Table 3. Summary of Post Hoc Analysis of Generalized Estimated Equation Analysis^a.

Measurement	Postintervention			Follow-up
	Mean adjusted change (95% CI)	P value	Effect size	Mean adjusted change (95% CI)	P value	Effect size
Primary outcomes (actigraphy-assessed)b
Sleep efficiency, %
Exercise vs control	3.5 (1.8 to 5.2)	<.001c	0.42	3.3 (1.4 to 5.2)	<.001c	0.35
Tai chi vs control	3.4 (1.6 to 5.1)	<.001c	0.39	6.2 (4.4 to 8.1)d	<.001c	0.69
Tai chi vs exercise	0.2 (−1.5 to 1.9)	.85	0.03	2.9 (1.1 to 4.8)	.002b	0.35
Wake time after sleep onset, min
Exercise vs control	−17.0 (−24.9 to −9.0)	<.001c	0.46	−23.0 (−31.7 to −14.2)d	<.001c	0.59
Tai chi vs control	−13.3 (−21.3 to −5.2)	.001c	0.31	−27.1 (−35.8 to −18.4)d	<.001c	0.65
Tai chi vs exercise	3.7 (−4.2 to 11.7)	.36	0.15	−4.1 (−12.6 to 4.4)	.34	0.06
No. of awakenings
Exercise vs control	−2.8 (−4.0 to −1.6)d	<.001c	0.48	−2.0 (−3.3 to −0.6)d	.004	0.44
Tai chi vs control	−2.2 (−3.5 to −1.0)	<.001c	0.42	−2.8 (−4.1 to −1.4)d	<.001c	0.55
Tai chi vs exercise	0.6 (−0.6 to 1.8)d	.35	0.05	−0.8 (−2.1 to 0.5)	.24	0.11
Sleep onset latency, min
Exercise vs control	−0.4 (−2.2 to 1.4)	.68	0.0002	−2.5 (−4.5 to −0.6)	.01	0.24
Tai chi vs control	−1.4 (−3.2 to 0.4)	.12	0.12	−3.8 (−5.8 to −1.9)	<.001c	0.47
Tai chi vs exercise	−1.1 (−2.8 to 0.7)	.24	0.12	−1.3 (−3.2 to 0.6)	.17	0.23
Secondary outcomese
PSQI
Exercise vs control	−1.1 (−1.9 to −0.3)	.01	0.31	−0.9 (−1.8 to −0.1)	.03	0.29
Tai chi vs control	−2.9 (−3.6 to −2.1)	<.001	0.83	−1.6 (−2.4 to −0.8)	<.001	0.49
Tai chi vs exercise	−1.8 (−2.5 to −1.0)	<.001	0.52	−0.6 (−1.4 to 0.2)	.17	0.2
ISI
Exercise vs control	−2.2 (−3.3 to −1.0)	<.001	0.35	−2.1 (−3.2 to −0.9)	<.001	0.37
Tai chi vs control	−4.1 (−5.2 to −3.0)	<.001	0.92	−2.6 (−3.8 to −1.5)	<.001	0.61
Tai chi vs exercise	−1.9 (−3.0 to −0.8)	.001	0.57	−0.5 (−1.7 to 0.6)	.43	0.24
Sleep efficiency in 7-d sleep diary, %
Exercise vs control	3.4 (−0.8 to 7.6)	.16	0.25	9.5 (5.1 to 13.9)	<.001	0.77
Tai chi vs control	9.0 (4.7 to 13.3)	<.001	0.51	10.2 (5.8 to 14.5)d	<.001	0.72
Tai chi vs exercise	5.6 (1.4 to 9.8)	.02	0.27	0.7 (−3.7 to 5.1)	.77	0.05
Wake time after sleep onset in 7-d sleep diary, min
Exercise vs control	−12.2 (−26.8 to 2.4)	.14	0.17	−23.9 (−39.1 to −8.6)	.004	0.43
Tai chi vs control	−40.4 (−55.2 to −25.7)d	<.001	0.63	−22.5 (−37.5 to −7.5)	.006	0.41
Tai chi vs exercise	−28.2 (−42.8 to −13.6)	<.001	0.46	1.3 (−13.8 to 16.5)	.87	0.02
Total sleep time in 7-d sleep diary, min
Exercise vs control	15.8 (−2.5 to 34.1)	.13	0.18	44.2 (25.2 to 63.3)d	<.001	0.55
Tai chi vs control	49.8 (31.3 to 68.3)d	<.001	0.46	36.1 (17.2 to 54.9)d	<.001	0.35
Tai chi vs exercise	34.0 (15.6 to 52.3)d	<.001	0.28	−8.1 (−27.2 to 10.9)	.48	0.2
Hypnotic medication usage recorded in 7-d sleep diary (weekly consumed LRD units)
Exercise vs control	−2.1 (−0.6 to −3.7)	.02	0.28	−3.3 (−1.7 to −4.9)	<.001	0.36
Tai chi vs control	−4.0 (−2.5 to −5.5)	<.001	0.76	−3.7 (−2.3 to −5.2)	<.001	0.73
Tai chi vs exercise	−1.9 (−0.2 to −3.6)	.05	0.48	−0.5 (−2.2 to 1.2)	.86	0.38

Abbreviations: ISI, insomnia severity index; LRD, lowest recommended dosage; PSQI, Pittsburgh sleep quality index.

^aHolm correction was used for post hoc analysis in order to account for the multiplicity within an actigraphy-assessed sleep parameter among difference groups and among the actigraphy-assessed sleep parameters. Closed test procedure with Holm-Bonferroni correction was used for the post hoc analysis of the secondary outcomes.

^bNo effect for interaction observed for total sleep time and average awaken time.

^cThe pairwise comparison remained statistically significant after Holm correction.

^dThe change in magnitude exceeded the clinical significance threshold defined by American Academy of Sleep Medicine (actigraphy tool: 5% in the sleep efficiency, 20 minutes in the wake time after sleep onset, and 2 times in the number of awakenings; subjective tool: 10% in the sleep efficiency, 30 minutes in the wake time after sleep onset, 30 minutes in the total sleep time).¹⁸

^eMean values are averages across 7-day sleep diary. No effect for interaction observed for number of awakenings, sleep onset latency, and average awaken time.