1

TITLE: Analysis of EMS Triage and Dispatch Errors by Registered Nurses in Italy 1

SHORT RUNNING TITLE: “Underestimation of medical emergency calls in Italy” 2

3

Abstract 4

Major elements of an effective Emergency Medical Services (EMS) system include a single telephone access 5

number, accurate assessment of the urgency of the health problem and timely dispatch of appropriate 6

personnel and equipment. In Italy, EMS calls are managed by Emergency Operations Centers (EOC). 7

Registered Nurses who have received specialized education in this function. The nurses determine the 8

criticality of the situations and assign an EMS response priority level identified by a color code, ranging 9

from red (very critical) to green (not critical). At times, the severity of a situation may be underestimated, 10

resulting in assignment of a lower EMS response priority, and the potential for patient death (code black). 11

Purpose: The purpose of this study was to identify the RN undertriage of EMS calls subsequently found to 12

be associated with fatality, termed “green-black code” cases. 13

Methods: We carried out a retrospective qualitative analysis of the telephone conversations utilizing Fele’s 14

conversation analysis method. We studied a sample of calls occurring in 2011 and we compared those calls 15

with the green-black cases occurred. Call characteristics were compared with the population of all EMS calls 16

during the study period. 17

Results: Study patients were older, with a mean age of 81.6 years. The callers were individuals calling on 18

behalf of the patients, rather than the patients themselves. Callers reported symptoms that were not life-19

threatening. Nurse operators did not always inquire about the vital signs as required by the Medical Priority 20

Dispatch System Protocol (MPDS). The phone conversations were considerably shorter than normal (54.26 21

vs. 65 seconds). 22

Discussion: Although the importance of dispatch system protocols is well-known, it is also important that 23

nurse triage operators have proper training in order to assure that major parameters such as vital signs and 24

symptomatology are obtained, and to reduce caller stress level. 25

26

2

Introduction 27

Efforts to improve health care outcomes while fostering cost containment through appropriate use of 28

resources have resulted in a proliferation in the provision of telephone assessment and consultation services 29

by Registered Nurses in a variety of settings. Handling a telephone call which may involve a request for 30

emergency care require substantial expertise. The operator is expected to quickly recognize the severity of 31

the medical condition, identify the possible etiology, and determine the resources required while reducing the 32

caller anxiety or aggressiveness in order to obtain their collaboration1. The operator should use time 33

efficiently, collecting only the necessary information without prolonging the phone call. Operators require 34

proficiency in effective communication skills in order to collect all the relevant data. The subsequent 35

decision-making process should lead to the best response (the right emergency care mobile resource, the 36

right staff, at the right time). In Italy Registered Nurses are responsible for both the triage of patients arriving 37

at the emergency department and the telephone EMS dispatch system. 38

Italian law specifies at least six months seniority in emergency nursing in order to perform medical triage. In 39

addition, Emergency Operating Centre (EOC), RN operators are emergency nurses with at least two years 40

seniority in emergency department and expertise in pre-hospital care. 41

At the time of the study, Italy did not have a codified system of data collection and analysis specifically 42

designed for pre-hospital emergencies2, although efforts have been made to standardize the collection of pre-43

hospital data in selected conditions, such as severe trauma and cardiac arrest3,4

. A recent survey by the Italian 44

Ministry of Health2 shows that EOC operators use a dual mode (color and alphanumeric) code system to 45

classify both the criticality and the severity of an emergency call (see Table. 1). The code assigned to the 46

prevailing pathology is the second most important information for emergency dispatchers. When a patient 47

has co-morbidities, the operator assigns a code that refers to the most relevant symptoms (Tab. 2). Finally, 48

the location code indicates where the event took place (Tab. 2). 49

Determining the Appropriate Response Resources 50

Emergency dispatching is a dynamic decision making process5, as well as the most important activity 51

performed by EOCs. It consists of four phases: taking incoming calls, instructing callers, dispatching the 52

appropriate EMS resources and instructing the ambulance crew. Appropriate conversation techniques enable 53

3

the operator to obtain collaboration from the caller. In addition, the use of a standardized interview protocol 54

allows for the collection of all relevant details while avoiding conflicts with the caller 6,7

. 55

Collaboration depends on three caller variables: emotional status, lack of knowledge, of the situation and 56

general behavior when reporting an emergency. Theoretically, the emotional status of the caller does not 57

really impact on collaboration, since a well-trained operator can guide a scared or angry caller using specific 58

interrogation techniques, as well as a calm voice. Another Italian study confirms that only four per cent of 59

callers are annoyed or irritated8. The caller may be the patient themselves (first-party caller), a person in the 60

patient’s direct vicinity (second-party caller), or a person who is not with the patient but is reporting from 61

some distance (third-party caller). About 55% of phone calls in Italy are made by first or second-party 62

callers9 who, if correctly interviewed, may provide all the relevant information. That is why the use of a 63

standardized interview protocol, together with an appropriate training on how to lead a succinct telephone 64

conversation is so important. When an emergency occurs, the caller has a distorted time perception. For this 65

reason, it is important that an emergency medical call is answered by at least the third ring, although such a 66

response time – about 12 seconds – would still prove rather lengthy for an emergency call10

. 67

In case of life-threatening situations such as sudden cardiac arrest, electrocution, drowning, and 68

suffocation11

, the telephone conversation should last less than one minute. The mobilization time (from call 69

end to EMS vehicle dispatch) generally varies from 75 to 90 seconds. This time interval is considered as part 70

of the standard ambulance response time in Italy, which is eight minutes for urban areas or life-threatening 71

emergencies, and twenty minutes for extra-urban areas such as isolated places. (Health Ministry Decree 15 72

May 1992). 73

Characteristics of an Effective Telephone Encounter 74

What are the main features of satisfactory communication when taking an emergency medical call? Although 75

research has increased in recent years, literature concerning the qualitative aspects of telephone dispatching 76

is still rather poor. Several studies appear to lack methodological rigor12

, with the 5-level priority system 77

used in France, Canada, the USA, the UK and Australia being more consensus based than evidence based13

. 78

Several studies focus on the efficacy of a single telephone number for all the emergency services, such the 9-79

11 in the United States or the 1-1-2, introduced in the European Community in 2001. 80

4

With the exception of the Province of Varese (Northern Italy) where a single emergency telephone number 81

has been available since 2010, Italy has a different emergency number for each of the various emergency 82

services (fire, medical, and police emergency). 83

Data from this experience show that EOC operators are able to screen about half of all calls, which ensures 84

that emergency services (police, fire department, EMTS) are activated only when needed14

. One Italian 85

study15

shows that EOC operators tend to overestimate the criticality and severity codes as compared to the 86

assessment performed by the emergency medical team on the scene. Several scene responder identified green 87

codes appeared to be overestimated as yellow, while red codes were not always properly identified. 88

In regard to telephone conversation content analysis, qualitative research is oriented toward defining the 89

local organizational structure1, while sociolinguistic research is more oriented toward analyzing telephone 90

conversation between people serving a formal organization/institution16

. A study analyzing “special 91

conversations” shows that intervention in emergency situations consists of five steps: dispatchers take the 92

incoming call and identify themselves, the caller asks for help, a brief conversation takes place, the caller’s 93

request is responded to, and the process is completed17

. 94

A systematic review of the literature concerning telephone triage performed by emergency medical system 95

operators identified 326 studies18

. The overall quality of these studies is modest, with only a few studies 96

seeming to consistently support the use of specific criteria to identify medical priorities in order to improve 97

the patient’s outcomes. Sometimes calls are not appropriate, e.g., patients that do not meet the specific 98

criteria for an immediate ambulance response. It seems that the accuracy of emergency response procedures 99

improves when operators assign a priority code using a Medical Priority Dispatch System (MPDS) protocol. 100

Statistically, an assessment utilizing the MPDS protocol is more concordant with the criticality code assigned 101

by the on-scene emergency medical team18,19

. The use of a MPDS protocol also seems to reduce emergency 102

medical technicians’ response time in the most critical cases20, 21, 22

. Despite the use of standard protocols, 103

EMS dispatch assessment errors (under triage) still occur, especially in complex situations involving patients 104

with multiple or chronic diseases. Underestimating an emergency situation may be significant, such as when 105

an emergency medical team, arriving on the scene of what they were told to be a green code (not particularly 106

critical), finds that the patient had died in the meantime. Such occurrences, dubbed “green-black code” cases 107

5

by Italian emergency health workers, are considered sentinel events. This study aims to detect, through a 108

retrospective qualitative analysis of the telephone conversations, the factors associated with undertriage. 109

Methods 110

We carried out a retrospective study on a sample of 839 EMS calls occurred during 2011 at the EOC of 111

Lecce – Italy. Permission to access the database of the EOC of Lecce was obtained from the General 112

Administrative Management of the EOC. The sample was randomly extracted using a table of random 113

numbers from a population of 62,392 EMS calls (<5%, IC 99%). We compared the characteristics of the 114

sample telephone conversations with the 15 green-black calls occurred in 2011. The phone calls were 115

analyzed with the aim of collecting information on conversational features. The following variables were 116

considered: details of the event, telephone conversation, EMS resource dispatched, and patient data. 117

Specific data analyzed for each sections include: 118

Details of the event: time interval from call receipt until the appropriate mobile care resource was 119

physically en route to the emergency, the dispatcher, the final criticality/priority assessment, and other data. 120

Telephone conversation: the criticality code assigned while interrogating the caller is the product of 121

a decision-making process that can be based on dispatchers’ experience or on application of the MPDS 122

protocol. This code is communicated to the ambulance crew immediately after the call. Conversations were 123

analyzed using the qualitative and phenomenological techniques implemented by Fele in his conceptual 124

model23

. Fele is an Italian sociologist who conducted several investigations in the field of the emergency 125

services (police, fire department, EMTS). His “Conversational Analysis Method” is specifically applied to 126

study institution-centered interactions, such as those occurring in EOCs. 127

EMS Care resources: information regarding the type of resource dispatched, and it’s response time. 128

In Italy the volunteers staffed vehicles provide basic life support and early defibrillation (BLSD), while 129

vehicle with physician and RN provide advanced life support (ALS). 130

Patient’s data: the data contained in this section permits to analyze the concordance between the 131

emergency code initially assigned by dispatchers and the medical team’s assessment on their arrival on the 132

scene. 133

Using Fele’s model, we analyzed the 15 telephone conversations focusing on 4 crucial aspects: incoming call 134

6

receipt (operators identify the service), problem description (operators collect all relevant data from the 135

caller), problem identification (operators assign a criticality code), and call ending (operators reassure the 136

caller and revise the information collected from them). Statistical analysis: data were exported to a Microsoft 137

Access database version 10.00 and analyzed using SPSS Version 17.0 for Windows. We performed a 138

factorial analysis considering all the variables of a dispatch. 139

In addition, the qualitative-quantitative variables of the “green-black code” cases were examined. 140

Results 141

The EOC is equipped with an Ericsson MD-110 PBX that routes emergency calls to operators (there are 142

generally six dispatchers per shift). During 2011, operators received 62,392 emergency phone calls, from 143

which we randomly extracted a sample of 839 calls. The number of incoming emergency calls per operator 144

while dispatchers were already handling phone calls varied from 2 to 47 (mean: 23.41, SD: 11.12). 145

Although the proficiency level of operators when creating a dispatch depends on several variables, a 146

regression analysis performed on shows that the variable “operator” is not strictly related to the whole 147

“green-black code” phenomenon. The linear regression analysis (R2 test) shows that the variable “operator” 148

accounts for only 14% of the entire process. 149

Dispatch creation time: The overall time to generate a dispatch has been calculated using a linear model. 150

The mean time is about 65 seconds (CI 99%: 61-69 sec.).The average duration of telephone conversations is 151

more or less the same in both the sample (range 43.8 – 64.8 sec.) and the “green-black code” subgroup 152

(range 61.5 – 69.4 sec.). 153

The average time necessary for sending the appropriate care mobile resource and complete the dispatch is 95 154

seconds (CI 99%: 84- 10 sec.). Phone calls generally last less than 120 seconds (82.5%) and, in some cases, 155

less than 60 seconds (43.3%). 156

Completeness of Data Collection: Operators failed to record all the details required for computer assisted 157

dispatching in 85.7% of the calls. This missing information could enable a better post hoc analysis. Some 158

data seem to be less neglected than others, as in the case of cardiovascular patients additional information 159

was almost always present. 160

7

Event location: Most phone calls were about medical emergencies occurring at home (76.9%) or in the 161

street (14.1%) as showed in Table 2. 162

Prevailing pathology: The vast majority of calls for service were about medical emergencies occurring at 163

home (76.9%). Cardiovascular pathologies represented the most common emergency cases (29.9%), 164

followed by trauma, musculoskeletal problems (21.9%), and respiratory diseases (13.8%). No prevailing 165

pathology was present in 17% of the cases. (Tab. 2). 166

Criticality score assigned by the operator: in the observed sample the color codes assigned were: red 167

(6.9%), yellow (56%), green (36.8%), and white - not urgent at all (0.3%). 168

Concordance in Criticality Score: The EOC operators and scene responders agreed in only 31.6% of green 169

patients, and in 26.7% of the yellow patients. 170

The “green – black code” cases: during 2011, 15 dispatches out of 62,392 (1:4000) classified as non-171

urgent (green) by the EOC operators were associated with a patient death (black code). The analyses 172

performed in the general population were repeated for this particular subgroup, however, the confidence 173

interval (CI) was set at 95%. since the subgroup was smaller (Tab. 3). 174

In addition, telephone conversations were analyzed to evaluate the quality of communication and the 175

occurrence of wrong assessments during calls. The response time in the “green – black code” subgroup was 176

significantly shorter (54.26 seconds, CI 95%: 44 -65 sec.) when compared to the sample. Most phone calls 177

lasted less than 60 seconds (66%). In those under triaged cases (green code), operators had not filled in all 178

the MPDS major items (consciousness, breathing, circulation) on the dispatch form, and a BLS staffed 179

ambulance (volunteers crew) was sent instead of an ALS staffed ambulance. 180

When the volunteers with BLS training assessed the criticality of the situation, they called the EOC for 181

medical support, so that a second ambulance with a nurse and a physician was sent. The first ambulance 182

arrived within 8 minutes (standard time for urban areas) in 25% of cases. Because of the non-urgent 183

emergency code, the mean response time of the first ambulance was about 17 minutes (CI 95%: 6 - 29 min.) 184

The mean response time for rural areas was 18 minutes (CI 95%: 9 -26 min.), which means that the 20-185

minute standard response time for rural areas was generally accomplished (86%). When compared to the 186

8

mean age of all the patients treated (61.8 years), the patients in the subgroup were significantly older (mean 187

age 81.6 years, CI 95%: 77 -86), as showed in Table 3. 188

Qualitative assessment of telephone conversations: As proposed by the “Fele” method, we divided 189

conversations into four parts: incoming call receipt, problem description, problem identification, and call 190

ending. 191

INCOMING CALL RECEIPT: When taking an incoming call, operators should immediately identify the 192

service as well as themselves by saying "Emergency Operations Center 118, I am a nurse". Operators 193

generally do identify the service (86.7%). 194

PROBLEM DESCRIPTION: When taking emergency calls, operators ask for details according to a set of 195

standard questions. They may go through the whole list of questions (accurate interrogation), ask only a few 196

questions (inaccurate interrogation) or just listen (simple listening). The interview was accurate in 4 cases 197

(26.7%), inaccurate in 8 cases (53.3%), and with operators simply listening in 3 cases (20%). 198

PROBLEM IDENTIFICATION: In order to quickly obtain vital information about patient status and scene 199

conditions, the operator should act according to the MPDS protocols, asking the caller if the victim is alert 200

and breathing. Operators asked both questions (alert and breathing) in one case (6.7%), only one question 201

(alert or breathing) in 5 cases (33.3%), and did not ask this question in 9 cases (60%). 202

CALL ENDING: At the end of the conversation the operator reassured the caller telling them that an 203

ambulance was on its way in 13 cases (86.7%). In order to evaluate the accuracy of telephone conversations, 204

we assigned a score between zero (0%) and 6 (100%) to the described variables (Tab. 4). The mean score of 205

the 15 telephone conversations was 54.4% (CI 95%: 43.7 – 65.2). 206

Discussion 207

This study focused on the qualitative aspects of EOC RN operators activity with the aim of detecting the 208

possible causes of underestimation of severity of health emergencies subsequently associated with so-called 209

“green-black code” cases. The 15 calls associated with the “green-black code” cases were routed to the 210

operators by automatic switching equipment. This study has several limitations, primarily related to the lack 211

of a common reference model for measuring the effectiveness of the dispatch process at EOCs. The few 212

existing studies on this topic refer to different organizational settings and are specifically focused on 213

9

indicators such as dispatch accuracy, ambulance response times, and concordance between severity codes 214

assigned by dispatchers and the actual emergency scenario found by medical teams on their arrival. 215

Assuming that the “green-black code” cases are sentinel events, we compared the 15 fatalities with the 216

overall activity of the EOC considered in this study via a retrospective analysis. The aim was to detect 217

possible differences in the activity of operators (dispatch appropriateness and effectiveness), the 218

demographic characteristics of the victims, or the telephone conversations. Our data refer to a local EOC, 219

which means that results cannot be generalized. 220

The mean length of telephone conversations was 65 seconds. As regards dispatch initiation, the average 221

length of time was 95 seconds, with the limit of 120 seconds – a standard accepted by experts9 - generally 222

being respected (82.5%). At times, collecting data from the caller was difficult and time-consuming, 223

therefore the delay was not attributable to operators. Although the proficiency level of operators when 224

creating a dispatch depends on several variables, a regression analysis shows that the variable “operator” is 225

not strictly related to the whole “green-black code” phenomenon. 226

Because of the small sample size, it is not possible to establish a relationship between the duration of a 227

dispatch and the criticality code assigned. As regards the dispatch form, we observed that items referring to 228

symptoms and events (chest pain, trauma, car accident, sudden illness) were often left blank (85.7%). 229

Symptoms/events that are more likely to be reported are “chest pain” (5.6%) and “car accident” (4.7%), 230

perhaps due to the fact that such occurrences require the implementation of specific protocols such as acute 231

myocardial infarction and pre-hospital trauma care protocols. Operators generally tend to overestimate the 232

condition of patients reporting chest pain20

. The vast majority of calls for service are about medical 233

emergencies occurring at home (76.9%), and this consistent with national statistics. Given their prevalence21

, 234

cardio circulatory problems are predominant (29.9%) and are followed by musculoskeletal problems (21.9%) 235

and a number of “other pathologies” (17.4%). The average duration of telephone conversations was more or 236

less the same in both the sample (range 43.8 – 64.8 sec.) and the “green-black code” subgroup (range 61.5 – 237

69.4 sec.). All the 15 “green-black codes” emergency calls were made by a relative and in one case the caller 238

was a caregiver. When compared to the sample, the 15 victims in the subgroup were older (mean age was 239

81.6 vs. 61.8 y.). The presence of chronic conditions and the old age of the patients may have led the callers 240

10

to probably underestimate the severity of the situation. Dispatch was not accurate in 11 cases. Operators 241

never filled in the dispatch form when interrogating the callers over the phone, with vital signs (awareness, 242

breathing) being only partially assessed, if not entirely neglected. All the above-listed factors could have 243

contributed to under-triage. When analyzing telephone conversations utilizing the “Fele method”, it appears 244

that in all cases callers described a vague and generally not alarming situation (dysuria, constipation, leg pain 245

since three days). If the emergencies reported, described as non urgent, probably strongly influenced the 246

operators decision on the type of criticality code to be assigned, our analysis reveals that callers had 310 not 247

been interrogated accurately enough, either. Such underestimations inevitably resulted in the dispatch of an 248

ambulance equipped only with a volunteer-based crew to a “green code” scene (therefore neither emergency 249

lights, nor sirens were activated). Volunteers, after assessing the emergency scenario, called the EOC for 250

medical support so that precious time was wasted waiting for a second ALS staffed ambulance with a nurse 251

and a physician. 252

Conclusions 253

Even if “green black code” cases are rare (1/4000 interventions), they represent a sentinel event that indicate 254

a failure in the EMS system. Nurses working at EOCs usually assess patients’ conditions and make important 255

decisions with very limited time (60 – 120 seconds). The introduction of a specific training based on the 256

“Fele method” could improve both quality and effectiveness of telephone triage. The training could involve 257

simulations of EMS calls, as well as the analysis and discussion of real cases of under triage EMS calls. 258

From a methodological point of view, the results of this research could be useful to create a near miss and 259

adverse events database. 260

Further research is needed to better understand the “green-black code” phenomenon and to analyze the role 261

of communication in the assignation of an underestimated priority code. 262

Nursing implications 263

Triage is essential for the early recognition and treatment of the seriously ill patients, which reduces 264

morbidity and mortality. In Italy, telephone triage of EMS calls is done by experienced nurses with at least 265

two years seniority in emergency department. 266

11

Regular retraining may help the RN operators to avoid undertriage by focusing on the MPDS protocol. It is 267

important that nurses act in accord to the MPDS protocols, asking the caller about patient status and scene 268

condition, particularly when the caller reports not life-threatening situations, in order to assign the 269

appropriate EMS response priority level. 270

271

12

References 272

1. Psathas G. Conversation analysis: The study of talk–in-interaction. Qualitative Research Methods: Series 273

35. California: A Sage University Paper; 1999. p. 58-61 274

2. Ministero della salute. Ricognizione della normativa, 334 delle esperienze, delle sperimentazioni, 275

relativamente a emergenza e 118, a livello regionale e nazionale. Mattoni SSN – Mattone 11 – Pronto 276

Soccorso e Sistema 118. Regione Lazio; 2006. 277

3. Cummins RO, Chamberlain DA, Abramson NS, Allen M, Baskett P, Becker L et al. Recommended 278

guidelines for uniform reporting of data from out-of- hospital cardiac arrest: the Utstein Style. A statement 279

for health professionals from a task force of the American Heart Association, the European Resuscitation 280

Council, the Heart and Stroke Foundation of Canada, and the Australian Resuscitation Council. Circulation. 281

1991; 20(8). p. 960 - 75. 282

4. Dick WF, Baskett PJ. Recommendations for uniform reporting of data following major trauma-- the 283

Utstein style. A report of a working party of the International Trauma Anesthesia and Critical Care Society 284

(ITACCS). Resuscitation. 1999; 42 (2). p. 81 -100. 285

5. Leprohon J, Patel VL. Decision-making strategies for telephone triage in emergency medical services. 286

Medical Decision Making. 1995; 15(3): 240-53. 287

6. DeGroot HA. Patient classification systems and staffing. Part 1, Problems and promise. Journal of 288

Nursing Administration. 1994; 24(9):43-51 289

7. Forslund K. Operators experiences of emergency calls. Journal of Telemedicine and Telecare 2004; 290

10(5): 5290-297 291

8. Ghiselli G, Calabrò P, Bono D. I codici gialli nell‟interpretazioni delle chiamate. Roma: Emergency, 292

Editore Key Communication; 2003. p. 4. 293

9. Thakore S, McGugan EA, Morrison W. Emergency ambulance dispatch: is there a case for triage? Journal 294

of the Royal Society of Medicine. 2002; 95(3): 126-9. 295

10. Wilde ET. Do emergency medical system response times matter for health outcomes? Health Economics 296

201;3 22(7):790-806. 297

11. Focarile F. Indicatori di qualità nell’assistenza sanitaria. Torino: Centro Scientifico Torinese. 1998. 298

13

12. Harden RD. (1999) „Critical appraisal of papers describing 359 triage systems.‟, Academic Emergency 299

Medicine. 1999; 6(11): 1166-71. 300

13. van der Wulp I, van Stel HF. (2009) „Adjusting weighted kappa for severity of mistriage decreases 301

reported reliability of emergency department triage systems : a comparative study. Journal of Clinical 302

Epidemiology. 2009; 62(11): 1196-201. 303

14. Corbetta C. Numero unico di emergenza 112: la sperimentazione di Varese. I quaderni di AREU. 2010; 304

1. Milano: Itacacomunicazione. 305

15. Greggio M, Scapparone P, Costanza R, Leigheb F, Panella M. I codici di gravità nell‟emergenza 306

sanitaria territoriale: comparazione tra i codici attribuiti dalla Centrale Operativa 118 dell‟ASO di 307

Alessandria e quelli assegnati sul luogo dell’evento. XII Conferenza Nazionale di Sanità Pubblica. Roma: 308

Edizioni Iniziative Italiane; 2011. p.623. 309

16. PeräKylä A, Vehvilainen S. Conversation analysis and the professional stocks of interactional 310

knowledge. Discourse & Society. 2004; 14(6): 727-50. 311

17. Zimmerman, H. (1992) The interactional organization of calls for emergency, talk at work: interaction 312

in institutional settings. Cambridge: Cambridge University Press, pp. 418-69. 313

18. Slovis CM, Carruth TB, Seitz WJ, Thomas CM, Elsea WR. (2000) „A priority dispatch system for 314

emergency medical services.‟, Annals of Emergency Medicine. 2000; 14(11): 1055-6. 315

19. Wilson S, Cooke M, Morrell R, Bridge P, Allan T. A systematic review of the evidence supporting the 316

use of priority dispatch of emergency ambulances. Emergency Medicine Research Group. 2002; 6(1): 42-9. 317

20. Clawson J, Olola CH, Heward A, Scott G, Patterson B. Accuracy of emergency medical dispatchers' 318

subjective ability to identify when higher dispatch levels are warranted over a Medical Priority Dispatch 319

System automated protocol's recommended coding based on paramedic outcome 385 data. Emergency 320

Medical Journal. 2007; 24(8): 560-3. 321

21. Clawson J, Olola C, Heward A, Patterson B, Scott G. The Medical Priority Dispatch System's ability to 322

predict cardiac arrest outcomes and high acuity pre-hospital alerts in chest pain patients presenting to 9-9-9, 323

Resuscitation. 2008; 78(3): 298-306. 324

14

22. Curka PA, Pepe PE, Ginger VF, Sherrard RC, Ivy MV, Zachariah BS. A priority dispatch system for 325

emergency medical services. Annals of Emergency Service. 1985; 22(11): 1055-60 326

23. Fele G. L’analisi della conversazione. Bologna: Il Mulino; 2007. 327