Deviation From National Dosing Recommendations for Children Having Out-of-Hospital Emergencies

We performed a retrospective cohort study using deidentified prehospital patient care records from the 2020 to 2021 ESO Data Collaborative (Austin, TX). The ESO Data Collaborative is a large dataset of electronic health records from approximately 2000 US-based EMS agencies.²⁶  The dataset includes data on dispatch, demographics, clinical presentation, vital signs, assessments, and interventions performed by EMS clinicians. A standard dataset is created annually for research purposes and is available at no cost following a research proposal process. This study was approved by ESO and by the Ann & Robert H. Lurie Children’s Hospital of Chicago Institutional Review Board. We included all pediatric (<18 years) EMS encounters during the study period who received at least 1 nonnebulized medication. We excluded encounters without a reported age.

We investigated the following 10 drugs: lorazepam, diazepam, midazolam, fentanyl, hydromorphone, morphine, ketorolac, epinephrine, diphenhydramine, and methylprednisolone. Of these, dosing for some medications (benzodiazepines, diphenhydramine, and epinephrine) could be used for multiple indications. For benzodiazepines, we limited our evaluations to encounters with an impression of seizures. For diphenhydramine, we limited our evaluation to patients with an impression of allergy or anaphylaxis. Given the substantially different uses of epinephrine based on indication, this medication was considered in 2 ways: as an intravenous (IV) and as an intramuscular (IM) drug. IM epinephrine was only evaluated for patients with a clinician impression of allergy or anaphylaxis.

We acquired the following descriptive variables of our sample: age, sex, type of service (advanced life support [ALS], basic life support, or critical care), transport disposition, agency status (volunteer, nonvolunteer, or mixed), agency type (community nonprofit, fire department, governmental nonfire, private nonhospital, and tribal), and primary and secondary clinician impressions.

We used 2 reference standards for medication dosing, accounting for differences in dosing by route and maximum doses. For the epinephrine, we considered appropriate dosing to be from the Pediatric Advanced Life Support (PALS) guidelines, which provides nationally recognized dosing recommendations of this drug for both cardiac arrest and anaphylaxis.²⁷  For the other 9 medications, we used dosing recommendations from the National Association of State EMS Officials (NASEMSO) Model EMS Clinical Guidelines.²⁸  The NASEMSO guidelines are produced by a multistakeholder panel, including the NAEMSP, the American College of Emergency Physicians, and the American Academy of Pediatrics Committee on Pediatric Emergency Medicine. Importantly, the NASEMSO guidelines do not provide a dosing range for medications and instead recommend a fixed weight-based dose with maximums (for example, instead of recommending 0.25–0.5 mg/kg to a maximum of 30 mg for IV ketorolac,²⁹  the NASEMSO guidelines recommend 0.5 mg/kg to a maximum of 15 mg without a range²⁸ ). This is consistent with EMS protocols that provide standing orders from EMS medical directors to EMS clinicians for specific indications (Table 1).^30,31 

Among all encounters, we identified the proportion who received at least 1 nonnebulized medication. We identified encounters who received a study drug and identified the number and percent of encounters who had a weight recorded among these. We calculated the ideal dose of each medication for each encounter in 2 ways. First, in encounters which had a weight (in kilograms) available within the ESO dataset, we calculated the appropriate weight-based dose. Second, as not all encounters had a documented weight, we estimated weight from age using the “Best-Guess” formula, which has been previously demonstrated to have greater accuracy for age-based weight estimation compared with other techniques.^32,33  We defined a medication dosing deviation as being ≥20% from the ideal weight-appropriate dose on the basis of previous literature evaluating dosing errors in children.^23–25,34  For drugs with different routes of delivery, we evaluated dosing deviations within these subgroups, combining together IV and intraosseous routes. We stratified our findings among the 3 major agency types (community nonprofit, fire, and government nonfire) and 2 major agency statuses (nonvolunteer and mixed). As post hoc exploratory analyses, we repeated our analysis for (A) methylprednisolone on the sample of patients with asthma and allergy or anaphylaxis on their primary or secondary impression and for (B) IV epinephrine on the sample of patients with a primary or secondary impression of cardiac arrest. Analyses were performed using R, version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

20 929 045 records were identified over the 2-year period. After applying exclusions, we identified 990 497 children. Of these, 63 963 (6.4%) received at least 1 nonnebulized medication (Fig 1). Among this sample, the median age was 13.4 years (interquartile range 5.5–16.2 years), and 54.9% were boys. Most encounters were treated by an ALS provider (93.8%) and were transported to the hospital (91.7%). Demographics of the included sample are provided in Table 2.

Among all nonnebulized doses administered, 53.9% were of the selected studied drugs. Most patients who received a study drug had a documented initial dose provided that could be used to compare it to the ideal weight-based dose (Table 3). A documented weight was present for 80.3% of children who received a study drug; this was highest for fentanyl (83.0%) and lowest for IM epinephrine (71.8%).

When using encounters with a recorded weight, the overall consistency with national guidelines was 42.6 per 100 administrations. The medications with the highest and lowest proportion dosed consistently with national guidelines were methylprednisolone (75.1%), IM epinephrine (67.9%), and ketorolac (56.4%). The medications with the lowest proportion dosed consistently with national guidelines were diazepam (19.5%), lorazepam (21.2%), and morphine (23.2%) (Fig 2). Most deviations represented an underdose (ranging from 5.4% for IV epinephrine to 74.7% for lorazepam). Overdosing occurred less frequently. The medication most frequently overdosed was IV epinephrine (49.3%) and least frequently was morphine (3.0%). When evaluating differences based on route, a higher proportion of children given IV diazepam were underdosed compared with those given PR diazepam, who were more frequently overdosed (Fig 3). A higher proportion of children given intranasal and IM midazolam were underdosed compared with those given IV midazolam. Dosing with ketorolac was more frequently low when given IM compared with IV. Overdosing with diphenhydramine was more common when given IV compared with orally. When comparing between agency types, findings for both overall and specific drugs were similar. Community nonprofit organizations were accurate in 41.5% of the time, fire-based services were accurate 42.7% of the time, and government nonfire based services were accurate 46.7% of the time (Supplemental Table 4). When comparing mixed and nonvolunteer agency statuses, our findings were again similar between both (Supplemental Table 5).

Our overall findings were similar when estimating dosages from weights calculated by age. The overall consistency with national guidelines was 35.5 per 100 administrations. Underdosing again was more common. The medication underdosed most frequently was lorazepam (84.4%) and least frequently was IM epinephrine (13.1%). The medication most frequently overdosed was IV epinephrine (53.5%) and least frequently was lorazepam (2.1%). Our findings when evaluating differences in dosing based on route were similar when using these calculated weights to when using actual weights (Supplemental Fig 4).

When limiting our evaluation of IV epinephrine to patients with cardiac arrest, dosing was appropriate in 47.2%, too high in 48.4%, and too low in 4.4% among patients with a recorded weight (n = 1918). When using estimated weights for age (n = 2504), the dose was calculated to be appropriate in 29.8%, too high in 51.8%, and too low in 18.3%. Dosing deviations were similar among patients with steroids among those with a primary or secondary impression of an allergic reaction or anaphylaxis and those with a primary impression of asthma (Supplemental Fig 5).

We used a multiagency dataset to evaluate drug dosing for commonly used medications provided to children in the prehospital setting. There was substantial variance in weight-based dosing from national guidelines for commonly used pediatric medications. The extent of dosing deviation varied by medication, with some (diazepam, lorazepam, and morphine) having substantially more deviation than others (methylprednisolone, IM epinephrine, and ketorolac). These results were stable across common agency types and statuses and when using age-based weight estimation techniques. Our findings suggest the need for rigorous research and quality improvement efforts to improve guideline adherence and to minimize dosing errors.

There are likely several drivers for the observed deviation from national guidelines. First, some of the deviations represent differences in prehospital protocols across EMS systems compared with national guidelines. Such variations have previously been highlighted for the prehospital management of pediatric seizures³⁰  and respiratory distress,³¹  and may explain why some medications had larger proportions of dosing deviations than others. For example, seizure medication dosing was consistently low compared with national guidelines. Dosing for midazolam across state guidelines varies widely, whereas some states use an IM dose of midazolam of 0.2 mg/kg, others use 0.1 mg/kg.³⁰  Only 3 states at the time of that publication explicitly used the NASEMSO guidelines. Even within the NASEMSO guidelines, it may be challenging for EMS provider clinicians to rapidly identify appropriate pediatric drug dosages in real time, an issue that may carry over to many individual EMS agency protocols given the number of conditions encountered by EMS personnel.^28,30,31  Although some EMS protocols are mandated statewide, others are optional, and some states do not provide any statewide guidance, leaving protocols to be developed locally. This finding suggests opportunities exist for improved education of EMS clinicians and EMS medical directors, along with opportunities to make EMS protocols more consistent with national guidelines.

Dosing errors are likely an additional driver for our findings. We build on existing knowledge of dosing variation found in prior regional or condition-specific research and in other pediatric care settings. In 1 study performed among 8 EMS agencies in Michigan, erroneous dosing occurred in 34.7% of administrations for 6 EMS-administered medications (albuterol, atropine, dextrose, diphenhydramine, epinephrine, and naloxone).²⁴  A more recent study performed on the statewide Michigan EMS system similarly demonstrated that medications given with a high error rate included morphine, fentanyl, and midazolam, though the exact percentages of error were limited by low counts for some of these drugs.²³  Other work has demonstrated a high proportion of dosing errors, specifically for seizure medications, with underdosing occurring more frequently than overdosing.^19,20,22  In contrast, a systematic review of pediatric and neonatal intensive care settings identified a median rate of medication errors of 14.6%, a sizeable but lower proportion compared with the variance seen in the current study.³⁵  Taken together, these findings suggest there may be even greater opportunities for improving dosing of pediatric medications in prehospital compared with in-hospital acute care settings.

We note differences in dosing deviations between drugs. Potentially, some of these findings may reflect hesitation among EMS clinicians or EMS medical directors to provide adequate dosing out of a concern for oversedation, a finding which may be reflected in the frequent underdosing of benzodiazepines and opioid medications.³⁶  In contrast, other medications, such as ketorolac, epinephrine, and methylprednisolone were more frequently appropriately dosed compared with these medications. An approach to address this will require greater implementation of expert, consensus-based guidelines combined with greater efforts toward training paramedics on the appropriate use of potentially sedating medications.

To improve care delivery, a strategy that incorporates research, quality improvement, and implementation-science based approaches will be required. As findings were similar across agency types and statuses, a comprehensive approach is warranted to address these gaps. Additional research and implementation efforts are needed to evaluate and monitor common-sense practice changes. Qualitative research studies have suggested specific challenges in pediatric medication dosing that include difficulties in weight estimation, incorrect use of the Broselow tape, and difficulty with calculations during stressful situations.¹⁹  A position statement from the NAEMSP suggests a multimodal approach to address this issue, including utilizing dose-derivation strategies that avoid use of calculations, using kilograms as a weight standard, confirmation of weight at the time of drug delivery, improving training through simulations, and use of standardized formularies.³⁷  One focus group study suggested that improvements can be made in simplifying dose delivery and that length-based tape, pediatric checklists, and dose cards may help reduce dosing errors.³⁸  A study performed in one urban EMS system demonstrated that requiring clinicians to report the color zone (which provides a weight approximation) of the Broselow tape to the base station was associated with an improvement in the dosing of IV epinephrine for cardiac arrest.¹⁷  Another study by the same investigators showed that errors in antiseizure medications could be reduced through the implementation of standardized formularies with precalculated medication dosages.¹⁶  A similar approach with prefilled syringes has been demonstrated to be effective in pediatric resuscitation scenarios.³⁹  Other investigators have reported success in the use of dedicated simulations or real-time aids to reduce dosing errors.^18,20,40,41  Although the optimal strategy to improve dosing accuracy in the prehospital setting is uncertain, our findings identify a clear need for the implementation of such strategies to address substantial variance in dosing of medications for prehospital pediatric patients.

Our findings are subject to limitations. This was a retrospective study that used previously charted data, and therefore may be subject to inaccuracies with drug dosages, timing, route of administration or weight. The ESO dataset does not contain geographic data or protocol information to account for weight-based dosing guidelines within local EMS protocols that differed from NASEMSO or PALS guidelines. Weights in the dataset may be obtained through varying ways and may not have been known before the time of drug delivery. We attempted to mitigate this issue by using 2 approaches toward ideal weight-based dose calculation. Despite these limitations, our findings suggest important deviations in pediatric drug dosing using a large, multiagency dataset, demonstrating important research, implementation, and quality improvement needs.

Using a nationally representative prehospital dataset, we identified a wide range of deviations in dosing of medications administered to prehospital pediatric patients compared with guideline-recommended weight-based doses. This was greater for some drugs, such as diazepam, lorazepam, and morphine, than for others, and was notable across different agency types and statuses. As these findings may have multiple causes, including differences in protocols and errors in dosing, addressing these deviations should be a target for future educational, quality improvement, and research activities.

We thank ESO (Austin, TX) for provision of the data used in this study.

ALS

advanced life support

EMS

emergency medical services

IM

intramuscular

IV

intravenous

NAEMSP

National Association of EMS Physicians

NASEMSO

National Association of State EMS Officials