A meta-analysis of diagnostic accuracy of three screening tools for insomnia

Background Insomnia is a highly prevalent health complaint in the modern societies; however, insomnia remains under-diagnosed and under-treated. Although screening tools, including the Insomnia Severity Index (ISI), Athens Insomnia Scale (AIS), and Pittsburg Sleep Quality Index (PSQI), are widely used for assessing the risk of insomnia, the diagnostic properties have yet to be summarized in a systematic manner. Objectives To estimate and to compare the diagnostic accuracy of the ISI, AIS, and PSQI for insomnia screening. Data sources We systematically searched EMBASE, PubMed, PsycINFO, CINAHL and Chinese Electronic Periodic Services for data from their inception to May 20, 2015. Data selection Original articles that had assessed the sensitivity and specificity of the ISI, AIS, or PSQI against a reference standard in adult participants (age > 18) were included. Results A total of 19 studies comprising 4693 participants were included. The pooled sensitivity for the ISI, AIS, and PSQI was 88% (95% confidence interval [CI] = 0.79 to 0.93), 91% (0.87 to 0.93), and 94% (0.86 to 0.98), respectively. The pooled specificity was 85% (0.68 to 0.94), 87% (0.68 to 0.95), and 76% (0.64 to 0.85); and the pooled DORs was 41.93 (8.77 to 200.33), 67.7 (23.4 to 196.1), and 53 (15.5 to 186.2), respectively. The summary estimates did not differ significantly among the ISI, AIS and PSQI (all P > 0.05). Conclusions The current evidence indicates that the ISI, AIS, and PSQI yield comparable diagnostic properties for insomnia screening.