Survival did not differ for patients whether they were in or out of the Coronary Artery Surgery Study (CASS)
ACP J Club. 1991 Mar-April;114:60. doi:10.7326/ACPJC-1991-114-2-060
Chaitman BR, Ryan TJ, Kronmal RA, et al., and the CASS Investigators Coronary Artery Surgery Study (CASS): comparability of 10 year survival in randomized and randomizable patients. J Am Coll Cardiol. 1990 Nov 1;16:1071-8.
To investigate whether 10-year survival of patients with stable angina pectoris or myocardial infarction differs between those taking part in the randomized Coronary Artery Surgery Study (CASS) and those refusing to participate.
An inception cohort of patients randomizable for CASS to surgical or medical treatment was followed for 10 years.
11 institutions participating in CASS from August 1975 to May 1979. The referral pattern was not described.
Of 16 626 patients having angiography, 2099 met randomization criteria. 780 patients were randomized: 390 to medical treatment and 390 to surgical treatment. 745 randomizable patients, who refused entry and did not have surgery within 90 days of cardiac catheterization or within the period in which 95% of patients in that institution had surgery, were considered to be in the eligible but not randomized ("randomizable") medical group; 570 randomizable patients who had surgery were included in the randomizable surgical group. 9.25 years after the end of randomization, survival status was known for > 99% of all patients.
Assessment of prognostic factors
Vessels were considered diseased if angiography showed ≥ 50% stenosis in the left main coronary artery or ≥ 70% stenosis in other main coronary arteries. (Of patients with left main coronary disease, ≥ 70% were excluded.)
Main outcome measure
10-year survival rate.
At entry the randomized group included more smokers and more patients with hypertension or diabetes than the randomizable group, which included more patients with proximal left coronary artery disease or single-vessel coronary disease.
No difference in survival existed between the randomized and randomizable patients in the medical groups (79% and 80%, respectively, P > 0.1) or surgical groups (82% and 81%, respectively, P > 0.1). Incidence of late coronary bypass surgery was similar for both medical groups, except that a greater percentage of randomized asymptomatic patients who had had a myocardial infarction required surgery (40% vs 28%, P = 0.02).
No differences existed in survival of patients who were free of angina pectoris after myocardial infarction or who had stable angina pectoris, whether treated surgically or medically. Participation in the randomized trial was not related to 10-year survival.
Source of funding: National Heart, Lung, and Blood Institute.
Address for article reprint: Dr. R.A. Kronmal, CASS Coordinating Center, University of Washington, 1107 NE 45th Street, Room 530, Seattle, WA 98105, USA.
The design and conduct of randomized trials have reached the point where a trial's results are accepted as constituting a valid representation of efficacy for patients within it; that is, the internal validity of a trial that meets current methodologic standards is seldom in doubt. Considerable interest now exists in the external validity of randomized trials: To what extent can the results of an internally valid trial be generalized and justifiably guide the treatment of patients outside the trial? One approach to studying (but not, I think, answering) this question, exemplified in this report from the CASS study, follows eligible but not randomized ("randomizable") patients and compares their outcomes with those of patients within the trial. When these are similar, as in this study, we feel reassured about the generalizability of its results. Indeed, the CASS investigators concluded that the follow-up of these eligible but not randomized patients "confirmed" the conclusions of their randomized trial.
As long as our understanding of human biology, disease mechanisms, and human behavior is imperfect (and if it were not, we would not need randomized trials!), neither this conclusion nor the general approach is on firm ground. That the results among "eligible but not randomized" CASS patients coincide with the trial results is clear, but I suggest that this can only constitute coincidence and never confirmation. More important, what if eligible but not randomized CASS patients fared better, or worse, than those in the trial? Would this mean that we should not use the trial to guide treatment decisions about our patients?
The most valid guides for deciding how to treat our patients come from patients inside, not outside, the CASS and other randomized trials, and the similar results reported here should no more comfort us than dissimilar results should discomfort us. Moreover, given the opportunity costs of tracking eligible but not randomized patients (which may divert energy from randomizing and following patients within trials), we need to reserve judgment on the validity and appropriateness of this approach.
David L. Sackett, MD, MSc
McMaster UniversityHamilton, Ontario, Canada