Hepatitis B immunization for indigenous adults, Australia

Abstract Objective To quantify the disparity in incidence of hepatitis B between indigenous and non-indigenous people in Australia, and to estimate the potential impact of a hepatitis B immunization programme targeting non-immune indigenous adults. Methods Using national data on persons with newly acquired hepatitis B disease notified between 2005 and 2012, we estimated incident infection rates and rate ratios comparing indigenous and non-indigenous people, with adjustments for underreporting. The potential impact of a hepatitis B immunization programme targeting non-immune indigenous adults was projected using a Markov chain Monte Carlo simulation model. Findings Of the 54 522 persons with hepatitis B disease notified between 1 January 2005 and 31 December 2012, 1953 infections were newly acquired. Acute hepatitis B infection notification rates were significantly higher for indigenous than non-indigenous Australians. The rates per 100 000 population for all ages were 3.6 (156/4 368 511) and 1.1 (1797/168 449 302) for indigenous and non-indigenous people respectively. The rate ratio of age-standardized notifications was 4.0 (95% confidence interval: 3.7–4.3). If 50% of non-immune indigenous adults (20% of all indigenous adults) were vaccinated over a 10-year programme a projected 527–549 new cases of acute hepatitis B would be prevented. Conclusion There continues to be significant health inequity between indigenous and non-indigenous Australians in relation to vaccine-preventable hepatitis B disease. An immunization programme targeting indigenous Australian adults could have considerable impact in terms of cases of acute hepatitis B prevented, with a relatively low number needed to vaccinate to prevent each case.


Introduction
Australia in 2016 has a population of approximately 24 million people. Indigenous Australians (Aboriginal and Torres Strait Islander peoples) constitute 3.0% of the population and have a younger age structure than non-indigenous Australians, similar to that of low-and middle-income countries. 1 Although health indicators for the general Australian population are comparable with those of other high-income countries, life expectancy and the burden of many diseases is greater among indigenous Australians. 2,3 The health disparities between indigenous and non-indigenous Australians have been acknowledged by the Australian government, which has implemented a whole-of-government strategy, entitled Closing the gap in indigenous disadvantage, aiming to completely remove these disparities by the year 2030. 4 Total viral hepatitis-related mortality in the Western Pacific Region of the World Health Organization (WHO) is now higher than deaths due to acquired immune deficiency syndrome, malaria and tuberculosis combined, constituting a critical public health challenge for the Region. 5,6 In Australia, hepatitis B vaccination was recommended (but not nationally funded) for infants and adults in high-risk groups, including indigenous Australians, in the late 1980s. 7 Universal vaccination of all infants commenced in the Northern Territory of Australia in 1990, followed by a funded national adolescent immunization programme starting in 1997. A funded universal infant hepatitis B immunization programme was introduced nationally in May 2000. 7 The seroprevalence of hepatitis B virus surface antigen (HBsAg) in indigenous Australian adults was estimated to be 17% in a meta-analysis of studies conducted before 2000, thus meeting the WHO definition for high endemicity. 8 Since then, HBsAg seroprevalence in indigenous Australians is estimated to have declined to 3.7% of the 548 366 population nationally in 2011. 9 This is a WHO-defined intermediate level of endemicity, but is still more than 10 times the rate in non-indigenous Australians born in Australia (0.3% of 13 836 559, excluding people who inject drugs and men who have sex with men). 9 Indigenous Australians also have a higher prevalence of comorbidities such as type 2 diabetes mellitus and alcoholrelated liver disease 10 which are associated with poorer prognosis and more rapid progression of chronic hepatitis B. 8,9,11,12 The incidence of liver cancer is up to 10 times higher compared with non-indigenous Australians. 8,9 In the light of this increased risk, indigenous Australians have been identified as a priority group for hepatitis B testing and immunization. [13][14][15] Universal hepatitis B infant immunization is funded under the Australian national immunization programme, with a school-based adolescent catch-up programme funded to 2013 (by which time the cohort immunized in infancy will have reached adolescence). However, funding and access to hepatitis B vaccination for adults at higher risk, including indigenous Australians, is limited and inconsistent across the eight Australian states and territories (jurisdictions). 16 Although data on hepatitis B vaccination coverage in indigenous adults are very limited, we believe that national coverage is likely to be low. This view is based on various evidence: from a sentinel study quantifying vaccination coverage from analysis of markers of hepatitis B infection; 17 from the poor coverage achieved in funded influenza and pneumococcal vaccination programmes Objective To quantify the disparity in incidence of hepatitis B between indigenous and non-indigenous people in Australia, and to estimate the potential impact of a hepatitis B immunization programme targeting non-immune indigenous adults. Methods Using national data on persons with newly acquired hepatitis B disease notified between 2005 and 2012, we estimated incident infection rates and rate ratios comparing indigenous and non-indigenous people, with adjustments for underreporting. The potential impact of a hepatitis B immunization programme targeting non-immune indigenous adults was projected using a Markov chain Monte Carlo simulation model. Findings Of the 54 522 persons with hepatitis B disease notified between 1 January 2005 and 31 December 2012, 1953 infections were newly acquired. Acute hepatitis B infection notification rates were significantly higher for indigenous than non-indigenous Australians. The rates per 100 000 population for all ages were 3.6 (156/4 368 511) and 1.1 (1797/168 449 302) for indigenous and non-indigenous people respectively. The rate ratio of age-standardized notifications was 4.0 (95% confidence interval: 3.7-4.3). If 50% of non-immune indigenous adults (20% of all indigenous adults) were vaccinated over a 10-year programme a projected 527-549 new cases of acute hepatitis B would be prevented. Conclusion There continues to be significant health inequity between indigenous and non-indigenous Australians in relation to vaccinepreventable hepatitis B disease. An immunization programme targeting indigenous Australian adults could have considerable impact in terms of cases of acute hepatitis B prevented, with a relatively low number needed to vaccinate to prevent each case.

Research
Hepatitis B virus immunization, Australia Andre Louis Wattiaux et al.
targeted on indigenous adults aged 18-49 years with relevant risk factors; 18 and on the basis of the ongoing higher rate of incident infections in indigenous than non-indigenous adults. 19 In this study, we aimed to review the epidemiology of newly acquired hepatitis B (the vaccine-preventable fraction of the burden of disease) in indigenous and non-indigenous people in Australia. We also estimated the potential impact of more systematic implementation of hepatitis B vaccination for all nonimmune indigenous Australian adults.

Data sources
Data on all cases of hepatitis B notified from 1 January 2005 to 31 December 2012 were obtained from the Australian National Notifiable Diseases Surveillance System (NNDSS). 20 The NNDSS compiles notification data from all eight Australian jurisdictions. Variables extracted for analysis included year of notification, age, sex, jurisdiction and indigenous status.
Data on the mid-year estimated national resident population for the same years (2005-2012) by jurisdiction; indigenous status and age were obtained from the Australian Bureau of Statistics. 1

Definitions
Indigenous notifications included all individuals with hepatitis B notification whose indigenous status was recorded in the NNDSS as Aboriginal or Torres Strait Islander, or both; non-indigenous notifications included all other individuals with hepatitis B notification, including those whose indigenous status was unknown. The indigenous population was individuals who self-identified in the Australian census as being Aboriginal or Torres Strait Islander, or both (with the numbers adjusted for net undercount measured by a post-enumeration survey); non-indigenous was the remaining resident Australian population.
Since 2004 the national surveillance case definition of newly acquired (i.e. acute) hepatitis B is based on laboratory confirmation of infection by one of the following criteria: (i) detection of HBsAg in a patient shown to be negative within the last 24 months; or (ii) detection of HBsAg and of immunoglobulin (Ig) M to hepatitis B core antigen (anti-HBc IgM), in the absence of prior evidence of hepatitis B virus infection; or (iii) detection of hepatitis B virus by nucleic acid testing and of anti-HBc IgM, in the absence of prior evidence of hepatitis B virus infection. 21 Confirmed cases of newly acquired hepatitis B are notifiable under public health legislation in each jurisdiction. Unspecified cases of hepatitis B (those with laboratory-definitive evidence but not meeting any of the criteria for a newly acquired case) 22 are assumed to be predominantly episodes of chronic hepatitis B and are notifiable.

Ethical considerations
Ethics approval was not required for this study as de-identified, aggregate population-based data were used for routine public health purposes only.

Data analysis
Analysis was restricted to newly acquired hepatitis B notification data. This was because the unspecified hepatitis B notification data do not differentiate between acute and chronic hepatitis B. Data were analysed by age group, sex, jurisdiction and indigenous status. While the completeness of recording indigenous status in notification data has improved, some variation exists between jurisdictions and over time. For this analysis, individuals whose indigenous status was unknown were classified as non-indigenous, according to established practice. 23 To assess any impact of under-identification of indigenous status, a separate analysis was undertaken excluding data from jurisdictions that had completeness of recording indigenous status below 95%.
Australian Bureau of Statistics' population data were used to calculate hepatitis B notification rates per 100 000 population and perform direct age standardization using the total Australian resident population as the standard. 1 Rate ratios comparing notification rates for indigenous and other Australians were calculated with 95% confidence intervals (CI). Analyses were performed using Stata statistical software version 12.0 (Stata Corp., College Station, United States of America [USA]).

Vaccination impact estimates
The proportion of indigenous Australians aged 15 years or older who were non-immune, by age group, was estimated from Australian Bureau of Statistics' population data 1 and the authors' (FB) expert opinion (Table 1). Expert opinion was based on a review of published HBsAg seroprevalence data 8,9,[24][25][26][27][28] and published 17 and unpublished hepatitis B vaccination coverage estimates, and was informed by previous experience implementing immunization programmes targeting indigenous Australians. While hepatitis B vaccination coverage in indigenous Australian infants has been consistently high (in the vicinity of 95%) after universal immunization was introduced in the year 2000, 23,29,30 coverage for adolescents was estimated to have been moderate and for adults was estimated to have been poor. We also accounted for the complexity involved in delivery of a hepatitis B immunization programme (including baseline serological testing and administration of three doses of vaccine).
On this basis we estimated the potential impact of hepatitis B immunization for two vaccination coverage scenarios: (i) low coverage, in which 25% of the susceptible population of 164 427 were vaccinated by the end of the 10-year programme, and (ii) high coverage, in which 50% of the susceptible population was vaccinated. These figures correspond to 10% and 20% respectively of the total indigenous adult population. A range of possible coverage, determined by author's (FB) expert opinion, was used for each of these scenarios, with cumulative coverage plotted for each year of the 10-year programme ( Table 2; available at: http://www.who. int/bulletin/volumes/94/11/16-169524).
We also used the model to estimate the vaccination coverage level at which hepatitis B incidence among indigenous Australians would be reduced to the current rate among non-indigenous Australians.
We estimated the impact of a hepatitis B immunization programme for non-immune indigenous Australians aged ≥ 15 years in terms of the number of cases of acute and additional chronic hepatitis B infections prevented, and the number needed to vaccinate to prevent each case. To do this we developed a Markov chain Monte Carlo simulation using a random walk (i.e. a computerized run-through of 100 000 scenarios, each with variables randomly selected from the range of parameter values outlined in Table 1), using Excel 2010 software (Microsoft Corp., Redmond, USA). Our model was built on a Markov chain with one-year cycles and allowed for agedependent transitions. This type of model

Hepatitis B virus immunization, Australia
Andre Louis Wattiaux et al.
structure conceptualizes health or disease status as a series of mutually exclusive and collectively exhaustive health states.
In each cycle of the model, individuals either reside in one of the health states, or transition probabilistically to another at the end of a given cycle. Health benefits are accrued based on the state occupied during a given cycle. A decision tree was used to compare the impact of a hepatitis B immunization programme targeting indigenous Australian adults, in terms of progression to acute or chronic infection, against the baseline of current practice (Fig. 1).
The variables incorporated into the model included: age group, estimated population, estimated proportion of people susceptible to hepatitis B, estimated baseline number of acute infections per year and vaccination seroconversion rate. Estimates of the risk of susceptible individuals developing acute infection and of progression to chronic infection were also incorporated into the model; these were derived from the literature from areas with intermediate or low endemicity for adolescents [33][34][35][36][37] and adults. [38][39][40]44 For the probability distribution of age group-specific parameters (Table 1), we assumed a normal distribution for the proportion of the population susceptible to hepatitis B infection, the seroconversion rate after vaccination and the projected cumulative additional vaccination coverage achieved in the hypothetical  [33][34][35][36][37] and adults. [38][39][40][41][42][43][44] The values of lower and upper limits were based on the author's (FB) expert opinion. e Estimate from Edmunds et al. 1993. 45 Notes: Risk of developing acute infection in susceptible individuals was calculated by dividing no. of newly acquired cases by age-specific population counts. Projected cumulative additional immunization coverage is presented in Table 2. 10-year immunization programme. We used a triangular distribution for the risk of progression to chronic infection in newly acquired cases.
The key assumptions used in our model were as follows. Records of newly acquired cases in the NNDSS were considered as cases of acute hepatitis B virus infection. To adjust for underreporting and misclassification of acute hepatitis B infection, we used previously modelled data from Australia (similar to estimates from the USA) 31,46 to estimate the true number of acute infections. We did this by applying a multiplier of 10 to the average annual number of NNDSS notifications of newly acquired hepatitis B for each age group over the period 2005-2012. Vaccine efficacy was considered all-or-nothing rather than leaky (partial protection). Vaccine efficacy for specific age groups was assumed to be equivalent to the seroconversion rate derived from overseas studies, as this is a well-established and robust surrogate of clinical protection. 32 Infected individuals or seroconverted vaccinees were assumed to stay immune indefinitely. Finally, individuals younger than 15 years at the programme start, but who subsequently entered the model when they reached 15 years of age during the modelled programme, were assumed to be immune due to high levels of vaccination coverage in this cohort.
The potential effects on herd protection and mother-to-child transmission were not factored into the model. Any additional vaccination coverage achieved over the 10-year period through existing mechanisms (characterized by inconsistent funding and poor promotion) was assumed to be low and was not factored into the model.  (Table 3).

Notifications
Notification rates for newly acquired hepatitis B between 2005 and 2012 showed no significant changes over time for indigenous Australians (Fig. 2). However, there was a significant downward trend in the annual notification rate for non-indigenous Australians, falling from 1.1 per 100 000 population in 2005 to 0.8 per 100 000 in 2012 (P < 0.001). The annual rate ratios for newly acquired hepatitis B showed no significant change over the period (Fig. 2).

Vaccination impact estimates
We estimated the size of the hepatitis B non-immune indigenous Australian population aged ≥ 15 years to be approximately 164 000 (38% of the total indigenous population aged ≥ 15 years of 429 261; Table 1).
With no additional adult vaccination coverage above that already occurring, modelling predicted an additional 1792 new acute hepatitis B cases in indigenous individuals aged ≥ 15 years over a 10-year period.
Potential health gains for each scenario are summarized in Table 4; the projected annual trends in acute  (Table 4). For chronic hepatitis B the projected numbers of cases prevented were proportionately lower and the number needed to vaccinate were higher for both scenarios (Table 4).

Discussion
Consistent with a previous study, 23 we found that rates of notification for newly acquired hepatitis B were significantly higher for indigenous than non-indigenous Australians. Low rates of acute hepatitis B infection in both indigenous and non-indigenous Australians younger than 15 years 23 reflect the success of the universal infant hepatitis B immunization programme, which began in Australia in 2000, building on targeted programmes during the preceding decade.
While current Australian guidelines recommend hepatitis B vaccination to be offered to all non-immune indigenous Australians, vaccination for indigenous adults is not funded under the national immunization programme and current uptake is thought to be poor. Our study shows that a hepatitis B immunization programme for indigenous Australians aged ≥ 15 years could have considerable impact in terms of cases of acute hepatitis B prevented, with a relatively low number of persons needed to vaccinate to prevent each case. Prevention of acute hepatitis B infection would also have an impact on the ultimate number of cases of chronic infection. Without such a programme, it will almost certainly take several decades for the disparity in rates of acute hepatitis B infection between indigenous and non-indigenous Austra-lian adults to reduce, as those vaccinated as infants gradually age into adulthood.
As of 2008, 177 countries had introduced hepatitis B vaccine into their national infant immunization programmes. 47 This is estimated to have prevented more than 80% of the 1 400 000 hepatitis-B-related deaths that would otherwise have occurred worldwide since WHO's initial recommen-dation in 1997. 48 Our findings support WHO's recommendation that catch-up campaigns for hepatitis B vaccination be considered for adolescents or adults in high-prevalence settings once infant immunization is established. 47 Modelling similar to that conducted in our study could be used to estimate the impact of catch-up programmes in such settings or in high-risk populations in interme-

Fig. 2. Trends in notification rates of newly acquired hepatitis B (left axis) and corresponding age-standardized rate ratios (right axis) by indigenous status, Australia, 2005-2012
Notification rate per 100 000 population Newly acquired cases for indigenous Australians Newly acquired cases for non-indigenous Australians Rate ratio between indigenous and non-indigenous Australians (secondary axis) Note: Rate ratios were calculated using direct age standardization with the total Australian resident population as the standard.  49 so too is a particular focus on offering screening, vaccination and treatment for hepatitis B to indigenous peoples. For example, a catch-up hepatitis B immunization programme for adolescents aged 15-19 years was estimated to be cost saving in China 50 and was subsequently implemented with good effect. 51 Another campaign targeting young adults aged 21-39 years in China has also been estimated to be cost saving. 51 Our study had several limitations. We did not use a dynamic model, primarily due to the lack of data on contact mixing patterns applicable to the indigenous population in Australia.
The static Markov chain Monte Carlo method we used is not able to capture herd protection effects, which could be substantial considering the welldocumented household overcrowding and heightened disease transmission that occurs within indigenous communities. 52 Also, we did not evaluate any potential incremental benefits on prevention of mother-to-child transmission. This route of transmission is uncommon in Australia, including in indigenous populations, due to the high quality of antenatal care and neonatal immunization. Nor did we estimate the protection afforded to additional individuals who may receive an incomplete course of vaccinations. Grouping notifications from people with unknown indigenous status with non-indigenous status likely underestimates the true disparity between indigenous and nonindigenous populations. 53 Lower access to health care may also contribute to underestimation of notification rates for indigenous Australians. However, this may be counteracted by recommendations in national guidelines to test all indigenous Australians for hepatitis B virus infection. We did not factor in any impact of vaccination coverage achieved through existing mechanisms over the 10-year period as this was assumed to be low. On balance, these limitations are likely to result in our estimates of potential impact being conservative.
Our study findings highlight the health disparity in hepatitis B infection between indigenous and non-indigenous Australians. This is connected to the overall disadvantage faced by indigenous Australians, which is due to a complex combination of inter-related socioeconomic, cultural and historical determinants. 4 It is also likely that hepatitis B was highly prevalent in the indigenous population before infant immunization programmes began. Multiple initiatives at many levels by the Australian government have been put in place to address the broader issues of indigenous disadvantage. The findings of our study suggest that this disparity in hepatitis B could be readily and rapidly eliminated through a modest increase in vaccination coverage among indigenous Australian adults, for example through a funded vaccination catch-up programme. ■

Hepatitis B virus immunization, Australia
Andre Louis Wattiaux et al.