TOPLINE

A non-invasive predictive model combining self-reported sleep measures, actigraphy, and polysomnography (PSG) was developed to effectively differentiate unipolar from bipolar depression during a major depressive episode.

METHODOLOGY

• Researchers conducted a comparative study of 159 patients with major depressive episodes, including 43 with bipolar depression and 116 with unipolar depression, recruited from a single routine care medical centre in Paris between 2022 and 2025.
• Participants underwent comprehensive assessments including clinical evaluation, validated questionnaires (Pittsburgh Sleep Quality Index [PSQI], Insomnia Severity Index [ISI], Hypersomnia Severity Index, Epworth Sleepiness Scale, morningness-eveningness questionnaire, and seven-item Generalized Anxiety Disorder scale), actigraphy for at least 14 days (performed in 42 patients with bipolar and 93 with unipolar depression), and PSG for one night (performed in 20 patients with bipolar and 44 with unipolar depression).
• Actigraphy was used to assess objective sleep parameters — including sleep efficiency — and circadian rhythm markers, including the percentage of rest over 24 hours and average activity during the least active 5-hour period (L5 average) and the most active 10-hour period (M10 average).
• PSG was used to assess sleep architecture, including percentages of sleep stages N2 and N3, total duration of non-REM (NREM) sleep, and total sleep time (TST).

TAKEAWAY

• Patients with unipolar depression reported poorer sleep quality (PSQI score, 12.83 vs 10.61; P = .001), more severe insomnia (ISI score, 18.98 vs 16.60; P = .026), and lower sleep efficiency on actigraphy (80.70% vs 83.62%; P = .026) than those with bipolar depression.
• Patients with bipolar depression showed reduced motor activity during both their least active (L5 average, 784.52 vs 1105.48; P = .004) and most active (M10 average, 10,028.09 vs 12,308.83; P = .024) periods of the day and spent more time at rest overall (49.10% vs 44.15%; P = .017).
• On PSG, patients with bipolar depression showed a higher proportion of N2 sleep (56.56% vs 48.30%; P = .005), a lower proportion of N3 sleep (19.38% vs 25.64%; P = .029), and a longer total NREM sleep duration (05:49:38 vs 04:59:16 hours:minutes:seconds; P = .013) than those with unipolar depression, with a statistically non-significant trend toward a longer TST.
• The final backward stepwise logistic regression model retained six variables (PSQI, ISI, actigraphic sleep efficiency, rest per 24 hours, L5 average, and N2% TST) and explained 49.4% of the variance, demonstrating excellent discriminative ability (area under the curve, 0.926; sensitivity, 0.882; specificity, 0.895; positive predictive value, 88.3%; negative predictive value, 89.5%).

IN PRACTICE

"[The study] results highlight meaningful differences in sleep profiles between the two disorders and suggest the potential of sleep-related features as diagnostic tools," the authors wrote.

SOURCE

The study was led by Jeanne Leseur, Bichat-Claude Bernard Hospital, Paris, France. It was published online on September 15, 2026, in Depression and Anxiety.

LIMITATIONS

Medication effects could not be fully controlled as antidepressants and mood stabilisers were used collinearly with unipolar and bipolar depression diagnoses, respectively. Not all participants completed actigraphy or PSG. As the model accounted for less than half of the variance, the authors called for larger studies with a wider range of predictors to clarify the mechanisms separating the two conditions.

DISCLOSURES

This study did not receive any specific funding. The authors reported having no conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.