This application seeks support for a team of statisticians and mental health scientists to collaborate on the development and validation of regression methods for multiple outcomes collected in longitudinal studies of mental health services. The specific methods to be developed will address the needs of ongoing health services research but also will have application in the basic mental health disciplines such as genetics, clinical trials and epidemiology. The methodologic advances from this research will enable mental health scientists in these disciplines to more efficiently investigate mental disorders and their patterns of care. The three specific aims are:1. To develop, validate, and disseminate new regression methods for multiple outcome measurements collected in longitudinal studies. The specific methods to be developed are for: multiple outcomes observed repeatedly through time; multiple survival (time-to-event) outcomes; and combinations of longitudinal and survival data.2. To compare the application of new and existing methods to mental health services data sets to assess their relative advantages and disadvantages and to disseminate our findings. 3. To develop stand- alone statistical software for new and existing methods unifying survival and longitudinal data analyses and also to integrate this software in Splus, a widely available statistical package with bridges to other popular software including SAS and SPSS. The co- investigators will work as a team that will: analyze diverse health services data sets using existing methods; identify and disseminate common methodologic barriers to scientific inferences; propose improved statistical methods; investigate their theoretical properties; implement the new procedures in computer software; compare the new and existing methods through analyses of our data sets; and disseminate the results to both the mental health services and statistical communities. The new methodologies can contribute to our understanding of mental disorders and patterns of care and ultimately to the improvement of mental health status.
Showing the most recent 10 out of 21 publications
