This review was prepared in accordance with PRISMA guidelines [14]. The systematic review research protocol is available online with the supporting information (See Text S1). Medline, Medline In-Process, EMBASE, and the Cochrane Database of Systematic Reviews were searched for studies reporting the seroprevalence of chronic HBV infection and immunity in immigrants and refugees, using no initial language restrictions, from the earliest date until November 1^st, 2011. The search strategy that was used for all databases is shown in Table 1. Additional studies were identified by examining the reference lists of eligible studies and the bibliography of review articles. The titles and abstracts of all identified articles were reviewed by two authors. For those studies that met the pre-defined eligibility criteria, full-text articles were obtained and reviewed by two authors. Reasons for exclusion were recorded for all full-text articles. Original studies published in English, French or Italian that reported data on the seroprevalence of chronic HBV infection and/or HBV immunity in immigrants or refugees originating from intermediate or high HBV prevalence countries and arriving in low HBV prevalence countries were included. Only studies that recruited migrants who were representative of the general migrant population were included. Studies that focused on migrant groups at increased risk for hepatitis B due to certain behaviours were excluded (i.e. sex workers, injection drug users, etc…) [15], [16].

Data were extracted independently and in duplicate, using a piloted and standardized data extraction form, by two authors for all articles in English or French. For the one study reviewed that was published in Italian, data was extracted by one author. Data were then entered independently and in duplicate into a Microsoft Access Database and imported into SAS (version 9.2, Cary, North Carolina) for comparing, cleaning and analysis. Discrepancies were corrected by consensus.

The primary study outcomes were 1) the proportion of subjects with chronic HBV infection among those screened and 2) the proportion of subjects with HBV immunity among those screened. Chronic HBV infection was defined as the presence of HBsAg, which is a serologic marker of either acute or chronic HBV infection. We assumed the presence of HBsAg represented imported chronic HBV infections rather than a new acute infection, given the fact that most immigrants and refugees from intermediate and high HBV prevalence countries acquire HBV infection during the perinatal period or in early childhood [5]. HBV immunity was defined as the presence of hepatitis B surface antibody (anti-HBs), with or without hepatitis B core antibody (anti-HBc). Anti-HBs are serologic markers that appear in individuals who have resolved an acute HBV infection and may also be found in individuals who have been immunized.

Data on study design, decade of study, country of landing, immigrant status, migrants’ region of origin, mean or median age, gender distribution, co-morbidities, method of participant identification for study, and serologic testing method used were also extracted. The method of participant selection was categorized as occurring in reception centers at the time of arrival, in the context of a clinic or hospital visit, screening of pregnant women or other situations [i.e. screening studies in general host populations that included a subset of migrants (i.e. the army, schools or community surveys) or studies which invited certain immigrant groups to be screened]. Data on immigrant status was classified into the following categories: immigrants, refugees, asylum seekers, and adopted children. A dichotomous variable of immigrant status combined immigrants and adopted children into ‘immigrants’ and refugees and asylum seekers into ‘refugees’, because of the very small number of studies with adopted children or asylum seekers. Region of origin was classified according to the World Bank country classification and included the following regions: Latin America and the Caribbean, Eastern Europe and Central Asia, Middle East and North Africa, Sub-Saharan Africa, South Asia, and East Asia and the Pacific [17]. If a study contained migrants from more than one region, then seroprevalences for infection and immunity were obtained separately for each region of origin, if possible.

For each study, the seroprevalence of chronic HBV infection and HBV immunity was calculated by using the reported numbers of subjects positive for HBsAg and anti-HBs serologic markers, respectively, divided by the total number of people screened for each of these markers. Proportions were transformed with the logit transformation and pooled using a random-effects model to account for the expected high heterogeneity between studies [18]. The logit-transformed proportions were back-transformed and results were presented as percentages. The logit transformation was used to avoid studies with few events from being weighted too heavily in the random-effects model [19] and because the multivariate analysis, which used a random-effects logistic regression model, is also based on the logit transformation (see below). Overall heterogeneity among studies was assessed using the I² statistic and estimates were stratified by region of origin and immigration status, as these variables were believed a priori to be important predictors of HBV infection [20]. The data were not stratified by age or gender as few studies, 40% and 26% respectively, included any form of information on these variables. The meta-analysis was performed using the metaprop command of the meta package in R (version 2.13.1) [21].

A random-effects logistic regression, using the glmer command of the lme4 package to fit generalized linear mixed-models in R, was used to further explore heterogeneity and to compare the odds of being infected among migrants from the different regions of origin after adjusting for potential confounding by immigrant status and decade of study [22]. Three studies that did not report estimates separately for immigrants and refugees and 18 studies that did not report estimates separately by region of origin were excluded from the random-effects logistic regression model.

To estimate the number of migrants who are chronically infected with hepatitis B in immigrant-receiving countries, data on the number of foreign-born residents from different regions of origin living in host countries was obtained from the national statistical agencies for each of these countries (see Appendix Table 1 in Text S2). This data and the pooled region-specific seroprevalence of chronic HBV infection in immigrants and refugees estimated in this review were multiplied to estimate the burden of chronic HBV infection in migrants living in traditional immigrant-receiving countries.

A total of 1,456 citations were identified in the electronic search and an additional 11 citations were identified through hand searching (Figure 1). After duplicates were removed, 926 citations were screened with the title and abstract and 757 were excluded. A total of 169 articles were assessed with predefined eligibility criteria in the full-text review and 53 were excluded. An additional six articles were excluded because the full-text could not be retrieved despite assistance from a McGill University librarian. A total of 110 studies were included in the systematic review and meta-analysis (Appendix Table 2 in Text S2).

All 110 studies reported the seroprevalence of HBV infection in immigrants or refugees, representing a total of 209,822 migrants from all world regions. Just over half of the seroprevalence studies of chronic HBV infection included refugees and asylum seekers and nearly a quarter studied migrants exclusively from East Asia and the Pacific, although almost half of all the studies screened migrants from more than one region (Appendix Table 3 in Text S2). All studies were conducted in the context of asymptomatic screening and almost 90% of study participants were recruited in settings with low risk for selection bias (i.e. mass screening in reception centers or among pregnant women). The number of studies published on the seroprevalence of HBV infection increased every decade after the initial studies first reported on the seroprevalence of infection in refugees from Southeast Asia.

The overall pooled seroprevalence of infection in all international migrants, among the 110 studies, was 7.2% (95% CI: 6.3%–8.2%). Higher seroprevalences were found among refugees and asylum seekers, compared immigrants (Table 2). Pooled HBsAg seroprevalence estimates by region of origin differed from one another and were similar to global estimates from the Centers for Disease Control and Prevention (CDC), the World Health Organization (WHO) and a recent systematic review of global seroprevalence of HBV infection [23]–[25]. These organizations divide the world into high (HBsAg seroprevalence ≥8%), intermediate (HBsAg seroprevalence between 2%–7%) and low (<2%) chronic hepatitis B endemic regions. Our region-specific pooled proportions for migrants fell within the same ranges for the HBV seroprevalence categories defined by the WHO and CDC (see Figure S1). Chronic HBV seroprevalence was high for migrants from East Asia and the Pacific and from Sub-Saharan Africa [11.3% (95% CI: 10.3%–12.4%) and 10.3% (95% CI: 9.1%–11.8%), respectively], intermediate for migrants from Eastern Europe and Central Asia and from South Asia [5.8% (95% CI: 4.3%–7.9%) and 4.6% (95% CI: 2.6%–7.8%), respectively], and low for migrants from Latin America and the Caribbean and the Middle East and North Africa [1.7% (95% CI 1.1%–2.7%) and 2.0% (95% CI: 1.6%–2.9%), respectively].

Results from the random-effects logistic regression analysis showing both unadjusted and adjusted odds ratios of chronic HBV infection are reported in Table 3. Different models, with and without interaction terms, using traditional model selection criteria were explored using region of origin, immigrant status and decade of study as variables. The best model without interaction terms, as measured with the lowest Akaike information criterion (AIC) value, included only region of origin, however models including immigrant status and decade of study were very similar (odds ratio differences by ≤0.06). Models that included interactions between variables were also explored. Although there were some statistically significant interactions in some models, there was no consistent significant trend of interactions across models. The model with all three of these variables without interactions was chosen as we felt that this was the most relevant for clinicians working with migrant populations. In the final model, migrants from all other world regions had significantly greater odds of being chronically infected with HBV as compared to migrants from Latin America and the Caribbean, the region with the lowest seroprevalence and the one most closely approximating the estimated prevalence in immigrant-receiving countries, after adjusting for immigrant status and decade of study. Refugees were found to have 1.42 (95% CI: 1.01–1.99) greater odds of being chronically infected compared to immigrants, after adjustment for region of origin and decade of study. Migrants in studies from the 1990s were found to have 1.58 (95% CI: 1.03–2.43) greater odds of being chronically infected, compared to migrants from studies reported in the 1980s, after adjusting for the other study-level covariates.

Of the 110 studies included in this systematic review, 39 reported data on the seroprevalence of HBV immunity in immigrants or refugees and included data on 40,330 migrants from all world regions. As with studies reporting the seroprevalence of chronic HBV infection, just over half of the studies examining HBV immunity screened refugees (Appendix Table 4 in Text S2). The overall pooled seroprevalence of immunity to HBV was 39.7% (95% CI: 35.7%–43.9%) and was highest in those regions that had the highest seroprevalence of chronic HBV infection (Table 4). Half of the migrants from East Asia & the Pacific [50.2% (95% CI: 45.8%–54.6%)] and almost half of Sub-Saharan African migrants [41.7% (95% CI: 37.6%–45.9%)] were immune to HBV. The seroprevalence of immunity was lower in regions where the seroprevalence of chronic HBV infection was lowest. However, the number of subjects screened for anti-HBs from the Middle East and North Africa (n = 131), South Asia (n = 364) and Latin America and the Caribbean (n = 29) were very small and results from these regions were imprecise. There were too few studies to examine differences in immunity between immigrants and refugees from each specific region of origin.

Almost 3.5 million immigrants and refugees (95% CI: 2.8–4.5) living in immigrant-receiving countries were estimated to be chronically infected with HBV (Table 5 and Figures S2 and S3). The proportion of all migrants chronically infected ranged from 3.7% to 9.7% in the different host countries with the largest estimated number residing in the United States (1.6 million), Canada (285,000), Germany (284,000), Italy (201,000), the United Kingdom (193,000), and Australia (176,000).