There are sparse epidemiologic data on coinfection of hepatitis B (HBV) and hepatitis C (HCV) in the United States. Therefore, the aim of this study was to determine the prevalence and predictors of HBV coinfection in a large U.S. population of HCV patients. We used the National Veterans Affairs HCV Clinical Case Registry to identify patients tested for HCV during 1997-2005. Patients were categorized based on HCV exposure (any two +HCV tests or one test with a diagnostic code), HCV infection (+RNA or genotype), HBV exposure (any +HBV test, excluding +HBsAb only), and HBV infection (+HBsAg, HBV DNA, or HBeAg). The prevalence of HBV exposure among patients with HCV exposure and that of HBV infection among patients with HCV infection were determined. Multivariate logistic regression evaluated potential demographic and clinical predictors of HBV coinfection. Among 168,239 patients with HCV exposure, 58,415 patients had HBV exposure for a prevalence of 34.7% (95% confidence interval [CI] 34.5-35.0). Among 102,971 patients with HCV infection, 1,431 patients had HBV coinfection for a prevalence of 1.4% (95% CI 1.3-1.5). Independent associations with HBV coinfection compared with HCV monoinfection were age ≤50 years, male sex, positive HIV status, history of hemophilia, sickle cell anemia or thalassemia, history of blood transfusion, cocaine and other drug use; there was decreased risk in patients of Hispanic ethnicity. Conclusion: This is the largest cohort study in the U.S. on the prevalence of HBV coinfection in HCV patients. Among veterans with HCV, exposure to HBV is common (∼35%), but HBV coinfection is relatively low (1.4%). Several possible risk factors were identified. (Hepatology 2013;58:538–545)
Hepatitis B virus (HBV) and hepatitis C virus (HCV) infections are the most common causes of liver disease worldwide, affecting ∼350 million and 170 million people, respectively.[1-3] In the United States, HCV infection affects 2.7-3.2 million and HBV infection affects 800,000-1.4 million people.[1, 4] It is estimated that 2%-10% of individuals with positive HCV antibodies also have HBV antibodies.[1, 5, 6] Reported risk factors of HBV HCV coinfection include older age, male sex, Asian ethnicity, injection drug use (IDU), a high number of sexual partners, and HIV infection, but not all of these risk factors have been consistently established.[1, 6-11] Studies indicate that HBV HCV coinfection is associated with increased risk of worse clinical outcomes such as advanced fibrosis or cirrhosis, decompensation, hepatocellular carcinoma, and transplantation compared with HCV monoinfection.[1, 3, 5, 7, 8, 12, 13]
Most information on the prevalence and predictors of HBV HCV coinfection comes from studies of populations with chronic HBV infection[10-12, 14, 15] conducted primarily in Europe and Asia.[1, 6-8, 16, 17] There is a paucity of data on the epidemiology of HBV coinfection in a U.S. population where chronic HCV infection is the more prevalent infection.[1, 4] One of the few studies in the U.S. to address this topic was conducted between 1998 and 2004 among 1,257 HCV-infected patients in two New York hospitals, but additional studies with a larger number of patients are needed. Awareness of the prevalence and predictors of HBV coinfection has implications for the screening and prevention of viral hepatitis. Furthermore, understanding the epidemiology of HBV coinfection is important given the potential for worse clinical outcomes in this cohort of patients.
The Department of Veterans Affairs (VA) has the largest integrated healthcare system in the U.S. and it collects clinical data on a large number of patients within the system who have HCV. Therefore, we conducted a retrospective study using a national VA cohort of HCV-infected patients to determine the prevalence and predictors of HBV coinfection.
This is the largest study in the U.S. to examine the prevalence and predictors of HBV coinfection in a cohort of patients with HCV. Approximately 35% of patients with exposure to HCV were also exposed to HBV in this study. Among HCV-infected patients, the prevalence of HBV coinfection defined by a positive test for HBsAg, HBV DNA, or HBeAg within 1 year before or after the HCV index date was 1.4%. The prevalence ranged from 1.0%-1.7% in the sensitivity analyses. The highest age-specific prevalence of HBV coinfection was 1.6% observed in the ≤50 age group, and the lowest prevalence was 0.8% in the ≥65 age group. Asians had the highest prevalence (2.1%) of HBV coinfection and Hispanics had the lowest prevalence (1.0%). The independent predictors for HBV coinfection were male sex, positive HIV status, a history of hemophilia, sickle cell anemia or thalassemia, history of blood transfusion, cocaine and other drug use.
The prevalence of HBV exposure among HCV-infected patients has been reported around 25% in a population-based study using the National Health and Nutrition Examination Survey data. In hospital-based studies also using a U.S. cohort with HCV from New York and Detroit, the prevalence of HBV exposure has been reported around 60%.[1, 21] The relatively high prevalence of exposure to HBV as found in our study (35%) and that reported in the literature (25%-60%) suggests a possible benefit of testing for HBV exposure among patients with exposure to HCV. This high prevalence may also warrant checking for HBsAb prior to vaccinating against HBV, as recommended by the American Association for the Study of Liver Diseases (AASLD), as a more cost-effective strategy given the likelihood of previous exposure to HBV.
The prevalence of HBV coinfection has been reported between 2%-10% in both U.S. and non-U.S. studies,[1, 5-7, 9] while the prevalence was only 1.4% in this study. Occult HBV infection (HBV DNA-positive and HBsAg-negative) was not captured in the definition used in our study for prevalence estimates and could have contributed to the relatively low prevalence.[5, 6] Although it is possible that the prevalence of HBV coinfection among HCV cohorts is actually lower than previously reported, because another large study from Australia with 117,547 patients with viral hepatitis identified 2,604 patients to have both HBV and HCV, a prevalence of 2.2%. HCV testing is likely to vary over time and among VA facilities; however, we are unable to account for changes in HCV or HBV testing in this study. It might be expected that with time newer and more sensitive tests would be used, resulting in an increased prevalence of HBV coinfection over time, but this was not observed. Based on our yearly prevalence estimates, HBV coinfection was more common in 1997, and steadily declined in subsequent years.
The VA has guidelines for HBV screening and vaccination among HCV-positive patients. However, we previously reported using the VA CCR data that in an HCV-infected cohort HBV vaccination was performed in only 22%. In the current study, the HBV vaccination rates among those with HCV exposure (any two +HCV tests or one test with a diagnostic code) and HCV infection (+RNA or genotype) were 26.4% and 32.6%, respectively. While Table 3 suggests that the temporal increase in HBV vaccination rates correspond with decreases in the prevalence of exposure to HBV and HBV coinfection, the relatively low vaccination rate is not likely to explain this drop.
The observed significant association between a history of hemophilia, sickle cell anemia, or thalassemia and a nearly 2-fold increased risk HBV coinfection has not previously been reported. One previous study has shown beta-thalassemia independent of unscreened blood consumption to be a risk factor for HCV infection, but HBV was not examined. Additionally, there has been a suggestion of patients with thalassemia being at risk for HCV not related to blood transfusions, but the source of infections was not identified.[24, 25] It is possible some risk factor(s) such as blood transfusions could not be completely accounted for in this study. While the observed association is strong, residual confounding may be present.
The findings of this study confirm the associations of cocaine, other drug use, and ethnicity with HBV coinfection.[1, 9, 11, 26] The finding that Asian ethnicity is likely associated with increased risk and Hispanic ethnicity associated with decreased risk of HBV coinfection is consistent with two smaller previous U.S. studies.[1, 9] In these studies HBV coinfection was also associated with a higher prevalence or increased risk in IDU.[1, 9] This study used cocaine and other drugs as surrogates for IDU since specific information on IDU was not available in the CCR. Both were associated with HBV coinfection, supporting the internal validity of this study, as IDU has been a consistent independent determinant of HBV coinfection.[1, 11, 26]
The findings of this study that provide insight into inconsistent reports in the literature include the association of age, sex, and blood transfusions with HBV coinfection. In non-U.S. studies there has been no significant difference in age or blood transfusions based on coinfection status.[6, 7] In another recent U.S. study by Bini and Perumalswami from two New York hospitals, male sex and blood transfusions were not independently associated with HBV coinfection. In both our study and the Bini and Perumalswami study, younger age (<50 years and <40 years, respectively) was associated with increased risk of HBV coinfection.
The prevalence estimates and predictors of HBV coinfection in our study is limited by incomplete HBV and HCV testing in this VA cohort and the inability to differentiate between acute versus chronic HBV infection. However, our sensitivity analyses indicate that our prevalence estimates are robust to these assumptions. Although we presented the findings from the multivariate model evaluating predictors of HBV coinfection in the HCV-infected cohort who had HBV testing within 1 year of the HCV index date, we also conducted a multivariate model evaluating predictors of HBV coinfection detected at any time during the study period (n = 119,323; HBV coinfection prevalence of 1.7%). The same significant predictors of HBV coinfection were identified in both models (data not shown).
We cannot account for the demographic and clinical differences observed between patients tested for HBV and those not tested for HBV. It is possible that those less likely to be tested (i.e., older patients, females, non-HIV-infected patients) may have been more likely to be HBV-infected, thus affecting our results about high-risk groups for coinfection (i.e., younger patients, male sex, HIV-infected patients). Yet the magnitude of the risk estimates and level of significance identified in the predictors of this study would suggest that the differences between patients tested for HBV and those not tested are unlikely to have affected the results. It is also possible that females may have been tested for HBV as part of obstetrical care provided outside of the VA, and this information would not be available in our dataset.
The generalizability of the findings may be limited since most patients were male. Nevertheless, more than 2,500 women were assessed. The use of secondary data from the CCR limited the examination of other possible risk factors such as number of sexual partners, detailed information on viral loads for HBV or HCV, and timing of viral infections. Additionally, evaluation of other known risk factors of HBV infection, like country of origin, could not be examined in this study, because that information is not available in the CCR. However, it is unlikely that the country of origin would have played an important factor among a veterans population that is typically comprised of American-born individuals.
A major strength of this study is the use of the HCV CCR, which provided information on over 190,000 patients with HCV. The CCR is more than an administrative database with ICD-9 codes, but also contains laboratory data, which allowed for more accurate classification of patients and variables. The use of several definitions of HBV coinfection resulted in prevalence calculations within a relatively tight range (1.0%-1.7%), supporting the robustness of the findings. This large dataset allowed the identification of a sufficient number of coinfected patients to perform detailed analyses on predictors of HBV coinfection.
The AASLD recommends screening of all persons with HCV for HBV infection. Our study shows that most persons with exposure to HCV were indeed tested for HBV (87%). While we did not perform a formal cost-effectiveness analysis of HBV screening in the study population, the 1.4% yield of actual HBV coinfection and the 35% yield of past exposure may make the screening strategy worthwhile. The utility of the AASLD recommendation is in testing both HBsAg and HBsAb as the best screening strategy.
Our study provides novel information on the prevalence of HBV exposure and coinfection among HCV patients in the U.S. Based on these findings, all patients with HCV exposure should be tested for HBV. Additionally, this study identified risk factors more frequent in patients with HBV coinfection than HCV monoinfection. These predictors of HBV coinfection can be used to target screening and prevention programs to those individuals who may be at greatest risk for coinfection.