Chronic hepatitis B virus (HBV) infection is a major source of morbidity and mortality worldwide. An estimated 257 million persons worldwide are chronically infected with HBV, which results in approximately 887,000 deaths annually (1). In the United States, 847,000 persons were living with the disease in 2011–2012 (2) and 1,698 deaths were attributed to HBV in 2016 (3). About two-thirds of persons with chronic HBV infection were unaware of their infection, and an estimated 15% of treatment-eligible patients are receiving treatment (4, 5).

Chronic HBV infection is defined as the presence of hepatitis B surface antigen (HBsAg) for more than 6 months. Prevalence is highest among persons born in the Pacific Islands, Asia and Africa (2, 6). Other high-burden populations include men who have sex with men (MSM), persons who inject drugs (PWID), persons who have been incarcerated, and sexual and household contacts of persons with HBV infection (7). The Centers for Disease Control and Prevention (CDC) recommends screening for hepatitis B among person at risk; however current guidelines do not include testing persons with a history of STI (8).

Since both sexually transmitted infections (STI) and HBV infections can be transmitted through unprotected sexual contact, persons with current or past STI are at elevated risk for HBV infection. As a result, CDC recommends hepatitis B vaccination in adults seeking evaluation or treatment for STIs (9). While CDC does not recommend pre-vaccination testing for persons with current or past STI, the American Association for the Study of Liver Disease recommends hepatitis B testing for these persons (10), and the American College of Physicians recommends hepatitis B testing for persons with a history of STI as a best practice (7). Hepatitis B is commonly transmitted through sexual contact (11); however, the hepatitis B prevalence among persons with current or past STI is unknown. We report the association of hepatitis B prevalence among persons with current and past STI through a comprehensive systematic review and meta-analysis.

Our methods are based on recommendations of the Cochrane Collaboration (12), and reporting is consistent with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and checklist; details available in Appendix A (13).

We undertook a systematic review and meta-analysis of the relationship between HBV infection and current or past STIs. We found a substantial body of evidence suggesting that there are positive and statistically significant associations between the prevalence of hepatitis B and STIs in the mix of countries, settings, and subpopulations for which we found usable studies. HBsAg prevalence among persons with STIs were derived from 11 of the 13 outcomes available for studies specifically reporting HBsAg prevalence with a median value of 4.17% and a range of 0.91% to 33.20% depending on the population, setting, and infection pathogen.

Among adults in the United States, the most common route of transmission for hepatitis B virus is sexual contact Among the 26% of persons with case-reports of acute HBV infection with information about risk behaviors or exposures, 22.9% reported having two or more sexual partners, 6.5% reported sexual contact with an HBV-infected person, and 12.2% of males reported having sex with another male (26). As many as 10%−40% of adults seeking treatment in STI clinics have evidence of current or past HBV infection, and 39% were screened or sought care for an STI prior to becoming infected with HBV (27). As a result, CDC recommends vaccination of adults at risk for HBV infection including persons at risk for infection by sexual exposure (28).

CDC does not currently recommend pre-vaccination serologic testing for HBV among persons with current or a history of STI (9), although the American Association for the Study of Liver Disease has recommended testing this group since 2009 (10). HBV infection and STI share common modes of transmission, and hepatitis B can be transmitted by sexual exposure. Transmission also occurs by close person-to-person contact by open cuts and sores, in hyperendemic areas (28). Therefore, persons with STI are likely at increased risk of HBV infection and should be vaccinated, and considered for pre-vaccination serologic testing (29). Screening, linkage to care, and vaccination programs for adults at risk for hepatitis B are a best practice (6). Our review suggests high hepatitis B prevalence among persons with current or past STI. As a result of this higher hepatitis B prevalence, person with a history of STI could benefit by receiving hepatitis B vaccination if they are negative or be linked to life saving care and treatment if they are found to be positive. In addition, those who are identified as positive can be counseled and is an opportunity to disrupt transmission of the virus to others.

While there was a large range of values for HBV infection and STI prevalence in the populations included in this review, the range of prevalence ratios is much more constrained. The pooled estimates vary from 1.65 to 6.76 for HBsAg, from 1.30 to 1.82 for anti-HBc and 1.15 to 1.89 unspecified HBsAg / anti-HBc. This suggests that, given a known prevalence of STIs within a potential target population, one might make reasonably precise estimates of the yield of HBV-positive persons.

We observed an overall clustering of effect sizes around the top of all three funnels reflecting the preponderance of large studies. This observation suggests that smaller studies could have been excluded. However, there is no reason to think that this exclusion was due to the magnitude or direction of their association. In addition, any potential exclusion of such studies is unlikely to affect the pooled effect sizes. For one of the three plots, we observed a few effect sizes on the right side of the plot but not on the left side at the bottom of the funnel; in addition to risk of publication bias, data heterogeneity chance, and other reasons may explain this asymmetry (17). (See Appendix F).

Among the strengths of this paper are its comprehensive look at the range of relevant HBV markers and STIs combined with a sufficiently large number of outcomes to test for statistical significance within each stratum. Thus, we were able to group ‘like with like’ and still obtain, for most pooled estimates, a fairly narrow confidence interval. The results were robust to sensitivity analyses in which a study with a different approach to portraying STI outcomes was omitted, and to confining the analyses to unadjusted outcomes only.

However, there are limitations to the methods employed in this study. We confined our literature search to PubMed, Embase and the Web of Science data bases, and did not exhaustively review the grey literature. It is thus possible that we excluded relevant information that may have affected findings. Because there are no other published systematic reviews pertaining to our research question, it is not possible to compare our results against those of other investigators who might have used different methods. As in all systematic reviews, we evaluated only studies that met inclusion criteria. Broader criteria, such as acceptance of earlier studies, or a different geographic or economic specification of included countries, might have yielded a different result, although the inclusion / exclusion criteria used are appropriate to the research question we addressed.

Given the existence of contextual heterogeneity (i.e., variation in population, study design, outcome reporting metrics) and statistical heterogeneity, we used random effects. Thus, the pooled point estimates provide an overall average effect size that may not be accurate for specific target populations and settings. Therefore, in addition to point estimates, confidence intervals should be taken into considerations for decision-making pertaining to disease screening programs or modeling. Given the cross-sectional nature of most of studies and the fact that both types of infections share common risk factors, we only assessed their association, not causal relationships. Related, during the quality assessment, we did not reduce the overall quality rating for studies that did not measure exposures prior to the outcome of interest since it would not be ethical in most instances to delay an effective vaccination in order to monitor outcomes overtime.

We derived PRs from adjusted measures of association when available and thus have a mix of adjusted and unadjusted PRs, Sensitivity analyses suggests this is not a major issue. Not all studies included HBsAg, which may have limited our estimations of the association between STI and infectious HBV. Including studies that reported on HBsAg or anti-HBc markers but did not specify which was used to determine hepatitis B status introduces an element of ambiguity in the interpretation of that sub-set of findings. However, these findings are consistent with findings pertaining to the known HBV marker types. Removing these results does not affect our overall findings. Even if we stratified data by HBV markers and STI pathogens, there is still a high degree of heterogeneity in some of the pooled analyses. This suggests that factors other than those for which we stratified may explain the heterogeneity.

This systematic review and meta-analysis found a substantial body of evidence suggesting robust and statistically significant associations between the prevalence of HBV infection and STIs for multiple infections and using various serologic tests. These findings may be useful in better understanding the potential benefits of pre-vaccination serologic testing for HBV among persons with or at risk for STIs and framing potential recommendations for expanded hepatitis B testing.