Interpreting score differences in the Insomnia Severity Index: using health-related outcomes to define the minimally important difference

Abstract Objective: To estimate the minimally important difference (MID) for the Insomnia Severity Index (ISI) by examining the association of score differences of the ISI with health-related outcomes including health-related quality of life, productivity, and fatigue. Methods: Data came from a randomized, placebo-controlled clinical trial evaluating the long-term efficacy of eszopiclone for primary insomnia (N = 828). Logistic regression models were used to characterize the relationship between ISI change scores (from baseline to 6 months post-treatment) and outcomes/anchors from the SF-36 Health Survey, Work Limitations Questionnaire (WLQ), and Fatigue Severity Scale (FSS). Odds ratios were derived from the regression coefficients to calculate the probability of a given outcome being associated with different ISI change scores. Convergence between anchor- and distribution-based estimates was assessed. Results: Higher ISI scores (indicating more severe insomnia) were significantly associated with higher probabilities of negative outcome in all models. Individuals with a 6-point score reduction in ISI scores (which corresponded to 1½ standard deviations) were 48% less likely to report 'feeling worn out' (SF-36) at 6 months, 46% less likely to be 'unable to think clearly' (WLQ), and 52% less likely to report 'feeling fatigued' (FSS). Similar results were found across a broad spectrum of all selected anchors. Conclusions: Based on results of the study, a 6-point reduction is recommended to represent a clinically meaningful improvement in individuals with primary insomnia. Research on generalizability of the recommended MID in this study to other patient populations and other type of treatment interventions is needed.