Effect of an Intervention Standardization System on Pediatric Dosing and Equipment Size Determination: Title and subTitle BreakA Crossover Trial Involving Simulated Resuscitation Events FREE

RECENT EFFORTS to gauge the frequency and gravity of medical errors in the United States have yielded alarming results, stimulating a national debate on error prevention.^{1 - 2} These studies^{2 - 6} have demonstrated that a large percentage of adverse events are drug related and that dosing errors contribute most heavily to adverse drug events. Pediatric medication dosing, because of complexities that are largely unique to the pediatric patient, is particularly fraught with error.^{6 - 8} Manual dose calculation using dosing equations, a practice nearly ubiquitous in pediatric treatment and comparatively rare in adult patient care, has been pinpointed as a high-error activity.^{7 ,9 - 13} Indeed, a well-reported longitudinal study^{7 ,14} that focused on in-hospital medication prescribing errors found the pediatric service to be the most error ridden of all hospital services and flawed dose calculations to be the most common type of pediatric prescribing error.

Several factors compound the risk of error when pediatric care must be delivered emergently. Arguably, the most significant of these is the need to estimate otherwise inaccessible patient data for input into medication dosing formulas. Patient weight is the most common of these variables, and evidence suggests that both physician and nurse estimates are often unreliable.^{15 - 16}

Limited opportunities for prescription monitoring and double-checking, lack of pediatrics-trained emergency physicians, inability to rely on standardized dosing (ie, adult ampule), and the stress of managing a life-or-death situation involving a child may further complicate emergency treatment. Significantly, the same study^{7 ,14} that found the pediatric service to be the most error ridden found the emergency department to be a close second.

In the early 1980s, James Broselow, a North Carolina–based emergency physician, developed a simple tool to improve weight estimation using established height-weight correlations.¹⁷ Today, the Broselow Pediatric Emergency Tape and accessories provide physician and emergency services personnel with standardized, precalculated medication doses, dose delivery volumes, and equipment sizes using color-coded zones based on similar height-weight correlations. The tools were developed in an attempt to alleviate or lessen each of the complicating factors listed herein.

Clearly, the improvement in care realized when using the Broselow tape and color-coded materials depends on the accuracy of the precalculated values and ease of system use. Accuracy in estimating patient weights and endotracheal tube size has been reported.^{16 ,18 - 20} To date, however, no study has attempted to estimate the value of this color-coded system in reducing error severity in the clinical situation it was devised to facilitate, the pediatric emergency. We hypothesized that use of the Broselow tape and color-coded materials would result in a decrease in deviation from recommended medication doses and equipment sizes and an increase in physician comfort level. We present the results of a clinical trial of color-coding use by resident physicians in simulated pediatric stabilization scenarios.

The study design was a 2-treatment, 2-period crossover trial with randomization of treatment. Each resident participated in 4 simulated stabilization scenarios. The first 2 scenarios (period 1) were randomly assigned to either color coding (intervention) or no color coding (control). The second 2 scenarios (period 2) were subsequently assigned to the opposite treatment (Figure 1). Investigators were not blinded to scenario assignments.

Learning effect, a potential problem in crossover trials, denotes enhanced performance in period 2 scenarios, resulting from participation in and experience gained from period 1 scenarios. The influence of learning effect in this study was minimized through both randomization and assignment (distribution) of scenarios. As noted herein, the treatment (intervention or control) was randomly assigned to periods 1 and 2 for each participant. Therefore, if learning was carried over to the second period, performance would be enhanced equally for intervention and control scenarios. Moreover, across all participants, scenarios were equally distributed among the periods and treatments. In this way, if level of difficulty varied among scenarios, there would be no consistent effect on performance. Last, the differences among the 4 scenarios, specifically the differences between the 4 simulated patients, and the inherent differences in medication dosing and equipment sizing with and without color-coding tools served to reduce the relevant learning carried over into the second period.

A crossover design was chosen for 2 reasons. First, with each participant serving as his or her own control, we were able to increase the statistical power available given the limited number of participants. A parallel design, in contrast, would have required splitting the study population into 2 groups, one assigned to the intervention and the other to the control. Second, intersubject variability with regard to clinical experience was kept from confounding the analysis, again because each participant served as his or her own control. Data were collected between December 1, 1999, and February 29, 2000.

All pediatrics and medicine-pediatrics combined program resident physicians at a tertiary, academic medical center were invited to participate. This population is likely more familiar with conventional pediatric dosing and equipment sizing than many emergency physicians. The involvement of this group was expected to yield a conservative estimate of the benefits of a color-coded system. First-year postgraduates were not included because of their relative inexperience with resuscitation situations. Institutional review board approval was secured, and all participants gave informed consent.

Several days before scheduled study involvement, participants were given sample color-coded materials and asked to review the organization of the listed, precalculated doses and equipment sizes. The participants were also asked to review traditional resuscitation reference materials (eg, Pediatric Advanced Life Support Survival Card,²¹ The Harriet-Lane Handbook²² ) and to bring those materials to the study session.

On arrival to the study session, participants were fitted with a Holter heart monitor for the collection of per minute heart rate data. Heart rate was used as a surrogate measure for comfort level. This method is modified from a study that describes the relationship between stress and heart rate in surgeons.²³ A standardized, 20-minute, interactive presentation on the development, intended benefits, and practical use of the Broselow tape and color-coded materials and rules for participation in the subsequent clinical scenarios followed. Materials used in the study included the Broselow tape and adjunct color-coded reference sheets (Figure 2). The participant was taught to (1) use the Broselow Tape to measure the patient from head to heel to accurately determine the patient's length-based color zone and (2) reference similarly colored sheets for precalculated doses and equipment sizes.²⁴

Each resident participated in 4 successive, simulated, and randomly ordered resuscitation scenarios: asthma-induced cardiopulmonary arrest in an adolescent, meningococcal septic shock in a toddler, 40% body surface area burns in an infant, and status epilepticus in a school-aged child. Across participants, similar numbers of each scenario were assigned to intervention and control and period 1 and period 2 (Figure 1). An appropriately sized, anatomically and developmentally correct mannequin was used to represent the patient in each scenario. The facilitator, a pediatric emergency medicine specialist with extensive experience in teaching through simulated resuscitation scenarios (K.F.), began the scenario with a standardized description of the patient's clinical situation. Thereafter, she supplied the participant with a continuous stream of standardized clinical data and answered any specific questions regarding the patient's clinical findings. The same physician facilitated all scenarios for all participants.

The participant assessed the patient given the clinical data and made treatment decisions, providing specific doses and equipment sizes. The participant was free to use references other than the Broselow tape in all scenarios. In addition, equipment (eg, nasogastric tube, endotracheal tube) of all sizes was available to aid the participant in selecting appropriate equipment sizes. All scenarios were videotaped for scoring purposes.

Immediately following participation, participants completed a 5-question survey that explored previous experience with the Broselow tape and accessories, perceived benefits of color coding, and likelihood of future system use. Participants were assured anonymity with regard to survey responses.

The primary outcome measure was deviation from recommended dose range (DRDR) during the management of simulated pediatric resuscitation events by resident physicians. Secondary outcome measures were deviation from recommended equipment size (DRES) and physician comfort level during the management of simulated pediatric resuscitation events by resident physicians.

For each prescribed medication, DRDR was calculated as the absolute value of the percentage deviation from the recommended dose or dose range. Median DRDR, largest deviation, and interquartile range were calculated for all medications. No attempt was made to grade the clinical severity of individual dosing deviations. A DRDR summary score was calculated for each scenario (4 per resident) by averaging the percent DRDR for each medication prescribed in the scenario. In calculating summary scores, no distinction was made between similar degrees of underdosing and overdosing. Additionally, similar dosing deviations (eg, 25% error) for dissimilar medications (eg, epinephrine and midazolam) were considered equal. Fluid dosing was included under medications.

Because equipment size is not a continuous variable, DRES was calculated as the absolute value of the number of sizes between the size prescribed by the participant and the recommended size or range of sizes. A summary score for each scenario (4 per resident) was calculated by averaging the size deviation for each piece of equipment ordered in the scenario. As with medication dosing, no distinction was made between similar degrees of undersizing and oversizing, and no attempt was made to classify the severity of individual equipment sizing deviations. Also, similar sizing deviations (eg, 1 size between prescribed value and acceptable value) for dissimilar equipment (eg, endotracheal tube and nasogastric tube) were considered equal.

Reference values were initially compiled using pediatric emergency medicine texts and guides.^{22 ,24 - 29} A panel of pediatric emergency medicine, pediatric intensive care, pediatric nursing, and respiratory therapy specialists then reviewed all values. The final recommended range for each medication and piece of equipment was made as broad as necessary to encompass varying beliefs and practices. All recommended values fell inside the dosing range indicated by the Broselow Pediatric Emergency Tape and color-coded sheets.

The DRDR summaries were grouped by treatment (intervention vs control). For each resident, 2 overall DRDR scores were calculated: the first by averaging the summary scores for the 2 intervention scenarios and the second by averaging the summary scores for the 2 control scenarios. The overall intervention score was then subtracted from the overall control score to generate a DRDR difference score for each resident. The median of the DRDR difference scores for all residents was calculated. Because of the small sample size, a Wilcoxon signed rank test was performed to examine whether the median DRDR difference was significantly different from 0.

Per minute heart rates were averaged to generate a single average heart rate for each scenario. Scenario averages were grouped and analyzed as described in the "DRDR Comparisons" section.

A total of 28 residents, representing 69% of all pediatrics residents and 50% of all medicine-pediatrics residents, were enrolled. Of the 20 pediatrics residents, 11 were second-year postgraduates and 9 were third-year postgraduates. Of the 8 medicine-pediatrics residents, 2 were second-year postgraduates, 2 were third-year postgraduates, and 4 were fourth-year postgraduates. Conflict between residents' clinical obligations and scheduled study sessions was the only reason given for refusal of participation. Fourteen participants were assigned to the intervention in period 1 and the control in period 2. The remaining 14 participants were assigned to the control in period 1 and the intervention in period 2 (Figure 1).

One third-year postgraduate pediatrics resident participated in a trial run of the study. Technical difficulties during the trial run prevented appropriate data collection for this resident, who was assigned to control in period 1. Of the 27 residents whose data were included in the analysis, more than 50% had never used the Broselow tape or any other color-coded materials in the resuscitation setting before study participation. More than 70% had not used color coding in the nonresuscitation setting.

The median DRDR for medications prescribed in intervention scenarios was 7.6% (95% confidence interval [CI], 4.5%-9.1%). The median DRDR for control scenarios was 36.3% (95% CI, 29.3%-51.2%). The median of the difference in DRDR was 25.4% (95% CI, 19.1%-32.5%). Results of the Wilcoxon signed rank test demonstrated that the median of the difference between the intervention and control scenarios was significantly different from 0 (P<.001) (Table 1).

Four times in intervention scenarios and 54 times in control scenarios, DRDRs exceeded 100% (Table 2 and Table 3). Five-fold or greater deviations (≥500%) were found 2 times in intervention scenarios and 4 times in control scenarios. The largest single DRDR (2226%) was found in an intervention scenario. Although the Broselow tape and corresponding color-coded reference sheets were made available to all residents participating in intervention scenarios, prescribed doses associated with deviations of more than 100% in intervention scenarios did not reflect doses suggested by the Broselow tape or the sheets.

The median DRES prescribed in intervention scenarios was 0.27 size (95% CI, 0.2-0.3 size). The median DRES for equipment prescribed in control scenarios was 0.42 size (95% CI, 0.3-0.45 size). The median of the difference in DRES was 0.12 size (95% CI, 0.03-0.22 size). A Wilcoxon signed rank test indicated that the median of the difference between the intervention and control scenarios was significantly different from 0 (P<.001) (Table 1). Deviations of 2 or more sizes were found once in intervention scenarios and 21 times in control scenarios (Table 4).

The median average heart rate for intervention scenarios was 91.84/min (95% CI, 85.45-97.05/min). The median average heart rate for control scenarios was 93.24/min (95% CI, 85.65-103.30/min). The median of the difference in heart rate was 1.99/min (95% CI, −1.83 to 6.57/min). A Wilcoxon signed rank test indicated that the median of the difference between the intervention and control scenarios was not significantly different from 0 (P = .07).

Eighty-one percent of participants believed that the Broselow tape and color-coded materials would be "extremely helpful" in reducing dosing errors. Ninety-two percent believed that color coding would be "extremely helpful" in reducing equipment-sizing errors. The remaining 19% and 8% believed that the system would be "helpful" in reducing dosing and sizing errors, respectively. All participants indicated that they would "definitely" use color-coded tools more routinely in the future.

This simulation study demonstrates that many large DRDRs (>100%) may be avoided when the Broselow tape and color-coded materials are made available to physicians making dosing decisions in the emergency setting. Moreover, a 25% median decrease in DRDR was seen with color-coding use. The narrow CI surrounding the estimate indicates a high probability of substantial reduction in DRDR. Large DRESs (≥2 sizes) were also far less common when the Broselow tape and sheets were made available. These findings suggest that a color-coded system may be useful in reducing the number of clinically significant errors made when treating emergently ill or injured children.

In recent years, authors have described substantial error reduction, resulting from the implementation of computerized order entry and decision support systems.^{30 - 32} The participation of a clinical pharmacist in physician rounds has also been shown to afford significant reduction in preventable medication dosing errors.³³ However, practical and financial considerations may prevent broad adoption of such valuable systems improvements in the acute care setting, especially in the near future. For example, a major benefit of order entry, the computer generation of sequential lists of options from which the physician may select order components (eg, medication, dose), requires ready system access to data that are specific to the patient. Cross-checking for drug-laboratory and drug-drug interactions requires access to still more data.^{30 - 31} In the acute care resuscitation setting, unlike the inpatient setting, patient-specific data may not be automatically available on the computer system. Therefore, the realization of error reduction may require extensive data entry, which is clearly a time-prohibited step in the emergency situation. In addition, both implementation of an integrated computer system and round-the-clock presence of a clinical pharmacist in the emergency department may be considered cost prohibitive by some hospitals.

As demonstrated in this study, a color-coded system does not rely on access to extensive patient data or additional ancillary support to provide dosing assistance. Moreover, after only a brief instructional session, trial participants demonstrated proper use of the Broselow tape and sheets, significant reduction in DRDR, and stable heart rates, suggesting a steep learning curve and easy system integration into clinical practice. Periodic refreshment training will likely be important for maintaining improved outcomes in the long-term. Importantly, participants themselves expressed overwhelming support of a color-coded system through survey responses. Finally, a color-coded system such as the one tested does not have the same cost considerations as other noted systems improvements.

The larger sizing deviations (≥2 sizes) were most often seen with equipment types with which pediatric residents at our institution have somewhat limited experience. Emergency physicians, intensivists, and respiratory therapists often determine endotracheal tube and nasogastric catheter sizes in clinical areas other than the delivery room and intensive care nursery, whereas nursing staff typically determines urinary catheter size. These equipment types also require greater precision than other types. For example, 1 of 3 self-explanatory face mask sizes (infant, pediatric, and adult) is typically adequate to provide oxygen. Conversely, endotracheal tubes require greater attention, are available in 12 sizes, and are more difficult to size appropriately. Once children are color coded using the Broselow tape, neither lack of familiarity with nor complexity of traditional size determinations affects sizing accuracy. Accordingly, large sizing deviations (≥2 sizes) for endotracheal tubes, nasogastric catheters, and urinary catheters were seen only in scenarios in which the Broselow tape was not available.

Of the 4 deviations greater than 100% seen in intervention scenarios, 3 were observed in analgesic dosing. Because the analgesic doses prescribed were clearly not those cited on the Broselow tape or color-coded reference materials, these deviations may point to an area where the system requires modification to enhance ease of use. Notably, the fourth deviation of more than 100%, associated with antiarrhymic dosing, was the largest in the study. In light of the small DRDRs observed with other antiarrhymic doses in intervention scenarios, we believe that this large DRDR represents an outlier value and is not indicative of a trend.

The median decrease in heart rate of 2/min in intervention scenarios most likely does not represent a clinically significant reduction in physician anxiety with use of the color-coded materials. It is noteworthy, however, that average heart rate in all scenarios was high, suggesting that the simulations were able to recreate some of the anxiety associated with a real resuscitation event.

Although we have demonstrated results that favor use of the Broselow tape and color-coded materials, in studying resident performance in simulated scenarios and in using deviations in dosing and equipment sizing as outcome measures, we can only estimate the impact on outcomes in the true clinical setting. This is a common limitation of simulation studies.³⁴ However, we believe that the transition from reduction in large deviations in the simulation setting to reduction in real clinical errors with the potential for adverse outcome is a logical and natural one. Further study of effectiveness of a color-coded system in the real clinical setting is necessary to bear out this progression.

This study examines the accuracy of decision making by individual physicians; no attempt was made to recreate the complex interactions among resuscitation team members (eg, physicians, nurses, respiratory therapists). Additionally, the amount of nonclinical patient information available to participants was strictly limited. This approach was adopted in an attempt to approximate resuscitation scenarios in which parents are not available to answer questions and physicians must rely on clinical acumen. Neither patient age nor weight, both common variables in medication dosing and equipment sizing formulas, was supplied to participants in any of the 4 scenarios. When the Broselow tape and sheets were not available, participants were forced to make estimations using clues such as patient size and physical development. The Broselow tape provides weight estimates and precalculated medication doses and equipment sizes. Therefore, the effect of the Broselow tape and sheets on error severity with regard to (1) facilitated weight estimation and (2) bypassed equation look-up, recall, and calculation were lumped together. Further study is required to evaluate specific benefits of the Broselow tape and sheets with regard to error reduction.

Similar degrees of deviation from recommended dosing (measured as percentages) may not be of similar clinical significance for different medications. Therefore, the limitations inherent in summary measures (eg, overall reduction in DRDR reported herein) should be recognized. Given the limited, noncontinuous nature of the deviation measurement for equipment sizing (ie, number of sizes), the unimpressive median DRES is perhaps an inadequate summary score, not fully reflecting the number of large deviations avoided when the Broselow tape and sheets are available.

Last, investigators were not blinded to assignment of scenarios to intervention vs control. The use of hard measures (eg, deviation of prescribed dose and size from reference range) likely helped to minimize the extent to which investigators introduced bias into outcome measurement.

Appropriate use of the Broselow tape and color-coded reference materials in the resuscitation of pediatric patients may afford significant reductions in the number of dosing and equipment sizing errors that have the potential to lead to adverse outcomes. Further testing in the real clinical setting is required to confirm our findings.

In emphasizing standardization and simplification, a color-coded system may also improve dosing and sizing in other hectic and time-constrained situations, such as the nonresuscitation management of acutely ill children. Physicians with limited pediatric experience who work in community emergency departments or urgent care settings may realize the greatest benefit from such a system. Recently, investigators exploring the benefits of a color-coded system outside the emergency department setting found considerable reductions in other medical errors.³⁵

Corresponding author and reprints: Karen Frush, MD, Pediatric Emergency Services, Children's Services, Duke University Medical Center, Box 3055 DUMC, Durham, NC 27710 (e-mail: frush002@mc.duke.edu).

Accepted for publication September 19, 2002.

Dr Wears has a 2% interest in Broselow Medical Inc, Hickory, NC. He has not received any funds from the shares (no dividends, etc) and has not received any honoraria or consulting fees. He was reimbursed by Broselow Medical Inc for travel to Philadelphia, Pa, in 1991 for a concept presentation. He has not profited from his involvement in the development of the Broselow-Luten Pediatric Concept.

We thank Ira Cheifetz, MD, Heather Keenan, MD, Robert Luten, MD, and Theresa Cromling, RN, for their input in matters requiring clinical or technical expertise. We also thank Jennifer Higgins, BA, for her assistance in data management and Michael Mallory, MD, Ricardo Pietrobon, MD, Eugene Oddone, MD, and Donald Frush, MD, for their support and review of the manuscript.

The problem of medical error in the United States has been the focus of many studies in the past several years. Pediatric medication dosing is particularly fraught with error because of complexities involving dosing equations. This study was undertaken to evaluate a system designed to reduce medication dosing error by avoiding the need to perform calculations.

This study is the first to our knowledge to evaluate the effectiveness of a simple color-coded system that uses zone dosing to eliminate the need to calculate medication doses. Results demonstrate a significant decrease in deviation from recommended doses in pediatric emergency situations. Widespread implementation of such a system may lead to decreased medical error in other clinical scenarios.