IPD from 12 RCTs were included (n=698).
Estimated effect at 2 years for studies with at least 2 years of follow-up (10 studies, n=646): all the studies demonstrated a marked beneficial effect of ACE inhibitors on the rate of change in albumin excretion rate. Using data from the patient-level regression model, the albumin excretion rate was 51.2% (95% confidence interval, CI: 33.6, 64.2) lower in those receiving ACE inhibitors when using the fixed-effect model and 50.2% (95% CI: 30.0, 64.6) lower when using the random-effects model.
Adjustment for baseline covariates: the separate addition of covariates such as age, gender, duration of diabetes, in addition to the initial adjustment of baseline albumin excretion, had no impact on the treatment effect.
Change in systemic blood-pressure and albumin excretion rate: when adjustment was made for changes in blood-pressure and albumin excretion rate, the treatment effect was attenuated from 50.7% (95% CI: 29.8, 65.4) to 45.1% (95% CI: 18.6, 63.1).
Effect of baseline albumin excretion rate: the estimated 2-year difference in albumin excretion rate was 74.1% in patients whose baseline albumin excretion rates were at the upper boundary of microalbuminuria (200 microg/minute), compared with just 17.8% in those who were at the lower boundary of microalbuminuria (20 microg/minute) at baseline (p=0.04).
Progression to macroalbuminuria and regression to normoalbuminuria: with the exception of one study, each trial demonstrated a beneficial effect and the fixed-effect model showed an overall OR of 0.38 (95% CI: 0.25, 0.57, p<0.001) in favour of ACE inhibitors. Regression to normoalbuminuria was greater in treated patients (fixed-effect OR 3.07, 95% CI: 2.15, 4.44, p<0.001))