Validity of two retrospective questionnaire versions of the Consensus Sleep Diary: the whole week and split week Self-Assessment of Sleep Surveys

Objective/Background: Prospective, daily sleep diaries are the gold standard for assessing subjective sleep but are not always feasible for cross-sectional or epidemiological studies. The current study examined psychometric properties of two retrospective questionnaire versions of the Consensus Sleep Diary. Participants/Methods: College students (N = 131, mean age = 19.39 ± 1.65; 73% female) completed seven days of prospective sleep diaries then were randomly assigned to complete either the Self-Assessment of Sleep Survey (SASS), which assessed past week sleep (n = 70), or the SASS-Split (SASS-Y), which assessed weekday/weekend sleep separately (n = 61). Participants also completed psychosocial/sleep questionnaires including the Pittsburgh Sleep Quality Index (PSQI). Sleep parameters derived from SASS, SASS-Y, PSQI, and sleep diaries were assessed via Bland Altman plots, limits of agreement, mean differences, and correlations. Results: SASS-Y demonstrated stronger correlations with prospective sleep diaries and slightly less biased estimates (r = 0.51 to 0.85, α = −0.43 to 1.70) compared to SASS (r = 0.29 to 0.84, α = −1.63 to 2.33) for terminal wakefulness (TWAK), sleep onset latency (SOL), sleep efficiency (SE), and quality (QUAL). SASS resulted in slightly less bias for total sleep time (TST) and wake after sleep onset (WASO) (α = −0.65 and 0.93, respectively) compared to SASS-Y (α = 14.90 and 1.05, respectively). SASS and SASS-Y demonstrated greater convergence with sleep diary than PSQI. Conclusions: Results demonstrated good psychometric properties for the SASS and SASS-Y. When prospective sleep diaries are not feasible, the SASS and SASS-Y are acceptable substitutes to retrospectively estimate sleep parameters. Retrospective estimation of sleep parameters separately for weekdays/weekends may offer advantages compared to whole week estimation.