Board (Ref IRB 00001052-17033). The board approved the verbal consent procedure. Informed consent from all respondents as a requirement for completing the survey.

1. Li X, Lu J, Hu S, et al. The primary health-care system in China. Lancet 2017;390:2584-94. Crossref

2. Liu Q, Wang B, Kong Y, Cheng KK. China’s primary healthcare reform. Lancet 2011;377:2064-6. Crossref

3. Li YN, Nong D, Wei B, Feng QM, Luo HY. The impact of predisposing, enabling, and need factors in utilization of health services among rural residents in Guangxi, China. BMC Health Serv Res 2016;16:592. Crossref

4. Sylvia S, Xue H, Zhou C, et al. Tuberculosis detection and the challenges of integrated care in rural China: A cross-sectional standardized patient study. PLoS Med 2017;14:e1002405. Crossref

5. Dong X, Liu L, Cao S, et al. Focus on vulnerable populations and promoting equity in health service utilization—an analysis of visitor characteristics and service utilization of the Chinese community health service. BMC public health 2014;14:503. Crossref

6. Li Z. Healthcare reform and the development of China’s rural healthcare: a decade’s experience [in Chinese]. Chinese Rural Economy 2019;9:1-22.

7. National Bureau of Statistics of China, PRC Government. China Statistical Yearbook 2019. Available from: http://www.stats.gov.cn/tjsj/ndsj/2019/indexch.htm. Accessed 20 Oct 2020.

8. Sun J, Luo H. Evaluation on equality and efficiency of health resources allocation and health services utilization in China. Int J Equity Health 2017;16:127. Crossref

9. China Statisitcs Press. China Health Statistical Yearbook 2018. Available from: https://data.cnki.net/area/Yearbook/Single/N2019030282?z=D09. Accessed 9 May 2020.

10. Zhang X, Wu Q, Shao Y, Fu W, Liu G, Coyte PC. Socioeconomic inequities in health care utilization in China. Asia Pac J Public Health 2015;27:429-38. Crossref

11. Wang M, Fang H, Bishwajit G, Xiang Y, Fu H, Feng Z. Evaluation of rural primary health care in Western China: a cross-sectional study. Int J Environ Res Public Health 2015;12:13843-60. Crossref

12. Shi Y, Xue H, Wang H, Sylvia S, Medina A, Rozelle S. Measuring the quality of doctors’ health care in rural China: an empirical research using standardized patients [in Chinese]. Studies in Labor Econ 2016;4:48-71.

13. van Zanten M, Boulet JR, McKinley DW. The influence of ethnicity on patient satisfaction in a standardized patient assessment. Acad Med 2004;79(10 Suppl):S15-7. Crossref

14. LaVeist TA, Nuru-Jeter A, Jones KE. The association of doctor-patient race concordance with health services utilization. J Public Health Policy 2003;24:312-23. Crossref

15. Li Z, Li Y, Cui J. Development history of Chinese minority traditional medicine [in Chinese]. Med Philosophy (Humanistic & Soc Med Ed) 2011;32:78-81.

16. Wu D, Lam TP. Underuse of primary care in China: the scale, causes, and solutions. J Am Board Fam Med 2016;29:240-7. Crossref

17. Sandberg J. Understanding human competence at work: an interpretative approach. Acad Manage J 2000;43:9-25. Crossref

18. Liu W, Ding Y, Yang X. The determinants of rural labor migration in minority regions [in Chinese]. Population J 2015;37:102-12.

19. Shi L, Sai D. An empirical analysis of income inequality between a minority and the majority in urban China: the case of Ningxia Hui Autonomous Region. Rev Black Polit Econ 2013;40:341-55. Crossref

20. Li L, Li XG. Ethnicity, social capital and income gap: a comparative study of the Northwest ethnic group and the Western Han nationality [in Chinese]. J Sun Yat-sen University (Social Science Edition) 2016;56:161-71.

21. Gustafsson B, Sai D. Temporary and persistent poverty among ethnic minorities and the majority in rural China. Rev Income Wealth 2009;55:588-606. Crossref

22. China National Knowledge Infrastructure. BMJ; Why are ethnic minority doctors less successful than white doctors? Available from: https://schlr.cnki.net/en/Detail/index/SPQDLAST/SPQD48BA99708EE31C6710E1A02B2C34186A. Accessed 12 Jun 2020.

23. Li Y, Chi I, Zhang K, Guo P. Comparison of health services use by Chinese urban and rural older adults in Yunnan province. Geriatr Gerontol Int 2006;6:260-9. Crossref

24. Anand S, Fan VY, Zhang J, et al. China’s human resources for health: quantity, quality, and distribution. Lancet 2008;372:1774-81. Crossref

25. Meng Q, Liu X, Shi J. Comparing the services and quality of private and public clinics in rural China. Health Policy Plan 2000;15:349-56. Crossref

26. Zhang Q, Chen J, Yang M, et al. Current status and job satisfaction of village doctors in western China. Medicine (Baltimore) 2019;98:e16693. Crossref

27. Chinese Statistics Press. National Accounts. In: China Statistical Yearbook 2017. Available from: http://www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm. Accessed 6 Jun 2020.

28. Wass V, Van der Vleuten C, Shatzer J, Jones R. Assessment of clinical competence. Lancet 2001;357:945-9. Crossref

29. Lever J, Krzywinski M, Altman N. Principal component analysis. Nat Methods. 2017;14:641-2. Crossref

30. Chinese Statistics Press. Population. In: China Statistical Yearbook 2017. Available from: http://www.stats.gov.cn/tjsj/ndsj/2017/indexch.htm. Accessed 14 Jun 2020.

31. Li X, Liu J, Huang J, Qian Y, Che L. An analysis of the current educational status and future training needs of China’s rural doctors in 2011. Postgrad Med J 2013;89:202-8. Crossref

32. Dias SF, Severo M, Barros H. Determinants of health care utilization by immigrants in Portugal. BMC Health Serv Res 2008;8:207. Crossref

33. Myers SL, Xiaoyan G, Cruz BC. Ethnic minorities, race, and inequality in China: a new perspective on racial dynamics. Rev Black Polit Econ 2013;40:231-44. Crossref

34. Esmail A, Everington S. Racial discrimination against doctors from ethnic minorities. BMJ 1993;306:691-2. Crossref

35. Iacobucci G. Specialty training: ethnic minority doctors’ reduced chance of being appointed is “unacceptable”. BMJ 2020;368:m479. Crossref

36. Zhang M, Wang W, Millar R, Li G, Yan F. Coping and compromise: a qualitative study of how primary health care providers respond to health reform in China. Human Resour Health 2017;15:50. Crossref

37. Tan X, Liu X, Shao H. Healthy China 2030: a vision for health care. Value Health Reg Issues 2017;12:112-4. Crossref

38. Windish DM, Reed DA, Boonyasai RT, Chakraborti C, Bass EB. Methodological rigor of quality improvement curricula for physician trainees: a systematic review and recommendations for change. Acad Med 2009;84:1677-92. Crossref

39. Yi H, Wu P, Zhang X, Teuwen DE, Sylvia S. Market competition and demand for skills in a credence goods market: evidence from face-to-face and web-based non-physician clinician training in rural China. PLoS One 2020;15:e0233955. Crossref

40. People Television. Training of ethnic minority grassroots doctors in the central and western regions launched [in Chinese]. Available from: http://tv.people.com.cn/n/2012/0915/c14645-19017341.html. Accessed 16 Jul 2020.

A substantial number of people infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) remain completely asymptomatic throughout the course of infection.1 In a recent prospective observational study of maternal and neonatal complications among 2130 pregnant women with and without SARS-CoV-2 infection, 44% of pregnant women diagnosed with coronavirus disease 2019 (COVID-19) were asymptomatic upon diagnosis. Despite this asymptomatic status, they exhibited increased risks of maternal morbidity (relative risk=1.24; 95% confidence interval=1.00-1.52) and pre-eclampsia (relative risk=1.63; 95% confidence interval=1.01-2.63), compared with pregnant women who had not been diagnosed with COVID-19.2 In a retrospective cohort study conducted in Spain, 3.1% of 759 pregnant women exhibited anti–SARS-CoV-2 antibodies and had been asymptomatic throughout pregnancy.3 Serological testing may serve as a useful tool to identify pregnant women who have recovered from a recent asymptomatic SARS-CoV-2 infection; this information can help guide the management of potential future complications.

Before the imposition of strict border control, a large number of people travelled between Hong Kong and the rest of the world each day. Pregnant women could have been infected without their knowledge because of asymptomatic or very mild disease. Furthermore, pregnancy symptoms can mask some COVID-19 symptoms, particularly if the COVID-19 symptoms are mild.4 In this study, we invited women who had undergone Down syndrome screening (DSS) since November 2019 to ascertain the seroprevalence of anti–SARS-CoV-2 antibodies in an unselected sample of pregnant women in Hong Kong. The findings were expected to provide insights concerning the asymptomatic infection rate among pregnant women.

This prospective cohort study included pregnant women who presented for routine DSS and underwent routine blood sample collection at 11 to 13 weeks of gestation between November 2019 and October 2020. All participants delivered at Kwong Wah Hospital or Prince of Wales Hospital, Hong Kong; the last delivery occurred in March 2021. Eligibility criteria included consent to serum storage for future research purposes and intention to deliver at the booking hospital. Eligible women who attended the booking hospital for delivery were invited to participate in the study. Women who delivered elsewhere, experienced pregnancy termination or miscarriage, or received a diagnosis of COVID-19 before the study were excluded. Women who agreed to participate in the study were asked to provide written informed consent for blood collection at delivery. Symptoms of COVID-19 throughout the pregnancy were evaluated at recruitment and at delivery.

Serum antibodies against SARS-CoV-2 were analysed using a qualitative serological assay. Qualitative detection of anti–SARS-CoV-2 antibodies (immunoglobulin G [IgG] and immunoglobulin M [IgM]) directed against the nucleocapsid protein (N-protein) of the virus was performed using the Elecsys Anti–SARS-CoV-2 assay (Roche, United States) on a Cobas^®e411 analyser. The result was provided as a cut-off index (COI). A positive anti–SARS-CoV-2 antibody result was defined as COI >=1.0. Individuals who had recovered from COVID-19 were recruited as positive control cases (COI >=1.0); individuals who had negative SARS-CoV-2 test results were recruited as negative control cases (COI <1.0) to ensure quality control in the anti–SARS-CoV-2 immunoassay.

Positive results based on qualitative detection of anti–SARS-CoV-2 antibodies were subsequently confirmed by quantitative measurements of IgG and IgM antibodies against the SARS-CoV-2 spike protein by using enzyme-linked immunosorbent assays (ImmunoDiagnostics Limited, Hong Kong). All tests were performed in duplicate, in accordance with the manufacturer’s instructions. Results were interpreted as negative when the optical density was <0.2 and as positive when the optical density was >=0.2. For samples with positive results, anti–spike protein concentrations (ng/mL) were calculated.

Continuous variables were expressed as medians (interquartile ranges or ranges). Categorical variables were summarised as counts and percentages. SPSS Statistics (Windows version 26.0; IBM Corp, Armonk [NY], United States) was used for data analyses.

The STROBE reporting guidelines were used during the preparation of this manuscript.

In total, 3219 consecutive pregnant women were approached; 306 declined participation and 1083 were excluded, including two with laboratory-confirmed SARS-CoV-2 infection, 69 who experienced miscarriage or pregnancy termination, and 1012 who planned delivery elsewhere. Thus, 1830 women were recruited to the study and provided written informed consent to participate. In total, 1810 (98.9%) women were Chinese, and 852 (46.6%) women were nulliparous. The median (interquartile range) maternal weight, height and age were 55.5 kg (50.3-62.5), 159 cm (155-163) and 33.0 years (30.2-36.4), respectively.

In total, six women (0.33%) were seropositive (COI >=1) at the DSS visit; this seropositivity persisted until delivery. Among these six women, one exhibited both anti–SARS-CoV-2 IgG and IgM antibodies; the IgM antibodies were undetectable at delivery. The remaining seropositive women exhibited only anti–SARS-CoV-2 IgG antibodies at both visits. All six of these women reported no relevant symptoms during pregnancy. Among the six women, one reported a travel history before her first DSS and one reported relevant contact history. The median COIs were 2.465 (interquartile range=1.430-3.178) and 1.680 (interquartile range=1.145-2.350) at DSS and delivery, respectively. The interval between sample collections was 177 days (range, 161-195). The COI at delivery was lower by a median of 31.8% (range, 12.0-35.1), compared with the COI at the DSS visit.

Characteristics of the study sample are presented in the Table. Fifty six women reported a travel history during pregnancy; one (1.79%) was seropositive at both visits. Two women reported relevant contact history during pregnancy; one (50.0%) was seropositive at both visits. Forty two women reported relevant symptoms during pregnancy; all were seronegative at both visits. Of the 1788 asymptomatic women, six (0.34%) were seropositive at both visits. In all, 259 women reported undergoing community SARS-CoV-2 testing during pregnancy; all tested negative. Among these 259 women, two (0.77%) were seropositive at both visits.

Our findings demonstrated a low seroprevalence (0.33%) of anti–SARS-CoV-2 antibodies in an unselected sample of pregnant women in Hong Kong. Among women with risk factors for SARS-CoV-2 infection, 1.79% and 50% with a travel history and relevant contact history, respectively, were seropositive. This finding suggests that targeted serological testing of pregnant women with a positive epidemiological link is useful for identifying women who have recovered from asymptomatic SARS-CoV-2 infection. A limitation of this study was that we relied on recruited individuals to recall COVID-19 symptoms throughout pregnancy, using only two recall time points: recruitment and delivery. This aspect may have introduced recall bias, particularly when COVID-19 symptoms could be non-specific and overlap with pregnancy symptoms. However, this limitation presumably did not have a large effect on the results because the seroprevalence of anti–SARS-CoV-2 antibody found in our unselected sample of pregnant women was very low. With a population of over 7 million, Hong Kong has largely been successful in controlling the transmission of SARS-CoV-2. Only 11 981 confirmed or probable cases have been recorded since the beginning of the epidemic.5 Thus, it is reasonable that the number of asymptomatic SARS-CoV-2 infections has also been low; this low number of asymptomatic infections has led to a low seroconversion rate in pregnant women.

Compared with other seroprevalence studies in pregnant women, the seroprevalence recorded in our study was substantially lower. The seroprevalence rates were 14% and 21% in Barcelona6 and southern Madrid3 (both in Spain), respectively. The low seroprevalence of anti–SARS-CoV-2 antibodies recorded in our study is presumably related to the implementation of a series of infection control strategies, including strict border control, mandatory quarantine for inbound travellers, mask wearing, and meticulous contact tracing. Hong Kong has learnt from its prior experience combating the SARS (severe acute respiratory syndrome) outbreak in 2003; accordingly, it implemented serious control measures early during the current epidemic. The Figure outlines the timeline of public health measures implemented during the first three waves of the epidemic in Hong Kong. By the end of March 2020, Hong Kong had closed its border to all incoming non-residents arriving from overseas and stopped transits through the city. All returning residents were subject to mandatory quarantine for 14 days; the quarantine period was extended to 21 days in December 2020. Locally, temporary closures of gyms, karaoke venues, clubs, and bars were periodically enforced, depending on the incidence of COVID-19. Dine-in service was forbidden from 6:00 pm to 5:00 am for several months, beginning in mid-July 2020. Mask wearing in public indoor areas and public transportation was also mandatory at that time. Notably, all seropositive cases in this study were first identified at the DSS visit between February 2020 and July 2020. This suggests that the seropositive cases acquired their infections during the first two waves of the epidemic; the implementation of stricter control measures by the local government during the third wave might have led to a lower transmission rate of asymptomatic infection among pregnant women, resulting in a lack of seroconversion during that period. Moreover, pregnant women are presumably more careful about social distancing and compliant with public health regulations. It would be useful to compare our seroprevalence results with the findings in other countries where strict measures were also implemented, such as Australia and New Zealand.

The humoral immune response is characterised by the production of virus-specific neutralising antibodies. Regardless of whether patients are symptomatic, IgG or IgM seroconversion has been observed in 65% to 100% of patients after infection with SARS-CoV-2.7 8 9 10 We previously demonstrated that 75% of pregnant women with laboratory-confirmed SARS-CoV-2 infection were seropositive at delivery.11 In addition to the strict control measures mentioned above, the low seroprevalence recorded in our study might have been related to undetectable antibody levels at the time of specimen collection—the timing of blood sample collection might not have been compatible with antibody detection. However, this is unlikely to have affected the findings among the large number of pregnant women in this study. A longitudinal study conducted in Wuhan, China, demonstrated that the median times from the first virus-positive test result to IgG or IgM seroconversion were 7 and 14 days in asymptomatic and symptomatic patients, respectively.12 In the asymptomatic cohort, all patients underwent seroconversion within 14 days from the first positive reverse transcriptase–polymerase chain reaction result.12 While waning immunity was observed at 5 months after infection,13 two studies showed that the antibody levels remained high and detectable at 8 months after infection.14 15 This finding is consistent with our results that the antibodies persisted until delivery in all women who had demonstrated seroconversion at the DSS visit. Because the two blood samples in this study were collected approximately 6 months apart, the timing of specimen collection was presumably adequate to identify most women who had contracted SARS-CoV-2 and developed detectable levels of antibodies.

Concerns regarding the usefulness of serological testing have been raised because of the uncertain onset and duration of humoral immunity; our study demonstrated a very low seroconversion yield in an unselected sample of pregnant women. Because of the ongoing vaccination programme, it is increasingly difficult to distinguish people who have acquired humoral immunity through natural infection from people who have acquired humoral immunity through vaccination. While the strict public health measures in Hong Kong have significantly reduced community transmission of SARS-CoV-2, these measures cannot be maintained indefinitely. The adverse effects of prolonged social distancing measures on the economy, education, mental health, and well-being of the population are immeasurable. There is increasing evidence that the COVID-19 mRNA vaccines are safe and effective; thus, pregnant women and women planning to become pregnant are encouraged to undergo vaccination at the earliest opportunity because pregnancy is a risk factor for severe COVID-19–related complications (eg, intensive care unit admission, invasive ventilation requirement, and death).16 17 The risk of preterm birth is also greater among pregnant women with COVID-19, compared with pregnant women who do not have COVID-19.18 19 20 Our study has demonstrated that anti–SARS-CoV-2 antibodies found at DSS persist until delivery. This result suggests that the presence of anti–SARS-CoV-2 antibodies in early pregnancy, preferably acquired through vaccination, would provide some protection for the pregnant woman and her baby during the remaining portion of the pregnancy.

Until a highly effective therapeutic drug targeting SARS-CoV-2 becomes readily available, mass vaccination remains the best solution to control the COVID-19 pandemic, avoid further lockdowns, and allow a return to “normal” pre–COVID-19 life, as well as saving lives. Our study highlights the importance of a successful vaccination campaign.

Concept or design: LCY Poon, DS Sahota.
Acquisition of data: HHY Leung, CYT Kwok, WC Leung, DS Sahota, MBW Leung, GCY Lui, SSS Ng.
Analysis or interpretation of data: HHY Leung, CYT Kwok, LCY Poon.
Drafting of the manuscript: HHY Leung, CYT Kwok, LCY Poon, DS Sahota.
Critical revision of the manuscript for important intellectual content: HHY Leung, CYT Kwok, LCY Poon, DS Sahota, SSS Ng, PKS Chan.

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.

LCY Poon has received speaker fees and consultancy payments from Roche Diagnostics and Ferring Pharmaceuticals. She has also received in-kind contributions from Roche Diagnostics. Other authors have disclosed no conflict of interest.

We thank Lijia Chen, Tracy CY Ma, Maggie Mak, Ching-man Mak, Angela ST Tai, Jeffery Ip, Phyllis Ngai, Andrea Chan, and Lisa LS Chan for making substantial contributions by involving in study coordination and patient recruitment.

This work was supported by Roche Diagnostic, United States, which provided reagents for the qualitative detection of anti– SARS-CoV-2 antibodies.

Approval for the study was obtained from the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (CREC Ref No. 2020.214). Patients provided written informed consent to participate in the study; this included consent for serum storage for research purposes.

The study is registered with ClinicalTrials.gov (Ref NCT04465474).

1. Byambasuren O, Cardona M, Bell K, Clark J, McLaws ML, Glasziou P. Estimating the extent of asymptomatic COVID-19 and its potential for community transmission: systematic review and meta-analysis. Off J Assoc Med Microbiol Infect Dis Can 2020;5:223-34. Crossref

2. Villar J, Ariff S, Gunier RB, et al. Maternal and neonatal morbidity and mortality among pregnant women with and without COVID-19 infection: the INTERCOVID multinational cohort study. JAMA Pediatr 2021;175:817-26. Crossref

3. Villala n C, Herraiz I, Luczkowiak J, et al. Seroprevalence analysis of SARS-CoV-2 in pregnant women along the first pandemic outbreak and perinatal outcome. PLoS One 2020;15:e0243029. Crossref

4. Afshar Y, Gaw SL, Flaherman VJ, el al. Clinical presentation of coronavirus disease 2019 (COVID-19) in pregnant and recently pregnant people. Obstet Gynecol 2020;136:1117-25. Crossref

5. Hong Kong SAR Government. Together, we fight the virus! Available from: https://www.coronavirus.gov.hk/eng/index.html. Accessed 28 Jul 2021.

6. Crovetto F, Crispi F, Llurba E, Figueras F, G mez-Roig MD, Gratac s E. Seroprevalence and presentation of SARS-CoV-2 in pregnancy. Lancet 2020;396:530-1. Crossref

7. W lfel R, Corman VM, Guggemos W, et al. Virological assessment of hospitalized patients with COVID-2019. Nature 2020;581:465-9. Crossref

8. Zhao J, Yuan Q, Wang H, et al. Antibody responses to SARS-CoV-2 in patients with novel coronavirus disease 2019. Clin Infect Dis 2020;71:2027-34. Crossref

9. Long QX, Liu BZ, Deng HJ, et al. Antibody responses to SARS-CoV-2 in patients with COVID-19. Nat Med 2020;26:845-8. Crossref

10. Qu J, Wu C, Li X, et al. Profile of immunoglobulin G and IgM antibodies against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Clin Infect Dis 2020;71:2255-8. Crossref

11. Poon LC, Leung BW, Ma T, et al. Relationship between viral load, infection-to-delivery interval and mother-to-child transfer of anti-SARS-CoV-2 antibodies. Ultrasound Obstet Gynecol 2021;57:974-8. Crossref

12. Jiang C, Wang Y, Hu M, et al. Antibody seroconversion in asymptomatic and symptomatic patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Clin Transl Immunology 2020;9:e1182. Crossref

13. Choe PG, Kang CK, Suh HJ, et al. Waning antibody responses in asymptomatic and symptomatic SARS-CoV-2 infection. Emerg Infect Dis 2021;27:327-9. Crossref

14. Hartley GE, Edwards ES, Aui PM, et al. Rapid generation of durable B cell memory to SARS-CoV-2 spike and nucleocapsid proteins in COVID-19 and convalescence. Sci Immunol 2020;5:eabf8891. Crossref

15. Choe PG, Kim KH, Kang CK, et al. Antibody responses 8 months after asymptomatic or mild SARS-CoV-2 infection. Emerg Infect Dis 2021;27:928-31. Crossref

16. UK Government. JCVI issues new advice on COVID-19 vaccination for pregnant women. Available from: https://www.gov.uk/government/news/jcvi-issues-new-advice-on-covid-19-vaccination-for-pregnant-women. Accessed 17 May 2021.

17. Hong Kong College of Obstetricians and Gynaecologists. HKCOG advice on Covid-19 vaccination in pregnant and lactating women (interim; updated on 21 April 2021). Available from: https://www.hkcog.org.hk/hkcog/Upload/EditorImage/20210423/20210423141055_6024.pdf. Accessed 17 May 2021.

18. Martinez-Portilla RJ, Smith ER, He S, et al. Young pregnant women are also at an increased risk of mortality and severe illness due to coronavirus disease 2019: analysis of the Mexican National Surveillance Program. Am J Obstet Gynecol 2021;224:404-7. Crossref

19. Martinez-Portilla R, Sotiriadis A, Chatzakis C, et al. Pregnant women with SARS-CoV-2 infection are at higher risk of death and pneumonia: propensity score matched analysis of a nationwide prospective cohort (COV19Mx). Ultrasound Obstet Gynecol 2021;57:224-31. Crossref

20. Allotey J, Stallings E, Bonet M, et al. Clinical manifestations, risk factors, and maternal and perinatal outcomes of coronavirus disease 2019 in pregnancy: living systematic review and meta-analysis. BMJ 2020;370:e3320. Crossref

Since the beginning of the human immunodeficiency virus (HIV) epidemic, stigma and discrimination have affected the provision of healthcare to patients with HIV. These factors have limited access to HIV testing and treatment; they have also prevented the uptake of interventions, such as pre-exposure prophylaxis.1 The global shift necessary in the biomedical response to acquired immunodeficiency syndrome (AIDS) thus heavily relies on reductions of both stigma and discrimination. The importance of stigma reduction has been recognised as a key priority in the Blueprint for Achieving an AIDS-Free Generation (established by The United States President’s Emergency Plan for AIDS Relief)2 and in the HIV investment framework (established by The Joint United Nations Programme on HIV/AIDS).3

For many years, stigma from healthcare professionals has remained a major barrier to access for HIV prevention and treatment services. According to aggregated data from the Stigma Index for 50 countries, healthcare has been denied to one in eight people living with HIV (PLHIV).4 Despite high antiretroviral therapy coverage, Hong Kong is no exception to this phenomenon. In the early and middle 2010s, 26.8% of PLHIV reported stigmatising experiences during treatment for non- HIV-related healthcare needs.5 Since the late 2000s, advances in HIV treatment have led to a decline in mortality.6 Increasing life expectancy among PLHIV has resulted in higher HIV prevalence.7 Thus, medical students have an increased likelihood of encountering patients with HIV in future clinical practice, irrespective of their specialties. Because HIV epidemiology has changed over the years, medical students’ attitudes towards the disease may also have changed. If attitudes have changed, the medical school curriculum should be adjusted to match such changes; this will ensure that future physicians can provide the best possible care. To our knowledge, no study has compared changes over time in medical students’ attitudes towards HIV. Thus, we performed a survey to compare medical students’ attitudes towards HIV/AIDS between the late 2000s (2007-2010) and middle 2010s (2014-2017).

This study was a descriptive cross-sectional survey to measure medical students’ attitudes towards HIV/AIDS, as well as their experiences within the HIV curriculum.

An assessment survey was administered to final-year medical students at the Chinese University of Hong Kong, one of the two medical schools in Hong Kong. Questionnaires were distributed to medical students at the end of their final-year attachment to an HIV specialist clinic. At the Chinese University of Hong Kong, HIV education is an integral part of the medical student curriculum. Over the course of 3 years, students attend microbiology and medicine lectures regarding HIV/AIDS, a community medicine module that includes the prevention and control of HIV infection, a half-day attachment to Hong Kong’s largest HIV specialist clinic, and hospital ward rounds that involve patients with HIV. The HIV curriculum was generally consistent between the late 2000s and middle 2010s. This study was based on the analysis of survey data collected from six cohorts of medical students: three from 2007 to 2010 and three from 2014 to 2017. All students were asked to complete the survey during their final teaching session. Hard copies of the self-administered structured questionnaire were distributed to the students and collected by the course instructors. All responses were anonymous and participation was voluntary.

The study questionnaire was originally developed as an assessment form by the HIV medicine teaching team at the Chinese University of Hong Kong; it was designed for completion by final-year medical students. The form was piloted in 2005 and became standardised in 2007, with minor modifications in subsequent years.8 The questionnaire was in English and consisted of 25 close-ended questions; all questions were completed by the students without assistance. The questionnaire contents slightly differed between the 2007-2010 and 2014-2017 survey periods. Seven questions that appeared in both questionnaires were selected for analysis. These questions focused on demographics, exposure to patients with HIV, and attitudes towards patients with HIV. Eight additional questions were analysed for cohorts from 2014 to 2017. These questions focused on a participant’s ability to recall various HIV learning experiences and their satisfaction with clinical exposure during attachment to an HIV clinic. We divided students into groups according to their willingness to provide HIV care in the future. An unwilling student chose “strongly agree” or “agree” (on a four-point scale for cohorts from 2007 to 2010 and a six-point scale for cohorts from 2014 to 2017) when asked whether they would refuse to perform treatment or surgical procedures for patients with HIV.

We calculated proportions of responses, along with exact binomial 95% confidence intervals (CIs). We used bivariable logistic regression to compare exposure and attitudes between participants in the 2007-2010 cohort (cohorts from 2007 to 2010) and the 2014-2017 cohort (cohorts from 2014 to 2017); each set of three cohorts was grouped together to maximise sample size. Factors associated with willingness to provide HIV care were analysed using bivariable logistic regression. SPSS software (Windows version 26; IBM Corp., Armonk [NY], United States) was used for data management and statistical analyses. Two-sided P values <0.05 were considered statistically significant.

In total, 1050 final-year students participated in the survey. Three cohorts of final-year medical students participated between 2007 and 2010 (n=546); three additional cohorts of final-year medical students participated between 2014 and 2017 (n=504). The response rates were 97% in the 2007-2010 cohort and 91% in the 2014-2017 cohort. Table 1 shows the detailed characteristics, exposure, and attitudes among the surveyed students. There was no difference in gender between the 2007-2010 and 2014-2017 cohorts (odds ratio [OR]=1.14; 95% CI=0.89-1.46).

Significantly fewer students (39.3%) in the mid-2010s initially encountered patients with HIV during attachment to an HIV clinic, compared with students in the late 2000s (72.1%; OR=0.25; 95% CI=0.18-0.34) [Table 1]. The proportion of students who personally knew HIV-positive friends or relatives remained low and did not significantly differ over time (OR=0.55; 95% CI=0.1-3.01). In the 2007-2010 and 2014-2017 cohorts, only four of 545 students (0.7%) and two of 495 students (0.4%), respectively, personally knew HIV-positive friends or relatives.

Student willingness to provide HIV care was similar between cohorts (Table 1). Approximately 78% of students were willing to provide care (OR=0.98; 95% CI=0.73-1.33). The proportion of students who preferred to avoid work in a field involving HIV/AIDS significantly decreased over time: 17.2% in the 2007-2010 cohort, compared with 10.6% in the 2014-2017 cohort (OR=0.57; 95% CI=0.4-0.83). An increasing number of students planned to specialise in clinical HIV treatment: 11 of 517 students (2.1%) in the 2007-2010 cohort, compared with 53 of 480 students (11%) in the 2014-2017 cohort (OR=5.71; 95% CI=2.95-11.07).

There was no difference in gender between willing and unwilling students (OR=1.24; 95% CI=0.92-1.68) [Table 2]. Unwilling students were more likely than willing students to have initially encountered patients with HIV during attachment to an HIV clinic (61.4% vs 50.9%) [OR=1.53; 95% CI=1.07-2.19]. Willingness was not associated with personally knowing HIV-positive friends or relatives. In the 2014-2017 cohort, most students (80.3%) recalled their attachment to an HIV clinic, whereas fewer than half could recall other components of the HIV curriculum (lectures and ward rounds; 14.5-48.7%). Notably, the ability to recall attachment to an HIV clinic was associated with willingness (OR=0.57; 95% CI=0.34-0.96) to provide HIV care. Ratings of the content and format of the attachment to an HIV clinic were not associated with willingness (OR=0.94; 95% CI=0.58-1.50 and OR=0.79; 95% CI=0.50-1.27, respectively). Overall, most students (97.5%) have encountered >1 patient with HIV during clinical attachment. Unwilling students (5/71, 7%) were less likely than willing students (4/295, 1.4%) to have encountered any patients with HIV during clinical attachment.

Human immunodeficiency virus prevalence has risen over time8 with advances in antiretroviral therapy that contribute to prolonged survival; this pattern has been observed in nearly all countries, irrespective of the initial HIV prevalence. Thus, it is unsurprising that exposure to patients with HIV/AIDS increased over time among medical students in our study. Nevertheless, Hong Kong remains a low prevalence setting for HIV/AIDS9; therefore, the number of students who personally knew HIV-positive friends or relatives has remained low. Despite predictable trends in exposure to patients with HIV since the late 2000s, some students maintained negative attitudes towards patients with HIV. Although exposure to patients with HIV has increased, the proportion of students unwilling to provide HIV care did not change between the late 2000s and middle 2010s. The proportion of unwilling students in our study is higher than that in a similar study conducted in 2011 in Malaysia, where 10% to 15% of students reported unwillingness to provide HIV care.10 Because medical students are future healthcare providers, their unwillingness to provide HIV care represents an extreme manifestation of stigma, which has been a problem since the HIV/AIDS epidemic began.11

Stigma can have subtle effects, which may influence career choices. A previous study found that these effects can diminish over time12; our results are consistent with that finding. While a large proportion of students did not plan to work in a field involving HIV, such plans may be related to personal preferences for medical disciplines that facilitate career development and job opportunities, although discrimination cannot be ruled out. However, the present study showed that, over time, fewer students have reported that they prefer to avoid working in a field involving HIV/AIDS. Encouragingly, increasing numbers of students are planning to specialise in clinical HIV treatment. “Interest” has been most frequently cited as the main reason for choosing a specialty; thus, interest in HIV/AIDS may be increasing among medical students.13 However, the proportion of students who intended to specialise in clinical HIV treatment was lower in our study than in a previous study in the United Kingdom, where 8% to 24% of students reported such an intention.14

Clinical attachment with patient exposure appears to be an effective learning experience. In the current system, some students may have overlooked and missed the opportunity to encounter a patient with HIV/AIDS. We found that, compared with willing students, a higher proportion of students unwilling to provide HIV/AIDS care had not encountered a patient with HIV/AIDS during their clinical attachment. This difference may be attributed to a lack of exposure to patients with HIV/AIDS during the medical school curriculum. Students may benefit from repeated exposure to patients with HIV/AIDS in different settings—willing students were more likely to have previously encountered a patient with HIV/AIDS. Further research is needed to determine whether students could be exposed to patients with HIV outside clinical settings (eg, through non-governmental organisations) and to understand the impacts of such exposure. Willing students were more likely to recall their attachment to an HIV clinic, compared with other teaching methods; this suggests that their emotions were aroused, which prompted recall.15

Previous research has shown that teaching methods with experiential, small group, or affective components and role models of positive attitudes can effectively change students’ attitudes.14 Such considerations may be useful in clinics where small numbers of students observe patient management by a specialist physician. If a student is emotionally affected by a patient or sees the clinician as a good role model, the clinical attachment may constitute a memorable experience. This may be more important than gaining technical knowledge and skills through the clinical attachment experience because content and format were not associated with willingness to provide HIV care. Our results are consistent with the findings of previous studies, which showed that frequent clinical exposure to patients with HIV/AIDS led to more positive attitudes.14 16 17 18 19 20 Furthermore, knowledge alone cannot effectively decrease HIV stigma21; similarly, we found that the ability to recall lectures was not associated with willingness to provide care for patients with HIV/AIDS. However, increased exposure alone may be insufficient to combat HIV stigma. Antibias information is also needed to reduceprejudice,22 such as homophobia, which has been associated with unwillingness to provide care for patients with AIDS.12 Medical training should address these issues that contribute to stigma towards patients with HIV/AIDS. Notably, medical students reportedly demonstrated increased willingness to provide care for patients with HIV/AIDS after they had attended an PLHIV sharing session or participated in experiential games that were designed to increase empathy towards PLHIV.23 In another study, HIV stigma levels decreased in medical students after exposure to an intervention that included discussion of HIV stigma and other pre-existing related stigmas (eg, homosexuality and illegal drug use).24

This study had some limitations. First, it was a comparison of cross-sectional surveys over a 10-year period and thus we were unable to assess potential changes in attitudes among medical students within a single cohort. Second, each cohort used in comparative analysis was composed of three annual cohorts; therefore, time intervals varied among cohorts (eg, the 2007 and 2017 cohort were separated by 10 years, whereas the 2010 and 2014 cohorts were separated by 4 years). Nevertheless, we grouped cohorts together to maximise sample size; our analysis clearly showed changes in attitudes among medical students over time.

Despite more positive attitudes towards HIV/AIDS in terms of career choices, the willingness of medical students to provide HIV care did not change between the late 2000s and middle 2010s.

Interventions to reduce stigmatising attitudes among towards PLHIV should be incorporated in medical training; however, the framework for medical school curriculum in Hong Kong makes no mention of such interventions, although it lists “attitudes and professionalism” as a core competency.25 The medical school curriculum could be updated to incorporate interventions that involve experiential and affective teaching components to adequately address HIV stigma; additional clinical attachments can ensure that medical students have adequate exposure to patients with HIV.

Concept or design: G Tam, SS Lee.
Acquisition of data: SS Lee.
Analysis or interpretation of data: NS Wong.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.

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

The authors declare that they have no competing interests.

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

This study was approved by the Survey and Behavioural Research Ethics Committee, The Chinese University of Hong Kong (Ref 12-01-2011). Participation was voluntary and completion of the survey implied consent to participate in the study. All data were anonymised and confidential.

1. Stangl AL, Lloyd JK, Brady LM, Holland CE, Baral S. A systematic review of interventions to reduce HIV-related stigma and discrimination from 2002 to 2013: how far have we come? J Int AIDS Soc 2013;16(3 Suppl 2):18734. Crossref

2. US Department of State. The United States President’s emergency plan for AIDS relief. Available from: https://www.pepfar.gov/documents/organization/201386.pdf. Accessed 7 Jun 2022.

3. Schwartländer B, Stover J, Hallett T, et al. Towards an improved investment approach for an effective response to HIV/AIDS. Lancet 2011;377:2031-41. Crossref

4. International Health Policy Program (IHPP), Ministry of Public Health, Thailand. Report of a Pilot: Developing Tools and Methods to Measure HIV-related Stigma and Discrimination in Health Care Settings in Thailand. Thailand: IHPP, Ministry of Public Health; 2014.

5. Scientific Committee on AIDS and STI (SCAS) Centre for Health Protection Department of Health. Recommended principles and practice of HIV clinical care in Hong Kong. Hong Kong, 2016. Available from: https://www.chp.gov.hk/en/static/24003.html. Accessed 8 Jun 2022

6. The Lancet. The global HIV/AIDS epidemic—progress and challenges. Lancet 2017;390:333. Crossref

7. Fettig J, Swaminathan M, Murrill CS, Kaplan JE. Global epidemiology of HIV. Infect Dis Clin North Am 2014;28:323-37. Crossref

8. Lee SS, Lam A, Lee KC. How prepared are our future doctors for HIV/AIDS? Public Health 2012;126:165-7. Crossref

9. Virtual AIDS Office of Hong Kong. HIV/AIDS Statistics in Hong Kong 1984-June 2020. Available from: http://www.info.gov.hk/aids/english/surveillance/cumhivaids.jpg. Accessed 8 Jun 2022.

10. Ahmed SI, Hassali MA, Bukhari NI, Sulaiman SA. A comparison of HIV/AIDS related knowledge, attitudes and risk perceptions between final year medical and pharmacy students: a cross sectional study. Healthmed 2011;5:317-25.

11. Radecki S, Shapiro J, Thrupp LD, Gandhi SM, Sangha SS, Miller RB. Willingness to treat HIV-positive patients at different stages of medical education and experience. AIDS Patient Care STDS 1999;13:403-14. Crossref

12. Kopacz DR, Grossman LS, Klamen DL. Medical students and AIDS: knowledge, attitudes and implications for education. Health Educ Res 1999;14:1-6. Crossref

13. Wu SM, Chu TK, Chan ML, Liang J, Chen JY, Wong SY. A study on what influence medical undergraduates in Hong Kong to choose family medicine as a career. Hong Kong Pract 2014;36:123-9.

14. Ivens D, Sabin C. Medical student attitudes towards HIV. Int J STD AIDS 2006;17:513-6. Crossref

15. Storbeck J, Clore GL. Affective arousal as information: how affective arousal influences judgments, learning, and memory. Soc Personal Psychol Compass 2008;2:1824-43. Crossref

16. Syed IA, Hassali MA, Khan TM. General knowledge & attitudes towards AIDS among final year medical and pharmacy students. Folia Medica 2010;45:9-13.

17. Mohsin S, Nayak S, Mandaviya V. Medical students’ knowledge and attitudes related to HIV/AIDS. Natl J Community Med 2010;1:146-9.

18. Tešić V, Kolarić B, Begovac J. Attitudes towards HIV/AIDS among four year medical students at the University of Zagreb Medical School—better in 2002 than in 1993 but still unfavorable. Coll Antropol 2006;30:89-97.

19. Sanchez NF, Rabatin J, Sanchez JP, Hubbard S, Kalet A. Medical students’ ability to care for lesbian, gay, bisexual, and transgendered patients. Fam Med 2006;38:21-7.

20. Umeh CN, Essien EJ, Ezedinachi EN, Ross MW. Knowledge, beliefs and attitudes about HIV/AIDS-related issues, and the sources of knowledge among health care professionals in southern Nigeria. J R Soc Promot Health 2008;128:233-9. Crossref

21. Liu X, Erasmus V, Wu Q, Richardus JH. Behavioral and psychosocial interventions for HIV prevention in floating populations in China over the past decade: a systematic literature review and meta-analysis. PLoS One 2014;9:e101006. Crossref

22. Aboud FE, Tredoux C, Tropp LR, Brown CS, Niens U, Noor NM, Una Global Evaluation Group. Interventions to reduce prejudice and enhance inclusion and respect for ethnic differences in early childhood: a systematic review. Dev Rev 2012;32:307-36. Crossref

23. Mak WW, Cheng SS, Law RW, Cheng WW, Chan F. Reducing HIV-related stigma among health-care professionals: a game-based experiential approach. AIDS Care 2015;27:855-9. Crossref

24. Varas-Díaz N, Neilands TB, Cintrón-Bou F, et al. Testing the efficacy of an HIV stigma reduction intervention with medical students in Puerto Rico: the SPACES project. J Int AIDS Soc 2013;16(3 Suppl 2):18670. Crossref

25. The Medical Council of Hong Kong. Hong Kong doctors. 2017. Available from: https://www.mchk.org.hk/english/publications/files/HKDoctors.pdf. Accessed 8 Jun 2022.

Early-stage rectal cancer is primarily treated with total mesorectal excision surgery, while ‘high-risk’ rectal cancers can be treated with neoadjuvant short-course radiotherapy alone or concurrent chemotherapy and long-course radiotherapy (neoadjuvant chemoradiotherapy; NCRT).1 High-risk rectal cancer is defined as the presence of T3 or T4 disease, node-positive disease, the presence of close or involved circumferential resection margin CRM) by staging magnetic resonance imaging (MRI) and/or low-lying tumours involving the anal sphincters.1 Randomised phase III trials have shown that neoadjuvant is more effective than adjuvant chemoradiotherapy, as it can improve disease-free survival (DFS), local tumour control, sphincter preservation and has better treatment compliance with fewer adverse drug effects.2 3 4 Furthermore, the addition of 5-fluorouracil (5FU) to neoadjuvant radiotherapy has been shown to be more effective than radiotherapy alone with higher rates of pathological complete response (pCR) and lower local relapse rate.5

Historically, NCRT has been associated with 15% to 27% pCR rates that have been associated with progression-free survival and overall survival (OS).6 Other prognostic markers such as the presence of tumour down-staging in terms of T stage and N stage,7 tumour regression grading based on pathological and radiological criteria6 8 and CRM status9 have all been evaluated in clinical studies and correlated with predict survival and risk of cancer recurrence. However, a recently published meta-analysis has failed to show pCR rate as a significant surrogate marker of 5-year OS—an important primary endpoint in randomised trials, in patients with locally advanced rectal cancer (LARC) undergoing NCRT.10 Therefore, a new endpoint known as the neoadjuvant rectal (NAR) score has been developed as a prognostic factor and study endpoint for clinical research in LARC. This is a composite endpoint consisting of both clinical and pathological information on T stage and N stage obtained before and after NCRT and has been validated in prospective clinical trials in Western populations.11 12 The NAR score has also been shown to better predict OS in clinical trials on rectal cancer than pCR.11

The primary objective of the present study was to validate the prognostic significance of NAR score and pCR in a cohort of Hong Kong Chinese patients with LARC in terms of OS, DFS, locoregional recurrence-free survival (LRFS) and distant recurrence-free survival (DRFS). The second objective was to investigate associations between NAR score (or pCR) and known prognostic factors such as CRM status, tumour location, extramural vascular invasion (EMVI) and other treatment-related factors. The third objective was to investigate factors that might predict a lower NAR score.

The data of patients with LARC who were referred to the local multidisciplinary Lower Gastrointestinal Tumour Board and then underwent NCRT at the Prince of Wales Hospital, Hong Kong, from August 2006 to October 2018 were extracted from hospital records and retrospectively evaluated. Data were also retrieved from the records of the Lower Gastrointestinal Tumour Board meetings and the surgical new case database from the Prince of Wales Hospital.

Eligible patients had histologically confirmed LARC as defined by the presence of T3 or T4 tumour; or node-positive disease, and/or the presence of threatened CRM, and/or low-lying tumours involving the anal sphincters. All eligible patients underwent MRI and whole-body computed tomography (CT) scan staging before and after NCRT. Patients were excluded from the study who had distant metastasis at the time of diagnosis; who were not fit for NCRT or surgery due to poor performance status and/or presence of serious medical co-morbidities; or who had not completed the full course of NCRT.

All treatment decisions were jointly made by the Lower Gastrointestinal Tumour Board. At baseline, all patients underwent MRI staging and also systemic staging with contrast CT scan and/or positron emission tomography–CT imaging. Magnetic resonance imaging staging was determined by MRI radiologists and reported in a standardised format that contained information on T stage and N stage, presence of EMVI, CRM status and tumour regression grade response criteria.13 For patients with MRI reports which did not contain the relevant data, the MRI scans were assessed retrospectively in order to obtain the study information.

All patients were treated according to the institutional radiotherapy protocol at the Prince of Wales Hospital, as represented by a long-course pelvic radiotherapy up to a total dose of 45 Gy at 1.8 Gy per day, five fractions per week for 5 weeks with boost 5.4 Gy at 1.8 Gy per day for three fractions. The majority of patients received concurrent chemotherapy with bolus intravenous 5FU and leucovorin that were given at week 1 and week 5 of radiotherapy, followed by adjuvant chemotherapy with 5FU and leucovorin or oxaliplatin-based chemotherapy.14 Some patients also received neoadjuvant (modified) FOLFOXIRI regimen followed by concurrent capecitabine during pelvic radiotherapy as part of a prospective clinical trial.15 All patients underwent total mesorectal excision surgery with curative intent, and pathologists at the New Territories East Cluster–affiliated hospitals performed pathological examination on all the resected surgical specimens. The presence of pCR was defined as the resolution of all tumour cells in all resected tissues including the lymph nodes.

The following data were collected: age, sex, location of tumour from anal verge (defined as the endoscopic distance from anal verge as ‘low’ [0-5 cm], ‘mid’ [5-10 cm], ‘high’ [>10 cm]), tumour histology, neoadjuvant and adjuvant chemotherapy and the overall TNM (tumour, node, and metastasis) stage, as defined by the American Joint Committee on Cancer, 8th version. The date at histological diagnosis, cancer progression, locoregional and/or distant recurrence and the date of last follow-up examination or death were collected.

Pre- and post-treatment MRI data were collected: T stage (T2, T3 or T4), N stage (node positive or node negative), CRM (non-involved margin is defined as ≥2 mm; involved margin is defined as <2 mm from the anticipated surgical margin). The presence of EMVI was determined in the MRI scans of 152 patients.

The NAR score was calculated according to the Valentini’s nomograms for survival based on the following formula16:

NAR = [5pN−3(cT−pT)+12]² / 9.61,

where cT = clinical T stage before NRCT; pN = pathological nodal stage after NCRT and surgery; and pT = pathological T stage after NCRT and surgery.

The relationship between NAR scores and clinical outcome were analysed with NAR score as a continuous variable (24 discrete scores by the nomograms)16 or in groups based on previous studies.12 17 The NAR scores were grouped as: ‘low’ (NAR score <8), ‘intermediate’ (NAR score 8-16), and ‘high’ (NAR score >16), as previously published in the National Surgical Adjuvant Breast and Bowel Project ‘R-04’ trial,11 or in quartiles according to the ‘FORWARC’ study.17

Overall survival was defined from the time of diagnosis to the time of death from any cause. Survival time will be censored at the last date the patient is known to be alive. Disease-free survival was defined from the time of diagnosis of rectal cancer to the time of disease recurrence and death from any cause. Locoregional recurrence-free survival was measured from the date of diagnosis to the date of locoregional recurrence and death from any cause. Distant recurrence-free survival was measured from the date of diagnosis to the date of distant metastasis and death from any cause.

Statistical analysis was performed using the SPSS (Window version 26; IBM Corp, Armonk [NY], United States). The Chi squared or Fisher’s exact test was used for analysing categorical variables, t test for continuous variables and logistic regression was used to analyse the relationship between continuous variables and disease recurrence. Time-to-event endpoints include OS, DFS, LRFS and DRFS were estimated using the Kaplan–Meier method and compared using the log-rank test. Cox proportional hazards model was used to evaluate any interaction between time-to-event endpoints and important covariates. The multivariable Cox regression with stepwise selection method was used to study NAR score and other prognostic factors. A value of P<0.05 was considered significant. The correlation between pCR and important covariates was obtained by using logistic regression. The odds ratio and the corresponding 95% confidence interval (CI) will be given.

A total of 209 patients were found to be eligible, 16 of whom had suboptimal response to NCRT as defined by one or more of the following factors: persistently positive CRM, absence of significant tumour regression on MRI, or frank radiological progression (Fig 1). These patients were treated with consolidation chemotherapy after NCRT and of whom eight patients responded and underwent surgery with curative intent. The characteristics of the remaining 193 patients who had optimal response after NCRT had a mean age of 62 years, with a male and female ratio of 2.94:1 (Table 1). The median follow-up duration for all patients was 47.7 months (range, 42.7-53.5).

When the NAR score was analysed as 24 discrete scores by Valentini’s nomograms,16 it was found to be associated with OS (hazard ratio [HR]=1.042, 95% CI=1.021-1.064; P<0.0001), DFS (HR=1.042, 95% CI=1.022-1.062; P<0.0001), LRFS (HR=1.070, 95% CI=1.039-1.102; P<0.0001) and DRFS (HR=1.034, 95% CI=1.012-1.056; P=0.002).

To evaluate the effect of NAR score on survival rates, patients were arbitrarily divided into three groups according to NAR score: low (score <8; n=50), intermediate (score 8-16; n=99) and high (score >16; n=44) [Table 2]. There was a significant difference among the OS curves of low, intermediate, and high NAR score groups (P=0.004, Fig 2). Similarly, there was a significant difference among the DFS rates of the low, intermediate, and high NAR score groups (P<0.0001, Fig 3). The DFS was lower for the intermediate NAR score group than for the low NAR score group (HR=4.50, 95% CI=1.35-14.95; P=0.014), whereas the risk of progression was higher for the high NAR score group than for the low NAR score group (HR=8.14, 95% CI=2.40-27.65; P=0.001).

There was a significant difference among the LRFS rates of the low, intermediate, and high NAR score groups as shown in Figure 4 (P=0.002). Similarly, as shown in Figure 5, the DRFS rates of the three NAR score groups showed a statistical difference (P=0.013). The intermediate NAR score group had a lower DRFS than the low NAR score group (HR=4.04, 95% CI=1.21-13.50; P=0.023), while the high NAR score group had a higher risk of distant recurrence than the low NAR score group (HR=5.65, 95% CI=1.61-19.84; P=0.007).

The NAR score was an independent prognostic factor for OS, DFS, LRFS and DRFS, irrespective of whether NAR score was analysed as a continuous variable or in groups of low, intermediate, and high NAR score (Tables 3 and 4). Other prognostic markers, such as age and MRI T stage, were predictive of OS, DFS and DRFS. The MRI tumour down-staging after NCRT was an independent prognostic factor for OS, DFS and LRFS. This study further evaluated the prognostic factors that might predict a low NAR score in subgroups of patients after NCRT. Of all the prognostic factors evaluated, only pCR was associated with a lower NAR score (NAR score ≤8 or >8) [Table 5].

In the 193 patients who had pCR to NCRT and surgery, MRI tumour down-staging was the only prognostic factor which was associated with the rate of pCR (P<0.0001) [Table 1].

In the present study, NAR score was found to be a more power prognostic factor than pCR. Furthermore, patients who achieved pCR post NCRT tend to have lower NAR scores. Furthermore, the results of the present study indicate significant differences in the rates of OS, DFS, LRFS and DRFS among patients with low, intermediate, and high NAR scores in a Hong Kong Chinese population, which is consistent with observations from a study in Western populations.12 Several interesting observations can be made in the survival rates among the low, intermediate, and high NAR score groups. The DFS and DRFS curves of the intermediate and high NAR score groups (Figs 3 and 5) crossed over around the 1-year mark, demonstrating that survival of the intermediate group was initially inferior to the high NAR score group. This trend might be explained by an imbalance in the sample size of patients were in the intermediate NAR score group (n=99) compared with the high NAR score group (n=44) [Fig 1]. The recurrence rate in the low NAR score group reached a plateau at around 3 years, whereas in the intermediate NAR score group, late recurrences (especially distant recurrence) could occur well over 72 months after diagnosis. Therefore, this study suggests that longer follow-up duration for a period beyond 72 months may be needed for the intermediate NAR score group. This is in contrast to the recommendation in the European Society of Medical Oncology guideline which suggests a follow-up duration of up to 60 months.18

In this study, the NAR score (not pCR) was found to be an independent prognostic marker for survival and disease recurrence. It is possible that NAR score could better reflect the magnitude and dynamics of tumour regression over time, whereas pCR could give only dichotomised results observed at a single time-point after surgery.

There are several limitations to this retrospective study. The sample size was relatively small and there was an imbalance in the number of patients in the intermediate NAR score group compared with the other groups (Fig 1). Given the prognostic significance of MRI EMVI in LARC,19 this study included this endpoint in the multivariate analysis. However, the MRI EMVI status could not be retrieved for some patients, especially those who had MRI imaging >5 years ago when this information was not captured at the time of imaging. Furthermore, the MRI N stage was only reported as either ‘positive’ or ‘negative’ in terms of nodal involvement without specifying the exact number of suspicious nodes. The CRM status and EMVI after NCRT and surgery has been shown in previous studies to affect prognosis and alter postoperative management.20 21 However, information on these two prognostic factors could not be traced retrospectively, therefore only the pretreatment MRI CRM and MRI EMVI were included in the analysis. Nevertheless, the findings of this study are significant given the multicentre nature and also relatively long follow-up duration. Furthermore, it is consistent with the results of previous studies.12 17

Although NAR score is a consistent and validated prognostic marker, its determination relies on the availability of radiological and pathological assessments after surgery. In clinical practice, surgeons and oncologists have to rely heavily on MRI and/or endoscopic findings on assessing response to NRCT when making decisions on operability and preoperative consolidation chemotherapy after NRCT. Nevertheless, the NAR score is useful in the decision-making process with regard to the need for intensifying adjuvant chemotherapy and also length of follow-up duration. A study in Japan showed a benefit in administering adjuvant chemotherapy to patients with low NAR score (<16), but not in those with higher NAR score (≥16).22 Further studies are needed to individualise adjuvant chemotherapy for Chinese patients using NAR scores after NCRT for LARC. Other more novel strategies such as personalised drug testing using rectal cancer organoid platforms in studying individual response to NCRT are on the horizon.23

The NAR score is an independent prognostic marker of survival and disease recurrence in a cohort of Hong Kong Chinese patients who received NCRT for LARC.

Concept or design: BBY Ma, SSK Ho, SSF Hon.
Acquisition of data: SSK Ho, SSF Hon, E Hung, JFY Lee, M Tong, C So, S Chu, DCK Ng, D Lam, C Cho, TWC Mak, SSM Ng, K Futaba, J Suen, KF To, AWH Chan, WWK Yeung, BBY Ma.
Analysis or interpretation of data: F Mo, SSK Ho.
Drafting of the manuscript: BBY Ma, SSK Ho, SSF Hon.
Critical revision of the manuscript for important intellectual content: BBY Ma, SSK Ho, SSF Hon.

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

All authors have disclosed no conflicts of interest.

This research was presented as a poster and abstract at the ESMO Asia Virtual Congress in 2020.

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

This study was approved by the New Territories East Cluster–The Chinese University of Hong Kong (NTEC-CUHK) Ethics Committee (Ref NTEC-2019-0086). The requirement for patient consent was waived by the ethics board owing to the retrospective nature of the study.

1. Johnson D, Li L, Lee KC, et al. Total neoadjuvant therapy for high risk rectal cancer in Western and Asian populations— current evidence and clinical applications. Clin Colorectal Cancer 2022;21:45-54. Crossref

2. Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. N Engl J Med 2004;351:1731-40. Crossref

3. Sauer R, Liersch T, Merkel S, et al. Preoperative versus postoperative chemoradiotherapy for locally advanced rectal cancer: results of the German CAO/ARO/AIO-94 randomized phase III trial after a median follow-up of years. J Clin Oncol 2012;30:1926-33. Crossref

4. Roh MS, Colangelo LH, O’Connell MJ, et al. Preoperative multimodality therapy improves disease–free survival in patients with carcinoma of the rectum: NSABP R-03. J Clin Oncol 2009;27:5124-30. Crossref

5. Gérard JP, Conroy T, Bonnetain F, et al. Preoperative radiotherapy with or without concurrent fluorouracil and leucovorin in T3-4 rectal cancers: results of FFCD 9203. J Clin Oncol 2006;24:4620-5. Crossref

6. Maas M, Nelemans PJ, Valentini V, et al. Long-term outcome in patients with a pathological complete response after chemoradiation for rectal cancer: a pooled analysis of individual patient data. Lancet Oncol 2010;11:835-44. Crossref

7. Fokas E, Liersch T, Fietkau R, et al. Downstage migration after neoadjuvant chemoradiotherapy for rectal cancer: the reverse of the Will Rogers phenomenon? Cancer 2015;121:1724-7. Crossref

8. Fokas E, Liersch T, Fietkau R, et al. Tumor regression grading after preoperative chemoradiotherapy for locally advanced rectal carcinoma revisited: updated results of the CAO/ARO/AIO-94 trial. J Clin Oncol 2014;32:1554-62. Crossref

9. Kelly SB, Mills SJ, Bradburn DM, Ratcliffe AA, Borowski DW, Northern Region Colorectal Cancer Audit Group. Effect of the circumferential resection margin on survival following rectal cancer surgery. Br J Surg 2011;98:573-81. Crossref

10. Petrelli F, Borgonovo K, Cabiddu M, Ghilardi M, Lonati V, Barni S. Pathologic complete response and disease-free survival are not surrogate endpoints for 5-year survival in rectal cancer: an analysis of 22 randomized trials. J Gastrointest Oncol 2017;8:39-48. Crossref

11. George TJ Jr, Allegra CJ, Yothers G. Neoadjuvant rectal (NAR) score: a new surrogate endpoint in rectal cancer clinical trials. Curr Colorectal Cancer Rep 2015;11:275-80. Crossref

12. Fokas E, Fietkau R, Hartmann A, et al. Neoadjuvant rectal score as individual-level surrogate for disease-free survival in rectal cancer in the CAO/ARO/AIO-04 randomized phase III trial. Ann Oncol 2018;29:1521-7. Crossref

13. Sclafani F, Brown G. Extramural venous invasion (EMVI) and tumour regression grading (TRG) as potential prognostic factors for risk stratification and treatment decision in rectal cancer. Curr Colorectal Cancer Rep 2016;12:130-40. Crossref

14. Yeung WW, Ma BB, Lee JF, et al. Clinical outcome of neoadjuvant chemoradiation in locally advanced rectal cancer at a tertiary hospital. Hong Kong Med J 2016;22:546-55. Crossref

15. Lam G, Tong M, Lee J, et al. A multicenter phase II study of neoadjuvant FOLFOXIRI followed by concurrent capecitabine and radiotherapy for high risk rectal cancer: a final report. Ann Oncol 2019;30(Suppl 9):ix30-41. Crossref

16. Valentini V, van Stiphout RG, Lammering G, et al. Nomograms for predicting local recurrence, distant metastases, and overall survival for patients with locally advanced rectal cancer on the basis of European randomized clinical trials. J Clin Oncol 2011;29:3163-72. Crossref

17. Deng Y, Cai Y, Zhang J, et al. Validation of neoadjuvant rectal cancer (NAR) score as a surrogate endpoint for disease free survival in Chinese FOWARC study. J Clin Oncol 2019;37(15 Suppl):e15162. Crossref

18. Glynne-Jones R, Wyrwicz L, Tiret E, et al. Rectal cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2017;28(suppl 4):iv22-40. Crossref

19. Cho MS, Park YY, Yoon J, et al. MRI-based EMVI positivity predicts systemic recurrence in rectal cancer patients with a good tumor response to chemoradiotherapy followed by surgery. J Surg Oncol 2018;117:1823-32. Crossref

20. Goffredo P, Zhou P, Ginader T, et al. Positive circumferential resection margins following locally advanced colon cancer surgery: risk factors and survival impact. J Surg Oncol 2020;121:538-46. Crossref

21. Song KS, Lee DW, Kim B, et al. Differences in prognostic relevance of rectal magnetic resonance imaging findings before and after neoadjuvant chemoradiotherapy. Sci Rep 2019;9:10059. Crossref

22. Maeda K, Shibutani M, Tachimori A, et al. Prognostic significance of neoadjuvant rectal score and indication for postoperative adjuvant therapy in rectal cancer patients after neoadjuvant chemoradiotherapy. In Vivo 2020;34:283-9. Crossref

23. Ganesh K, Wu C, O’Rourke KP, et al. A rectal cancer organoid platform to study individual responses to chemoradiation. Nat Med 2019;25:1607-14. Crossref

Despite dramatic improvement in the prognosis of paediatric acute lymphoblastic leukaemia (ALL), the age at diagnosis remains a major prognostic factor: adolescents with ALL have worse outcomes than their younger counterparts.1 2 3 4 This is partly related to differences in disease biology, such that older children with ALL more frequently have a T-cell phenotype and less frequently have high hyperdiploidy or ETV6-RUNX1 translocation.1 1 1 1 1 9 Therefore, older children constitute a distinct subgroup for which an optimal treatment strategy has not been determined. Although intensive treatment protocols for paediatric ALL reportedly improve outcomes among adolescents,3 5 10 11 12 limited data are available from East Asian countries regarding the characteristics of adolescents with ALL who underwent treatment in clinical trials.13 The National Cancer Institute criteria, used for risk stratification in most international ALL trials, define age ≥10 years as a risk factor for B-cell precursor ALL1 2 3 4 5 10 11 12 14; however, most treatment-related toxicities occur with significantly greater frequency in older adolescents (aged ≥15 years).1 2 3 4 5 10 11 12 14 To our knowledge, there is limited available information regarding the differences in clinical characteristics and long-term treatment outcomes between adolescents (younger adolescents aged 10-14 years and older adolescents aged 15-18 years) and young children (aged 1-9 years) who receive intensive paediatric treatment protocols for ALL.13 15 Additionally, because ALL is a comparatively uncommon disorder in older adolescents, specific treatment outcome data for such patients are limited. We aimed to study the territory-wide outcome of adolescents with ALL treated by uniform chemotherapy protocols in Hong Kong, and tried to identify the treatment response and toxicity profile in the adolescents, and also the causes of treatment failure in particular older adolescents who shared similar characteristics of young adults.

In total, 711 patients (aged 1-18 years) newly diagnosed with ALL were enrolled in consecutive clinical trials during the period from 1993 to 2015; these trials were HKALL 9316 (1993-1997, n=144), HKALL 9717 (1997-2002, n=170), ALL IC-BFM 200218 (2003-2008, n=169), CCLG-ALL 200819 (2008-2015, n=221), and EsPhALL20 (2008-2014, n=7).

Detailed treatment stratification and therapy protocols used in the five trials have been described elsewhere. Briefly, stratification in the HKALL 93, HKALL 97, and ALL IC-BFM 2002 trials was performed using the following information: initial white blood cell count, central nervous system (CNS) status, immunophenotype, age at diagnosis, molecular-genetic abnormalities (t[9;22]/BCR-ABL1, ETV6-RUNX1, t[1;19]/TCF3-PBX1, and KMT2A-rearranged), and early response to chemotherapy (day 8 prednisone response and post-induction bone marrow status). Thus, patients were stratified into three risk groups within the respective trials: standard-risk, intermediate-risk, and high-risk. In the CCLG-ALL 2008 trial, therapy stratification was performed using flow cytometry and polymerase chain reaction–based analyses of minimal residual disease (MRD).19 Definitive risk assignment (for provisional standard- or intermediate-risk cases based on presenting features) was performed after MRD evaluation during therapy. In the EsPhALL trial, patients were stratified into good and poor risk groups according to their early response to induction therapy (day 8 prednisone response and post-induction bone marrow status).

Characteristics were compared among age-groups using the Chi squared test or Fisher’s exact test for categorical variables; the Wilcoxon rank-sum test was used for comparisons of continuous variables. We used the following age-group definitions: young children were patients aged 1 to 9 years and adolescents were patients aged 10 to 18 years; younger adolescents were patients aged 10 to 14 years and older adolescents were patients aged 15 to 18 years. Complete remission (CR) was defined as <5% bone marrow lymphoblasts and the absence of peripheral lymphoblasts or extramedullary disease. Event-free survival (EFS) was defined as the length of time from diagnosis to the last follow-up or first event (relapse, secondary malignancy, or death from any cause). Overall survival (OS) was defined as the length of time from diagnosis to the last follow-up or death from any cause. The probabilities of EFS and OS were estimated by Kaplan–Meier analysis; they were compared between groups using the log-rank test. Time to relapse was defined as the length of time from the end of remission induction chemotherapy (for patients who achieved CR) to relapse. The cumulative incidence of relapse was estimated according to time period; death from any cause before relapse was regarded as a competing event. Time to treatment-related death (TRD) was defined as the length of time from the date of diagnosis until death from non-progressive disease. The cumulative incidence of TRD was estimated by regarding leukaemia-related death and relapse as competing risk factors. Gray’s methods were used to assess the effects of age-group on the cumulative incidences of relapse and TRD. Univariable and multivariable Cox proportional hazard regression models were used to identify predictors of survival; univariable and multivariable competing risks regression models were used to identify predictors of TRD. Predictors with P values <0.1 in univariable analyses were included in the corresponding multivariable model. All tests were two-sided, and P values <0.05 were considered statistically significant. Stata Statistical Software (version 12.0; StataCorp, College Station [TX], United States) was used for all statistical analyses. The STROBE checklist was followed to ensure standardised reporting.

The characteristics of the 711 patients analysed in this study are shown in Table 1. There were 530 young children, 136 younger adolescents, and 45 older adolescents. Sex distribution did not differ between young children and adolescents, but the proportion of male patients tended to be higher among older adolescents. The proportion of patients with white blood cell count ≥50 × 10⁹/L at presentation was greater among adolescents than among young children (29.8% vs 19.8%, P=0.005). The proportion of patients with a B-cell immunophenotype was greater among young children (91.3% vs 72.9%), while the proportions of patients with a T-cell immunophenotype were significantly greater among older and younger adolescents than among young children (31.1% vs 23.5% vs 7.5%, P<0.001). The incidences of CNS involvement at diagnosis (CNS2/3 status) were 11.1%, 4.4%, and 4.2% among older adolescents, younger adolescents, and young children, respectively; these values did not significantly differ (P=0.102). Concerning the karyotypes of leukaemic cells, the proportion of patients with high hyperdiploidy (≥51 chromosomes) was significantly greater among young children than among older or younger adolescents (P=0.001). ETV6-RUNX1 fusion was also significantly more common among young children (P<0.001).

In total, 471 patients underwent evaluations of blast count in peripheral blood after 7 days of prednisone therapy. The proportion of patients with poor prednisone response (blast count >1.0 × 10⁹/L after 7 days of prednisone therapy) was greater among older adolescents than among younger adolescents or young children (22.9% vs 13.5% vs 6.9%, P=0.003). Additionally, the CR rate was significantly lower among older adolescents than among younger adolescents or young children (80.0% vs 92.6% vs 98.3%, P<0.001). The early death rate during induction therapy was higher among older adolescents than among younger adolescents or young children (6.7% vs 0.7% vs 1.1%, P=0.008). In total, 288 patients underwent MRD assessment at the end of remission induction; the proportion of patients with MRD ≥1% was greater among adolescents than among young children (16.7% vs 5.2%), while the proportion of patients with MRD <0.01% was lower among adolescents than among young children (47.4% vs 70.5%, P<0.001). However, MRD response did not differ between younger adolescents and older adolescents.

The treatments and outcomes of older adolescents with ALL are shown in the online supplementary Figure. Three patients died during induction (two had TRD and one had leukaemia-related death). Among the 36 older adolescents who achieved CR, three patients underwent allogeneic hematopoietic stem cell transplantation (HSCT) during CR1; one died of transplant-related infection, one relapsed (they achieved CR2 after salvage chemotherapy and remained in continuous CR), and one remained in continuous CR. The remaining 33 patients received only chemotherapy; 28 remained in continuous CR, one died of treatment-related infection, and five relapsed. Among the patients who relapsed, one was lost to follow-up, two died of progressive leukaemia, and two received allogeneic HSCT during CR2; one of the two transplant patients died of transplant-related infection, while the other remained in continuous CR.

Among the six patients who failed to achieve CR after remission induction chemotherapy, two died of progressive leukaemia, while four achieved CR after salvage chemotherapy. Among the four patients who achieved CR, three received allogeneic HSCT during CR1 and remained in continuous CR; the other patient relapsed and received allogeneic HSCT after achievement of CR2, then died of transplant-related infection. In summary, six of the 11 deaths among older adolescents were treatment-related; the main cause of TRD was infection.

The median follow-up interval (for all groups) was 12.78 years (interquartile range=6.73-19.09). Young children had significantly better 10-year OS and EFS rates, compared with adolescents (87.6% [95% confidence interval (CI)=84.4%-90.2%] vs 77.9% [95% CI=71.0%-83.4%], P=0.0003; 76.5% [95% CI=72.6%-79.9%] vs 69.7% [95% CI=62.3%-76.0%], P=0.0117; Fig 1a and b). Ten-year relapse rates were similar between young children and adolescents: 20.6% (95% CI=17.3%-24.4%) for young children vs 22.8% (95% CI=16.9%-30.4%) for adolescents (P=0.479; Fig 1c). The 10-year incidence of TRD was significantly greater among adolescents (7.2% [95% CI=4.1%-12.4%]) than among young children (2.3% [95% CI=1.2%-4.1%]) [P=0.002; Fig 1d]. Subgroup analysis revealed that OS and EFS rates, as well as cumulative incidences of relapse and TRD, were similar between younger adolescents and older adolescents (Fig 2).

Predictors of OS and EFS are shown in Tables 2 and 3, respectively. Univariable analysis showed that both younger and older adolescent age-groups (vs young children) were associated with poor OS (P=0.003 and P=0.009). Additionally, univariable analysis showed that more recent time periods and treatment protocols (ALL IC-BFM 2002 and CCLG-ALL 2008), as well as favourable cytogenetics (high hyperdiploidy and/or ETV6-RUNX1), were significantly associated with better OS. After adjustments for parameters with P values <0.1 in univariable analysis, multivariable Cox regression analysis revealed that both younger and older adolescent age-groups remained independent predictors of OS (hazard ratio=1.79 [95% CI=1.07-3.00], P=0.026; hazard ratio=2.98 [95% CI=1.41-6.30], P=0.004). Favourable cytogenetics also remained an independent predictor of OS (P=0.002). Similarly, univariable analysis showed that the younger adolescent age-group (vs young children) was significantly associated with poor EFS (P=0.029); the older adolescent age-group (vs young children) tended to show an association with poor EFS, although this was not statistically significant (P=0.111). Upon inclusion of all parameters with P values <0.1 in univariable analysis, multivariable Cox regression analysis revealed that both younger and older adolescent age-groups (vs young children) were significantly associated with poor EFS (hazard ratio=1.57 [95% CI=1.02-2.41], P=0.039; hazard ratio=2.18 [95% CI=1.16-4.09], P=0.016).

Predictors of the cumulative incidence of TRD are shown in Table 4. Univariable analysis showed that only younger and older adolescent age-groups (vs young children) were significantly associated with a greater incidence of TRD (hazard ratio=3.25 [95% CI=1.35-7.83], P=0.009; hazard ratio=4.50 [95% CI=1.43-14.13], P=0.010). Furthermore, favourable cytogenetics (high hyperdiploidy and/or ETV6-RUNX1) tended to show an association with lower incidence of TRD, although this was not statistically significant (P=0.088). After adjustments for parameters with P values <0.1 in univariable analysis, multivariable competing risks regression analysis revealed that both younger and older adolescent age-groups remained independent predictors of a greater incidence of TRD [hazard ratio=3.16 (95% CI=1.11-9.01), P=0.031; hazard ratio=4.69 (95% CI=1.28-17.20), P=0.020].

In this retrospective study, we combined five clinical trials of paediatric ALL treatment in Hong Kong to compare characteristics and outcomes among young children, younger adolescents, and older adolescents with ALL; we specifically focused on the outcomes of older adolescents. Among the overall cohort of patients with ALL in this study, which covered a 20-year period and included 711 non-infant patients, 6.3% were older adolescents; this proportion was comparable with the findings in previous studies.1 4 8 21 22 Additionally, our results are consistent with published literature in that adolescents with ALL were more likely to have a T-cell immunophenotype and less likely to have favourable genetic features (eg, high hyperdiploidy or ETV6-RUNX1), compared with young children who had ALL.1 4 5 6 7 8 9 13 These findings are consistent with the results of previous studies conducted in Western countries.1 4 5 6 7 8 9

Over the past two decades, several comparative analyses have shown that adolescents with ALL experience better outcomes when they receive paediatric treatment protocols, rather than adult treatment protocols.6 10 23 24 Adult protocols for ALL (eg, hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone) only achieved 5-year OS rates of 40% to 60% in adolescents and young adults with ALL.25 Although most adult treatment programmes for ALL have evolved from the multi-agent approach used in paediatric protocols, there are some notable differences in treatment design. Paediatric ALL protocols generally use more intensive dosing of several key therapeutic agents, including corticosteroids, vincristine, asparaginase/PEG-asparaginase, and anti-metabolites (eg, methotrexate and 6-mercaptopurine); they also use more intensive and prolonged CNS prophylaxis with intrathecal chemotherapy.25 26 27 In the present study, the 10-year EFS (70.2% vs 68.6%) and OS (78.8% vs 75.4%) rates for younger and older adolescents confirm the favourable outcomes of paediatric ALL protocols for adolescents aged ≤18 years.4 13 15 21 22 28 29 30 There are some important challenges involved in the treatment of adolescents with intensive chemotherapy protocols; these include a greater frequency of treatment-related complications (eg, liver derangement and thrombosis) than in young children who receive similar treatment. Drug compliance is also challenging in adolescents; poor adherence to long-term maintenance treatment may lead to worse outcomes.31

Notably, the long-term OS and EFS rates remained worse in adolescents with ALL than in young children (aged 1-9 years) with ALL. Our results indicate that this difference is not related to an increased rate of relapse; it arises from an increased risk of TRD. An age-related increase in treatment-related toxicity has been reported in almost all cohorts of patients with ALL who have received paediatric treatment protocols. Most studies have shown that, compared with young children, adolescents have greater risks of severe adverse events.28 32 The use of paediatric intensive combination chemotherapy is effective for preventing relapse in adolescents with ALL, but these patients may not tolerate the toxicity of intensive multi-agent chemotherapy (eg, myeloablative allogeneic HSCT). For example, among older adolescents in the present study, the high incidence of TRD was mainly attributed to two TRDs in 45 patients who received remission induction chemotherapy, one TRD in 33 patients who received post-induction chemotherapy during CR1, and three TRDs in nine patients who received allogeneic HSCT during CR1 or CR2. Further studies are needed to identify optimal treatment adjustments that can improve toxicity profiles among adolescents with ALL who receive paediatric treatment protocols.

Consistent with previous findings,1 33 34 the present study showed that poor early response to treatment was more common in adolescents, a greater proportion of whom had poor day 8 prednisone response and did not achieve CR. Minimal residual disease response after induction is an important prognostic indicator of treatment failure. In our more recent treatment protocols, MRD was included in the disease monitoring. A greater proportion of adolescents had MRD ≥1% after remission induction, but the relapse rate was not greater in adolescents than in young children. Adolescents received higher intensity consolidation, reinduction, and continuation therapy; some received allogeneic HSCT during CR1. The higher intensity of post-induction treatment led to a lower relapse rate but resulted in greater treatment-related mortality; thus, the OS and EFS rates were worse in adolescents than in young children. To improve survival outcomes among adolescents with ALL, clinical trials have been initiated with a focus on new agents that might achieve better survival without excessive toxicity; these agents include the proteasome inhibitor bortezomib, as well as antibody- or cell-mediated immunotherapy (eg, rituximab, inotuzumab, blinatumomab, or tisagenlecleucel).35 36 37 38

This study had some limitations. First, it used a retrospective design, which might have allowed incomplete reporting bias and missing data. For example, cytogenetic information at diagnosis was missing for 172 (24.2%) of 711 patients because of culture failure or poor bone marrow blast growth. Individuals with missing data were excluded during overall outcome analyses. However, our estimates might have been biased because of this restricted statistical analysis approach.39 Second, confounding factors (eg, selection bias and enrolment bias) might have been present. For example, the distributions of high-risk ALL subgroups (eg, Ph-like ALL and early-T-precursor ALL) were not examined in our analysis because of limited data. Therefore, caution is needed when interpreting the results of this study.

In conclusion, our analysis of children with ALL suggested that long-term EFS and OS rates were favourable among adolescents who received intensive paediatric treatment protocols. However, ALL treatment outcomes were worse among adolescents than among young children; further optimisation is needed to reduce treatment-related mortality. Novel targeted agents for patients with poor early response to ALL treatment may overcome treatment resistance, eradicate MRD, and improve clinical outcomes.

Concept or design: CK Li.
Acquisition of data: FWT Cheng, AKS Chiang, GKS Lam, TTW Chow, SY Ha, CW Luk, CH Li, SC Ling, PW Yau, KKH Ho, AWK Leung.
Analysis or interpretation of data: J Feng, FWT Cheng, AWK Leung, NPH Chan, MHL Ng, CK Li.
Drafting of the manuscript: J Feng.
Critical revision of the manuscript for important intellectual content: FWT Cheng, AKS Chiang, GKS Lam, TTW Chow, SY Ha, CW Luk, CH Li, SC Ling, PW Yau, KKH Ho, AWK Leung, NPH Chan, MHL Ng, CK Li.

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

All authors have disclosed no conflicts of interest.

The authors thank Ms H Wong for contributing to the data collection.

The Children’s Cancer Foundation provided technical support for data management and funding for minimal residual disease testing.

This study was approved by The Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee (CRE2008.007T).

1. Pichler H, Reismüller B, Steiner M, et al. The inferior prognosis of adolescents with acute lymphoblastic leukaemia (ALL) is caused by a higher rate of treatmentrelated mortality and not an increased relapse rate—a population-based analysis of 25 years of the Austrian ALL-BFM (Berlin-Frankfurt-Munster) Study Group. Br J Haematol 2013;161:556-65. Crossref

2. Pole JD, Alibhai SM, Ethier MC, et al. Adolescents with acute lymphoblastic leukemia treated at pediatric versus adult hospitals. Ann Oncol 2013;24:801-6. Crossref

3. Boudestein K, Kamps WA, Veerman AJ, Pieters R. Different outcome in older children with acute lymphoblastic leukemia with different treatment protocols in the Netherlands. Pediatr Blood Cancer 2012;58:17-22. Crossref

4. Pui CH, Pei D, Campana D, et al. Improved prognosis for older adolescents with acute lymphoblastic leukemia. J Clin Oncol 2011;29:386-91. Crossref

5. Boissel N, Baruchel A. Acute lymphoblastic leukemia in adolescent and young adults: treat as adults or as children? Blood 2018;132:351-61. Crossref

6. Boissel N, Auclerc MF, Lheritier V, et al. Should adolescents with acute lymphoblastic leukemia be treated as old children or young adults? Comparison of the French FRALLE-93 and LALA-94 trials. J Clin Oncol 2003;21:774-80. Crossref

7. Plasschaert SL, Kamps WA, Vellenga E, de Vries EG, de Bont ES. Prognosis in childhood and adult acute lymphoblastic leukaemia: a question of maturation? Cancer Treat Rev 2004;30:37-51. Crossref

8. Moricke A, Zimmermann M, Reiter A, et al. Prognostic impact of age in children and adolescents with acute lymphoblastic leukemia: data from the trials ALL-BFM 86, 90, and 95. Klin Padiatr 2005;217:310-20. Crossref

9. Usvasalo A, Raty R, Knuutila S, et al. Acute lymphoblastic leukemia in adolescents and young adults in Finland. Haematologica 2008;93:1161-8. Crossref

10. Stock W, La M, Sanford B, et al. What determines the outcomes for adolescents and young adults with acute lymphoblastic leukemia treated on cooperative group protocols? A comparison of Children’s Cancer Group and Cancer and Leukemia Group B studies. Blood 2008;112:1646-54. Crossref

11. Ribera JM, Oriol A, Sanz MA, et al. Comparison of the results of the treatment of adolescents and young adults with standard-risk acute lymphoblastic leukemia with the Programa Espanol de Tratamiento en Hematologia pediatric-based protocol ALL-96. J Clin Oncol 2008;26:1843-9. Crossref

12. Ramanujachar R, Richards S, Hann I, et al. Adolescents with acute lymphoblastic leukaemia: outcome on UK national paediatric (ALL97) and adult (UKALLXII/E2993) trials. Pediatr Blood Cancer 2007;48:254-61. Crossref

13. Kato M, Manabe A, Koh K, et al. Treatment outcomes of adolescent acute lymphoblastic leukemia treated on Tokyo Children’s Cancer Study Group (TCCSG) clinical trials. Int J Hematol 2014;100:180-7. Crossref

14. Chiaretti S, Vitale A, Cazzaniga G, et al. Clinico-biological features of 5202 patients with acute lymphoblastic leukemia enrolled in the Italian AIEOP and GIMEMA protocols and stratified in age cohorts. Haematologica 2013;98:1702-10. Crossref

15. Testi AM, Attarbaschi A, Valsecchi MG, et al. Outcome of adolescent patients with acute lymphoblastic leukaemia aged 10-14 years as compared with those aged 15-17 years: long-term results of 1094 patients of the AIEOP-BFM ALL 2000 study. Eur J Cancer 2019;122:61-71. Crossref

16. Li CK, Chik KW, Chan GC, et al. Treatment of acute lymphoblastic leukemia in Hong Kong children: HKALL 93 study. Hematol Oncol 2003;21:1-9. Crossref

17. Li CK, Chik KW, Ha SY, et al. Improved outcome of acute lymphoblastic leukaemia treated by delayed intensification in Hong Kong children: HKALL 97 study. Hong Kong Med J 2006;12:33-9.

18. Stary J, Zimmermann M, Campbell M, et al. Intensive chemotherapy for childhood acute lymphoblastic leukemia: results of the randomized intercontinental trial ALL IC-BFM 2002. J Clin Oncol 2014;32:174-84. Crossref

19. Cui L, Li ZG, Chai YH, et al. Outcome of children with newly diagnosed acute lymphoblastic leukemia treated with CCLG-ALL 2008: the first nation-wide prospective multicenter study in China. Am J Hematol 2018;93:913-20. Crossref

20. Biondi A, Schrappe M, De Lorenzo P, et al. Imatinib after induction for treatment of children and adolescents with Philadelphia-chromosome-positive acute lymphoblastic leukaemia (EsPhALL): a randomised, open-label, intergroup study. Lancet Oncol 2012;13:936-45. Crossref

21. Barry E, DeAngelo DJ, Neuberg D, et al. Favorable outcome for adolescents with acute lymphoblastic leukemia treated on Dana-Farber Cancer Institute Acute Lymphoblastic Leukemia Consortium Protocols. J Clin Oncol 2007;25:813-9. Crossref

22. Nachman JB, La MK, Hunger SP, et al. Young adults with acute lymphoblastic leukemia have an excellent outcome with chemotherapy alone and benefit from intensive postinduction treatment: a report from the children’s oncology group. J Clin Oncol 2009;27:5189-94. Crossref

23. Siegel SE, Stock W, Johnson RH, et al. Pediatric-inspired treatment regimens for adolescents and young adults with Philadelphia chromosome-negative acute lymphoblastic leukemia: a review. JAMA Oncol 2018;4:725-34. Crossref

24. de Bont JM, Holt B, Dekker AW, van der Does-van den Berg A, Sonneveld P, Pieters R. Significant difference in outcome for adolescents with acute lymphoblastic leukemia treated on pediatric vs adult protocols in the Netherlands. Leukemia 2004;18:2032-5. Crossref

25. Siegel SE, Advani A, Seibel N, et al. Treatment of young adults with Philadelphia-negative acute lymphoblastic leukemia and lymphoblastic lymphoma: hyper-CVAD vs. pediatric-inspired regimens. Am J Hematol 2018;93:1254-66. Crossref

26. Larsen EC, Devidas M, Chen S, et al. Dexamethasone and high-dose methotrexate improve outcome for children and young adults with high-risk B-acute lymphoblastic leukemia: a report from Children’s Oncology Group Study AALL0232. J Clin Oncol 2016;34:2380-8. Crossref

27. Carobolante F, Chiaretti S, Skert C, Bassan R. Practical guidance for the management of acute lymphoblastic leukemia in the adolescent and young adult population. Ther Adv Hematol 2020;11:2040620720903531. Crossref

28. Hough R, Rowntree C, Goulden N, et al. Efficacy and toxicity of a paediatric protocol in teenagers and young adults with Philadelphia chromosome negative acute lymphoblastic leukaemia: results from UKALL 2003. Br J Haematol 2016;172:439-51. Crossref

29. Pieters R, de Groot-Kruseman H, Van der Velden V, et al. Successful therapy reduction and intensification for childhood acute lymphoblastic leukemia based on minimal residual disease monitoring: Study ALL10 from the Dutch Childhood Oncology Group. J Clin Oncol 2016;34:2591-601. Crossref

30. DeAngelo DJ, Stevenson KE, Dahlberg SE, et al. Long-term outcome of a pediatric-inspired regimen used for adults aged 18-50 years with newly diagnosed acute lymphoblastic leukemia. Leukemia 2015;29:526-34. Crossref

31. Schmiegelow K, Heyman M, Gustafsson G, et al. The degree of myelosuppression during maintenance therapy of adolescents with B-lineage intermediate risk acute lymphoblastic leukemia predicts risk of relapse. Leukemia 2010;24:715-20. Crossref

32. Gupta A, Matloub Y, Damania R, O’Riordan M, Ahuja SP. Increased toxicity among adolescents and young adults treated for acute lymphoblastic leukemia at US Children’s Hospitals [abstract]. Blood 2017;130(Suppl 1):222.

33. Conter V, Bartram CR, Valsecchi MG, et al. Molecular response to treatment redefines all prognostic factors in children and adolescents with B-cell precursor acute lymphoblastic leukemia: results in 3184 patients of the AIEOP-BFM ALL 2000 study. Blood 2010;115:3206-14. Crossref

34. Moricke A, Zimmermann M, Valsecchi MG, et al. Dexamethasone vs prednisone in induction treatment of pediatric ALL: results of the randomized trial AIEOP-BFM ALL 2000. Blood 2016;127:2101-12. Crossref

35. Kantarjian H, Stein A, Gokbuget N, et al. Blinatumomab versus chemotherapy for advanced acute lymphoblastic leukemia. N Engl J Med 2017;376:836-47. Crossref

36. Lo Nigro L, Pulvirenti G, Cannata E, Bonaccorso P, Andriano N, Russo G. “Feasible and effective administration of Bortezomib with Rituximab in children with relapsed/resistant B-cell precursor acute lymphoblastic leukemia (BCP-ALL): a step toward the first line”. Pediatr Hematol Oncol 2019;36:438-44. Crossref

37. Maude SL, Laetsch TW, Buechner J, et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N Engl J Med 2018;378:439-48. Crossref

38. Kantarjian HM, DeAngelo DJ, Stelljes M, et al. Inotuzumab ozogamicin versus standard therapy for acute lymphoblastic leukemia. N Engl J Med 2016;375:740-53. Crossref

39. Mallinckrodt CH, Sanger TM, Dube S, et al. Assessing and interpreting treatment effects in longitudinal clinical trials with missing data. Biol Psychiatry 2003;53:754-60. Crossref