Re-estimating prevalence of hepatitis B virus immunity among adults in the United States: self-reported vaccination, immunologic markers, and bias correction

Background: Two instruments used to measure adult hepatitis B vaccination coverage in the United States are self-report and antibody to hepatitis B surface antigen (anti-HBs). Estimates based on either of these measures are subject to misclassification when used to determine immunity to hepatitis B. This study presents misclassification-corrected estimates of hepatitis B immune prevalence in the US and compares them to self-report- and antibody-based estimates. Methods: We used cross-sectional data from the 2015-2016 NHANES cycle on 5,151 adults in the US age 18 and older. Existing literature on long-term immunity after vaccination informed anti-HBs sensitivity as a measure of immunity. Our model incorporated literature-based distributions for sensitivity and specificity using a Bayesian approach to correct for misclassification of true immune status by anti-HBs. Results: After correcting for misclassification, overall adult immune prevalence was estimated at 31.0% (95% credible interval, 27.9% to 34.1%). Anti-HBs prevalence was 6.4 (3.9 to 8.8) and self-report prevalence 2.6 (-0.6 to 5.8) percentage points lower than overall immune prevalence. Among Asian Americans, anti-HBs and self-report underestimated immune prevalence by 15.8% (11.5% to 20.9%) and 25.1% (17.3% to 33.2%), respectively. Among 19 to 25-year-olds, anti-HBs and self-report underestimated immune prevalence by 26.5% (20.7% to 32.5%) and 21.0% (12.6% to 28.9%). Conclusions: Both self-reported vaccination and antibody-based measures underestimate hepatitis B immunity among adults. This underestimation was especially large among younger adults and Asian Americans. The consequences of treating these surrogates as unbiased measures of vaccination or immunity may only increase as more vaccinated children age into adulthood. Keywords: hepatitis B, hepatitis B vaccine, hepatitis B antibodies, patient reported outcome measures, outcome measurement errors, immunological memory