OBJECTIVE: Cost-effective data collection is a critical component of pharmacoeconomic and outcomes studies, and is often a key limiter of study design and scope. The objectives of this presentation are to: 1) present a new method of clinical data collection in the community-based physician setting, and 2) present statistics on data collected in 1998. METHODS: This presentation focuses on collecting clinical information in the community-based physician setting. This information has broad utility, including support of the conduct of pharmacoeconomic and outcomes studies. The data collection process is built around community-based physicians’ preferred means of documenting patient encounters -- dictation and transcription. The data collection process begins with templated physician dictation to support discrete data element capture. New transcription technologies that provide for discrete data element capture, in addition to free text typing, are used to document the encounter and capture the data elements. Back-end data scrubbing, natural language encoding, and data mapping are then used to transform physician natural language into useable data. Techniques for validating data through source document sampling will also be presented. RESULTS: Data collected from cardiologists through September 1998 will be presented. Preliminary statistics include the following: 5,865 coronary artery disease patients were seen; 54.1% of these patients were on a statin. Of these patients 42.1% were on atorvastatin, 36.7% were on simvastatin, 11.2% were on pravastatin, and 10.0% were on other statins. Switching patterns between statins will be presented. Common switching reasons include formulary changes, ineffectiveness and side effects. CONCLUSIONS: Existing clinical documentation processes can be leveraged to collect more detailed and meaningful information for pharmacoeconomics and outcomes studies than traditional, transaction-based information sources.