It should surprise no one who works in an emergency department that the sepsis protocols were recently shown to perform abysmally. (JAMA Intern Med. 2021 Jun 21; doi.10.1001/jamainternmed.2021.3333.) Alerts were called on almost one of every five patients (18%), yet they failed to detect sepsis in 67 percent of septic patients.
We must do more than admit the failure of these protocols (and other means of medical “datafication”) that are enacted to capture CMS reimbursements and manipulate metrics that damage the scientific and therapeutic fabric of emergency medicine. We should explore why they were so wrong.
When arguments are made against the accuracy of converting complex qualitative aspects of health into simplified, often binary, quantitative data points, the metric and medical artificial intelligence progressives argue that these are only tools to assist medical professionals in their craft. This is a nefarious response by those employing the medical provider to behave in a particular way that maximizes the company’s profit.
Calling a sepsis protocol a clinical decision aid does not make it function any less robotically or blindly in the emergency department. As Sinclair Lewis wrote, “It is difficult to get a man to understand something when his salary depends upon his not understanding it.”
Consider for a moment what would happen if a dutiful provider-in-triage or a machine actually accepted an individual patient’s answer prima facie for the following questions:
- What number would you put on your pain?
- What was your fever at home?
- How many times did you vomit?
- How many pads have you used in the past hour?
- How many beers have you had tonight?
- Is there a chance you could be pregnant?
- Have you recently done any recreational drugs?
- If I administer a narcotic, do you have a ride home?
- Do you feel safe where you live?
- Or even, when do you think these symptoms first began?
These are questions we ask multiple times a shift, thousands of times a year. What is the difference between getting that information up front by a nurse, a dutiful provider-in-triage, or a machine to start a protocol or workup instead of waiting for a medical professional to interview the patient with their loved ones and friends in a private room?
Electronic data are born of a nonelectronic environment. Data that have been input must be validated by nonverbal and communally collected information. When the patient is sitting up in bed bright-eyed, messaging on his cell phone, and his mother is slumped in a chair looking at the wall, the patient’s verbal statement that he hasn’t kept anything down for three days is difficult to believe because of the nonverbal information you have collected implicitly in the interview. This is different from the patient who is diaphoretic or has roving, slow-blinking eyes with a tearful loved one nearby.
Which information is more accurate: the farmer who says his chest pain started that day, doesn’t take any meds, and hasn’t had any surgeries in a long time or his wife who looks over her glasses and barks, “He has been complaining about this for weeks. He is supposed to be taking medications for diabetes and hypertension, and he had a quadruple bypass five years ago!” (COVID reminded us how poor the interview is when the patient is alone.)
Epistemology is the study of how we decipher true statements from false ones. Cognitive decision-making is a field of neuroscience in which one considers how emotions affect memory and what may feign as reasonable answers. (Daniel Kahneman, PhD, Thinking Fast and Slow. 2011; https://amzn.to/3jC5kan.) Explicit and implicit biases, inference, confirmation bias, and other complex machinations make communication and decision-making extremely complex. Inputted data collection is not the same as information gathering, which is not the same as the awareness of an accurate reality.
Nonverbal communication is the mother of acquiring knowledge. Communally deciphered stories are the home of accuracy. Emergency medicine professionals have a massive, treasured neuro-visual file from which we draw every shift, often intuitively. We use multiple viewpoints and nonverbal responses from others in the room to humbly recheck the accuracy of our intuitive knowledge. It’s called the art of medicine, but that makes it sound unscientific. Actually, the highest form of healing science is acquiring this suprarational medical knowledge.
This is why datafication at a booth with a dutiful individual in a busy emergency department (or on an app or telemedicine screen) is the death of good emergency medical care. This is why protocols, provider-in-triage, and various forms of artificial intelligence where the patient is empowered to input his own data become garbage out.
We should seriously reconsider the idea of self-derived health data as a tool that guides emergency care. How many unnecessary, expensive visits have already been generated by home blood pressure cuffs, Fitbits with heart rates of 200 bpm, and iPhones that said it might be sepsis? What about every automated phone line that says, “If you have chest pain, hang up and call 911.” Allowing everyone to cast wider nets does not catch the particular fish you are looking for; we simply catch a lot more fish you don’t want in the net. Please offer emergency physicians the space and time to plumb the depths of a medical abyss we have fished for years.
Some may cry paternalism as a vestige of an old, bad school, but I would argue that our current modes of datafication are mechanized and industrial-strength paternalism for which we wash our hands guilt-free in the sink of consumerism. We are not being paternalistic when we hone the deep arduous craft of the medical interview, humbly checking neuro-intuitive knowledge with communally collected information. Instead, we are skillfully nurturing what has been handed down in the medical interview as the best kind of maternalism.
Access the links in EMN by reading this on our website, www.EM-News.com.
Comments? Write to us at [email protected].
Dr. Mosleyis an emergency physician in Wichita, KS. Read his past articles athttps://bit.ly/EMNHealingWords.