This is the third submission of a application to conduct analyses of data collected by the NIMH collaborative study of the psychobiology of depression. The investigator has worked closely with clinical investigators in this study, and now proposes a further extension of the statistical work to accomplish three aims: 1) to estimate the effects of different levels of somatic treatment during the maintenance phase; 2) to estimate the effects of starting and continuing treatment during episodes of illness; and 3) to adjust estimates of predictive value of prognostic factors for the confounding of treatment and prognosis. Methods include the proportional hazards model with a time-varying treatment indicator, with a model-based adjustment for the propensity score. This strategy is being used to address the problem of imbalance in prognostic factors across non-randomly chosen treatments.The importance of this work is in providing information on treatment effects on depression in the real world during continuation and maintenance phases. The information from this study could prove very useful to clinicians in making decisions about how long to continue medication, at what dosage level, and about when and how to restart medication which has been discontinued.In addition, the analytic strategies developed and tested for this dataset will have implications for other nonexperimental, observational studies of increasing interest to services researchers.
Showing the most recent 10 out of 25 publications
