Presentation (CTI)

OBJECTIVES: Incremental Cost Effectiveness Ratios (ICERs) are used in Cost-Effectiveness Analyses (CEAs). With two interventions, the calculation is straightforward, but is less obvious with additional interventions. We sought to determine the frequency and type of ICER miscalculation/misinterpretation in published CEAs comparing more than two alternatives.
METHODS: A PubMed search identified CEAs published in 2017. Article titles and abstracts were screened by at least two reviewers and excluded if they compared only two interventions, were not a CEA, or lacked an abstract. Remaining articles were reviewed in full-text by two reviewers using a sequential protocol to identify errors in ICER calculation/interpretation. Errors assessed, in order, were: 1) calculating average cost-effectiveness ratios (ACERs), 2) calculating ICERs comparing all alternatives to a common one, 3) calculating ICERs combining multiple disease states or other overlapping populations, 4) failing to provide a willingness-to-pay (WTP) value in deciding cost-effectiveness, 5) misapplying the WTP, and 6) making other ICER calculation/interpretation errors. Only the first error identified in this sequence was recorded. If none were found, the article was coded as correct.
RESULTS: Our search identified 815 publications. After exclusions, 132 articles were reviewed and 62% (82) contained an error. Of these 132 articles, 35.6% compared all interventions to a single comparator; 15.9% used ACERs. Failing to include, or misapplying, a WTP were less common (3.8% each) as were ICER calculations using overlapping populations or other errors in mathematical or dominance calculations (1.5% each).
CONCLUSIONS: Publications using incorrect economic methods may have implications for health system efficiencies. Our findings likely underestimate total errors in the CEA literature because our protocol focused exclusively on ICER calculation and interpretation. The high frequency of error may cast doubt on the usefulness of CEAs for healthcare decision-making and should be addressed so that they have the requisite integrity to inform decisions.