Can cost-effectiveness results be combined into a coherent league table? Case study from one high-income country.
Wilson N., Davies A., Brewer N., Nghiem N., Cobiac L., Blakely T.
BACKGROUND: Doubts exist around the value of compiling league tables for cost-effectiveness results for health interventions, primarily due to methods differences. We aimed to determine if a reasonably coherent league table could be compiled using published studies for one high-income country: New Zealand (NZ). METHODS: Literature searches were conducted to identify NZ-relevant studies published in the peer-reviewed journal literature between 1 January 2010 and 8 October 2017. Only studies with the following metrics were included: cost per quality-adjusted life-year or disability-adjusted life-year or life-year (QALY/DALY/LY). Key study features were abstracted and a summary league table produced which classified the studies in terms of cost-effectiveness. RESULTS: A total of 21 cost-effectiveness studies which met the inclusion criteria were identified. There were some large methodological differences between the studies, particularly in the time horizon (1 year to lifetime) but also discount rates (range 0 to 10%). Nevertheless, we were able to group the incremental cost-effectiveness ratios (ICERs) into general categories of being reported as cost-saving (19%), cost-effective (71%), and not cost-effective (10%). The median ICER (adjusted to 2017 NZ$) was ~ $5000 per QALY/DALY/LY (~US$3500). However, for some interventions, there is high uncertainty around the intervention effectiveness and declining adherence over time. CONCLUSIONS: It seemed possible to produce a reasonably coherent league table for the ICER values from different studies (within broad groupings) in this high-income country. Most interventions were cost-effective and a fifth were cost-saving. Nevertheless, study methodologies did vary widely and researchers need to pay more attention to using standardised methods that allow their results to be included in future league tables.