Edwin Lok Yan Wong, Harvey Neal, Christopher Jones, Andrew Surtees
Fifteen included studies with 572 participants were evaluated with a quality assessment framework, and pooled for five separate meta-analyses on discrepancies and/or correlations between objective and subjective sleep parameters. Overall, significant standardised mean differences were found between sleep measured by primarily actigraphy-based methods versus sleep diaries and questionnaires. Total sleep time showed modest concordance, as subjective values were higher but positively associated with objective values. There was poor concordance for sleep onset latency (longer subjective estimates), sleep efficiency (higher subjective estimates) and wake after sleep onset (shorter subjective estimates). No significant differences were detected among bipolar disorder presentations or diagnoses granted that subgroup analyses were underpowered and inconclusive.
BACKGROUND: Poor sleep is an important symptom linked with the onset, deterioration and relapse of bipolar disorder, and accurate measurement of sleep is important for clinical research and management. This meta-analysis aimed to review the concordance between objective and subjective sleep measures for the bipolar disorder population.
METHOD: Four databases were searched with terms for sleep variables, bipolar disorder alongside objective and subjective sleep measures. Records were screened using pre-determined inclusion and exclusion criteria to identify research that directly examined sleep concordance or reported data for statistical comparison.
RESULTS: Fifteen included studies with 572 participants were evaluated with a quality assessment framework, and pooled for five separate meta-analyses on discrepancies and/or correlations between objective and subjective sleep parameters. Overall, significant standardised mean differences were found between sleep measured by primarily actigraphy-based methods versus sleep diaries and questionnaires. Total sleep time showed modest concordance, as subjective values were higher but positively associated with objective values. There was poor concordance for sleep onset latency (longer subjective estimates), sleep efficiency (higher subjective estimates) and wake after sleep onset (shorter subjective estimates). No significant differences were detected among bipolar disorder presentations or diagnoses granted that subgroup analyses were underpowered and inconclusive.
DISCUSSION: Main implications included recommendations on utilising objective and subjective sleep measures for bipolar disorder individuals across clinical and research contexts (e.g., questionnaires for rapid screening; actigraphy for follow-up). Limitations of available evidence were high heterogeneity, small samples and inconsistent/inadequate statistical reporting. Low numbers of included studies restricted further analyses for confounding elements.