The review concluded that incident stroke risk was not significantly reduced with intensive glucose control versus standard care, but evidence indicated reduced risk for patients with a BMI above 30kg/m². Overall conclusions are likely to be reliable, but conclusion on the subgroup of obese patients should not be considered definitive due to limitations in the review analyses.

To evaluate the efficacy of glucose control in the prevention of incident stroke.

MEDLINE, EMBASE and The Cochrane Library were searched up to June 2012 for studies in English. References of non-randomised trials and ClinicalTrials.gov were searched for additional published and unpublished references. Search terms were reported.

Randomised controlled trials (RCTs) that compared the effects of intensive control of glucose with standard care on incident stroke (placebo, standard care, or glucose control of reduced intensity) were eligible for inclusion. More than ten incident cases of stroke had to be recorded during the study, and follow-up duration had to be a minimum of 12 months.

Most patients were male. Where reported, mean patient age ranged between 53.3 and 67.5 years, mean body mass index (BMI) ranged from 27.0 to 32.2 kg/m², mean total cholesterol ranged from 178.2 to 210.0 mg/dL, and glycosylated haemoglobin ranged from 5.8% to 9.4%. Most studies included patients with Type 2 diabetes. Types of treatments used varied.

Two reviewers independently selected studies for inclusion and disagreements were settled by a third reviewer.

Studies scored on a scale of 0 to 5; higher scores indicated higher quality. The following quality criteria were assessed: randomisation, concealment of treatment allocation, blinding, completeness of follow-up, and use of intention-to-treat analysis.

Quality was assessed by two reviewers, with disagreements resolved with a third reviewer.

Outcome data (number of incident stroke) were extracted to calculate risk ratios and 95% confidence intervals. Data were extracted by two reviewers, with disagreements resolved with a third reviewer.

Outcomes data were pooled using meta-analysis (fixed-effect and random-effects models). Heterogeneity was assessed using Ι². The effect of potential sources of heterogeneity (patient characteristics, follow-up duration) on the overall estimates were explored using meta-regression and subgroup analyses. Sensitivity analyses were performed to assess the effect of each individual study on the pooled estimates. Publication bias was assessed using Egger's test.

Nine RCTs (59,197 patients, range 620 to 12,537) were included. Duration of follow-up ranged from one to 10 years. One trial had a quality score of 5 out of 5, five scored 4, two scored 3 and one scored 2.

There was no statistically significant difference in risk of stroke between intensive control of glucose and standard care (RR 0.96; 95% CI 0.88 to 1.06; nine RCTs). There was no evidence of significant heterogeneity (Ι²=6.8%). Subgroup analyses showed a statistically significant effect on incident stroke in trials where BMI was 30 or higher (RR 0.86; 95% CI 0.75 to 0.99; number of RCTs unclear). No other significant differences were detected between the effect of intensive control of glucose and standard care when based on other subsets of variables. Results of fixed-effect analyses were not reported.

Overall, the risk of incident stroke was not significantly reduced using intensive glucose control compared with standard care, but evidence indicated reduced risk for patients with a BMI above 30kg/m².

The review question and selection criteria were clearly stated. Several bibliographic sources were consulted, but language restrictions were applied to searches. Study selection, data extraction and quality assessment were performed in duplicate, which reduced risk of reviewer bias and error. Results of the quality assessment were reported, although only in aggregate form.

The decision to pool the studies appeared appropriate. Although the authors stated that results of analyses using different meta-analytical models were similar, only random-effects results were reported. The main analysis was based on large and very large trials. Estimates of heterogeneity were low and sensitivity analyses confirmed the robustness of results. A large number of subgroup analyses were performed, and the upper limit of the confidence interval for the significant subgroup result nearly reached the line of non-significance. This raises concerns about the risk of false positive results associated with repeated testing. Therefore, conclusions about the subgroup of obese patients should be interpreted with caution and should not be considered definitive.

Practice: The authors did not state any implications for practice.

Research: The authors stated that future trials should record reported normatively the type of stroke and the extent of glycosylated haemoglobin lowering of treatments. They stated that the role of intervention duration and dosage should be taken into consideration before evaluating clinical outcomes.

None.

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.