Table 3.
Multivariable Adjusted Odds Ratios for Menopausal Status, Vasomotor Symptoms, Estradiol and Follicle Stimulating Hormone Associated with 3 Types of Sleep Difficulties§
| Trouble falling asleep: 17692∼18748 obs from 2997∼3006 women OR (95% CI) | Wake up several times: 18070∼19670 obs from 3032∼3043 women OR (95% CI) | Wake up early: 18234∼19809 obs from 3033∼3043 women OR (95% CI) | |
|---|---|---|---|
| Menopausal Status | P = 0.0006 | P < 0.0001 | P = 0.0041 |
| Premenopausal (no HT) | 1.00 | 1.00 | 1.00 |
| Early perimenopausal (no HT) | 1.16 (0.93, 1.45) | 1.26 (1.08, 1.47)** | 1.06 (0.88, 1.28) |
| Late perimenopausal (no HT) | 1.49 (1.08, 2.06)* | 1.96 (1.56, 2.46)*** | 1.53 (1.17, 2.01)** |
| Postmenopausal (no HT) | 1.77 (1.29, 2.44)*** | 1.78 (1.42, 2.23)*** | 1.19 (0.91, 1.56) |
| Surgically menopausal (no HT) | 2.05 (1.21, 3.49)** | 2.04 (1.38, 3.03)*** | 1.50 (0.96, 2.33) |
| Unknown† | 1.55 (1.13, 2.14)** | 1.22 (0.97, 1.54) | 1.39 (1.07, 1.82) * |
| Postmenopausal (HT) | 0.95 (0.57, 1.57) | 1.13 (0.82, 1.57) | 1.03 (0.70, 1.53) |
| Surgically menopausal (HT) | 1.05 (0.58, 1.88) | 1.48 (0.95, 2.30) | 1.60 (0.98, 2.61) |
| Vasomotor symptoms | P < 0.0001 | P < 0.0001 | P < 0.0001 |
| None | 1.00 | 1.00 | 1.00 |
| < 6 days/2 weeks | 1.37 (1.16, 1.62)*** | 1.31 (1.17, 1.47)*** | 1.35 (1.17, 1.55)*** |
| 6–14 days/2 weeks | 2.60 (2.14, 3.14)*** | 2.93 (2.55, 3.37)*** | 2.16 (1.84, 2.54)*** |
| E2 (log scale) ‡ | |||
| Baseline | 1.09 (0.94, 1.27) | 1.04 (0.93, 1.17) | 0.99 (0.87, 1.12) |
| Change since baseline | 1.12 (1.04, 1.21)** | 1.08 (1.02, 1.14)** | 1.03 (0.97, 1.10) |
| FSH (log scale) | |||
| Baseline | 1.01 (0.86, 1.18) | 1.01 (0.90, 1.14) | 0.98 (0.86, 1.12) |
| Change since baseline | 1.05 (0.95, 1.16) | 1.11 (1.03, 1.19)** | 1.02 (0.94, 1.11) |
Abbreviations: OR, odds ratio; CI, confidence interval; FSH, follicle stimulating hormone; E2, estradiol; HT, hormone therapy; CES-D, Center for Epidemiologic Studies Depression Scale; obs, observations (note: numbers of obs and women differ slightly across models due to small differences in missing information between dependent variables and independent variables and covariates).
Ethnicity, site, age, and current HT use were forced into all the final multivariable models. Estimates also were adjusted for the following covariates (P < 0.05) in Menopausal Status models for all 3 sleep difficulties: CES-D score, anxiety symptom score, urinate at night, number of physical symptoms, bodily pain score, very upsetting life events, and psychotropic/sleep medication use. For Trouble Falling Asleep: model also was adjusted for difficulty paying for basics, body mass index, smoking status, and education. Education overall effect was not statistically significant but was included because it was an important confounder for the smoking effect. For Wake Up Several Times: model also was adjusted for pain medication use and self-assessed overall health. In all Vasomotor Symptoms models the sleep difficulties were adjusted for the same set of covariates as in the respective Menopausal Status models. Covariate adjustment in the Hormone (E2 and FSH) models − for Trouble Falling Asleep: same as Menopausal Status model; for Wake Up Several Times: same as Menopausal Status model, plus number of medical conditions; for Wake Up Early: same as Menopausal Status models. Only Hormone models were additionally adjusted for whether the blood level was drawn within the cycle day 2–5 window. In multivariable hormone models, compared with women who had blood drawn in cycle day 2–5 window, women who did not have blood drawn in window were more likely to report sleep difficulties with odds ratio of 1.18–1.38.
“Unknown” category of menopausal status indicates those women who used hormone therapy but were not postmenopausal yet.
Estimates presented for decreasing E2.
Note: *P < 0.05; **P < 0.01; ***P < 0.001.