Assessing Medication Safety Technology in the Intensive Care Unit

Understanding the strengths and weaknesses of a technology in the context of the distributed system in which it is working is critical to assessing and improving the performance of that system. Taking a systems approach requires knowledge about how all agents in a system work together to achieve the goals of that system. With these aims, the alerting mechanism of infusion pumps containing Dose-Error Reduction Software (DERS) was studied to determine its effectiveness in the Intensive Care Units (ICU's) of three hospitals. In 1,146 of the 9,557 pump alerts (12.0%), the alert caused the clinician to change the input. Of these, 1,030 were changed to within the hospital's recommended dosing limits. The alert was overridden for 8,400 (88.0%) of the alerts. The data show that this technology successfully informed clinicians over 1000 times that unintended doses had been inputted and stopped those doses from reaching the patient, thereby averting potential Medication Events. The data also suggest that, because nearly 90% of the alerts were overridden, a well-intended and valuable alert may be perceived by the clinicians as a false alarm and may be overlooked. Another key finding from this analysis was that clinicians may have used potentially unsafe workarounds to administer intravenous drug boluses (i.e., more rapid infusion of a defined dose or volume) and to keep the patient's line active between infusions. In a separate parallel study, clinician self-report of potentially harmful medication events was studied. During 559 hours of direct observation, clinicians detected 27 (IV and non-IV) medication events. All of the reported events were outside of the scope of what DERS technology was designed to detect. In addition, during the same time period the technology detected five potentially harmful IV medication events that the clinicians did not report. The results of these two studies indicate two possible classes of solutions that could reduce the impact and likelihood of medication administration errors. One class of solutions involves the procedures and policies of the hospital, ensuring that process and technology implementations are optimally tuned, taking human performance and the current practice of the clinicians into account. The other class of solutions involves using new strategies and technologies to ensure that each system agent has access to other agents' perspectives, and the broader system's perspective. Studies such as these can provide insight into the use of safety technology during critical care processes and provide direction for future research, including more effective design of alerting mechanisms of ICU devices.