Projected Economic Impact Of Current And Future Chronic Hepatitis C Treatment Regimens And Recent Screening Guidelines In The United States

Abstract

published NMAs have accounted for this variable when assessing the relative efficacy of CHB interventions. We undertook baseline-adjusted and -unadjusted analyses using the same dataset to explore the impact of BVL on UVL achievement. METHODS: The NMA was undertaken in WinBUGs using standard fixed/random effects approaches. A systematic literature review was the source of efficacy data for eight interventions. Individual patient data from an entecavir RCT was used to quantify the impact of different baseline characteristics on the likelihood of achieving UVL at one year. Study level mean baseline values from all identified papers and the results from the abovementioned analysis were used to inform prior distributions in the adjusted analysis. Results were presented as relative risks (RRs) at one year and 95% credible intervals. Entecavir was the reference treatment. Sensitivity analyses surrounding the inclusion/exclusion of individual studies were performed. RESULTS: Entecavir and tenofovir studies included heterogeneous patient populations in terms of BVL (mean entecavir value 9.29 log10copies/ml, tenofovir value 8.65 log10copies/ml). After adjusting the UVL for baseline viral load using an informative prior, the RR for entecavir versus tenofovir went from statistically significant to non-significant [unadjusted: RR 1.43 (1.30 to 1.54), adjusted: RR 1.21 (0.48 to 1.51)]. Relative efficacy results for all other interventions were similar across adjusted and unadjusted analyses. Adjusted results were robust to the inclusion/exclusion of additional studies. Adjusted results for all interventions were consistent with clinical trial data. CONCLUSIONS: Adjusted and unadjusted NMA, using the same dataset, demonstrated the importance of adjusting for BVL when assessing the relative efficacy of CHB interventions. Failing to account for differences in BVL across studies may result in flawed relative efficacy estimates.