Developing a Mental Health Data Resource

Specific Objectives for the Proposed Research Network – The proposed network will establish the infrastructure necessary to efficiently conduct effectiveness research across the full range of mental health conditions and treatments. We will consider all stages of the process from initial observational studies, to selection of priority research questions, to efficient conduct of effectiveness and cost-effectiveness trials, to disseminating research results and implementing effective treatments. Long-term objectives include:

1) Develop large linked databases characterizing treatment exposures and outcomes – While observational studies have long used insurance claims records and pharmacy refill records, the growing use of electronic health records will allow access to rich data regarding treatment quality and outcomes for large, diverse populations. Moreover, rapid access to records data will allow research recruitment and data collection as treatment is occurring. New data extraction methods such as Natural Language Processing can identify and extract discrete data regarding clinical characteristics from text of clinical notes. As demonstrated in our previous research, linked records databases can be used for:

• Rapid and efficient enrollment into effectiveness trial

• Observational studies of treatment patterns and treatment effectivenes

• Identification of priority effectiveness questions for randomized trial

• Assessment of economic outcomes of treatme

• Studies of implementation, guideline dissemination, and policy evaluatin

2) Develop an efficient infrastructure for clinical outcome assessment – Many effectiveness questions can only be addressed by direct collection of clinical outcomes data. Efficiency of assessment is especially important in effectiveness research in order to reduce costs of research and minimize burden on study participants. We anticipate that effectiveness research will often rely on brief assessments administered via new communication technologies such as web-based questionnaires, electronic secure messaging, or interactive voice response. Methodologic research will be necessary to examine potential biases due to differential participation and use of brief assessments.

3) Use population-based data to select priority effectiveness questions – Representative data from real-world samples can identify the treatment decisions most often encountered in practice. In many cases, this population-based approach will identify different questions than those identified by efficacy researchers or commercial sponsors. For example, this approach would focus depression research on improving treatment selection and promoting treatment adherence rather than testing efficacy of individual drugs or resolving interdisciplinary debates regarding the efficacy of medication and psychotherapy.

4) Develop an efficient infrastructure for targeted clinical trials – If randomized trials, rather than market forces, are to guide dissemination of new treatments, findings must be available rapidly. Given the large number of effectiveness questions we must address, costs of study recruitment, treatment delivery, and outcome assessment must be dramatically reduced. New research models are necessary to:

• Rapidly and efficiently identify and enroll representative samples of patients treated in community settings using electronic medical records and population-based clinical registries

• Embed random assignment into care delivered by community providers and covered by health insurance

5) Develop infrastructure and capacity for assessing cost and economic impact of mental health treatment – As discussed above, a comprehensive assessment of treatment cost-effectiveness should include:

• Development of input cost assessment methods to estimate true costs of intervention delivery

• Use of health system claims and cost accounting records to assess effects on health services costs

• Use of survey and/or computerized re

<< Back to the list of CRN projects