On the clinically important difference
ACP J Club. 1993 May-June;118:94. doi:10.7326/ACPJC-1993-118-3-094a
To the Editor
We write to add a consideration to the recent editorial by Cook and Sackett (1), which presented an approach to deciding whether or not to change one's practice on the basis of new information about a therapeutic intervention. It is generally recognized that the beneficial effects of an intervention delivered in everyday clinical practice are less than those observed in the ideal setting of the randomized control trial (the efficacy-effectiveness differential) (2). It is less often noted, however, that the frequency of adverse effects of an intervention may be higher in routine clinical practice than in the closely supervised and highly structured research setting.
This phenomenon has been observed for many invasive procedures, such as angioplasty and coronary artery bypass surgery, and is also a concern with certain pharmacologic interventions. For example, the 24-month cumulative incidence of bleeding complications in a recent placebo-controlled randomized trial of warfarin in chronic atrial fibrillation was 6.3% (3). By contrast, the 24-month incidence of bleeding was 25% in a large hospital-based observational study of outpatients taking warfarin (). It is therefore not sufficient for readers to decide on the basis of the research report alone whether the benefits of a new treatment are offset by the risks. We suggest that before implementing a new treatment, readers should estimate what the rate of adverse events is likely to be in their practice or in settings similar to theirs. If no data from either setting are available, readers should presume that the adverse event rate in the research report represents the lower bound of the complication rate they are likely to observe in their own practice. This is especially true if the intervention requires the practitioner to learn a new technique or if strict monitoring of patients is known to reduce complication rates.
Carol M. Ashton, MD, MPH
Nelda P. Wray, MD, MPH
Department of Veterans Affairs Medical Center
The thoughtful letter from Drs. Ashton and Wray provides an opportunity for us to re-emphasize the importance of incorporating local factors into the calculus that clinicians use to decide whether to start down a particular therapeutic path. These local factors include the clinical competency of the local health professionals (e.g., the surgeons doing carotid endarterectomy), the accuracy and responsiveness of the local laboratory (e.g., in detecting and reporting prolonged international normalized ratios), and the compliance of patients in one's own practice (which, if it is lower than the quite high compliance usually attained in randomized trials, will reduce both the effectiveness and the risk for dose-dependent adverse effects). Above all, clinicians must scrutinize their own ability to apply, monitor, and, if necessary, modify the treatment under consideration to maximize the potential for good and minimize the risks for harm.
Deborah J. Cook, MD, MSc
David L. Sackett, MD, MSc
Hamilton, Ontario, Canada
1. Cook D, Sackett DL.On the Clinically Important Difference. ACP J Club. 1992 Sep-Oct:A16-7 (Ann Intern Med. vol 117, suppl 2).
3. Pedersen P, Boysen G, Godfredson J, Andersen ED, Andersen B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. Lancet. 1989;1:175-9.