Author + information
- Anthony N. DeMaria, MD, MACC, Editor-in-Chief, Journal of the American College of Cardiology⁎ ()
- ↵⁎Address correspondence to:
Dr. Anthony N. DeMaria, Editor-in-Chief, Journal of the American College of Cardiology, 3655 Nobel Drive, Suite 630, San Diego, California 92112
Cardiovascular specialists have become justifiably proud of their commitment to evidence-based medicine and clinical trials. More large-scale randomized prospective clinical trials (RCTs) are performed in cardiology than in any other field, and contemporary practice is nearly always based upon at least one such study. By controlling variables, minimizing subconscious bias, and insuring data of adequate power and statistical validity, megatrials provide results that are seen as definitive in areas of equipoise. They convey guidance to physicians whose individual practices could not reveal the subtle benefits (or detriments) of specific management alternatives that can be delineated by large study populations. We have come to rely on trial data so much that trainees sometimes now feel paralyzed to make a decision in the absence of data from an RCT.
The proliferation of clinical trials has raised questions regarding the role of clinical judgment in contemporary medicine and how best to incorporate the results of trials into clinical practice. In fact, compared with the wealth of data from RCTs, there has been a paucity of studies regarding how to convert these findings into “real-world” settings (1). For all of their attributes, clinical trials have a number of limitations. In my view, the time has come to acknowledge these limitations and re-emphasize the important role of clinical judgment in interpreting research findings and applying them to individual patients.
As previously stated, clinical trials have a number of limitations. The end points listed are not always optimal. Patients are interested in survival, lack of morbidity, and quality of life; society also values cost effectiveness. However, trials often use surrogate end points. In addition to often being of little importance to patients, surrogates may lead to misleading and erroneous conclusions. The classic example is CAST (Cardiac Arrhythmia Suppression Trial), where suppression of premature ventricular contractions resulted in more mortality rather than less (2). More recently, torcetrapib decreased survival despite increasing high-density lipoprotein (3). Trials often employ composite end points that, although they enable assessment of nonfatal events, entail a number of shortcomings. The individual events are not always of similar magnitude (e.g., death vs. rehospitalization) and some events, such as the need for a subsequent procedure, are significant primarily in regard to timing. Clinical trials often employ extensive subgroup analysis. However, multiple testing can lead to chance findings; the ISIS (International Study of Infarct Survival) trial demonstrated variability in the efficacy of aspirin in groups identified by their astrological sign (4). To be of value, analysis of subgroups should be pre-specified and involve adequate sample size.
A number of difficulties arise when applying the findings of clinical trials to everyday practice. Therapies may have multiple effects that may not be tested. The cyclooxygenase-2 inhibitor VIOXX (Merck, Whitehouse Station, New Jersey) had adverse cardiovascular effects that may have been somewhat obscured by the focus upon gastrointestinal end points and would be of particular importance in patients with coronary artery disease. Drug–drug interactions are common and may be encountered more frequently in the “real world” than in trials. Not all agents in a given class may share the same efficacy, a characteristic exhibited by beta-blockers. In addition, the benefit of therapy may be time limited, and this interval may not be examined by the trial. The risk of late stent thrombosis of drug-eluting stents presents such an example.
The biggest limitation of clinical trials is that they deal with rigidly-defined patient populations in a highly-controlled fashion. My colleague (and JACC Associate Editor) Barry Greenberg has likened trials to “Muzak” (bland background music sometimes referred to as “elevator” or “telephone on hold” music) as compared with clinical practice, which is analogous to Mozart. Clinical practice deals with broad patient populations from all strata of society, with diverse etiologies, variable durations and degrees of disease, and frequent comorbidities. Like Mozart, it is heterogeneous and expansive. In contrast, RCTs usually deal with narrow populations with restricted etiologies and severity of disease in whom most comorbidities have been excluded. Like Muzak, they tend to be homogenous and confined. It is not surprising, therefore, that the results of clinical trials are frequently not easily translated to individual patients.
The confined nature of clinical trials can result in some important limitations. For example, the SOLVD (Studies of Left Ventricular Dysfunction) trial (5), which demonstrated the benefit of angiotensin-converting enzyme inhibition with enalapril, excluded patients with an ejection fraction ≥35%, age >79 years, and creatinine >2 mg/dl. Heart failure patients encountered in clinical practice often manifest these characteristics. The SOLVD trial enrolled 80% males. In the recent landmark COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) trial, 36,000 patients were screened, of which 3,071 were eligible to participate and 2,287 (6.3%) were ultimately enrolled (6). The implications of such potential enrollment bias are obvious.
Another factor complicating the application of clinical trial results to practice is the issue of lack of publication of negative trials. Studies in which the intervention examined is not of benefit are less attractive to publish for both authors and editors, and in fact often are never published. Thus, these negative results may not be available to offset positive findings, resulting in a skewed impression on the part of the physicians. While mandated registration of trials may alleviate some of the problems of negative studies, it remains uncertain whether such results will have comparable impact to well-publicized positive data.
In addition to the aforementioned issues involved in translating trial findings to practice, problems may be encountered when physicians apply results in a fashion not followed in the study. After publication of the RALES (Randomized Aldactone Evaluation Study) trial, an increase in the use of spironolactone was observed in heart failure patients who were treated outside of the boundaries specified in the protocol (7). Such management resulted in an increase in complications compared with those encountered by patients in the RALES trial.
Given the issues delineated above, it appears that an important role for clinical judgment should exist in applying the data from clinical trials. In fact, evidence exists to validate this concept. In a clever experiment recently published in JACC, Pereira et al. (8) from MASS (Medicine, Angioplasty, or Surgery Study) II had 2 cardiologists indicate their treatment preference prior to randomizing patients to percutaneous or surgical revascularization. Patients who were randomized to treatment concordant with the preference of the cardiologists had a lesser event rate than those in whom the form of revascularization was discordant with the wishes of the cardiologists. These data are consistent with findings from the EAST (Emory Angioplasty versus Surgery Trial) (9) and BARI (Bypass Angioplasty Revascularization Investigation) (10) trials, in which patients qualifying for enrollment but who participated only in the registry did slightly better than those patients who were randomized. In aggregate, these data support the value of clinical judgment in synthesizing trial data and patient factors to select optimal patient management.
At this point, it should be emphasized that I am a proponent of clinical trials and the evidence-based medicine that can result. It is clear that retrospective observational studies are subject to the influence of uncontrolled variables, and that no single physician can catalogue sufficient observations from his or her own practice to detect modest changes in benefit or risk. Numerous studies, such as the ACC GAP (American College of Cardiology Guidelines Applied in Practice) project, have demonstrated that guidelines-based standardized care is associated with a reduction in mortality from acute myocardial infarction (11). Neither would I contest that therapies of proven efficacy are often underutilized. I believe that clinical trials provide very valuable data to guide patient care decisions. I just believe that the data from RCTs represents the beginning of the decision-making process, not the end.
As I said at the beginning of this piece, cardiovascular specialists should be justifiably proud of their leadership in performing clinical trials. However, the importance and infallibility of clinical trials has likely been overemphasized at the expense of clinical judgment. The megatrials are far from perfect, and at best provide guidance for the care of individual patients. The experience and wisdom of a thoughtful physician can make an important contribution to the application of the evidence base that is available. Although this may not be welcomed by some outside of medicine, it should be of great comfort to those of us inside the profession.
- American College of Cardiology Foundation
- DeMets D.L.,
- Califf R.M.
- Cardiac Arrhythmia Suppression Trial (CAST) Investigators
- Bozkurt B.,
- Agoston I.,
- Knowlton A.A.
- Pereira A.C.,
- Lopes N.H.M.,
- Soares P.R.,
- et al.
- Eagle K.A.,
- Montoye C.K.,
- Riba A.L.,
- et al.