Tutorial on monitoring time to next medication error.

Alemi and colleagues in this issue of the journal have proposed that rare events can be monitored by shifting from frequency of the event to the examination of the time to the event. This article examines their claim with data obtained from an acute care hospital in the United States. We examined the data on medication omissions to see whether changes in underlying process can be detected through control charts. Medication errors are rare; the article examines medication errors due to omission, which makes the phenomena rarer. The empirical question was whether changes in process of care could be detected using control charts from data on medication omissions. Two different types of control chart, the XmR and Tukey charts, were used to analyze the data. The control chart with the tightest control limits was chosen for further interpretation. The XmR chart showed that there was sufficient power to detect unusual days in which the time to omission error was higher than historical norm. This article suggests that even rare events can be monitored through judicious use of time to the event. It shows the viability of safety teams using time to sentinel events to monitor progress in reducing frequency of sentinel events.

Probabilistic risk analysis of medication error.

UNLABELLED: Over the years, the United States has spent billions of dollars in its quest to improve the quality and safety of health care through the development of new drugs and technologies. Using the probabilistic risk analysis of medication error process, we demonstrate the application of Bayesian Causal Network Model to assess the probabilities of occurrence of rare events related to medication errors. This article summarizes the methodology involved in the process. METHODS: A convenience sample of annual incident reports from a Northeast acute care community hospital was used for the study. Importance sampling was used to improve the accuracy of estimates of the rare events so that we can better understand the relationship of causes within the reported events. DISCUSSION: The Bayesian Causal Network Model provided contextual maps of the behaviors and errors that lead to medication delivery process failures, including unanticipated risks associated with actual errors as well as near misses and common deviations from standard procedures and policies. Health care administrators, clinicians, regulators, and educators can prospectively identify and prioritize risk reduction interventions using the Bayesian Causal Network Model. CONCLUSIONS: The Bayesian Causal Network Model can identify behavioral and systemic factors that can enhance or reduce the risk of wrong drug, wrong frequency, wrong dose, omitted dose, drug interactions, and wrong patient medication errors in hospitals.