Enabling physicians with clinical data harvested from Internet accessible resources

Problem. Timely and effective delivery of information to health care providers may facilitate more informed decisions and more accurate practice performance assessments. This research involved harvesting clinical data, delivering information to providers, and evaluating the information delivery system.; Aims. The research determined the information needs of health care providers, constructed a strategic conceptual data model for a warehouse of electronic medical record (EMR) data, designed an implementation prototype model for demonstration, deployed and implemented the system through secure Web technologies for a population of patients with diabetes mellitus, and conducted usability evaluations of the system.; Methods. A survey for primary health care providers was constructed, pilot tested and distributed to determine knowledge, information needs, and beliefs regarding specific components of guidelines for diabetes mellitus care. A conceptual data model for the clinical research data was constructed. A database conceptual schema was designed and implemented using the MySQL database management system. Data from the Practice Partner Research Network (PPRNet) were extracted from the PPRNet SAS data files and used to populate the new system. A secure Web interface to the database system was created and deployed using Apache web server, HTML forms, PERL scripts, the CGI protocol, and .htaccess files. Queries of recommended guidelines for blood pressure, hemoglobin AlC and LDL cholesterol were implemented. Users evaluated the system and were asked their perceptions about the usefulness of the system and ideas for improvement.; Results and significance. The Diabetes Mellitus Information Survey was completed by 36 respondents. Seventy-eight percent of the respondents indicated that diabetes care ranked among the three principal reasons for office visits. Respondents were knowledgeable about the recommended guidelines for diabetes care. PPRNet members were more familiar with the targets for HDL cholesterol (p=0.02) and influenza vaccine (p=0.039) than non-members. The tests that received highest importance ratings were blood pressure (92% of respondents), Lipids (94%) and Hemoglobin A1C (94%). These were also considered the most essential data elements for treatment of diabetes and were included in the implementation model. The conceptual data model was constructed. The implementation model was successfully employed. Formative evaluation by eight PPRNet members in seven practices found the system to be easy to use, intuitive, and functional. This research demonstrated the practicality of harvesting data captured in EMR systems during the course of daily practice and involved potential users of an electronic tool in the development and continual evaluation and refinement of the system.