OBJECTIVES

Casual observation suggested weak adherence to methods standards in CEAs, including calculation of ICERs, particularly when more than 2 alternative treatments were considered (e.g. comparing all alternatives to a single common one). This research sought to formally assess adherence to methods standards for several commonly prescribed attributes in CEAs. This study investigates CEAs published in 2017 with respect to their conforming to methods quality standards on ICER calculations, use of discounting, a lifetime time horizon, probabilistic sensitivity analysis (PSA) and a quality adjustment for outcomes.

METHODS

We performed a PubMed search for 2017-published CEAs and, after abstract review, extracted those evaluating more than 2 alternatives. These 119 CEAs were assessed in five dimensions: (1) the ICER was appropriately calculated/interpreted; (2) discounting was applied to both costs and effects; (3) PSA was done, (4) a lifetime time horizon was used, and (5) a quality-adjusted outcome was used. Assessment was done for each attribute with outcomes as YES/NO/UNCLEAR. Each CEA was reviewed by at least two of the study authors. In cases where the designation was UNCLEAR, the field was either marked as such or the CEA was put forward to a third study author attempt resolution.

RESULTS

For appropriateness of ICERs assessment showed 48/61/10 as YES/NO/UNCLEAR. For discounting: 76/33/10. PSA results were 80/32/7. Lifetime horizon results were: 39/67/13 and quality adjustment applied to outcomes results showed responses of 81/38/0.

CONCLUSIONS

These findings suggest significant room for improvement in analysis and reporting of approaches used in CEAs. In these important attributes a lack of clarity (UNCLEAR) may, arguably, be considered equivalent to a NO. Additional work will attempt to determine whether cases of NO/UNCLEAR responses are justifiable (e.g. a less than 1 year time horizon was appropriate, perhaps obviating the need for discounting).