Irregular sleep linked to lower fluid intelligence in UK Biobank cohort

People whose sleep timing, duration and movement patterns varied more from night to night were more likely to score lower on a subsequent test of fluid intelligence in an exploratory UK Biobank study. The association was observed among 6,081 participants who wore wrist accelerometers and later completed two cognitive assessments. Researchers found no evidence that the variability measures predicted cognitive decline or performed better than the established Sleep Regularity Index.

The study examined four dimensions of sleep variability calculated from within-person standard deviations: sleep-onset time, waking time, sleep duration and the number of movement-defined sleep interruptions. These measures were combined into an equally weighted Composite Variability Score, abbreviated CVS. The investigators also evaluated SRI1, an existing measure of sleep regularity. Because the analysis was observational, the results demonstrate associations rather than showing that irregular sleep caused lower cognitive performance.

At the first subsequent cognitive assessment, all four individual variability measures and the composite score were associated with lower fluid-intelligence scores. For CVS, the estimated association was −0.1444 points per exposure-standard-deviation increase, with a 95% confidence interval from −0.1970 to −0.0918. After correction for six exposure tests across two outcomes using the Benjamini–Hochberg procedure, the result remained statistically significant at q < 0.0001. The researchers used type 3 heteroscedasticity-consistent, or HC3, inference to account for heteroscedasticity in the statistical models.

The models accounted for estimated age at accelerometry, sex, within-country regional socioeconomic scores, country, body mass index, smoking, history of alcohol consumption, average sleep duration and assessment timing. In separate longitudinal analyses, the researchers additionally adjusted for earlier performance in the same cognitive domain. Under this stricter specification, waking-time variability, sleep-duration variability and CVS continued to show small negative associations with later fluid intelligence, with estimates of −0.0591, −0.0483 and −0.0506, respectively. However, their 12-test false-discovery-rate q-values ranged from 0.0610 to 0.0704 and did not cross the prespecified corrected significance threshold.

Sleep variability also added little information beyond SRI1: the increase in explained variance was no greater than 0.00079, and that incremental contribution did not meet the study’s multiplicity-corrected threshold. Estimates became weaker after selection weighting, an indication that the findings were sensitive to factors affecting which participants were retained in the analysis. Together, the first-assessment results consistently linked greater sleep variability with lower fluid intelligence, while the longitudinal findings were smaller, directionally similar and less robust. The study therefore supports variability as a possible marker of cognitive performance, but it does not establish that sleep irregularity predicts future cognitive decline or that measuring it offers a clinical advantage over the Sleep Regularity Index.

Subject of Research: Associations between night-to-night sleep variability and subsequent fluid-intelligence performance in a UK Biobank cohort

Article Title: Night-to-night sleep variability and subsequent cognitive performance: a UK Biobank cohort study

Article References: Cai, W.-M., Rui, D., Yang, Y.-Y., Zhang, M.-R., Xue, X., Ma, Y., Gao, Y.-H., & Liu, L. (2026). Night-to-night sleep variability and subsequent cognitive performance: a UK Biobank cohort study. BMC Public Health. https://doi.org/10.1186/s12889-026-29996-1

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29996-1

Keywords: sleep variability, fluid intelligence, UK Biobank, cognitive performance, wrist accelerometry, sleep regularity, cohort study, sleep duration, Night-to-night, sleep, variability, subsequent