1. Aschenbrenner, Diane S. MS, RN


* Smart infusion pumps have reduced but not eliminated medication errors. These errors are frequently related to misuse or underuse of the safety features of the pump.


* The Institute for Safe Medication Practices offers safety considerations for using smart pumps and suggestions for maintaining an institution's smart pump drug library.



Article Content

In the United States, 7,000 to 9,000 people die yearly because of a medication error.1 Smart drug infusion pumps, developed by Nathaniel Sims and colleagues in 1992 and implemented in 1997 at Massachusetts General Hospital,2 are considered a major advancement in safe medication administration. The devices contain drug libraries for each drug infused at a particular hospital or health care setting. These libraries use dose error reduction software and contain preset dosing limits to prevent under- or overdosing infusion errors. Each institution creates and maintains its own library. Use of smart infusion pumps, however, has not led to an error-free environment for patients receiving IV infusions.


As recently as 2022, the ECRI Institute included infusion pump errors on its list of "Top 10 Health Technology Hazards."3 The Joint Commission warns that many errors occur because the dose error reduction software is not optimized.4 Some smart pump errors are human errors, such as working around drug library safeguards or failing to use them at all, while others result from software or hardware design flaws, such as key bounce (one touch sensed as multiple touches) or data stored via one technology being unavailable to access using another. Ideally, all health care technology should communicate, meaning that computerized order entry systems, bar coding, and smart pumps should work together to prevent medication errors.


In a review of the literature to determine human-based smart pump errors, five main categories of errors were identified: undocumented, drug library, programming, administration, and ancillary equipment errors.5 The authors further mapped error prevention strategies to each error category. These strategies included providing training to end users on how and why to use the safety features of the smart pump; standardizing drug libraries to limit the number of medications available and to include bolus doses; updating the drug library on a regular basis; implementing technical integration of smart pumps with barcode scanning and computerized order entry systems; and increasing end user compliance by decreasing unnecessary warnings and minimizing workarounds.5


In its January Medication Safety Alert!, the Institute for Safe Medication Practices (ISMP) offered several safety considerations for using smart infusion pumps. The ISMP points out that drug library customization requires a dedicated multidisciplinary team to maintain and update library content. The team also needs to consider that a drug's safety parameters may vary based on the clinical practice site and the patient population served. For example, the concentration and dose of a drug is likely to be different for adults in critical care than for newborns.


The ISMP identified difficulties health care providers may encounter when trying to optimize drug libraries and prevent medication errors from smart pumps. Some drugs do not have a universal maximum dose; instead, they have variable doses based on patient-specific parameters, tolerability, or clinical response. Wide dosing parameters or variable rates of infusion can pose problems when trying to create dose range limits in the smart pump library. Without a preprogrammed maximum dose, the pump cannot prevent programming errors such as decimal errors that can increase the dose 10-fold. Drugs that have several indications, each with a different suggested dose, are another challenge when creating the drug library.


Although using the drug library can be cumbersome and confusing, nurses should avoid trying to save time by skipping it and programming every medication infusion as if it was a main line of basic fluids. Doing so places the patient at increased risk for medication error.4 When possible, nurses should avoid administering a bolus dose of a medication from a continuous infusion unless the smart pump is designed to accommodate this. To avoid programming errors when a drug's limits are not preset into the drug library-when titrating a drug dose based on a patient's response, for example-a second nurse should be present to confirm the programming of the smart pump. Each warning message should be taken seriously. If nurses are frequently overriding a warning for a particular drug, they should inform the committee responsible for the smart pump drug library and seek resolution.


Maintaining and updating the drug library can pose challenges, and the ISMP encourages nurses to be active on their organization's smart infusion pump team. The ISMP suggests that the team establish and approve dose limits and use computerized order entry and the smart pump in tandem, where the order entry system offers safe dose choices while the smart pump provides alerts or hard stops in the drug administration process, acting as a final layer of protection against medication errors. To read all of the ISMP's suggestions for the interdisciplinary team responsible for the smart pump's drug library, go to Please note, parts of this website require a subscription.




1. Tariq RA, et al Medication dispensing errors and prevention. In: StatPearls [internet]. Treasure Island, FL: StatPearls Publishing; 2023. [Context Link]


2. Massachusetts General Hospital, Mass General Research Institute. Nathaniel M. Sims, M.D. n.d. [Context Link]


3. ECRI Institute. Top 10 health technology hazards for 2022: executive brief. Plymouth Meeting, PA; 2022. [Context Link]


4. Joint Commission. Sentinel event alert 63: optimizing smart infusion pump safety with DERS. Jt Comm J Qual Patient Saf 2021;47(6):394-7. [Context Link]


5. Kirkendall ES, et al. Human-based errors involving smart infusion pumps: a catalog of error types and prevention strategies. Drug Saf 2020;43(11):1073-87. [Context Link]