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

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

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

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

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

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

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

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

Standard errors (SEs) of the 50th percentile were estimated using the expression , assuming that the SE of the percentile of interest can be expressed as a multiple of the standard deviation (SD).13 14

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

All authors have disclosed no conflicts of interest.

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

The International Study of Asthma and Allergies in Childhood (ISAAC) is considered a landmark international investigation of the global burden of three major atopic diseases—asthma, allergic rhinitis, and eczema1 2 3 4—utilising standardised methodology to enable accurate estimation of prevalence trends in allergic diseases. Phase One examined the prevalence and severity of the three diseases in 1994-1995, Phase Two investigated the corresponding risk factors, and Phase Three constituted a follow-up study examining global prevalence and severity trends in 2001-2003.1 2 4 The Global Asthma Network updated global burden information using data from 27 Global Asthma Network Centres in 14 countries in 2017-2020,3 but Hong Kong data were excluded from that study.

Asthma remains a major medical burden responsible for nearly 500 000 deaths in 20175 and substantial direct, indirect, and intangible costs,6 ranging from mortality and hospitalisations to school- or work-related loss (in the form of absenteeism from school or loss of working days), and quality of life impairments. Thus, accurate and current asthma prevalence data are essential for guiding public health initiatives and formulating healthcare policies. Recent estimates suggest global asthma prevalences of 9.1% in children, 11.0% in adolescents, and 6.6% in adults7; asthma is often cited as the most common chronic disease in children.6

Trends regarding the three major atopic diseases have considerably varied among countries and regions since 1994-1995. Asthma has shown the greatest variation, peaking in some countries but continuing to increase in others, especially developing countries.2 3 4 5 6 8 9 10 11 12 13 14 15 16 Most studies of asthma prevalence were conducted before the coronavirus disease 2019 (COVID-19) pandemic, and statistics considerably differ even within the same region. In 2019, a large-scale study examining the global adult and paediatric prevalences of asthma in >200 territories showed a 24% decrease in age-standardised point prevalence in most countries5; however, increases were observed in Oman, Saudi Arabia, and Vietnam.5 This discrepancy highlights the need for further investigation of the risk and protective factors for asthma in modern societies.

Hong Kong is an urban-centric port city in southeast China with a unique blend of Chinese and international influences, which have led to trends that considerably differ from those of neighbouring regions.17 Cultural practices17 18 19 and urbanisation (eg, cooking methods, joss stick burning, and high population density in urban areas)9 17 20 21 increase the complexity involved in estimating asthma prevalence within Hong Kong based on data from neighbouring regions. This study presents contemporary epidemiological data regarding the prevalence and severity of asthma and wheezing among children in Hong Kong, which provides an update to the results of the ISAAC conducted in 1994-1995 and 2001-2003.

This cross-sectional study, based on the ISAAC study protocol,2 was performed using traditional Chinese versions of the validated ISSAC measurements.22

In total, 2148 children aged 6 to 7 years or 13 to 14 years residing in Hong Kong were recruited from primary and secondary schools between September 2020 and August 2021. Inclusion criteria for children were enrolment in Primary 1-2 (Grades 1-2) or Secondary 2-3 (Grades 8-9) in Hong Kong and parental provision of informed consent to participate. The primary caregiver of the student (for the primary school group) or the student himself/herself (for the secondary school group) was asked to complete questionnaires in accordance with the ISAAC protocol. Only written questionnaires were used, considering reports of high agreement with their video counterparts.23 24 Inclusion criteria for parents were: (1) being either the father or mother of the child, (2) bearing primary responsibility for the child’s care for ≥6 months, and (3) provision of informed consent to participate. Exclusion criteria for parents and children were the presence of a learning disability or organic disorder that would impair the ability to understand and respond to the questionnaires, and inability to understand Chinese. Additionally, children attending Special Educational Needs schools were excluded.

All schools in Hong Kong (n=900) were invited by phone and mail to participate in the survey. Contact persons were identified in each school. The investigators made at least three attempts per school (by phone, mail and/or email) to obtain responses from participating students. A target sample size of 1000 to 3000 participants was established, in accordance with the ISAAC study protocol.2

Questions were adopted from the ISAAC study protocol, and disease definitions were based on the original ISAAC Phase Three questionnaire and handbook.2 ‘Current severe asthma’ was defined as affirmative responses to one or more of the following items in the past 12 months: (1) ≥4 wheezing episodes, (2) woken from sleep by wheezing ≥1 night per week, or (3) limitation of speech during wheezing. Other questions and their definitions adhered to the standardised ISAAC Coding and Data Transfer Manual2 22 to ensure comparability with previous studies utilising the ISAAC protocol.

Descriptive statistics, including means and standard deviations, were used for continuous variables; frequency distributions were used for categorical variables. Sex differences within each age-group were assessed using Chi squared and independent t tests. Demographic information was entered into hierarchical logistic regression models to identify potential predictive factors associated with asthma severity. Annual change in prevalence was calculated via division of the prevalence by the number of years between surveys. Logistic regression modelling was conducted to identify risk factors associated with asthma severity. Potential risk factors included parent gender, parent education, and household monthly income. All analyses were performed using SPSS software (Windows version 26.0; IBM Corp, Armonk [NY], United States). P values <0.05 were considered statistically significant.

Nineteen primary schools and 25 secondary schools accepted the invitation to participate, representing 3748 and 5771 eligible students in the respective groups. Overall, 1970 (52.6%) questionnaires were returned for the primary school group, whereas 1968 (34.1%) questionnaires were returned for the secondary school group; 1100 and 1048 questionnaires were analysed in the respective groups after exclusion due to invalid consent, incomplete data, or ineligibility (Fig). Boys comprised 62.51% and 42.64% of the primary and secondary school groups, respectively (Tables 1 and 2). The mean ages were 7.02 years (standard deviation [SD]=0.76) in the primary school group and 14.09 years (SD=0.89) in the secondary school group.

The estimated prevalences of current wheezing were 6.19% and 4.97% in the primary school and secondary school groups, respectively. The respective prevalences of exercise-induced wheezing and night cough were 7.46% and 16.74% in the primary school group and 17.88% and 10.61% in the secondary school group (Tables 1 and 2).

The prevalences of asthma ever were 5.55% and 6.12% in the primary school and secondary school groups, respectively; significantly more boys reported asthma ever in the secondary school group. The prevalences of bronchial hypersensitivity ever were 33.03% and 11.09% in the primary and secondary school groups, respectively (Tables 1, 2, 3). In total, 85.25% and 78.13% of primary and secondary school participants with asthma exhibited comorbid eczema and/or allergic rhinitis, respectively (Table 4)

Indicators of severe asthma were also examined. The incidences of ≥4 wheezing episodes in the past year were 2.18% (boys vs girls: 3.20% vs 0.49%) in the primary school group and 1.91% (boys vs girls: 2.24% vs 1.67%) in the secondary school group. Severe wheezing that limited speech in the past year was reported by 0.64% and 1.43% of participants in the primary and secondary school groups, respectively. Waking from sleep due to wheezing in the past year occurred in 2.82% and 1.24% of participants in the primary and secondary school groups, respectively. Hospitalisation due to shortness of breath was reported by 7.37% (boys vs girls: 8.30% vs 5.83%) and 4.30% (boys vs girls: 4.93% vs 3.83%) of participants in the primary and secondary school groups, respectively (Tables 1 and 2). The mean ages of wheezing onset in the primary and secondary school groups were 2.37 years (SD=1.52) and 7.16 years (SD=4.38), respectively.

Compared with the previous Hong Kong ISAAC data,1 25 26 27 the prevalence of wheezing ever increased from 16.80%26 in 1994-1995 to 20.38% in the primary school group but decreased from 20%25 to 12.05% in the secondary school group. Conversely, the prevalence of asthma ever decreased from 7.80%26 to 5.55% in the primary school group and from 11%26 to 6.12% in the secondary school group. The prevalence of wheezing in the past year decreased from 9.20% in 1994-199526 and 9.40% in 2000-200127 to 6.19% in our study (2020-2021) in the primary school group; it decreased from 12% in 1994-1995 to 4.97% in our study in the secondary school group. The annual changes in wheezing prevalence were -0.16% and -0.27% in the primary and secondary school groups, respectively. The aforementioned indicators of severe asthma, including ≥4 wheezing episodes in the past year and severe wheezing that limited speech in the past year, decreased in prevalence compared with the 1994-1995 figures. However, the prevalence of waking from sleep due to wheezing in the past year decreased in the primary group and increased in the secondary school group (Table 5).

Overall, both the prevalence and severity of asthma declined compared with the 1994-1995 data. We observed a statistically significant male predominance for wheezing ever (P=0.003), current wheezing (P=0.012), and bronchial hypersensitivity diagnosis (P=0.002) in the primary school group (Table 1), as well as for asthma ever in the secondary school group (P=0.005) [Table 2].

In the logistic regression model, demographic characteristics, including parent gender, household income, and parent education, were not significantly predictive of wheezing episodes in the past year. Hierarchical logistic regression using demographic characteristics along with concomitant current eczema and rhinitis also did not show significantly predictive effects for wheezing episodes in the past year.

This is the first study since 2001 to investigate trends in asthma prevalence and severity among school children in Hong Kong. The global variability in these trends is complicated by the lack of consensus regarding exact definitions of asthma entities, the heterogeneity of the disease itself, changes in community awareness, the absence of a ‘gold standard’ diagnostic test, and the non-specific nature of symptoms shared with other diagnoses.13 This study showed a decrease in asthma prevalence, consistent with findings from the neighbouring region of Taiwan.11

The 2020 study by Asher et al4 investigating international symptom trends showed annual current asthma prevalence increases of 0.06% in the secondary school group (from 13.2% to 13.7%) and 0.13% in the primary school group (from 11.1% to 11.6%), and increases in the prevalence of asthma ever by 0.18% and 0.28% in the primary and secondary school groups, respectively. However, trends varied across regions; in general, asthma prevalence decreased in higher-income regions but increased in low- and middle-income regions. Considering that Hong Kong exhibits higher levels of income and gross domestic product per capita, it is unsurprising that the local asthma prevalences showed a decreasing trend.

A decrease in asthma severity was also observed in most regions, reflected by the three indicators of severe asthma in our study and parameters such as asthma-related hospital admissions and mortality. These findings suggest that, regardless of overall asthma trends, milder forms of asthma have become more prevalent in recent years.7 28 Our study revealed similar trends in terms of fewer severe asthmatic exacerbations and flares. However, according to the ISAAC protocol, centres with 1000 to 2999 participants are considered appropriate for comparisons of prevalence, but not severity, with centres in other regions.2

Safiri et al5 identified the highest asthma prevalence among 5- to 9-year-olds, which then decreased and remained stable until adulthood. A proportion of young children with wheezing may have ‘transient wheezing’ that does not progress to asthma.4 This notion is consistent with our finding of less frequent symptoms in the past year among older children.5 The prevailing view is that approximately 50% of preschool children with wheezing will progress to asthma by the time of primary school entry.4 The prevalences of 12.05% and 6.12% for wheezing ever and asthma ever observed in our secondary school group are consistent with this view.

Our results showed a male predominance for asthma and wheezing in the primary school group, consistent with existing literature.29 Various mechanisms have been proposed to explain the post-pubertal shift towards female predominance due to sex hormone changes, genetic and epigenetic differences, co-morbidities, and socio-economic factors.14 The significantly older age of wheezing onset reported in the secondary school group, along with the lower prevalence of wheezing ever in the secondary school group, may be attributable to recall bias due to the use of questionnaires, a limitation noted in other studies.30

Co-morbidities with other atopic diseases were common among our asthma participants; most participants in both groups had concomitant atopic dermatitis, allergic rhinitis, and asthma, consistent with previous reports.8 16 Otherwise, our study did not identify relevant demographic characteristics or risk factors through statistical analysis that could predict wheezing episodes. Many studies have investigated various factors potentially associated with asthma diagnosis, wheezing, exercise-induced symptoms, and nocturnal cough. These factors include, but are not limited to, family history, recurrent respiratory infections, early-life severe respiratory syncytial virus infection, exposure to cigarette smoke, exposure to pets, incense burning, maternal education level, sex, race, vaccination rates, humidity, air pollution index (particulate matter) and exposure (particularly nitrogen dioxide and sulphur dioxide), exposure to indoor mould, farm residence, exposure to indoor endotoxins, socio-economic status, obesity, open fire cooking, nutritional levels, neonatal antibiotic use, delivery mode, urban living environment, psychosocial environment (including maternal stress), current paracetamol use, maternal antibiotic use during pregnancy, maternal vitamin D consumption during pregnancy, maternal weight gain during pregnancy, maternal paracetamol use during pregnancy, proton-pump inhibitor and H2-receptor antagonist use, and new immigrant status.2 3 4 5 6 8 9 10 11 12 13 14 15 29 The next phase of our study will examine these predictive or protective factors; it will also explore risk factors unique to our population, including cultural perceptions,18 feeding and weaning practices,19 31 joss stick burning,15 and exposure to traditional Chinese herbal medicine.32 33

Due to its survey approach, this study has limitations of recall bias and cross-sectional design. The estimations of symptom prevalence also lack objective confirmation through medical assessments and objective tests, which could lead to over- or underestimation of the asthmatic population; this aspect is further complicated by diagnostic discrepancies among medical professionals across regions and time periods.4 34 Although our study population exceeded 1000 students per group, it was smaller than the original Hong Kong ISAAC studies, which included >300026 and 400025 participants, respectively. The paediatric population size has also significantly changed, according to the Hong Kong Population Census data: 86 000 13- to 14-year-olds in 1994 compared with 61 800 in 2021.35 The male-to-female ratio of 1.67 and 0.74 in the primary and secondary school groups, respectively, differed from the 2021 census figures (1.063 and 1.067, respectively)35; therefore, our findings may not reflect the true prevalence and severity of asthma in the Hong Kong paediatric population, which represents a key limitation of the study. Although the ISAAC protocol acknowledges the impracticality of video questionnaires due to logistic or technical factors,2 exclusive use of the written questionnaire may have impacted our results; notably, a study has shown that video and written questionnaires are comparable.23 The relatively large population size and the use of validated and standardised questionnaires may partially mitigate these limitations.

In this study, response rates were lower than in the previous ISAAC studies (where rates >80% for most centres),36 likely due to ‘survey fatigue’ and challenges in motivating participants during the COVID-19 pandemic. Similarly low response rates were observed in studies requesting parental completion of home questionnaires.37 Although no significant differences in airway disease or symptom prevalences have been identified between non-responders and responders,38 the lower response rates may have introduced selection bias into our study.

The increased prevalence of wheezing ever in the primary school group, decreased prevalence of wheezing ever in the secondary school group, decreased prevalences of asthma ever in both groups, and decreased symptom severity all reflect enhanced awareness and modified diagnostic practices among the general population and medical professionals.4 11 13 Doctors are less likely to classify patients as ‘asthmatic’ without collecting a thorough clinical history and performing diagnostic testing, leading to a larger difference in the prevalences of wheezing and asthma.11 13

The local COVID-19 policy, commonly known as the mask mandate, along with social isolation, increased awareness of infection control, changes in drug compliance, and enhanced hygiene practices, may have significantly reduced triggers for infectious and allergic airway diseases.39 40 However, mechanisms linking COVID-19 to severe asthma risk have also been proposed.41 These policies and social practices may explain the overall decreases in asthma prevalence and severity observed in our study. The timing of the study coincided with the COVID-19 pandemic, where the modified local health practices may have influenced trends concerning current asthma and airway symptoms.

This study provides an essential update regarding the prevalences of asthma and other respiratory symptoms among school children in Hong Kong. Our findings indicate overall decreasing trends in asthma severity and prevalence. A follow-up study will explore the protective and risk factors contributing to these trends.

Concept or design: JWCH Cheng, YP Tsang, YY Lam, AKY Chu, CHY Chan, YL Fung, PSY Chau, DCK Luk.
Acquisition of data: All authors.
Analysis or interpretation of data: All authors.
Drafting of the manuscript: JWCH Cheng, CSY Ng.
Critical revision of the manuscript for important intellectual content: JWCH Cheng, CSY Ng, CHY Chan, YL Fung.

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 all the children and their families for their participation in this research.

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

This research was approved by the Human Research Ethics Committee of The University of Hong Kong, Hong Kong (Ref No.: EA2002007). All study participants provided written consent for publication of their data which were de-identified in this article.

1. Asher MI, Montefort S, Björkstén B, et al. Worldwide time trends in the prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and eczema in childhood: ISAAC Phases One and Three repeat multicountry cross-sectional surveys. Lancet 2006;368:733-43. Crossref

2. Asher MI, Keil U, Anderson HR, et al. International Study of Asthma and Allergies in Childhood (ISAAC): rationale and methods. Eur Respir J 1995;8:483-91. Crossref

3. Asher MI, Rutter CE, Bissell K, et al. Worldwide trends in the burden of asthma symptoms in school-aged children: Global Asthma Network Phase I cross-sectional study. Lancet 2021;398:1569-80. Crossref

4. Asher MI, García-Marcos L, Pearce NE, Strachan DP. Trends in worldwide asthma prevalence. Eur Respir J 2020;56:2002094. Crossref

5. Safiri S, Carson-Chahhoud K, Karamzad N, et al. Prevalence, deaths, and disability-adjusted life-years due to asthma and its attributable risk factors in 204 countries and territories, 1990-2019. Chest 2022;161:318-29. Crossref

6. Ferrante G, La Grutta S. The burden of pediatric asthma. Front Pediatr 2018;6:186. Crossref

7. The Global Asthma Report 2022 [editorial]. Int J Tuberc Lung Dis 2022;26(Suppl 1):1-104 Crossref

8. Kansen HM, Le TM, Uiterwaal C, et al. Prevalence and predictors of uncontrolled asthma in children referred for asthma and other atopic diseases. J Asthma Allergy 2020;13:67-75. Crossref

9. Rodriguez A, Brickley E, Rodrigues L, Normansell RA, Barreto M, Cooper PJ. Urbanisation and asthma in low-income and middle-income countries: a systematic review of the urban–rural differences in asthma prevalence. Thorax 2019;74:1020-30.Crossref

10. Castro-Rodriguez JA, Forno E, Rodriguez-Martinez CE, Celedón JC. Risk and protective factors for childhood asthma: what is the evidence? J Allergy Clin Immunol Pract 2016;4:1111-22. Crossref

11. Chen WY, Lin CW, Lee J, Chen PS, Tsai HJ, Wang JY. Decreasing ten-year (2008-2018) trends of the prevalence of childhood asthma and air pollution in Southern Taiwan. World Allergy Organ J 2021;14:100538. Crossref

12. Akinbami LJ, Simon AE, Rossen LM. Changing trends in asthma prevalence among children. Pediatrics 2016;137:1-7. Crossref

13. Sears MR. Trends in the prevalence of asthma. Chest 2014;145:219-25. Crossref

14. Chowdhury NU, Guntur VP, Newcomb DC, Wechsler ME. Sex and gender in asthma. Eur Respir Rev 2021;30:210067. Crossref

15. Lin TC, Krishnaswamy G, Chi DS. Incense smoke: clinical, structural and molecular effects on airway disease. Clin Mol Allergy 2008;6:3. Crossref

16. Arasi S, Porcaro F, Cutrera R, Fiocchi AG. Severe asthma and allergy: a pediatric perspective. Front Pediatr 2019;7:28. Crossref

17. Wong GW, Ko FW, Hui DS, et al. Factors associated with difference in prevalence of asthma in children from three cities in China: multicentre epidemiological survey. BMJ 2004;329:486. Crossref

18. Leo HL, Wang A, Gong M, Clark N. Cultural perceptions on identification and treatment of childhood asthma in the US and China. J Allergy Clin Immunol 2006;117 Suppl:S53. Crossref

19. Mihrshahi S, Ampon R, Webb K, et al. The association between infant feeding practices and subsequent atopy among children with a family history of asthma. Clin Exp Allergy 2007;37:671-9. Crossref

20. Wong GW, Brunekreef B, Ellwood P, et al. Cooking fuels and prevalence of asthma: a global analysis of phase three of the International Study of Asthma and Allergies in Childhood (ISAAC). Lancet Respir Med 2013;1:386-94. Crossref

21. Asher MI. Urbanisation, asthma and allergies. Thorax 2011;66:1025-6. Crossref

22. Chan HH, Pei A, Van Krevel C, Wong GW, Lai CK. Validation of the Chinese translated version of ISAAC core questions for atopic eczema. Clin Exp Allergy 2001;31:903-7. Crossref

23. Lai CK, Chan JK, Chan A, et al. Comparison of the ISAAC video questionnaire (AVQ3.0) with the ISAAC written questionnaire for estimating asthma associated with bronchial hyperreactivity. Clin Exp Allergy 1997;27:540-5. Crossref

24. Becerir T, Akcay A, Duksal F, Ergin A, Becerir C, Guler N. Prevalence of asthma, local risk factors and agreement between written and video questionnaires among Turkish adolescents. Allergol Immunopathol (Madr) 2014;42:594-602. Crossref

25. Leung R, Wong G, Lau J, et al. Prevalence of asthma and allergy in Hong Kong schoolchildren: an ISAAC study. Eur Respir J 1997;10:354-60. Crossref

26. Lau YL, Karlberg J. Prevalence and risk factors of childhood asthma, rhinitis and eczema in Hong Kong. J Paediatr Child Health 1998;34:47-52. Crossref

27. Lee SL, Wong W, Lau YL. Increasing prevalence of allergic rhinitis but not asthma among children in Hong Kong from 1995 to 2001 (Phase 3 International Study of Asthma and Allergies in Childhood). Pediatr Allergy Immunol 2004;15:72-8. Crossref

28. Ebmeier S, Thayabaran D, Braithwaite I, Bénamara C, Weatherall M, Beasley R. Trends in international asthma mortality: analysis of data from the WHO Mortality Database from 46 countries (1993-2012). Lancet 2017;390:935-45. Crossref

29. Estela DB, Arturo B, Nayely RN, et al. Have asthma symptoms in Mexico changed in the past 15 years? Time trends from the International Study of Asthma and Allergies in Childhood to the Global Asthma Network. Allergol Immunopathol (Madr) 2021;49:1-10. Crossref

30. Strachan DP. The prevalence and natural history of wheezing in early childhood. J R Coll Gen Pract 1985;35:182-4.

31. Yung J, Yuen JW, Ou Y, Loke AY. Factors associated with atopy in toddlers: a case-control study. Int J Environ Res Public Health 2015;12:2501-20. Crossref

32. Li XM. Traditional Chinese herbal remedies for asthma and food allergy. J Allergy Clin Immunol 2007;120:25-31. Crossref

33. Li XM, Brown L. Efficacy and mechanisms of action of traditional Chinese medicines for treating asthma and allergy. J Allergy Clin Immunol 2009;123:297-306; quiz 307-8. Crossref

34. Hederos CA, Hasselgren M, Hedlin G, Bornehag CG. Comparison of clinically diagnosed asthma with parental assessment of children’s asthma in a questionnaire. Pediatr Allergy Immunol 2007;18:135-41. Crossref

35. Census and Statistics Department, Hong Kong SAR Government. Population estimates. Table 110-01002: Population by sex and age. Available from: https://www.censtatd.gov.hk/en/web_table.html?id=110-01002#. Accessed 14 Oct 2024.

36. Worldwide variations in the prevalence of asthma symptoms: the International Study of Asthma and Allergies in Childhood (ISAAC). Eur Respir J 1998;12:315-35. Crossref

37. Ellwood P, Ellwood E, Rutter C, et al. Global Asthma Network Phase I surveillance: geographical coverage and response rates. J Clin Med 2020;9:3688. Crossref

38. Rönmark EP, Ekerljung L, Lötvall J, Torén K, Rönmark E, Lundbäck B. Large scale questionnaire survey on respiratory health in Sweden: effects of late- and nonresponse. Respir Med 2009;103:1807-15. Crossref

39. Izquierdo-Domínguez A, Rojas-Lechuga MJ, Alobid I. Management of allergic diseases during COVID-19 outbreak. Curr Allergy Asthma Rep 2021;21:8. Crossref

40. Carr TF, Kraft M. Asthma and atopy in COVID-19: 2021 updates. J Allergy Clin Immunol 2022;149:562-4. Crossref

41. Hosoki K, Chakraborty A, Sur S. Molecular mechanisms and epidemiology of COVID-19 from an allergist’s perspective. J Allergy Clin Immunol 2020;146:285-99. Crossref

Atopic eczema (AE) is a common childhood skin disease associated with pruritus and sleep disturbance.1 2 3 4 5 Childhood AE can substantially influence quality of life (QOL) among affected children and their parents. The extent of QOL impairment is often correlated with eczema severity, skin dehydration, and staphylococcal infection.6 Additionally, many affected children and their parents develop depression, anxiety, and stress symptoms.1 These mental health issues are correlated with disease severity, impaired QOL, and therapeutic non-compliance.1 7 8 A study using the 42-item Depression, Anxiety, and Stress Scales (DASS-42) found that depression, anxiety, and stress symptoms were present in 21%, 33%, and 23% of Hong Kong adolescent patients with AE, respectively.1 These psychological symptoms were significantly correlated with poor QOL.1 A study of Japanese children showed eczema severity was associated with mental health.9 Furthermore, a retrospective, cross-sectional population-based survey of childhood eczema in the United States revealed that increased eczema severity was associated with a higher risk of mental disorders.10 School-aged children with moderate and severe AE have a higher risk of psychosocial problems that can influence their quality of sleep and cognitive development.11 Emotion, attention, interpersonal relationships, and conduct can also be affected by AE.10 12 Moreover, parents are often unaware of potential psychosocial health issues in their children with eczema.13 Most affected children and their parents do not receive appropriate psychological help and support; they also exhibit low symptom recognition. The impacts of childhood eczema on the parent-child dyad have not been extensively studied in terms of healthrelated quality of life (HRQOL), eczema severity, or mental health status.14

This study was performed to examine associations between mental health issues and disease severity in children or adolescents with AE and their parents using the concise validated 21-item Chinese version of DASS-42 (DASS-21).1 15 16 17

This 2-year cross-sectional survey was conducted between November 2018 and October 2020. Participants with a diagnosis of AE and their parents were recruited at the paediatric dermatology clinic of a university-affiliated hospital in Hong Kong. Eczema was clinically diagnosed in accordance with the United Kingdom Working Party’s Diagnostic Criteria for Atopic Dermatitis.18 Participants and their parents received information about the study including objectives, procedures, voluntary participation, and right of withdrawal. Non-Chinese participants and individuals aged <11 years who were not accompanied by a parent during recruitment were excluded from the study.

The questionnaires were self-administered and supervised by research staff. Parents would help complete the Children’s Dermatology Life Quality Index (CDLQI), Infants’ Dermatitis Quality of Life Index (IDQOL), Nottingham Eczema Severity Score (NESS), and Patient-Oriented Eczema Measure (POEM) for their younger children. The DASS-21 was individually administered to all parents and to children aged >11 years.

The validated Chinese version of the three-item NESS, completed by children aged >11 years or the parents of children aged ≤11 years, was used to determine eczema severity in participating children.19 20 The presence of eczema and number of nights affected by skin itchiness each week in the past 12 months were rated from 1 to 5. A higher score indicated greater eczema severity. Additionally, areas of skin with eczematous lesions (eg, rash, lichenified skin, and/or bleeding) were recorded. Scores on the NESS were categorised as mild (3 to 8), moderate (9 to 11), and severe (12 to 15) eczema. Subjective measurements were determined using the validated seven-item Chinese translation of the POEM, which was also completed by children aged >11 years or the parents of children aged ≤11 years.21 22 Each item was scored from 0 to 4, with a maximum aggregate score of 28. A higher score indicated greater eczema severity in the past week (ranges of 0-2, 3-7, 8-16, 17-24, and 25-28 correspond to clear, mild, moderate, severe, and very severe levels of eczema, respectively).

Health-related QOL was evaluated using the Chinese version of the 10-item CDLQI23 and the 10-item IDQOL.24 The CDLQI was completed by children aged ≥4 years with guidance from parents, whereas the IDQOL was completed by parents of children aged <4 years. Each item on the two scales was scored from 0 to 3, with a maximum aggregate score of 30. A higher score indicated greater eczema-related HRQOL impairment in the past week (CDLQI ranges of 0-1, 2-6, 7-12, 13-18, and 19-30 correspond to no, small, moderate, very large, and extremely large effects on HRQOL, respectively; these ranges for IDQOL are 0-1, 2-5, 6-10, 11-20, and 21-30, respectively).

Mental health was assessed by measuring depression, anxiety, and stress in children with eczema and their parents using the validated Chinese version of the DASS-21. This scale has been used to examine symptoms of depression, anxiety, and stress among individuals with dermatitis or eczema.1 17 The DASS-21 was individually administered to all parents and to children aged >11 years. The DASS-21 composite score can be divided into the DASS Depression, DASS Anxiety, and DASS Stress domains. The total score for each domain ranges from 0 to 42.25 A higher score indicates greater emotional distress in that domain (DASS Depression ranges of 0-9, 10-13, 14-20, 21-27, and ≥28 correspond to normal, mild, moderate, severe, and extremely severe levels, respectively; these ranges for DASS Anxiety are 0-7, 8-9, 10-14, 15-19, and ≥20, whereas they are 0-14, 15-18, 19-25, 26-33, and ≥34 for DASS Stress).

Clinical data were de-identified and analysed using SPSS (Windows version 25.0; IBM Corp, Armonk [NY], United States). Frequency distributions were used to describe the demographic and clinical characteristics of participants. Continuous variables with normal distributions were expressed as means±standard deviations (corrected to 1 decimal place). Nominal and ordinal variables were expressed in numbers with percentage (corrected to 1 decimal place). Independent samples t tests were used to explore sex differences regarding age, education level, disease severity, QOL, and emotional distress in parents and children. Pearson correlation analysis was utilised to examine associations among mental health, eczema severity, and HRQOL in parents and children of both sexes. Multiple linear regression was performed to adjust for variations in parental and child DASS scores and HRQOL according to demographic and clinical variables. P values <0.05 were considered statistically significant.

Among 380 parents (mean age=41.13±6.52 years), 49 were fathers (mean age=42.50±7.49 years) and 331 were mothers (mean age=40.95±6.36 years). Parents’ education levels were primary to secondary (n=124, 13 males), post-secondary (n=26, 1 male), and undergraduate or above (n=81, 14 males) [Table 1]. Parents reported moderate to extremely severe depression (n=58, 2 males), moderate to extremely severe anxiety (n=101, 5 males), and moderate to extremely severe stress (n=90, 6 males). Parents with a higher education level had lower levels of depression, anxiety, and stress (Fig 1). Compared with fathers, mothers generally had higher overall DASS-21 depression, anxiety, and stress scores (P<0.001-0.005) [Table 1].

Among 432 children (mean age=9.61±5.41 years), 218 were boys (mean age=9.15±5.44 years) and 214 were girls (mean age=10.06±5.35 years). Most children had moderate to severe/very severe disease according to the POEM (n=290) and NESS (n=291). Over half of the children displayed a moderate to extremely large impact on QOL in the CDLQI (n=171, 50.4%) and IDQOL (n=56, 62.9%). Small numbers of children had moderate to extremely severe depression (n=36), anxiety (n=43), and stress (n=30). There were no significant sex differences in disease severity, HRQOL, or emotional distress in the DASS (Table 2).

Disease severity in terms of NESS, POEM, and HRQOL (ie, CDLQI and IDQOL) was generally worse among infants than among older children (Fig 2). Thus, eczema severity and QOL generally appeared to improve with age. Correlation analysis demonstrated that depression, anxiety, and stress levels were significantly associated with NESS, POEM, and CDLQI, regardless of sex (Table 3).

Correlation analysis revealed that eczema severity (NESS and POEM) and HRQOL (CDLQI) were associated with depression, anxiety, and stress levels (DASS-21) among children and parents, regardless of sex (Table 3 and Fig 3). Moreover, depression, anxiety, and stress levels in mothers were significantly correlated with NESS, POEM, IDQOL, and CDLQI. Paternal anxiety and stress levels were correlated with NESS, POEM, and CDLQI (P<0.001 to 0.034). However, paternal depression was only correlated with POEM (P=0.014) [Table 3].

Maternal depression showed a marginal association with higher anxiety levels in boys with eczema (n=42, r=0.311; P=0.045) [Table 3]. However, considering the small number of pairs, no clinical or statistical inferences should be made regarding sex differences in mental health among children and parents. Additionally, there were no statistically significant associations between the mental health of children and parents concerning depression, anxiety, and stress levels in the DASS-21 (Table 3). Regression analysis showed that the child’s HRQOL and parental age mostly explained variation in parental anxiety and stress, whereas parental education level explained variation in parental depression (Table 4). Younger parents had higher risk of developing more anxiety and stress compared with the older parents. Depression level of parents with primary to secondary education was 58% higher than their counterparts with post-secondary education or above. Conversely, the child’s eczema severity and HRQOL mostly explained the child’s emotional distress. Eczema severity and parental emotional distress significantly affected HRQOL in children of all ages (Table 4).

There was no psychological or physiological discomfort resulted from administration of the surveys.

Psychological symptoms of depression, anxiety, and stress were prevalent among children with AE and their parents. Our findings indicate associations between the mental health of children and parents and the eczema severity in those children. Increased eczema severity in children and adolescents led to greater emotional distress in parents and children, regardless of sex. Similarly, psychological symptoms in children and their parents were negatively correlated with the child’s eczema severity (NESS and POEM) and HRQOL impairment (CDLQI or IDQOL), regardless of sex. These strong correlations suggest that psychological symptoms, eczema severity, and impact on QOL have mutually detrimental effects. The DASS depression, anxiety, and stress scores were generally higher among mothers than among fathers, suggesting that mothers (the primary caregivers for children with eczema) were more strongly affected. The present study showed that eczema severity can adversely affect emotions and QOL among parents and children, highlighting the need for further exploration of biopsychosocial interactions among children and adolescents with eczema. Children with severe disease reportedly have more problems with depression and internalising behaviour.26 Behavioural issues can lead to adverse social interactions with peers, further reducing self-esteem and HRQOL. Therefore, interactions among parental perception of the child’s disease severity, the child’s treatment adherence, the child’s social influence by peers, and the child’s school environment should be considered when clinicians make comprehensive decisions about holistic treatments.

Our results using the DASS-21 are consistent with findings in previous studies1 27 that used the more comprehensive DASS-42. As in previous studies,1 27 we found that caregivers were especially likely to experience anxiety related to care provision in the home.28 29 In the present study, maternal depression was associated with a higher anxiety level, particularly in relation to boys with eczema. Accordingly, the Harmonising Outcome Measures for Eczema initiative recommends documentation of disease severity and QOL impairment in eczema cases.25 30 However, there have been few international initiatives and clinical trials regarding the psychological symptoms of caregivers and patients, particularly in the context of childhood eczema. Therefore, we suggest that clinicians should consider these important measurable domains in terms of therapeutic interventions and psychological support. Childhood eczema treatments mainly focus on pharmacological control of physical symptoms, but they often completely neglect the psychological symptoms of affected children and their parents. A more holistic treatment approach is needed for this potentially devastating common childhood disorder. Given the increasing numbers of proposed assessment tools, we advocate a holistic and comprehensive approach for eczema management that considers children and their families. This treatment tool should use a composite score to continuously evaluate disease severity (in objective and subjective manners), QOL impairment, psychological symptoms, and miscellaneous disease surrogates in affected children and their parents.1 16 21 26

A strength of this study was that compared with the DASS-42, the DASS-21 demonstrated better performance with 50% fewer questions and a shorter completion time. Findings from the DASS-21, but not the DASS-42, were correlated with disease severity as measured by the NESS and POEM.1 These discrepancies could have arisen because the sample size in the present study (using the DASS-21) was threefold greater than the sample size in the previous DASS-42 study.1 In the present study, the DASS-21, especially in child and mother, was moderately to strongly correlated with the CDLQI, IDQOL, NESS, and POEM. Thus, the DASS-21 can effectively represent the degree of emotional distress among parents and children or adolescents with eczema. This questionnaire is available in different languages, potentially allowing it to be used for assessment of patients with other ethnicities. To our knowledge, this is the first study to use the DASS-21 to assess the mental health of parents and children with eczema in a paediatric setting. This study revealed the presence of childhood eczema-related depression, anxiety, and stress in affected children and their parents.

This study had a few limitations including its relatively small sample size, especially concerning father-child pairs. A greater proportion of mothers participated in this study, which is expected because mothers are the main caregivers for children with eczema; they typically accompany their children during medical consultations. Considering that paediatric dermatological clinics also cater adolescent patients aged ≥16 years, a few participants aged 16 to 19 years completed the CDLQI on their HRQOL; although these participants exceeded the suggested age range of ≤16 years, the overall results were not affected.

Another limitation is that the number of recruited mothers, who are normally regarded as the main child caregiver, much outweighs that of recruited fathers. In addition, compared with fathers, mothers may know their child’s health more and get anxious or depressed as the eczema severity of their child escalates over time. Thus, the difference of the role in childbearing, sample size and the understanding of child’s health may affect the findings in parental-child correlations. It should be cautious when the results regarding parental-child correlations are studied and presented. The CDLQI (n=339) is a questionnaire for children, and the IDQOL (n=89) is for infants. The different numbers of participants who completed each of these questionnaires is consistent with the CDLQI coverage of a broader age range, whereas the IDQOL is only suitable for children aged <4 years. Although maternal depression was correlated with boys with anxiety, it is important to note that statistical significance should not be used to infer that there is a sex difference between parent and child groups in terms of mental health; such an inference would constitute overgeneralisation.

Children with eczema and their parents demonstrated mental health impairment, which was correlated with disease severity. Eczema-induced anxiety, stress, and other mental health issues in affected children and their parents should be considered by healthcare professionals during comprehensive assessments for the treatment of eczema. In addition to primary eczema, possible secondary psychiatric symptoms should be monitored in children with moderate to severe eczema and their parents. Childhood eczema severity and the mental health of affected children and their parents should be simultaneously evaluated to prevent and manage secondary psychological problems.

Concept or design: KL Hon.
Acquisition of data: PH Lam.
Analysis or interpretation of data: PH Lam, P Ip.
Drafting of the manuscript: PH Lam, KL Hon.
Critical revision of the manuscript for important intellectual content: KL Hon, S Loo, MJ Koh, CHY Chan, CK Li, P Ip.

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

The authors thank all children and parents who participated in this research.

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

This research was approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: CRE.2018.401). Written informed consent was obtained from participants and parents prior to the research.

1. Hon KL, Pong NH, Poon TC, et al. Quality of life and psychosocial issues are important outcome measures in eczema treatment. J Dermatolog Treat 2015;26:83-9. Crossref

2. Leung AK, Hon KL, Robson WL. Atopic dermatitis. Adv Pediatr 2007;54:241-73. Crossref

3. Leung TN, Hon KL. Eczema therapeutics in children: what do the clinical trials say? Hong Kong Med J 2015;21:251-60. Crossref

4. Hon KL, Lam PH, Ng WG, et al. Age, sex, and disease status as determinants of skin hydration and transepidermal water loss among children with and without eczema. Hong Kong Med J 2020;26:19-26. Crossref

5. Hon KL, Wong KY, Leung TF, Chow CM, Ng PC. Comparison of skin hydration evaluation sites and correlations among skin hydration, transepidermal water loss, SCORAD index, Nottingham Eczema Severity Score, and quality of life in patients with atopic dermatitis. Am J Clin Dermatol 2008;9:45-50. Crossref

6. Holm EA, Wulf HC, Stegmann H, Jemec GB. Life quality assessment among patients with atopic eczema. Br J Dermatol 2006;154:719-25. Crossref

7. Slattery MJ, Essex MJ, Paletz EM, et al. Depression, anxiety, and dermatologic quality of life in adolescents with atopic dermatitis. J Allergy Clin Immunol 2011;128:668-71. Crossref

8. Magin PJ, Pond CD, Smith WT, Watson AB, Goode SM. Correlation and agreement of self-assessed and objective skin disease severity in a cross-sectional study of patients with acne, psoriasis, and atopic eczema. Int J Dermatol 2011;50:1486-90. Crossref

9. Kuniyoshi Y, Kikuya M, Miyashita M, et al. Severity of eczema and mental health problems in Japanese schoolchildren: the ToMMo Child Health Study. Allergol Int 2018;67:481-6. Crossref

10. Wan J, Takeshita J, Shin DB, Gelfand JM. Mental health impairment among children with atopic dermatitis: a United States population-based cross-sectional study of the 2013-2017 National Health Interview Survey. J Am Acad Dermatol 2020;82:1368-75. Crossref

11. Fishbein AB, Cheng BT, Tilley CC, et al. Sleep disturbance in school-aged children with atopic dermatitis: prevalence and severity in a cross-sectional sample. J Allergy Clin Immunol Pract 2021;9:3120-9.e3. Crossref

12. Schmitt J, Apfelbacher C, Chen CM, et al. Infant-onset eczema in relation to mental health problems at age 10 years: results from a prospective birth cohort study (German Infant Nutrition Intervention plus). J Allergy Clin Immunol 2010;125:404-10. Crossref

13. Walker C, Papadopoulos L, Hussein M. Paediatric eczema and psychosocial morbidity: how does eczema interact with parents’ illness beliefs? J Eur Acad Dermatol Venereol 2007;21:63-7. Crossref

14. Ali F, Vyas J, Finlay AY. Counting the burden: atopic dermatitis and health-related quality of life. Acta Derm Venereol 2020;100:adv00161. Crossref

15. Wong KC. Psychometric investigation into the construct of neurasthenia and its related conditions: a comparative study on Chinese in Hong Kong and Mainland China [dissertation]. Hong Kong: The Chinese University of Hong Kong; 2009.

16. Lam PH, Hon KL, Leung KK, Leong KF, Li CK, Leung TF. Self-perceived disease control in childhood eczema. J Dermatolog Treat 2022;33:1459-64. Crossref

17. Gong X, Xie XY, Xu R, Luo YJ. Psychometric properties of the Chinese versions of DASS-21 in Chinese college students [in Chinese]. Chinese J Clin Psychol 2010;18:443-6.

18. Williams HC, Burney PG, Pembroke AC, Hay RJ. The U.K. Working Party’s Diagnostic Criteria for Atopic Dermatitis. III. Independent hospital validation. Br J Dermatol 1994;131:406-16. Crossref

19. Emerson RM, Charman CR, Williams HC. The Nottingham Eczema Severity Score: preliminary refinement of the Rajka and Langeland grading. Br J Dermatol 2000;142:288-97. Crossref

20. Hon KL, Ma KC, Wong E, Leung TF, Wong Y, Fok TF. Validation of a self-administered questionnaire in Chinese in the assessment of eczema severity. Pediatr Dermatol 2003;20:465-9. Crossref

21. Hon KL, Kung JS, Tsang KY, Yu JW, Cheng NS, Leung TF. Do we need another symptom score for childhood eczema? J Dermatolog Treat 2018;29:510-4. Crossref

22. Gaunt DM, Metcalfe C, Ridd M. The Patient-Oriented Eczema Measure in young children: responsiveness and minimal clinically important difference. Allergy 2016;71:1620-5. Crossref

23. Salek MS, Jung S, Brincat-Ruffini LA, et al. Clinical experience and psychometric properties of the Children’s Dermatology Life Quality Index (CDLQI), 1995-2012. Br J Dermatol 2013;169:734-59. Crossref

24. Lewis-Jones MS, Finlay AY, Dykes PJ. The Infants’ Dermatitis Quality of Life Index. Br J Dermatol 2001;144:104-10. Crossref

25. Gerbens LA, Prinsen CA, Chalmers JR, et al. Evaluation of the measurement properties of symptom measurement instruments for atopic eczema: a systematic review. Allergy 2017;72:146-63. Crossref

26. Hon KL, Kam WY, Lam MC, Leung TF, Ng PC. CDLQI, SCORAD and NESS: are they correlated? Qual Life Res 2006;15:1551-8. Crossref

27. Duran S, Atar E. Determination of depression, anxiety and stress (DAS) levels in patients with atopic dermatitis: a case-control study. Psychol Health Med 2020;25:1153-63. Crossref

28. Shelley AJ, McDonald KA, McEvoy A, et al. Usability, satisfaction, and usefulness of an illustrated eczema action plan. J Cutan Med Surg 2018;22:577-82. Crossref

29. Rork JF, Sheehan WJ, Gaffin JM, et al. Parental response to written eczema action plans in children with eczema. Arch Dermatol 2012;148:391-2. Crossref

30. Spuls PI, Gerbens LA, Simpson E, et al. Patient-Oriented Eczema Measure (POEM), a core instrument to measure symptoms in clinical trials: a Harmonising Outcome Measures for Eczema (HOME) statement. Br J Dermatol 2017;176:979-84. Crossref

Despite high-quality care, many patients admitted to the intensive care unit (ICU) do not survive; therefore, management of the dying process is a required skill among modern healthcare professionals.1 Life-sustaining technology has advanced sufficiently that it is possible to maintain vital organ function despite the knowledge that the patient’s return to health and an acceptable quality of life is no longer feasible. In these situations, a decision to limit life-sustaining treatment (LST) has become a common clinical practice in most countries worldwide.2 3 4 5 6 In recent decades, attempts to define desirable principles for end-of-life care according to a global professional consensus have achieved considerable success.3 Nevertheless, decision-making processes for death and dying are likely to be heavily influenced by regional and cultural norms and expectations; thus, it is reasonable to expect different medical practices related to end-of-life decisions. Several local and international surveys of healthcare professionals have revealed regional differences in attitudes towards end-of-life ethical concerns, as well as substantial differences in clinical practices.7 8 9 10 11 Limited prospective observational data from international studies support the existence of regional variability in end-of-life practices.5 12 13

Hong Kong is a special administrative region of China with an overwhelmingly Chinese population; nevertheless, it maintains an independent fiscal budget and healthcare system. The Hong Kong Hospital Authority, funded by the Hong Kong SAR Government, provides >90% of hospital-based services available for the local population; although nearly all healthcare workers in the public health services exhibit Chinese ethnicity, health services are based on Western medical conventions.14 Hong Kong is considered a high-income region, and recently published patient outcomes data indicate that the Hong Kong Hospital Authority provides high-quality intensive care services.15 The juxtaposition of a Western medical system and a culturally Chinese population creates a situation where Western medical practices (driven by Western cultural and ethical values) may conflict with Chinese cultural values, particularly at the end of life when deep-rooted cultural beliefs may become more relevant. A small number of studies have explored end-of-life care practices in Hong Kong ICUs; these include a survey of ICU physicians’ ethical attitudes concerning end-of-life care8 and a prospective observational study regarding end-of-life practices at a single tertiary university hospital.16 No observational territory-wide data have been published thus far. Additionally, end-of-life practices in Europe have substantially changed in recent decades17; similar changes may have occurred in Hong Kong, although previous comparative data are sparse.16 Multiple Hong Kong ICUs participated in the recent worldwide Ethicus-2 study,13 18 with the understanding that the Hong Kong data would be accessible for secondary analysis. The aim of this study was to provide a detailed description of current end-of-life care practices in Hong Kong.

This study constituted a secondary analysis of the Ethicus-2 database, focusing on the Hong Kong data. The Ethicus-2 study was a prospective, multicentre, global observational study of end-of-life practices in 199 ICUs across 36 countries.13 17 All 15 adult ICUs in publicly funded hospitals in Hong Kong were invited to participate by the Hong Kong study coordinator, representing the Hong Kong Society of Critical Care Medicine. Eight ICUs in Hong Kong participated.

Consecutive adult patients admitted to the ICU over an individual ICU-selected 6-month period between 1 September 2015 and 30 September 2016 with LST limitation or death were included. Follow-up continued until death or 2 months from the initial decision to limit LST. End-of-life categories included withholding LST, withdrawing LST, active shortening of the dying process, failed cardiopulmonary resuscitation (CPR), and brain death. These categories were mutually exclusive; if more than one limitation was triggered in a particular case, the most stringent limitation was chosen (ie, active shortening of the dying process was considered more stringent than LST withdrawal, followed by LST withholding).

Data were collected by the senior physician, or a representative, responsible for making end-of-life decisions. De-identified patient data were entered into a secure online database. Collected data included age; sex; religion; end-of-life category; admission date, time, and diagnoses; chronic disorders; use of ventilation and vasopressors, sedatives, or analgesics; date and time of hospital and ICU admission; and date and time of death or discharge from the ICU or hospital. End-of-life process data collected included type, date, and time of LST; presence of information about patient wishes; discussions with the patient or their family; degree of concurrence between the decision and patient/family wishes; and reasons for treatment decisions.

Data quality was monitored by concurrent audit and feedback, with a quality review involving 5% of all patients.17 Categorical variables were reported as numbers and percentages within end-of-life groups. After normality assessment using the Shapiro–Wilk test, continuous variables were reported as means (standard deviations) or medians (interquartile ranges [IQRs]), as appropriate. Differences among LST withholding, LST withdrawal, and no LST limitation groups were compared using analysis of variance, the Kruskal–Wallis H test, or the Chi squared test, as appropriate. Subsequent pairwise group comparisons were performed with Bonferroni correction for multiple tests. All analyses were performed using SPSS software (Windows version 27.0; IBM Corp, Armonk [NY], United States).

This prospective observational study has been reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist for observational studies.

The eight participating ICUs were distributed across Hong Kong; at least two ICUs represented each of the New Territories, Kowloon, and Hong Kong Island. Two ICUs were located in academic university hospitals (comprising 20 and 25 acute ICU beds, respectively), and the remainder were located in medium-to-large regional hospitals (ranging from 12 to 22 acute ICU beds per unit).

Among the 4922 consecutive patients screened during the study period, 548 (11.1%) patients with LST limitation (withholding or withdrawal) or death (failed CPR or brain death) were included in the study. Life-sustaining treatment limitation occurred in 455 (83.0%) patients, including 353 (77.6%) with decisions to withhold LST and 102 (22.4%) with decisions to withdraw LST. Of the 93 patients who died without LST limitation, 80 (86.0%) had failed CPR, and 13 (14.0%) experienced brain death (Fig 1). No patients underwent shortening of the dying process.

Patient characteristics are summarised in Table 1; knowledge of patient and family/surrogate wishes, as well as the timing of end-of-life processes, are described in Table 2. Patients without LST limitation had a shorter duration of ICU stay (median: 3 days, IQR=1-6) compared with patients who had decisions to withhold (median: 4 days, IQR=2-13) or withdraw (median: 6 days, IQR=3-11) [P<0.001].

The prevalences of treatments withheld or withdrawn at the initial and final decisions to limit LST are shown in Figure 2. Higher percentages of patients had endotracheal tube (P=0.009), renal replacement therapy (P<0.001), and sedation/analgesia (P=0.002) withheld at the final decision, compared with the initial decision. Similarly, higher percentages of patients had endotracheal tube, mechanical ventilation, vasopressor, and renal replacement therapy withdrawn at the final decision (all P<0.001), compared with the initial decision.

Information about decision-making practices for patients with LST limitation is provided in Table 3. In the majority of cases, the ICU physician was involved in key aspects of end-of-life decision-making and implementation. The responsible ICU physicians’ explanations of the reasons and considerations for supporting end-of-life decisions are provided in Table 4. The primary clinical reason for limiting LST was unresponsiveness to maximal therapy; the patient’s best interest, perceived good medical practice, and autonomy were key decision-making considerations.

This is the first large, multicentre, prospective, observational study of end-of-life care practices in Hong Kong ICUs. Our main findings were that LST limitation preceded >80% of patient deaths, and that death occurred in the vast majority of patients with LST limitation; only 4% of patients with LST limitation were alive at 2 months. Only 15% of ICU patients died after failed CPR (ie, without any LST limitations). Advance directives were rarely available, and no cases of active shortening of the dying process (euthanasia) were reported. Life-sustaining treatment limitation occurred in the majority (83.0%) of patients, predominantly via withholding (77.6%); withdrawal was less common (22.4%) [Fig 1]. High rates of LST limitation, such as those observed in this study, are generally presumed to reflect good end-of-life practices and have been associated with the presence of written end-of-life guidelines,19 such as those provided by the Hong Kong Hospital Authority.20

Regarding treatments that were withdrawn or withheld, the withholding of CPR universally accompanied all limitation decisions. Nutrition, hydration, and sedation were rarely withheld or withdrawn at any time, consistent with guidance from professional bodies in Hong Kong that additional safeguards are necessary when considering these actions.20 At the time of the initial limitation decision, there was relatively frequent withholding of vasopressors and renal replacement therapy; withholding or withdrawal of endotracheal tubes was less common. Although the patterns of LST limitation were similar between the initial and final decisions, such that withholding remained more prevalent than withdrawal, a substantial increase was observed in the prevalence of LST withdrawal at the time of the final decision. This finding may reflect the common Chinese cultural perspective that LST withholding and withdrawal are not ethically equivalent, with a documented preference for withholding over withdrawal as an end-of-life care strategy.8 11 The increase in withdrawal prevalence at the time of the final decision across key treatment categories (eg, vasopressors, mechanical ventilation, and renal replacement therapy) suggests that, with increasing prognostic certainty and clear progression towards death, LST withdrawal becomes more acceptable. There also appeared to be a greater reluctance to adopt withdrawal strategies early in the ICU stay, evidenced by the longer interval between ICU admission and initial limitation, if the initial limitation was withdrawal. This tendency may also reflect the need for greater prognostic certainty prior to the implementation of a withdrawal strategy. Comparisons with international data indicate that although the high rate of LST limitation prior to death is similar to practices in other countries, the early and more frequent use of withholding (rather than withdrawal) remains distinct from practices reported in North America, North and Central Europe, and Asia.13

Comparative historical data from Hong Kong are limited. A single-centre observational study conducted between 1997 and 1999 showed that LST limitation occurred in 59% of patients,16 although its LST limitation categories are not fully aligned with those of the current study. Notably, the mean interval between ICU admission and LST limitation in this previous study was nearly 8 days,16 whereas the median interval in the current study was 1.8 days (IQR=0.5-7); this difference suggests that recognition of the need for LST limitation is occurring much earlier in Hong Kong, consistent with a pattern observed in Europe during the same period.17 21 When LST limitation is indicated, earlier intervention leads to a shorter duration of patient discomfort; the observed reduction in time to limitation may represent a meaningful practice improvement over time.

Despite similar rates of LST withholding/withdrawal, the low rate of survival after LST limitation in Hong Kong (3%-4%)—comparable to the findings in a previous pan-European study12—contrasts with current European ICU outcomes, where the combined survival rate after LST withdrawal or withholding was 20%.17 This difference may possibly be attributed to implementing LST in patients at the very end-of-life when prognostic certainty is greater. The earlier implementation of end-of-life interventions may represent an area for further exploration to improve end-of-life ICU practices and minimise suffering.

Key practices in end-of-life decision-making included the initiation of discussions to limit LST by ICU physicians in the vast majority of cases; when such discussions began, ICU physicians were always involved in end-of-life decision-making processes. Notably, shared decision-making between ICU physicians and families was the predominant model reported. These findings align with the best practices described in recent international expert consensus documents.1 3 Despite frequent use of the shared decision-making model, direct or indirect knowledge of the patient’s wishes regarding LST was available for fewer than half of patients (40.3%); in the vast majority of cases (94.6%), this information was transmitted by relatives rather than by the patient themselves. Only 33 (6.0%) patients had decision-making capacity during the decision-making process, and only two (0.4%) patients had advance directives (Table 2). These results highlight the need to encourage patients to discuss their wishes regarding future end-of-life care preferences with relatives, or communicate such wishes through the use of advance directives, ensuring that patients receive the preferred level of care at this critical time. Nevertheless, levels of agreement among all parties regarding end-of-life decisions were high, and delays in decision-making due to disagreement were uncommon.

Advance directives in Hong Kong ICUs were rarely available, possibly due to selection bias; individuals with advanced disease and a greater likelihood of advance directives may have lower ICU admission priority. However, the current rate of advance directive use in North American ICUs at the end of life is nearly 50%.18 A relatively recent population-based study demonstrated very low public awareness of advance directives in Hong Kong, such that 86% of participants reported no previous knowledge of the advance directive concept.22 However, once informed of this concept, the majority of participants indicated a willingness to consider using such directives. The legislative process to formalise advance directive use in Hong Kong has substantially progressed, and there is a recognised need for public education and healthcare professional–specific guidance to promote the use of these directives.23 24

In the present study, the most common diagnostic categories at ICU admission were respiratory (43.4%) and sepsis-related (38.0%) [Table 1], similar to reported findings in most other regions worldwide.18 There were no substantial age or sex differences regarding LST limitation, but there were distinct differences in ICU admission diagnoses, such that limitation was less likely in patients with cardiovascular conditions and more likely in patients with sepsis or gastrointestinal disease. The vast majority of patients exhibited at least one co-morbidity, again similar to recently reported findings in other regions.18 Intriguingly, no patients with cancer were among those who died without LST limitation.

The primary clinical reasons for initiating LST limitation included unresponsiveness to maximal therapy, multiorgan failure, and neurologic failure; in few cases, the limitation arose from a family request or mainly in relation to quality of life (Table 4). Overwhelmingly, the primary consideration for decision-making was the patient’s best interest, followed by the principle of good medical practice, defined as the recognition that continued maximal therapy would not be beneficial for the patient (Table 4). These observations closely match the responses recently provided by a group of international experts who were asked to rank the triggers they would likely use in clinical practice to initiate discussions about LST limitation.1

Two questions related to decision-making and patient treatment wishes revealed an interesting observation. Across all end-of-life categories, approximately 70% of physicians in charge of end-of-life decision-making reported that if the patient’s wishes were known, they were followed. In contrast, when a surrogate’s treatment wishes were known, they were followed in nearly every case (Table 2). These responses indicate that the family’s treatment preferences are respected more frequently compared with known patient preferences, in contrast to guidelines from the Medical Council of Hong Kong25 and the Hospital Authority.20 Both guidelines clearly state that treatment preferences should be sought via communication with patients and family when possible, and a consensus should be reached; however, when conflicting views cannot be reconciled, the patient’s treatment preferences should supersede the family’s preferences.20 25 It is possible that physicians prioritised the family’s preferences because few patients were capable of direct communication; there was low certainty regarding perceived patient wishes when communicated through third parties. Nevertheless, this finding warrants further investigation and reflection among Hong Kong ICU healthcare professionals.

Most communication related to end-of-life decision-making occurred between ICU physicians and family/surrogates; nurses, primary physicians, and consulting physicians were involved in fewer than half of the reported cases. It has been suggested that this relatively low percentage of nurse involvement is an underestimate because most data were reported by physicians who may be unaware of nurse involvement.26 Decision agreement between healthcare staff and family members, as well as among family members, was reportedly very high (>95%) [Table 3]. Disagreements between family and staff were rare, as were delays in implementing end-of-life care because of disagreement, indicating a high level of acceptance of the decision-making process by the public and healthcare professionals in Hong Kong.

This study’s strengths included its involvement of a large number of patients over a 6-month period, provision of detailed follow-up data for up to 2 months, prospective design, and representation of most public ICUs in Hong Kong. Moreover, the data were provided by the physician in charge of end-of-life decision-making, with support from clear definitions and uniform collection across ICUs; they were also subjected to external quality control measures, minimising measurement bias. The main limitations were the lack of random ICU allocation and inclusion of consenting ICUs only, which may have introduced selection bias.

Data from the majority of Hong Kong ICUs, spanning the entire territory and representing both academic and non-academic ICUs, revealed that LST limitation occurs in most patients prior to death in ICU. Practices generally align with recommendations from local professional bodies and key international consensus documents. Although decision-making is usually initiated by ICU physicians, shared decision-making between medical staff and family/surrogates is the predominant model. Because a loss of decision-making capacity is common in the ICU, patients should be encouraged to communicate their wishes regarding end-of-life care through dialogue with relatives or more formal methods. Certain practices and outcomes observed in Hong Kong are more similar to those reported in North America and Europe than to patterns in other parts of Asia.

Concept or design: CL Sprung, A Avidan, GM Joynt.
Acquisition of data: GM Joynt, SKH Ling, LL Chang, PNW Tsai, GKF Au, DHK So, FL Chow, PKN Lam.
Analysis or interpretation of data: A Lee, GM Joynt.
Drafting of the manuscript: GM Joynt.
Critical revision of the manuscript for important intellectual content: All authors.

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

All authors have disclosed no conflicts of interest.

No funding was obtained to conduct this research from any funding agency in the public, commercial, or not-for-profit sectors. The Walter F and Alice Gorham Foundation funded the international co-ordination of the Ethicus-2 study. The Foundation had no part in the design and conduct of the research; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

This research was approved by the relevant Research Ethics Committee for each of the participating centres, including:
(1) Tuen Mun Hospital—The New Territories West Cluster Research Ethics Committee of Hospital Authority, Hong Kong (Ref No.: NTWC/CREC/15078);
(2) Prince of Wales Hospital and North District Hospital—The Joint Chinese University of Hong Kong—New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: 2015.080);
(3) Caritas Medical Centre—The Kowloon West Cluster Research Ethics Committee of the Hospital Authority, Hong Kong [Ref No.: KW/EX-15-103(88-02)];
(4) Kwong Wah Hospital—The Kowloon West Cluster Research Ethics Committee of the Hospital Authority, Hong Kong [Ref No.: KW/EX-15-105(88-04)];
(5) Pamela Youde Nethersole Eastern Hospital—The Hong Kong East Cluster Research Ethics Committee of the Hospital Authority, Hong Kong (Ref No.: HKEC-2015-028);
(6) Princess Margaret Hospital—The Kowloon West Cluster Research Ethics Committee of the Hospital Authority, Hong Kong [Ref No.: KW/EX-15-104(88-03)]; and
(7) Queen Mary Hospital—Institutional Review Board of The University of Hong Kong / Hospital Authority Hong Kong West Cluster, Hong Kong (Ref No.: UW 15-361).

The requirement for informed patient consent was waived by the relevant Clinical Research Ethics Committees as the study was observational only, where all collected data were anonymised at source and only de-identified data were passed on to the co-ordinating centre for analysis, and risk to participants was minimal.

Gary KF Au, Department of Intensive Care, Kwong Wah Hospital, Hong Kong SAR, China
Alexander Avidan, Department of Anesthesiology, Critical Care and Pain Medicine, Hadassah Medical Organization and Faculty of Medicine, Hebrew University of Jerusalem, Jerusalem, Israel
KC Chan, Department of Intensive Care, Tuen Mun Hospital, Hong Kong SAR, China
WM Chan, Adult Intensive Care Unit, Queen Mary Hospital, Hong Kong SAR, China
LL Chang, Department of Intensive Care, Pamela Youde Nethersole Eastern Hospital, Hong Kong SAR, China
FL Chow, Department of Intensive Care, Caritas Medical Centre, Hong Kong SAR, China
Gavin Matthew Joynt, Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China
Gladys WM Kwan, Department of Intensive Care, Tuen Mun Hospital, Hong Kong SAR, China
Philip KN Lam, Department of Intensive Care, North District Hospital, Hong Kong SAR, China
Anna Lee, Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China
E Leung, Adult Intensive Care Unit, Queen Mary Hospital, Hong Kong SAR, China
Steven KH Ling, Department of Intensive Care, Tuen Mun Hospital, Hong Kong SAR, China
CH Ng, Department of Intensive Care, Kwong Wah Hospital, Hong Kong SAR, China
HP Shum, Department of Intensive Care, Pamela Youde Nethersole Eastern Hospital, Hong Kong SAR, China
Dominic HK So, Department of Intensive Care Unit, Princess Margaret Hospital/Yan Chai Hospital, Hong Kong SAR, China
Charles L Sprung, Department of Anesthesiology, Critical Care and Pain Medicine, Hadassah Medical Organization and Faculty of Medicine, Hebrew University of Jerusalem, Jerusalem, Israel
L Sy, Department of Intensive Care, North District Hospital, Hong Kong SAR, China
Polly NW Tsai, Adult Intensive Care Unit, Queen Mary Hospital, Hong Kong SAR, China
HH Tsang, Department of Intensive Care, Kwong Wah Hospital, Hong Kong SAR, China
WT Wong, Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China

1. Joynt GM, Lipman J, Hartog C, et al. The Durban World Congress Ethics Round Table IV: health care professional end-of-life decision making. J Crit Care 2015;30:224-30. Crossref

2. Vincent JL. Forgoing life support in western European intensive care units: the results of an ethical questionnaire. Crit Care Med 1999;27:1626-33. Crossref

3. Sprung CL, Truog RD, Curtis JR, et al. Seeking worldwide professional consensus on the principles of end-of-life care for the critically ill. The Consensus for Worldwide End-of-Life Practice for Patients in Intensive Care Units (WELPICUS) study. Am J Respir Crit Care Med 2014;190:855-66. Crossref

4. Phua J, Joynt GM, Nishimura M, et al. Withholding and withdrawal of life-sustaining treatments in low-middle-income versus high-income Asian countries and regions. Intensive Care Med 2016;42:1118-27. Crossref

5. Lobo SM, De Simoni FH, Jakob SM, et al. Decision-making on withholding or withdrawing life support in the ICU: a worldwide perspective. Chest 2017;152:321-9. Crossref

6. Wong WT, Phua J, Joynt GM. Worldwide end-of-life practice for patients in ICUs. Curr Opin Anaesthesiol 2018;31:172-8. Crossref

7. Attitudes of critical care medicine professionals concerning forgoing life-sustaining treatments. The Society of Critical Care Medicine Ethics Committee [editorial]. Crit Care Med 1992;20:320-6. Crossref

8. Yap HY, Joynt GM, Gomersall CD. Ethical attitudes of intensive care physicians in Hong Kong: questionnaire survey. Hong Kong Med J 2004;10:244-50.

9. Yaguchi A, Truog RD, Curtis JR, et al. International differences in end-of-life attitudes in the intensive care unit: results of a survey. Arch Intern Med 2005;165:1970-5. Crossref

10. Bito S, Asai A. Attitudes and behaviors of Japanese physicians concerning withholding and withdrawal of life-sustaining treatment for end-of-life patients: results from an Internet survey. BMC Med Ethics 2007;8:7. Crossref

11. Weng L, Joynt GM, Lee A, et al. Attitudes towards ethical problems in critical care medicine: the Chinese perspective. Intensive Care Med 2011;37:655-64. Crossref

12. Sprung CL, Cohen SL, Sjokvist P, et al. End-of-life practices in European intensive care units: the Ethicus study. JAMA 2003;290:790-7. Crossref

13. Avidan A, Sprung CL, Schefold JC, et al. Variations in end-of-life practices in intensive care units worldwide (Ethicus-2): a prospective observational study. Lancet Respir Med 2021;9:1101-10. Crossref

14. Kong X, Yang Y, Gao J, et al. Overview of the health care system in Hong Kong and its referential significance to mainland China. J Chin Med Assoc 2015;78:569-73. Crossref

15. Ling L, Ho CM, Ng PY, et al. Characteristics and outcomes of patients admitted to adult intensive care units in Hong Kong: a population retrospective cohort study from 2008 to 2018. J Intensive Care 2021;9:2. Crossref

16. Buckley TA, Joynt GM, Tan PY, Cheng CA, Yap FH. Limitation of life support: frequency and practice in a Hong Kong intensive care unit. Crit Care Med 2004;32:415-20. Crossref

17. Sprung CL, Ricou B, Hartog CS, et al. Changes in end-of-life practices in European intensive care units from 1999 to 2016. JAMA 2019;322:1692-704. Crossref

18. Feldman C, Sprung CL, Mentzelopoulos SD, et al. Global comparison of communication of end-of-life decisions in the ICU. Chest 2022;162:1074-85. Crossref

19. Mentzelopoulos SD, Chen S, Nates JL, et al. Derivation and performance of an end-of-life practice score aimed at interpreting worldwide treatment-limiting decisions in the critically ill. Crit Care 2022;26:106. Crossref

20. Hospital Authority, Hong Kong SAR Government. Working Group on review of Hospital Authority Clinical Ethics Committee (HKCEC) guidelines related to EOL decisionmaking. HA Guidelines on Life-Sustaining Treatment in the Terminally Ill. 2020. Available from: https://www.ha.org.hk/haho/ho/psrm/LSTEng.pdf. Accessed 4 Jul 2023.

21. Lesieur O, Herbland A, Cabasson S, Hoppe MA, Guillaume F, Leloup M. Changes in limitations of life-sustaining treatments over time in a French intensive care unit: a prospective observational study. J Crit Care 2018;47:21-9. Crossref

22. Chung RY, Wong EL, Kiang N, et al. Knowledge, attitudes, and preferences of advance decisions, end-of-life care, and place of care and death in Hong Kong. A population-based telephone survey of 1067 adults. J Am Med Dir Assoc 2017;18:367.e19-27. Crossref

23. Food and Health Bureau, Hong Kong SAR Government. End-of-Life Care: Legislative Proposals on Advance Directives and Dying in Place Consultation Report. July 2020. Available from: https://www.healthbureau.gov.hk/download/press_and_publications/consultation/190900_eolcare/e_EOL_consultation_report.pdf. Accessed 4 Jul 2023.

24. Fong BY, Yee HH, Ng TK. Advance directives in Hong Kong: moving forward to legislation. Ann Palliat Med 2022;11:2622-30. Crossref

25. Medical Council of Hong Kong. Code of professional conduct: for the guidance of registered medical practitioners. Revised October 2022. Available from: https://www.mchk.org.hk/english/code/files/Code_of_Professional_Conduct_(English_Version)_(Revised_in_October_2022).pdf. Accessed 4 Jul 2023.

26. Benbenishty J, Ganz FD, Anstey MH, et al. Changes in intensive care unit nurse involvement in end of life decision making between 1999 and 2016: descriptive comparative study. Intensive Crit Care Nurs 2022;68:103138. Crossref

The advent of highly active antiretroviral therapy (HAART) in 1996 led to a substantial decrease in the incidence of opportunistic infections among people living with human immunodeficiency virus (HIV) in many regions.1 2 3 Nonetheless, HIV-associated tuberculosis (TB) remains an important global health challenge. The World Health Organization estimated that 10.6 million people were living with TB worldwide in 2021, 6.7% of whom were living with HIV.4 In the same year, 1.6 million people died of TB, including 187 000 people who were living with HIV.4 The burden of HIV-associated TB considerably varies across countries and regions.5 6 The prevalence of HIV-associated TB in individual areas reportedly changes over time.7 8 9 Awareness of changes in the epidemiology and clinical manifestations of patients with HIV-associated TB can help policymakers formulate timely relevant prevention and control measures. It can also help improve treatment outcomes for patients with HIV-associated TB.

In Hong Kong, the TB case notification rate has exhibited an overall decreasing trend over the past few decades.10 In 2021, the provisional number of TB cases reported to the Department of Health was 3741.11 The corresponding TB notification rate was 50.5 per 100 000 inhabitants, a substantial decrease from 84.1 per 100 000 inhabitants in 2006.11 The overall prevalence of HIV infection in the general adult population has remained low (<0.1%).12 The epidemiology and clinical manifestations of HIV-associated TB during the period 1996 to 2006 in Hong Kong have been reported.13 Notably, the report showed that TB had become an increasingly important acquired immunodeficiency syndrome (AIDS)–defining illness in Hong Kong, surpassing Pneumocystis jirovecii pneumonia as the most common primary AIDS-defining illness in 2005; the two illnesses represented 39% and 31% of all such illnesses, respectively, in 2005.13 The presentation of HIV-associated TB is often atypical.13 Considering the declining prevalence of TB in the general population and accompanying decrease in TB transmission in Hong Kong, the implementation of strategies to enhance screening for latent TB infection, the increased use of molecular tests for TB diagnosis, and the expansion of HAART in recent years, we conducted a retrospective review of data regarding patients with HIV-associated TB that were reported to the TB-HIV Registry of the Department of Health during the period 2007 to 2020. We aimed to identify temporal trends in the epidemiology and clinical manifestations of HIV-associated TB during that period.

We retrospectively reviewed data contained within the TB-HIV Registry, which captured information regarding nearly all cases of HIV-associated TB diagnosed in the Tuberculosis and Chest Service and Special Preventive Programme (SPP) under the Department of Health, as well as cases referred from regional hospitals of the Hong Kong Hospital Authority, during the period 2007 to 2020. The details of data contained in the TB-HIV Registry, along with the criteria for TB as a primary AIDS-defining illness, were described in a previous report.13 There have been no changes in the criteria for TB as a primary AIDS-defining illness since the last report. Where necessary, both clinic records and hospital discharge records were reviewed. All data were imported into Epi Info14 and exported to statistical software SPSS (Windows version 26.0; IBM Corp, Armonk [NY], US) for analysis. The Cochran–Armitage trend test in XLSTAT software (Lumivero, Denver [CO], US) was used to identify trends in the proportion of reported AIDS cases with TB as a primary AIDS-defining illness during the study period. The Cochran–Armitage trend test was also used to examine changes in demographic features and clinical manifestations during the same period. Where relevant, we compared the demographic features of patients reported to the TB-HIV Registry during the study period with the features of a historical cohort from the period 1996 to 200613 using the Chi squared test. P values <0.05 were considered statistically significant.

This study was an extension of a previous study designed to evaluate the public health programme for HIV-associated TB in Hong Kong15; it did not constitute research on human participants. Throughout the review process, we implemented all reasonable precautions to protect the confidentiality of personal data and excluded personally identifiable information from the electronic database.

All 390 cases reported to the TB-HIV Registry from 1 January 2007 to 31 December 2020 were included in this retrospective analysis. Information about whether TB constituted a primary AIDS-defining illness was available for 363 of 390 (93.1%) patients, where TB was listed as a primary AIDS-defining illness in 251 of those cases (69.1%). Overall, TB as a primary AIDS-defining illness represented a decreasing trend of 18.3% of 1375 reported AIDS cases during the period 2007 to 202012 (Cochran–Armitage trend test, P<0.001) [Fig], compared with 28.2% (192/680; Chi squared test, P<0.001) among the historical cohort of patients reported during the period 1996 to 2006.13

Trends in demographic features, including age, sex, case category, ethnicity, residence, primary source of care (first presentation), and smoking status, for the 390 HIV-associated TB cases reported to the TB-HIV Registry during the study period are shown in Table 1. The proportion of female patients significantly increased whereas that of ever-smokers significantly decreased (Cochran–Armitage trend test, P=0.035 and P=0.029, respectively). No significant trends were detected in other variables examined. Additionally, the proportions of Chinese individuals and permanent residents were lower in the present cohort than in the historical cohort of 1996 to 200613 (260/390, 66.7% vs 152/190, 80.0%; Chi squared test, P=0.001 and 258/390, 66.2% vs 144/190, 75.8%; Chi squared test, P=0.018, respectively). The proportion of female patients was significantly higher in the present cohort than in the historical cohort13 (82/390, 21.0% vs 21/190, 11.1%; Chi squared test, P=0.003).

Trends in clinical manifestations, including symptoms, presence of pulmonary TB, radiographic features (for cases with abnormalities on chest radiographs), presence of extrapulmonary TB (EPTB), most common EPTB sites, sputum smear positivity status, drug susceptibility patterns, presence of TB risk factors, CD4 cell count at TB diagnosis, presence of other AIDS-defining illnesses at the time of co-infection, and antiretroviral therapy (ART) status (among patients with a diagnosis of HIV infection before TB), among the 390 TB cases are presented in Tables 2 and 3. The proportions of patients presenting with site-specific symptoms (other than chest-related symptoms) and with EPTB both significantly increased during the period 2007 to 2020 (Cochran–Armitage trend test, P=0.029 and P=0.008, respectively) [Table 2]. The most common EPTB sites were lymph nodes (42.8%), pleura (21.5%), and abdomen (13.8%). Among patients who underwent sputum smear tests, the proportion of patients with sputum smear positivity significantly decreased (Cochran–Armitage trend test, P=0.006) [Table 2]. Among patients with lung parenchymal lesions on chest radiographs, a decreasing trend was observed in the proportion of patients in which the lower zone was the predominant lesion site (Cochran–Armitage trend test, P=0.045) [Table 3]. Among patients with a diagnosis of HIV infection before TB, the proportion of patients receiving ART at TB diagnosis significantly increased (Cochran–Armitage trend test, P=0.003) [Table 2].

This study revealed a decreasing trend in the proportion of reported AIDS cases with TB as a primary AIDS-defining illness during the period 2007 to 2020. The overall proportion (18.3%) was also lower than the proportion (28.2%) in the historical cohort of cases reported during the period 1996 to 2006.13 The proportions of Chinese individuals and permanent residents were lower, whereas the proportion of female patients was higher, in our cohort compared with the historical cohort.13 The proportions of female patients and patients with extrapulmonary involvement significantly increased, whereas the proportions of ever-smokers and the proportion with sputum smear positivity among pulmonary TB cases significantly decreased during the period 2007 to 2020. A decreasing trend was observed in the proportion of patients with pulmonary TB in which the lower zone was the predominant site of lung parenchymal lesions. Among patients with a diagnosis of HIV infection before TB, an increasing trend was observed in the proportion of patients receiving ART.

In Hong Kong, TB is considered an AIDS-defining illness when the disease is extrapulmonary. Pulmonary TB and cervical lymph node TB are considered AIDS-defining illnesses only when the CD4 cell count at the time of TB diagnosis is <200/μL, as recommended by the Scientific Committee of the Advisory Council on AIDS in 1994.16 Since then, there have been no changes in the criteria for TB as an AIDS-defining illness among individuals infected with HIV. The decreasing trend in Hong Kong regarding the proportion of reported AIDS cases with TB as a primary AIDS-defining illness might be due to decreased community transmission of TB through better TB control, the expansion of HAART since its introduction in 1997, and (perhaps to a lesser extent) increased acceptance of testing for latent TB infection and preventive treatment for TB among HIV-infected individuals since the early 2000s. Similar decreasing trends related to HAART-induced improvements in immune status among HIV-infected individuals have also been identified during studies conducted in some other countries.7 9 17 In an observational, retrospective study of AIDS cases included in the Barcelona AIDS register between 1994 and 2005, decreases were observed regarding the incidence of TB as an AIDS-defining illness among both native and immigrant populations.7 Another study examining trends in the incidence of AIDS-defining opportunistic illnesses over a 25-year period in Brazil showed a reduction in TB incidence from 1991-1993 to 2009-2012.9 A prospective cohort study of participants in the HIV Outpatient Study at 12 HIV clinics within the US indicated that TB incidence decreased after HAART introduction.17 Conversely, among participants in the HOMER cohort study (HAART Observational Medical Evaluation and Research) conducted during the period 1996 to 2007 in Canada, no statistically significant trends were observed in the proportion of cases with TB as the AIDS-defining illness, probably because the small number of TB cases reported in each time period limited the ability to detect significant changes in the reported cases.18 Another study examining AIDS notification data in Australia during the period 1993 to 2000 revealed that the proportion of AIDS cases with TB as the AIDS-defining illness was higher during 1996 to 2000 (post-HAART era) than during 1993 to 1995 (pre-HAART era); the authors attributed this difference to the increasing proportion of Australian patients with AIDS who had been born in sub-Saharan Africa and Asia during the 1990s, among whom the risk of TB was considerably higher.19 Further studies examining trends in TB as an AIDS-defining illness, as well as location-specific factors that may influence such trends in the post-HAART era, are warranted to facilitate control strategies for HIV-associated TB.

Further attention is needed regarding the observation of higher proportions of non-Chinese individuals (mostly Asians and Africans, who have much higher TB incidence than the rate among Hong Kong Chinese individuals) and non-permanent residents of Hong Kong in the 2007-2020 cohort compared with the historical 1996-2006 cohort. Similar findings of higher incidences of AIDS-associated TB in foreign-born populations from countries with much higher TB incidence compared with the native population have been reported on the basis of some studies conducted in developed countries.7 20 21 These observations highlight the need for TB screening and prophylaxis for people living with HIV who were born in countries with a high background prevalence of TB.

Intriguingly, we observed an increasing trend in the proportion of women during the period 2007 to 2020. The reason for this increase is unclear; it may be related to changes in the societal roles of men and women that influence exposure risk. The role of an increased proportion of EPTB (reportedly associated with female sex and observed throughout our cohort, as discussed below) requires further investigation.

The proportion of ever-smokers significantly decreased during the period 2007 to 2020, consistent with the findings of some studies conducted in the US.22 23 In an analysis of patients from a HIV surveillance system in the US, the prevalence of current smoking declined from 37.6% in 2009 to 33.6% in 2014.22 In another prospective cohort study that examined smoking trends among HIV-positive patients in the US, a decline in the annual prevalence of current smoking from 1984 to 2012 was also reported; however, disparities were noted according to race, ethnicity, and education.23 Nonetheless, because smoking increases HIV-related and non–HIV-related morbidity and mortality among people living with HIV, smoking cessation interventions remain an essential component of routine care for such individuals.

The predominance of the lower zone as the site of lung parenchymal lesions on chest radiographs is a relatively common feature among patients with HIV-associated pulmonary TB, according to our previous report on the 1996 to 2006 cohort13 and some other reports.24 25 The present study showed that the lower zone was less frequently the predominant site of lung parenchymal lesions during the period 2007 to 2020. The proportion of patients in the 2007 to 2020 cohort with the lower zone as the predominant site (16.7%) was also lower compared with the proportion of patients in the historical 1996 to 2006 cohort (32.4%).13 This difference may be related to the higher CD4 cell count at TB diagnosis among patients in the current cohort compared with patients in the historical cohort (median CD4 cell counts at TB diagnosis: 100/μL and 78/μL, respectively). Nonetheless, lower zone involvement was present in approximately one-sixth of pulmonary TB cases reported during 2007 and 2020. A high index of suspicion is required for the accurate and timely diagnosis of pulmonary TB in people living with HIV.

A decreasing trend was observed in the proportion of patients with sputum smear positivity during the period 2007 to 2020. The overall proportion of patients with sputum smear positivity in the 2007 to 2020 cohort (36.6%) was also lower than the proportion in the historical cohort of 1996 to 2006 (42.2%)13; it was similar to the proportion identified during a population database study in South Korea (36.4%).26 These results suggest that TB cases have been diagnosed at increasingly earlier stages due to enhanced active TB screening efforts among people living with HIV, as well as the increased use of molecular testing that enhanced the diagnosis of smear-negative cases in Hong Kong. These results highlight the need for continued TB screening efforts and early detection of TB among people living with HIV.

Our findings indicate that EPTB became more common among HIV-associated TB patients during the period 2007 to 2020. The overall proportion of patients with EPTB was also higher in the present cohort compared to that reported in a local study that examined risk factors for mortality in an earlier cohort (2006 to 2015) and that in the historical 1996 to 2006 cohort (71.0%, 64.9%,15 and 62.6%,13 respectively). These differences may have arisen through enhanced diagnosis of EPTB with the increased use of molecular testing in Hong Kong. An increasing trend of extrapulmonary involvement among TB patients has also been reported in some other studies, although such studies are mostly population-based.27 28 29 Few reports have been published regarding temporal trends in EPTB specifically among HIV-associated TB patients.30 Further studies are needed to examine these trends and associated factors.

Strengths of this study included its use of cohort data from the TB-HIV Registry covering a relatively long period (14 years) to study temporal changes in the epidemiology and clinical manifestations of HIV-associated TB. However, some limitations should be considered when interpreting the results of this study. First, the TB-HIV Registry may not capture all HIV-associated TB cases—some patients encountered in the SPP were not referred to a chest clinic but underwent anti-TB treatment in private clinics or other countries. The total number of HIV-associated TB cases in the HIV Surveillance Report of SPP was approximately 10% higher than the number in the TB-HIV Registry; the difference mostly comprised non-permanent residents temporarily staying in Hong Kong. Nonetheless, data from the HIV Surveillance Report showed a similar decreasing trend in TB as a primary AIDS-defining illness during the study period (data not shown). Second, this study utilised a retrospective design, and data present in the database of the TB-HIV Registry may be incomplete. Information regarding some parameters such as case category, co-morbidities, and CD4 cell count was unavailable for some patients. To overcome this limitation, we traced and reviewed relevant clinical records from chest clinics and hospitals when necessary. Therefore, we expect minimal bias due to missing data. Finally, the sample size may have led to insufficient statistical power for detecting temporal changes in some less common parameters.

This study showed that TB has become less important as a primary AIDS-defining illness in Hong Kong over the 14 years of the study period. Nonetheless, it remains the second most common primary AIDS-defining illness after P jirovecii pneumonia. Important temporal changes were also observed in the patterns of demographic features and clinical manifestations. Continued surveillance regarding the patterns of demographic features and clinical manifestations is needed to inform policymakers during the formulation of TB control strategies to improve patient care and treatment outcomes among people living with HIV. This surveillance is especially important in situations such as the coronavirus disease 2019 era, during which resources from TB programmes may be diverted to management of the global pandemic.

Concept or design: ACK Chan, SS Huang.
Acquisition of data: ACK Chan, SS Huang.
Analysis or interpretation of data: ACK Chan, SS Huang.
Drafting of the manuscript: ACK Chan, SS Huang.
Critical revision of the manuscript for important intellectual content: KH Wong, CC Leung, MP Lee, TY Tsang, WS Law, LB Tai.

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 all of their colleagues in the Tuberculosis and Chest Service and the Special Preventive Programme of the Department of Health who provided assistance and support to make this paper possible. The authors also thank Ms Ida KY Mak, Research Officer at the Tuberculosis and Chest Service of the Department of Health, for her dedicated efforts in maintaining the Tuberculosis–Human Immunodeficiency Virus Registry and assisting with the analysis.

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

This research was a retrospective analysis of observational data routinely collected in a local Registry as part of the ongoing evaluation of the public health programme for human immunodeficiency virus–associated tuberculosis in Hong Kong. Approval for the evaluation and exemption from obtaining informed patient consent has been granted by the Ethics Committee of the Department of Health of the Hong Kong SAR Government (Ref No.: L/M 416/2017).

1. Low A, Gavriilidis G, Larke N, et al. Incidence of opportunistic infections and the impact of antiretroviral therapy among HIV-infected adults in low- and middleincome countries: a systematic review and meta-analysis. Clin Infect Dis 2016;62:1595-603. Crossref

2. B-Lajoie MR, Drouin O, Bartlett G, et al. Incidence and prevalence of opportunistic and other infections and the impact of antiretroviral therapy among HIV-infected children in low- and middle-income countries: a systematic review and meta-analysis. Clin Infect Dis 2016;62:1586-94. Crossref

3. Rubaihayo J, Tumwesigye NM, Konde-Lule J. Trends in prevalence of selected opportunistic infections associated with HIV/AIDS in Uganda. BMC Infect Dis 2015;15:187. Crossref

4. World Health Organization. Global Tuberculosis Report 2022. Geneva: World Health Organization; 2022. Available from: https://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2022. Accessed 4 Jul 2024.

5. World Health Organization. WHO releases new global lists of high-burden countries for TB, HIV-associated TB and drug-resistant TB. 2021. Available from: https://www.who.int/news/item/17-06-2021-who-releases-new-global-lists-of-high-burden-countries-for-tb-hiv-associated-tb-and-drug-resistant-tb. Accessed 4 Jul 2024.

6. Benito N, Moreno A, Miro JM, Torres A. Pulmonary infections in HIV-infected patients: an update in the 21st century. Eur Respir J 2012;39:730-45. Crossref

7. Martin V, García de Olalla P, Orcau A, Caylà JA. Factors associated with tuberculosis as an AIDS-defining disease in an immigration setting. J Epidemiol 2011;21:108-13. Crossref

8. González-García A, Fortún J, Elorza Navas E, et al. The changing epidemiology of tuberculosis in a Spanish tertiary hospital (1995-2013). Medicine (Baltimore) 2017;96:e7219. Crossref

9. Coelho L, Cardoso SW, Amancio RT, et al. Trends in AIDS-defining opportunistic illnesses incidence over 25 years in Rio de Janeiro, Brazil. PLoS One 2014;9:e98666. Crossref

10. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Annual Report. Tuberculosis and Chest Service. 2020. Available from: https://www.info.gov.hk/tb_chest/doc/Annual_Report_2020.pdf. Accessed 16 Nov 2022.

11. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Notification & death rate of tuberculosis (all forms), 1947-2022. Available from: https://www.chp.gov.hk/en/statistics/data/10/26/43/88.html. Accessed 17 Nov 2022.

12. Centre for Health Protection, Department of Health, Hong Kong SAR Government. HIV Surveillance Report—2020 Update. 2021. Available from: https://www.chp.gov.hk/ files/pdf/aids20.pdf. Accessed 16 Nov 2022.

13. Chan CK, Alvarez Bognar F, Wong KH, et al. The epidemiology and clinical manifestations of human immunodeficiency virus–associated tuberculosis in Hong Kong. Hong Kong Med J 2010;16:192-8.

14. Centers for Disease Control and Prevention, United States Government. Epi Info. Available from: https://www.cdc.gov/epiinfo/index.html. Accessed 4 Jul 2024.

15. Chan CK, Wong KH, Lee MP, et al. Risk factors associated with 1-year mortality among patients with HIV-associated tuberculosis in areas with intermediate tuberculosis burden and low HIV prevalence. Hong Kong Med J 2018;24:473-83. Crossref

16. Scientific Committee of the Advisory Council on AIDS Hong Kong. Classification system for HIV infection and surveillance case definition for AIDS in adolescents and adults in Hong Kong. 1995. Available from: https://www.aids.gov.hk/pdf/g40.pdf. Accessed 16 Nov 2022.

17. Buchacz K, Baker RK, Palella FJ Jr, et al. AIDS-defining opportunistic illnesses in US patients, 1994-2007: a cohort study. AIDS 2010;24:1549-59. Crossref

18. Jafari S, Chan K, Aboulhosn K, et al. Trends in reported AIDS defining illnesses (ADIs) among participants in a universal antiretroviral therapy program: an observational study. AIDS Res Ther 2011;8:31. Crossref

19. Dore GJ, Li Y, McDonald A, Ree H, Kaldor JM; National HIV Surveillance Committee. Impact of highly active antiretroviral therapy on individual AIDS-defining illness incidence and survival in Australia. J Acquir Immune Defic Syndr 2002;29:388-95. Crossref

20. Camoni L, Regine V, Boros S, Salfa MC, Raimondo M, Suligoi B. AIDS patients with tuberculosis: characteristics and trend of cases reported to the National AIDS Registry in Italy—1993-2010. Eur J Public Health 2012;23:658-63. Crossref

21. Dore GJ, Li Y, McDonald A, Kaldor JM. Spectrum of AIDS-defining illnesses in Australia, 1992 to 1998: influence of country/region of birth. J Acquir Immune Defic Syndr 2001;26:283-90.

22. Frazier EL, Sutton MY, Brooks JT, Shouse RL, Weiser J. Trends in cigarette smoking among adults with HIV compared with the general adult population, United States—2009-2014. Prev Med 2018;111:231-4. Crossref

23. Akhtar-Khaleel WZ, Cook RL, Shoptaw S, et al. Trends and predictors of cigarette smoking among HIV seropositive and seronegative men: the multicenter AIDS cohort study. AIDS Behav 2016;20:622-32. Crossref

24. Affusim C, Abah V, Kesieme EB, et al. The effect of low CD4+ lymphocyte count on the radiographic patterns of HIV patients with pulmonary tuberculosis among Nigerians. Tuberc Res Treat 2013;2013:535769. Crossref

25. Chamie G, Luetkemeyer A, Walusimbi-Nanteza M, et al. Significant variation in presentation of pulmonary tuberculosis across a high resolution of CD4 strata. Int J Tuberc Lung Dis 2010;14:1295-302.

26. Kim JM, Kim NJ, Choi JY, Chin BS. History of acquired immune deficiency syndrome in Korea. Infect Chemother 2020;52:234-44. Crossref

27. Ben Ayed H, Koubaa M, Marrakchi C, et al. Extrapulmonary tuberculosis: update on the epidemiology, risk factors and prevention strategies. Int J Trop Dis 2018;1:006.

28. Peto HM, Pratt RH, Harrington TA, LoBue PA, Armstrong LR. Epidemiology of extrapulmonary tuberculosis in the United States, 1993-2006. Clin Infect Dis 2009;49:1350-7. Crossref

29. Sandgren A, Hollo V, van der Werf MJ. Extrapulmonary tuberculosis in the European Union and European Economic Area, 2002 to 2011. Euro Surveill 2013;18:20431. Crossref

30. Mohammed H, Assefa N, Mengistie B. Prevalence of extrapulmonary tuberculosis among people living with HIV/AIDS in sub-Saharan Africa: a systemic review and meta-analysis. HIV AIDS (Auckl) 2018;10:225-37. Crossref