Traditionally, randomized clinical trials (RCTs) have sought to ascertain whether an experimental treatment is superior to standard treatment or placebo in improving quality of life or preventing morbid or mortal events—what we will refer to as effectiveness outcomes. In these superiority trials, the primary objective is to determine the magnitude of increased benefit of the experimental intervention over standard therapy on effectiveness outcomes.
You are an internist seeing a 51-year-old woman with severe osteoarthritis and limited mobility who presents with progressive dyspnea for a 3-day period. She is subjectively in distress, with a pulse of 105/min, a respiratory rate of 28/min, and an arterial oxygen saturation of 85% while breathing room air. Aside from her arthritis, the physical examination findings are unremarkable, and a lower-extremity examination reveals no sign of deep venous thrombosis. A computed tomographic (CT) pulmonary angiogram reveals unequivocal clot in 2 lobar arteries.
Recently, you have been treating patients with deep venous thrombosis without hospital admission using low-molecular-weight heparin (LMWH) administration in the outpatient setting. You are less comfortable not admitting a patient in the more dangerous setting of pulmonary embolism. You recall receiving, from the updating service to which you subscribe (see Chapter 5, Finding Current Best Evidence), a recent randomized trial that addressed this issue. Before discussing the issue of inpatient vs outpatient treatment with your patient, you quickly review the article.1 In doing so, you find that the trial tested for noninferiority and you wonder, as you begin to read the methods and results, if there are special issues you should consider when using this article to guide your clinical care.
Recently, another paradigm has emerged that offers novel experimental treatments not on the basis of superiority in effectiveness outcomes but instead because they reduce harms or other treatment burdens relative to standard treatment. In modern medicine, clinicians are fortunate to have many effective treatments; unfortunately, these treatments are often associated with harms, inconvenience, or excessive cost. For these interventions, reducing treatment burden, including limitations and inconvenience, becomes a legitimate goal of innovative therapy.
In such instances, a question arises: can clinicians be confident that the experimental treatment's impact on effectiveness outcomes—the prime reason for wanting to prescribe it—is sufficiently close to that of standard treatment that they are comfortable substituting it for the existing standard? In technical terms, is the novel treatment noninferior to the standard treatment?
Noninferiority trials provide an alternative to equivalence trials, which endeavor to establish that an experimental treatment is neither better nor worse (beyond a specified margin) than the standard. In contrast, the noninferiority trialist is unconcerned if the experimental treatment is better as long as it is “not much worse.” Perhaps illustrating the limitations of the term, a noninferior treatment may thus be inferior, just not so inferior that it would cause concern. How much worse (ie, how much less ...