We identified 429 randomised controlled trials: 73 trials did not compare percutaneous transluminal coronary angioplasty with control, 312 trials evaluated various procedures related to angioplasty (for example, cotreatments and different techniques for the prevention of restenosis), eight trials compared angioplasty with coronary artery bypass grafting, and 28 trials included patients with acute coronary disease. Eight trials proved eligible, of which two had to be excluded because allocation to angioplasty in the intervention groups was not random.11,12
Of the six remaining trials, three included patients with multivessel disease and pre-existing Q wave myocardial infarction (table )13–15
and three were restricted to single vessel disease and patients without previous Q wave infarction.16–18
The success rates for percutaneous transluminal coronary angioplasty varied between 80% and 100%. Rates of complications varied between 0.01% and 2.8% for myocardial infarction and between 1.5% and 2.8% for immediate coronary artery bypass grafting. There was one death related to percutaneous transluminal coronary angioplasty in one trial.14
Antithrombotic prophylaxis in the angioplasty study groups reflects practice at the time the studies were conducted. During percutaneous transluminal coronary angioplasty patients received only heparin, and stents were used in a minority of patients in only one study.14
In all trials medical treatment included administration of antiplatelet agents, β blockers, nitrates, and calcium channel blockers, but only one trial used aggressive lipid lowering treatment.15
The mean follow up time in the 953 patients treated with angioplasty and 951 with medical treatment ranged between 6 and 57 months, and quality scores varied between 2 and 4 points. Low quality scores were mainly due to lack of concealment and documentation of blinded outcome assessment for clinical end points.
Baseline characteristics and criteria for inclusion in randomised controlled trials of percutaneous transluminal coronary angioplasty (PCTA) compared with medical treatment in non-acute coronary heart disease
Table provides information about event rates in single trials. The pooled risk ratio for angina in patients with percutaneous transluminal coronary angioplasty compared with medical treatment was 0.70 (95% confidence interval 0.50 to 0.98), but there was significant heterogeneity (P<0.001) (figure). The pooled absolute risk difference for angina was 0.17 (95% confidence interval 0.00 to 0.32; test of heterogeneity P<0.001). The risk ratio for fatal and non-fatal myocardial infarction in patients treated with percutaneous transluminal coronary angioplasty compared with medical treatment was 1.42 (0.90 to 2.25; test of heterogeneity P>0.10), and that for death was 1.32 (0.65 to 2.70; test of heterogeneity P>0.20). There was an increased risk of coronary artery bypass grafting (risk ratio 1.59; 1.09 to 2.32; test of heterogeneity P>0.10) and an increased risk of repeated angioplasty (1.29; 0.71 to 3.36), though the results for repeated angioplasty proved variable across studies (test of heterogeneity P<0.001).
Table 2 Number of clinical events and relative risks (95% confidence intervals) for different end points in randomised controlled trials of percutaneous transluminal coronary angioplasty (PCTA) compared with medical treatment in non-acute coronary heart (more ...)
We explored heterogeneity and examined risk ratios and 95% confidence intervals in trials with single versus multivessel percutaneous transluminal coronary angioplasty and trials with different quality scores (
3 or higher) and duration of follow up (
24 months or longer). For all end points (angina, death, myocardial infarction, coronary artery bypass grafting, and repeated percutaneous transluminal coronary angioplasty) we found similar summary estimates, and P values for the difference in summary estimates in each pair of subgroups were all above 0.10, indicating no significantly different effect sizes in subgroup summary estimates (data not shown).