The report by Shapiro et al1 in this issue of the Archives raises a number of issues regarding multicenter collaborative studies and their analyses. Our general comments pertain primarily to publicly funded collaborative trials. Multicenter studies sponsored by pharmaceutical companies or other private sources raise a different set of issues that are discussed elsewhere.2
See also pp 397 and 401.
The report by Shapiro et al1 stemmed from a request from the authors to reanalyze data from the National Institute of Mental Health (NIMH) (Rockville, Md) Collaborative Study on Maintenance Drug Therapy in Affective Disorders3 by using a life-table regression model. The authors had not participated in the collaborative study or in the preparation of earlier reports of findings. Their results demonstrate that large collaborative studies can benefit from infusion of different analytic approaches from outside the collaborative study group. Unfortunately, some multicenter collaborative projects often