Five RCTs were included in the review (27,802 patients). Study size ranged from 753 to 11,140 participants. All of them reported the blinded outcome assessment and independent adjudication for outcomes. All RCTs also reported intention-to-treat analysis and had a randomised open-label design. Follow-up rates varied from 85.5 to 99.8%.
Disease: Intensive glucose control was associated with reduced risk of cardiovascular disease (relative risk 0.90, 95% CI: 0.83 to 0.98; risk difference -15, 95% CI: -24 to -5; five RCTs) and chronic heart disease (relative risk 0.89, 95% CI: 0.81 to 0.96; risk difference -11, 95% CI: -7 to -5; five RCTs) compared with conventional treatment. Subgroup analyses of early versus more recent trials showed similar results. No significant statistical heterogeneity was found.
Mortality: Intensive glucose control had no significant effect on cardiovascular mortality or all-cause mortality compared with conventional treatment, but there was significant heterogeneity between results of the subgroup analyses; p values for heterogeneity between subgroups were 0.095 (cardiovascular) and 0.105 (all-cause). Early trials (two RCTs) showed non-statistically significant protective effects of intensive glucose control on cardiovascular and all-cause mortality, whereas later trials (three RCTs) indicated non-significant increased risks for these outcomes.
Other outcomes: Intensive glucose control had no significant effect on overall risk of stroke or chronic heart failure compared with conventional treatment (no significant statistical heterogeneity found). It was associated with a 16% reduced risk of non-fatal myocardial infarction (five RCTs) and absolute risk reductions of nine events per 1000 patients over five years of treatment in overall and subgroup analyses. No association was found between intensive glucose control and fatal myocardial infarction (five RCTs), non-fatal stroke (five RCTs), fatal stroke (five RCTs) or peripheral artery disease (four RCTs).
Hypoglycaemia: Intensive glucose control was associated with a two-fold increase in severe hypoglycaemia (an absolute increase of 39 events per 1000 patients). Evidence of statistical heterogeneity was found (p<0.001, I2=85.4%). In subgroup analyses there was no association between intensive glucose control and severe hypoglycaemia in the early trials (two RCTs) and a 2.5-fold increase in the more recent trials (absolute increase of 54 events per 1000 patients over five years; three RCTs).
Sensitivity analyses reported similar results and did not substantially alter the main findings.