Predictors of short-term survival after helicopter rescue

RESUSCITATION

ELSEVIER

Resuscitation30 (1995)133-140

Predictors of short-term survival after helicopter rescue Johannes Bonatti*a’b, Oskar GGschlavb,Peter Larcherapb,Ralf W6dlingerayb, Gerhard Floraalb ‘Department of Surgery I/University of Innsbruck School of Medicine, Anichstrajle 35, A-6020 Innsbruck. Austria bdAMTC-Helicopter Emergency Medical System “Christophorus I”, Flughafen, A-602# Innsbruck, Austria

Received8 April 1995;revision received18July 1995;accepted20 July 1995 Abstract The purpose of the present retrospective study was to identify easily obtainable predictors of short-term outcome for emergencyvictims treated by a physician-staffed helicopter emergencymedical system(HEMS). The study was conducted at the HEMS unit ‘Christophorus 1’ based at Innsbruck, Austria. Outcomes for 2139 patients rescued in primary missions during a 3-year period from 1 January 1989 to 31 December 1991 were included in the study. The majority of missions were in responseto sports accidents, although missions included a wide spectrum of emergencies. Data were obtained from the ‘Christophorus 1’ operation protocols and by written, personal, or telephone request from admitting hospitals. Eleven parameters selectedfrom the HEMS fli&t logs were tested for their predictive value on survival following helicopter rescue.In a univariate analysis, the causeof the emergency,time at the scene,total duration of the emergencymission, patient age,patient gender, severity of the emergencyusing the National Advisory Committee of Aeronautics (NACA) scoring system,state of consciousness,respiratory status and patient circulatory status each had a statistically significant influence on survival up to 90 days following the emergency.Flight time to the scene and the original specialty of the additionally trained emergency physician had no significant influence on outcome. Multivariate analysis using the Cox proportional hazards model revealed that severity of the emergencyby the sevenlevel NACA scale (P = O.OOOl),initial respiratory status (P = O.OOOl),time at the scene (P = O.OlOS),patient age (P = 0.0047) and patient gender (P= 0.0477) were each independent predictors of short-term survival following physician-staffed helicopter rescue.We conclude that the parametersdescribed above can be used in an initial predictive assessmentby the flight physician and the admitting institution. Keyword:

Helicopter; Survival; NACA score; Emergency medical system; HEMS; Cox proportional hazards model

1. Introduction

Throughout Europe, physician-staffed helicopter emergency medical systems (HEMS) have Abbreviations: EMT, emergency medical technician; HEMS, helicopter emergency medical system; NACA, National Advisory Committee of Aeronautics. * Corresponding author, Tel: +43-512-5042529; Fax: +43-512-5042528.

been introduced recently in order to transport qualified emergency physicians rapidly to emergency scenes[I]. Several attempts have been made to demonstrate superior clinical outcome for emergency victims treated by emergency physicians at the scene compared with those victims rescued by conventional emergency medical technician (EMT) or paramedic-staffed emergency medical systems [2]. Reduced mortality rates for emer-

0300-9572B51SO9.050 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0300-9572(95)00883-V

134

J. Bonatti er al. /Resuscitation

gency victims transported by helicopter as compared with more conventional transport techniques have been reported [3] and the transport mortality of patients suffering multiple trauma has been shown to be decreasedfrom 25% to 1% by using physician-staffed rescue systems [4]. Further, a reduction in the average intensive care unit length of stay has been documented for patients transported via helicopter [4] and this reduction advanced in support of the technique in discussions of cost-benefit ratios for helicopter rescue systems.In order to demonstrate further the medical benefits of physician-staffed helicopter rescue, however, identification of those parameters affecting patient outcome is necessary. For the helicopter-based emergencyphysician a predictive model would be a useful tool for the initial assessment of the victim at the scene.Few current studies provide information regarding the specific parameters influencing survival outcome of patients who are treated by the HEMS rescue team

PI.

The aim of the present retrospective study was therefore to identify easily obtainable predictors of patient outcome which might demonstrate the efficacy of physician staffed HEMS not only for trauma cases, which are the focus of most studies concerning helicopter rescue [2,4,6,7] but also for a broad spectrum of medical and surgical emergencies. 2. Methods We studied 2139 patients rescued in primary missions by a physician-staffed helicopter during a 3-year period from 1 January 1989to 31 December 1991. Data were obtained from the HEMS operation protocols and by written, personal, or telephone request from medical records of the admitting hospitals (Innsbruck University Hospital and nine local hospitals). Flight logs were recorded in a computerized documentation system immediately following each mission and follow-up data were added to the database as hospital discharge summaries were received. The HEMS ‘Christophorus 1’ is based in western Austria at the Innsbruck airport and serves the alpine area in a 50 km radius around

30 (1995) 133-140

Innsbruck extending to the German border in the north and the Italian border in the south. The helicopter used during the study period was an Aerospatiale AS 350 B2 staffed by a pilot, an emergencyphysician, and an EMT. The whole rescue team was specially trained in alpine rescue techniques. Emergency physicians on call were recruited from the Departments of Surgery and Anesthesiology at Innsbruck University Hospital after additional training in helicopter emergency medicine. Primary missions included sporting accidents (811 ski accidents = 37.9% of total missions; 272 hiking and climbing accidents = 12.7%; 66 paragliding crashes= 3.1%), motor vehicle accidents (237 = 11.l%), occupational injuries (119 = 5.6%), medical emergencies (344 = 16.I%), neurological emergencies (60 = 2.8%) and other missions (230 = 10.7%).Severity of the emergency was rated using the seven-levelNational Advisory Committee of Aeronautics (NACA) scale in its modification by Tryba (Table 1) [8]. Assessmentof initial vital signs (state of consciousness,respiratory status, circulatory status) was documented by the flight physician on a fourpoint scale:state of consciousnessby clinical exam was assessedas ‘patient alert’ (A), ‘patient responding to vocal stimuli’ (V), ‘patient responding to painful stimuli’ (P), or ‘patient unresponsive’

Table 1 The sevenlevel NACA scaleextending from level 0 (no injury) to level 7 (death) is used by most Austrian, German and Swiss emergencymedical systems NACA 0 NACA 1 NACA 2 NACA 3 NACA 4 NACA 5 NACA 6 NACA 7

No injury or disease Injuries/diseaseswithout any need for acute physicians care Injuries/diseasesrequiring examination and therapy by a physician but hospital admission is not indicated Injuries/diseaseswithout acute threat to life but requiring hospital admission Injuries/diseaseswhich can possibly lead to deterioration of vital signs Injuries/diseaseswith acute threat to life Injuries/diseasestransported after successfulresuscitation of vital signs Lethal injuries or diseases(with or without resuscitation attempts)

J. Bonatti et al. /Resuscitation

(U); respiratory status was graded as ‘sufficient (Grade I), ‘insufficient’ (Grade II), ‘patient ventilated’ (Grade III), or ‘respiratory arrest’ (Grade IV); circulatory status was described as ‘stable’ (Grade I), ‘early shock’ (Grade II), ‘manifest shock’ (Grade Ill), or ‘circulatory arrest’ (Grade IV). All descriptions were basedon the initial clinical status of the emergency victim. For univariate survival analysis the life-table method and the Wilcoxon (Gehan) test were employed. For multivariate analysis the Cox proportional hazards model was applied. Statistical analyseswere performed on SPSSfor Windows’s’. Eleven parameters were examined for possible influence on survival: cause of injury or emergency; time factors (flight time to the scene, time at the scene,total duration of the mission); demographic factors (patient age, patient gender); patient status at the scene(NACA score, state of consciousness, respiratory status, circulatory status); original specialty (surgery vs. anesthesiology) of the emergency physician. P I 0.05 was considered to be statistically significant. 3. Results

Overall patient survival at 30 days for all missions flown was 87.9% by univariate analysis. Cumulative survival rates at 30 days assorted by type of injury or emergency were as follows: ski injuries 99.1%; alpine climbing or hiking accidents 96.8%; paragliding crashes 98.1%; motor vehicle accidents 86.1%; occupational injuries 90.9%. Survival rates for medical and neurologic emergencies at 30 days were 56.2%and 44.6%, respectively. All of thesedifferences were statistically significant by univariate analysis at the P < 0.0001 level. There was no significant association found between short-term survival and flight time to the scene(P = 0.1256). The actuarial survival rates by flight time to the sceneare illustrated in Fig. 1. For flight times between 0 and 10 min, cumulative survival at 30 days was 81.6%. With flight times between 1l-20 min and 21-30 min, 30-day survival rates were 91.1% and 93.7%, respectively. For flight times greater than 30 min, the survival rate was 88.7%at 15days (data on 30-day survival were not available). The rates of patients who showed

30 (1995) 133-140

135

Fig. 1. Actuarial survival by flight time to the scene. For flight times between 0 and 10 min probability of survival at 30 days lies in the 80% range. For longer flight times survival rates are higher. This difference reaches no statistical significance.

no vital signs upon arrival of the helicopter team were 10.1%for flight times between 0 and 10 min, 4.8% for flight times between 11 and 20 mitt, 4.2% for flight times between 21 and 30 min and 10.3% for flight times greater than 30 min (P < 0.0001, Chi-square test). The actuarial survival rates by time at the scene are illustrated in Fig. 2. For times between 0 and 10 min and between 11 and 20 min, survival at 30 days was 97.5% and 95.5%, respectively. With increasing lengths of time at the scene,30day survival rates declined: for times between 21 and 30 min

Fig. 2. Actuarial survival illustrated for scene times of O-10 min, 1l-20 min, 21-30 min, 31-40 min and above 40 min. For times at the scene up to 20 min, probability of survival lies above 95%. The survival rate drops significantly with increasing time at the scene.

136

J. Bonatti et al. /Resuscitation 30 (1995) 133-140

survival was 63.2%, while with on-scenetimes between 31 and 40 mm survival was 30.8%. For time at the scenegreater than 40 min, survival dropped to just 8.1% at 30 days (P < 0.0001). The total duration of the emergencymission was inversely associatedwith survival at 30 days. For missions lasting O-20 min, 30-day survival was 95.5%. For missions lasting between 21 and 40 min, survival at 30 days was 91.7%. Survival rates decreasedto 87.6% and 86.8% for total mission times between 41 and 60 min and 61 and 80 min, respectively. For missions lasting longer than 80 min, survival at 30 days further decreasedto 78.8% (P = 0.0001). Turning to patient demographic factors, younger patients had significantly higher shortterm survival rates compared with older patients, as is illustrated in Fig. 3. Survival at 30 days was 93.6% for patients O-25 years old and 92.1% for those 26-50 years old. By comparison, survival at 30 days was 75.6% for patients 51-75 years old, further declining to 64.4% for those patients over 75 years old (P < 0.0001). No significant association was found between patient gender and survival at 30 days; however, there was a significant relationship between gender and survival in the first 15 days following the emergency: females had a 15-day survival rate of 93.7%compared with 88.9% for males (significant at P = 0.0035). Survival rates at 30 days were near-

Fig. 3. Actuarial survival illustrated for age groups O-25 years, 26-50 years, 51-75 years and above 75 years. For patients under 51 years old, probability of survival is approximately 90%. Survival rates drop significantly for the older age groups.

ly identical for females and males, at 86.9% and 87.9%, respectively. These results are depicted in Fig. 3. Severity of the emergency, as classified by the seven-level NACA scoring system, was strongly and inversely associated with short-term survival (Fig. 4). There were no deaths for casesclassified at NACA levels 1,2, or 3. NACA level 4 caseshad a 30-day survival rate of 97.2%.NACA 5 caseshad a survival rate of 85.6% at 30 days, while NACA level 6 emergency patients had a survival rate of only 12.8% at 30 days (P < 0.0001). The initial patient status parameters - state of consciousness,respiratory status and circulatory status - were each found to have a statistically significant association with patient survival by univariate analysis For patients found to be ‘alert (A)’ at the scene,survival at 30 days was 97.4%. Patients ‘responding to vocal stimuli (V)’ had a 30day survival rate of 96.7%. For patients ‘responding to painful stimuli (P)’ the 30-day survival rate was 75.4%, whereas patients found ‘unresponsive (U)’ had a 30-day survival rate of 14.1% (P < 0.0001). Patients with initial respiratory status assessedas ‘suffkient’ at the scenewere found to have 30-day survival rates of 99.5%. Survival fell to 57.0%at 30 days for patients assessedas having ‘insufficient’ respiratory status at the scene. Patients artificially ‘ventilated’ at the sceneprior to

Fig. 4. Actuarial survival by NACA score. Survival curves for NACA levels O-3 and NACA level 7 cases are not illustrated (survival 100% and O%, respectively). The survival rate for patients classified NACA level 4 lies above 95%. For higher NACA levels the rate drops significantly.

137

J. Bonatti et al. /Resuscitation 30 (1995) 133-140

arrival of the rescue helicopter had survival rates of 32.7% at 30 days, while those patients found at the scenein ‘respiratory arrest’ without ventilatory support had survival rates of 7.5% at 15 days (Fig. 5, P < 0.0901, data on 30&y survival were not available). Initial circulatory status at the scene had a similar impact on survival rates: patients found in ‘early shock’ had a 97.2% 30-day survival rate; those in ‘manifest shock’ had a 78.2% 30day survival rate and in those patients already in ‘circulatory arrest’ upon arrival of the rescueteam the survival rate dropped to 10.8% at 15 days (P < 0.0001, data on 30-day survival were not available). There was no significant association demonstrated between survival outcomes and original specialty (surgery vs. anesthesiology) of the additionally trained flight physician. Patients treated by surgeons had 30-day survival rates of 87.4%, compared with 88.0%when treated by anesthesiologists (P = 0.4437). Multivariate analysis of the data using the Cox proportional hazards model (seeTable 2) revealed that five of the 11parameters assessedin this study

Fig. 5. Actuarial survival compared with initial patient respiratory status as defined in the operation protocol (‘suffkient,’ ‘insuffkient,’ ‘patient ventilated’ = patient intubated and artificially ventilated by a local practitioner or bystander physician, ‘respiratory arrest’ = respiratory arrest occurred prior to HEMS arrival). A survival rate above 95% for patients with adequate respiratory status is illustrated. The survival rate drops significantly for patients with insufficient respiration. There are further drops in survival rate for patients ventilated prior to arrival of the HEMS team and those patients found in full respiratory arrest.

Table 2 P values of assessedparameters by univariate and multivariate analysis. Except for flight time to the sceneand original specialty of the emergencyphysician, all other parameters investigated were signiticantly associated with survival outcome in univariate analysis. Multivariate analysis revealed that five of the I1 parameters assessedwere independently associated(P < 0.05) with short-term survival following physician-staffed helicopter rescue:NACA score, initial respiratory status, time at the scene,patient age and patient gender P value

Cause of injury/emergency Flight time to the scene Scenetime Total duration of mission Patient age Patient gender NACA score State of consciousness Respiratory status Circulatory status Emergency physician

Univariate analysis

Multivariate analysis

O.oool ns 0.0901 0.0001 0.0001 0.0035 0.0001 0.0001 0.0001 0.0001 ns

ns ns 0.0108 GO47 0.0477 0.0001 ns 0.0001 ns ns

were independently associated with short-term survival following physician-staffed helicopter rescue: severity of the emergency by the seven-level NACA scale (P = O.OOOl),patient respiratory status (P = O.OOOl),time at the scene(P = O.OlOS), patient age (P = 0.0047) and patient gender (P = 0.0477). The remaining six parameters (cause of the emergency flight time to the scene, total duration of the mission, initial state of consciousness,initial circulatory status and original specialty of the physician on call) were not significantly associatedwith survival by multivariate analysis. 4. Discussion The present study found an overall 30-day survival rate of 87.9% for physician-staffed helicopter primary rescue missions. Becauseother studies of HEMS outcomes have examined different patient populations, helicopters staffed by either physicians or paramedics and systems with significant

138

J. Bonatti et al. /Resuscitation

differences in resource allocation [2,3,6,7], direct comparison between studies may not be valid. Mortality rates following HEMS rescueare known for populations with specific types of emergencies [9-l 11,while mortality rates for the phasesbefore and after helicopter transport have been reported for unselected populations of emergency patients I1,41.

The type of emergency which the flight physician faces in action is a statistically significant predictor of survival outcome by univariate analysis in the present study. When confronted with sports accidents, one can expect an excellent survival rate above 95%. For motor vehicle accidents and occupational injuries one would predict survival near 85%. In medical and neurologic emergencies, however, only one in two patients will likely survive (45-55% range). These rough estimations of outcome may be made even prior to departure of the helicopter rescue team, since typically the nature of the emergency is described in the initial call. Surprisingly, our study found no significant association between flight time to the emergency scene and overall outcome. Rather than the expected trend toward improved survival outcomes with shorter response time, univariate analysis of our data showed a trend toward improved survival in those missions with longer flight times. This trend might be explained by increased mortality rates prior to arrival of the helicopter at distant sites, or it is possible that flight time to the emergency sceneis not as critical for survival outcome as has been thought. The former hypothesis, which could not be supported by our data, would suggest a need for a higher density of HEMS bases or faster helicopters, together with improved groundbased local emergency medical systems;the latter hypothesis would de-emphasize speed, which might also be an important step toward improved flight safety in helicopter rescue missions [ 121. These results clearly deserve further evaluation. Turning to the time spent on patient treatment at the emergencyscene,our results suggestthat 20 min representsa critical period. Up to 20 min, excellent survival can be expected, but if patient stabilization at the scenerequires 30 min only two out of three victims will likely survive. After 40

30 (1995) 133-140

min of treatment at the scenejust one in three victims will have a realistic chance of survival. The importance of such time factors has already been documented in the HEMS literature [ 11,131.Attention to the length of time spent to treat the patient can yield valuable prognostic information during the actual mission. As one might expect, we found that patient demographics have a significant impact on outcome following HEMS rescue. Younger emergency victims fare quite well, with over 90% survival in those under 50 years. For victims aged 5l-75 three out of four can be expectedto survive, while only two out of three victims over 75 years old will likely survive HEMS rescuein our population. We also found that patient gender is an independent predictor of outcome, though with a marginal P value in multivariate analysis. Univariate analysis of the data demonstrated that during the first 2 weeksfollowing HEMS rescuefemale patients can expect a better outcome than male patients. This relative survival advantage for femalesseemsto be due to a higher rate of serious emergenciesamong male patients in the study population, with statistically significant higher mean NACA scores(3.6 for malesvs. 3.3 for females).In addition, the distribution of types of emergency was different for malescompared with females:occupational injuries and medical emergencies,with mean NACA scores of 3.7 and 4.1, respectively, occurred predominately among male patients. Patient respiratory and circulatory status, as well as state of consciousness, are assessedimmediately by the flight physician upon arrival at the emergencysceneas primary components of the initial examination. Not surprisingly each of these parameterswas found to have a statistically significant influence on outcome in univariate analysis of our data, although in multivariate analysis initial respiratory status was found to have the strongest independent influence on survival. Our data suggest that when the flight physician finds the patient with adequate initial respiratory status, nearly 100% survival can be anticipated. If initial respiratory status is found to be compromised (‘insufficient,’ in our grading system), then the survival rate drops to below 60% at 30 days. The very poor prognosis of those patients already receiving

J. Bonatti et al. /Resuscitation

artificial ventilation upon arrival of the HEMS team seemsprimarily due to their more serious condition: usually only patients with agonal respirations or respiratory arrest were intubated and ventilated by local or bystander physicians (mean NACA score for these patients was 5.9). The NACA score modified by Tryba in 1980[8], which is used by most Austrian, German and Swissemergencymedical systems,was found to be a statistically significant predictor of survival by both univariate and multivariate analyses of the data. Probability of survival based on NACA scorein our population perfectly met definitions of the scoring system itself, with no deaths anticipated in patients classified as NACA levels 1 through 3. Among patients classified by the flight physician as NACA level 4, there is a small but real chance of death. Among NACA level 5 cases, four out of five can be expected to survive. Only one out of six patients classified as NACA level 6, which by definition is a condition that results in death without emergency medical treatment, can be expected to survive at 1 month following the emergency. Sefrin and Sellner, in a 1993 study, reported 8.7%, 15.3% and 63.2% mortality rates, respectively, for NACA levels 4, 5 and 6 casesin an unselected population of emergency patients rescued by physician-staffed helicopter or ground transport [5]. Their results are quite similar to those reported in the current study; however, differing statistical methods and follow-up intervals makes direct comparison of the two studies difficult. The level of training of HEMS personnel, especially the distinction between physicianstaffed versus paramedic-staffed HEMS rescue teams, has been cited as an important parameter affecting outcome in emergency medicine. Results in the literature are conflicting in this regard [2,7]. In the present study we note that the original specialty (surgery vs. anesthesiology) of our additionally trained emergency physicians had no significant influence on patient survival. In Austria no paramedic-staffed emergency medical systems operate at present, so a direct comparison of physician-staffed versus paramedic-staffed helicopter rescueis not possible in our setting. We suggest that the similar outcomes achieved by the two

30 (1995) 133-140

139

groups of physicians in our system is explained by nearly identical training in emergency medical skills prior to recruitment for HEMS missions. The present retrospective study of 2139 emergency patients rescued by HEMS over a 3-year period was conducted in order to identify predictors of short-term survival following physicianstaffed helicopter rescue. This study might thus further clarify the benefits of helicopter rescuesystems, which are often the subject of major public criticism [12]. For the flight physician, identification of parameters influencing clinical outcome can be of immediate practical value as the basis for a rough prognostic assessmentat the emergency scene. The parameters identified in this study, however, are derived from retrospective analysis of data from a large group and, thus, should be applied with great caution to clinical decisions concerning individual cases. Acknowledgements

We would like to thank Mag. Markus Falk from the Institute of Biostatistics at Innsbruck University for his assistancewith statistical calculations, Univ. Prof. Dr. Norbert Mutz from the University Clinic of Anesthesiology and General Intensive Care Medicine Innsbruck for his help in structuring the manuscript and R. Lawrence DePalma MD from the Division of Medical/Surgical Psychiatry at the University of Alabama at Birmingham for correction of the English translation. References [1] Demartines N, Kiener A, Scheidegger D, Harder F. Hubschrauber-Notatzt System in der Basler Region. Chirurg 1990;61: 457-461. [2] Baxt WG, Moody P. The impact of a physician as part of the aeromedical prehospital team in patients with blunt trauma. J Am Med Assoc 1987;257: 3246-3250. [3] Mackenzie CF, Shin B, Fisher R, Cowley RA. Two year mortality in 760 patients transported by helicopter direct from the road accident scene.Am Surg 1979;45: 101- 108 [4] Muhr G, Kayser M. Mehrfachverletzung - Rettungssystem.Bergung und Erstversorgung. Chirurg 1987; 58: 625. [S] Sefrin P, Sellner J. Qualit~tssicherung in der praklinischen Notfallmedizin. Notfallmedizin 1993; 19: 267-274.

140

J. Bonatti et al. /Resuscitation

(61 Moylan JA, Fitzpatrick KT, Breyer AJ, Georgiade GS. Factors improving survival in multisystem trauma patients. Ann Surg 1987;207(6): 680-683. [7] S&mid U, Muggia-Sullam M, Holch M et al. On scene helicopter transport of the polytraumatized patient. Comparison of a German and American system. UnfaJlchirurg 1993;96: 287-291. [8] Ttyba M, Briiggemann H, Echtermeyer V. Klassifmierung von Erkrankungen und Verletzungen im Notarztrettungssystem. Notfalhned 1980; 8: 725. [9] Kaplan L, Walsh D, Bumey RE. Emergency aeromedical transport of patients with acute myocardial infarction. Ann Emerg Med 1987; 16: 55-57.

30 (1995) 133-140

[lo] Black RE, Mayer T, Walker ML. Air transport of pediatric emergency cases. N Engl J Med 1988; 307: 1465-1468. 1111 Anderson TE, Rose WD, Leicht MJ. Physician-staffed helicopter scene response from a rural trauma center. Ann Emerg Med 1987; 16: 58-61. [12] Bumey RE. Efficacy, cost and safety of hospital-based emergency aeromedical programs. Ann Emerg Med 1987; 16: 227-229. [13] Fischer RP, Flen PC, Miller PW et al. Urban helicopter responseto injury. J Trauma 1984;24: 946-951.