0551 Longitudinal Measurement Invariance of the Insomnia Severity Index in Veterans with Sleep Apnea

Abstract

Abstract Introduction The Insomnia Severity Index is a commonly used instrument to assess the presence of insomnia symptoms as well as an outcome measure following an intervention. Longitudinal measurement invariance is a necessary property of an assessment instrument when it is repeated over time. The validity of conclusions regarding change in the construct ‘insomnia severity’ depend on scale equivalence at each measurement timepoint. Assessment of measurement invariance of the ISI in sleep apnea patients has never been performed. Methods Veterans with sleep apnea (n=654; AHI=36±28; 93% male; age=52±12; BMI=33±6) completed the ISI on the night of their overnight PSG and again when they picked up their PAP device. Invariance was determined by imposing a series of more restrictive equivalence constraints on a 2-factor model of the ISI. The series of constraints tested for configural, weak, strong and strict invariance. Invariance testing was modeled with exploratory structural equation modeling in Mplus (v. 7.0). Results The 2-factor model that emerged from the analysis showed items relating to nighttime symptoms loading on factor 1 and daytime symptoms loading on factor 2. The sleep ‘satisfaction’ item, however, had weak but similar loadings on both factors. The increasingly restrictive constraints imposed on the model revealed no decrement in model fit (RMSEA=.039 to.043; CFI=.987 to .980; TLI=.981-.977; SRMR=.027-.041). Conclusion The ISI met strict criteria for longitudinal measurement invariance demonstrating that it is a valid instrument to be used in repeated measures study designs of insomnia in sleep apnea patients. Change over time on the ISI is not due to the changing measurement characteristics of the ISI but to true changes in the ‘insomnia severity’ construct. Support None