Twenty-one studies were included in the review: 19 RCTs (1,440 participants) and two retrospective cohort studies (13,452 participants). All RCTs were of low quality. Most trials used intention-to-treat analyses. There was lack of reporting on allocation concealment and blinding in most trials. Follow-up ranged from 1.0 to 4.5 years. Baseline characteristics were comparable between groups. Analyses were conducted to adjust for confounders.
Reduction of fasting plasma glucose from the baseline was significantly larger for sustained-release glipizide than for immediate-release glipizide (MD -0.26mmol/L, 95% CI -0.52 to -0.01; 17 RCTs; Ι²=35%). There were no significant differences in HbAlc and two-hour postprandial plasma glucose between the two groups.
Sustained-release glipizide reduced two-hour postprandial insulin from baseline by 2.94IU/mL and immediate-release formulation increased its secretion from baseline by 0.24IU/mL (WMD -3.18IU/mL, 95% CI -5.47 to -0.90; five trials; Ι²=39%). There was a non-significant reduction in fasting insulin but there was significant heterogeneity on this outcome (Ι²=83%).
Compared with immediate-release glipizide, sustained-release glipizide was associated with less hypoglycaemia (Peto OR 0.21, 95% CI 0.08 to 0.52; 14 trials) and lower missed dosing (RR 11.42, 95% CI 6.47 to 20.18; seven trials). Where reported, no significant heterogeneity was found for the outcome of hypoglycaemia (Ι²=0%). Consistent results on patient compliance were found in the cohort studies.
There was no evidence of publication bias. Subgroup analyses failed to explain the heterogeneity in efficacy outcomes. Results for other outcomes were also reported.