Eight phase III open-label trials with a total of 2,361 participants were identified, although only six were included in the main analysis. Three of these trials enrolled patients with type I diabetes (n=1,344), whilst the other 3 enrolled patients with type II diabetes (n=1,017). The other two trials were of a crossover design, one in type I diabetics and one in type II diabetics.
Pooled studies (patients with type I and type II diabetes; 6 studies).
Using a fixed-effect model, one outcome (a decrease in postprandial blood glucose levels to less than or equal to 8 mmol/L) showed a statistically-significant difference in favour of insulin lispro (p<0.00001). Using a random-effects model, a significantly higher proportion of patients receiving lispro achieved at least one therapeutic success (59.4 versus 49.3%; OR 1.68, 95% confidence interval, CI: 1.34, 2.12). No significant differences were found for the other two variables (a 2-hour postprandial blood glucose level within 20% of the pre-meal level or at least a 50% decrease from baseline in 2-hour postprandial glucose excursion).
Continuous variables (1- and 2-hour postprandial blood glucose, and 1- and 2-hour glucose excursion) showed statistically-significant differences in favour of insulin lispro (p<0.02, p<0.001, p<0.001 and p<0.001, respectively). The weighted mean differences for measures of long-term glycaemic control, fasting blood glucose, and hypoglycaemic rate per 30 days showed no distinctions between insulins.
Studies in patients with type I diabetes (3 studies).
Overall, a significantly higher proportion of patients receiving lispro achieved at least one therapeutic success (OR 1.35, 95% CI: 1.32, 1.83), although only one third of the individual measurements showed a statistically-significant benefit. Significantly more patients had a decrease in the postprandial blood glucose to less than or equal to 8 mmol/L with insulin lispro (OR 1.52, 95% CI: 1.27, 1.83, p<0.00001).
Studies in patients with type II diabetes (3 studies).
Overall, a significantly higher proportion of patients receiving lispro achieved at least one therapeutic success (OR 1.45, 95% CI: 1.20, 1.75), although only one third of the individual measurements showed a statistically-significant benefit (OR 1.79, 95% CI: 1.39, 2.32). Significantly more patients had a decrease in the postprandial blood glucose level to less than or equal to 8 mmol/L with insulin lispro (p<0.00001). Statistically-significant differences in the weighted mean differences were seen in the 1- and 2-hour glucose excursion (p<0.05 and p<0.02, respectively) but not in the 1- and 2- hour postprandial blood glucose levels.
The results obtained when using the random-effects model with all 8 studies reached similar conclusions to the fixed-effect model.