Twelve parallel open label trials were included (n=4,385); two trials were included only in the subgroup analysis. Mean sample size was 366 (range 24 to 756) participants. There was no evidence of publication bias. No statistical heterogeneity was detected in any of the analyses.
There were no statistically significant differences between NPH and glargine in terms of changes in insulin dose, FPG, and HbA1c. The result for body weight favoured NPH and this was statistically significant (-0.33 kg, 95% CI -0.61 to -0.06; five trials).
Patient-reported symptomatic and nocturnal hypoglycaemia was reported to be lower following glargine (p<0.0001), but there was no statistically significant difference in pooled rates between the groups. Subgroup analysis revealed that the mean net change in FPG was significantly lower following treatment with glargine in trials with a duration greater than or equal to 24 weeks (0.261, 95% CI 0.041 to 0.481; eight trials) and in trials that contained a higher proportion of men (0.415, 95% CI 0.175 to 0.655; five trials).
Less weight gain was reported following NPH treatment in trials with sample sizes greater than 422 (-0.599, 95% CI -1.099 to -0.099; two trials) and those with a higher proportion of men (-0.410, 95% CI -0.742 to -0.77; three trials).
In sensitivity analysis, glargine was superior in terms of net change in FPG when two trials were excluded and in terms of lowered HbA1c when another trial was excluded.