“M10, this is a Delta response for a 25-year-old male who is complaining of feeling shaky after drinking 4 Red Bulls.”
Or is it the people who specify whether the response is Code 2 or 3 who should re-think their policies?
Ever wonder just how accurate dispatch information is? What dying person are we driving to with lights and sirens? I happen to have 232 calls to show you. This is a relatively small sample size, but the data collection is ongoing. By the time I get bored, I may have a thousand calls logged. We’ll talk about the details in a second.
Thanks to this neat and simple graph-making tool at the National Center for Education Statistics, here are some of the numbers expressed in pie charts. Yes, it is for kids. A simple tool for a simple fool.
In June 2011, I wrote in total disbelief about how the CDC and ACEP claimed that only 8 percent of ED patients were considered “non-urgent,” the lowest level of acuity under their classification, which by extension meant 92 percent of patients should be seen within 2 hours.
That’s a weird classification, to put it nicely.
That very week, a quick tally of 19 patients resulted in me (and most normal providers, I’m sure) subjectively labeling 12 of them – 63 percent – “non-urgent,” as defined by normal English usage.
Over the past several years, as more agencies adopt the ubiquitous MPDS, I have noticed that the overtriage* based on common response models is quite significant. For those of you who aren’t familiar with MPDS, in the interest of keeping this post short, its main goals are to sensibly dispatch the appropriate type and amount of responders (or not dispatch anything at all) to medical calls based on pre-arrival information and to quickly sort out and dispatch units to the cardiac arrest calls. Think taxis for the least-serious Omegas and Alphas, and anyone with an AED for the non-breathing Echoes. One main feature of MPDS is that the determinants (Omega, Alpha, Bravo, Charlie, Delta, Echo) are set based on RPs’ responses to predetermined dispatch questions while individual jurisdictions are free to customize their response models (Code 2, Code 3, first response only, ambulance only, both, etc.) to the determinants. I like San Francisco’s old bus token idea.
For instance, for the data that I’ve collected, responses to Omegas and Alphas are Code 2, the rest are Code 3. This is a fairly common setup in other jurisdictions.
In the United States, because the typical staffing configuration in most emergency response jurisdictions only consists of engines, trucks and ambulances, and because the typical disposition options with which most oversight agencies (and their lawyers) are comfortable are “transport” and “no transport,” in practical terms, the only real noticeable result is the increase in Code 2 responses in place of the riskier yet not more fruitful Code 3 responses, which isn’t a bad thing. It just doesn’t go far enough.
Judging by the details involved in the implementation of EMS systems as well as MPDS, there are plenty of signs out there that even many EMS jurisdictions and the people who think they know enough to make decisions do not understand the purpose of MPDS, which is exacerbated by the fact that most administrators have absolutely no or pathetically little time working in actual field-level EMS capacities, but this seems to be the norm and not something that will go away anytime soon. It’s the Dilbert principle as applied to EMS.
A common way jurisdictions misuse MPDS is how they use determinants to establish response time standards. For example, 20 minutes to Alphas, 15 to Bravos, 12 to Charlies, 10 to Deltas, 8:15 to Echoes.
Really? With the exception of Echoes, I’m wondering if this is set up for the sole purpose of generating fines for late responses. Based on what I’ve seen and done, there is no way anyone can convince me a Delta is 2 minutes more serious than a Charlie, and a Bravo is 5 minutes more serious than an Alpha.
Worse, units get re-directed from Bravos to Charlies, Charlies to Deltas, etc. This is colossally stupid. The only diversions should be to Echoes.
We once were dispatched to a Bravo for someone who fell off a horse on the outskirts of town, and nearly 10 minutes into the call, less than a half mile from it, we were re-routed to a Delta for chest pain on the other side of town because we were the closest unit. I tried to insist on continuing to the Bravo but was overruled by dispatch. We turned around, and after about 6 minutes, we were canceled by fire, implying total bullshit. We then were put back on the Bravo since we were still the closest unit, and drove back to the original call. There were several factors at play here, but none of them was common sense.
Anecdotally, my co-workers are quite often upset when they are dispatched Code 2 to a call and they find a patient legitimately sick. Frankly, I’m not much bothered by this. There has not been any substantial evidence demonstrating the benefits of shorter response times (or transport times) in the form of clinical outcome in any cases other than salvageable out-of-hospital cardiac arrest. On the other hand, I am more unhappy about being dispatched Code 3 to a call and finding some complete bullshit. There certainly have been plenty of actual negative outcomes of unwarranted Code 3 responses, usually in the form of wrecks. This is not an easy problem to solve for the simple fact that (a) callers are notoriously unreliable and sometimes downright dishonest, and (b) it is difficult to definitively determine (a) over the phone.
- Quantify patient acuity as determined by MPDS and response modes
- Evaluate patient acuity as observed on scene
- Compare the related distributions
Initially there were suggestions to include information on first responders’ (or lack of) actions and interventions. I did not include this because ultimately this project was about MPDS and response modes. We hardly ever went to any good calls, and I spent most of my time writing paperwork and not providing any treatment anyway, so I can hardly expect first responders to provide any treatment either, let alone single them out for the lack of treatment.
I also did not include details on interventions. The main reason is that there simply aren’t many interventions to record. Seriously, look at the shit calls we went to. Besides, I think that subjective acuity is sufficient to express the disparity between MPDS, response modes and actual disposition.
All 911 calls for medical assistance during which a patient presenting with a complaint of illness or injury is found.
- Canceled en route
- Canceled on scene by first unit on scene
- Unable to locate
- No complaint of illness or injury
- Psych holds and self-committals
As you well know, these incidents make up a sizable portion or our responses. I did not record the numbers, and sometimes I wish I did. Every canceled or excluded call means that any response was for the most part unnecessary in the first place. As for the psych calls, well, what a waste of time and resources**, from the perspective of 911 ambulances. If it’s not some guy who wants a place to crash and a sandwich to eat, it’s another guy who’s going for the fourth time this week because he’s “going to kill himself” again. If it’s not those guys, then it’s the guy who’s trying to get out of an arrest, and he knows how to play the system. If you didn’t actually hang yourself or legitimately slice your wrists down the bone or ingested 230 pills, psych calls are absolutely not 911 ambulance calls. When are we going to grow a spine and make the cops take the psych calls?
Of the information recorded, some of the relevant data are used:
- MPDS card, determinant, subcode
- Response mode
- Chief complaint as reported to the call-taker
- Chief complaint on scene, provider impression and relevant information
- Poor historian(s)
- Psychiatric component
- Frequent 911 caller
- Subjective acuity
Subjective acuity is categorized into “Very Minor,” “Minor,” “Urgent” and “Emergent.” It’s just what I decided to label these patients, based on their presentation, their history, my experience and common sense. Their meanings are very plain. It is nothing like what the CDC and ACEP did. In all of these cases, my subjective acuity was broadly in agreement with how the ED decided to manage them, in terms of resources assigned. The EDs use triage categories more in line with what the CDC and ACEP use than my subjective ones, but in many instances this has more to do with theory than practice. For example, “chest pain” may automatically be triaged as something more serious even if it’s been ongoing “for 2 years.” As nosy as I am, I made quite some effort to follow up unofficially just to see what patients ended up with. I don’t remember any ED finding something that made a patient more acute than I rated, but if that happens in the future, I fully intend to adjust the subjective acuity I assign a patient.
To give you a better idea of this so-called subjective acuity, while dependent on specific cases and presentations, and not absolute, here are some examples:
Very Minor – no complaint, toothache, not feeling well for days, obvious death, etc.
Minor – abdominal pain, vomiting, anxiety, s/p seizure, etc.
“Urgent” and “Emergent” are self-explanatory and do not interest me much since they are not the source of our collective aggravation, anger, eye-rolling and dash-pounding.
I also want to point out that the main description of these subjective categories is best defined by the most reasonable and least self-centered way to go to the ED. For example, a few years ago when I ended up needing emergency surgery, I found a ride. Just because someone ends up being diagnosed with a legitimate entity doesn’t mean at the time this person called 911 it was absolutely necessary for us to respond Code 3. In fact, far from it. Studies typically use discharge diagnosis as the typical measure of patient acuity, because it is objective. However, I think there is some value in using such subjective measures as providers’ impressions. Seriously, how often do we get patients who elicit, from EMS to RN to PA to MD, a continuous chain of eye-rolling, accompanied by, “This is bullshit.”?
Findings (no shit, Sherlock):
- MPDS and common response models produce significant overtriage
- Large majority of 911 calls for medical assistance are mind-numbingly routine
- Please keep all of this in mind when you drive to a call
Less obvious findings (based on this relatively small sample size):
- Alphas tend to be quite accurately triaged
- The ratios of Very Minor/Minor to Urgent/Emergent are quite similar in Bravos, Charlies and Deltas, implying there is no practical difference in these determinants. Given that most jurisdictions dispatch different equipment traveling in different modes to scenes depending on whether it’s Bravo, Charlie or Delta, and given that units are often re-directed from Bravos to Charlies and Charlies to Deltas, this seems pointless. I think we’re dicing these calls too fine by differentiating between Bravos, Charlies and Deltas.
- Charlies and Deltas have similar respective breakdowns to the overall breakdown in acuity.
- Deltas include all the “unknown medicals,” “non-triaged medicals,” PD requests, etc. The lack of information automatically makes these calls Deltas, but there were very few legitimate patients found as a result of these dispatches. This makes no sense when contrasted with medical alarms, which usually have no information as well, and are basically “indoor unknown medicals.” These calls are automatically categorized as Bravos, but they are in essence no different from the Deltas.
*Other overtriages: Patients who meet trauma center criteria and the use of spinal immobilization, but those are entirely separate posts.
**Some years back when we were self-reporting call data on Scantron forms on every single call, I took it very seriously because they’d give us our numbers at the end of the year. One year I think I went to more than 1400 calls, and the most common chief complaint I had marked was “Behavioral/Psychiatric” at 12% of all complaints. “Shortness of Breath” was second at 11%. That should have been a huge red flag as far as 911 ambulance utilization is concerned. But obviously nothing has changed.