Derivation of emergency medical services dispatch codes associated with low-acuity patients

DERIVATION OF EMERGENCY MEDICAL SERVICES DISPATCH CODES ASSOCIATED WITH LOW-ACUITY PATIENTS Manish N. Shah, MD, Paul Bishop, MA, EMT-P, E. Brooke Lerner, PhD, EMT-P, Tim Czapranski, EMT-P, Eric A. Davis, MD assumes that all patients who call 911 are critically ill and in need of immediate care.2 Operational regulations frequently require that EMS respond within a short response interval, often less than 10 minutes. Additionally, some systems require ambulances to always respond to calls using emergency lights and sirens, even through previous studies have shown that up to 50% of requests for EMS assistance are not emergencies3–7 (Czapranski T, personal communication, 2002). This rapid, lights-and-sirens response for all patients has significant costs because ambulances using emergency lights and sirens are at greater risk of crashing without significantly shortening response intervals.8–12 Furthermore, this philosophy of universal rapid response increases the cost of health care because maintaining and deploying sufficient EMS units to meet short response intervals are expensive. The Medical Priority Dispatch System (MPDS) is a widely used 911 call center-based EMS dispatch system developed by Priority Dispatch Corporation (Salt Lake City, UT).13 This system allows EMS dispatchers to categorize the severity of patient complaints and to assign proper resources to each request. This system works by asking callers a series of specific, scripted questions identifying ‘‘priority’’ complaints that are potentially life threatening. By completing the protocol-driven questions, each case is categorized into pre-existing dispatch codes. Each code has three components. The first component is a number that reflects the protocol number, which is based on the primary complaint such as chest pain. The second component is a letter between ‘‘A’’ and ‘‘E’’ that reflects the patient acuity and level of resources required. ‘‘A’’ represents the lowest level and ‘‘E’’ the highest level, but in a nonlinear fashion. The third part is a subcode to further classify the patient and complaint.13 Based on the code assigned, the dispatcher can triage the severity of each patient’s illness and assign a proper response. If accurate, this 911 call center-based prioritization tool can improve the EMS system efficiency and quality by providing the appropriate level of EMS care within an appropriate time period.14 Studies have been performed using locally derived (non-MPDS) dispatch protocols.2,15–18 These studies have shown that EMS dispatchers can better assign EMS resources using preestablished protocols. To our knowledge, only 3 studies have been performed describing the predictive power of the MPDS system

ABSTRACT Objective. To identify emergency medical services (EMS) dispatch codes associated with basic life support (BLS) level of prehospital care, a proxy for low illness acuity. Methods. This retrospective cohort study was conducted in an urban city with a single advanced life support level EMS provider. The 911 center was certified in using dispatch protocols from Priority Dispatch Corporation (Salt Lake City, UT). Dispatch data on all transported EMS patients from August 2001 to April 2002 were abstracted. The authors prospectively defined a low-acuity patient as one who received BLS-level care and defined a low-acuity dispatch code as one in which at least 90% of coded patients required only BLS care. For each dispatch code or code group, the authors calculated the fraction of patients who received BLS-level care. For each ‘‘A’’-level (lowest category) dispatch code group, the fraction of patients receiving BLS-level care was also evaluated. Results. A total of 19,332 calls met inclusion criteria and were categorized into 118 dispatch codes or code groups. Twenty-eight codes or code groups with 7,801 patients met the authors’ definition of low acuity. Overall, 7,394 patients received only BLS care (94.8%, 95% confidence interval: 94.3%–95.3%). Analysis of ‘‘A’’-level dispatch code groups found BLS use rates of 52.8% to 99.3%. Conclusions. Certain dispatch codes are associated with the delivery of BLS-level care, indicating identification of patients likely to be low acuity. These codes are not necessarily ‘‘A’’-level dispatch codes, which are commonly considered to represent the lowest-acuity patients. Future studies are needed to prospectively validate that these codes do represent low-acuity patients. Key words: ambulance dispatch; emergency medical services; triage; prehospital care; emergency medicine.

PREHOSPITAL EMERGENCY CARE 2003;7:434–439

The emergency medical services (EMS) system exists to respond to members of the community who request emergency assistance for illness or injury.1 To minimize the risk of undertriage, the system traditionally Received February 25, 2003, from the Departments of Emergency Medicine (MNS, EBL, EAD), and Community & Preventive Medicine (MNS, EBL), University of Rochester School of Medicine & Dentistry, Rural Metro Medical Services (PB, TC), and Monroe Community College (PB), Rochester, New York. Revision received April 15, 2003; accepted for publication April 16, 2003. Presented at the National Association of EMS Physicians annual meeting, Panama City, Florida, January 2003. Address correspondence and reprint requests to: Manish N. Shah, MD, 601 Elmwood Avenue, Box 655, Rochester, NY 14642. e-mail: . doi:10.1197/S1090-3127(03)00213-2

434

Shah et al.

435

LOW-ACUITY DISPATCH CODES

despite its wide use.19–21 However, none of these studies used the current version of the MPDS protocols and none analyzed individual MPDS dispatch codes. The objective of this study was to identify MPDS dispatch codes associated with basic life support (BLS)level prehospital care, a proxy for low-acuity patients, at least 90% of the time. A secondary objective was to evaluate the proportion of calls that were assigned ‘‘A’’level MPDS codes that received BLS care. We hypothesized that many MPDS codes included patients who received BLS care more than 90% of the time. We additionally hypothesized that calls assigned to certain ‘‘A’’-level codes frequently received advanced life support (ALS)-level prehospital care.

METHODS Study Setting and Population This study took place in Rochester, New York, an urban northeastern U.S. city with 219,000 residents. All emergency calls were answered by the county-operated, tax-supported 911 call center. This call center has been recognized by the National Academy of Emergency Medical Dispatch as a Center of Excellence.22 The 911 call center assigned each call a dispatch code by using the MPDS system (version 11). For patients assigned certain MPDS codes, first responders, provided by the fire department, were activated. For all patients, the dispatch information was transmitted electronically to the contracted city EMS provider. An ambulance was then dispatched to the location of the request for aid. EMS was provided by Rural Metro Medical Services, a Commission for Accreditation of Ambulance Services (CAAS)-certified agency that staffed an all-ALS system. Each year, approximately 50,000 requests for EMS assistance are made through the 911 call center (Chiumento S, personal communication February 1, 2003). Each ambulance was staffed with at least one emergency medical technician (EMT)–paramedic or EMT–critical care (a level of training recognized by New York State that includes less didactic training than the EMT–paramedic curriculum, but allows similar level of clinical activity), allowing each unit to provide ALS evaluations and, if needed, ALS-level care to all patients. All transported patients were taken to one of four hospital emergency departments within the region. EMS had to respond to all requests for assistance using emergency lights and sirens during the study period. Additionally, paramedic-initiated refusals of transport or transports to alternative sites were not allowed according to local protocols. Rural Metro Medical Services used a computerassisted dispatch system that automatically recorded extensive data regarding each request for EMS assis-

tance. For each call, the system recorded the dispatch code, the EMS dispatch time, and the level of prehospital care provided to the patient (ALS vs. BLS). Additionally, on arrival at the scene, departure from the scene, and arrival at the hospital, the EMTs reported to dispatch so that these times were also recorded in the computer dispatch system.

Study Design and Protocol This retrospective cohort study examined all subjects who requested EMS care through the 911 dispatch center in Rochester, New York, between August 1, 2001, and April 1, 2002. Subjects were excluded if the EMS call did not result in a transport to an emergency department; if the call was assigned a local (nonMPDS) dispatch code; or if the call was cancelled, a duplicate response, or reassigned to another unit. This study was approved by the University of Rochester Research Subject Review Board. We defined a low-acuity patient before initiation of this study as a subject who required only BLS care as defined by the regional protocols. Thus, all patients receiving ALS care were considered high-acuity. In this region, any patient who received pulse oximetry measurement, fingerstick blood glucose measurement, defibrillation, any medication, or placement of an intravenous catheter was considered to have received ALS care. We defined a low-acuity dispatch code as an MPDS code in which at least 90% of patients received BLS care. As a result of clinical similarities between MPDS ‘‘A’’level subcodes, these codes were combined. For example, animal bites would be categorized as 3A1 if they occurred in a nondangerous body area, but would be categorized as 3A2 if they were not recent and 3A3 if they were superficial. In this study, all of these codes were combined into one code group, 3A, and then analyzed together. Codes at other levels that were not felt to be so clinically similar did not need to be grouped.

Data Management and Analysis All data were abstracted from the computer dispatch records, including the dispatch code assigned, the level of care provided, and the final disposition of the patient. These data were then evaluated in a Microsoft Access database and statistical analysis was performed using Microsoft Excel (Redmond, WA) and Stata 7.0 (College Station, TX). Before data analysis, MPDS codes or ‘‘A’’-level code groups with less than 25 calls transported were eliminated from analysis as a result of the small number of calls. Then, for each dispatch code, the authors performed descriptive statistics to determine the proportion of calls that received BLS care. The 95%

436

PREHOSPITAL EMERGENCY CARE OCTOBER / DECEMBER 2003

confidence interval (CI) was calculated for each code or code group. We also analyzed all of the groupings of the ‘‘A’’-level codes to characterize the proportion of calls that received BLS care. The 95% CI was calculated for each group with at least 25 calls transported.

RESULTS During the study time period, a total of 30,075 calls for EMS assistance were received by the 911 call center. Figure 1 shows the data set and the number of patients in each exclusion category. After applying the exclusion criteria, 19,332 calls remained for analysis. These calls were assigned to a total of 118 dispatch codes and code groups that had at least 25 calls. Twenty-eight MPDS codes and groups met the study’s a priori criteria as low-acuity dispatch codes. These codes included a total of 7,801 patients, or 39% of the total included calls. Of these patient calls, 7,394 (94.8%, 95% CI = 94.3%–95.3%) received BLS care. Table 1 shows the identified low-acuity codes and the number of patients assigned to each code who received BLS care. It is interesting to note that although the majority of the codes are ‘‘A’’-level codes, many codes from other MPDS levels did meet criteria. Table 2 combines the low-acuity codes based on clinical chief complaint, perhaps a clinically simpler and more logical method of viewing the dispatch information. Table 3 lists each ‘‘A’’-level MPDS code group and the number of patients assigned to each code grouping who received BLS care. The five codes that do not meet our definition of low acuity are shown in boldface type.

DISCUSSION Low-acuity Dispatch Codes The 911 dispatcher is the first member of the EMS system to interact with a patient requesting emergency

FIGURE 1. Derivation of the study set. EMD = emergency medical dispatch.

VOLUME 7 / NUMBER 4

assistance. An effective, accurate, and consistent system that can help the dispatcher identify low-acuity patients is critical for the efficient and safe operation of the EMS system because it can allow the appropriate delivery of medical resources within an appropriate time period. This study shows that it was possible, using the MPDS system, a nationally recognized dispatch center, and a CAAS-certified EMS agency, to identify low-acuity (not needing ALS-level care) patients from the initial 911 call. This study significantly builds on previously completed research. Studies performed using other dispatch protocols have shown that rigorous protocols can differentiate between patients requiring ALS and BLS care.2,16,20 Studies performed using older versions of the MPDS protocols have shown that the MPDS system can identify patients experiencing cardiac arrest with a sensitivity of 90% and specificity of 50%,19 that the MPDS system can better assign the proper level of EMS response required,17 and that assignment to a level ‘‘A’’ MPDS code reduces the probability of the patient being critically ill.21 However, unlike these previous studies, our study identifies individual dispatch codes that were associated with low-acuity patients and defined a low-acuity patient as one needing only BLS care. The combination of our study with these previous studies shows that MPDS dispatch codes do have discriminative ability to identify low-acuity patients. The potential system-wide benefits of using such a prioritization protocol are tremendous. First, because these low-acuity codes include a sizable fraction of all 911 calls requesting aid, the impact on emergency responses will be significant. Second, because these individuals are found to have lower-acuity illnesses, system managers can argue to regulatory and public agencies that responding to these calls using emergency lights and sirens is unnecessary because the additional time saved will not help the patient but will put the public at additional risk.8–11 Third, EMS systems that are organized in two tiers, with ALS and BLS responding separately, can use these codes to determine which calls initially need only a BLS response. However, having a paramedic present for certain high-risk categories that do not usually require ALS care may be safer for the community, thus limiting the number of codes receiving BLS response. Finally, and potentially most interesting, these codes, when used in conjunction with other criteria such as a physical examination, may have value in identifying patients who can be safely transported to a medical care facility by alternate means or who can be safely transported to facilities other than an emergency department for care.14 However, these codes need to be validated before systemwide use.

Shah et al.

437

LOW-ACUITY DISPATCH CODES

TABLE 1. Performance of Low-acuity Medical Priority Dispatch System (MPDS) Codes Dispatch Code

1A 1C2 1C3 1C4 4A 4B1A 4B3A 4D3A 4D4A 5A 7A 16A 17A 18A 18C4 21A 24A 24B1 24B2 24D3 24D5 25A 25B1 25B2 25B4 26A 30A 30B1

Descriptor

Remarks

Abdominal pain/problems Abdominal pain/problems Abdominal pain/problems Abdominal pain/problems Assault/sexual assault Assault/sexual assault Assault/sexual assault Assault/sexual assault Assault/sexual assault Back pain Burns (scalds)/explosion Eye problems/injuries Falls Headache Headache Hemorrhage/lacerations Pregnancy/childbirth/miscarriage Pregnancy/childbirth/miscarriage Pregnancy/childbirth/miscarriage Pregnancy/childbirth/miscarriage Pregnancy/childbirth/miscarriage Psychiatric/abnormal behavior Psychiatric/abnormal behavior Psychiatric/abnormal behavior Psychiatric/abnormal behavior Sick person Traumatic injuries, specific Traumatic injuries, specific

Abdominal pain Females with fainting or near fainting, aged 12–50 yr Males with pain above navel aged $35 yr Females with pain above navel aged $45 yr Possibly dangerous body area Unknown status Abnormal breathing Dangerous body area

Sudden onset of severe pain (,3 hours) First trimester hemorrhage or miscarriage Labor (delivery not imminent, .5 months) Unknown status Imminent delivery (.5 months) High-risk complications Nonviolent and nonsuicidal (alert) Violent Threatening suicide Unknown status No priority symptoms Possibly dangerous body area

Included Calls

Percent BLS

95% Confidence Interval

633 69 95 120 146 297 92 27 44 318 39 31 547 71 52 288 92 103 37 276 101 605 485 830 88 1,828 354 133

94.9 95.7 93.7 90.8 99.3 94.9 91.3 96.3 90.9 95.9 92.3 93.5 92.9 93 90.3 95.5 95.7 97.1 97.3 96.4 91.1 97.4 98.1 96.9 96.6 91.9 96.9 94.7

92.9–96.5 87.8–99.1 86.8–97.6 84.2–95.3 96.2–100 91.8–97.1 83.6–96.2 81–99.9 78.3–97.5 93.1–97.8 79.1–98.4 78.6–99.2 90.3–94.9 84.3–97.7 79–96.8 92.4–97.6 89.2–98.8 91.7–99.4 85.8–99.9 93.4–98.4 89.8–95.8 95.7–98.5 96.5–99.1 95.4–97.9 90.4–99.3 90.6–93.1 94.5–98.4 89.5–97.9

BLS = basic life support.

MPDS Level ‘‘A’’ Codes In the MPDS system, the ‘‘A’’-level dispatch codes are considered to identify the patients needing the least aggressive response.13 This would seem to imply that ‘‘A’’-level codes identify the lowest-acuity patients. A cursory review of the codes would lend support to this notion. For example, 10A1 (chest pain, breathing normally, age , 35) seems less critical than 10C4 (chest pain, breathing normally, age $35). It may be true that within each priority complaint (protocol card), the ‘‘A’’level codes identify the lowest-acuity patient as compared with, for example, the ‘‘C’’-level codes.

However, this study shows that all patients assigned to different ‘‘A’’-level dispatch codes do not necessarily need the same level of resources and are not equally healthy. Thus, this study supports the comments by Clawson and Dernocoeur cautioning that the levels do not linearly relate to patient acuity.13 In fact, we found that dispatch codes for diabetic problems, chest pain, and allergies were associated with a high ALS utilization rate in our system (Table 3).

LIMITATIONS

AND

FUTURE DIRECTIONS

This study had a number of limitations that must be considered. First, because it was a retrospective study,

TABLE 2. Low-acuity Medical Priority Dispatch System (MPDS) Codes Grouped by Clinical Complaint Complaint

Dispatch Codes

Included Patients

BLS

95% Confidence Interval

Abdominal pain Assault/sexual assault Back pain Burns Eye problems Falls Headache Pregnancy-related Psychiatric Sick person Trauma/hemorrhage

1A, 1C2–4 4A, 4B1, 4B3, 4D3, 4D4 5A 7A 16A 17A 18A, 18C4 24A, 24B1, 24B2, 24D3, 24D5 25A, 25B1, 25B2, 25B4 26A 21A, 30A, 30B1

917 606 318 39 31 547 123 609 2,008 1,828 775

94.3 95.2 95.9 92.3 93.5 92.9 91.9 95.6 97.4 91.9 96

92.6–95.7 93.2–96.7 93.1–97.8 79.1–98.4 78.6–99.2 90.3–94.9 85.6–96 93.6–97.1 96.6–98 90.6–93.1 94.4–97.3

BLS = basic life support.

438

PREHOSPITAL EMERGENCY CARE OCTOBER / DECEMBER 2003

VOLUME 7 / NUMBER 4

TABLE 3. Performance of ‘‘A’’-level Medical Priority Dispatch System (MPDS) Codes* Dispatch Code

1A 2A 4A 5A 7A 10A 12A 13A 16A 17A 18A 21A 24A 25A 26A 30A 31A

Descriptor

Calls Included

BLS

95% Confidence Interval

Abdominal pain/problems Allergies (reactions)/envenomations Assault/sexual assault Back pain (nontraumatic or nonrecent) Burns (scalds)/explosion Chest pain Convulsions/seizure Diabetic problems Eye problems/injuries Falls Headache Hemorrhage/lacerations Pregnancy/childbirth/miscarriage Psychiatric/abnormal behavior Sick person Traumatic injuries, specific Unconscious, now alert

633 40 146 318 39 158 144 125 31 547 71 288 92 605 1,828 354 96

94.9 70 99.3 95.9 92.3 86.1 54.2 52.8 93.5 92.9 93 95.5 95.7 97.4 91.9 96.9 68.8

92.9–96.5 53.5–83.4 96.2–100 93.1–97.8 79.1–98.4 80–91.1 45.7–62.5 43.7–61.8 78.6–99.2 90.3–94.9 84.3–97.7 92.4–97.6 89.2–98.8 95.7–98.5 90.6–93.1 94.5–98.4 58.5–77.8

BLS = basic life support. *Dispatch codes with fewer than 25 calls were excluded and are not presented. Those shown in boldface type are the five that did not meet the authors’ definition of ‘‘low acuity.’’

we were dependent on data recorded and were unable to evaluate the accuracy of the data. Second, we used a proxy, the provision of BLS level care, to identify lowacuity patients. We were unable to further evaluate the extent of ALS care provided and we did not use any direct outcomes data after the EMS care was delivered. Third, our results may not be generalizable because our EMS dispatch center is certified as a Center of Excellence. Other regions, without certified dispatch centers, may not find the same results. Finally, we do not know the accuracy of the paramedic evaluation. Patients requiring ALS care may have received only BLS care and, conversely, patients receiving ALS care may have required only BLS care. However, we believe that the prehospital bias is to aggressively care for patients, meaning patients are more likely to receive ALS care when it is not needed than BLS care when ALS is needed. Thus, our results are likely to be conservative in identifying low-acuity dispatch codes. The most important future study that must be performed is a prospective validation study to confirm that these low-acuity MPDS dispatch codes are truly associated with low-acuity patients. Furthermore, this study needs to include an evaluation of the ALS care delivered to determine the impact of the care provided. A second study that is also needed is to evaluate using these codes, in combination with patient historical and clinical characteristics, to perform field triage and identify patients who can safely be transported to an emergency department by a vehicle other than an ambulance, transported to an alternative destination, such as an urgent care center, or even left at the scene to follow up with their primary care physicians. Hopefully, with the guidance of dispatch codes, we can develop rigid protocols that can reduce the undertriage

rate identified in other studies that allowed paramedics to determine the need for treatment and transport.23

CONCLUSIONS Certain MPDS codes were associated with the delivery of BLS level care, indicating identification of patients likely to be low acuity. MPDS ‘‘A’’-level codes, although seemingly representative of low-acuity patients, do not necessarily identify low-acuity patients. Future studies are needed to prospectively validate that these codes do represent low-acuity patients. Ultimately, these dispatch codes have the potential to improve the ability of paramedics to determine whether a patient needs transportation, the mode of transportation, and the patient destination. The authors thank the Monroe County Emergency Communications Center and Mike Kuder, AEMT-P, for assisting with this study.

References 1. Roush WR (ed). Principles of EMS Systems. Washington, DC: American College of Emergency Physicians, 1994. 2. Palumbo L, Kubincanek J, Emerman C, et al. Performance of a system to determine EMS dispatch priorities. Am J Emerg Med. 1996;14:388-90. 3. Narad RA, Driesbock KR. Regulation of ambulance response times in California. Prehosp Emerg Care. 1999;3:131-5. 4. Gibson G. Measures of emergency ambulance effectiveness: unmet need and inappropriate use. J Am Coll Emerg Physicians. 1977;6:389-92. 5. Kost S, Arruda J. Appropriateness of ambulance transportation to a suburban pediatric emergency department. Prehosp Emerg Care. 1999;3:187-90. 6. Richards JR, Ferrall SJ. Inappropriate use of emergency medical services transport: comparison of provider and patient perspectives. Acad Emerg Med. 1999;6:14-20. 7. Sosnin M, Young D, et al. A study of emergency ambulance utilisation. Aust Fam Physician. 1989;18:233-4.

Shah et al.

439

LOW-ACUITY DISPATCH CODES

16. Curka PA, Pepe PE, Ginger VF, et al. Emergency medical services priority dispatch. Ann Emerg Med. 1993;22:1688-95. 17. Bailey ED, O’Connor RE, Ross RW. The use of emergency medical dispatch protocols to reduce the number of inappropriate scene responses made my advanced life support personnel. Prehosp Emerg Care. 2000;4:186-9. 18. Neely KW, Eldurkar JA, Drake MER. Do emergency medical services dispatch nature and severity codes agree with paramedic field findings? Acad Emerg Med. 2000;7:174-80. 19. Klassen G. The use of priority medical dispatch to distinguish between high- and low-risk patients [abstract]. Ann Emerg Med. 1990;19:458-9. 20. Slovis CM, Carruth TB, Seitz WJ, et al. A priority dispatch system for emergency medical services. Ann Emerg Med. 1985;14:105560. 21. Glasser J. Reliability of the medical priority dispatch system in a large urban center [abstract]. J Emerg Med. 2001;20:321-2. 22. National Academies of Emergency Dispatch. Academy accredited centers. Available at: http://www.naemd.org/scripts/ db.cgi?db=centers&uid=default&view_records=1&sb=2&ID=*. Accessed February 1, 2003. 23. Schmidt T, Atcheson R, Federiuk C, et al. Evaluation of protocols allowing emergency medical technicians to determine need for treatment and transport. Acad Emerg Med. 2000;7:663-9.

8. Brown LH, Whitney CL, Hunt RC, et al. Do warning lights and sirens reduce ambulance response times? Prehosp Emerg Care. 2000;4:70-4. 9. Hunt RC, Brown LH, Cabinum ES, et al. Is ambulance transport time with lights and siren faster than without? Ann Emerg Med. 1995;25:507-11. 10. Ho J, Lindquist M. Time saved with the use of emergency warning lights and siren while responding to requests for emergency medical aid in a rural environment. Prehosp Emerg Care. 2001;5:159-62. 11. O’Brien DJ, Price TG, Adams P. The effectiveness of lights and siren use during ambulance transport by paramedics. Prehosp Emerg Care. 1999;3:127-30. 12. Davis R. Speeding to the rescue can have deadly results. USA Today. 2002; Apr 4. Available at: www.usatoday.com/news/ nation/2002/03/21/usat-ambulance(acov).htm. Accessed June 24, 2002. 13. Clawson JJ, Dernocoeur KB. Principles of Emergency Medical Dispatch, 2nd ed. Salt Lake City, UT: Priority Press, 1998. 14. Dunford JV. Emergency medical dispatch. Emerg Med Clin N Am. 2002;20:859-75. 15. Wilson S, Cooke M, Morrell R, et al. A systematic review of the evidence supporting the use of priority dispatch of ambulances. Prehosp Emerg Care. 2002;6:42-9.

d

-In MemoriamPeter Safar, MD Distinguished Professor of Resuscitation Medicine The worlds of emergency medicine, critical care medicine, and anesthesia lost a great leader, innovator, teacher, friend, and colleague on August 3, 2003, when Peter Safar died in his home from complications of cancer. His loving wife Eva and son Paul were at his side. His legacy will endure. He will be missed.