Original Contribution

Prehospital intervention probability score:

a Novel method for determining necessity of emergency medical service units^?

Denise Livingston MD, PhD ^a, Andreia Marques-Baptista MD ^a, Richard Brown MBS ^b,

Junfeng Liu PhD ^c, Mark A. Merlin DO, EMT-P ^d,?

^aDepartment of Emergency Medicine, University of Medicine and Dentistry of New Jersey, Graduate School of Biological Sciences, Piscataway, New Jersey 08901

^bUniversity of Medicine and Dentistry of New Jersey, Graduate School of Biomedical Sciences at the Osteopathic School of Medicine, Stratford, New Jersey 08084

^cBiostatistics and Emergency Medicine, University of Medicine and Dentistry of New Jersey-School of Public Health, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, New Brunswick, New Jersey 08901 ^dEmergency Medicine and Pediatrics, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, New Brunswick, New Jersey 08901

Received 25 December 2008; revised 3 February 2009; accepted 3 February 2009

Abstract

Objective: This article models use of emergency medical services (EMS) within a defined geographical area. Our goal was to develop an original quantitative method to delineate the need for EMS units within a geographical population.

Methods: Use of the EMS system within 11 municipalities was analyzed in 2007. The geospatial distributions of interventions during this year were examined, as well as the population dynamics of the region. A statistical model to determine the probability of an individual within the call area requiring an intervention was proposed using weighted population statistics and the application of an intervention probability.

Results: The observed interventional probability increased exponentially with age, notably after the age of 75. Areas with higher proportions of elderly residents had substantially higher rates of intervention and EMS use. Municipality H had the largest age-group of 20 to 24 years with an intervention probability of 0.34% (95% confidence interval [CI], 0.24-0.44), their more than 85-year age-group also had the largest intervention probability of 19.54% (95% CI, 15.60-23.48).

Conclusions: Contrary to current practice patterns of placing Paramedic units in regions of greatest population density, we established a formula based on population vs intervention probability. We found the actual numbers of interventions performed are not dependent solely on population size but also are

^? This study received no grants or financial support. It has not been presented at any meeting or previously published. No conflict of Interest exists or copyright constraints with any author.

* Corresponding author. UMDNJ-Robert Wood Johnson Medical School, Department of Emergency Medicine, New Brunswick, NJ 08901. Tel.: +1 732

235 8717.

E-mail address: [email protected] (M.A. Merlin).

0735-6757/$ - see front matter (C) 2010 doi:10.1016/j.ajem.2009.02.002

affected by the age of the population being served. This is particularly relevant to growing elderly communities. This determination will aid in the disbursement of limited preHospital resources in regions by improving availability of EMS personnel.

(C) 2010

Introduction

Background

The decision to place an emergency medical services (EMS) unit within a town or geographic region is traditionally based upon population density information [1]. This practice has been called into question, and a more reliable method has been sought [2,3]. The addition of EMS units is often based upon best patient care practices, financial reimbursement, and demand from regional administrators [4,5]. The geographical approach to EMS localization has been most commonly based on population distributions, with the assumption that greater population will produce greater dispatch volume. Some EMS systems use a dynamic approach of system status management that changes with time of day, day of the week, or time of year [6,7]. This involves deployment of ambulances based upon historic demand patterns. Other EMS systems consider Response times into the equation [8]. Unfortunately, no single approach has been uniformly successful in providing consistent coverage, and thus, further study is needed in this area [9-11].

Table 1 Potential LSI procedures determining PIP score

Importance

The best method for determining ambulance need while providing evidence-based prehospital care remains elusive [12-15]. Operational concerns such as minimizing total fleet mileage, crew safety from exposure to emergency mode driving, EMS skills retention, and Job satisfaction are addressed by optimizing placement of EMS units [11]. In areas of rapid growth, the long range planning of loci for EMS units and possible colocation with other public safety resources intensify the need for a more refined parameter to better situate components of EMS systems [15].

Goals of this investigation

The goal of this investigation is to estimate where units should be placed based on past use of the EMS system and the estimated probability of where an intervention will need to be performed. We describe a new prehospital intervention probability (PIP) score to help determine best practice methods for supplying a region with an EMS unit. Furthermore, the investigators suggest a statistical model that can be applied to multiple types of EMS systems and different levels of Prehospital care providers. To this end, the term lifesaving intervention unit (LSIU) is used instead of Advanced life support or basic life support (BLS) so that various EMS systems can determine themselves what

constitutes a lifesaving intervention (LSI) within their EMS system (Table 1). From this PIP score, we evaluate the placement of LSI units and assess the need for future placement. This approach can then be applied by EMS system administrators to determine the placement of lifesaving

Procedures

12-lead electrocardiogram Cardioversion central venous access chest decompression Combitube continuous positive airway pressure Defibrillation External pacemaker Intraosseous Intubation, nasal intubation, oral Intravenous access-infusion Intravenous access-lock Military Anti-shock Trousers inflation nasogastric tube insertion Medications

Adenosine (Adenocard) Lidocaine bolus Astellas, Deerfield, IL, USA Amiodarone (Cordarone) Lidocaine infusion bolus Wyeth-Ayerst, Philadelphia, PA, USA Amiodarone (Cordarone) Lorazepam (Ativan) infusion Wyeth-Ayerst Philadelphia, PA, USA Aspirin (acetylsalicylic acid) magnesium sulfate (MgSO4) 81mg atropine sulfate Methylprednisolone (SoluMedrol) Pfizer New York, NY, USA Calcium chloride Metoclopramide (Reglan) Wyeth-Ayerst Philadelphia, PA, USA Dextrose 50% Metoprolol (Lopressor) Hospira, Lake Forest, IL, USA Diazepam (Valium) Midazolam (Versed) Roche Roche Laboratories, Laboratories, Nutley, NJ, USA Nutley, NJ, USA Diltiazem (Cardizem) morphine sulfate (MSO4) Hospira Lake Forest, IL, USA

(continued on next page)

Diphenhydramine (Benadryl) Ortho-McNeil, Raritan, NJ, USA

Procedures

Table 1 (continued)

Naloxone (Narcan) Endo Pharmaceuticals, Chaddds Ford, PA, USA

Setting

We evaluated 11 municipalities of one county within the primary coverage area of our EMS system. Our 2-tiered system is composed of 6 ALS units housed at four locations that contain either 2 paramedics or 1 paramedic and a prehospital

Dopamine Nitroglycerine IV (Tridil)

Schwarz Pharma, Mequon, WI, USA

Epinephrine 1:1000 Nitroglycerine paste

Epinephrine 1:10 000 Nitroglycerine tablets

registered nurse. The county population of approximately 800 000 residents is made up of 68.4% white, 13.9% Asian, 13.6% Hispanic (of any race), and 9.1% African American residents [16]. The county occupies 323 square miles with a combina-

Etomidate (Amidate) Hospira, Lake Forest, IL, USA

Fentanyl (Duragesic/ Sublimaze) Janssen Pharma, Titusville, NJ, USA

Furosemide (Lasix) Sanofi-Aventis, Bridgewater, NJ, USA

Procainamide HCl (Pronestyl) Bristol-Myers-Squibb Princeton, NJ, USA

Albuterol (Proventil) GlaxoSmithKline Research Triange Park, NC, USA Sodium bicarbonate

tion of urban cities with a high-crime region, suburban

Glucagon Succinylcholine Sandoz Pharma, Princeton, NJ, USA (Anectine) Sandoz, Princeton, NJ, USA

Haloperidol (Haldol) Ortho-McNeil, Raritan, NJ, USA

Ipatropium bromide (Atrovent) Boehringer Ingelheim Ridgefield, CT, USA

Ketamine

Labetalol HCl (Normodyne) Schering-Plough Kenilworth, NJ, USA

Terbutaline (Brethine) Novartis Pharma Princeton, NJ, USA

Vecuronium (Norcuron) Organon, West Orange, NJ, USA

resources such as ALS units or publicly accessible automatic external defibrillators.

Methods

This article examines the probability that certain age-groups within a population will require more frequent potential LSIs by EMS units. We hypothesize that an older population has an increased need for advanced prehospital services. This concept should govern the geographical placement of LSIU opposed to their placement in areas with the greatest overall population density. The LSIU and PIP score was derived from the experienced opinion of the authors.

This study, approved by the institutional review board, recorded and analyzed all LSIs from January 1, 2007, to December 31, 2007. Interventions were defined as the use of one or more of 55 procedures or medications in the prehospital setting administered by paramedics Table 1. These procedures include all medications and any physical intervention including intravenous access, airway access or maintenance, needle decompression, and intraosseous access. Each procedure was weighted equally.

Fig. 1 PIP score derivation through universal intervention probabilities.

Fig. 2 Geospacial Map of Region. (A) life saving Intervention plotted by age. (B) Median Age by Census Block. (C) LSI Call Locations by mean age census data. (D) Magnified view of concentrated data.

2.4. Primary data analysis

Our model used 2 specific parameters as follows: weighted census data and the probability of an individual requiring an intervention during the year. A regression model was designed with predicted age and response intervention probability (IP) based on the 12 pairs of age intervals provided in the census data. The predicted IP for each municipality was the estimated total number of patients who required an LSIU divided by total population size within this town, given the universal IP P_j (age- group “j”) that was from county-level population; the IP estimation for municipality “i” is as follows:

j = 1

i,j

j

j = 1

X12

j = 1

i,j

n / X12

n p = X12

w p ,

j

j

Fig. 3 Population sizes vs age-groups based upon municipalities.

communities with multiple nursing home, and rural farmlands A large University exists within the county.

Methods of measurement

Patients in this study were broken down into 12 age- groups based on the intervals given for the 2000 US census. For each patient who received an intervention, their age was recorded, and the total number of patients that received interventions in the corresponding age-group was documen- ted. We analyzed the total number of interventions in each age-group to determine an overall LSI probability for each of the 12 Age subgroups.

Data collection and processing

An investigator trained in Microsoft Access and the Emergency Department Information Management database collected all data. Patient care reports generated on paper during the course of patient interactions were transferred to a Microsoft Access database after completion. All EMS patient care reports are managed similarly in our region. A standardized data dictionary was used for our data elements. Accuracy of data abstraction and entry was assessed by 3 of the authors individually. From this Access database of EMS responses, a query was performed to list all patients who received interventions in 2007. If a patient received 10 interventions, it was only counted as one because higher acuity patients may need more procedures.

Geospatial relationships and figures were plotted using ArcGis 9.2 software, using Census 2000 Tiger/line data (ESRI, Redlands, Calif). Statistical analysis was carried out using SAS

9.1 TS level 1M0, XP_PRO platform (SAS Institute Inc, Cary, NC) and Minitab 15 (Minitab Inc, State College, Pa).

which was a weighted average of age-group specific universal intervention probabilities where n_ij is the population size for age-group “j” within municipality “i” (Fig. 1). We created a PIP score range of 0 to 5 to simplify the usefulness of these IP scores. The formula uses the weighted population size (w) based upon each population interval multiplied by the IP of a single individual (p).

Results

Four thousand five hundred eighty-five LSIs were performed during the study period. The region studied covered a large geographic area including urban, suburban, and Rural settings. A geospatial map of the region studied is shown in Fig. 2. Each LSI, coded for age, was plotted (panel A). Few EMS calls in this region were covered by

Fig. 4 Directly observed intervention probabilities vs age-groups based upon municipalities.

other EMS providers. Median age of each census block, as reported by the 2000 census, was mapped in groupings of

20 years. The LSI call locations were then shown with respect to the census median age data (panel C). Location of current ALS units within our system was also included. A magnified view of one area of concentration of LSIs is shown in panel D. This geospatial representation of the data shows that median age of the population increases within areas of increased LSI calls.

Census data for the 11 municipalities studied were examined, and the distribution of ages per municipality is shown in Fig. 3. Most of the towns studied had their population within the 35 to 45 age range. However, municipality H had a peak in population in the 25 to 35 age range. This could be accounted for by a large university located within this municipality. Another outlier to the trend is municipality E, which has its population peak within the 75 to 85 age range. This difference could be explained by the large number of age-restricted communities located within this area.

The observed IP for each municipality was calculated and graphed (Fig. 4). The observed intervention probabilities for each municipality remain stable until the age-group of 75 to 84, where they increase exponentially. In all municipalities studied, the IP significantly increases from age 85 (Table 2). This increase in IP was seen regardless of the population size, suggesting that age is an independent but critical marker for use of the EMS system.

To examine our hypothesis that the number of interven- tions performed is not solely dependent on the population size, we compare, in further detail, 2 municipalities, H and E. In municipality H, the total population consists of 48 573 persons and received 644 interventions during 2007. The greatest proportion of the population in this municipality is ages 20 to 24, with a population size of 12 304. This age- group makes up 25.33% of the population. This municipa- lity’s elderly population (N65 years old) is composed of 3146 people or 6.5% of its total residents.

In contrast, municipality E with a total population of 27 999 had 1350 interventions. Municipality E’s largest population group falls within the 65 to 84 years of age bracket, with the 65 to 74 years of age population making up 19.67% of the total population and the 75 to 84 years of age-group are comprising 18.92% of the population. Municipality E has almost half the population of munici- pality H, but LSIs are needed more than twice as often as the municipality with a larger population.

In examining census data for the area covered by our LSIUs, we have found that in the county as a whole, most of the population is within the 30 to 40 age range. The greatest proportion of the population in the one major urban area covered includes residents in the 20 to 30 age range. Reasons for this include a high migrant population and the presence of a large university with undergraduate and graduate students living in the region. However, when the interven- tions for 1 year were analyzed, these 20-to-30-year-olds were

not the bulk of the population but the elderly who were using the system with greater frequency. This trend held true in both urban and suburban towns within our purview, despite the inclusion of college students and persons of Lower socioeconomic status.

When using our formula to calculate the IP, the largest IP of 22.27% (95% confidence interval [CI], 23.77-28.40) is seen in municipality E for people 85 years or older. The largest age-group in municipality E was 65 to 74 years old with a predicted probability of 21.38% (95% CI, 18.78- 23.99). Although municipality H’s largest age-group is 20 to 24 years with an IP of 0.34% (95% CI, 0.24-0.44), their 85 and older age-group also has the largest IP within the region of 19.54% (95% CI, 15.60-23.48). In less populated areas such as municipalities J and K, it is similar with the highest IPs among the residents older than 85 years. In municipality J, this age-group’s IP is 9.62% (95% CI, 6.03- 13.20). In this same age-group, municipality K’s IP is 9.40% (95% CI, 4.11-14.69). Municipality C also demon- strates its highest IP in its eldest group with 12.27% (95% CI, 7.94-16.61).

Intervention probability results in a PIP score for each individual municipality (Table 3). The observed probability represents the statistical analysis based on interventions and weighted census data as previously described. This break- down by age results in confirmation that elderly populations have an increased IP. We assigned a cutoff number for the 5-point PIP score of 3 to signify a need for an additional unit in the area. In our study, only municipality H, which has an IP percentage of 4.82 and a subsequent PIP score of 3, qualified for an additional individual LSIU.

Limitations

This study uses the most recent census data and the most recent LSI data in our county; however, they are not from the same year. The most recent population data available are from the 2000 census, whereas the LSI studied is from 2007. There is the potential that population shifts have occurred for the proceeding 7 years that are not being accounted for or are being improperly represented. This limitation was also seen in the geospatial analysis of the data where the most recent line data available are from 2000. Therefore, some interventions were not represented in our geospatial study because new locations built after 2000 could not be included in our geospatial mapping analysis. A further limitation in the mapping analysis was that some locations were recorded incorrectly by the LSI personnel at the time of the call. These errors in recording were not addressed at the time of collection and therefore are unable to be corrected in the future.

Another confounding factor in the analysis of these data involves cross coverage when LSI units from our EMS system are not available. In these instances, calls

Table 2 Directly estimated IP (%) with 95% CI from observed data (municipalities A-K)
	Age-group	Municipality
		A	B	C	D	E
	1 (0-4)	0.58 (0.30-0.86)	1.61 (0.00-3.83)	0.36 (0.00-0.76)	0.22 (0.00-0.65)	0.72 (0.22-1.22)
	2 (5-9)	0.14 (0.02-0.27)	0.89 (0.00-2.64)	0.11 (0.00-0.33)	0.23 (0.00-0.69)	0.17 (0.00-0.41)
	3 (10-14)	0.13 (0.02-0.25)	0.85 (0.00-2.50)	0.24 (0.00-0.57)	0.52 (0.00-1.23)	0.16 (0.00-0.38)
	4 (15-19)	0.72 (0.42-1.02)	1.14 (0.00-3.35)	0.84 (0.17-1.52)	0.89 (0.00-1.89)	0.70 (0.27-1.13)
	5 (20-24)	0.55 (0.23-0.88)	2.53 (0.00-6.00)	0.10 (0.00-0.30)	0.31 (0.00-0.90)	1.57 (0.65-2.49)
	6 (25-34)	0.41 (0.24-0.59)	0.85 (0.00-1.81)	0.07 (0.00-0.17)	0.10 (0.00-0.29)	1.09 (0.60-1.57)
	7 (35-44)	0.51 (0.36-0.66)	0.22 (0.00-0.65)	0.63 (0.31-0.95)	1.52 (0.80-2.24)	1.60 (1.14-2.07)
	8 (45-54)	1.23 (0.99-1.48)	1.65 (0.05-3.25)	0.90 (0.46-1.34)	2.60 (1.53-3.67)	2.09 (1.55-2.63)
	9 (55-64)	2.56 (2.12-3.01)	2.10 (0.00-4.45)	1.76 (0.98-2.55)	5.09 (3.06-7.11)	4.18 (3.45-4.90)
	10 (65-74)	3.10 (2.49-3.71)	2.90 (0.00-6.86)	1.66 (0.77-2.56)	3.64 (1.53-5.76)	3.70 (3.21-4.20)
	11 (75-84)	9.30 (7.97-10.62)	6.25 (0.00-14.64)	5.68 (3.93-7.43)	6.17 (3.04-9.30)	9.65 (8.85-10.44)
	12 (85-)	24.17 (20.34-28.00)	33.33 (6.66-60.01)	12.27 (7.94-16.61)	9.40 (4.11-14.69)	26.09 (23.77-28.40)

from one system are taken by LSI units of other systems and would therefore not be recorded in our data set. As a result, there were calls that occurred during 2007 in the studied region that were not recorded by our EMS system although they occurred in our coverage area. To address this, only municipalities in which more than 95% of the calls from the town were taken by our units were included in our analysis.

Finally, a theoretical limitation to this study is that it examines the number of interventions by LSI providers not the total Number of calls for LSIU providers. This model was chosen because it removed the variable of times where LSIU were called, but not needed, and were either cancelled before arriving on scene or cancelled at the scene. However, in using this approach, the model misses instances where EMS providers are needed but are not performing an intervention before arrival at the hospital. This “grab-and-go” philosophy versus remaining on the scene to perform interventions, may delay more important surgical intervention; however, this is rarely observed in our system secondary to large distances from the 3 available state level I trauma centers. Therefore, we believe this is a minor limitation and does not significantly affect our analysis.

In designing the PIP score, another theoretical limitation encountered in this study was the unique geospatial limitations of the area studied and the system in which our LSIUs operate. Every area has its own unique traffic patterns, town distributions, and geospatial limitations. Therefore, the PIP score can only be used to examine IP on an individual municipality basis. One must then examine a municipality within the unique geographic area in which it is located to determine if placement in that town or between it and an adjacent town would be most optimal. This calculation must also take into account the inherent limitations within the EMS system of the area such as total number of units available or governance of the system on a regional or other type of system.

Discussion

The placement of LSIUs by EMS administrators is a complex task, attempting to distribute precious resources while maintaining excellence in patient care. Traditional placement of LSIUs has focused mainly on geographical placement in areas of high population concentration [2,3]. We suggest, however, that a more efficient and cost-effective use of prehospital resources should take into account the history of LSIU use while attempting to predict the future needs. In doing so, we suggest that other variables such as age of the population being served should be considered. We, therefore, have devised a PIP score that identifies the likelihood of a person’s need for an LSI. We suggest that this PIP score be factored into the future placement of LSIU for more effective use of prehospital resources.

We chose a very broad definition of LSI on purpose so individual EMS system administrators could determine for themselves which procedures should be included. We felt this was important because EMS systems within the United States are so diverse. For example, some systems may include albuterol as a BLS skill, whereas other ALS systems may not. Specifically, we decided to include intravenous access in our LSI for this discussion; however, other systems will probably decide to exclude this in their definition. Our decision was based on our type of EMS system that is 100% on line. Because our paramedics speak to a physician while treating patients, they rarely place intravenous fluids in patients who are discharged. Subsequently, we included intravenous fluids in our LSI definition because it was felt by the physician and paramedic that the patient warranted the procedure rather than just because they were being transported by ALS.

We specifically examined 11 municipalities within our system with less than 5% cross coverage from neighboring systems and examined their age distribution vs call volume. We found that despite varying town size and composition, use of the EMS system increased proportionally with age. From

F	G	H	I	J	K
0.52 (0.00-1.23)	0.53 (0.24-0.82)	0.09 (0.00-0.19)	0.30 (0.10-0.49)	0.30 (0.00-0.63)	0.64 (0.00-1.35)
0.42 (0.00-1.01)	0.42 (0.16-0.68)	0.11 (0.00-0.23)	0.30 (0.12-0.49)	0.00 (0.00-0.00)	0.00 (0.00-0.00)
0.41 (0.00-0.97)	0.18 (0.00-0.35)	0.23 (0.03-0.43)	0.24 (0.06-0.41)	0.32 (0.00-0.68)	0.21 (0.00-0.62)
0.77 (0.00-1.64)	0.68 (0.32-1.03)	0.50 (0.31-0.68)	0.62 (0.29-0.96)	0.44 (0.01-0.88)	0.88 (0.02-1.74)
1.47 (0.19-2.76)	0.93 (0.53-1.34)	0.34 (0.24-0.44)	0.78 (0.34-1.21)	0.42 (0.01-0.84)	0.52 (0.00-1.25)
0.73 (0.15-1.31)	0.44 (0.28-0.60)	0.62 (0.45-0.79)	0.40 (0.24-0.57)	0.36 (0.13-0.60)	0.10 (0.00-0.29)
1.05 (0.48-1.62)	0.92 (0.69-1.15)	1.56 (1.22-1.90)	0.68 (0.50-0.86)	0.95 (0.58-1.32)	0.30 (0.01-0.59)
1.53 (0.79-2.27)	1.30 (0.99-1.61)	2.49 (1.97-3.01)	2.11 (1.72-2.49)	1.65 (1.08-2.22)	1.38 (0.68-2.07)
2.10 (1.05-3.16)	2.51 (1.95-3.06)	4.30 (3.42-5.18)	3.09 (2.45-3.72)	3.01 (2.06-3.97)	2.26 (1.20-3.32)
2.14 (0.94-3.33)	5.20 (4.23-6.17)	6.02 (4.84-7.21)	3.65 (2.76-4.55)	4.96 (3.65-6.28)	2.42 (1.31-3.52)
9.64 (6.80-12.48)	8.12 (6.62-9.63)	7.25 (5.79-8.71)	10.47 (8.36-12.57)	6.07 (4.53-7.61)	6.11 (4.03-8.20)
15.70 (9.22-22.19)	17.38 (13.28-21.48)	19.54 (15.60-23.48)	25.40 (20.02-30.77)	9.62 (6.03-13.20)	9.40 (4.11-14.69)

these data, a PIP score was devised to determine the likelihood of an individual within a given area using the EMS system. Although it is intuitive to think that the elderly are more likely to use the medical system, current EMS placement considera- tions are focused more on population rather than on age.

Municipality

Furthermore, we have shown that patients requiring LSIs are not uniformly integrated into the population within our region. There are concentrated areas within our county that contain a disproportionate amount of elder patients. These areas are not well staffed by our LSIUs because they are not within large population areas. Placement of these adult communities suggests a clustering of aging adults in age- restricted adult communities in more rural areas. Our data show that individuals living in these communities have a statistically higher likelihood of using the EMS system. The growing risk and concern for this trend is that the current population-based placement of LSIUs does not account for clustering of potential patients in this manner. The result of understaffing in this type of area leads to a great strain on the prehospital system, with units traveling greater distances for calls and an increasing probability of missed calls due to

Municipality	Probability (%)	PIP score
A	1.55	2
B	1.42	2
C	1.02	2
D	1.78	2
E	4.82	5
F	1.93	2
G	1.54	2
H	1.33	2
I	1.43	2
J	1.63	2
K	1.36	2

unavailability. With the increasing popularity of the age- restricted adult communities and the aging of the baby boomer generation, we suggest that allocation of LSIU resources should closely monitor shifts in location of this particular population and that our PIP scoring system allows for such monitoring.

This study of our prehospital system suggests a simplified formula for determining the necessity of an LSI unit. It allows individual administrators to determine for themselves what a critical LSI is and allows multiple levels of providers to be used in this formulation. If prospectively validated in other EMS systems, this method can be incorporated into other geospatial modeling systems that include additional factors such as response time to optimize patient outcomes. Further study on the use of prehospital resources is needed and should include predictive planning analyses of all Prehospital personnel, including BLS and air transport providers. Together, these methods can provide cost-efficient powerful tools to enhance patient care and optimize patient outcomes. Further validation of this model needs to be done in various EMS systems.

References

Table 3 Model predicted IP and observed IP comparison

1. Peleg K, Pliskin JS. A geographic information system simulation model of EMS: reducing Ambulance response time. Am J Emerg Med 2004;22:164-70.
2. Su S, Shih CL. Modeling an emergency medical services system using Computer simulation. Int J Med Inform 2003;72:57-72.
3. Su S, Shih CL. Resource reallocation in an emergency medical service system using computer simulation.
4. Edwards DP, Ornato JP, Overton JL, et al. Geotemporal demand analysis further improves system status management in high-performance emergency medical services system. Ann Emerg Med 2001;34 [Abstract].
5. Lerner EB, Nichol G, Spaite DW, et al. A comprehensive framework for determining the cost of an emergency medical services. Ann Emerg Med 2007;3:304-13.
6. Baker JR, Clayton JR, Taylor BW. A non-linear multi-criteria programming approach for determining county emergency medical service ambulance allocations. J Oper Res Soc 1989;5:423-32.
7. Overton J. Reimbursement in emergency medical services: how to adapt in a changing environment. Prehosp Emerg Care 2002;1:137-40.
8. Davis J, Kuhns R, Watson DJ. Evaluating emergency medical services: controlling the rising cost of saving lives. J Health Jum Serv Adm 2004;4:485-512.
9. Bledsoe BE, Dean S, Post C. System status management lacks scientific proof. The origins of system status management. Emergency Medical Services 2004;33:32, 38, 40.
10. Scout J, Pepe PE, Mosesso VN. All-advanced life support vs. tiered- response ambulance systems. Prehosp Emerg Care 2000;4:1-6.
11. Billittier AJ, Lerner EB, Moscati RM, et al. Triage, transportation, and destination decision by Out-of-hospital emergency care providers. Prehosp Disaster Med 1998;13:98-103.
12. Carter AJ, Grierson R. The impact of Ambulance diversion on EMS resource availability. Prehosp Emerg Care 2007;11: 421-6.
13. Patel PB, Derlet RW, Vinson DR, et al. Ambulance diversion reduction; the Sacramento solution. Am J Emerg Med 2006;24:206-13.
14. Cone DC, Davidson SJ, Nguyen Q. A time motion study of emergency medical services turnaround interval. Ann Emerg Med 1998;31:241-6.
15. Fischer AJ, O’Halloran P, Littlejohns P, et al. Ambulance economics. J Public Health Med 2000;22:413-21.
16. U.S. Census Bureau. Profile of general demographic characteristics: 2000. Data set: Census 2000 Summary File 1 (SF 1) 100-percent data Geographic Area: Middlesex County (NJ). Available at http:// factfinder.census.gov/servlet/QTTable?_bm=n&_lang=en&qr_name= DEC_2000_SF1_U_DP1&ds_name=DEC_2000_SF1_U&geo_id= 05000US34023. Accessed June 5, 2008.