POSA262 The Use of Akaike Information Criterion and Bayesian Information Criterion Rules of Thumb in Nice Oncology Appraisals: A Targeted Review

Abstract

NICE Decision Support Unit’s Technical Support Document 14 (2011) recommends using Akaike information criterion (AIC) and Bayesian information criterion (BIC) to assess the statistical fit of parametric survival models to observed data for use in economic models. However, the document includes no clear guidance on interpretation of AIC and BIC values aside from the lowest AIC and BIC being the best statistical fit, despite “rules of thumb” for categorising models according to relative statistic fit versus the distribution with the lowest AIC and BIC being available in the literature.