Incorporating uncertainty in Health Benefits Package Design – Empowering local decision-makers in putting limited healthcare resources to their best use

Context: There has been an increasing emphasis on developing rational evidence-based approaches to health benefits package (HBP) design to ensure scarce resources are prioritised to generate the most population health. The evidence base informing expected costs and benefits of interventions is, however, inevitably uncertain due to limitations in data quality, quantity and relevance to the context of interest. There may therefore be a risk that including certain interventions in a package could reduce overall population health, once the opportunity costs of using limited health care resources are considered.  In this context, additional research could valuably contribute to improve allocations choices.

Method: We have developed a framework, implemented by a freely available stand-alone tool, for supporting HBP design by identifying priorities for research and conditions when research should be prioritized over immediate implementation of interventions. Since the evidence base informing HBP is essentially constituted by secondary cost-effectiveness data,  the tool was specifically developed to quantify uncertainty around investment decisions and the potential value from research using only reported information on the expected variation around costs and outcomes. This information may be presented under various mediums such as tornado plots, cost-benefits scatterplots, incremental cost effectiveness ratio histograms, cost-effectiveness acceptability curves or confidence intervals. To illustrate how the framework can inform data-collection priorities, and when these may take precedence over service provision, we applied the tool to the evidence base for the Malawi Health Sector Strategic Plan for 2017-2022.

Results: We found there was value in undertaking research for a third of the interventions considered. For one intervention, the health gains to be generated by improved evidence were large enough to potentially justify delaying implementation for up to 4 years.