T EMPORAL T RENDS IN A MBULANCE D IVERSION IN A M ID-SIZED M ETROPOLITAN A REA

Objectives. To evaluate the amount of ambulance diversion in an emergency medical services (EMS) system and to investigate potential predictive factors. Methods. Ambulance diversion status of hospitals in the four-county metropolitan Portland, Oregon, area has been recorded for approximately 15 years. These data are used by EMS transporting agencies to determine appropriate hospital destination for their patients. The authors calculated the total yearly hospital ambulance diversion time for “Total Ambulance Divert (TAD)” and “Critical Care Divert (CCD)” for the time period between January 1, 1996, and December 31, 1999. Yearly EMS 9-1-1-generated patient transport volume, hospital emergency department (ED) census volume, total population, amount of health maintenance organization (HMO) penetration, and number of licensed and available hospital beds were calculated for each yearly interval. Kendall's tau-b correlation was used to determine significant secular trends. Potential predictive factors for the amount of ambulance diversion were tested using Pearson's correlation. Results. Total TAD increased 122.5% (p = 0.04), total CCD increased 64.4% (p = 0.50), total EMS transport volume increased 16.1% (p = 0.04), total ED census increased 9.4% (p = 0.04), total licensed beds decreased 5.7% (p = 0.17), total available beds decreased 15.8% (p = 0.17), HMO penetration increased 4.7% (p = 0.04), and total population increased 9.7% (p = 0.04) over the four-year study period. CCD and TAD were not significantly related to each other (p = 0.50). The only significant factor associated with the increase in TAD was number of available beds (p = 0.03). There were no significant factors associated with CCD. Conclusion. TAD increased significantly over time and was associated only with the decrease in available hospital beds.     

A MBULANCE,F IRE,AND P OLICE D ISPATCH T IMES C OMPARED WITH THE A TOMIC C LOCK

Objectives. To quantify any differences between the times used by public safety answering points (PSAPs) in a multijurisdictional county compared with the atomic clock and to determine whether there was consistency in any time differences. Methods. All 25 ambulance, fire, and police PSAPs were contacted by telephone. The current time in hours, minutes, and seconds on the dispatch center's timepiece was -requested. The atomic clock time was simultaneously recorded. Time differences between the reported and atomic clock times were calculated and the absolute values were used to calculate the mean difference. The procedure was repeated one week later. Consistency in time deviation was evaluated by subtracting the time differences between weeks 1 and 2 for each center. Results. All 25 centers were contacted and three declined to participate. Time differences ranged from ?551 to 117 seconds (mean difference: 61.2 ± 120.3) for week 1 and ?103 to 79 seconds (mean difference: 36.9 ± 33.4) for week 2. Time deviations between weeks 1 and 2 were: 0 seconds for one center, 1 to 30 seconds for 12 centers, 31 to 60 seconds for four centers, and more than 60 seconds for five centers. Conclusions. The maximum time difference between dispatch center and atomic clock times was 551 seconds. This difference may be clinically significant for time-dependent research, quality improvement tasks, or medical legal reviews when multiple PSAPs are involved. Lack of consistency in time deviation over one week suggests systematic adjustment for these differences may not be possible.     

R EDUCTION OF THE C ALL -R ESPONSE I NTERVAL WITH A MBULANCE B ASE P AGING

Objective. To determine whether the call-response interval for an emergency medical services (EMS) system would be decreased through the introduction of ambulance base paging. Methods. The study community included a mixture of urban and rural areas with a total population of approximately 400,000. The EMS system is composed of two ambulance services and one central ambulance communication center with computer-aided dispatching capabilities. Approximately 30,000 calls are responded to yearly by the combined ambulance services. A before-and-after study design was used. In a retrospective review of one ambulance service, there were 224 calls collected in the period before base paging and 200 calls collected in the period after base paging was introduced. In the other ambulance service, there were 571 calls captured in the period before base paging and 515 calls captured in the period after base paging. Results. The call-receipt-to-crew-notified interval was reduced from the before period to the after period in both ambulance services: Cambridge—61.8 to 49.8 seconds (p < 0.0001); Kitchener—66.6 to 46.2 seconds (p < 0.0001). The crew-notified-to-vehicle-mobile interval was reduced from the before period to the after period in both ambulance services: Cambridge—91.8 to 73.2 seconds (p < 0.0001); Kitchener—80.4 to 66.0 seconds (p < 0.0001). Conclusions. The introduction of ambulance base paging reduced components of the call-response interval in this EMS system. Overall, the reduction in time was approximately 30 seconds, which was found to be statistically significant.     

A PPROPRIATENESS OF I NTRAVENOUS C ANNULATION BY P ARAMEDICS : A L ONDON S TUDY

Introduction. The number of patients undergoing intravenous (IV) cannulation by paramedics has increased dramatically over recent years in the UK. Treatment protocols for cannulation in the field are loosely defined. Variation in practice may lead to patients' receiving differential treatment according to customary practice, rather than according to their clinical conditions. Objectives. To explore variations in practice and assess level of appropriatenesss of IV cannulation by London Ambulance Service (LAS) paramedics; to revise treatment protocols and work toward clinical guidelines, if indicated by study findings. Methods. Skill usage data were analyzed for all LAS paramedics for 1995–96. All patients who were IV-cannulated and transported to three hospitals by LAS during March 1996 were identified. A panel of accident and emergency (A&E) and prehospital specialists judged each case for appropriateness. Results. Variation during the year was wide, with a range of 1 to 221 (mean 47) patients cannulated per paramedic, although the majority showed some consistency in frequency of skill usage. A sample of 183 cases was reviewed. The majority judged 149 (81.4%) to be appropriate, although there was considerable disagreement between reviewers (κ = 0.43, p < 0.001). Data suggested that those paramedics who cannulate more frequently cannulated less appropriately during the study period (lowest 30%: 73.9% appropriate; highest 30%: 45.8% appropriate, p = 0.05). Conclusion. Despite wide variation between paramedics, the panel judged overall appropriateness of cannulation to be high. The audit advisory group judged that new clinical guidelines might not achieve an improvement in practice and were not supported by study findings. It was recommended that variations be addressed through individual practice review.     

E MERGENCY M EDICAL S ERVICES AND C ULTURAL D ETERMINANTS OF AN E MERGENCY IN K ARACHI,P AKISTAN

Objectives. The study was conducted to understand the prehospital system in Karachi, the mode of transport that adult inpatients use to reach the emergency departments (EDs), and the barriers to the use of ambulances. Methods. The study consisted of two parts. The first part involved interviewing the administrators of major ambulance services in Karachi. The second part consisted of a structured interview of randomly selected adult inpatients admitted to one government and one private hospital. Results. Seven ambulance service administrators were interviewed. The interviews revealed that ambulances in Karachi are mainly involved in transporting patients from hospital to hospital or to home. A large number of calls are for transporting dead bodies. A total of 92 patients were interviewed (58 male, 34 female). Admission complaints included abdominal pain (22), blunt trauma (11), penetrating trauma (3), chest pain (6), shortness of breath (4), hematemesis (3), acute focal weakness (4), high fever (4), and other (32). The most common mode of transport to the ED was taxi (53, 58%), followed by private car (21, 23%). Specific reasons for not using ambulances included a perception that the patient was not sick enough (34, 45%), slow response of the ambulance services (17, 23%), not knowing how to find one (8, 11%), and the high cost (6, 8%). Conclusion. In case of a medical emergency, most people in Karachi do not use ambulances. The reasons for this low usage include not only poor accessibility, but also cultural barriers and lack of education in recognition of danger signs.     

P ROSPECTIVE D ETERMINATION OF M EDICAL N ECESSITY FOR A MBULANCE T RANSPORT BY P ARAMEDICS

Objective. It has been estimated that between 11% and 61% of ambulance transports to emergency departments are not medically necessary. This study's objective was to analyze paramedic ability to determine the medical necessity of ambulance transport to the emergency department. Methods. Paramedics prospectively assessed adult patients transported to an emergency department during a six-week period. The setting was an urban, all advanced life support, public utility model emergency medical services (EMS) system with 58,000 transports per year. Paramedics determined medical necessity of patient transport based on the following five criteria: 1) need for out-of-hospital intervention; 2) need for expedient transport; 3) potential for self-harm; 4) severe pain; or 5) other. On arrival in the emergency department, the emergency physician made a blinded determination based on the same criteria. Kappa statistics were used to assess agreement. Results. Data forms were completed on 825 of 1,420 (58%) patients transported. Emergency physicians determined 248 (30%) transports were not necessary, paramedics 236 (29%), with agreement in 76.2% (K = 0.42) of cases. Paramedics undertriaged 92 patients (11%). Rates of agreement on the five criteria were: 1) 71.9% (K = 0.43); 2) 77.7% (K = 0.22); 3) 89.6% (K = 0.40); 4) 89.6 (K = 0.32); and 5) 82.2% (K = 0.29). Conclusions. Paramedics and emergency physicians agreed that a significant percentage of patients did not require ambulance transport to the emergency department. Despite only moderate agreement regarding which patients needed transport, the undertriage rate was low.     

P REDICTED U TILIZATION OF E MERGENCY M EDICAL S ERVICES T ELEMEDICINE IN D ECREASING A MBULANCE T RANSPORTS

Objective. To determine predicted utilization, decrease in ambulance transports, and target population for emergency medical services (EMS) if telemedicine capabilities were available to the medic units in the field. Methods. A retrospective chart review of 345 consecutive ambulance transports to four hospitals (Level I urban trauma center, urban tertiary care center, children's hospital, and suburban community hospital) was performed by a panel of three board-certified emergency medicine physicians experienced and credentialed in emergency telemedicine. They independently reviewed the emergency department (ED) and EMS records and were asked to determine whether patients required ambulance transport for evaluation or whether disposition could be made following paramedic and emergency physician assessment via telemedicine. A five-point Likert scale was used to grade feasibility of telemedicine disposition (definitely yes, probably yes, maybe, probably no, definitely no). Other variables analyzed included age, sex, race, chief complaint, phone, private medical doctor, and call location by patient zip code, call site, and receiving hospital. Results. In 14.7% of cases (6% definitely yes and 8.7% probably yes), disposition could be made without transport using telemedicine. The age range for eliminating transport was 2 weeks through 92 years, with mean age of 26.6 years. Under the age of 50 years, 46 out of 238 patients (19.3%) could have possibly been managed by telemedicine. Conclusion. Use of EMS telemedicine could result in an approximately 15% decrease in ambulance transports when it alone is added to the prehospital care provider's armamentarium. Emphasis for implementation should be placed on younger patients and an identified subset of chief complaints conducive to management using telemedicine.     

A C ASE S ERIES A NALYSIS OF M ASS C ASUALTY I NCIDENTS

Introduction. Mass casualty incidents (MCIs) are infrequent but potentially overwhelming events that can stress the capabilities of even the most organized emergency medical services (EMS) system. The Maryland EMS system has been identified as a pioneer and leader in the field of prehospital emergency care and, as with many states, Maryland's regional preparation for MCIs has been integrated into its overall EMS systems planning. Objective. To determine how successful this integration has been by examining a three-year history of response to MCIs in Maryland. Methods. A three-year case series of MCIs in Maryland was obtained from a Nexis national news publications search. These MCIs were cross-referenced with U.S. postal ZIP codes and the U.S. Census Bureau's ZIP code files. They were then mapped and summary statistics were prepared for analysis. Data obtained through the Maryland Health Services Cost Review Commission for all severely injured patients discharged from Maryland hospitals were obtained over the same three-year period for comparison. Results. Eight MCIs occurred over a three-year period, resulting in a total of 203 injuries. An average of 25.4 ± 10.7 injuries occurred per MCI. A total of 158 (77.8%) of injuries necessitated ambulance transportation. An average of 3.1 ± 1.1 hospitals were involved per MCI. Conclusions. The Maryland EMS system was effective in responding to MCIs ranging in size from 10 to nearly 40 injuries. Analyzing MCIs that reoccur on a year-to-year basis should figure into the planning process for EMS systems.     

P ARAMEDIC U SE OF E NDOTRACHEAL T UBE I NTRODUCERS FOR THE D IFFICULT A IRWAY

Objective. To evaluate the ability of paramedics to learn and apply the skill of introducer-aided oral intubation in the setting of the simulated “difficult airway.” The authors hypothesized that, following a brief introduction to the device, intubation success rates would not differ for traditional and introducer-aided intubations of an immobilized airway mannequin. Methods. During a paramedic recertification class, experienced paramedics were given a brief didactic introduction to the “bougie-like” Flex Guide endotracheal tube introducer (ETTI). The participants were then asked to intubate adult mannequins immobilized in the head-neutral position, with and without the ETTI. “Successful placement” was defined as completion of the procedure within 30 seconds and endotracheal tube position confirmed by the investigator with direct visualization. Results. For both traditional and ETTI intubations, 34 (97%) of the 35 paramedics successfully intubated within 30 seconds. The two unsuccessful intubation attempts were recognized by the paramedic as esophageal intubations, and correct tube placement was obtained within an additional 30 seconds. Conclusion. In this study, use of the ETTI was mastered by the participants after only a brief didactic introduction to the device, with their ability to intubate an immobilized mannequin using the ETTI being equal to their ability to perform traditional intubation. These results suggest that use of the ETTI is easily learned, and may support the device's role in the prehospital management of the difficult airway.     

A GITATED D ELIRIUM AND S UDDEN D EATH : T WO C ASE R EPORTS

Objective. To determine the chemicals involved in fire department hazardous materials (hazmat) responses and analyze the concomitant emergency medical services' patient care needs. Methods. The setting was a mid-sized metropolitan area in the southwestern United States with a population base of 400,000 and an incorporated area of 165 square miles. The authors conducted a retrospective evaluation of all fire department hazmat reports, with associated emergency medical services patient encounter forms, and in-patient hospital records from January 1, 1992, through December 31, 1994. Results. The fire department hazardous materials control team responded to 468 hazmat incidents, involving 62 chemicals. The majority of incidents occurred on city streets, with a mean incident duration of 46 minutes. More than 70% of the responses involved flammable gases or liquids. A total of 32 incidents generated 85 patients, 53% of whom required transport for further evaluation and care. Most patients were exposed to airborne toxicants. Only two patients required hospital admission for carbon monoxide poisoning. Conclusion. Most hazmat incidents result in few exposed patients who require emergency medical services care. Most patients were exposed to airborne toxicants and very few required hospitalization. Routine data analysis such as this provides emergency response personnel with the opportunity to evaluate current emergency plans and identify areas where additional training may be necessary.     

T HE E FFECTS OF P REHOSPITAL A DVANCED L IFE S UPPORT D RUG T REATMENT ON P ATIENT I MPROVEMENT AND I N-HOSPITAL U TILIZATION

Objective. There is little published evidence to support the benefits of prehospital drug administration by ambulance personnel in reducing subsequent hospital utilization by the medical patients receiving such drugs. The authors studied the outcome of patients treated by Ontario's Emergency Health Services “Symptom Relief Drug Program,” which was developed to relieve patient symptoms in the field for specific medical emergencies. Methods. A retrospective study spanning a three-year period from January 1996 to December 1998 was undertaken in a mid-sized southern Ontario community. From a review of ambulance call reports (ACRs), eligible patients were recruited by mail and divided into two groups: those treated before the introduction of the program (pre) and those treated after (post). Out-of-hospital data were retrieved from ACRs and in-hospital data were gathered from medical chart reviews. Outcomes included emergency department (ED) length of stay (LOS), frequency of admissions, and departmental use. Secondary endpoints included differences in prehospital improvement, ED interventions, and ambulance scene times. Results. For the unpaired analysis, 406 patients provided consent (pre: 215 vs post: 191). Ambulance time on scene was longer in the post group, 14.2 minutes (95% CI 13.7–14.8), versus the pre group, 12.3 minutes (95% CI 11.7–12.9), p < 0.001. A larger proportion of patients receiving prehospital drug treatment were judged to have improved on ED arrival (pre: 19.5% vs post: 48.2%, χ² p < 0.0001). The ED LOSs did not differ between groups (pre: 206.9?min, 95% CI 185.9–230.4, vs post: 220.9?min, 95% CI 196.9–247.7, p = 0.42) but were shorter within the post group for hypoglycemic patients receiving glucagon. The overall proportion of admissions was significantly lower in the post group (pre: 145 [67.4%] vs post: 102 [54.3%], χ² p < 0.01), and this was driven by chest pain patients. Conclusions. The lower rate of admissions for chest pain patients is the first published evidence of prehospital drug treatment's reducing hospital utilization in a subgroup of such medical patients. The “Symptom Relief Drug Program” is effective in improving patients' field conditions and can decrease ED LOS in hypoglycemic persons receiving glucagon injections. More outcome research pertaining to ambulance-administered prehospital drug treatment is warranted.     

U SE OF E MERGENCY M EDICAL S ERVICES FOR S USPECTED A CUTE C ARDIAC I SCHEMIA AMONG D EMOGRAPHIC AND C LINICAL P ATIENT S UBGROUPS : T HE REACT T RIAL

Objective. Barriers to the use of emergency medical services (EMS) and patient delay in seeking care can limit the receipt or effectiveness of reperfusion therapies and the availability of prehospital emergency cardiac care. The Rapid Early Action for Coronary Treatment (REACT) trial was designed to determine the impact of a community intervention on use of EMS among demographic and clinical subgroups of patients with suspected acute cardiac ischemia. Methods. A randomized controlled community trial was conducted in 20 pair-matched communities in the United States. One community from each pair received an 18-month, multicomponent community education program. Data were collected at 44 participating hospitals during a four-month baseline period and throughout the 18-month trial, using medical record abstracts to collect information on mode of transport to the hospital and other sociodemographic and clinical variables. Eligible patients were persons aged ≥30 years presenting with chest pain or discomfort to emergency departments (EDs) who were admitted and discharged with a cardiac-related diagnoses (ICD 410-414, 427-429, 440, 786.9). Results. The net change in the odds of EMS use was an increase of 34% in intervention compared with control communities [adjusted odds ratio (OR) 1.34, 95% CI 1.07–1.67]. We observed greater increases in the odds of EMS use among patients who had chronic or other cardiac diagnoses (adjusted OR 1.53, 95% CI 1.18–1.99, and adjusted OR 1.52, 95% CI 1.17–1.97, respectively) than in those diagnosed as having acute ischemia (adjusted OR 1.14, 95% CI 0.91-1.44). We observed greater increases in odds of EMS use in those who were retired (adjusted OR 1.62, 95% CI 1.29–2.04) or had systolic blood pressure (SBP) at or below 160?mm Hg upon presentation to the ED (adjusted OR 1.55, 95% CI 1.26–1.91 for SBP 100-160 mm Hg; 1.61, 95% CI 0.88–2.97 for SBP <100?mm Hg). Conclusions. The REACT trial demonstrated a significant impact on the use of EMS among patients admitted to the hospital for suspected acute myocardial infarction, with greater increases among patients with chronic or other cardiac ICD-9 discharge diagnoses, those presenting with lower SBP, and retired persons.     

P REHOSPITAL C ARE R ESEARCH F ORUM /NAEMSE 2002

Objectives. Emergency medical dispatch (EMD) protocols are intended to match response resources with patient needs. In a small city that previously sent a first-responder basic life support (BLS) engine company lights-and-siren response to every emergency medical services (EMS) call, regardless of nature or severity, an EMD system was implemented in order to reduce the number of such responses. The study objectives were to determine the effects of the EMD system on first-responder call volume andto assess the safety of the system. Methods. This was a prospective, before–after trial. Using computer-assisted dispatch (CAD) records, all EMS calls in the 120 days before implementation of the EMD protocol andthe 120 days after implementation were identified (excluding a one-month wash-in period). In the “after” phase, patient care reports of a random sample of cases in which an ambulance was dispatched with no first responders was manually reviewed to assess whether there might have been any benefit to first-responder dispatch. Given the lack of accepted clinical criteria for need for first responders, the investigators' clinical judgment was used. Paired t-tests were used to compare groups. Results. There were 9,820 EMS calls in the “before” phase, with 8,278 first-responder engine runs (84.3%), and9,943 EMS calls in the “after” phase, with 3,804 first-responder engine runs (39.1%). The first-responder companies were dispatched to a median of 5.65 runs/day (range 1.1–12.7) in the “before” phase, and3.17 runs/day (range 0.6–5.0) in the “after” phase (p = 0.0008 by paired t-test). Review of 1,816 “after” phase ambulance-only patient care reports (PCRs) found ten (0.55%) in which first-responder dispatch might have been beneficial, but review of EMS andemergency department (ED) records found no adverse outcomes in these ten patients. Conclusions. This study suggests that a formal EMD system can reduce first-responder call volume by roughly one-half. The system appears to be safe for patients, with an undertriage rate of about one-half of one percent.     

T HE L AY P UBLIC ' S E XPECTATIONS OF P REARRIVAL I NSTRUCTIONS W HEN D IALING 9-1-1

Objective. To determine whether the lay public expects public safety answering points (PSAPs) to provide prearrival instructions. Methods. Two thousand telephone numbers were randomly generated from all listed residential numbers in a county containing urban, suburban, and rural communities served by 26 enhanced 9-1-1 PSAPs. Only a minority of the PSAPs provided prearrival instructions. Research assistants made two attempts to contact an individual at each telephone number. A survey was administered to any person who answered the telephone provided the person was at least 18 years of age and gave verbal consent. The respondents were asked their age, level of education, and gender. They were also asked what number they would call for first aid or an ambulance and whether they would expect telephone instructions from the dispatcher if a close relative was choking, not breathing, bleeding, or giving birth. Results. One thousand twenty-four individuals were successfully contacted; and 524 (51%) were at least 18 years of age and agreed to participate. The respondents' mean age was 50 (standard deviation 19 years). Sixty-five percent of the respondents were female; and 90% had at least a high school diploma. Only 37% had previously called 9-1-1 (nine-one-one) for an emergency. Ninety-seven percent said they would dial either 9-1-1 (85%) or 9–11 (nine-eleven) (12%) in an emergency. Seventy-six percent (95% CI: 73%–80%) expected prearrival instructions for all four medical conditions. Specifically, prearrival instructions were expected by: 88% for choking (95% CI: 85%–90%), 87% for not breathing (95% CI: 84%–90%), 89% for bleeding (95% CI: 86%–91%), and 88% for childbirth (95% CI: 86%–91%). Ninety-nine of 117 respondents (81%) served by a PSAP that did not provide prearrival instructions expected to receive phone instructions for all four emergencies. Logistic regression revealed that knowing to dial 9-1-1 or 9–11 in an emergency was the only significant predictor of prearrival instruction expectation [p < 0.03, odds ratio 3.4 (95% CI: 1.16–9.78)]. Age, gender, service by a PSAP providing prearrival instructions, and level of education were not predictive. Conclusion. The lay public expects prearrival instructions when calling 9-1-1, although they may not currently receive this service.     

A N E MERGENCY M EDICAL S ERVICES P ROGRAM OF A LTERNATE D ESTINATION OF P ATIENT C ARE

Objective: The emergency department (ED) is ideally reserved for urgent health needs. The ED, however, is often the site of care for nonurgent conditions. The authors investigated whether emergency medical technicians could decrease ED use by patients with nonurgent concerns who use 911 by appropriately identifying and triaging them to alternate care destinations. Methods: From August 2000 through January 2001, two King County fire-based emergency medical services (EMS) agencies participated in an alternate care destination program for patients with specific low-acuity diagnosis codes (intervention group). Eligible patients were offered care at a clinic-based destination as an alternate to the ED (n = 1,016). The frequency of the destination of care (ED, clinic, or home) for the intervention group was compared with a matched control group that was comprised of a preintervention historical cohort of EMS encounters from the same two fire-based agencies and with the same acuity and diagnosis criteria and seasonal interval (n = 2,617). Results: Compared with the preintervention group, a smaller proportion of patients in the intervention group received care in the ED (44.6% vs. 51.8%, p = 0.001), while a greater proportion of patients in the intervention group received clinic care (8.0% vs. 4.5%, p = 0.001) or home care (no transport) (47.4 vs. 43.7%, p = 0.043). Results were comparable when adjusted for other patient characteristics. Similar relationships were not evident among nonparticipating King County EMS agencies. Based on physician review and patient interview, the alternate care intervention appeared to be safe and satisfactory. Conclusion: An EMS-based program may represent one approach to limiting nonurgent ED use.     

Abstracts for the 2007 NAEMSP Scientific Assembly

Objective. To determine the number of air medical helicopter accidents in the United States during a five-year period beginning January 1, 1997, and ending December 31, 2001. Methods. The National Transportation Safety Board's (NTSB's) Accident Synopses database was accessed to determine the number of accidents involving air medical helicopters during the five-year study period. The NTSB reports for each accident were downloaded. Results. The NTSB records revealed 47 accident files pertaining to air medical helicopters during the five-year study period. These were analyzed for: date of accident, time of accident, air ambulance operator, location of accident, type of aircraft, number of persons, number of fatalities, number of injuries, cause of accident, and other factors the NTSB investigators deemed appropriate. Of the 47 accidents, there were 40 fatalities and 36 injuries. Overall, there were 13 helicopter types involved. The majority of accidents (70%) were attributed to pilot error. The number of accidents increased from a low of 4 in 1997 to a maximum of 12 in both 2000 and 2001. Conclusions. This study has examined 47 U.S. air medical helicopter accidents for a five-year period (1997–2001). There was an increase in the number of accidents during the study period. However, this study is limited by the fact that it presents only raw data and does not reflect the actual incidence of accidents per hours flown. Various factors related to these accidents have been described. These factors should be considered when strategies to improve air medical helicopter safety are developed.