. 2018 Jan 26;25(2):145–153. doi: 10.1097/GME.0000000000000971

TABLE 2.

PSQI global and individual domain scores

Measurement	Placebo (PBO)	Transdermal (t-E2)	Oral (o-CEE)	Overall P	PBO/t-E2	Contrasts PBO/o-CEE	t-E2/o-CEE
PSQI score and subscales (n = 653)
Global score
Baseline score^a	6.17 ± 3.06	6.10 ± 3.12	6.57 ± 3.36
Mean follow-up score	5.69 ± 2.95	4.92 ± 2.63	5.23 ± 2.71
Mean change (95% CI)^b	−0.60 (−0.89, −0.31)^c	−1.32 (−1.64, −1.00)^c	−1.27 (−1.58, −0.95)^c	0.001	0.002	0.001	0.975
Satisfaction/quality
Baseline score^a	1.12 ± 0.66	1.17 ± 0.75	1.20 ± 0.72
Mean follow-up score	0.97 ± 0.55	0.85 ± 0.57	0.84 ± 0.55
Mean change (95% CI)^b	−0.19 (−0.25, −0.12)^c	−0.32 (−0.39, −0.25)^c	−0.34 (−0.41, −0.27)^c	0.001	0.008	<0.001	0.449
Sleep latency
Baseline score^a	1.02 ± 0.92	0.95 ± 0.88	1.06 ± 0.93
Mean follow-up score	0.97 ± 0.79	0.74 ± 0.70	0.79 ± 0.73
Mean change (95% CI)^b	−0.08 (−0.15, 0.00)	−0.27 (−0.36, −0.19)^c	−0.28 (−0.36, −0.19)^c	<0.001	0.002	<0.001	0.822
Sleep efficiency
Baseline score^a	0.46 ± 0.78	0.45 ± 0.72	0.61 ± 0.91
Mean follow-up score	0.40 ± 0.64	0.35 ± 0.47	0.39 ± 0.64
Mean change (95% CI)^b	−0.10 (−0.16, −0.03)^c	−0.15 (−0.23, −0.08)^c	−0.16 (−0.24, −0.09)^c	0.451
Sleep disturbances
Baseline score^a	1.52 ± 0.54	1.44 ± 0.61	1.52 ± 0.60
Mean follow-up score	1.43 ± 0.47	1.26 ± 0.41	1.39 ± 0.47
Mean change (95% CI)^b	−0.08 (−0.13, −0.03)^c	−0.22 (−0.27, −0.16)^c	−0.12 (−0.18, −0.07)^c	0.001	<0.001	0.166	0.029
Sleep duration
Baseline score^a	0.54 ± 0.68	0.51 ± 0.70	0.57 ± 0.77
Mean follow-up score	0.47 ± 0.64	0.43 ± 0.60	0.41 ± 0.58
Mean change (95% CI)^b	−0.08 (−0.14, −0.01)^c	−0.10 (−0.18, −0.03)^c	−0.15 (−0.22, −0.07)^c	0.377
Daytime dysfunction
Baseline score^a	0.91 ± 0.75	0.89 ± 0.70	0.84 ± 0.68
Mean follow-up score	0.79 ± 0.59	0.71 ± 0.58	0.73 ± 0.55
Mean change (95% CI)^b	−0.11 (−0.17, −0.05)^c	−0.19 (−0.26, −0.12)^c	−0.14 (−0.21, −0.08)^c	0.258
Vasomotor symptom scores (n = 662)
Hot flashes severity
Baseline score^a	1.39 ± 0.79	1.38 ± 0.82	1.37 ± 0.88
Mean follow-up score	0.89 ± 0.68	0.44 ± 0.60	0.39 ± 0.46
Mean change (95% CI)^b	−0.48 (−0.55, −0.42)^c	−0.93 (−1.00, −0.85)^c	−0.98 (−1.05, −0.90)^c	<0.001	<0.001	<0.001	0.343
Night sweats severity
Baseline score^a	1.08 ± 0.92	1.12 ± 0.95	1.08 ± 0.98
Mean follow-up score	0.62 ± 0.67	0.33 ± 0.54	0.33 ± 0.46
Mean change (95% CI)^b	−0.47 (−0.54, −0.41)^c	−0.76 (−0.83, −0.69)^c	−0.76 (−0.83, −0.69)^c	<0.001	<0.001	<0.001	0.919

PSQI results are reported on a total of 653 respondents (n = 243 [PBO], 201 [t-E2], 209 [o-CEE]), whereas vasomotor symptom results are presented on 662 respondents (n = 243 [PBO], 205 [t-E2], 214 [o-CEE]); to describe the treatment responses, repeated measurements were summarized by the average score per woman during the treatment period, whereas comparisons between treatments utilized all serial data points available via repeated measures analysis (see “Methods” section for details).

Treatment groups were tested for a difference in response in ANCOVA models that included site and baseline score as covariates, with the GEE method used to account for within-participant correlation from repeated measurements. From each ANCOVA, the three-level treatment effect was screened for significance and only when such evidence was revealed did we proceed with post hoc testing of pair-wise group differences.

CI, confidence interval; o-CEE, oral conjugated equine estrogen; PBO, placebo; PSQI, Pittsburgh Sleep Quality Index; t-E2, transdermal estradiol.

^aThere were no significant differences in baseline scores between treatments (P > 0.05 from analysis of variance [ANOVA]). To avoid case-wise deletion in analyses of treatment responses, the mean baseline estimate of each PSQI scale was imputed on n = 38 women who lacked a baseline measurement.

^bBaseline-adjusted mean change over follow-up assessments was estimated via analysis of covariance [ANCOVA] modeling, and is reported as mean and 95% confidence interval.

^cIndicates that the baseline-adjusted change within corresponding group was statistically significant (P < 0.05).