Tai Chi Chuan for Subjective Sleep Quality: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Table 3

Meta-regression and additive models: sleep quality change response to Tai Chi exercise.

Moderator category/level	β	Adjusted R^2 (%)	value	Effect size (95% CI)	value
Geographic population	−0.563	15.09	0.038
American (k = 11)				−0.213 (−0.558, 0.132)	0.226
Asian (k = 13),				−0.778 (−1.097, −0.460)	<0.001
European (k = 1)				−0.527 (−1.128, 0.075)	0.086
Methodological quality (score)	−0.675	28.96	0.012
Lowest tier, ≤5 (k = 1)				−1.276 (−1.781, −0.770)	<0.001
Middle tier, 6 to 7 (k = 16)				−0.718 (−0.983, −0.453)	<0.001
Highest tier, ≥8 (k = 8)				−0.041 (−0.441, 0.385)	0.839
Training intensity (min/session)	−0.930	17.22	0.026
Low, 30 ≤ time < 60 (k = 10)				−0.596 (−0.975, −0.218)	0.002
Moderate, 60 ≤ time < 90 (k = 12)				−0.647 (−0.946, −0.348)	<0.001
Vigorous, 90 ≤ time < 120 (k = 3)				0.286 (−0.726, 1.298)	0.579
Additive models:
(1) Among Asian population; (2) among studies with methodological quality of middle tier (PEDro score: 6 to 7 points).
Low intensity: 30 ≤ time < 60 (min/session)				−0.674 (−1.215, −0.133)	0.015
Moderate intensity: 60 ≤ time < 90 (min/session)				−0.795 (−1.246, −0.344)	0.001

β, regression coefficient; k, number of trials; adjusted R², the proportion of the source of heterogeneity explained by each moderator; value, the statistical significance of each moderator in the meta-regression model; value, the statistical significance of subgroup analyses; effect sizes are generated by subgroup analyses based on the three moderators, respectively.