Nineteen studies (nine randomised and 10 observational studies) were included in the review (35,425 participants, range 25 to 9,314). The quality of the randomised trials was fair: four reported allocation concealment, five reported no loss to follow-up and one reported blinding of outcome assessors. Eight out of 10 observational studies reported adequate adjustments for confounders and prognostic balance at baseline. Blinding was not reported in any of the observational studies.
Meta-analysis showed that intensive glycaemic control did not reduce the risk of death (11 studies), myocardial infarction (three studies), stroke (three studies) and hypoglycaemia (11 studies). The authors indicated that there was significant statistical heterogeneity (Ι²=60%) for these outcomes. Intensive glycaemic control was associated with decreased risk of infection (RR 0.41, 95% CI 0.21 to 0.77; four studies, Ι²=33%).
There was an increased risk of hypoglycaemia in surgical ward patients (one study) and when the glycaemic target was achieved. The remaining planned subgroup analyses did not reveal statistically significant interactions.