Randomized clinical trials have had an immense effect on the practice of medicine. However, in order to answer the questions posed in such trials, relevant and sufficient patient populations and treatments must be identified.
Large RCTs, or megatrials, can identify small differences in populations but tend to exaggerate their significance. Several megatrials have questionable relevance to clinical care.
The recent COMMIT/CCS-2 study examined the role of early intravenous metoprolol in nearly 46,000 Chinese patients with Killip class I-III ST-segment-elevation MI (STEMI). But in contrast to U.S. treatment practices, fibrinolysis was common and percutaneous coronary intervention (PCI) was uncommon. In addition, the trial design included intravenous metoprolol for patients with Killip class III with heart failure, a treatment that many U.S. physicians would have been reluctant to give. The data indicated that this reluctance was well founded. Metoprolol caused an increase in death due to shock and heart failure in the Killip class III patients, which counterbalanced the decrease in arrhythmic deaths observed in the Killip I and II patients. Overall, there was no benefit associated with intravenous metoprolol in STEMI patients.
The GUSTO I trial, reported in 1993, randomized 41,021 patients with STEMI to compare the benefit of thrombolysis with streptokinase with accelerated tissue plasminogen activator (TPA), both combined with intravenous heparin. The 30-day mortality was 7.4% in streptokinase-treated patients and 6.3% in TPA patients. Despite this meager absolute difference of 1.1% (P = .001) and in the face of increased hemorrhagic strokes in the TPA-treated patients (P = .03), TPA, at a cost 10 times that of streptokinase, became the U.S. standard of therapy, while streptokinase remains the most common thrombolytic therapy in the rest of the world.
The HOPE trial enrolled 9,297 patients to test the benefit of the ACE inhibitor ramipril in patients at a high risk of CAD on the composite end point of ischemic events including death. In 2000, after 5 years of follow-up, the event rates were 17.8% in the placebo group and 14.0% in the ramipril group (P < .001). These results led to the rapid inclusion of ACE inhibitor therapy in any patients with or at risk of CAD. But meanwhile, another RCT, PEACE, had enrolled 8,290 similar patients to test the benefit of the ACE inhibitor trandolapril in patients who were being treated with β-blockers, statins, and PCI. PEACE reported its findings in 2004 and found no benefit of ACE inhibitors, largely due the more aggressive concomitant therapy, which resulted in a lower placebo event rate. In just a few short years, therapy had changed so rapidly that ACE inhibitors no longer appeared to have an impact on the outcome in patients at risk of ischemic events. In retrospect, HOPE was dated even before it was reported.
RCTs continue to impact on our bedside decisions. These experiences with megatrials, however, give reason to be critical of their importance in the care of our patients. It is best to remember that today's scientific “truths” may be shown to be “false” tomorrow.