Date of Award


Degree Type


Degree Name

Honors Thesis



First Advisor

Joseph Jordan


Background: Medication errors comprise a significant proportion of medical errors, and are abundant, costly, and associated with causing harm to patients via adverse drug events. The most serious medication errors often involve IV medications. Smart pumps were developed to improve patient safety by reducing medication errors. While some studies have found that smart pumps do not decrease medication errors, most have found they are effective to some degree. It is believed that routinely analyzing data on smart pump alerts, making corresponding adjustments in the drug libraries, and analyzing those adjustments can reduce alarm fatigue, which may then decrease medication errors by resulting in less smart pump users overriding the alerts and utilizing workarounds of smart pump safety features.

Objective: The objective of this study is to assess if changes made to the Indiana University Health system smart pump drug library decreased nuisance alerts by comparing the actions taken in response to alerts before and after the changes were made.

Methods: For a given change made to the Indiana University Health smart pump drug library on April 1, 2016, actions taken in response to alerts corresponding to that change three months prior to and three months after the change were analyzed. The primary outcome was the percent of total alerts that were overrides. Using data from the smart pumps, the number of overrides, reprograms, cancels, and total alerts for each drug in the first and second quarter were recorded. The percentage of total alerts that were overrides, the percentage of total alerts that were reprograms, and the ratio of overrides to reprograms for each quarter were calculated.

Results: Analysis was conducted on 8 drugs: carboplatin, fentanyl PCA, hydromorphone PCA, morphine PCA, morphine PCA 10-24kg, morphine PCA >40kg, naloxone, and octreotide. From the first quarter to the second quarter, the percent of overrides increased for 3 drugs, but for all 3, the number of overrides and total alerts decreased. Of the 5 drugs that had a decrease in the percent of overrides, 3 had an increase in the number of overrides and total alerts. Only 2 drugs had a decrease in the percent of overrides and the number of overrides and total alerts. Statistical significance was achieved only for hydromorphone PCA and morphine PCA. The difference between the first and second quarters in the all the measured outcomes varied between the drugs.

Conclusions: Forming any definitive conclusions was difficult due to the results containing a significant amount of variation. The literature suggests methods to improve smart pump usage, and improve medication safety by extension. These methods are interfacing smart pumps with computerized physician order entry, clinical decision support systems, electronic medical record/electronic medication administration record, pharmacy information systems, bar-coded medication administration, and laboratory data, as well as improving smart pump safety features compliance through education of smart pump users, leadership support, including/consulting smart pump users in drug library design, and routinely using the event log data as a component of a continuous quality improvement program. These methods are all in line with the current, trending belief that the best method for preventing medication errors is making changes to the medication use system as a whole to correct underlying systems failures instead of addressing a single point, such a smart pump alerts.