Systematic Errors in the Evaluation of Side Effects
CHARLES W. AVERY M.D.1,
BERTRAM P. IBELLE PH.D.,
BETTY ALLISON R.N., , and
NANCY MANDELL R.N.
1 85 Jefferson St., Hartford 14, Conn. 06103
Twenty-three inpatients at Hartford Hospital were selected for the pilot study of NIMH Collaborative Study III, testing the feasibility of high-dosage chlorpromazine in depressed patients. After random assignment to either a nonchecklist group or a checklist group, both groups were asked the following question: "Have you noticed any change in bodily function or had any physical complaints in the past week?" The nonchecklist group was asked no further questions, and only spontaneously reported symptoms were recorded. The checklist group was also asked specific questions from a list of possible drug side effects.
Statistical analysis in terms of both quantity and severity of side effects showed significantly greater numbers and greater severity of side effects reported in the checklist group. On the basis of these resuits the authors feel that unless reports of medication side effects are obtained in a uniform manner, the data collected under varying conditions of inquiry will not be comparable. Research findings based on such data could be invalid.
