A Statewide Analysis of EMS’ Pediatric Transport Destination Decisions

Abstract

Introduction:

Deciding where to transport a patient is a key decision made by emergency medical services (EMS), particularly for children because pediatric hospital resources are regionalized. Since evidence-based guidelines for pediatric transport destinations are being developed, the purpose of this study was to use a large statewide EMS database to describe current patterns of EMS providers’ transport destination decisions for pediatric patients.

Methods:

This is a retrospective study of pediatric transports from 2011–2016 in EMS Tracking and Reporting System (EMSTARS), Florida’s statewide EMS database. We included patients greater than 1 day and less than or equal to 18 years who were primary EMS scene transports. Our primary outcome variable was ‘reason for choosing destination.’ We performed descriptive and comparative analysis between closest facility and all other ‘reason for choosing destination’ choices. We used geospatial analysis to examine destination choice in urban and rural counties.

Results:

Our final study sample was 446,274, and 48.2% of patients had closest facility as their ‘reason for choosing destination.’ The next largest category was patient / family choice (154,035 patients, 35.7%). Closest facility patients were older (median age 12 versus 10 years, p<0.0001) and had shorter median EMS transport times (11.3 versus 15 minutes, p<0.0001) compared to all other destination decisions. Notably, 60% of respiratory distress patients’ and 44% of seizure patients’ reason for choosing destination was something other than closest facility. Geospatial analysis revealed that fewer rural patients were documented as closest facility compared to urban (43.9% versus 47%, p< 0.0001). Correspondingly, more rural patients’ destination decision was patient / family choice than urban patients (36.3% versus 34.3%, p<0.0001).

Conclusions:

This large, statewide study describes EMS’ reason for choosing destination for pediatric patients. We found that just under half of patients were documented as closest facility, and over one-third as patient / family choice. Significant differences in destination reasons were noted for rural versus urban counties. This study can help those currently developing pediatric EMS destination guidelines by revealing a high proportion of patient / family choice and identifying conditions with high proportions of destination reasons other than closest facility.

Introduction

Deciding where to transport a patient is a key decision made by emergency medical services (EMS). That decision is particularly important for children because regionalization of care has concentrated pediatric resources in select hospitals.¹ Despite regionalization, there are no widely-accepted evidence-based guidelines for EMS’ transport destination for pediatric patients, save for those with trauma.²

In order to safely transport children to an optimal destination without overburdening EMS resources, work has begun to develop transport guidelines for non-trauma pediatric patients. A multidisciplinary group of EMS and pediatric stakeholders constructed a 13-item consensus-based criterion for EMS to identify children in need of higher-level pediatric services such as specialty and/or critical care.³ Retrospective studies have identified specific non-trauma pediatric conditions at risk for requiring interfacility transport to definitive care after initial EMS transport to a local hospital.⁴ Since prehospital identification of specific medical conditions is not always feasible, an abnormal Pediatric Assessment Triangle (PAT) has shown promise in identifying children in need of specialty care.⁵

Meanwhile, while those studies help formulate future guidelines for pediatric medical patients, the current patterns of EMS providers’ transport destination decision-making for those patients are not yet described. For adult trauma patients, studies indicate that EMS providers use “gut instinct” or patient/family input to guide transport decisions despite evidence-based trauma transport guidelines.^6–8 Understanding current EMS pediatric destination decision-making is important for the development, dissemination, and implementation of the guidelines currently in development. Therefore, our objective was to describe EMS providers’ current transport destination decisions for pediatric patients using a large statewide EMS database.

Methods

Study Setting and Data Source

This is a retrospective study of pediatric transports from EMS Tracking and Reporting System (EMSTARS), Florida’s statewide EMS database. Florida is the state with the fourth-largest pediatric population.⁹ During the study period, EMSTARS captured approximately 74% of 9–1-1 calls in Florida.¹⁰ EMSTARS’ characteristics and data elements have been described in-depth in a prior publication.¹¹ EMSTARS uses National EMS Information Systems-National Highway Traffic Safety Administration (NEMSIS) Version 2.2.1 Data Dictionary elements,¹² as well as data elements custom to Florida.

Inclusion and Exclusion Criteria

Our original dataset was EMS encounters in EMSTARS from 2011–2016. Included encounters were primary EMS scene transports of patients ages greater than 1 day and less than or equal to 18 years, excluding those whose age was unknown. We excluded interfacility transports. Our reason for excluding patients aged 1 day or less was the assumption that those patients would likely be out-of-hospital births. Since out-of-hospital births would likely be transported with a second patient (the mother), we reasoned that EMS’ transport destination decision would also be influenced by the medical needs of that second patient.

Data Variables

Our primary outcome variable was ‘reason for choosing destination’ (a NEMSIS 2.2.1 compliant variable).¹² EMSTARS only provides for one selection for that variable, meaning that the options (closest facility, patient choice, family choice, protocol, specialty resource center, patients’ physicians’ choice, other, law enforcement choice, diversion, online medical direction, insurance status, not known, not recorded, and not applicable) are mutually exclusive.^10,12 We merged patient choice and family choice into one variable - ‘patient / family choice’ for simplicity since we reasoned that patient choice and family choice were interchangeable for pediatric patients.

Other data variables included patient demographics, clinical characteristics, and EMS variables. We categorized the EMS agency’s home county as urban or rural based on the Florida Department of Health’s classification.¹³ We used the first recorded vital sign value by EMS since evidence suggests many EMS decisions are made based on the initial assessment, including vital sign measurements.¹⁴ We created categories of abnormal or normal vital sign values based on patient age using Pediatric Advanced Life Support ‘General Vital Signs and Guidelines’ table.¹⁵ We grouped EMS-administered medications and procedures into a priori-determined clinical categories, detailed in Online Supplement e1. We also included EMSTARS’ ‘possible injury’ variable, which is defined as “…the reason for the EMS encounter was related to an injury or traumatic event…based on mechanism and not on actual injury.”¹⁰

Statistical Analysis

The primary outcome and other variables were analyzed with descriptive statistics (frequencies, percentages, mean and standard deviation or median and interquartile ranges, as appropriate). A priori based on our experience and studies finding that closest facility is a common choice,^7,8 we decided to compare patients whose ‘reason for choosing destination’ was closest facility versus those with any other ‘reason for choosing destination’ (‘Other’). A priori, we chose to include patients whose reason for choosing destination was marked as not known, not recorded, or not applicable in the ‘Other’ category. We performed univariate analysis comparing characteristics of closest facility versus other patients using unpaired t test or Wilcoxon Rank Sum tests (as appropriate for continuous variables), and Chi-square test for categorical variables. We used the Kolmogorov-Smirnov test for normality given that our sample size was much greater than 2,000.¹⁶ Missing data are noted in the results and were excluded from individual variable univariate analysis (e.g., patients with missing data for ethnicity were excluded from the chi-square analysis of ethnicity between closest facility and other patients). Statistical analysis was performed using SAS^® version 9.4 (Cary, NC).

Geospatial Analysis

Using ArcGIS Desktop 10.4.1 (Redlands, CA), we mapped each patient to the home county of the treating EMS agency. We created shaded thematic (“choropleth”) maps showing descriptive statistics of patient demographic and EMS clinical variables aggregated to county-level units, as well as each county’s rural/urban classification.¹³ We mapped the spatial patterns of pediatric admitting hospitals in relation to the study populations’ county-level statistics. We identified pediatric admitting hospital facilities from a previous study of asthma exacerbations.¹⁷ Since asthma is a common pediatric condition requiring inpatient admission, we assumed the pattern of admitting children for asthma should mimic the pattern for pediatric inpatient admissions in general.

Results

From 2011–2016 there were 3,491,241 unique patient encounters of all ages in EMSTARS. After application of inclusion and exclusion criteria, our final study sample was 446,274 patient encounters (Figure 1). Nearly half of patients (48.2%, N=208,323) had closest facility as their ‘reason for choosing destination.’ The next largest category was patient / family choice (35.7%, N=154,035). Table 1 details all ‘reason for choosing destination’ categories for the study sample with select demographics and EMS transport times. Of note, patient / family choice, patients’ physicians’ choice, specialty resource center, and online medical direction had younger median ages. The longest median transport times were seen with specialty resource center, patients’ physicians’ choice, and diversion. Nearly half of ‘possible injury’ patients were marked as closest facility.

Figure 1:

Inclusion and Exclusion Criteria Producing Final Study Sample

Table 1:

Reason for Choosing Destination for Ages > 1 day and ≤ 18 years, Florida EMSTARS 2011–2016

Reason for Choosing Destination	Overall N=446,274	Age in Years Median (IQR)	EMS Transport Time in minutes Median (IQR)	Possible Injury N=130,369
Closest Facility	208,323 (48.2%)	12 (4–16)	11.3 (7.8–17)	62,042 (47.6%)
Patient / Family Choice	154,035 (35.7%)	10 (3–16)	14 (9.2–22)	41,788 (32.1%)
Protocol	35,567 (3.2%)	11 (3–16)	14.7 (9–22)	11,793 (9.0%)
Specialty Resource Center	23,119 (5.4%)	9 (2–15)	20.5 (13–31)	9,158 (7.0%)
Patients Physicians Choice	4,381 (1%)	6 (1–14)	23.9 (14–37.4)	432 (<1%)
Other	3,322 (<1%)	12 (4–16)	17.3 (10.7–26.3)	1,359 (1.0%)
Law Enforcement Choice	2,212 (<1%)	16 (15–17)	14 (9–20)	670 (<1%)
Diversion	671 (<1%)	13 (5–17)	21 (14–30)	231 (<1%)
Online Medical Direction	410 (<1%)	10 (3–15)	17.3 (9–33)	181 (<1%)
Insurance Status	63 (<1%)	14 (4–17)	16.7 (9.4–26)	11 (<1%)

Reason for Choosing Destination = Not Known, Not Recorded, or Not Applicable = 14,171 (3.2% of 446,274)

Demographic, clinical, and EMS encounter characteristics are detailed in Table 2. About half of patients were male (48.8%), and the median age was 11 years. Age was bimodally distributed (data not shown). A plurality of incidents occurred at a home or residence (45.4%), and 29.2% overall were categorized as a possible injury. A minority of patients had abnormal vital signs.¹⁵ Table 2 also displays differences between the closest facility and other destination choice patients. Notably, closest facility patients were significantly older and had significantly shorter EMS transport times. Most vital signs were statistically, but not clinically, different between the two groups. Overall most patients did not receive any medications or procedures from EMS, save for oxygen (15.3% overall; 16.4% other vs 14% closest facility, p<0.0001).

Table 2:

Demographics, EMS Encounter, Vital Signs, and EMS-Administered Medications and Procedures for EMSTARS patients < 1 day and ≤ 18 years, 2011–2016

Variable	Overall N=446,274	Missing	Closest Facility N=208,323	Other^ N=237,951	p value
Patient Demographics
Gender – Male	217,545 (48.8%)	259 (<1%)	102,546 (47.1%)	114,999 (52.9%)	< 0.0001
Race		36,847 (8.3%)
White	194,674 (43.6%)		92,404 (44.3%)	102,270 (43.0%)	< 0.00011
Black or African American	143,090 (32%)		63,666 (30.6%)	79,424 (33.4%)
Other2	71,663 (16.1%)		34,698 (16.7%)	36,965 (15.5%)
Ethnicity – Not Hispanic or Latino3	291,462 (65.3%)	89,567 (20.1%)	135,699 (65.1%)	155,763 (65.5%)	0.024
Age in years, Median (IQR)	11 (3–16)	0	12 (4–16)	10 (3–16)	< 0.0001
Age < 5 years	133,966 (30.0%)	0	57,374 (27.5%)	76,592 (32.2%)	< 0.0001
Age < 10 years	199,299 (44.7%)	0	86,972 (41.7%)	112,327 (47.2%)	< 0.0001
Age < 15 years	260,921 (58.5%)	0	115,513 (55.4%)	145,408 (61.1%)	< 0.0001
EMS Encounter Characteristics
Incident at Home / Residence4	202,768 (45.4%)	21,712 (4.9%)	87,630 (42.1%)	115,138 (48.4%)	< 0.0001
EMS Response Mode to Scene: Lights and Sirens5	360,274 (79.0%)	0	165,119 (79.3%)	195,155 (82.0%)	< 0.0001
EMS Transport Mode from Scene: Lights and Sirens5	177,762 (39.8%)	0	82,492 (39.6%)	95,270 (40.0%)	0.0028
Time Intervals in Minutes6 Median (IQR)
EMS Arrival	5.8 (4 – 8.1)	803 (<1%)	5 (3.8 – 8)	6 (4 – 9)	< 0.0001
Scene	12 (8 – 16)	911 (<1%)	12 (8 – 16)	12 (8 – 16)	0.27
Transport	13 (8.3 – 20)	428 (<1%)	11.3 (7.8 – 17)	15 (9.5 – 23)	< 0.0001
ED Turnaround	19 (13 – 27)	1023 (<1%)	19 (12.7 – 26)	20 (13.7 – 28)	< 0.0001
Possible Injury – Yes	130,369 (29.2%)	259 (<1%)	62,042 (29.8%)	68,327 (28.7%)	< 0.0001
Vital Sign7
Heart Rate (Mean (SD))	109.5 (28.9)	4,353 (<1%)	107.8 (28.5)	110.9 (29.2)	< 0.0001
Tachycardia	138,600 (31.1%)		63,118 (30.3%)	75,482 (31.7%)	< 0.0001
Respiratory Rate (Mean (SD))	22.2 (9.9)	11,226 (2.5%)	21.9 (10.1)	22.5 (9.7)	< 0.0001
Tachypnea	75,415 (16.9%)		34,209 (16.4%)	41,206 (17.3%)	< 0.0001
Pulse Oximetry (Median (IQR))	99 (98–100)	41,553 (9.3%)	99 (98–100)	99 (98–100)	< 0.0001
Glasgow Coma Score (Median (IQR))	15 (15–15)	31,206 (7%)	15 (15–15)	15 (15–15)	< 0.0001
<15	59,712 (13.4%)		26,487 (12.7%)	33,225 (14.0%)	< 0.0001
<14	46,428 (10.4%)		20,133 (9.7%)	26,295 (11.1%)	< 0.0001
Systolic Blood Pressure (Mean (SD))	119.2 (21.8)	65,650 (14.7%)	119.8 (21.1)	118.7 (22.4)	< 0.0001
<90	16,086 (3.6%)		6,765 (3.2%)	9,321 (3.9%)	< 0.0001
<80	7,634 (1.7%)		3,098 (1.5%)	4,536 (1.9%)	< 0.0001
Diastolic Blood Pressure (Mean (SD))	74.5 (16)	77,543 (17.4%)	74.9 (15.4)	74.1 (16.4)	< 0.0001
<40	5,518 (1.2%)		2,142 (1.0%)	3,376 (1.4%)	< 0.0001
<30	4,042 (1.0%)		1,495 (<1%)	2,547 (1.1%)	< 0.0001
EMS-Administered Medications & Procedures8
Airway Procedure	4,945 (1.1%)	n/a	2,454 (1.2%)	2,491 (1.0%)	< 0.0001
Other Critical Procedure	5207 (1.2%)	n/a	2,760 (1.3%)	2,447 (1.0%)	< 0.0001
Contact Medical Control	831 (<1%)	n/a	58 (<1%)	773 (<1%)	< 0.0001
Resuscitation Medication	3846 (<1%)	n/a	2,242 (1.1%)	1,604 (<1%)	< 0.0001
Nebulized Medication	12,016 (2.7%)	n/a	5,353 (2.6%)	6,663 (2.8%)	< 0.0001
Benzodiazepine	2,445 (<1%)	n/a	1,208 (<1%)	1,237 (<1%)	0.0067
Oxygen	68,105 (15.3%)	n/a	29,175 (14.0%)	38,930 (16.4%)	< 0.0001
Fluids	15,605 (3.5%)	n/a	6,126 (2.9%)	9,479 (4.0%)	< 0.0001
Dextrose	636 (<1%)	n/a	297 (<1%)	339 (<1%)	0.99
Pain Medication	6,240 (1.4%)	n/a	2,582 (1.2%)	3,658 (1.5%)	< 0.0001

^{^}Denotes all other destination choices besides closest facility as per Table 1, includes ‘Not Known, Not Applicable, and Not Recorded’

¹Chi Square test performed as White vs all others

²Other includes American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, and Other Race

³Chi Square test performed as Not Hispanic or Latino vs Hispanic or Latino

⁴Other includes: Bathtub, Bucket, Farm, Forest/Trail/Wilderness, Health Care Facility, Industrial Place and Premises, Lake/River/Ocean, Mine or Quarry, Other, Place of Recreation or Sport, Pool, Public Building, Railroad, Residential Institution, Spa, Street or Highway, Trade or service

⁵Chi Square test done as lights and sirens versus all other modes of response/transport

⁶EMS Arrival = Time from EMS unit en route to scene arrival, Scene = Time from EMS scene arrival to EMS scene departure, Transport = Time from EMS scene departure to EMS emergency department (ED) arrival, Turnaround = Time from EMS ED arrival to EMS unit back in service, Total = Time from EMS notification by dispatch to EMS unit back in service

Negative time intervals were treated as errors and classified as missing

⁷Created specific for study using PALS 2019 as a reference

⁸See online Supplement e1 for list of included medications and procedures

Table 3 displays ‘reason for choosing destination’ by EMS provider’s primary impression. Most cardiac arrest patients’ (73.5%) were marked as closest facility. However, most patients with other critically ill provider primary impressions (e.g., sepsis, airway obstruction, hypovolemia/shock) had other than closest facility as their reason for choosing destination. Notably, 60% of respiratory distress patients and 44% of seizure patients’ reason for choosing destination was other, rather than closest facility.

Table 3:

Reason for Choosing Destination by Provider Primary Impression

Impression	Overall N=446,274	Closest N=208,323	Other N=237,951
Trauma / Injury
Traumatic Injury	122,702 (27.5%)	59,523 (28.6%)	63,179 (26.6%)
Stings / venomous bites	840 (<1%)	401 (<1%)	439 (<1%)
Electrocution	64 (<1%)	25 (<1%)	39 (<1%)
Respiratory
Respiratory Distress	34,715 (7.8%)	13,896 (6.7%)	20,819 (8.7%)
Neurologic
Seizure	42,785 (9.6%)	23,969 (11.5%)	18,816 (7.9%)
Behavioral/psychiatric disorder	14,269 (3.2%)	6,730 (3.2%)	7,539 (3.2%)
Stroke/Cerebrovascular accident	501 (<1%)	278 (<1%)	223 (<1%)
Altered level of consciousness	13,188 (3.0%)	6,373 (3.1%)	6,815 (2.9%)
Alcohol Related Problem / Delirium tremens	1,406 (<1%)	844 (<1%)	562 (<1%)
Intentional Drug Use / Related Problems	2,010 (<1%)	1,213 (<1%)	797 (<1%)
Endocrine
Diabetic Symptoms	2,186 (<1%)	969 (<1%)	1,217 (<1%)
Critically Ill
Cardiac Arrest	1,835 (<1%)	1,349 (<1%)	486 (<1%)
Cardiac Rhythm Disturbance	1,792 (<1%)	753 (<1%)	1,039 (<1%)
Airway Obstruction	2,478 (<1%)	1,080 (<1%)	1,398 (<1%)
Hypovolemia / shock	384 (<1%)	149 (<1%)	235 (<1%)
Obvious	122 (<1%)	92 (<1%)	30 (<1%)
Respiratory arrest	550 (<1%)	281 (<1%)	269 (<1%)
Sepsis	403 (<1%)	61 (<1%)	342 (<1%)
Other
Syncope / fainting	11,171 (2.5%)	5,335 (47.8%)	5,836 (52.2%)
Vaginal Hemorrhage	706 (<1%)	300 (42.5%)	406 (57.5%)
Pregnancy / Obstetric delivery	4,211 (1.0%)	1,342 (31.9%)	2,869 (68.1%)
Hyperthermia	4,071 (<1%)	1,968 (48.4%)	2,103 (51.6%)
Hypothermia	93 (<1%)	48 (51.6%)	45 (48.4%)
Chest pain	7,739 (1.7%)	3,334 (43.1%)	4,405 (56.9%)
Abdominal pain / problems	28,616 (6.4%)	14,607 (51.0%)	14,009 (49.0%)
Sexual assault / rape	351 (<1%)	171 (48.7%)	180 (51.3%)
Poisoning / drug ingestion	10,287 (2.3%)	5,007 (48.7%)	5,280 (51.3%)
Inhalation injury (toxic gas)	159 (<1%)	77 (48.4%)	82 (51.6%)
Smoke inhalation	227 (<1%)	55 (24.2%)	172 (75.8%)
Allergic reaction	10,075 (2.3%)	5,039 (50.0%)	5,036 (50.0%)
Blood pressure related problems	254 (<1%)	101 (39.8%)	153 (60.2%)
Fever Related Symptoms / Problems	10,798 (2.4%)	3,516 (32.6%)	7,282 (67.4%)
General Malaise	16,787 (3.7%)	7,511 (44.7%)	9,276 (55.3%)
Heart Related Illness	702 (<1%)	241 (34.3%)	461 (65.7%)
Other, Non Traumatic Pain	24,988 (5.6%)	16,423 (65.7%)	8,565 (34.3%)
Sickle Cell Crisis	148 (<1%)	49 (33.1%)	99 (66.9%)

Missing Provider Primary Impression = 72,661 (16.3%)

We were able to map patients to 65 of Florida’s 67 counties based on EMS agency home county data. The two counties without data were Hardee and Columbia, both rural. Most (93.2%, N=415,956) patient encounters were in urban counties and 6.2% (N=27,490) in rural counties (N=2,828, <1%, missing EMS agency county information). The maps in Figure 2 highlight rural counties, displaying locations of pediatric admitting hospitals and county-level percentages for the closest facility (Figure 2a), and patient / family choice ‘reason for choosing destinations’ (Figure 2b). Overall, fewer rural patients’ transport destination decision was closest facility compared to urban (43.9% versus 47%, p<0.0001). Correspondingly, more rural patients’ destination decision was patient / family choice than urban patients (36.3% versus 34.3%, p<0.0001). EMS median transport times were materially different between urban and rural locations (13 minutes urban versus 18 minutes rural, p< 0.0001, Online Supplement e2).

Figure 2.

a: Reason for Choosing Destination: Percentage Closest Facility Among EMSTARS Patients 0–18 Years of Age by County in Florida. b: Reason for Choosing Destination: Percentage Patient / Family Choice among EMSTARS Patients 0–18 Years of Age by County in Florida.

Discussion

This large, statewide study describes EMS’ reason for choosing destination facilities for pediatric patients. We found that just under half of patients were documented as closest facility, and a large percentage (35.7%) were designated as patient / family choice. Our results include both trauma and non-trauma patients, which may explain differences compared to a previous study of adult and pediatric trauma patients’ destinations.⁸ That study found that 50.6% of patients’ destinations were patient or family choice, with 20.7% being closest facility.⁸ However, there is geographic variation in those results,⁸ which we also found with more rural patients having destination decisions other than closest facility compared to urban, including patient / family choice. It is notable that our and others results from the United States contrast greatly with a study from Australia which found that 82% of rural pediatric prehospital patients were taken to the closest facility.¹⁸

Given widely operative pediatric trauma protocols, possible injury may also influence EMS’ destination choice. A previous analysis from three EMS agencies whose catchment areas included large children’s hospitals found that patients with injuries were more likely to be transported to higher-level pediatric resources than non-injury patients.¹⁹ We cannot compare our study directly to those results as our data does not include whether closest facility or patient / family choice led to transport to a higher-level pediatric resource or trauma center. However, compared to that study of three EMS agencies,¹⁹ this study has a much larger sample size from a state where 30 counties are rural and 37 are urban.¹³ Perhaps reflecting that more heterogeneous sample, we found most possible injury patients’ reasons for choosing destination were closest facility or patient / family choice (80%), and not protocol (e.g., trauma protocol) or specialty resource center. This may reflect areas where there is no tertiary care children’s hospital within a feasible EMS transport time. Such areas may be rural, suburban, or even urban, as not all cities have a tertiary care children’s hospital.

It appears that age may also influence destination decisions, with more marked effects for younger children. We found younger median ages in the other group compared to closest facility. Similarly, the study by Lerner et al found that patients above 15 years of age were more likely to go to general EDs than pediatric EDs.¹⁹ Further to this, as age decreased, increasing proportions of patients had destination decisions other than closest facility. At a certain teenage age range, EMS likely treats pediatric patients as they would adults and transports to the closest facility.

One of our most notable findings was that 60% of respiratory distress and 44% of seizure patients’ destination decisions were other than closest facility. Although we were unable to examine the free text narratives in the EMS records to see if respiratory distress patients were hemodynamically unstable or if seizure patients were actively seizing (both of which would prompt concern for needing to transport to the closest facility), this finding is notable for several reasons. First, both of those conditions / provider impressions have been previously identified as high-risk for secondary transport and interfacility transport.^4,20 This study provides additional evidence that a significant proportion of pediatric patients with respiratory distress and seizures are thought by EMS to need transport to specialized care. EMS’ “gut instinct” or prior experience may be that certain high-risk respiratory distress patients will require intensive care. Seizure patients’ families may express to EMS that they wish to be cared for in a center which knows their child’s epilepsy history, prompting transport to a specialty resource center (i.e., children’s hospital). What is not revealed by this study is EMS’ comfort and proficiency in caring for children in respiratory distress and with seizures over longer transport distances to get to those specialty centers. Additionally, only 3.5% of destination decisions overall were marked as protocol, which may reflect the need for more protocolized destination guidance for pediatric conditions other than trauma. Conversely, this may also reflect that more common destination choices selected by EMS providers (e.g., closest facility), are also specified by protocols.

Whether a pediatric patient goes to the closest facility or not affects EMS operations, particularly if the non-closest facility choice is much further away. Indeed, we found significantly longer transport times for other versus closest facility patients. Those longer transport times could stem from bypassing the closest facility to an alternative destination such as a pediatric specialty hospital. A study of non-trauma pediatric EMS bypass of closest facilities in Maryland also discovered a high (43%) rate of bypass of the closest facility and corresponding longer transport times and distances for bypass patients.²¹ We also found longer emergency department (ED) turnaround times for other versus closest facility patients, which may reflect patient complexity or acuity resulting in longer EMS-to-ED handoffs. Similarly, a study from a large EMS system in North Carolina found longer turnaround times for those transported to academic or tertiary care facilities versus community hospitals.²²

A combination of individual patient, caregiver, EMS, and geographic factors likely coalesce to inform the destination decisions for pediatric patients. Regardless, this study reveals a large proportion of other than closest facility destination decisions and significant differences between urban and rural counties. Therefore, it seems appropriate for future work to combine clinical and geospatial analysis from multiple studies and locations to produce evidence-based non-trauma pediatric transport guidelines.

Limitations

This study has limitations that merit consideration. Overall, interpretation of closest facility versus other should be tempered by our lack of precise coordinates for EMS scene and destination locations. Therefore, we are unable to say if closest facility patients went to the actual closest facility, and if other patients bypassed a closer facility in favor of an alternative destination. Patient / family choice may also be the geographically closest facility, specialty resource center may not be a children’s hospital, etc. Therefore, the reader should interpret the variable as the reason for choosing destination, and not the actual destination itself. Additionally, since ‘reason for choosing destination’ choices in EMSTARS are mutually exclusive, we are unable to discern if EMS providers’ choice was multifactorial. Since we did not have EMS scene and hospital locations, we were unable to determine if any transports were bypasses of the closest facility. Online medical direction is a value for ‘reason for choosing destination’, however, the EMSTARS data dictionary does not specify if online medical direction was attempted, provided, and/or whether the EMS providers transported to the advised destination.¹⁰

Regarding the 29% of patients noted by EMS to have a possible injury, there are widely operative guidelines for trauma patient destination nationwide and in Florida. However, only 3.2% of patients’ reason for choosing destination in this study was protocol, and only 5.4% was specialty resource center. A limitation of this study is that the ‘possible injury’ variable is a poor surrogate for a trauma activation patient, and our IRB-approved study design did not include a specific trauma patient analysis, therefore we were precluded from a pediatric trauma sub-group analysis.

EMSTARS, while containing a wealth of data, is a database including only Florida; as such, our results may not be generalizable to all areas. However, Florida is a very populous state with a mix of urban and rural counties, therefore our results (including geospatial analysis) and large sample size can provide more generalizable information than single-agency studies from urban locations. However, Florida DOH’s method of classifying urban and rural is at the unit of a county, which limits identification of areas in urban counties that may actually be rural or vice versa. Since our dataset was limited to county-level geographic variables, we were unable to use more spatially resolute descriptions of urban and rural areas. EMSTARS does not contain information from all EMS agencies in Florida, which is reflected in our geospatial analyses, and may impact how certain county-level variables appear on the maps presented. As with other large EMS databases, there are variables which contain large amounts of missing data. For our univariate comparisons between closest facility and other patients, we did not include patients with missing values for those data (e.g., for comparing transport time, those whose transport times were not recorded were excluded for only the transport time comparison). Some variables have higher degrees of missing data (e.g., ethnicity), which may bias the results of univariate comparisons for those variables. Additionally, for some counties, there was an unusual breakdown of gender amongst transported patients. Specifically, an overwhelming majority of transported patients were female (Collier 92.8%, Lee 78.3%, Hernando 78%, and Seminole 75%). We did not have access to the free text narrative in the EMS patient care reports, which would provide additional information on provider’s reason for destination choice.

Conclusions

In this large statewide study of EMS’ reason for choosing destination for pediatric patients, nearly one-half of patient encounters were documented as closest facility, and over one-third as patient / family choice. Significant differences in destination decision-making were noted for rural versus urban counties. Additionally, in this study of nearly a half-million pediatric EMS transports, there was a high proportion of patient / family choice, and specific conditions with high proportions of reason for choosing destination other than closest facility. Those results should be noted within the context of regionalization of pediatric care into specialized urban centers. These results can therefore aid EMS agencies in operations planning and those currently developing pediatric EMS destination guidelines.