Ten studies, reporting 14 comparisons, were included in the meta-analysis. Seven studies, including 8 comparisons (n=1,800), were in patients with type 1 DM; 3 studies, including 6 comparisons (n=4,472), were in patients with type 2 DM. Nine comparisons were performed in Europe, three in North America and two in Asia.
Quality.
Eight comparisons reported adequate allocation concealment. The degree of blinding in 7 comparisons was unclear. An intention-to-treat analysis was carried out in 11 comparisons.
Macrovascular events and mortality.
A total of 134 events (of any type) was recorded in type 1 DM, whilst a total of 1,587 events (of any type) was recorded in type 2 DM. Intensive glycaemic control was associated with a risk reduction in macrovascular events in both type 1 DM (IRR 0.38, 95% CI: 0.26, 0.56; based on 8 comparisons) and type 2 DM (IRR 0.81, 95% CI 0.73, 0.91; based on 6 comparisons) compared with conventional treatment. A substantial amount of variance across studies was found for type 2 DM. The NNT for 10 years to prevent one macrovascular event was 16 for type 1 DM, 14 for type 2 DM low risk and 7 for type 2 DM high risk.
Cardiac events. A total of 40 cardiac events was recorded in type 1 DM, whilst a total of 1,197 cardiac events was recorded in type 2 DM. A risk reduction was found in both type 1 DM (IRR 0.41, 95% CI: 0.19, 0.87; based on 7 comparisons) and type 2 DM (IRR 0.91, 95% CI: 0.80, 1.03; based on 6 comparisons) compared with conventional treatment, although this was not statistically significant in the latter.
Peripheral vascular events.
Totals of 88 and 87 events were recorded in types 1 and 2 DM, respectively. A risk reduction was found in both type 1 DM (IRR 0.39, 95% CI: 0.25, 0.62; based on 4 comparisons) and type 2 DM (IRR 0.58, 95% CI: 0.38, 0.89; based on 6 comparisons) compared with conventional treatment.
Strokes.
A total of 6 strokes was recorded in type 1 DM, whilst a total of 303 strokes was recorded in type 2 DM. A risk reduction was found in both type 1 DM (IRR 0.34, 95% CI: 0.05, 2.57; based on 2 comparisons) and type 2 DM (IRR 0.58, 95% CI: 0.46, 0.74; based on 6 comparisons) compared with conventional treatment, although this was not statistically significant in the former. A substantial amount of variance across studies was found for type 2 DM.
Studies that achieved a greater reduction in HbA1c demonstrated a greater reduction in risk for macrovascular events in type 1 DM; no interaction was found in type 2 DM. The beneficial effect of improved glycaemic control reduced with longer diabetes duration in type 2 DM. A reduction in the beneficial effect of improved glycaemic control in older age was shown in both type 1 DM and type 2 DM, although this was not statistically significant in the former. No statistically significant interaction of gender, year study began, year study was published, or study quality was found. There was no evidence of publication bias.