Self-reported versus actigraphy-assessed sleep duration in the ELSA-Brasil study: analysis of the short/long sleep duration reclassification.

Abstract

This study was aimed to determine the magnitude and predictors of self-reported short/long sleep duration (SDUR) reclassifications using objective measurements. Adult participants from the ELSA-Brasil study performed self-reported SDUR, 7-day wrist actigraphy, and a portable sleep study. We explored two strategies of defining self-reported SDUR reclassification: (1) short and long SDUR defined by <6 and ≥8h, respectively; (2) reclassification using a large spectrum of SDUR categories (<5, 5-6, 7-8, 8-9, and >9 h). Data from 2036 participants were used in the final analysis (43% males; age: 49±8 years). Self-reported SDUR were poorly correlated (r=0.263) and presented a low agreement with actigraphy-based total sleep time. 58% of participants who self-reported short SDUR were reclassified into the reference (6-7.99 h) or long SDUR groups using actigraphy data. 88% of participants that self-reported long SDUR were reclassified into the reference and short SDUR. The variables independently associated with higher likelihood of self-reported short SDUR reclassification included insomnia (3.5-fold), female (2.5-fold), higher sleep efficiency (1.35-fold), lowest O2 saturation (1.07-fold), higher wake after sleep onset (1.08-fold), and the higher number of awakening (1.05-fold). The presence of hypertension was associated with a 3.4-fold higher chance of self-reported long SDUR reclassification. Analysis of five self-reported SDUR categories revealed that the more extreme is the SDUR, the greater the self-reported SDUR reclassification. In adults, we observed a significant rate of short/long SDUR reclassifications when comparing self-reported with objective data. These results underscore the need to reappraise subjective data use for future investigations addressing SDUR.