Sponsorship Bias in Base-Case Values and Uncertainty Bounds of Health Economic Evaluations? A Systematic Review of Herpes Zoster Vaccination

BACKGROUND: New health technologies are more likely adopted when they have lower incremental cost-effectiveness ratios (ICERs) and/or when their ICER is presented with more certainty. Industry-funded (IF) health economic evaluations use often more favorable base-case values, leading to more favorable conclusions. PURPOSE: To study whether IF health economic evaluations of varicella-zoster virus vaccination in the elderly use more favorable base-case values and account for less uncertainty than non-industry-funded (NIF) evaluations. Methods: DATA SOURCE: PubMed. Data extracted: funding source; incremental cost per quality-adjusted life year (QALY) gained; vaccine price; study quality score; base-case values, uncertainty ranges, and data sources for influential parameters: duration of vaccine protection, utility loss due to herpes zoster (HZ) disease, percentage of HZ patients developing postherpetic neuralgia (PHN), and duration of PHN. DATA SYNTHESIS: qualitative comparisons; Fisher exact test for differences in study quality score and 1-sided Mann-Whitney U tests for differences in base-case values and uncertainty ranges. RESULTS: Despite using the same data sources, IF studies ( n = 10) assume a longer duration of vaccine protection ( U = 56, P = 0.03), have a higher percentage of HZ patients developing PHN ( U = 22/33, P = 0.02/0.03 for ages 60-64/65-69), and tend to use higher HZ utility loss than NIF studies ( n = 11) for their baseline. IF studies show lower ICERs given similar or even higher vaccine prices than NIF studies, consider less uncertainty around the duration of vaccine protection ( U = 8, P < 0.001), and tend to use less uncertainty around the duration of PHN. Yet their quality has been rated equally well, using current standard quality rating tools. CONCLUSION: Researchers and decision makers should be aware of potential sponsorship bias in health economic evaluations, especially in the way source data are used to specify base-case values and uncertainty ranges.