While many effective depression treatments (medications and psychotherapies) are available, individuals vary widely in response to specific treatments. Unfortunately, our ability to match any individual to the most effective treatment is hardly better than chance. Research on personalizing treatment has typically used data from randomized trials, attempting to identify stable individual characteristics (biomarkers, genetic variations, etc) that predict differential response to specific treatments. Previous research to predict treatment response may have yielded disappointing results because the response patterns studied are not stable within individuals across time. Observational data from large population-based samples offer a new model for research to personalize depression treatment. This approach would allow study of tens of thousands of individuals, rather than the hundreds typically enrolled in clinical trials. Most important, data for multiple treatment episodes per person could identify patterns of response that are stable within individuals across episodes. Identifying stable response patterns or phenotypes is a prerequisite to identifying genotypic or other predictors of response. Pilot work is necessary to examine the feasibility of this approach and to address key methodolgic questions. Data from Group Health Cooperative, a large prepaid health plan, will be used to identify 2 samples of patients: 400 adults completing acute-phase antidepressant treatment (Sample #1) and 200 patients from recent depression trials who have since initiated another treatment episode (Sample #2). Patients in both samples will be asked to complete a detailed assessment regarding treatment response in current and past episodes and to contribute saliva samples for future genetic studies. Computerized records will be used to characterize all depression treatment over the past 10 years. These data will be used to:
Aim 1 - Evaluate the feasibility of proposed study methods - Data from sample #1 will be used to examine: """""""" Use of medical records data to facilitate identification of favorable and unfavorable treatment response """""""" Participation rate and biases due to non-participation in either clinical assessments or genetic testing Aim 2 - Examine sources of bias in observational studies - Data from sample #2 will be used to examine: """""""" Potential bias due to influence of past treatment experience on subsequent treatment selection """""""" Accuracy of recall for past treatment outcome and influence of current mood and time lapsed on recall Aim 3 - Develop statistical methods to identify person-level predictors of treatment response - Data from both samples will be used to develop and evaluate hierarchical modeling approaches to distinguish episode-level, person-level, and treatment-level predictors of favorable and unfavorable treatment response.
Aim 4 - Plan for a multisite study using population-based data from large health plans - If pilot work supports the feasibility and utility of this new research model, we anticipate developing a research consortium of large health systems to conduct multi-site studies evaluating personalized approaches to depression treatment. Public Health Relevance: Large observational studies using longitudinal data have the potential to transform research on personalizing depression treatment. We propose a set of targeted methodologic studies to prepare for a multi-site research consortium focused on predicting response to specific depression treatments.
Large observational studies using longitudinal data have the potential to transform research on personalizing depression treatment. We propose a set of targeted methodologic studies to prepare for a multi-site research consortium focused on predicting response to specific depression treatments.
| Simon, Gregory E; Rutter, Carolyn M; Stewart, Christine et al. (2012) Response to past depression treatments is not accurately recalled: comparison of structured recall and patient health questionnaire scores in medical records. J Clin Psychiatry 73:1503-8 |
| Simon, Gregory (2011) What little we know about tailoring depression treatment for individual patients. Depress Anxiety 28:435-8 |
| Simon, Gregory E; Perlis, Roy H (2010) Personalized medicine for depression: can we match patients with treatments? Am J Psychiatry 167:1445-55 |
