(This commentary was not written by CRD, but by the authors of the DH Register.)
1) The confidence intervals of the meta-analysis could have been used in the sensitivity analysis to illustrate the range of uncertainty for each drug. 2) The median duration of the trials is 10 weeks, but benefits have been estimated for 20 years of use. 3) Assigning a QOL of 0.98 to those on therapy resulted in net negative benefits for all treatments. 4) Compliance is not discussed. 5) Benefits of intervention derive from epidemiological rather than trial data. 6) The reductions in blood pressure by class of drug and initial BP level are based on a meta-analysis of placebo-controlled trials (randomisation is not specified). 7) These reductions are based on small numbers and thus have wide confidence intervals. 8) The sensitivity analysis was not adequate.