News|Articles|July 28, 2026

Sleep Variability Differs Across Psychosis Spectrum

Listen

0:00 / 0:00

Key Takeaways

Elevated variability in time in bed, wake after sleep onset, and sleep efficiency characterized both clinical high risk and schizophrenia spectrum cohorts versus controls.
Stage-specific signatures emerged for total sleep time, remaining control-like in clinical high risk yet markedly more heterogeneous in schizophrenia spectrum disorders.
Exploratory moderation suggested antipsychotic use may amplify total sleep time variability in clinical high risk, although evidence was limited by few contributing studies.
Findings support actigraphy-derived variability as a candidate digital biomarker for stratified interventions targeting insomnia vs hypersomnia phenotypes across psychosis trajectories.
Case-control designs, inconsistent race/ethnicity reporting, heterogeneous CHR-P instruments, and aggregate-level analyses limit causal inference and subgroup confirmation without individual-participant data.

This eta-analysis shows sleep variability—not averages—signals psychosis risk, with stage-specific patterns to guide digital biomarkers and care.

Interindividual variability in sleep patterns, rather than average sleep differences alone, distinguishes individuals across the psychosis spectrum from healthy controls, with distinct variability profiles emerging at different illness stages, according to a systematic review and meta-analysis published in JAMA Network Open.¹

Study Explores Variability, Not Just Averages

Sleep abnormalities are already well documented across the psychosis spectrum: a prior meta-analysis found that disrupted sleep is present from clinical high risk through chronic illness and that altered sleep often precedes a psychotic episode rather than simply co-occurring with it.² Most actigraphy research in this space, however, has focused on group-level mean differences, an approach that can obscure clinically meaningful heterogeneity within diagnostic groups, the current study's authors noted.¹ A cohort with equal numbers of very short and very long sleepers could produce a mean sleep duration indistinguishable from healthy controls, masking 2 potentially distinct clinical subpopulations.

To address this gap, researchers pooled data from 18 case-control studies encompassing 1358 participants: 202 individuals at clinical high risk for psychosis (CHR-P), 574 with schizophrenia spectrum disorders (SSDs), and 582 healthy controls. Studies were identified through PubMed, Embase, MEDLINE, and several trial registries, with searches updated through April 2026. The team calculated the natural logarithm of the variability ratio (lnVR) and coefficient of variation ratio for total sleep time (TST) and secondary actigraphy measures, including time in bed (TIB), sleep latency, wake after sleep onset (WASO), sleep efficiency, and number of awakenings.¹

Findings Point to Shared and Distinct Patterns

Both the CHR-P and SSD groups showed significantly greater interindividual variability than healthy controls in TIB, WASO, and sleep efficiency. For example, WASO variability was elevated in CHR-P (lnVR, 0.45; 95% CI, 0.07-0.83) and in SSDs (lnVR, 0.52; 95% CI, 0.15-0.89).

TST variability, however, differed by illness stage. Participants at CHR-P did not differ significantly from controls (lnVR, 0.16; 95% CI, −0.05 to 0.37), while those with SSDs showed markedly greater variability (lnVR, 0.46; 95% CI, 0.28-0.64). The authors suggested this may indicate that early-stage disruptions primarily affect sleep opportunity (TIB) without a corresponding change in actual sleep duration, with TST variability emerging later as illness progresses.

Moderator analyses found that antipsychotic use was associated with greater TST variability in the CHR-P group, although the authors cautioned this exploratory finding was based on only 5 studies. No moderation effects for antipsychotic use, age, or sex were found in the SSD group. Sensitivity analyses excluding studies at high risk of bias preserved the direction and significance of all primary findings, and results were consistent across sampling epochs.

Implications for Stratified Care

The authors framed the findings as support for further investigation of sleep as a candidate digital biomarker to inform stratified care in psychosis, potentially helping identify subpopulations—such as those with hypersomnia vs insomnia patterns—who might benefit from targeted interventions. They pointed to existing work, including a feasibility trial of cognitive behavioral therapy for sleep problems in at-risk youth that also reduced depression, anxiety, and psychotic-like symptoms, as an example of a stabilizing intervention that could be tested against these variability signatures.

Limitations

The authors acknowledged several constraints. The case-control design precluded causal inference about whether sleep variability precedes clinical changes, and race and ethnicity data were inconsistently reported across primary studies. Heterogeneous diagnostic instruments across the CHR-P literature limited harmonization, and moderator analyses were underpowered given the small number of contributing studies. Because the analysis relied on aggregated study-level statistics, the authors noted that individual-participant data will be needed to confirm whether the observed variability reflects distinct clinical subgroups or a single broader distribution.

References

Aronica R, Torous J, Minichino A, Mills M, McGuire P, Oliver D. Interindividual sleep variability across the psychosis spectrum: a systematic review and meta-analysis. JAMA Netw Open. 2026;9(7):e2624358. doi:10.1001/jamanetworkopen.2026.24358
Bagautdinova J, Mayeli A, Wilson JD, et al. Sleep abnormalities in different clinical stages of psychosis: a systematic review and meta-analysis. JAMA Psychiatry. 2023;80(3):202-210. doi:10.1001/jamapsychiatry.2022.4599