Abstract
Objective:
The epidemiology accompanying helicopter emergency medical services (HEMS) transport has evolved as agencies have matured and become integrated into regionalized health systems, as evidenced primarily by nationwide systems in Europe. System-level congruence between Europe and the United States, where HEMS is geographically fragmentary, is unclear. In this study, we provide a temporal, epidemiologic characterization of the largest standardized private, nonprofit HEMS system in the United States, STAT MedEvac.
Methods:
We obtained comprehensive timing, procedure, and vital signs data from STAT MedEvac prehospital electronic patient care records for all adult patients transported to UPMC Health System hospitals in the period of January 2012 through October 2021. We linked these data with hospital electronic health records available through June 2018 to establish length of stay and vital status at discharge.
Results:
We studied 90,960 transports and matched 62.8% (n = 57,128) to the electronic health record. The average patient age was 58.6 years (19 years), and most were male (57.9%). The majority of cases were interfacility transports (77.6%), and the most common general medical category was nontrauma (72.7%). Sixty-one percent of all patients received a prehospital intervention. Overall, hospital mortality was 15%, and the average hospital length of stay (LOS) was 8.8 days (10.0 days). Observed trends over time included increases in nontrauma transports, level of severity, and in-hospital mortality. In multivariable models, case severity and medical category correlated with the outcomes of mortality and LOS.
Conclusion:
In the largest standardized nonprofit HEMS system in the United States, patient mortality and hospital LOS increased over time, whereas the proportion of trauma patients and scene runs decreased.
Helicopter emergency medical services (HEMS) systems are responsible for approximately 1% of emergency medical services (EMS) transports annually in the United States, leveraging speed and access to specialty care not possible in ground-based systems.1 Unimpeded by the traffic and terrain, a typical HEMS helicopter can cover at least 3 times the distance of a ground ambulance in the same amount of time. It can project centralized, advanced resources like extracorporeal membrane oxygenation far outside the ground capture area of many tertiary care receiving hospitals.2,3 HEMS can aid regionalization by facilitating rapid interfacility transfer between outlying community hospitals and centralized specialty facilities.4–6
Except in rare circumstances in which geography necessitates HEMS for all emergent transports, the generally high threshold for the use of HEMS and relatively stringent HEMS triage rules tend to concentrate the critically ill among their patients. Moreover, in many cases, the high acuity necessitated by the composition of these cohorts leaves little room for error, resulting in sustained interest in quality improvement among HEMS systems.2,7,8 Taken together, these features create opportunities among such systems for investigating potentially high mortality/morbidity conditions, including major trauma, sepsis, acute respiratory failure, stroke, myocardial infarction, and cardiac arrest.9–15 HEMS trauma care also serves as a valuable research analog for military casualty evacuation (medevac), both sharing a common transport modality and a common philosophy of rapid transport from low-skill, low-resourced care in forward battlefield positions to definitive care accompanied by advanced air medical care providers.16,17
Much of what is known about HEMS at a systems level comes from national HEMS in Europe. We sought to provide a temporal, epidemiologic characterization of the largest standardized private, non-profit HEMS system in the United States, STAT MedEvac, to describe trends over time. We hypothesized that patient acuity, illness, injury, and outcomes would vary over time reflective of process-specific and local secular events.
Methods
This study was conducted with approval from the University of Pittsburgh Institutional Review Board (19020113 & 20050368).
Data Acquisition and Processing
Prehospital patient data were obtained from the STAT MedEvac air medical service (Pittsburgh, PA). STAT MedEvac currently has 18 bases (increasing from 17 to 18 in 2019) located in 4 states and the District of Columbia and employs 275 crewmembers, serving a catchment area of 37,000 square miles and a population of approximately 8.7 million people. The service maintains capabilities for ground transportation in the event of inclement weather using the same personnel and protocols.
Cases were included for the period of January 1, 2012, through October 6, 2021, and included all qualifying air or ground transports. We excluded patients younger than 18 years of age, cases in which no patient was transported, and cases in which the receiving hospital was missing. The service uses a single electronic health record (EHR) system, EMSCharts (ZOLL Medical, Chelmsford, MA), which collects National Emergency Medical Services Information System–compliant data and supplementary data.18 EHR data were exported from the EMSCharts SQL backend system and accessed through Stata (version 17; StataCorp LLC, College Station, TX). The prehospital EHR included time-stamped intervention data fields for event timing, medications, procedures, vital signs, and titration/adjustment of ongoing interventions.
The prehospital EHR was linked to the hospital EHR for patients delivered to UPMC Health System tertiary/quaternary care hospitals in Western Pennsylvania. Hospital record linkage was achieved through direct multifield matching of key patient-identifying variables, including name, date of birth, sex, and date/time of hospital arrival; hospital length of stay (LOS); and vital status at discharge. Importantly, linkage was limited to only cases delivered to hospitals in the health system in the period through the end of June 2018 because of constraints of the available data systems at the time of analysis. All other cases were only analyzable in the prehospital phase.
To categorize the seriousness of cases, we determined the Revised Trauma Score (RTS), which incorporates the Glascow Coma Scale, systolic blood pressure, and respiratory rate into a formula to produce a scale with an absolute range of 0 to 7.8408.19 Previous studies have demonstrated that RTS is predictive of mortality with moderate accuracy when measured at the scene or upon admission among injured patients.20–22 We calculated both the minimum and maximum observed RTS from all available, complete vital sign data across each case.
To describe the breadth and complexity of patient procedural needs, we classified interventions performed by the flight crew into several broad categories for tabulation: cardiac, airway, thoracic, cardiac assist devices, invasive vascular access, vasoactive medications, crystalloids, blood products, and bleed control. Among these, a core group of procedures was also subclassified as lifesaving interventions (LSIs) in line with the existing field norms.
Analysis
We summarized case characteristics overall and by year. We also summarized key interventions as frequencies and the proportion of patients receiving them. The outcomes of hospital mortality and LOS were summarized as frequency and percent or mean (± standard deviation), respectively, overall and stratified by general medical category (trauma/nontrauma) and higher specificity subcategories; t-tests, analysis of variance, and chi-square tests or nonparametric equivalents as appropriate were used to compare characteristics by category and over time.
We constructed multivariable regression models incorporating an a priori set of predictors likely to correlate with mortality and morbidity. These included age, sex, EMS time to arrival, time of day, season, general medical category, medical subcategory, need for prehospital LSI, and severity summarized by RTS. Multivariable logistic regression models were constructed for the outcome of hospital mortality, and multivariable linear regression models were constructed for the outcome of LOS. Results of the logistic regression models were reported as odds ratios (ORs) with 95% confidence intervals (CIs), whereas results of the linear regression model were reported as linear coefficients with 95% CIs. For all statistical analyses, an alpha level of 0.05 was used.
Results
Demographics
The consort diagram for this study is shown in Figure 1. After exclusions, a total of 90,960 cases were included for characterization at the prehospital level. Between the years 2012 and 2020 (the last complete year in the capture period), the total volume of cases transported by STAT MedEvac increased by approximately 20%, and the mean (standard deviation) transport distance was 43.1 (33.7) miles. The physical footprint for all transports is shown in Figure 2, with ranges constrained to the mid-Atlantic region of the United States, wherein lie the complete paths of more than 95% of STAT MedEvac transports. Most transports (70%) had a destination at 1 of the 5 core hospitals in the UPMC Health System.
Figure 1.
Consort Diagram. Patient flow is shown as a fraction of overall dispatches of the STAT MedEvac service during the capture period. Cases were excluded in two tiers, reflecting excludable prehospital and in-hospital characteristics.
Figure 2.
Map of the STAT MedEvac Service Footprint. The spatial density map shows in tones from dark/blue to light/yellow the relative density of transport event referring locations for the capture period.
The general demographic characteristics of the cohort are presented in Table 1. STAT MedEvac executed both direct scene run and interfacility transports serving over 480 hospitals and 880 EMS agencies, with scene runs making up 22.4% of all transports. Of the interfacility transports, 50% came from 20 hospitals in Western Pennsylvania; the remainder came from more than 250 additional facilities. Additionally, 9.7% of transports were executed by ground units. Ground units, in turn, were infrequently used for scene transports (2.7%) and were used for those interfacility runs in which flight was not available (23.2%). The overall caseload was lowest in winter, representing 21.2% of annual cases compared with a summer peak of 28.2%. Although the crewmembers work 12- and 24-hour shifts, when the day was divided into a 3-shift schedule for analysis, the highest burden was seen during shift 2 between 3 pm and 11 pm.
Table 1.
Demographics of the STAT MedEvac Cohort
| All Patients N = 90,960 | Hospital Linked N = 57,128 | |
|---|---|---|
|
| ||
| Sex | ||
| Male | 51,898 (57.9) | 33,303 (58.7) |
| Female | 37,804 (42.1) | 23,460 (41.3) |
| Age, y | 58.6 (19.0) | 59.8 (18.5) |
| Weight group, kg | ||
| Mean (SD) | 87.4 (27.0) | 87.0 (26.3) |
| 0–79 | 36,538 (40.4) | 23,222 (40.7) |
| 80–199 | 53,420 (59.0) | 33,596 (58.8) |
| 200–500 | 533 (0.6) | 303 (0.5) |
| Type of service | ||
| Interfacility | 70,246 (77.6) | 45,077 (78.9) |
| Scene | 20,257 (22.4) | 12,051 (21.1) |
| Transport mode | ||
| Ground | 8,733 (9.7) | 5,537 (9.7) |
| Medevac | 81,771 (90.4) | 51,591 (90.3) |
| Medical category | ||
| Trauma | ||
| Adult trauma (motor vehicle) | 9,236 (10.2) | 5,625 (9.9) |
| Adult trauma (other) | 12,672 (14.0) | 8,736 (15.3) |
| Adult trauma (recreational vehicle) | 1,677 (1.9) | 1,022 (1.8) |
| Burn | 982 (1.1) | 579 (1.0) |
| Nontrauma | ||
| Cardiac | 13,230 (14.6) | 6,987 (12.2) |
| Medical | 29,372 (32.5) | 18,251 (32.0) |
| Neurologic | 17,937 (19.9) | 13,211 (23.1) |
| Obstetric | 1,667 (1.8) | 36 (0.1) |
| Surgical | 2,358 (2.6) | 1,667 (2.9) |
| Toxicology | 1,305 (1.4) | 1,011 (1.8) |
| Time of day | ||
| Morning | 26,872 (30.0) | 16,675 (29.6) |
| Afternoon | 41,597 (46.4) | 26,353 (46.7) |
| Midnight | 21,236 (23.7) | 13,390 (23.7) |
| Seasons | ||
| Winter | 19,259 (21.2) | 12,096 (21.2) |
| Spring | 23,250 (25.6) | 14,695 (25.7) |
| Summer | 25,680 (28.2) | 15,981 (28.0) |
| Fall | 22,771 (25.0) | 14,356 (25.1) |
| Times | ||
| Transport time | 32.9 (23.9) | 34.4 (25.0) |
| Response time | 25.7 (20.8) | 25.3 (21.4) |
| Scene time | 27.5 (18.4) | 27.2 (18.1) |
| Vital Signs | ||
| Systolic blood pressure, mm Hg | 130 (27) | 131 (27) |
| Heart rate, beats/min | 90 (20) | 90 (20) |
| Respiratory rate, breaths/min | 19 (5) | 19 (5) |
| SpO2, % | 97 (4) | 97 (4) |
| Minimum RTS, median (IQR) | 7.6 (5.7–7.8) | 7.6 (5.4–7.8) |
RTS = Revised Trauma Score; SD = standard deviation; SpO2 = oxygen saturation.
The age distribution of adult cases is shown in Figure 3 stratified by sex. It showed a peak between 58 and 77 years old. The age distribution within this cohort was relatively stable during the capture period, increasing from an average of 55.9 years in 2012 to 60.2 years in 2021. Overall, females made up 42.1% of the cohort.
Figure 3.
Sex by Age Category. Case counts are shown stratified by age and sex for all cases not meeting prehospital exclusion criteria.
Medical Category and Severity
Grossly classified, 27.2% of incidents were categorized as trauma compared with 72.8% classified as nontrauma. Moreover, trauma was much more common among males than females (32.1 vs. 20.2%, P < .001). The distribution of specific medical subcategories over time is shown in Figure 4. The overall proportion of trauma transports fell, accompanied by a fall in adult trauma (motor vehicle). The catchall medical category “medical” was the most common category, comprising 32.5% of all cases and 44.6% of nontrauma cases. See Supplemental Figure 1 for details on this category.
Figure 4.
Medical Category by Year. Case counts of paramedic assigned medical categories are shown for each year of the study period.
Vital signs and RTS summary statistics for all transports are shown in Table 1. The overall cohort showed a mild level of severity as measured by RTS. Stratified by medical subcategory, the minimum RTS was lowest among the categories “medical” (OR = 7.1; 95% CI, 4.1–7.8) and “toxicology” (OR = 4.1; 95% CI, 4.1–6.9) and highest among “obstetric” (OR = 7.8; 95% CI, 7.8–7.8) cases. RTS differed marginally between interfacility and scene runs (7.5 vs. 7.8, P < .001).
Interventions
Common procedures performed during care are tabulated in Table 2. At least 1 procedure was administered to 61.1% (n = 55,567) of patients. STAT MedEvac crew intubated 4,588 (5.0%) patients, equating to approximately 1.3 intubations per day. Additionally, 7,867 (8.7%) patients were transported with a central venous catheter in place, and 14,568 (16.0%) patients received some vasopressor support. The elapsed time to LSIs from arrival at the patient retrieval site is shown in Table 2. The median (interquartile range [IQR]) time from arrival at the patient’s location to arrival at the receiving hospital was 51 minutes (IQR: 38–72 minutes), and the overall median elapsed time to key LSIs was 8 minutes (IQR: 2–15 minutes).
Table 2.
Select Procedures and Time to Lifesaving Intervention During Prehospital Care
| Trauma |
Nontrauma |
Overall |
||||
|---|---|---|---|---|---|---|
| Patient Count (%) | Minutes, Median (IQR) | Patient Count (%) | Minutes, Median (IQR) | Patient Count (%) | Minutes, Median (IQR) | |
|
| ||||||
| Cardiac | 378 (1.5) | 7.5 (0–23) | 1,732 (2.6) | 18 (5–35) | 2,101 (2.3) | 15 (4–32) |
| CPR | 326 (1.3) | 7 (0–20) | 663 (1) | 14 (2–39) | 989 (1.1) | 11 (0–30.667) |
| Shock | 35 (0.1) | 22 (11–35) | 276 (0.4) | 24 (9–46) | 311 (0.3) | 23.5 (9–43.75) |
| Pacing | 23 (0.1) | 17 (5–29) | 606 (0.9) | 15 (7–31) | 629 (0.7) | 15 (7–30) |
| Cooling | 36 (0.1) | 14.5 (4–32.5) | 479 (0.7) | 25 (14–40) | 515 (0.6) | 25 (13–40) |
| Airway | 4,928 (19.8) | 11 (5–19) | 23,394 (35.4) | 21 (13–31) | 28,322 (31.1) | 20 (11–30) |
| BVM | 2,146 (8.6) | 5 (0–12) | 2,593 (3.9) | 10 (0–28) | 4,739 (5.2) | 7 (0–19) |
| OPNP | 583 (2.4) | 6 (1–11) | 503 (0.8) | 11 (3.02–24) | 1,086 (1.2) | 8 (2–15) |
| SGA | 97 (0.4) | 13 (8–19) | 42 (0.1) | 23 (17–38) | 139 (0.2) | 16 (9–23) |
| Cricothyrotomy | 7 (0) | 17 (14–17.0833) | 6 (0) | 21 (18–55) | 13 (0) | 17.08 (15–21) |
| ETI | 2,508 (10.1) | 10 (7–14) | 2,080 (3.2) | 15 (10–25) | 4,588 (5) | 11 (8–18) |
| NIPPV | 19 (0.1) | 16.5 (13–25) | 2,601 (3.9) | 17 (11–25) | 2,620 (2.9) | 17 (11–25) |
| Ventilator | 4,482 (18) | 22 (16–29) | 20,706 (31.3) | 24 (16–34) | 25,188 (27.7) | 24 (16–33) |
| Thoracic | 957 (3.9) | 8 (2–18) | 663 (1) | 11 (4–32) | 1,620 (1.8) | 9 (2–22) |
| Chest tube | 517 (2.1) | — | 654 (1) | — | 1,171 (1.3) | — |
| Needle decompression | 452 (1.8) | 10 (3–20) | 20 (0) | 36.5 (20–81) | 472 (0.5) | 10 (4–21) |
| Pericardiocentesis | 0 (0) | NA (NA) | 0 (0) | NA (NA) | 0 (0) | NA (NA) |
| Escharotomy | 3 (0) | 29 (10–29) | 0 (0) | NA (NA) | 3 (0) | 29 (10–29) |
| Cardiac assist devices | 15 (0.1) | — | 1,716 (2.6) | — | 1,731 (1.9) | — |
| ECMO | 3 (0) | — | 258 (0.4) | — | 261 (0.3) | — |
| IABP | 2 (0) | — | 1,164 (1.8) | — | 1,166 (1.3) | — |
| VAD | 10 (0) | — | 285 (0.4) | — | 295 (0.3) | — |
| Impella (Abiomed, Inc., Danvers, MA) | 0 (0) | — | 141 (0.2) | — | 141 (0.2) | — |
| Invasive vascular access | 433 (1.7) | — | 9,166 (13.9) | — | 9,599 (10.6) | — |
| Aterial line | 175 (0.7) | — | 3,489 (5.3) | — | 3,664 (4) | — |
| Central line | 351 (1.4) | — | 7,516 (11.4) | — | 7,867 (8.7) | — |
| PA line | 5 (0) | — | 445 (0.7) | — | 450 (0.5) | — |
| Vasoactive medications | 1,013 (4.1) | 13.02 (5–25) | 13,555 (20.5) | 13 (6–22) | 14,568 (16) | 13 (6–23) |
| Epinephrine | 692 (2.8) | 11 (4–22) | 3,071 (4.6) | 18 (6.5–36) | 3,763 (4.1) | 16 (6–34) |
| Norepinephrine | 499 (2) | 20 (11–35) | 11,544 (17.5) | 14 (7–25) | 12,043 (13.2) | 14 (7–25) |
| Dopamine | 40 (0.2) | 16 (9–25) | 1,582 (2.4) | 15.25 (10–25) | 1,622 (1.8) | 15.5 (10–25) |
| Dobutamine | 3 (0) | 12 (7–20) | 473 (0.7) | 18 (11–28) | 476 (0.5) | 18 (11–28) |
| Phenylephrine | 19 (0.1) | 15.5 (10.08–18) | 986 (1.5) | 19 (12–28) | 1,005 (1.1) | 19 (12–28) |
| Vasopressin | 49 (0.2) | 31.5 (24–40.5) | 2,180 (3.3) | 28 (19–41) | 2,229 (2.5) | 28 (19–41) |
| Crystalloids | 8,855 (35.7) | 8 (4–15) | 22,847 (34.6) | 11 (5–21) | 31,704 (34.9) | 10 (5–19) |
| NS | 8,688 (35) | 8 (4–15) | 21,761 (32.9) | 11 (5–21) | 30,451 (33.5) | 10 (5–19) |
| LR | 184 (0.7) | 13 (6–23) | 1,140 (1.72) | 10 (5–18) | 1,324 (1.5) | 10 (5–19) |
| Blood products | 1,576 (6.3) | 18 (10–28) | 3,594 (5.44) | 15 (7–28) | 5,171 (5.7) | 15 (8–28) |
| RBC | 1,376 (5.5) | 18 (10–30) | 3,098 (4.69) | 15 (8–29) | 4,475 (4.9) | 16 (8–29) |
| FFP | 197 (0.8) | 20 (11–29) | 616 (0.93) | 18 (9–31) | 813 (0.9) | 19 (10–30) |
| Platelets | 42 (0.2) | 13.5 (9–26) | 130 (0.2) | 14 (8–25) | 172 (0.2) | 14 (9–25) |
| WB | 23 (0.1) | 20 (7–28) | 8 (0.01) | 31 (14–54) | 31 (0) | 20 (8–32) |
| Cryoprecipitate | 2 (0) | 6.5 (2–11) | 21 (0.03) | 20 (15–45) | 23 (0) | 20 (15–45) |
| Bleed control | 2,029 (8.2) | 6 (2–15) | ||||
| Bleed control | 806 (3.2) | 4 (1–10) | — | — | — | — |
| Tourniquet | 314 (1.3) | 0 (0–5) | — | — | — | — |
| Pelvic | 591 (2.4) | 6 (4–10) | ||||
| TXA | 581 (2.3) | 17 (12.75–25) | — | — | — | — |
| Any above | 12,476 (50.2) | 7 (3–13) | 43,885 (66.37) | 8 (1–16) | 55,567 (61.1) | 8 (2–15) |
BVM = bag valve mask; CPR = cardiopulmonary resuscitation; ECMO = extracorporeal membrane oxygenation; ETI = endotracheal intubation; FFP = fresh frozen plasma; IQR = interquartile range; LR = lactated Ringer’s; NA = not applicable; NIPPV = noninvasive positive pressure ventilation; NS = normal saline; OPNP = Oropharyngeal/Nasopharyn-geal; PA = pulmonary artery; RBC = red blood cell; SGA = supraglottic airway; TXA = tranexamic acid; WB = white blood; VAD = ventricular assist device.
Outcomes
Hospital linkage with survival outcome was available for 57,128 cases that were transported to UPMC hospitals, and outcomes are summarized in Figure 5. Among those patients with hospital EHR linkage, 15% died before hospital discharge. Over the capture period, the overall mortality increased from 12.3% to 17.4% (P < .001). Death before hospital discharge was more common among nontrauma cases than those with trauma (17.5 vs. 6.5%, P < .001) and was most common among the “cardiac” (22.9%) and “medical” (20.0%) subcategories and least common among “adult trauma–recreational vehicle” (2.1%) and “obstetric” (2.9%) subcategories. In unadjusted analyses, scene run cases had a lower in-hospital death rate than interfacility transports (6.9 vs. 16.4%, P < .001). Mortality was also lower among men than women (13.7 vs. 15.1%; P < .001). Among those patients who required an LSI, mortality was 4-fold higher than those who did not receive LSIs (20.7 vs. 5.1%, P < .001). RTS was concentrated in a markedly lower range among those who died in hospital (IQR: 3.36–4.09) than those who were discharged alive (IQR: 4.09–7.55) (P < .001).
Figure 5.
Hospital Outcomes by Year and Medical Category. Mortality (%) and average length of stay are shown on the same scale for each medical category of each year of the study period. It is noteworthy that obstetric cases received by the included hospitals (trauma centers) reflect both low overall counts and distinctly high severity.
Among those who did not die in hospital, the median (IQR) LOS was 5 days (IQR: 2–10 days), increasing over the capture period from a median of 5 to 6 days (P < .001). This was higher than the median LOS of the baseline general hospital population (median = 4 days [IQR: 2–7 days]). LOS was higher among nontrauma cases (9.4 vs. 7.6; P < .001), interfacility transports (9.4 vs. 7.2, P < .001), and those who needed prehospital LSIs (12.6 vs. 8.0, P < .001). LOS was highest among the “burn,” “surgical,” and “medical” subcategories; lowest for “obstetric”; and differed significantly across all categories (P < .001). For those discharged alive, the most common destinations were home (59.7%), a skilled nursing facility (17.4%), and a rehabilitation facility (16.0%).
The results of multivariable regression models are shown in Table 3. Hospital mortality and LOS were both independently associated with the need for prehospital LSIs, RTS, scene run, “medical” subcategory, season, and time of day after full model adjustment. The strongest independent predictor of mortality was receipt of a prehospital LSI (OR: 1.74, 95% CI, 1.61–1.88). Post hoc receiver operating curve analysis of the logistic model yielded an area under the curve of 0.81. An LSI was also a significant contributor to LOS in the linear regression model (coefficient: 0.94; 95% CI, 0.74–1.15). By comparison, several medical subcategories had larger effect sizes, with “cardiac” having the highest direct impact on mortality and “burn” having the highest direct impact on LOS.
Table 3.
Multivariable Regression Models for Mortality and Length of Stay (LOS)
| Mortality |
LOS |
|||||
|---|---|---|---|---|---|---|
| Odds Ratio | P Value | 95% CI | Coefficient | P Value | 95% CI | |
|
| ||||||
| Age | 1.03 | <.001 | 1.03–1.03 | 0 | .783 | 0–0.01 |
| Sex | 1.08 | .004 | 1.03–1.14 | 0.25 | .008 | 0.06–0.43 |
| Air transport | 0.97 | .404 | 0.89–1.05 | −0.77 | <.001 | −1.09 to −0.46 |
| Scene | 0.8 | <.001 | 0.73–0.88 | −1.02 | <.001 | −1.28 to −0.77 |
| LSI | 1.74 | <.001 | 1.61–1.88 | 0.94 | <.001 | 0.74–1.15 |
| RTS | 0.61 | <.001 | 0.6–0.62 | −1.48 | <.001 | −1.55 to −1.41 |
| Time of day | ||||||
| Afternoon | 0.92 | .005 | 0.86–0.97 | −0.05 | .63 | −0.26 to 0.16 |
| Midnight | 0.89 | .001 | 0.82–0.96 | −1.16 | <.001 | −1.41 to −0.92 |
| Season | ||||||
| Spring | 0.94 | .095 | 0.87–1.01 | −0.37 | .005 | −0.63 to −0.11 |
| Summer | 0.97 | .406 | 0.9–1.05 | −0.3 | .022 | −0.56 to −0.04 |
| Fall | 0.99 | .831 | 0.92–01.07 | −0.14 | .312 | −0.4 to 0.13 |
| Medical category | ||||||
| ATO | 1.36 | <.001 | 1.15–1.6 | −1.57 | <.001 | −1.93 to −1.21 |
| ATRV | 0.65 | .063 | 0.41–1.02 | −2.03 | <.001 | −2.72 to −1.34 |
| Burn | 0.92 | .642 | 0.65–1.3 | 3.67 | <.001 | 2.76–4.58 |
| Cardiac | 2.88 | <.001 | 2.45–3.38 | −0.67 | .002 | −1.09 to −0.25 |
| Medical | 2.25 | <.001 | 1.93–2.62 | 1.34 | <.001 | 0.97–1.71 |
| Neurologic | 1.65 | <.001 | 1.41–1.93 | −1.75 | <.001 | −2.12 to −1.38 |
| Obstetric | 1.47 | .717 | 0.18–11.65 | −4.4 | .013 | −7.86 to −0.93 |
| Surgical | 1.88 | <.001 | 1.52–2.33 | 2.37 | <.001 | 1.77–2.98 |
| Toxicology | 0.59 | .001 | 0.44–0.8 | −6.68 | <.001 | −7.4 to −5.97 |
Pseudo R2 = 0.204 Adjusted R2 = 0.106
ATO = adult trauma other; ATRV = adult trauma recreational vehicle; CI = confidence interval; LSI = lifesaving intervention; RTS = Revised Trauma Score.
Discussion
Our study provides the first system-level description of the characteristics of the largest standardized, nonprofit, air medical service in the United States and its patients over a nearly decade-long period. The scope of this database is prohibitive for deeply granular investigation in a single study; however, we believe that this general summary is not only useful as a benchmark for the understanding of HEMS and transport of the critically ill and injured but also creates a foundation for identifying and addressing a diverse array of future research questions at a system-level scale.
Purely descriptive epidemiologic characterizations of individual US HEMS systems are rare. A general query of the National Emergency Medical Services Information System registry for incidents in which patients were transported by helicopter (EMS Data Cube v3 [National Emergency Medical Services Information System, Salt Lake City, UT]; 2017–2022, N = 1,091,993) indicates that age (17.9% 61–70 years old), sex (58.7% male), and trauma (31.6%) were moderately comparable, but interfacility transport (61.2%) was much less common nationally. Considering only trauma patients, Michaels et al23 separately found lower age but a similar proportion of males, mortality, and LOS among more than 44,000 patients surveilled by the National Trauma Data Bank in 2014. Standardized national systems in Northern Europe may provide more direct comparisons. Denmark, excluding Greenland, has a land area (16,600 square miles) roughly half the size of the STAT MedEvac coverage footprint, a population two thirds as large (5.8 million), and a national HEMS system with 4 bases strategically located to provide coverage to the entire country. Sørensen et al24 summarized nearly 4 years of data from the Danish national HEMS system comprising more than 5,000 missions, excluding interfacility transports. Twenty-three percent of patients in the Danish study were classified as trauma, far fewer than the 73% of scene runs attended by STAT MedEvac, and patients were, on average, approximately 7 years older. The same study reported 30-day mortality rates between 7% (trauma) and 19% (nontrauma) compared with roughly 4% and 12% in our study. Finland, although much larger than Denmark, has a similar population size. Björkman et al25 reported similar Finnish demographic characteristics and comparable mortality among patients encountered by the national HEMS service. A core tenant of many European systems is to bring the doctor to the patient, allowing for triage on scene to other modes of transport, and, in some cases, including in Finland, patients treated by HEMS may not be transported by HEMS, complicating the comparison. Depending on the actual ratio of ground to flight transport, this may be a convenient comparison to STAT MedEvac, given the mixed modalities of this service.
We additionally found that mortality increased over time in this system, partly driven by increasing interfacility transport of critically ill nontrauma patients. For comparison, Saviluoto et al26 reported trends in the Finnish system over an overlapping 7-year period, demonstrating relative stability in case medical category distribution and mortality. Some of the influences on the characteristics and trends seen in STAT MedEvac may not be generalizable. These include the phenomenon of regionalization and patterns of patient movement driven by the parent health system. For instance, the decision to transport trauma patients by HEMS might be made based on an increasing threshold of severity as regional centers are made more robust and the most complex patients are secondarily transferred. It is also possible that the increased mortality is a consequence of regional consolidation occurring over a wide area without concomitant increases in helicopters. This may partly explain the differences seen between the Finnish system and STAT MedEvac.
Prehospital factors have long been associated with in-hospital mortality and LOS. Studies, including those of our own group, have demonstrated potential for interventions to impact these outcomes.27 The association with prehospital LSIs and mortality may be both a consequence of injury severity and an opportunity to identify early interventions that reduce mortality and morbidity across patient types. Associations of scene and interfacility transports with mortality beg for a comparison of direct transfer models for trauma, stroke and ST-segment elevation myocardial infarction and hub-and-spoke models for critical medical issues such as sepsis. Differences in secular trends such as seasonality and time of day may be resource related and reflect the availability of hospital services at off-peak hours or increased out-of-hospital times during inclement weather.28,29
Importantly, we found the severity of cases using the RTS equally predictive of mortality for both trauma and medical cases, which may be a novel generalization of this score. Comparison to other HEMS systems is difficult because severity among comparable European studies is often expressed in terms of the National Advisory Committee for Aeronautics score.30 Even so, the similarity in outcomes suggests that the severity levels were comparable.
Our study has several limitations. First, we describe a single HEMS agency that is largely focused around a single regional health system, limiting external generalizability. Second, although we do provide some temporal stratifications of our results, there are likely important unaddressed secular influences, including changes in guidelines/protocols, patterns in major public health phenomena (eg, the coronavirus disease 2019 pandemic, the opioid epidemic, and long-term trends in motor vehicle accidents), and insurance reimbursement. Third, we were only able to report outcomes for patients transported to certain hospitals. Although this results in large outcome incompleteness, we have no reason to believe that the destination hospital would impact the quality of care by the flight crews. Lastly, as with any record review study, the results are only as good as the accuracy of the source data. Measures that meet or exceed industry best practices are in place within the STAT MedEvac agency and in our research data management team to mitigate recall and reporting biases.
Conclusions
In the largest standardized nonprofit HEMS system in the United States, patient mortality and hospital LOS increased over time, whereas the proportion of trauma patients and scene runs decreased. Our cohort has a comparatively strong focus on interfacility transport predicated by the HEMS function within an integrated health care delivery system.
Supplementary Material
Acknowledgments
The present study received support from a National Heart, Lung, and Blood Institute grant (5R01HL141916) awarded to Dr. Pinsky. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute or the National Institutes of Health.
The authors thank the personnel of the STAT MedEvac air medical service for their dedication.
Footnotes
Declaration of Competing Interest
The author(s) have no relevant disclosures. There was no grant funding or financial support for this manuscript.
CRediT authorship contribution statement
All authors listed have made substantial contributions to the conception, design, execution, and write-up of this study.
Supplementary materials
Supplementary material associated with this article can be found in the online version at https://doi.org/10.1016/j.amj.2023.11.004.
References
- 1.EMS Data Cube. Version 3, 2017–2021. Available at: https://nemsis.org/view-reports/public-reports/ems-data-cube/. Accessed August 29, 2022. [Google Scholar]
- 2.Lyng JW, Braithwaite S, Abraham H, et al. Appropriate air medical services utilization and recommendations for integration of air medical services resources into the EMS system of care: a joint position statement and resource document of NAEMSP, ACEP, and AMPA. Prehosp Emerg Care. 2021;25:854–873. [DOI] [PubMed] [Google Scholar]
- 3.Chen X, Gestring ML, Rosengart MR, et al. Speed is not everything: identifying patients who may benefit from helicopter transport despite faster ground transport. J Trauma Acute Care Surg. 2018;84:549–557. [DOI] [PubMed] [Google Scholar]
- 4.Elliott JP, O’Keeffe DF, Freeman RK. Helicopter transportation of patients with obstetric emergencies in an urban area. Am J Obstet Gynecol. 1982;143:157–162. [DOI] [PubMed] [Google Scholar]
- 5.Konstantopoulos WM, Pliakas J, Hong C, et al. Helicopter emergency medical services and stroke care regionalization: measuring performance in a maturing system. Am J Emerg Med. 2007;25:158–163. [DOI] [PubMed] [Google Scholar]
- 6.Clemmensen P, Schoos MM, Lindholm MG, et al. Pre-hospital diagnosis and transfer of patients with acute myocardial infarction–a decade long experience from one of Europe’s largest STEMI networks. J Electrocardiol. 2013;46:546–552. [DOI] [PubMed] [Google Scholar]
- 7.Balazs KT, Thompson CB. Quality assurance and continuous quality improvement within air transport programs. Air Med J. 1996;15:104–107. [DOI] [PubMed] [Google Scholar]
- 8.Aspiotes CR, Gothard MQ, Gothard MD, Parrish R, Schwartz HP, Bigham MT. Setting the benchmark for the ground and air medical quality in transport international quality improvement collaborative. Air Med J. 2018;37:244–248. [DOI] [PubMed] [Google Scholar]
- 9.Galvagno SM Jr, Sikorski R, Hirshon JM, et al. Helicopter emergency medical services for adults with major trauma. Cochrane Database Syst Rev. 2015;12:CD009228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kashyap R, Anderson PW, Vakil A, Russi CS, Cartin-Ceba R. A retrospective comparison of helicopter transport versus ground transport in patients with severe sepsis and septic shock. Int J Emerg Med. 2016;9:15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Meuli L, Zimmermann A, Menges AL, et al. Helicopter emergency medical service for time critical interfacility transfers of patients with cardiovascular emergencies. Scand J Trauma Resusc Emerg Med. 2021;29:168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wejnarski A, Leszczyński P, Świeżewski S, et al. Characteristics of aeromedical transport, both interhospital and directly from the scene of the incident, in patients with acute myocardial infarction or acute trauma between 2011–2016 in Poland: a case-control study. Adv Clin Exp Med. 2019;28:1495–1505. [DOI] [PubMed] [Google Scholar]
- 13.Chesters A, Harris T, Hodgetts TJ, Keefe N. Survival to discharge after cardiac arrest attended by a doctor-paramedic helicopter emergency medical service: an Utstein-style multiservice review of 1085 activations. J Emerg Med. 2015;49:439–447. [DOI] [PubMed] [Google Scholar]
- 14.Florez-Perdomo WA, Garcia-Ballestas E, Konar SK, et al. Effect of helicopter transportation of acute ischemic stroke patients on mortality and functional outcomes: a systematic review and meta-analysis. Air Med J. 2022;41:476–483. [DOI] [PubMed] [Google Scholar]
- 15.Brown JB, Gestring ML, Guyette FX, et al. Helicopter transport improves survival following injury in the absence of a time-saving advantage. Surgery. 2016;159:947–959. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Mabry RL, Apodaca A, Penrod J, Orman JA, Gerhardt RT, Dorlac WC. Impact of critical care-trained flight paramedics on casualty survival during helicopter evacuation in the current war in Afghanistan. J Trauma Acute Care Surg. 2012;73(suppl 1): S32–S37. [DOI] [PubMed] [Google Scholar]
- 17.Bailey JA, Morrison JJ, Rasmussen TE. Military trauma system in Afghanistan: lessons for civil systems? Curr Opin Crit Care. 2013;19:569–577. [DOI] [PubMed] [Google Scholar]
- 18.ZOLL. Build dynamic patient care reports that tell a complete story. Available at: https://www.zolldata.com/ems-software/epcr. Accessed August 29, 2022.
- 19.Champion HR, Sacco WJ, Copes WS, Gann DS, Gennarelli TA, Flanagan ME. A revision of the Trauma Score. J Trauma. 1989;29:623–629. [DOI] [PubMed] [Google Scholar]
- 20.Raux M, Sartorius D, Le Manach Y, David JS, Riou B, Vivien B. What do prehospital trauma scores predict besides mortality? J Trauma. 2011;71:754–759. [DOI] [PubMed] [Google Scholar]
- 21.Galvagno SM Jr, Massey M, Bouzat P, et al. Correlation between the Revised Trauma Score and Injury Severity Score: implications for prehospital trauma triage. Prehosp Emerg Care. 2019;23:263–270. [DOI] [PubMed] [Google Scholar]
- 22.Farzan N, Foroghi Ghomi SY, Mohammadi AR. A retrospective study on evaluating GAP, MGAP, RTS and ISS trauma scoring system for the prediction of mortality among multiple trauma patients. Ann Med Surg (Lond). 2022;76:103536. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Michaels D, Pham H, Puckett Y, Dissanaike S. Helicopter versus ground ambulance: review of national database for outcomes in survival in transferred trauma patients in the USA. Trauma Surg Acute Care Open. 2019;4:e000211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Sørensen LM, Rogna s L, Alstrup K. Trauma versus nontrauma patients treated by the Danish helicopter emergency medical service: a register-based study of patient characteristics, hospitalization, and mortality. Air Med J. 2021;40:325–330. [DOI] [PubMed] [Google Scholar]
- 25.Björkman J, Laukkanen-Nevala P, Olkinuora A, Pulkkinen I, Nurmi J. Short-term and long-term survival in critical patients treated by helicopter emergency medical services in Finland: a registry study of 36 715 patients. BMJ Open. 2021;11:e045642. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Saviluoto A, Björkman J, Olkinuora A, et al. The first seven years of nationally organized helicopter emergency medical services in Finland - the data from quality registry. Scand J Trauma Resusc Emerg Med. 2020;28:46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Sperry JL, Guyette FX, Brown JB, et al. Prehospital plasma during air medical transport in trauma patients at risk for hemorrhagic shock. N Engl J Med. 2018;379:315–326. [DOI] [PubMed] [Google Scholar]
- 28.Cantwell K, Morgans A, Smith K, Livingston M, Spelman T, Dietze P. Time of day and day of week trends in EMS demand. Prehosp Emerg Care. 2015;19:425–431. [DOI] [PubMed] [Google Scholar]
- 29.Ramgopal S, Dunnick J, Owusu-Ansah S, Siripong N, Salcido DD, Martin-Gill C. Weather and temporal factors associated with use of emergency medical services. Prehosp Emerg Care. 2019;23:802–810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Raatiniemi L, Liisanantti J, Tommila M, et al. Evaluating helicopter emergency medical missions: a reliability study of the HEMS benefit and NACA scores. Acta Anaesthesiol Scand. 2017;61:557–565. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.