One or two serological assay testing strategy for diagnosis of HBV and HCV infection? The use of predictive modelling

Initial serological testing for chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infection is conducted using either rapid diagnostic tests (RDT) or laboratory-based enzyme immunoassays (EIA)s for detection of hepatitis B surface antigen (HBsAg) or antibodies to HCV (anti-HCV), typically on serum or plasma specimens and, for certain RDTs, capillary whole blood. WHO recommends the use of standardized testing strategies – defined as a sequence of one or more assays to maximize testing accuracy while simplifying the testing process and ideally minimizing cost. Our objective was to examine the diagnostic outcomes of a one- versus two-assay serological testing strategy. These data were used to inform recommendations in the 2017 WHO Guidelines on hepatitis B and C testing.

Few published studies have compared diagnostic outcomes for one-assay versus two-assay serological testing strategies for HBsAg and anti-HCV. Therefore, the principles of Bayesian statistics were used to conduct a modelling exercise to examine the outcomes of a one-assay versus two-assay testing strategy when applied to a hypothetical population of 10,000 individuals. The resulting model examined the diagnostic outcomes (true and false positive diagnoses; true and false negative diagnoses; positive and negative predictive values as a function of prevalence; and total tests required) for both one-assay and two-assay testing strategies. The performance characteristics assumed for assays used within the testing strategies were informed by WHO prequalification assessment findings and systematic reviews for diagnostic accuracy studies. Each of the presumptive testing strategies (one-assay or two-assay) was modelled at varying prevalences of HBsAg (10%, 2% and 0.4%) and of anti-HCV (40%, 10%, 2% and 0.4%), aimed at representing the range of testing populations typically encountered in WHO Member States. When the two-assay testing strategy was considered, the model assumed the independence of the two assays.

Modeling demonstrated that applying a single assay (HBsAg or anti-HCV), even with high specificity (99%), may result in considerable numbers of false positive diagnoses and low positive predictive values (PPV), particularly in lower prevalence settings. Even at very low prevalences shifting to a two-assay testing strategy would result in a PPV approaching 1.0. When test sensitivity is high (>99%) false negative reactions are rare at all but the highest prevalences; but a two-test strategy might yield more false negative diagnoses. The order in which the tests are used has no impact on the overall accuracy of a two-assay strategy though it may impact the total number of tests needed to complete the diagnostic strategy, incurring added cost and complexity. HBsAg assays may have a low sensitivity (<90%), and result in large numbers of false negative diagnoses, particularly in high prevalence settings, which would be exacerbated in the two-assay testing strategy. In contrast, most anti-HCV assays have high sensitivity and lead to fewer false negative results, both in the one-assay and two-assay testing strategies. At prevalences ≤2% the number of tests needed using a second assay was nearly always small, at <300 per 10,000 individuals tested, making sustainability of a second assay uncertain in such a setting.

A key public health objective of an effective testing strategy is to identify all individuals who would benefit from treatment. Therefore, a strategy that prioritizes a high NPV (minimal false negatives) may be acceptable even if the PPV is suboptimal (some false positives) as the implementation of such a public health programme must also take account of other factors such as costs, feasibility, impact on testing uptake and linkage to care, and consequences of a false-positive test. This rationale informed the development of the WHO Viral Hepatitis Testing Guidelines, with a conditional recommendation for a one-assay serological testing strategy in most testing settings and populations (≥0.4% prevalence in population tested). A one-test strategy results in few failures to diagnose infection and, although it is associated under most assumptions with a sub-optimal PPV, benefits include greater simplicity, easier implementation, lower costs and better feasibility, uptake and linkage to care. Furthermore, prior to antiviral therapy all those diagnosed either HBsAg or anti-HCV positive will require confirmation of viræmia, preventing unnecessary treatment of those who may be false positive on serology. For HBsAg, in low-prevalence settings (≤0.4%), a second recommendation was made to consider a two-assay testing strategy, using a confirmatory neutralization step or a second different HBsAg assay.