Introduction Classes of blood pressure lowering drugs (BPLD) result in unexplained differences in stroke risk, independent of mean BP. We did a systematic review of BPLD on consistency of control of BP in randomised controlled trials (RCTs), related to stroke risk.

Methods Baseline and follow-up mean (SD) BP were extracted from all relevant RCTs. Effect of BPLD on variance in BP (SD-squared) was expressed as the variance ratio (VR), by random-effects meta-analysis. In trials with ≥100 patient-years per treatment group, VR was correlated with stroke risk, adjusted for mean BP.

Results Of 1372 eligible trials, 389 reported mean BP and SD at follow-up. Of the substantial heterogeneity between trials in VR (p<1×10^-40), 68% was due to drug class. Compared to other drugs, SD-SBP was lower on calcium antagonists (VR=0.81 95% CI 0.76 to 0.86, p<0.001) and diuretics (0.87, 0.79–0.96, p=0.007), and higher on ACE-inhibitors (1.08, 1.02–1.15, p=0.008), ARBs (1.16, 1.07–1.25, p<0.001) and β-blockers (1.17, 1.07–1.28, p<0.001). Across all classes, reduced VR was associated with a reduced risk of stroke (VR<1.0-OR 0.87, 0.78–0.97, p=0.012; VR≤0.80-OR 0.79, 0.71–0.87, p<0.0001), which was significant (p=0.0016) when adjusted for change in mean BP.

Conclusion Substantial drug-class effects on consistency of control of BP, independent of mean, account for differences in stroke risk. Consistency of control of SBP is as important as extent of control (mean BP) in preventing stroke.