Characterising the potential for recall bias in anchor-based MCID calculation of patient-reported outcome measures for chronic rhinosinusitis.

Abstract

Anchor-based methods to calculate the minimal clinically important difference (MCID) of a patient-reported outcome measure (PROM) may suffer from recall bias. This has never been investigated for otolaryngic PROMs. We sought to identify evidence of recall bias in calculation of MCIDs of PROMs for patients with chronic rhinosinusitis (CRS). Retrospective analysis of data from two previous studies calculating the MCID of the 22-item Sinonasal Outcome Test (SNOT-22) and 5-dimensonal EuroQol questionnaire (EQ-5D) in CRS patients. Tertiary rhinology clinic. Adults with CRS. SNOT-22 score, and EQ-5D visual analog scale scores (EQ-5D VAS) and health utility values (EQ-5D HUV) before and after medical treatment for CRS. After treatment, participants were asked to rate the change in sinonasal symptoms and general health (the anchor question) as "Much worse," "A little worse," "About the same," "A little better" or "Much better." Participants' responses to the anchor question were checked for association with post-treatment and pre-treatment scores using ordinal regression. On univariate association, post-treatment SNOT-22 and EQ-5D scores were associated with respective participants' anchor question responses (P<.001 in all cases). Only pre-treatment SNOT-22 score was associated with anchor question responses (P=.017) on univariate association, in contrast to pre-treatment EQ-5D scores. Pre-treatment EQ-5D scores only associated with anchor question responses when controlling for post-treatment scores. The anchor-based MCIDs of the SNOT-22, which reflects disease-specific QOL, and the EQ-5D, which reflects general health-related QOL, appear to be largely free of recall bias.