How important is data to the success of your system and the outcome of the patient care you deliver? At Montgomery County (Texas) Hospital District (MCHD), we consider data one of our most valuable resources. Being able to measure something gives us the ability to understand and manage it. Without data, we’re in the dark about our performance, which means we can’t improve it.
So if having data is so important, how much do you need to make informed clinical decisions? You need exactly as much data as required to accurately reflect reality of what you’re looking at. When changes occur slowly, the data frequency requirement is low—and vice versa.
The right measure of performance is critical for balancing the amount of data with the process you’re measuring. EMS examples of appropriate matching include measuring dispatch process times in seconds, vital signs every few minutes and hospital length of stay in days. Using the wrong unit of measure could lead to an inability to draw meaningful results.
With the most critical of patients, most notably those in whom we’re immediately affecting pulse, respirations and blood pressure, how much data is needed? Many EMS protocols call for vital signs every five minutes in the sickest of patients. Although this may have been acceptable in the past, we believe the bar has been raised.
MCHD records, trends and analyzes every second of vital sign data to properly assess vital human physiologic parameters. Only through the retrospective review of this data have we come to fully understand how the care we provide affects patients; the National Association of EMS Physicians (NAEMSP) agrees.1
How we got there is a bit of a journey, so let us explain.
How It Started
In 2003, researchers in San Diego began presenting and publishing the results of their landmark trial of paramedic rapid sequence intubation (RSI) in patients with severe traumatic brain injury. Although we, like many other agencies, were surprised that the study concluded, “paramedic RSI protocols to facilitate intubation of head-injured patients were associated with an increase in mortality and a decrease in good outcomes,” we were fascinated by the details.2
We had an opportunity to see James V. Dunford, MD, FACEP, present on the researchers’ findings, where Dunford demonstrated that there were events happening that hadn’t been previously reported. Dunford used recordings of RSI cases containing high-resolution data to show transient periods of hypoxia occurring during intubation.
These hypoxic episodes were going unnoticed by the paramedics doing the intubation. Traditional, “once every five minutes” vital signs were missing these episodes as well. During debriefing sessions immediately after the cases, paramedics consistently described the intubations as easy, and none noted the hypoxia occurring.
Our immediate reaction to learning about this situation was to ask ourselves, “Do we have that problem in our service?” To find out, we had to ask whether we had the data and whether we could measure similar cases in the same way. Fortunately, and unfortunately, the answer to both questions was “yes.”
The ZOLL M-Series monitors that we used at the time recorded discrete readings of heart rate, SpO2, respirations and EtCO2 every second. Through a somewhat cumbersome and time-consuming process, we were able to create graphs that showed us this critical trend data at a one-second sample rate. We used the published definitions and methodology as our standard to benchmark our process.
After developing a process to create graphs of this second-by-second trend data, we were able to identify that we did indeed have a similar issue to the one uncovered in San Diego. More than 50% of our RSI patients were experiencing hypoxic episodes, with duration and depth at comparable rates to those published in their series of articles on the subject.2–4 A finding that should concern every EMS physician was that even though our number of paramedics was smaller and our intubation success rate higher, the extent of the problem was eerily similar (see Figure 1, p. 11).
By this point, Dunford and Daniel P. Davis, MD, had also shown that hyperventilation was perhaps an even larger issue than hypoxia. We quickly confirmed once again that the ability to measure our performance in this newfound way showed us a problem we had that we weren’t aware of. Our paramedics and firefighters were hyperventilating our patients. The muscle memory of most providers caused them to squeeze the BVM up to 30 times per minute.
Immediately after discovering the problem, we started a journey toward improvement. It wasn’t easy because our first, second and even third attempts failed. We initially did what we had been trained to do, and that meant looking away from individual providers as the root cause of the problem. We focused tremendous energy on training, equipment and protocol issues. We had some sort of airway training at every quarterly mandatory training session for a long time.
We continued to show graphs of hypoxia and hyperventilation. Crews would ask, “Was that one of my patients?” We didn’t want to embarrass anyone by singling them out as having a problem because it was happening across the system. The effect was that everyone reached the same conclusion: Everyone else had a problem, not them.
They continued to see graphs of different situations, and although they found them interesting and informative, no one really changed their behavior. In our attempt to address this as a system issue, we weren’t identifying specifics. This meant our message wasn’t effective, and we weren’t learning from our mistakes. It took a while, but we finally found a process to improve the rates of hypoxia and hyperventilation.
A New Workflow
Ultimately, we had to find a way to use this information to get better. We looked at how the best agencies enhanced their processes, and we formally adopted Six Sigma as our improvement methodology. Using this tried-and-true, quality process, we were able to develop a process that ultimately proved effective.
We created a workflow to make and analyze the graphical representation of the one-second trend data as soon as possible after each critical patient care encounter. Our goal after that was to get this information in front of the staff that had cared for the patient by their next shift. Our plan was to share the right information at the right time with the right people. This would allow us to turn our data into information, so that the staff could gain knowledge about their performance.
It worked. This knowledge, when provided to the staff immediately after a critical airway incident, allowed them to gain wisdom about the effect of their interventions. After learning how their decisions and actions affected the patient, crews changed their treatment plans on the next call to better manage the situation. They finally put those lessons we’d been providing during that airway continuing education to use. Such things as “first attempt equals best attempt,” bag-valve mask (BVM) application prior to intubation, gum elastic bougie use and getting the patient off of the floor to intubate finally started to pay off (see Figure 2, above).
After literally years of minimal improvement, we developed a method to drastically reduce the hypoxia and hyperventilation associated with airway procedures. It was time to celebrate!
We learned many lessons from this technique. By analyzing the cases in such high resolution, we were able to identify additional trends and uncover multiple other details about our care.
We learned about the impatience of paramedics and how they were quick to attempt intubation when SpO2 didn’t immediately rise with BVM ventilations. We were able to show how it can take 45 seconds or longer to see increased oxygenation using a fingertip SpO2 sensor.
Another critical improvement came when our paramedics witnessed clear feedback that use of a transport ventilator allowed much more consistent control of respirations. This, in turn, allowed improved management of EtCO2. We also witnessed the hyperdynamic phase following cardiac arrest, as well as subsequent cardiovascular collapse—something we had read about but didn’t fully appreciate until seeing the phenomenon in our patients (go online for Figure 3).
Beyond the Noise
One of the most interesting cases we identified involved the successful decompression of a tension pneumothorax. The particular hemodynamic footprint of this case is unique: It includes tachycardia, low amplitude ECG, low SpO2, tachypnea and unexpectedly high EtCO2 (more than 99 mgHg). Go online for Figure 4, which clearly illustrates the dramatic improvement of all physiologic parameters following chest decompression.
One thing you’ll note is the frequency of the data points can sometimes give a noisy or messy appearance. We initially considered trying to smooth these lines out, or somehow reduce the number of readings, thereby lowering the resolution of the information. Thankfully, we decided early on not to reduce or filter that data in any way. In much the same way that researchers didn’t initially realize that aberrant readings in ozone levels were telling them about a hole in the ozone layer of the atmosphere, we found that there’s information within this noise.
The fact that some lines appear thick or fuzzy shows us there’s inherent irregularity in the parameter being measured. Rapidly fluctuating readings indicate the device may be searching for an accurate reading and that the data may not be 100% accurate. Our ultimate conclusion is clear: More data equals higher resolution, which equals more information, which equals more learning, ultimately equaling improved patient care.
The NAEMSP stated in its 2003 Position Paper on Uniform Reporting of Data from Out of Hospital Airway Management that oxygen saturation should be recorded within five minutes before intubation and again within five minutes after intubation. That could leave up to a 10-minute gap in what happened during intubation. Now, combining information from San Diego with our own experience, we know that what happens during these critical few minutes is of the utmost importance.
The NAEMSP discusses drug-assisted intubation (DAI) in its most recent position paper. It now states in Drug-Assisted Intubation in the Prehospital Setting Position that all agencies should have “resources for continuous monitoring and recording of heart rate and rhythm, oxygen saturation and EtCO2 before, during, and after DAI.” This is exactly what we’re demonstrating here. To be considered state of the art, you must have and use this data.
If you publish research on airway management, your data is incomplete if you don’t demonstrate exactly what’s happening during intubation attempts. How many patients experience hypoxia, to what extent and for how long? How many are hyperventilated and for how long? At MCHD, we believe this standard should apply not only to EMS agencies, but also hospital emergency department quality and research efforts as well.
It should come as no surprise that when the time came to replace our cardiac monitors, one of our essential requirements was the ability to monitor and record data at one-second intervals.
At the time, only one major monitor manufacturer included this feature on its devices. However, once we explained our desire for this feature and demonstrated the value of this data, Philips was convinced and changed the software in its device. The company’s willingness to make this change, along with our staff’s positive field trial of the Philips HeartStart MRx, made it our winning bidder. In March 2011, MCHD became the first agency to record and analyze one-second data using a Philips device.
Today, using the Philips monitors and software provided by ESO Solutions, we’re only a few clicks and seconds away from graphical analysis of a wide variety of cases.
One-second trend analysis should be an industry standard that’s available to every provider on every call, directly from the monitor. “Vital signs trend” analysis in real time should be available during the care of the patient and printed, analyzed and discussed by the providers after each call.
Disclosure: The authors have reported receiving honoraria and/or research support, either directly or indirectly, from the sponsor of this supplement.
1. National Association of EMS Physicians. Drug-assisted intubation in the prehospital setting position statement of the National Association of Emergency Physicians. Prehosp Emerg Care. 2006;10(2):260. www.naemsp.org/pdf/Drug%20Assisted%20Intubation%20New.pdf.
2. Davis DP, Dunford JV, Poste JC, et al. The impact of hypoxia and hyperventilation on outcome after paramedic rapid sequence intubation of severely head-injured patients. J Trauma. 2004;57(1):1–8; discussion 8–10.
3. Davis DP, Hoyt DB, Ochs M, et al. The effect of paramedic rapid sequence intubation on outcome in patients with severe traumatic brain injury. J Trauma. 2003;54(3):444–453.
4. Davis DP, Stern J, Sise MJ, et al. A follow-up analysis of factors associated with head-injury mortality after paramedic rapid sequence intubation. J Trauma. 2005;59(2):486–490.
This article originally appeared in an editorial supplement to the September 2011 JEMS as “One-Second Data: Analysis is only a few clicks away in Montgomery County, Texas.”