As our EMS systems become awash in data, it’s important to use this information to create clinical and operational wisdom. We hope to demonstrate how to measure commonly available data that will allow the development of unique operational changes; to show the utility of merging dispatch data with prehospital clinical records; to reveal the optimal method of presenting this data; and to discuss real-world operational changes that have been made using this linked clinical data.
Emergency medical dispatch (EMD) is a system that:
- Categorizes and prioritizes emergency calls;
- Predicts patients who require rapid care;
- Has a goal to provide appropriate and timely prehospital response; and
- Allows local effectiveness to be measured when linked with electronic patient care records.
Almost all mature EMS organizations use some form of EMD. The system categorizes 9-1-1 calls—by analyzing things like the likelihood of the patient requiring a time-dependent treatment such as defibrillation or the need for an ALS crew—in order to send a quick and appropriate response.
Prior studies in differing EMS configurations have used a variety of EMD programs with both health- and non-health-trained dispatchers as well as different clinical measures to gauge success. The overall consensus, however, is that organizations using EMD usually have improved response times to urgent calls and a better ability to predict needed levels of service.1–15
Medical Priority Dispatch System
Some large cities use their own locally developed EMD system; only a few communities use a system developed by the Association of Public Safety Communication Officers. The Medical Priority Dispatch System (MPDS) is a proprietary EMD system used by 71% of major U.S. cities and commonly used internationally.17 Because of its prevalence in EMS, it’s been studied the most.
With the MPDS, 9-1-1 callers are asked a series of scripted questions that include the patient’s level of consciousness, age, chief complaint and other complaint-specific questions. In the days before computers were widely available, each call would be assigned a card determining the level of care needed. Today, most dispatch centers use a computer program, which can also document the dispatcher’s decisions for a robust quality improvement process.
There are 33 standardized, complaint-based categories a 9-1-1 call can fall into, which are further generally classified as Alpha (BLS cold), Bravo (BLS hot), Charlie (ALS cold), Delta (ALS hot), or Echo (ALS hot with AED support). Calls may be further assigned a numerical subgroup and a modifier, which provide responders with more specific details. Together, the numerical protocol, priority (Alpha through Echo), subgroup and modifier (when present) make up the MPDS category.
For example, a call may be assigned to the MPDS category 12D3E. The number 12 is the complaint-based category for seizure, D (or Delta) represents priority, 3 is a subcategory that informs prehospital providers that the patient has irregular breathing, and E is a modifier that indicates the patient has a history of epilepsy.
EMD data allows you to see if there are certain lower priority subgroups that require more interventions than you would expect.
Computer-aided dispatch (CAD) systems are general computer programs that can be paired with any EMD system. They usually record information about each call, including date, time and location of call; dispatch time; dispatch code; and disposition (e.g. “transported code 2”). Disposition codes are assigned when the on-scene unit has updated the call status. Most CAD systems generate a unique number for that run and another number used to identify the patient care record (PCR), which allows for matching between these two data sets.
This combination of EMD, CAD and electronic PCRs can give our industry an unprecedented opportunity to measure the effectiveness of local EMD to predict the need for time-dependent care. The data allows you to see if there are certain lower priority (Alpha) subgroups that require more interventions than you would expect for calls with a relatively long response time (or BLS-only response in some systems). In our system, we noted 27 Bravo calls (e.g. penetrating trauma calls from a third party) had a high rate of non-transport but still had an unacceptably high rate of needing critical interventions. The data can also demonstrate other categories, such as man down or falls, that have truly low rates of lifesaving interventions and should safely remain a low-priority response despite that one case you heard about last week at the fire station.
Sporer has been the medical director for three distinct EMS systems in the Bay Area and noticed significant local differences between them. For example, there was a large difference in the rate of Echo respiratory distress between two counties as well as a significant difference in the Man Down categories. The rate of benzodiazepine administration to 12As ranged from 2–7% in differing counties. This variation leads us to conclude it would be prudent to use your local experience to make any substantive decisions about your
After years of attempting to make clinical sense of this linked CAD and PCR information, we have developed an optimal list of the information we believe is needed to assess the function of the EMD system in any jurisdiction.15,17–21 (See Table 1, below.) When analyzing data to create this list, we decided not to analyze any categories that have fewer than 100 calls in a year. We also chose to include morphine in the list of critical interventions. This decision is highly debatable and an important one because it’s the most common medication given and it might be reasonable to assign morphine its own special category. We didn’t include some common procedures such as the placement of IV lines, pulse oximetry or the performance of blood glucose measurement, because they’re so commonly performed and are a poor proxy for illness.22
We’ve used this data-driven, decision-making process to help fine-tune our individual systems. In both San Mateo and San Francisco Counties, we had multiple complaints from providers about low-priority calls to seizure patients. Instead of having anecdote drive our response patterns, we examined those calls that were coded as 12A1, seizure stopped and breathing verified. We found that among these patients, 9% ultimately received midazolam for a recurrent or persistent seizure. Our committee thought this was a time-dependent treatment and that this percentage was too high for a non-code 3 response. However, EMD data showed it was best to keep 12A1 calls as low priority despite provider complaints. This same category in Alameda County had a much lower rate of midazolam administration (3%) and has kept it at a non-lights and siren response.
Figures 1a and 1b
Alameda County posts all EMD statistics on its website, www.alcoems.org/mpds-data. The above screenshots show data for 26C calls.
In San Francisco, approximately 65% of EMS calls were sent with a lights-and-sirens response. The San Francisco Fire Department (SFFD) repeatedly defended its practice from a variety of neighborhood groups concerned about noise pollution. SFFD examined categories that were commonly used, were currently a code 3 response and were found to have a low rate of time sensitive ALS interventions such as CPR, advanced airway, or defibrillation. Some categories such as Diabetic Problem (13C1–C3) were felt to not be appropriate for downgrading and falls were maintained due to manpower issues (17D3–D5). It was agreed to change the response to a variety of categories such as abdominal pain, headache, sick person, heart problem, and man down resulting in a 15% drop in lights and siren calls in our community. In particular, the category of man down, vital signs unknown (32D1) was dispatched several times a day as a possible resuscitation with several apparatus. Our analysis demonstrated a rate of cardiac arrest of three out of a thousand such patients (0.003%) and we were able to change this to a non lights and siren response.
Alameda County has recently undertaken an extensive evaluation of our experience with two years of MPDS from our two Centers of Excellence. We have undertaken the unprecedented move to post our results on our website, www.alcoems.org/mpds-data, for all of our providers and our community to review. We’re hoping to demystify some of the issues that exist in any system about the function of any EMD system. This will be helpful for the ambulance medics, who are generally skeptical of the dispatch data that they receive each day. It’ll be useful to the engine medic who’s convinced the sick patient who was mis-triaged by dispatch last month isn’t a systemic problem.
It becomes clear very quickly that the system works reasonably well but it is far from perfect. There’s always a role for eyes on the patients, boots on the ground.
After publicizing this data locally, we plan on developing a working group to decide on the optimal EMS response for our system. Many of our local fire departments are currently overwhelmed with call volume and this will allow our group to see if there are categories that can safely wait for their transporting ambulance without an engine-first response.
This working group may also identify groups of patients with a small rate of critical interventions that could safely be sent without lights and siren, making all of our other citizens a little bit safer on the roads. We may also find that some categories would benefit from an engine-only response in categories with a high cancellation rate and low incidence of time-sensitive ALS interventions.
Our current EMS system generates most of the lights and siren responses in our cities and these are for commonly perceived but not real threats to life. It behooves us to use the data generated by our EMD systems to use our ambulances and our paramedics in the smartest and safest way possible. This resourceful use of local data will allow us to tailor specific responses to our citizens in the safest and appropriate manner.
1. Palumbo L, Kubincanek J, Emerman C, et al. Performance of a system to determine EMS dispatch priorities. Am J Emerg Med. 1996;14(4):388–390.
2. Neely KW, Eldurkar JA, Drake ME. Do emergency medical services dispatch nature and severity codes agree with paramedic field findings? Acad J Emerg Med. 2000;7(2):174–180.
3. Feldman MJ, Verbeek PR, Lyons DG, et al. Comparison of the medical priority dispatch system to an out-of-hospital patient acuity score. Acad J Emerg Med. 2006;13(9):954–960.
4. Craig A, Schwartz B, Feldman M. Development of evidence-based dispatch response plans to optimize ALS paramedic response in an urban EMS system (abstract). Prehosp Emerg Care. 2006;10:114.
5. Neely KW, Eldurkar J, Drake ME. Can current EMS dispatch protocols identify layperson-reported sentinel conditions? Prehosp Emerg Care. 2000;4(3):238–244.
6. Calle P, Houbrechts H, Lagaert L, et al. How to evaluate an emergency medical dispatch system: A Belgian perspective. Eur J Emerg Med. 1995;2(3):128–135.
7. Clawson J, Olola C, Heward A, et al. Cardiac arrest predictability in seizure patients based on emergency medical dispatcher identification of previous seizure or epilepsy history. Resuscitation. 2007;75(2):298–304.
8. Clawson J, Olola C, Heward A, et al. The Medical Priority Dispatch System’s ability to predict cardiac arrest outcomes and high acuity pre-hospital alerts in chest pain patients presenting to 9-9-9. Resuscitation. 2008;78(3):298–306.
9. Bailey ED, O’Connor RE, Ross RW. The use of emergency medical dispatch protocols to reduce the number of inappropriate scene responses made by advanced life support personnel. Prehosp Emerg Care. 2000;4(2):186–189.
10. Flynn J, Archer F, Morgans A. Sensitivity and specificity of the Medical Priority Dispatch System in detecting cardiac arrest emergency calls in Melbourne. Prehosp Disaster Med. 2006;21(2):72–76.
11. Shah MN, Bishop P, Lerner EB, et al. Derivation of emergency medical services dispatch codes associated with low-acuity patients. Prehosp Emerg Care. 2003;7(4):434–439.
12 Myers JB, Hinchey P, Zalkin J, et al. EMS dispatch triage criteria can accurately identify patients without high-acuity illness or injury. Prehosp Emerg Care. 2007;11(1):42–48.
13. Shah MN, Bishop P, Lerner EB, et al. Validation of using EMD dispatch codes to identify low-acuity patients. Prehosp Emerg Care. 2005;9(1):24–31.
14. Michael GE, Sporer KA. Validation of low-acuity emergency medical services dispatch codes. Prehosp Emerg Care. 2005;9(4):429–433.
15. Sporer KA, Youngblood GM, Rodriguez RM. The ability of emergency medical dispatch codes of medical complaints to predict ALS prehospital interventions. Prehosp Emerg Care. 2007;11(2):192–198.
16. Cady G. (1999) The medical priority dispatch system—A system and product overview. National Association of Emergency Dispatch. Retrieved Aug. 24, 2011, from www.emergencydispatch.org/articles/ArticleMPDS(Cady).html.
17. Sporer KA, Craig AM, Johnson NJ, et al. Does emergency medical dispatch priority predict delphi process-derived levels of prehospital intervention? Prehosp Disaster Med. 2010;25(4):309–317.
18. Sporer KA, Johnson NJ. Detailed analysis of prehospital interventions in medical priority dispatch system determinants. West J Emerg Med. 2011;12(1):19–29.
19. Sporer KA, Johnson NJ, Yeh CC, et al. Can emergency medical dispatch codes predict prehospital interventions for common 9-1-1 call types? Prehosp Emerg Care. 2008;12(4):470–478.
20. Johnson NJ, Sporer KA. How many emergency dispatches occurred per cardiac arrest? Resuscitation. 2010;81(11):1499–1504.
21. Hodell EM, Sporer KA, Brown JF. Which emergency medical dispatch codes predict high prehospital nontransport rates in an urban community? Prehosp Emerg Care. 2014;18(1):28–34.
22. Kuzma K, Sporer KA, Michael GE, et al. When are prehospital intravenous catheters used for treatment? J Emerg Med. 2009;36(4):357–362.