Wearables may add insight into perimenopausal sleep complaints

Perimenopausal women over-reported total sleep time (TST) by a mean of 35.3 minutes and under-reported wakefulness after sleep onset (WASO), with a mean difference of −30.8 minutes, compared with Fitbit measurements, despite strong correlation between self-reported and Fitbit-measured TST (ρ=0.67), according to results of a poster presented at The Menopause Society 2026 Annual Meeting.

Investigators evaluated correlations between self-reported and Fitbit-measured sleep in 190 perimenopausal women aged 36–60 years (mean 47.1 years, SD ±4.3; range 36–57 years) who reported vasomotor symptoms or sleep disturbances via the Evidation consumer health platform. Women were classified as perimenopausal if they were not pregnant and reported in the last 12 months having ≥1 menstrual period and either missing ≥1 menstrual period or experiencing new-onset menstrual irregularities (≥7 days difference).

Pairing daily surveys with 14 nights of Fitbit data

Self-reported TST, total sleep period duration (TSPD), WASO, and sleep quality were collected from daily morning surveys over 14 days. Fitbit devices measured TST, TSPD, WASO, and sleep efficiency (TST/TSPD) over 14 consecutive nights. All 190 women had ≥11 days of paired daily survey and Fitbit sleep data. Investigators calculated repeated measures correlation coefficients to assess correlations between paired measures and conducted day-level Bland-Altman analysis to assess agreement.

Fitbit data indicated worse sleep than self-reports

Investigators analyzed 2537 paired samples. Fitbit-measured median values were:

• TST: 6 hours 49 minutes
• TSPD: 7 hours 42 minutes
• WASO: 46 minutes
• Sleep efficiency: 90%

Variability was high between participants and nights. Self-reported median values were 7 hours 25 minutes for TST, 7 hours 45 minutes for TSPD, and 10 minutes for WASO. Overall, women reported 78% of nights as very or fairly good sleep quality.

Strong correlations were seen between self-reported and Fitbit-measured TST (ρ=0.67; P < 0.001) and TSPD (ρ = 0.68; P < 0.001). Self-reported WASO was moderately positively correlated with Fitbit WASO (ρ = 0.35; P < 0.001) and moderately negatively correlated with Fitbit sleep efficiency (ρ = −0.36; P < 0.001).

Mean differences (self-report minus Fitbit) were:

• TST: 35.3 minutes (95% limits of agreement [LoA], −81.4 to 152.0 minutes), indicating over-reporting
• WASO: −30.8 minutes (95% LoA, −96.0 to 34.4 minutes), indicating under-reporting
• TSPD: 4.3 minutes (95% LoA, −114.1 to 122.7 minutes)

Proportional bias, the correlation between the difference and the mean, was minimal for TST (slope: 0.0968, r² = 1.2%) and TSPD (slope: −0.0394, r² = 0.2%) and moderate for WASO (slope: −0.1479, r² = 1.1%) (all P < 0.05).

Moderate correlations were seen between self-reported sleep quality and Fitbit measures (P < 0.001 for all):

• TST: ρ = 0.30
• TSPD: ρ = 0.19
• Sleep efficiency: ρ = 0.31
• WASO: ρ = −0.23

Wake time shows the greatest discrepancy

The investigators reported that Fitbit-derived and self-reported sleep measures were minimally-to-moderately associated in perimenopausal women, with Fitbit-derived measures indicating worse sleep.

“WASO shows the greatest discrepancy, particularly at higher levels of WASO, suggesting complementary utility of wearables for gaining deeper insights into this hallmark sleep complaint in perimenopausal women,” the study authors concluded.

Reference

Joffe H, Baker F, Lee J, et al. Concordance between self-reported and wearable device-measured sleep in perimenopausal women. Poster. Presented at: The Menopause Society Annual Meeting. October 20-24, 2026. San Diego, California.