Article Text

Download PDFPDF

The frequency of intravenous medication administration errors related to smart infusion pumps: a multihospital observational study
  1. Kumiko O Schnock1,
  2. Patricia C Dykes1,
  3. Jennifer Albert2,
  4. Deborah Ariosto3,
  5. Rosemary Call4,
  6. Caitlin Cameron5,
  7. Diane L Carroll2,
  8. Adrienne G Drucker6,
  9. Linda Fang7,
  10. Christine A Garcia-Palm8,
  11. Marla M Husch9,
  12. Ray R Maddox10,
  13. Nicole McDonald5,
  14. Julie McGuire8,
  15. Sally Rafie11,
  16. Emilee Robertson12,
  17. Deb Saine7,
  18. Melinda D Sawyer4,
  19. Lisa P Smith13,
  20. Kristy Dixon Stinger13,
  21. Timothy W Vanderveen9,
  22. Elizabeth Wade14,
  23. Catherine S Yoon1,
  24. Stuart Lipsitz1,
  25. David W Bates1
  1. 1Division of General Internal Medicine & Primary Care, Brigham and Women's Hospital, Boston, Massachusetts, USA
  2. 2Munn Center for Nursing Research, Massachusetts General Hospital, Boston, Massachusetts, USA
  3. 3Vanderbilt University Medical Center, Nashville, Tennessee, USA
  4. 4The Johns Hopkins Hospital, Baltimore, Maryland, USA
  5. 5Department of Pharmacy, Maricopa Medical Center, Phoenix, Arizona, USA
  6. 6Partners eCare, Partners Healthcare, Boston, Massachusetts, USA
  7. 7Winchester Medical Center, Winchester, Virginia, USA
  8. 8Central DuPage Hospital, Winfield, Illinois, USA
  9. 9Solutions Management and Business Development, BD, San Diego, California, USA
  10. 10College of Pharmacy, University of Georgia, Savannah, Georgia, USA
  11. 11Department of Pharmacy, UC San Diego Health, San Diego, California, USA
  12. 12St Joseph's/Candler Health System, Savannah, Georgia, USA
  13. 13Department of Nursing, Western Connecticut Health Network, Danbury, Connecticut, USA
  14. 14Concord Hospital, Concord, New Hampshire, USA
  1. Correspondence to Dr Kumiko O Schnock, Division of General Internal Medicine & Primary Care, Brigham and Women's Hospital, 1620 Tremont Street, OBC-3, Boston, MA 02120-1613, USA; kschnock{at}partners.org

Abstract

Introduction Intravenous medication errors persist despite the use of smart pumps. This suggests the need for a standardised methodology for measuring errors and highlights the importance of identifying issues around smart pump medication administration in order to improve patient safety.

Objectives We conducted a multisite study to investigate the types and frequency of intravenous medication errors associated with smart pumps in the USA.

Methods 10 hospitals of various sizes using smart pumps from a range of vendors participated. Data were collected using a prospective point prevalence approach to capture errors associated with medications administered via smart pumps and evaluate their potential for harm.

Results A total of 478 patients and 1164 medication administrations were assessed. Of the observed infusions, 699 (60%) had one or more errors associated with their administration. Identified errors such as labelling errors and bypassing the smart pump and the drug library were predominantly associated with violations of hospital policy. These types of errors can result in medication errors. Errors were classified according to the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP). 1 error of category E (0.1%), 4 of category D (0.3%) and 492 of category C (excluding deviations of hospital policy) (42%) were identified. Of these, unauthorised medication, bypassing the smart pump and wrong rate were the most frequent errors.

Conclusion We identified a high rate of error in the administration of intravenous medications despite the use of smart pumps. However, relatively few errors were potentially harmful. The results of this study will be useful in developing interventions to eliminate errors in the intravenous medication administration process.

  • Medication safety
  • Quality improvement
  • Near miss
  • Healthcare quality improvement
  • Information technology
View Full Text

Statistics from Altmetric.com

Introduction

Medication errors that occur during the administration phase may now be the most common type of error in hospitals.1 ,2 Among these, intravenous medication errors have been identified as the most dangerous, with the potential to cause considerable patient harm.3 In the USA, the Association for the Advancement of Medical Instrumentation (AAMI) and the Food and Drug Administration (FDA) held the AAMI/FDA Infusion Device Summit in 2010, in part because of 56 000 reported incidents related to intravenous infusions.4 Finding solutions to prevent intravenous medication errors represents a pressing patient safety priority.

Smart pumps or computerised patient infusion devices include features for administration error reduction and data collection. These represent transformational clinical tools with the potential to greatly decrease the rate of intravenous medication errors in hospitals.5 This technology provides medication error reduction capabilities via a preprogrammed drug library with audiovisual feedback in response to entries programmed beyond predetermined dose, concentration and duration thresholds. According to the American Society of Health-System Pharmacists’ (ASHP) national survey of pharmacy practice in hospitals in 2012, 77% of hospitals in the USA used smart pumps.6 The smart pump adoption rate in the USA has doubled since 2005 and correlates with the implementation of other technologies for quality and safety improvement such as electronic health records (EHRs), computerised provider order entry (CPOE) and barcode-assisted medication administration (BCMA).6

Implementing new technologies such as smart pumps can be challenging. A study by Rothschild et al7 evaluated an early version of a smart pump with limited decision support that identified but did not reduce the rate of serious intravenous medication errors. In addition, the authors noted that maximising the benefits of smart pumps requires continual training of users, maintenance of pumps and updates to the drug libraries. More recently, Husch et al8 assessed the frequency of intravenous medication errors and how preventable those errors would be with the use of smart pump technology. They identified issues related to infusions without orders, infusions continued by the nurse after discontinuation by prescribers, wrong infusion rates, wrong doses, incomplete labelling, incorrect or incomplete documentation, missing or incorrect patient identification bracelets and pump programming issues. Until now,9 ,10 there is no compelling evidence of the effectiveness of smart pumps for preventing all types of medication errors.

To identify the key issues related to intravenous medication administration using smart pumps across a variety of hospital settings, we conducted a multisite study using the prospective point prevalence approach as described by Husch et al.8 The objective of this study was to investigate the frequency and types of intravenous medication errors associated with the use of smart pumps. We report results of our data collection from 10 hospitals in the USA.

Methods

Intravenous medication errors data collection

Medication error is defined as an error occurring at any stage in the medication-use process, including prescribing,

transcribing, dispensing, administering or monitoring.11 A single error has the potential to have negative consequences at multiple stages of the medication-use process if it is not identified and corrected at an earlier stage. Smart pumps are designed to support the safe administration of intravenous medications, so our study focused on medication errors that occurred during administration. Medication errors that occurred in the administration phase associated with wrong patient, medication, dose, route and time, and additional error types as investigated by Husch et al8 were included in our investigation with minor modifications. A complete list of error type definitions and modifications is provided in online supplementary appendix 1. All violations of hospital policies were identified as errors regardless of the capacity to directly contribute to patient harm; examples include insufficient labelling of intravenous products and bypassing the use of the smart pump entirely. These policy violations were included to fully assess the intravenous medication administration process and to aid the study team in developing comprehensive strategies to improve safety and efficiency. Each error was categorised and the severity was assigned based on the status of the error at the time of observation. Errors were not followed up to identify actual harm or whether the error type changed. For example, delay or omission of a medication was categorised at the time of observation only. The potential severity of each error was ranked according to the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) Harm Score Index12 (see online supplementary appendix 2). The NCC MERP Harm Score Index was modified to describe the potential of an error to cause harm if allowed to continue versus actual harm after the error occurred.13 Through an iterative participatory software development process, a web-based data collection tool using Redcap (Research Electronic Data Capture, Vanderbilt University, 2009)14 was developed by a multidisciplinary group of study team members.15

Supplementary Appendix

The primary study outcomes were the overall intravenous medication administration error rate and an assessment of their potential harm. Multiple errors could occur during a single administration. The assigned severity was based on the potential for the error to have resulted in patient harm if it had not been intercepted.

Measurement of intravenous medication errors

Study setting and samples

Participating hospitals were recruited at AAMI Foundation's Healthcare Technology Safety Council meeting in January 2012, where interdisciplinary patient safety experts gathered. Hospitals that volunteered were screened for their ability to participate in a large research study and were required to have smart pumps in use at their institution. The hospital size, geographic location in the USA, hospital type and smart pump vendor were identified as described in online supplementary appendix 3. The investigators selected hospitals representing a variety of smart pump vendors, hospital sizes and geographic locations. Once a hospital was included, if a subsequent hospital of similar description volunteered the second volunteer was excluded. A total of 10 hospitals (seven academic medical centres and three community hospitals) were included in the study. Study coordinators at each hospital included clinical leaders who lead local, regional and national patient safety and quality improvement initiatives in the USA.

We evaluated one medical intensive care unit (ICU), one surgical ICU, one general surgical unit and one medical unit at each institution. Patients were included in the study if receiving any intravenous fluid or medication on the day of observation. All intravenous medications and fluids delivered via smart infusion pumps, syringe pumps and patient-controlled analgesia pumps were included in the investigation if these were hanging from the pump at the time of evaluation. Patient-controlled epidural analgesia, total parenteral nutrition and blood products were excluded. All medication orders were reviewed by a pharmacist prior to administration to the patient, and in urgent situations, verbal orders were permitted. At the time of the study, none of the hospitals had interoperability between smart pumps and their EHRs.

This study was approved by the institutional review boards at all study sites. All data collected were de-identified in the Redcap application and did not include protected health information.

Data collection procedure

Two trained observers (nurse and/or pharmacist) from each site collected data as a pair during day shifts for 2–4 days between February and August 2013. Observers compared the intravenous medication, attached labelling, dose and infusion rate programmed in the smart pump with the prescribed medication, dose and rate in the EHR. The presence of correct patient identification bracelet and name verification were recorded for each patient. Actual use of a smart pump (ie, the intravenous fluid was infusing through the pump) and the use of the drug library were also assessed.

Compliance with placement of tubing change tags on intravenous tubing and hospital labels on intravenous medications was assessed. A ‘tubing change tag’ was defined as an identifier on intravenous tubing indicating when the tubing should be replaced. A ‘hospital label’ was defined as a label attached to an intravenous medication by patient care area staff. A ‘pharmacy label’ was defined as a label attached in the pharmacy before dispensing medications from the hospital pharmacy to the patient care area. Labelling policies varied across institutions. For some institutions, hospital policy required either a hospital label or a pharmacy label for all intravenous products, including all medications and electrolyte solutions, while others did not. Data collection observers at each institution reviewed only the hospital labels for compliance with their own hospital policies.

In order to confirm that errors or deviations from hospital standardised practices and policies were present, both observers at the site had to agree that an error had occurred. All observers received training on assessing potential harm using the NCC MERP Index using sample cases. Observers entered all data into the Redcap data collection tool and each error was rated by the observers using the NCC MERP Index.12 ,13 ,15 If an error that had the potential to cause harm was identified, the nurse caring for that patient was discreetly informed so that it could be immediately corrected.

Data analysis

The results were analysed as frequency of intravenous medication errors, broken down by types of intravenous medication errors and their NCC MERP severity rating. Error rate (percentage) was calculated as the number of identified errors per the number of observed medication administrations. In addition, the error frequency of intravenous medication error by type within each site was calculated. The observed errors were divided into two categories: violation of hospital policy errors and medication errors. Labelling and tubing tag compliance errors were categorised as violation of hospital policy. Medication errors included any medication error reaching the patient (excluding violation of hospital policy). The error of bypassing the use of the smart pump or drug library was, in part, violation of policy; however, these errors were categorised as medication errors because of the high potential risk of harm of these types of errors. Errors with potential seriousness of Harm Index D or greater were rated retrospectively by observers at the lead site as to whether smart pumps with closed loop interoperability between smart pumps and the EHR would have prevented the error. In order to answer this question, closed loop interoperability was defined as barcode scanning which triggers the transmission of physician-ordered, pharmacist-reviewed infusion parameters from the EHR to prepopulate the smart pump, and time-coded infusion data that are automatically sent from the smart pump to the EHR for the nurse to review and document.

Results

Frequency, type and potential severity of errors

During the data collection period, 800 inpatients across 10 hospitals were evaluated for inclusion in the study. Of the 800 patients evaluated, 478 were receiving and/or prescribed intravenous medication at the time of observation. These patients received a total of 1164 intravenous medications or fluid infusions that were assessed. In total, 1691 errors were identified and 699 (60%) infusions had one or more errors associated with their administration. The frequency and types of intravenous medication errors associated with the use of smart pumps across sites are summarised in table 1. Violations of intravenous labelling and tubing change policies were the most frequent error types at 60% and 35%, respectively. Excluding these violations of hospital policy, 1 error of NCC MERP category E (0.1%), 4 errors of category D (0.3%) and 492 errors of category C (excluding violations of hospital policy) (42%) were identified. Of these errors, unauthorised medication (24%), bypassing the smart pump or drug library (10%), wrong rate (5%) and omission of intravenous fluids and medications (5%) were the most frequent errors.

Table 1

Total error frequency, type and potential severity rating

The intravenous medication error rate varied widely by site, ranging from 6% to 78% of the observed infusions (table 2).

Table 2

Medication error frequency by type and by site*

Labelling errors

The non-compliance rate of tubing change tags varied from 10% to 92% (table 3), reflecting varying rates of deviation from policy among each institution's nursing practices. Hospital label policy violation was the most frequent error identified. This type of error was common in all institutions, with frequency rates ranging from 13% to 96% of all observed infusions (table 3). Across all hospitals, 24% of the observed intravenous medications or fluids had no label and 36% had incomplete labels. The most frequent types of missing information on hospital labels were the hang time/date and/or expiration date (table 3).

Table 3

Violation of hospital policy error frequency by type and by site

Unauthorised medications

Unauthorised medications accounted for 281 errors, occurring in 24% of the observed infusions. Of these errors, 245 (87% of unauthorised medications) were due to the administration of normal saline at low infusion rates to keep the vein open (KVO), 22 (8%) were other maintenance solutions and 14 (5%) were intravenous medications. Among the 14 intravenous medication cases, 10 were discontinued medications that were still connected to the smart pump (ie, medication contents remained in the bag and the smart pump was off and/or not infusing into the patient). In all four remaining cases, intravenous medications were infusing when they should have been discontinued. However, one case had a permissible physician's undocumented verbal order to continue the medication.

Smart pump use errors

Smart pump non-compliance was 10% (n=120) of the observed infusions across all sites. This result includes both bypassing the smart pump (n=16) and bypassing the available drug library (n=104). Fourteen of the infusions administered without the use of a smart pump were maintenance fluids and two were medications (magnesium sulfate and cefazolin). Drug libraries were primarily bypassed when administering maintenance fluids (n=91) rather than medications (n=13). Of the 13 medication cases, the more frequent bypassed drug library entries were piperacillin/tazobactam (n=4), vancomycin (n=3) and magnesium sulfate (n=2). Pump setting errors included selection of the wrong drug library entry (eg, morphine programmed in for hydromorphone) (n=3) and programming the wrong concentration (eg, vancomycin 1750 mg selected instead of 1500 mg) (n=3).

Wrong rate errors

Wrong rate errors were observed in 26 intravenous fluids and 28 medications. The most common medications were norepinephrine (n=4), propofol (n=4), vancomycin (n=3), piperacillin/tazobactam (n=3) and magnesium sulfate (n=2). In 47 cases, medications or fluids were infused slower or faster than ordered rate. For example, a patient who had a current order for piperacillin/tazobactam at 200 mL/h was actually receiving the medication at 25 mL/h. Upon further investigation, three different types of causes were identified such as slip of the finger errors, order changes or deviations of individual nursing practices. The remaining seven cases were administration errors, which occurred when the rate was set at outside the titration parameter range specified in the order.

Patient identification errors

There were three errors related to not using patient identification bracelets (0.6% of observed patients). There were two errors attributed to the wrong patient name on the label of the intravenous medication bags. Two patients had the same medication ordered (vancomycin) and the intravenous bags were switched (rated as Harm Index C).

Severity ratings of medication errors

Examples of errors rated D or greater are summarised in table 4. The most severe error identified was the delay of administration and was rated as Harm Index E, defined as likely to cause temporary harm. This incident was related to delayed dextrose 5% in water (D5W) administration for a patient who was hypoglycaemic. There were four errors rated D, including delay of administering norepinephrine, wrong rate of norepinephrine and magnesium sulfate and omission of ordered piperacillin/tazobactam.

Table 4

Examples of errors with potential seriousness of Harm Index D or greater and judged to be preventable

The most frequent medication errors rated C were unauthorised medication and wrong rate followed by omission of an ordered intravenous fluid or medication, expired drug, wrong dose and delay of administration. Thirty-one per cent of errors rated C pertained to continuous or intermittent intravenous medications, which include high-alert medications16 such as epinephrine, propofol or hydromorphone.

Discussion

Overall error rates

A high rate of errors was found in the administration of intravenous medications despite the use of smart pumps, but relatively few were harmful errors. One of the main contributing factors to the high error rate was non-compliance with hospital policies on medication labelling and tubing change tags. Infusion rate errors were the leading type of serious medication error followed by unauthorised medications and omission of medications. Current smart pumps cannot intercept these types of errors shown in table 4 without interoperability between smart pumps and EHRs. These five intravenous medication errors potentially required additional monitoring or actions to preclude harm. More severe errors that would result in prolonged hospitalisation were not observed. Though the severity of the observed errors was relatively low, the study team identified issues of current nursing practice, which could disrupt safe intravenous medication administration. For example, KVO orders tend to be administered without a smart pump or drug library and this was rated as Harm Index C. Some hospitals did not enforce the use of smart pumps for all intravenous medications and solutions despite hospital policies requiring their use for every infusion; this allows nurses to administer intravenous medications without smart pumps as a standard practice. It is essential to use smart pumps for every single intravenous administration to maximise smart pump safety features and to intercept medication errors. Therefore, our study focused on not only eliminating harmful errors but also common practice errors that bypass important safety features provided by smart pumps when correctly employed. In addition, our results show that there were a number of errors which were not likely to be intercepted by smart pumps in the absence of interoperability (table 4). Most of the example cases were deemed preventable if closed loop interoperability between the smart pump and the EHR was implemented. Some errors identified by our study were not directly related to smart pump use; however, they identify an opportunity to streamline the intravenous medication administration process to avoid unintended consequences.

Labelling errors

Errors due to incomplete hospital labels were the most frequent labelling errors observed across the hospitals. Each hospital had its own intravenous labelling policy; therefore, the observation data were collected according to each institution's policy. The differences in hospital policies were substantial and contributed to a wide variation in error rates across hospitals. Some policies required that hospital labels contain comprehensive information, while others required only a few items. According to the Joint Commission National Safety Goals in 2015 (NPSG.03.04.01 and MM.05.01.09),17 medication name, concentration, amount, diluents, date prepared and expiration date and time (if expiring within 24 h) are required for medications prepared in a patient ward. Only expiration date and time are required for intravenous solutions removed from a medication cabinet. Our results showed that the required labelling information tended to be omitted, while some unrequired information was included.

Through the process of reviewing hospital labelling policies, the study team found that some information needed prior to the implementation of CPOE, BCMA and the electronic medication administration record (eMAR) is no longer as vital to document on the paper label. For example, in most institutions, intravenous medication bags are documented via integrated bar code scanning and an eMAR where nurses can access order information. The time the infusion was started as well as the identification of the person who started it may be included in the eMAR. Since this information is already present in the medical record, nurses may omit it from hospital labels, particularly when non-critical medications (eg, maintenance fluids) are administered. While these types of labelling practices were necessary when manual and paper medication administration documentation processes were in place, they may have less relevance when CPOE, eMAR, BCMA and intravenous smart pumps are used.

We found tubing change tags were often omitted even though all hospitals require these labels as a nursing practice policy. Tubing change tags are necessary in order to comply with the Center for Disease Control and Prevention requirements to reduce the risk of catheter-related blood stream infections.18 The various sites were using different approaches to labelling tubing as no standard tubing change tag label exists. By reviewing all participating hospitals’ labels and discussing the results with the study team, the team recognised the benefits of using standardised tubing change tag labels to distinguish the day of the week on which the nurse should change the tubing. These results highlight the importance of reviewing existing practices and policies when implementing technologies such as smart pumps.

Unauthorised medication errors

Unauthorised medications represented 24% of the observed infusions. This was higher than Husch et al's8 study (16%), although there was large variation among institutions. The error percentage ranged from 3% to 53%. The most frequent type of unauthorised medication error was the lack of provider orders for KVO maintenance fluids (87%). In the four institutions with higher rates of unauthorised medication errors, all were associated with missing KVO infusion orders. According to the hospitals’ policies reviewed in this study, all intravenous medications and solutions, including KVO infusions, require written orders. During informal discussions with staff nurses, we found that these errors were not recognised as critical risks by staff, and nurses often infuse normal saline at low infusion rates without any written order. A proposed resolution for this issue would be to create a standard order set for KVO infusions when secondary intravenous medications are ordered or add an automatic KVO order to existing order sets. In addition, not all institutions had policies specifying the standard rates of KVO infusions and nurses independently set rates from 5 to 20 mL/h. These inconsistencies have potential to result in miscommunication of fluid balance. The importance of having recommendations for standardised KVO rates clearly defined in hospital policies was acknowledged by the study team members.

Another factor associated with unauthorised intravenous medication errors pertained to discontinued medications and intravenous medication bags being left at the bedside. For example, the study team considered a discontinued medication as unauthorised when an intravenous medication order was discontinued in CPOE but the medication was still attached to the patient, even if it was not infusing. The study team felt that if a discontinued medication was not taken down by the intended time, there was a risk of unintended administration. We believe it is important to educate nurses to not only stop the infusion of intravenous medications when ordered to discontinue, but to also remove completed or discontinued intravenous bags from the bedside. The expired medication order error occurred when the medication was to be continued but the active medication order had expired and the new order had not yet been processed. Other types of unauthorised medication errors were due to medications being administered without an active order (undocumented verbal order error). Five per cent of all unauthorised medications were related to medications other than intravenous fluids and these medications could pose a higher risk of potential harm. We recommend limiting verbal medication orders to exceptional circumstances and including an approved list of medications that can be overridden in an emergency from medication dispensing cabinets in hospital policies.

Severity ratings of medication errors

Although Harm Index rated D or greater errors were observed in only a few cases, our study team pointed out that preventing the frequent errors rated C could have a great impact of improving medication administration safety because the study design did not allow us to confirm that there was no harm in errors rated C. In fact, some errors rated C pertained to high-risk medications, which are most often involved in patient harm.16 Our results allowed us to identify possible points of failure in the medication administration process that are opportunities to prevent potentially harmful errors in future. The fact that these errors reached the patient suggests that there are opportunities within the system to improve. It is important to identify a strategy to eliminate errors rated C as well as those rated D or greater.

Comparison with previous studies

The rate of errors in our study is similar to that found in Husch et al's8 previous single-centre study. With regard to labelling compliance, the error rate in our study appears to be higher than Husch et al's8 study. This could be due to the fact that our study definition was expanded to include not only the omission of rate information on labels but also the absence of any information required by hospital policy. The tubing change tag policy compliance rate was not evaluated by Husch et al8 and thus was not available for comparison. We identified a low rate of compliance with tubing change tag policies (65%) in our study. The greatest difference between our findings and those of Husch et al8 was that we found very few patient identification errors. The compliance rate with the attachment of patient identification bracelets was 99.9%. This may be related to the fact that most of the participating institutions had BCMA and these institutions had a high compliance rate with the scanning of patient identification bracelets and BCMA.

When comparing our results with Rothschild et al's7 study, wrong dose and concentration error rates in our study were substantially lower. Rothschild et al's study was conducted at a single site in 2002 and the hospital had just implemented an early version of the smart pump. Rothschild et al found that even when a drug library was available for a particular drug, nurses bypassed the drug library for about 25% of intravenous medication administrations and used the manual entry mode. Additionally, override alerts were acknowledged but not acted upon, further limiting the efficacy of the smart pumps in preventing errors and adverse drug events. Our study found that out of the 8% of medication administrations where the use of the available drug library entry was bypassed, most were for maintenance fluids and no harm was identified. Our findings suggest that compliance with using the drug library has dramatically improved in the past 10 years. Though the compliance rate of using the smart pump was relatively high in all sites, we recommend that all medications and fluids be administered using a smart pump to maximise the benefits of the smart pump for patient safety. Therefore, room for improvement exists to achieve 100% compliance at all sites.

Significant reductions in the medication error rate compared with earlier studies may be due to soft and hard limit functions, that is, dose range alerts. Early versions of smart pumps did not provide a comprehensive list of medications with strict hard limits. This functionality prevents miscalculation of dose errors and key entry errors. We observed a very low rate of errors since the majority of the institutions had implemented multiple error reduction technologies, including smart pumps with dose range alerts, CPOE, BCMA and eMAR. These technologies reduce errors in multiple steps of the medication-use process: ordering, dispensing and administration.19 However, there is still a gap between existing BCMA, eMAR and CPOE integration and smart pump technology. In most institutions, smart pumps are not integrated with CPOE or BCMA and require independent programming by the user. None of the identified errors rated D or greater could have been prevented using current smart pump technology. Thus, implementing interoperability of smart pumps may help eliminate errors and close the safety gap.

Limitations

This study was a point prevalence study and thus included a limited number of observations. Observers rated the potential harm at the time of the observation. The actual harm caused by an error was not investigated after the observation and was not recorded as part of the study protocol. This may have resulted in assigned severity ratings being higher or lower than the actual harm incurred. In addition, the likelihood of an error to cause certain types of harm, such as prolonged hospitalisation, could not be assessed due to the point prevalence methodology. We recognised this as a study limitation and decided to include any identified error that reached the patient (NCC MERP of C, D, E, F, G, H or I) as a medication error.

Additionally, as the study was conducted across multiple institutions and relied on human observation, interobserver variability may have occurred, even though all observers were trained to limit this factor. Selection bias may have affected results as hospitals included in the study were not chosen randomly. Not all types of intravenous medication errors were captured in this study. Only errors that could be observed during intravenous administration were included. This study also assumes that the medication order information is correct.

Conclusion

Our study found a high rate of errors in the administration of intravenous medications with the majority of errors having relatively low potential for harm. Overall, violations of hospital policy for labelling intravenous bags, tubing change tags and administration of unauthorised medications were especially frequent. Since these errors are not directly related to the use of smart pumps, these findings suggest that smart pump technology alone cannot fully prevent errors associated with intravenous infusions. This study further emphasises the need for interoperability between currently implemented healthcare information technologies in order to make meaningful improvements in intravenous medication safety. Further research is warranted to understand the full impact of interoperability and its potential unintended effects. The findings from this study can be used to further improve the safety of the intravenous medication process by providing additional insight into the types and frequencies of errors that still exist. Our next steps include developing and implementing an intervention plan to reduce errors associated with intravenous medications and fluids administered using smart pumps. The effectiveness of an intervention bundle will be assessed by collecting additional data post-intervention.

Acknowledgments

We gratefully acknowledge Association for the Advancement of Medical Instrumentation (AAMI) and Carefusion Foundation for funding for this study as well as Mary Logan, JD, CAE, Marilyn Neder Flack, Sarah Fanta Lombardi, MPH for supporting the study and. We also acknowledge our national smart pump project collaborators, Jo Anna Lamott, RPh, MBA, Anne Bane, RN, MSN, Elizabeth Buckley, RN, Amy Chouinard, RN, Moreen Donahue, DNP, RN, NEA-BC, FAAN, Bob Feroli, PharmD, Beverly King, RN, Carol Luppi, B.S. RN, Kathleen McIntosh, RN Johnston S. Morrison, MSN RN, CPPS, Katie Outten, MSN, Julie Zimmerman, MS, RN, CNS, CCRN for conducting the study and sharing insights. We also thank Ann Blandford, FBCS, CEng, PhD and Imogen Lyons for their thoughtful review of the manuscript.

References

View Abstract

Footnotes

  • Contributors KOC, PCD, CSY, SL and DWB: contributed to conception and design, analysis and interpretation of the data and results, drafting and writing the article, revising it critically for important intellectual content and final approval of the version to be published. JA, DA, RC, CC, DLC, AD, LF, CAG-P, MMH, RRM, NM, JM, SR, ER, DS, MDS, LPS, KDS and EW: contributed to supervision of the data collection at each site, developed hypotheses, analysis and interpretation of the data and results, as well as drafting and writing the article, revising it critically for important intellectual content and final approval of the version to be published. TWV: contributed to conception and revising the paper critically for important intellectual content.

  • Funding Carefusion Foundation; Association for the Advancement of Medical Instrumentation (AAMI).

  • Competing interests None declared.

  • Ethics approval Partners Healthcare Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles