Medication Preparation in Pediatric Emergencies: Comparison of a Web-Based, Standard-Dose, Bar Code–Enabled System and a Traditional Approach

Abstract

OBJECTIVES: Increased acuity within the pediatric emergency department increases the risk of medication-related adverse events, despite the availability of validated dosing references. The eBroselow system is a standardized, web-based, bar code–enabled dosing system that eliminates the need for mathematic calculations. This study was designed to assess the accuracy of the eBroselow system and the time needed to prepare medications during pediatric simulated resuscitations compared with standard dosing references.

METHODS: This is a two-treatment, two-period crossover trial in which 13 nurses from the adult emergency department who had had pediatric advanced life support training within the previous 3 years, carried out medication dosing during pediatric code simulations. Nurses were randomized to the eBroselow system or to traditional dosing references during period one and transitioned to the opposite treatment group during period two.

RESULTS: Use of the eBroselow system resulted in a 24.6% increase in the accuracy of prepared medications, with a complete elimination of clinically significant errors (those ≥20% deviation from the recommended dose). In addition, on average, medications were prepared 8 minutes faster with the eBroselow system versus standard dosing references.

CONCLUSIONS: Use of the eBroselow system, a standardized, bar code–based, electronic medication dosing reference, increased the accuracy of medication doses prepared during pediatric code simulations by nearly 25%, with no errors being considered clinically significant.

INTRODUCTION

Medication errors are a common cause of iatrogenic adverse effects and often involve children.¹ Preventable errors are significantly more common in emergency departments (EDs) than in other hospital departments.² Kozer and colleagues³ performed a retrospective chart review to estimate the incidence and type of drug errors in a pediatric ED in order to identify factors related to an increased risk of medication errors. A review of charts for 1532 children treated in the ED of a pediatric tertiary care hospital revealed that prescribing errors occurred in 10% of all reviewed charts. In addition, trainees were more likely to commit prescribing errors, with most errors involving seriously ill patients.

Increased acuity within the pediatric ED increases the risk of medication-related adverse events. A study reviewing medication preparation during mock code resuscitations revealed that 17% of medication orders were incomplete, 16% of prepared doses deviated by at least 20% of that expected, and 7% of prepared doses deviated by 50% or more.⁴

In 2010, in response to such high percentages of medication errors reported in the literature, Drs James Broselow and Robert Luten partnered with the Pediatric Pharmacy Advocacy Group in the development of the eBroselow Initiative, which serves as the underlying standard for the eBroselow software.⁵ The goal of the initiative is to develop a simple, safe, and effective international standard for acute pediatric administration of drugs. The eBroselow emergency care software contains peer-reviewed acute care content, including information for medication dosing and medication preparation, as well as infusion and monitoring parameters, for more than 1000 medications. The integration of bar code scanning of medication vials allows for the quick population of weight-specific dosing information. Several hospitals have adopted eBroselow; however, focused studies regarding improved outcomes and safety with the eBroselow system in pediatric patients have not yet been published in the medical literature. The purpose of this study was to compare the accuracy and timeliness of medication preparation during pediatric code simulations when using the eBroselow software versus traditional dosing resources.

MATERIALS AND METHODS

This was a prospective, two-treatment, two-period crossover, Institutional Review Board–approved trial comparing the accuracy and timeliness of medication preparation during pediatric code simulations. The study was conducted at an academic medical center consisting of an adult ED and an adjacent pediatric ED, with a total volume of approximately 65,000 patients per year, with 90,000 visits in 2012. Participants included adult emergency medicine nurses who had limited pediatric nursing experience and had completed pediatric advanced life support training within the previous 3 years. The institution’s emergency medicine pharmacist identified nurses who met inclusion criteria and requested their participation in the study. All participants were required to give informed consent prior to randomization. Once informed consent was complete, a 10-minute, standardized tutorial on the use of the eBroselow system was provided. Participants were given instruction on how to search for specific medications, how to find key features regarding medication preparation, and how to complete bar code scanning of medications. Participants were then randomized to one of two treatment groups for period one. Participants randomized to the intervention group were allowed to obtain medication information from the eBroselow software or by consultation of a pharmacist. Participants randomized to the control group were allowed to obtain medication information from an electronic dosing spread-sheet routinely used in the institution’s pediatric ED, from Lexicomp’s Pediatric & Neonatal Dosage Handbook, 19th edition,⁶ or by consultation of a pharmacist. The institution’s electronic dosing spreadsheet allows users to input the patient’s weight into a cell. Once the input of the weight is complete, dosages in mg and mL for select, common rapid-sequence intubation and resuscitation medications at standard concentrations are populated for patient-specific dosing. The list of medications is not all-inclusive; however, it reflects medications commonly used in emergencies in our pediatric ED. In period one, participants completed a simulated asthma-induced respiratory failure case. The scenario facilitator, a pediatric advanced life support–trained pharmacist, began the scenario with a standardized description of the patient’s clinical situation. Thereafter, he or she supplied the participant with a continuous stream of standardized clinical data and provided requests for 5 medications to be prepared by the participants (Table 1). Medications were to be prepared using the dosing references available to them based on their randomization. After the completion of period one, participants switched randomization groups for period two and completed a simulated hyperkalemia-induced ventricular fibrillation case, preparing 5 additional medications using the references to which they were randomized (Table 1).

Table 1.

Medications Prepared by Participants ^*

All simulations were conducted in the patient care exam rooms that were equipped with digital video recording to assist with time sequencing and review of medication preparation. A trained laboratory assistant was responsible for the recording of each scenario. During each simulation, prepared medications were delivered to the scenario facilitator, who was responsible for recording the medication used to prepare the dose, the concentration used, the volume of medication in the syringe, and the diluents and volume, if applicable. Using the digital video recording of each scenario, the primary investigator recorded the time each medication request was made by the scenario facilitator and the time the prepared medication was delivered to the facilitator, in order to determine the time it took participants to prepare each requested medication. If the participant requested a pharmacist consultation, 1 minute was added to the time it took the participant to prepare the medication. One minute was chosen as an average time it would take to contact a pharmacist to make an inquiry about medication dosage or preparation.

The primary objective was to compare the accuracy of medication preparation, including drug selection and drug dosing, when using traditional reference materials versus eBroselow software. Prepared doses deviating ≥20% from the recommended dose were considered clinically significant deviations. The secondary objective was to compare the timeliness of medication preparation between each group. Descriptive statistics were used to describe all study variables. Medications chosen by each nurse, including concentration, were recorded as correct or incorrect. Prepared dosages were recorded as percentages. All times were recorded in minutes and reported as means for comparison.

RESULTS

A total of 13 nurses were enrolled, completed informed consent, and were randomized. Participants had a combined average of 4.8 years of nursing experience, with an average of 1.5 years being devoted to pediatric patient care (Table 2). A total of 6 of the 13 nurses spent ≥4 hours per week exclusively providing emergency care to pediatric patients. Eight participants were assigned to the intervention in period one and the control in period two. The remaining 5 participants were assigned to the control in period one and the intervention in period two (Figure 1). Nurses were randomized as they were recruited, leading to an odd number of participants in each group because the intended goal of 20 participants was not reached.

Table 2.

Summary of Study Population ^*

Figure 1.

Overview of randomization. Thirteen nurses participated in the study. In period one, 8 were randomized to eBroselow, with the remaining 5 randomized to traditional dosing references. Participants switched randomization groups in period two.

A total of 130 medication orders were given; 40 orders were prepared using the intervention dosing reference in period one and 25 orders were prepared this way in period two. The control dosing references were used to prepare 25 orders during period one and 40 orders during period two. A total of 22 of 130 prepared medication orders (17%) deviated by any amount from the recommended dosage range. A total of 6 of the 22 dosage errors (27%) were considered clinically significant because they deviated by ≥20% of the recommended dose (Table 3). All 6 clinically significant dosage errors were made during period two in the control group. Clinically significant dosage errors ranged from 50% to 900% deviation from the recommended dose (Table 4).

Table 3.

Medication Preparation Errors ^*

Table 4.

Clinically Significant Medication Preparation Errors *

The largest error—900%—resulted from a participant selecting an incorrect concentration of epinephrine to prepare an intravenous dose for a patient in ventricular fibrillation. The order was prepared using a concentration of epinephrine 1 mg/mL rather than the correct 0.1 mg/ mL concentration, leading to a 10-fold overdose. A total of 4 of the 6 errors resulted in underdosing due to the participant entering a weight of 3 kg rather than 33 kg in the electronic dosing spreadsheet routinely used in the institution’s pediatric ED; therefore, all doses were based on the lower weight.

The smallest error (50%) occurred because of an error in a manual calculation performed by a participant when attempting to calculate a dosage of intravenous dextrose for the treatment of hyperkalemia. A total of 16 of the 22 incorrectly prepared medications were considered non-significant errors because they deviated from the recommended dosage range by <20%. A total of 13 of the non-significant errors were prepared using the control dosing references, whereas the remaining 3 were prepared using the intervention dosing reference. All 3 errors in the intervention group occurred as a result of the participant preparing a 0.36-mg dose of epinephrine versus a 0.33-mg dose. This error can likely be attributed to the participants using 5-mL syringes with graduates of 0.2 mL to prepare these doses, with the assumption that each graduate was 0.1 mL, leading to preparation of a slightly increased dose.

The use of the intervention dosing reference resulted in faster medication preparation for nearly all prepared medications. Medication preparation took an average of 1.9 minutes per medication during period one for participants using control dosing references, compared with an average of 1.4 minutes per medication for those participants using the intervention dosing reference. One of the 5 medications prepared during period one, dexamethasone, was not included on the electronic dosage spreadsheet, which forced participants to use the pediatric dosage handbook and/or consult a pharmacist for medication dosing. This resulted in the largest time difference for medication preparation in period one, an average of 3.4 minutes versus 1.8 minutes in the control group and intervention group, respectively.

Medications that were included on the electronic dosage spreadsheet were prepared slightly faster than those for the control group during period one, compared with the intervention group (mean 1 minute versus 1.2 minutes, respectively; Figure 2). All medications in period two were prepared faster by participants in the intervention group compared with those using control dosing references (mean 1.8 minutes versus 2.8 minutes, respectively; Figure 3). Medications prepared significantly faster by participants using the intervention reference included complex medications, such as dextrose and dopamine. An intravenous dose of dextrose for treatment of hyperkalemia was prepared an average of 3.2 minutes faster and without the need for pharmacist consultation when participants were able to prepare the medication using the intervention reference. Similarly, a dopamine drip was prepared an average of 1.6 minutes faster and without the need for pharmacist consultation in the intervention group. Interestingly, pharmacist consultation was not needed by participants in the intervention group during period two; however, a pharmacist consultation was required for 35% of medication orders in the control group.

Figure 2.

Timeliness of medication preparation, respiratory case (period one). Average time (in minutes) of medication preparation by participants, for each medication requested during period one.

Control; Intervention

Figure 3.

Timeliness of medication preparation, cardiac case (period two). Average time (in minutes) of medication preparation by participants, for each medication requested during period two.

Control; Intervention

DISCUSSION

For adults, the reported incidence of errors in treatment with medications ranges from 1% to 30% of all hospital admissions, or 5% of orders written.⁷ In pediatrics, however, this number has been reported to be as high as 1 in 6.4 orders. Children vary in weight, body surface area, and organ system maturity, affecting their ability to metabolize and excrete medications. In addition, there are few standardized doses for children, with most medication dosing requiring body weight calculations. Therefore, the causes of drug errors in pediatric patients are multifactorial. The American Academy of Pediatrics (AAP) is committed to decreasing medication errors in the treatment of children and to the development of systems designed to identify and learn from errors. As part of working toward reducing medication errors, the AAP and the Journal of Pediatric Pharmacology and Therapeutics recommend, when reasonable, the use of a computerized system to check medication doses and dosing schedules.^8,9

Computer-assisted dosing in pediatric medicine has increased, with the goal of reducing medication dosing errors and preventing adverse medication reactions. In 2002, Shannon and colleagues¹⁰ developed a web-based computer program in order to increase accuracy and speed up calculations during resuscitations. They performed a validation study comparing accuracy and speed of the computerized calculator with the conventional paper-based calculation methods. The computerized program requires input of the patient’s age, which is then used to calculate an average weight based on the 50th percentile for children. The system was tested on 20 medical staff members in a controlled setting, in which each participant was asked to calculate the weight and then a set of resuscitation requirements for 3 patients. On average, using the computer program afforded 21.4% greater accuracy than the paper-based method. Additionally, participants completed tasks 11.5 minutes quicker, on average, when using the computer program.

In 2010, Yamamoto and Kanemori¹¹ performed a paired, unblinded sequenced trial involving ED and pediatric intensive care unit nurses completing medication dosing with a computer method and conventional methods, including calculators, dosing tools, and handbooks. Using these methods, nurses were asked to calculate drug administration volume, dose per kilogram, drug dose when given a specific weight, and infusion rates. To complete all tasks, the mean conventional method total time was 1243 seconds versus the mean computer program total time of 879 seconds. In addition to significantly reducing the time taken to complete each task, the number of errors for all tasks was significantly reduced, with a mean of 1.8 errors in the conventional manual method versus a mean of 0.7 errors for the computer program. The group concluded that the use of computerized assistance reduced errors and the time required for drug administration calculations. Each of these studies led to implementation of computer-based dosage calculation programs at the respective study center.

Our study is similar to these previously performed trials, with the inclusion of emergency medicine nurses and the use of a web-based, computerized dosing tool. All medical content included in the reference is peer reviewed prior to inclusion. After selection of weight, either using the validated Broselow-Luten color code or, if weight is known, selecting from the appropriate weight range, medication selection is possible and will result in an mg and mL weight-specific dose. Medication selection can be further simplified by the use of a bar code scanner. Once the medication bar code is scanned, an mg and mL standard, weight-specific dose is populated based on the concentration of the medication that was scanned. During our study, nurses were provided with varying concentrations of epinephrine and dextrose; however, they were not required to use bar code scanning to populate medication doses. Therefore, the feature of bar code scanning was not able to be fully evaluated and we were unable to determine whether errors such as the one in the control group, caused by preparation of an epinephrine dose using the incorrect product concentration, could be fully eliminated by mandating scanning of all medication vials when using eBroselow. This feature of the program is important and should be evaluated in future studies, especially during an age in which frequent drug shortages often require variation of medication concentrations supplied, based on availability. This in turn could lead to incorrect preparation of doses if concentration is not taken into consideration.

On average, using eBroselow afforded 24.6% greater accuracy than the conventional method, with the elimination of preparation errors with >20% deviation from recommended doses. Additionally, participants performed all 5 medication scenarios 8 minutes quicker, on average, when using the computer program. The <20% errors that eBroselow was unable to eliminate were those in which nurses estimated graduates on a 5-mL syringe to be 0.1 mL versus the actual 0.2 mL, resulting in a larger volume of medication being drawn up than necessary. The most significant improvement in time required to prepare medications was seen with those medications requiring dilution to be administered through a peripheral intravenous line, such as dextrose 50% and dopamine for infusion.

Without the use of the eBroselow program, nurses required pharmacist consultation in 7 of 8 scenarios in order to prepare an intravenous dose of dextrose, and in 6 of 8 scenarios in order to prepare a dopamine infusion. With the use of the eBroselow program, the preparations of dextrose and dopamine did not require pharmacist consultation and were prepared on average 3.2 minutes and 1.6 minutes quicker, respectively, compared with conventional methods. It is speculated that the time to prepare these medications would have been significantly longer if pharmacist consultation was not available to participants. This is of clinical significance, given that nurses may have to leave the bedside to gather information on the preparation of these types of medications and may not be readily available for continued clinical assessment of the patient or for preparation of other vital medications. In contrast, all medications that required no dilution and could be given as straight drug from the vial were prepared, on average, in 1.4 minutes per medication in the conventional and eBroselow groups. Although the time to prepare more simple medications is not significantly faster with the eBroselow system compared with conventional methods, these medications are prepared with increased accuracy, and the time for preparation of more complex medications is reduced.

We do realize there are particular limitations of this study. One limitation is that participants could not be blinded to treatment, which could introduce some bias because they were made aware of the outcomes being measured during the informed consent process. Additionally, as we were performing simulated resuscitation scenarios, we can only estimate the impact on outcomes in the true clinical setting.

CONCLUSION

Prescribing errors in pediatrics remains high despite the use of validated dosing references and tools. However, the use of a standardized, bar code–based, electronic dosing system, such as eBroselow, increases the accuracy of medication doses prepared during pediatric code simulations by 24.6%, with the elimination of clinically significant errors, and also reduces the time required for medication preparation.

ABBREVIATIONS

AAP

American Academy of Pediatrics

ED

emergency department