Heterogeneous effect of an emergency department expert charting system

Study objective: We compare results from different modules (occupational exposure to blood and body fluids, low back pain, and fever in children) of the Emergency Department Expert Charting System. Each module of this electronic medical record provides real-time advice based on clinical guidelines embedded in the software. Methods: We used a staggered off-on-off interrupted time-series design with an intent-to-treat analysis to implement the Emergency Department Expert Charting System in a university hospital emergency department for the treatment of fever in children, low back pain, and occupational exposure to blood and body fluids. We measured the quality of documentation as the percentage of essential items contained in the medical record and discharge instructions, the percentage of appropriate testing and treatment decisions, median charges per patient visit, physician satisfaction by pre-experiment and postexperiment questionnaires, and patient satisfaction by telephone questionnaire. Results: The Emergency Department Expert Charting System improved documentation rates for all modules. The Emergency Department Expert Charting System consistently improved the appropriateness of diagnostic testing and treatment decisions for patients with occupational exposure to blood and body fluids while decreasing median charges. For the low back pain and fever in children modules, improvements in appropriateness of testing and treatment were less consistent and did not result in a decrease in charges. Although physicians were generally supportive of the intervention, the physicians' use of the Emergency Department Expert Charting System and satisfaction with the modules were greatest with occupational exposure to blood and body fluids and least with fever in children. For each presenting complaint, mean patient satisfaction was highest during the Emergency Department Expert Charting System phase. Conclusion: The delivery of guidelines through an electronic medical record with background decision support improved documentation, patient care, and patient satisfaction, although effects were heterogeneous across presenting complaints. The optimal guideline implementation strategy likely varies with the nature of the clinical problem and the type of health care delivery setting. For selected problems, delivering guidelines in this format through the Internet holds great promise for modifying physician behavior and improving care (see http://www.needlestick.mednet.ucla.edu).