Journal of Rehabilitation Medicine 51-6 | Page 14

408 Z. Zhou et al. other sources (Fig. 1). A total of 292 articles remained after removing duplicates. After removal of irrelevant articles through titles and abstracts, full-text reading of 50 articles was performed for further assessment. Of these, 25 articles were excluded for various reasons (see Fig. 1). Therefore, a final total of 23 studies reported in 25 articles (32–56) were included in this systematic review and meta-analysis. Characteristics and risk of bias assessment in individual study One RCT reported in 3 articles (49–51) was regarded as a study, which together with another study (48) originated from Australia (Table II). One of 23 studies originated from South Korea (43), one was conducted in Thailand (42) and the rest in China. The sample size ranged from 16 to 200, and mean age ranged from 35.6 to 69.5 years. The average duriation of disease a ranged from 1 to 23 years, and the mean HbA1c before treat- ment from 6.9% to 11.9%. Four studies (33, 34, 36, 38) applied Chen style tai chi, and one (34) applied Yang style tai chi. A simplified style was used in 5 studies (45–47, 55, 56). Sun style and Yang style tai chi was used in 2 studies (4 articles) (48–51), Lin style was used in 1 study (42), and Da-yuan-jiang-tang style was used in 1 study (54). The mean time per session ranged from 15 to 120 min, mean intervention frequency from 2 to 14 sessions per week, and the mean duration of the intervention from 4 to 24 weeks. The majority of studies reported inclusion criteria, except for 2 studies (38, 55). The treatment group and the control group were comparable at baseline for all studies, and point estimates and variability were all reported. All studies except 4 (32, 36, 43, 56) were considered high quality. The risk of bias assessment is shown in Table III. Effects of tai chi in patients with type 2 diabetes Fasting plasma glucose. A total of 21 studies (32–45, 47, 51–56) evaluated the effects of tai chi on FPG. With high heterogeneity across studies, a random-effects pooled analysis was performed to synthesize the data. The results indicated that tai chi was potentially effective in reducing FPG (SMD –0.67; 95% CI –0.87 to –0.47; p < 0.001; heterogeneity, I 2  = 53.2%, p = 0.001; Fig. 2). Glycosylated haemoglobin (HbA1c). A total of 14 studies (35, 36–43, 45, 48, 49, 52–54) compared the effects of tai chi on HbA1c with a control group. Meta- analysis showed that tai chi was beneficial in lowe- ring HbA1c (MD –0.88; 95% CI –1.45% to –0.31%; p = 0.002; heterogeneity, I 2  = 97.8%, p < 0.001; Fig. 3A). Sensitivity analysis showed that 2 studies (37, 45) exerted great influence on the stability of synthesized results. The results after excluding these 2 studies also indicated that tai chi was effective in decreasing HbA1c (MD –0.53; 95% CI –0.62% to –0.44%; p < 0.001; heterogeneity, I 2  = 43.4%, p = 0.033; Fig. 3B). Literature retrieved from database searching (n=487): Wan Fang (n=159), CNKI (n=98), CBM (n=39), VIP (n=86), Pubmed (n=33), Embase (n=54), Cochrane Library (n=18) Literature obtained from other resources (n=0) Literature after removing duplicates (n=292) Excluded articles (n=242): Review (n=68); Healthy individuals or patients with other diseases (n=54); Non-tai chi treatment (n=57); Clinical guidelines (n=8); Non-RCTs (n=26); Pathological mechanism (n=29) Articles assessed by reading titles and abstracts (n=292) Articles assessed through full-text reading (n=50) Excluded articles (n=25): Reference paper (n=1); Non-RCTs (n=1); Non-T2DM participants (n=3); Inappropriate control (n=13); Non-targeted outcomes (n=5); Unavailable data (n=2) Studies (reported in 25 articles) included in the systematic review (n=23) Fig. 1. Flow chart of study selection. T2MD: type 2 diabetes mellitus; RCT: randomized controlled trial. www.medicaljournals.se/jrm