1
VOL. 62, NO. 8, AUGUST 2001 Editorial Where Did We Go Wrong? It is easy to explain why a clinical trial worked. You may have demonstrated a true difference, you had an unconscious bias, someone committed fraud, or you were just lucky. Determining why you were unable to demonstrate a scientific truth requires much more work. One system for evaluating clinical trials uses a checklist. 1'2 Because so many aspects of a clinical trial depend on the initial scientific question (eg, the study population, selection of the study drug and comparator agents, other treatment modalities allowed, bias control, and assessment of whether the goals of the intervention have been reached), the first step is to ask yourself whether you formulated the question properly. Did you have all the necessary background information and did you analyze it correctly? Second, consider the end points. Did they represent the answer you intended to obtain? Could they be measured reliably? Could other factors have obscured the end points or somehow made them more difficult to assess? Third, consider the study design. Was it appro- priate for the question under study? Some clinical trial investigators like to propose arcane designs, often to illustrate their knowledge of the field. Rather, the study design should be the most appropriate one for the initial scientific query. It may decrease the likelihood of losing focus and making errors. It is hoped that investigators will rarely have to perform an autopsy on a failed clinical trial. A careful initial analysis of the scientific question, end points, and study design can help avoid it. Michael Weintraub, MD Editor-in-Chief REFERENCES 1. Weintraub M. How to evaluate reports of clinical trials. Pharm Ther. 1990;15:1463- 1476. 2. Weintraub M. Revisiting clinical trial evaluations: A more advanced checklist. Pharm Ther. 1999;24:15--32. 557

Where did we go wrong?

Embed Size (px)

Citation preview

Page 1: Where did we go wrong?

VOL. 62, NO. 8, AUGUST 2001

Editorial

Where Did We Go Wrong? It is easy to explain why a clinical trial worked. You may have demonstrated a true difference, you had an unconscious bias, someone committed fraud, or you were just lucky. Determining why you were unable to demonstrate a scientific truth requires much more work.

One system for evaluating clinical trials uses a checklist. 1'2 Because so many aspects of a clinical trial depend on the initial scientific question (eg, the study population, selection of the study drug and comparator agents, other t reatment modalities allowed, bias control, and assessment of whether the goals of the intervention have been reached), the first step is to ask yourself whether you formulated the question properly. Did you have all the necessary background information and did you analyze it correctly? Second, consider the end points. Did they represent the answer you intended to obtain? Could they be measured reliably? Could other factors have obscured the end points or somehow made them more difficult to assess? Third, consider the study design. Was it appro- priate for the question under study? Some clinical trial investigators like to propose arcane designs, often to illustrate their knowledge of the field. Rather, the study design should be the most appropriate one for the initial scientific query. It may decrease the likelihood of losing focus and making errors.

It is hoped that investigators will rarely have to perform an autopsy on a failed clinical trial. A careful initial analysis of the scientific question, end points, and study design can help avoid it.

Michael Weintraub, MD Editor-in-Chief

REFERENCES 1. Weintraub M. How to evaluate reports of clinical trials. Pharm Ther. 1990;15:1463-

1476. 2. Weintraub M. Revisiting clinical trial evaluations: A more advanced checklist. Pharm

Ther. 1999;24:15--32.

557