Fleming and DeMets [1] do an excellent job of pointing out the inadequacy of surrogate markers as appropriate outcomes for phase III trials. Surrogate markers are also inappropriate for clinical practice, yet they serve as the foundation on which many of the daily decisions in medicine are made.

Managed care is forcing clinicians to re-evaluate their decision making with well-focused questions on outcomes of care. Many clinicians try to deflect these queries as merely being based on finances. These questions, however, force us to examine our practices to determine whether our interventions truly are in the best interests of our patients.

When we try to decide whether a treatment offers a true clinical benefit, we should be able to turn to the literature to find "patient-oriented evidence that matters" (POEMs) [2]. As Fleming and DeMets point out, however, POEMs are unavailable to support most of what we do for patients. And when we do find clinically appropriate information, we tend to ignore it when it contradicts what we "know" is correct, a failing termed "reverse gullibility" [3].

A major change in our belief system must occur before our paradigm changes from basing patient care decisions on what should work, as established by effects on surrogate end points, to what does work. Designers of clinical trials are not the only ones who need to refocus their sights; all of medicine has to let go of the reductionistic approach to patient care (focusing on patients' component parts) that has been embraced for the past 80 years. Instead, we need to adopt the goal that medicine must provide care that has been shown, through actual study, to help patients live longer, healthier, more productive, symptom-free lives.