Abstract
INTRODUCTION: Prioritising health benefits packages (HBPs) that specify which health services are covered under insurance is sometimes done in disease-specific clusters. Cancer is a good candidate for this, given its high cost and rising disease burden, particularly in low- and middle-income countries.The Government of Rwanda assessed 49 cancers against nine criteria to inform the design of its HBP. Each cancer had a basic, core and enhanced package of services, and one preventive intervention was assessed, totalling 148 interventions. This paper focuses on the results of one criterion: cost-effectiveness. The objectives were to specify which cost-effectiveness methods were selected and why; to assess the cost-effectiveness of 148 cancer interventions; and to recommend how to strengthen the global cost-effectiveness evidence base. METHODS: Methods were selected using an adaptive health technology assessment approach, by considering the trade-offs between available time, data and capacity. The assessment undertook a review of the Tufts cost-effectiveness assessment (CEA) registry and filled evidence gaps with structured expert elicitation. Analysts summarised lessons learnt to recommend improvements to the global cost-effectiveness evidence base. RESULTS: Of the 148 cost-effectiveness ratios (CERs) sought, 39 were from the Tufts registry and 83 were expert elicited. Limited availability of CERs from the literature resulted in a disproportionate number of CERs being elicited from experts. Analysts recommend better reporting and improved consistency in the extraction of CEAs to support HBP design. CONCLUSION: This is the first study to assess the cost-effectiveness of many cancers simultaneously for HBP design. It highlights the strengths and limitations of existing evidence and demonstrates the feasibility of combining rapid review with expert elicitation to obtain replicable CER estimates. These cost-effectiveness findings were used to prioritise a package of cancer services for Rwanda alongside several other criteria (reported separately).