Medical errors occur frequently when clinicians fail to respond to abnormal laboratory test results. This problem plays an especially important role in emergency departments, and other ambulatory settings because in many cases patients have been discharged by the time test results become available. An automated system to ensure follow-up for late-arriving test results may decrease rates of this type of medical error. This project will evaluate the effectiveness of a computerized system called the Automated Late-Arriving Results Monitoring System (ALARMS) in an emergency department. ALARMS gathers data from the hospital information system to generate a list of all late-arriving abnormal laboratory results. ALARMS prompts the user to respond to each result and then generates a note documenting the response, which becomes part of the medical record. The study hypothesis is that introduction of ALARMS will significantly improve rates of follow-up for late-arriving abnormal laboratory results. In the first phase, this project will evaluate the accuracy of ALARMS for identifying all late-arriving abnormal laboratory results, as compared to a manual review by researchers as the criterion standard. In the second phase, the emergency department's current rate of appropriate follow-up for late-arriving test results will be evaluated by running ALARMS in the background, with its output not yet available to clinicians. For each result identified by ALARMS, researchers will contact the patient individually to determine whether notification regarding the test result has occurred. In the third phase, ALARMS will be introduced for clinical use. Again, for each result identified by ALARMS, researchers will contact the patient to determine whether notification has occurred. The principal outcome measures will be the rate at which notification have occurred, comparing rates of medical errors (in the form of inadequate notification) before and after clinical implementation of the ALARMS system. As a secondary measure, rates of adverse events attributed to these errors will be measured as well. Finally, in the fourth phase of the study, some costs of using ALARMS, specifically, the time spent by clinicians addressing lab results that do not require follow-up, will be measured.
