Infusion Medication Error Reduction by Two-Person Verification: A Quality Improvement Initiative
OBJECTIVE: Errors made in the administration of intravenous medication can lead to catastrophic harm. The frequency of hospital settings in which medication pumps are being used are increasing. We sought to improve medication safety by implementing a 2-person verification system before medication administration.
METHODS: Our quality improvement initiative took place in an anesthesia radiology imaging service at a tertiary pediatric hospital. Key drivers included frequent educational meetings with clinicians, written reminders, display of visual reminders, constant feedback in the clinical areas that carried out the processes, and sharing of knowledge on displayed run charts. A multidisciplinary team conducted a series of tests of changes to address the interventions. Data were collected and entered into a database by an independent and impartial data collector. Data were analyzed via run charts and statistical process control methods.
RESULTS: The team ran 24 plan–do–study–act ramps. The rate of 2-person verification of infusion pump programming increased from 0% to 90% and was sustained. Overall, 4 errors were rectified before the medication was administered to the patient. There was no delay in case starts (>90% before and during the project). This project played a key role, as part of a larger initiative within the department of anesthesia, in reducing medication errors.
CONCLUSIONS: A brief 2-person verification approach can reduce medication errors due to inaccurate infusion pump programming. This improvement was achieved with the use of plan–do–study–act cycles. The impact can be significant and will promote a hospital safety culture.
- CCHMC —
- Cincinnati Children’s Hospital Medical Center
- CRNA —
- certified registered nurse anesthetist
- EMR —
- electronic medical record
- FMEA —
- failure modes and effects analysis
- PDSA —
- QI —
- quality improvement
- RN —
- registered nurse
The anesthesia radiology imaging service at Cincinnati Children’s Hospital Medical Center (CCHMC) is a typical brief procedure, multilocation area. There were 2 infusion pump programming errors (resulting in no patient harm) in our radiology imaging service area related to incorrect entry of patient weight or drug dosage on the infusion pump. These errors were an impetus for this quality improvement (QI) project. Many high-risk infusion medications (eg, propofol, dexmedetomidine) are used for sedation, and errors in programming and operation can be catastrophic.1 Common reasons for intravenous medication infusion errors are incorrectly programming weight or drug dosage, programming the wrong medication, tampering with infusion pumps by unauthorized users, and overriding alerts without recognizing an error.
More than 56 000 adverse events, 710 deaths, and 87 recalls associated with the use of infusion devices alone were reported to US Food and Drug Administration in the 5-year span from 2005 to 2009.2 Children are at higher risk for medication errors than adult patients because of the need for precise weight-based dosing. Most intravenous medication errors occur when drugs are administered that require multiple-step preparation.3 Although technological advances have greatly improved patient safety through the use of smart pumps, the risk of intravenous medication infusion errors related to programming has not been eliminated.4,5 A variety of approaches to reduce these errors have been described1; finding safe ways to prevent intravenous medication infusion errors remains a priority.6
Two-person verification is a process in which the work of the first person is checked independently by another person.7 Independent checking is crucial in reducing any bias that may occur when the second checker sees what he or she expects to see, regardless of any errors.7 Two-person verification has successfully reduced wrong-patient or wrong-study radiology events in a pediatric hospital8 and has been proposed to reduce medication errors.9 We organized a multidisciplinary taskforce consisting of anesthesiologists, certified registered nurse anesthetists (CRNAs), and radiology registered nurses (RNs) to reduce these errors. Our overall objective was for ≥90% of the infusion pump programming to be verified by 2 people before being administered to the patient in the anesthesia radiology imaging division.
This project was considered a local QI project. There was no direct contact with patients or families, and it was considered non–human subject research. Data were obtained from a data collection form created for this project and were deidentified. The data were stored on a password-protected computer.
Standards for Quality Improvement Reporting Excellence (SQUIRE) guidelines were followed in the preparation of this manuscript.10
The project was conducted at CCHMC, an urban tertiary academic care center with round-the-clock anesthesia service availability to provide anesthesia or sedation for radiologic imaging (∼5000 cases per year). The Department of Anesthesia consists of 60 anesthesiologists, 40 CRNAs, and 29 anesthesia imaging RNs. The setting for this study consisted of 4 MRI scanners, 2 computed tomography scanners, 1 nuclear scanner, and 1 auditory brainstem response testing unit.
The Department of Anesthesia provides anesthesia for children undergoing radiologic imaging by a standardized method. Anesthesia is induced with sevoflurane in a mixture of oxygen and nitrous oxide, and an intravenous line is placed. An infusion of propofol or dexmedetomidine is started, based on the clinical indication, to achieve motion control.
The QI team consisted of 3 anesthesiologists, 2 radiology imaging nurses, 3 CRNAs, and 1 administrative assistant. More than half of the team have had previous formal QI training. A QI consultant and the hospital QI leaders provided input during the project.
Planning the Intervention
The QI project was conducted between August 2014 and February 2015. The impetus for this project was 2 infusion pump programming errors (resulting in no patient harm) that occurred in our hospital before August 2014.
Data were collected via standardized questionnaires developed by the QI team to ascertain whether an infusion pump was used and whether an error was found and rectified. The answers were collected as binary responses. An independent administrative assistant who was blinded to the QI project collected the data forms and entered them into the database. No patient-related health information data were collected. An “error” occurred when an infusion pump was used for that particular patient and an error with programming was found and rectified (answered “yes” to both questions). Pilot data were collected for a period of 4 weeks in August 2014. Elective, urgent, and emergent cases were included in the project if they entailed the use of infusion medications.
The method used in the project started with small tests of change or plan–do–study–act (PDSA) cycles.11 The QI team initially used 2 QI methods: first, process mapping (Fig 1) to lay out the whole process of administering infusion medications and, second, failure modes and effects analysis (FMEA) (Fig 2) to identify the failures that occurred in the process. FMEA helped clarify the lack of process that could potentially result in a medication error.
The team identified operational factors or key drivers and their associated interventions (Fig 3) based on the pilot data and from the FMEA. The team set a goal of increasing compliance with 2-person verification from 0% to 90%. A national benchmark was not available for this goal. Because this is an important safety issue, we had the goal to achieve at least level 1 reliability, where there is 80% to 90% success (1 or 2 failures out of 10).12 The PDSA cycle began with 1 imaging location and then spread to other imaging locations.
Key Drivers and Interventions
In this key driver, education was targeted to the small percentage of anesthesiologists and CRNAs who regularly provided anesthesia for radiologic imaging. The first step to make this process work was to provide education about the importance of a safety check to reduce patient medication errors. Education about the use of standardized pump programming was provided to stakeholders with a job aid (anesthesiologists, CRNAs, RNs), and education about anesthesia medications was provided to RNs. The specific roles of each person were clarified. A data form was created, and the RN who was involved in caring for the patient completed the data form. One challenge during this process was resistance to a culture change from old processes to the 2-person verification process. This challenge was discussed at departmental meetings.
The first visual aid was a sticker, which was to be signed by 2 people to confirm that the 2-person verification was completed. Although considered a significant advance, this process met with a lot of resistance from the staff because of space constraints in the anesthesia workspace. Therefore, this sticker was abandoned.
The second visual aid was developed midway through the improvement project, and acted as a reminder. The visual reminder was placed in locations where failures most often occurred to remind the clinical staff to complete the 2-person verification.
The initial resistance to culture change was converted successfully with buy-in from stakeholders. The key driver diagram and the weekly updated run charts were continuously displayed in the work area to achieve support, provide a constant reminder of the project, and share knowledge. Constant reminders and a presentation of the process during the monthly QI meeting were also performed. During the QI project, 4 programming errors were detected and were rectified before they could affect the patient. These instances of error rectification were a major impetus for stakeholder buy-in.
Modifications to Electronic Medical Records and Procedural Timeout
The electronic medical record (EMR) was modified to accurately document 2-person verification of the infusion pump programming. The procedural timeout is a routine check performed before the procedure and start of anesthesia or sedation.13 The QI team representatives worked to locally add the 2-person verification process for infusion pump programming to the procedural timeout to ensure sustainability.
Performance of the Verification Process
The first person to perform the verification process (anesthesiologist or CRNA) fully programmed the pump. Subsequently, the second person (another anesthesiologist or CRNA, RN) walked through each step in the infusion pump programming and cross-verified the drug, with particular emphasis placed on patient weight and drug dosage programming. These steps were taken before the medication was administered to the patient and before procedural timeout.
Study of Interventions
The QI team assessed how well the interventions were implemented and their impact on the goal after the completion of the project. The data collection was continued after completion of the project to ensure sustainability. After changes in the EMR documentation process and the procedural timeout, impact on the radiology anesthesia service was assessed. On a monthly basis, the QI team reviewed the reports, and the information was conveyed to perioperative staff through regular meetings and poster boards.
Methods of Evaluation and Analysis
The primary (process) measure was the proportion of 2-person verification of infusion pump programming (goal ≥90%). The secondary outcome measure was the reduction in medication errors related to administration of infusion pump medications (goal = 0%). Balancing measure refers to no delays in first case starts in radiology (goal >90%). Case start was defined as the time at which the anesthetic or sedative medication administration was started. In addition, a cross-sectional analysis was done via Survey Monkey to assess delays in case starts.
The number of data forms available each day was cross-verified with the number of cases where infusion medication was used that day by 1 of the 4 independent patient flow coordinator RNs to ensure that no data were overlooked. The number of cases conducted per week was monitored to ensure that the data captured were representative of the sample population. The process measure directly collected the data for the outcome measure (eg, if the nurse completed the double-check form, that form was used to check whether the first person made an infusion pump medication “potential” error).
Data obtained were analyzed each week in terms of proportions. The primary and secondary outcomes were represented with statistical control charts and presented to the QI team on a monthly basis. Median was applied to assess the statistical process control. These charts helped display and analyze any variations in the time series data. Special cause variation was used to evaluate the effectiveness of interventions. Special cause variation was considered to be present based on a shift (≥8 consecutive points), a trend (6 consecutive points), or alternating points (14 consecutive data points).14 Any special cause variation was investigated to learn about the reason for the change, and the charts were annotated. The process was regarded as having a change, and a new baseline was created when the special cause coincided temporally with a plausible explanation.
A total of 24 PDSA ramps were tested. The baseline data are shown for 4 weeks. In the first few months of testing the key interventions, the run chart showed special cause variations, specifically 8 consecutive points above the centerline (Fig 4). The centerline remained consistently above the goal of 90% during the rest of the time period. The stakeholder buy-in was remarkable, and there was minimal variation with the process. The 2-person verification process was incorporated during the procedural timeout during the time annotated in the run chart (Fig 3). Fifteen months after the project started, the centerline was still above the goal, with minimal variation. The EMR was modified to record the 2-person verification. Given the success, we were able to scale the process to other anesthesia locations of the hospital. The stakeholders were eager to emulate this process in other areas.
During the initial period of the project, 4 programming errors were identified and rectified before medication administration (Fig 5). Two errors were due to incorrect dosage or weight programming, and the other 2 resulted from programming the incorrect infusion pump module. The “no delay in case starts” rate was >90% before project start, during the project, and after project completion. In the cross-sectional analysis performed via Survey Monkey, 95% of stakeholders indicated there was no delay in case starts.
A standardized, team-based approach to reduce the number of intravenous medication infusion administration errors in a high-turnover area of the hospital where anesthesia and sedation are administered is described. Implementation of 2-person verification resulted in >90% medication programming being double-checked before medication administration. This change was sustained through human factor–dependent processes, such as reminders and procedural timeouts, and human factor–independent processes with inclusion in the EMR. This standardized patient-centered approach decreased the number of medication errors by early identification of programming errors. This project played a key role, as part of a larger initiative in the department of anesthesia, to reduce medication errors. Over the course of 2 years, this initiative decreased medication errors from a rate of 4 errors per month to 1 error per month.
The US Pharmacopeia Medication Errors Reporting Program showed a significantly higher rate of medication error resulting in harm or death of 31% in pediatric patients as compared with 13% in adult patients.15 The occurrence of potential adverse drug events (those not causing harm) is also 3 times higher in pediatric patients as compared with adult patients.16 Intravenous fluids are the most common product implicated in pediatric medication errors.15 Errors with high-hazard drugs can be catastrophic. Although the infusion pumps used to administer pediatric medications are “smart pumps,” they are not completely error-free because of their dependence on human factors.4 In our project, we found that although there were hard limits in the infusion pump drug library, there were also soft limits that allowed high override rates. Therefore, a continuous QI process was important to improve their safe use.4 An infusion pump informatics analytics system can be used to design a smart pump drug library.17 For a system transitioning from a paper ordering system to a computer-based ordering, it is important to know that this transition can introduce new types of medication errors, and the system should be designed to recognize such errors.18 Integration of barcodes on a smart infusion pumps and creation of closed-loop systems will also reduce intravenous medication infusion errors.
Two-person verification is used by high-reliability organizations such as the nuclear industry, the US Department of Defense, and the airline industry. In health care, 2-person verification has been used for a long time with blood component transfusion and is mandated by the Joint Commission to achieve patient safety. In the radiology imaging service, this 2-person verification was deemed a value-based approach because 2 anesthesia personnel or 1 anesthesia staff person and an imaging nurse care for every patient. Because infusion pump errors related to programming and operation are common,1 we deemed 2-person verification a reasonable approach to ensure patient safety, akin to reducing clerical errors with transfusion.
Although this project is done in the setting of an anesthesia radiology imaging service, the implications of the project relate to all areas of a hospital where infusion medications are used. With the increasing use of sedatives and other infusion medications across the hospital, reducing medication errors related to infusion pump programming is critical. The sustainability of 2-person verification was reviewed and analyzed on a weekly basis.19 The culture of change has been sustained to date by incorporating the process into the procedural timeout. Procedural timeout is a checklist created by World Health Organization to address surgical and procedural safety.13 The checklist helps with effective communication and teamwork and helps health care teams remember critical information about the patient before administering anesthesia.20 The use of a checklist is imperative for patient safety, and the radiology anesthesia service at CCHMC is 100% compliant with the use of this surgical checklist. Checklist compliance is monitored by a radiology anesthesia nurse and is overseen by the nursing clinical director of radiology anesthesia imaging. Therefore, incorporating the 2-person verification as part of the procedural timeout checklist also enabled new staff to comply with the verification process.
One limitation of this QI work is that a few PDSA cycles were introduced in a period of 6 months; it is hard to discern which of those had the highest impact. Different aspects of our approach, including constant communication and the display of the outcome measure run chart with reduction in medication errors, had the greatest impact, given the interaction with stakeholders. Another limitation is the nonavailability of QI infrastructure at hospitals trying to replicate the QI method.
Our QI project was successful in implementing a 2-person verification process to enable safe medication administration by identifying a set of key drivers and testing interventions. Analysis-driven modifications to the 2-person verification process and reduction of medication errors, which was refined via a model for improvement, led to a safe culture for intravenous medication infusion administration at a large tertiary children’s hospital. Education, constant communication, impact demonstration, and sustainability were critical to the QI effort.
Pediatricians can follow this method to acquire competence in QI21 and to apply this competence in their own practices. The safety impact is generalizable and can be implemented in any busy location of the hospital where critical intravenous medication infusion is used for emergency, urgent care, or routine elective procedures. We suggest that as a part of continuous QI, health care facilities develop a local policy to maintain and address medication errors.
We thank all the members of the QI team: Lois Curtwright (radiology anesthesia imaging, nursing clinical director), John Holcomb, Kelly Buczak, anesthesiologists, CRNAs, radiology imaging nurses, and radiology technicians.
- Accepted June 2, 2016.
- Address correspondence to Rajeev Subramanyam, MD, MS, Department of Anesthesia, Cincinnati Children’s Hospital Medical Center, University of Cincinnati School of Medicine, 3333 Burnet Ave, Cincinnati, OH 45229. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Patient Safety Movement: Zero Preventable Deaths by 2020
- Association for the Advancement of Medical Instrumentation
- Taxis K,
- Barber N
- Institute for Safe Medication Practices
- Davidoff F,
- Batalden P,
- Stevens D,
- Ogrinc G,
- Mooney S; SQUIRE Development Group
- Langley GJMR,
- Nolan KM,
- Nolan TW,
- Normal CL,
- Provost LP
- NHSScotland Quality Improvement Hub
- World Alliance for Patient Safety
- Wheeler DJ
- Cowley E,
- Williams R,
- Cousins D
- Walsh KE,
- Adams WG,
- Bauchner H, et al
- Walker IA,
- Reshamwalla S,
- Wilson IH
- Catlin AC,
- Malloy WX,
- Arthur KJ, et al
- Copyright © 2016 by the American Academy of Pediatrics