Introduction
Hepatitis B virus (HBV) infection carries the risk of chronicity, with threats of progression to cirrhosis, hepatocellular carcinoma (HCC) and associated mortality [Reference Zamor, deLemos and Russo1]. As of 2019, some 300 million people worldwide were living with chronic HBV infections [2]. In 2016, the World Health Organization (WHO) called for the elimination of viral hepatitis [3], the achievability of which hinges on the prevention of new infections and the reduction of chronic diseases. Prevention of HBV infections can be achieved largely through the implementation of universal vaccination, while prompt diagnosis and antiviral treatment of chronic HBV infection is needed in parallel to achieve elimination in the population [4]. Globally, the population burden of HBV infections varies geographically, as reflected by the seroprevalence of hepatitis B surface antigen (HBsAg) [Reference Hou, Liu and Gu5]. The HBsAg prevalence is highest at 5–10% among adult populations in sub-Saharan Africa and East Asia [3, Reference Paik6], and lower in most industrialised countries. The variability highlighted the differences in the scale of the public health challenges posed to each city or country.
Since the early 1980s, many cities/countries in highly endemic Asia had developed vaccination programmes as a key strategy for reducing HBV population burden. After prioritising higher risk populations in the initial phase, universal vaccination of all newborns rapidly became the standard in many places in the Asia Pacific region [Reference Lok7–Reference Park, Chung and Lee10]. With the universal neonatal vaccination programme in place for over 3 decades, it is anticipated that the prevalence of HBV infection, especially that of the younger adult population, should have fallen to a very low level. As for patients with chronic HBV infections, antiviral treatment has, in the past decade, become increasingly adopted in the region [Reference Shan, Cui and Jia11, Reference Liaw12]. Located in the Asia Pacific and physically linked with Mainland China, Hong Kong was one of the high endemicity jurisdictions with a high HBsAg prevalence of about 10% among healthy donors in the 1980s [Reference Paik6]. Similar to the timeline of neighbouring countries, hepatitis B vaccination was introduced in Hong Kong in 1982 [Reference Lok13], with universal vaccination provided to all newborns since 1988, a strategy subsequently strengthened by mop-up vaccination of pre- and primary school children [Reference Lee14]. (Supplementary Fig. 1) With a 3-dose schedule, vaccinations are given at birth, at 1 month and at 6 months of age. The HBV 3-dose vaccination coverage in children aged 2–5 were stable at above 99% each year from 2001 to 2018. [15]. Since 1980s, universal antenatal screening for HBsAg has been in place with above 99% coverage [Reference Poon16]. In 2020, the government launched an action plan to eliminate viral hepatitis with reference to the WHO's recommendations (https://www.hepatitis.gov.hk/english/action_plan/content.html). This study was designed to determine the current population burden of HBV infection in this previously high endemicity city of Hong Kong, following long history of universal vaccinations.
Materials and methods
Study subjects and study design
This is a territory-wide study with the recruitment of subjects from geographically random households, using a cluster sampling frame based on building groups in Hong Kong, the protocol of which has been published [Reference Poon17]. With informed consent, each household representative (index participant) was contacted for questionnaire completion and blood-taking arrangement. To compensate for the travel expenses a HKD25 voucher (USD1~HKD7.8) was given to each participant who had completed the study. Consent forms of participants aged below 18 were signed by their legal guardians. Ethical approval from the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee (CREC 2017.697) had been obtained.
The main outcome variables were HBsAg and anti-HBs test results (positive and negative). Independent variables included (i) socio-demographics – age, year of birth, gender, marital status, education level, ethnicity, place of birth, economic activity status; (ii) history of viral hepatitis testing, diagnosis and treatment, self-reported as HBV carriers and family history of infection; (iii) history of potential exposure to HBV including behavioural risk; and (iv) history of vaccination.
Laboratory testing
All blood samples were collected through venesection, except for six children and two adults who had dried blood spots (DBS) collected because of difficulty with venesection. Due to low sensitivity and specificity, DBS test results were excluded from analysis in this study. Screening of serological markers of HBV (HBsAg, anti-HBs Ab and anti-HBc Ab) in each subject and testing of HBeAg in subjects tested HBsAg positive were performed by ELISA on an automated immunoassay system (MINI Vidas®, BioMérieux). All tests were performed and interpreted in accordance with the manufacturers' recommendations.
Statistical analysis
We estimated the proportion of households testing HBsAg positive, with 95% confidence interval (CI) using binomial exact method. Birth year-gender specific prevalence of HBsAg in the population was estimated by selecting the smallest member ID of the household in the respective stratum. Sensitivity analysis was performed by selecting the youngest member in each stratum. Crude HBV prevalence was the proportion of participants with valid HBsAg test result testing positive. The prevalence was further adjusted by the distribution of population by gender and age groups from the government's Census and Statistics Department.
In the analyses, we examined factors associated with HBV exposure, vaccination coverage and chronic diseases. In evaluating the correlates of HBsAg positivity, bivariable and multivariable logistic regression models with age as a confounder were performed. To account for the association of HBsAg status of household members, a sub-analysis with age and HBsAg positivity of household members as confounders was performed after exclusion of single participant households. Among HBsAg positive participants, those who self-reported as HBV carriers at the time of survey were compared with others unaware in bivariable logistic regression models. For HBsAg negative participants, comparison was made between anti-HBs negative (undetectable immunity) and positive (detectable immunity) subjects, while among participants testing negative for both HBsAg and anti-HBs, factors associated with anti-HBc positive (inferring exposure) results were examined by performing bivariable and multivariable logistic regression models with age as a confounder. To examine the overall serologies, comparison was made between susceptible (negative for both HBsAg and anti-HBs), immune (anti-HBs positive, HBsAg negative), and chronically infected (HBsAg positive, anti-HBs negative) persons, after exclusion of cases with unclear status (HBsAg negative, anti-HBc positive, and anti-HBs negative) in multivariable multinominal logistic regression with variable of year of birth after 1990 as a confounder in SPSS25. Complete case analyses were performed.
Results
General characteristics of study population
Between September 2018 and October 2020, a total of 38 020 invitation letters were dispatched by post spanning over 2 years. We received 1497 (4%) household replies indicating consent to join the study. Ultimately, 1143 (76%) responding households completed the study, resulting in the enrolment of 2085 participants with a participants-to-household ratio of 1.82. About half (52%) of the recruited households had at least two members joining the study. Of all participants, 44% were male, with a median age of 54 (IQR = 39–63, range = 4–89 years old), almost all (99%) were ethnic Chinese and local permanent residents (98%) while 71% were born in Hong Kong (Supplementary Table S1). Comparing with population by-census data of 2016, a higher proportion of recruited participants were Chinese, female, aged 35 or above and with post-secondary education attainment (Supplementary Table S2). However, the distribution of recruited households by residential district was within 2% difference to that of residential building groups (Supplementary Table S3).
By hepatitis B serology, the different proportions of participants who were chronic carriers (HBsAg + ), immune due to vaccination (HBsAg-, anti-HBs + ) and immune from natural infection (HBsAg-, anti-HBs + , anti-HBc + ) by birth cohort are shown in Figure 1. The youngest people (born after 1990) had the lowest proportion with HBsAg + and highest proportion with presumptive susceptibility to HBV infection because of absent anti-HBs. The proportion with natural immunity increased with age, while that from vaccination peaked for the birth cohort of 1981–1990 for male and 1971–1980 for female.
Distribution of participants' HBV serology by birth cohort and gender, after the exclusion of unclear status (HBsAg-, anti-HBs-, anti-HBc+) (selection for more than 1 member per household in age-gender strata: smallest member number).
Chronic hepatitis B infection
Overall, 9.2% (95% CI 7.5%–10.9%) of the index household participants tested HBsAg positive, while 12.8% (95% CI 10.8%–14.7%) of the households had participant testing positive (Supplementary Table S3). Among households with at least two members joining the study (n = 598), 16% had one or more member who tested HBsAg positive. Geographically, the proportion of households with index participants testing HBsAg positive varied by district (a total of 18 districts) (Supplementary Fig. 2), but the difference was not statistically significant (χ 2 test, P = 0.2). Spatial statistical comparison could not be made with only 18 districts as geographic unit.
Stratified by birth year group and gender, HBsAg prevalence was the highest among participants born in 1961–1970 for both female (10.9%, 95% CI 7.1%–14.6%) and male (9.6%, 95% CI 5.3%–13.8%) (Fig. 2). A declining prevalence was observed in both gender after this birth year stratum, with the lowest prevalence in participants born after 1990 (female at 0.9%, male at 0.8%). Sensitivity analysis was performed by selecting the youngest member in the birth year-gender strata, showing no difference in the count of HBsAg positive cases. On an individual level, 155/2076 participants tested HBsAg positive, giving a crude prevalence of 7.5% (95% CI 6.3%–8.6%), and age- and sex-adjusted prevalence of 6.3% (95% CI 5.3%–7.4%) (Supplementary Table S4). Comparing between subjects with HBsAg results, the proportion positive for HBsAg was higher among those born in Mainland China than Hong Kong (12% vs. 5%, odds ratio (OR) 0.40, 95% CI 0.29–0.57).
Birth year-gender stratified prevalence of HBsAg (selection for more than 1 member per household in age-gender strata: smallest member number).
The HBsAg positive population
There were altogether 155 (7%) HBsAg positive persons out of 2070 participants. Comparing with HBsAg negative participants, HBsAg positive participants were older (OR 1.01, 95% CI 1.004–1.03) or more likely born in or before 1990 (OR 11.38, 95% CI 2.80–46.21) (Table 1). Participants having attended local childhood immunisation programme (adjusted OR (aOR) 0.14, 95% CI 0.05–0.36), who had a post-secondary education level (aOR 0.59, 95% CI 0.41–0.86), born in Hong Kong (aOR 0.41, 95% CI 0.29–0.58), and being local residents (aOR 0.38, 95% CI 0.15–0.93) were less likely to be HBsAg positive after adjusting by age. All HBsAg positive subjects were ethnic Chinese, and none giving a history of illicit drug use or dialysis. In the sub-analysis, the same variables were significantly associated with HBsAg positive results after adjusting by age and by household member testing HBsAg positive, except for being local residents, history of fatty liver and number of lifetime sex partners which were insignificant factors (Supplementary Table S5).
Comparison between HBsAg negative (N = 1915) and positive participants (N = 155)
a age as the confounder in multivariable logistic regression.
b Born in Hong Kong in or after 1984 or migrate to Hong Kong at the age 12 or below.
*P < 0.05.
Only 59% (91/155) of the HBsAg positive participants were aware of their status and self-reported as HBV carriers (referred as ‘known carriers’ in this study) (Table 2). Among known carriers, 11% (9/80) were diagnosed within 5 years prior to the survey, while 34% (27/79) have ever been in care, and 10% (12/88) on antiviral treatment. Ten out of 150 HBsAg positive participants (6.7%, 95% CI 2.6%–10.7%) were HBeAg positive: four male, with median age of 47.5 (IQR = 36–54 years old), six being known carriers, and none in HBV care or treatment. There was no difference in HBeAg positive status and age between known carriers and those unaware of their infection status (unknown carriers). Known carriers were more likely to have family member with HBV infection (OR 2.34, 95% CI 1.17–4.65), and past history of surgery (OR 3.17, 95% CI 1.41–7.10; Never as reference). A lower proportion of ‘known carriers’ had household members who tested HBsAg positive (OR 0.28, 95% CI 0.09–0.88).
Characteristics of participants who tested HBsAg positive in the study (n = 155) with comparison between known carriers (n = 91) and unknown carriers who were unaware of their status (n = 64)
*P < 0.05.
Population immunity from HBV infection
Overall 1915 (93%) tested negative for HBsAg. The proportion of participants with self-reported history of vaccination by year of birth was: After 1990: 57%; 1981–1990: 56%; 1971–1980: 38%; 1961–1970: 33%; before 1960: 22%. Serologically, a total of 1037 (50%) participants tested negative for HBsAg (HBsAg-) and positive for anti-HBs (anti-HBs + ) (Supplementary Table S6), whose characteristics were compared with 872 anti-HBs negative (HBsAg- anti-HBs-) participants (Table 3). The median age of all HBsAg- individuals was 54 (IQR = 39–63 years old), with an older age for participants who tested anti-HBs + (OR 1.01, 95% CI 1.003–1.01). Participants born in 1990 or before (OR ranging between 2.04 for ≤1950 and 2.68 for 1971–1980; with >1990 as reference) were more likely to test for anti-HBs. Adjusted by age, locally born participants (aOR 0.66, 95% CI 0.53–0.81), local permanent residents (aOR 0.36, 95% CI 0.17–0.78), and those having attended local childhood immunisation programme (aOR 0.54, 95% CI 0.38–0.78) were less likely to test anti-HBs + . Participants with self-reported history of HBV vaccination (aOR 2.70, 95% CI 2.19–3.32) were more likely to be anti-HBs + . Overall comparisons between immune and susceptible, and between immune and chronically infected are shown in Supplementary Table S7.
Characteristics of HBsAg negative participants (n = 1909) with comparison between anti-HBs negative (n = 872) and positive (n = 1037) cases
a adjusted by age in multivariable logistic regression.
b Born in Hong Kong in or after 1984 or migrate to Hong Kong at the age 12 or below.
*P < 0.05.
Natural infection vs. HBV vaccination
Excluding the 7.5% HBsAg positive participants with chronic HBV infection, some 17% (144/864) of those who tested negative for both HBsAg and anti-HBs, were anti-HBc positive, reflecting natural infection following exposure. Anti-HBc positive participants were older than anti-HBc negative counterparts (median age 64 vs. 49, OR 1.08, 95% CI 1.06–1.09) (Table 4). They were less likely to be born locally (aOR 0.27, 95% CI 0.18–0.41), and have a post-secondary education level (aOR 0.55, 95% CI 0.35–0.87), but more likely to be married (aOR 2.24, 95% CI 1.12–4.48) after adjusting by age, as compared to anti-HBc negative participants.
Characteristics of participants who tested negative for both HBsAg and anti-HBs (n = 872), with comparison by anti-HBc test result (720 anti-HBc −ve, 144 anti-HBc + ve)
a adjusted by age in multivariable logistic regression.
b Born in Hong Kong in or after 1984 or migrate to Hong Kong at the age 12 or below.
*P < 0.05.
Discussion
In this population-level HBV study conducted in 2020, the crude HBsAg prevalence was 7.5%, which places Hong Kong in the high-intermediate endemicity category by epidemiologic state [Reference Hou, Liu and Gu5]. With age/sex adjustment, the population prevalence was 6.5%, which is slightly lower than that of 7.2% in an earlier study of 2015/2016 [Reference Liu18], further distancing from the 8.8% in a population survey conducted 20 years ago in 2001 [15], and the 9.5% prevalence among blood donors some 30–40 years ago [Reference Lok13]. Our study provided evidence that the HBV population burden has been falling over the last decades. The family clustering and the higher prevalence in older population was similar to the pattern observed before the universal vaccination programme was in place [Reference Lok13]. The universal neonatal vaccination programme and the interventions for reducing mother-to-child transmission (MTCT) of HBV have contributed to the continued reduction of new infections. The MTCT strategy includes hepatitis B immunoglobulin prophylaxis for newborns of HBsAg positive mothers, and more recently, antivirals to mothers with high HBV viral load and post-vaccination serology testing. The neonatal vaccination and MTCT programmes have contributed to a dramatic fall of HBsAg prevalence to below 1% for people age below 30. Nevertheless, the overall reduction of HBV carriage at population level had occurred only very gradually, the pattern of which was seen in the results of a population cohort in Taiwan where the HBsAg prevalence was similarly high in the past [Reference Chen19].
With a median age of 56, the HBsAg positive persons in the current study were not just older but that over 40% were unaware of their infection status, of which <20% had been on any form of HBV therapy. Taking HBeAg positivity as a marker of chronic active hepatitis [Reference Kawanaka20], this subpopulation accounted for less than 10% of the HBsAg positive carriers, but none was on any treatment. The estimated half a million persons with positive HBsAg status in Hong Kong is a cause for concern. Epidemiologically, it appears unlikely that the burden will decrease precipitously given the pace of population ageing and the small proportion of HBsAg positive patients on any form of antiviral therapy [Reference Wong21].
In this study, population immunity as inferred from the presence of anti-HBs could be shown in only 50% of the local residents after excluding all HBsAg positive carriers. The low self-reported history of vaccination was related to the uncertainty of participants' knowledge of immunisation status which was not always available in health records. With a high vaccination coverage of over 90% in Hong Kong [15], protective immunity can be demonstrated serologically in only 35%. This suggested the lack of immunity in residents born before implementation of the universal vaccination programme and the waning of vaccination effects over time, [Reference Lu22, Reference Lao23] which was consistent with a previous local study's findings [Reference Lin and Wong24]. As our recruitment has targeted adults in households, we may not have included a high proportion of children and adolescents with higher anti-HBs following vaccination. There were a higher proportion of anti-HBc negative persons among the younger anti-HBs negative adults, suggesting the absence of natural infection following vaccination. While anamnestic responses could be elicited 20 years or more following full vaccination in childhood [Reference Bagheri-Jamebozorgi25, Reference Cocchio26], the absence of anti-HBs in such cases does pose a challenge in assessing population immunity. On the other hand, natural HBV infection might have occurred in 17% of the anti-HBs negative persons, as shown by the presence of anti-HBc. The phenomenon of isolated anti-HBc is however a technical challenge. Apart from being a possibly false positive result, it may imply occult HBV infection or occurrence during the window period of acute HBV infection and the late-stage infection after anti-HBs has fallen to undetectable levels [Reference Wu, Kwok and Tran27].
Our study carries a few limitations. First, the return rate was low (4%), and the recruitment rate was slowed by the COVID-19 epidemic. Second, self-selection bias might have arisen as we did not exclude known infected individuals and their household members, who might be attracted to join the study for checking their status. Conversely, some might have self-deferred as they have already known their infection status. Third, there might be recall bias in survey items relating to history of vaccination and potential risk exposure, while some participants were uncertain about their vaccination history in childhood. Also, we have collected minimal data for epidemiological purpose and had not covered clinical histories.
In conclusion, the introduction of universal HBV vaccination in Hong Kong has contributed to the reduction of population burden with evolvement from a high to high-intermediate endemicity status. Whereas half the population has demonstrable immuno-protection against HBV infection, the prevalence of chronic infection has fallen only slightly. The continuation of high coverage universal neonatal vaccination programme would undoubtedly further reduce the HBV burden in the coming decades. The low prevalence of HBsAg and anti-HBc among the vaccinated people, the continuing presence of anti-HBs alone or with anti-HBc supports the hypothesis of an anamnestic response and for the proposition of not giving booster doses in late adult life. Booster HBV vaccination remains controversial [Reference Le28], which is generally not recommended by international guidelines including that of the WHO [29], though single booster dose administration has been advocated [Reference Wang30]. With the high number of HBsAg positive older adults, the WHO target of providing treatment to 80% eligible hepatitis B patients by 2030 [3] would be a challenge. HBV infection in elderly people is a neglected issue the importance of which is poorly appreciated in public health context [Reference Loustaud-Ratti31]. Innovative approach to population screening in conjunction to practical strategy for linkage to care is yet to be developed.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S095026882300002X.
Acknowledgements
The authors tank CY Poon and KY So for their assistance in the laboratory work, CQ Ye and SF Wong for project coordination and data entry, HY Wong for blood sample collection and the nurses at the Parkes Medical Centre for blood taking.
Author contributions
Conceptualisation: SSL and CMP; Methodology: CMP, NSW and CPC; Formal analysis and investigation: NSW and DPCC; Data curation: LL; Result interpretation: all authors; Writing – original draft preparation: SSL and NSW; Writing – review and editing: EKY and all other authors; Funding acquisition: SSL and CMP; Supervision: SSL. All authors have read and approved the final manuscript.
Financial support
This work was supported by Health and Medical Research Fund (Grant number: 17160962).
Conflict of interest
None.
Data availability statement
The data that support the findings of this study are available from the corresponding author, SS Lee, upon reasonable request.
Open access
