Expect Higher Standards in Research

Article excerpt

Peer-reviewed research should provide clear guidance to clinical practice. However, there is considerable information that we as physicians have a right to expect from peer-reviewed research but are not getting. This deficiency makes it harder for us to practice evidence-based medicine.

We are not paying enough attention to research methods, either as investigators or as skeptical consumers of the literature.

A great deal of what we do in medicine depends on the results of randomized, controlled trials. Unfortunately, many of the researchers performing trials today fail to ask important questions before beginning their studies. For example, researchers need to ask themselves what they think will happen in the study and articulate their ideas as clearly stated hypotheses. Also, they need to determine what would constitute a clinically significant result. It is not enough to report that an outcome is statistically significant. Physicians need to know if the information they read in a journal is meaningful enough to change the way they practice.

The Food and Drug Administration could help to improve the information that is reported by changing its requirements for drug approval. Currently, the FDA requires two positive trials. That means two trials where the experimental drug differs from the control at a P value of less than .05, a relatively easy result to obtain if enough subjects are included in the trial and the sample of subjects is sufficiently "purified" to keep noise out of the data. The agency could ask for more information that gets to the question of clinical relevance. It could even begin to base its approval decisions on such information.

But even absent a change in government policy, there are some relatively simple definitions that researchers could use to judge success. One such measure is the number needed to treat. This measure allows the clinician to see how many patients would have to be treated to see one additional positive outcome. …


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.