This review concluded that no recommendations could be made for physicians to encourage patients with type II diabetes to practice yoga; large well-designed clinical trials are needed. The authors' conclusions appeared to reflect the evidence and are appropriately cautious given the poor quality and small number of included trials and the limited evidence available.

To assess the effectiveness of practicing yoga for the management of type II diabetes mellitus.

Published trials in any language were identified through a search of MEDLINE, EMBASE, CINAHL, AMED, Cochrane Central Register of Controlled Trials (CENTRAL) and Cochrane Metabolic and Endocrine Disorders Group up to May 2007. Search terms were reported. Trial registrations from www.controlled-trials.com, symposium, conference and colloquium proceedings were also searched. Journals not available electronically were handsearched. Reference lists of included studies were searched for relevant trials. Experts and authors in the diabetes and yoga fields were contacted for any unpublished data.

Randomised or quasi-randomised trials that compared yoga with any other intervention in adult in-patients or out-patients (aged at least 18 years of age) with a confirmed diagnosis of type II diabetes mellitus were eligible for inclusion. Patients were required to be on medication, either oral or insulin injection. All types of yoga practice (including Ayurveda) were eligible in any comparison with diabetes symptom or complication management strategy (such as usual management, complementary or alternative therapy). Patients in critical care, not on medical management for diabetes, or with a physical limitation to their ability to participate in the intervention were excluded.

The primary outcome was glucose level control, either fasting plasma glucose and/or glycated haemoglobin (HbA1c). Secondary outcomes included: change in anti-diabetic medication use (as an indication for diabetes control); development of diabetic complications (including neuropathies, retinopathy, nephropathy and cardiovascular diseases); weight (kg) or body mass index; lipids profile: total cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides; diabetes-related mortality; and any adverse effects resulted from yoga practice.

In the included trials, interventions included yoga alone, or in combination with lifestyle modification and/or diet and herbal intake. Yoga was conducted in group settings and by individuals. The frequency of yoga practice varied from 20 minute sessions twice a day to three, four or five 90 minute sessions. The mean trial duration was 14 weeks (range six to 24 weeks). The average age of patients ranged from 49.9 years to 57.0 years.

The authors did not state how the studies were selected for the review.

Methodological quality was assessed by two independent reviewers using the Cochrane Collaboration Criteria that evaluated randomisation, allocation concealment, blinding, withdrawals or loss to follow-up and intention-to-treat. Each trial was graded as A where all criteria were met (low risk of bias), B where one or more criteria were partly met (moderate risk of bias), or C where one or more criteria were not met (high risk of bias). Disagreements were resolved by consensus.

Numbers of participants and events for dichotomous outcomes were extracted to calculate of relative risk (RR); means and standard deviations were extracted to calculate weighted or standardised mean differences (WMD/SMD) for continuous outcomes.

The authors did not state how many reviewers performed the data extraction.

The authors had planned to calculate the pooled relative risk and corresponding 95% confidence intervals for dichotomous outcomes using a fixed-effect meta-analysis, where there was no evidence of statistical heterogeneity. A random-effects model was planned if statistically significant heterogeneity was observed. For continuous outcomes the authors had planned to pool the weighted mean differences or standardised mean differences (and corresponding 95% confidence intervals) using a fixed-effect meta-analysis where there was no evidence of statistical heterogeneity. A random-effects model would be used if statistically significant heterogeneity was observed.

The authors decided not to pool the trials due to clinical and methodological heterogeneity and presented a narrative synthesis grouped by outcome.

Publication bias was assessed using Egger’s test and the fail-safe N.

Five trials (n=362 patients) were included in the review. The mean number of participants was 72 (range 21 to 154). Overall trial quality was poor; two trials were graded B (moderate risk of bias) and three studies were graded C (high risk of bias).

Fasting plasma glucose: Yoga lowered fasting plasma glucose, with mean differences that ranged from -28.8 (95% CI -0.55 to -2.60) to 41.10 (95% CI -53.80 to -28.40) in three trials. No difference was found in one trial (mean difference -3.60, 95% CI -19.92 to 12.72).

Glycated haemoglobin: Yoga lowered the level of glycated haemoglobin in one trial (mean difference -1.20, 95% CI -2.23 to -0.17). No differences were found in two trials.

Cholesterol: Yoga lowered the level of cholesterol in two trials, with mean differences of -19.49 (95% CI -28.85 to -10.13) and -35.90 (95% CI -44.18 to -27.62). No difference was found in one trial.

High-density lipoprotein: Yoga lowered the level of high-density lipoprotein in two trials, with mean differences of 8.49 (95% CI 6.58 to 10.40) and 10.90 (95% CI 8.80 to 13.00). No difference was found in one trial.

Triglycerides: Yoga lowered the level of triglycerides in two trials, with mean differences of -31.01 (95% CI -40.05 to -21.97) and -8.00 (95% CI -13.71 to -2.29). No difference was found in one trial.

No differences were reported for lipid profile low-density lipoprotein (two trials), weight and body mass index (two trials) or diabetes related complications (two trials).

No definitive recommendations could be made for physicians to encourage their patients to practice yoga. The important recommendation that was drawn from the review was the need for well-designed large randomised clinical trials to assess the effectiveness of yoga on type II diabetes.

This review addressed a clear question supported by appropriate inclusion criteria. Relevant databases were searched; no limits were placed on language and attempts were made to locate unpublished data. Publication bias was considered in the report but, given the small number of included trials, did not add value to the review. Suitable methods to minimise risk of reviewer error and bias were reported for quality assessment, but it was unclear whether steps to reduce reviewer bias and error were undertaken for study selection or data extraction.

The decision to present included trials narratively and not to pool them in a meta-analysis was appropriate given the clinical heterogeneity between trials. The authors acknowledged some of the methodological difficulties with the review; specifically, the poor methodological quality of included trials and their small sample sizes.

The authors' conclusions appeared to reflect the evidence and are appropriately cautious given the poor quality of included trials, the small number of included trials and limited evidence available.

Practice: The authors stated that no recommendations can be made for physicians to encourage their patients to practice yoga at present.

Research: The authors stated that well-designed, large, randomised clinical trials that assess the effectiveness of yoga on type II diabetes are needed, and that the long term impact of yoga and its side effects should be assessed.

Not stated

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.