We’re taking a short break from our series on transfers to discuss a recent post on the EMT-Medical Student blog. One of the issues he brought up is the old saw, “Treat the patient, not the machine.” Rogue Medic struck on this as well.
What do people mean when they say this? They mean that if you attach a diagnostic tool like a pulse oximeter, and it gives you a result that is at odds with the rest of your assessment, then it is probably wrong, and you should not base your decisions on it. It can be broadened to the BLS level, including findings like vital signs, by saying: “Treat the patient, not the number.”
And it’s essentially true. In fact, something I frequently harp on is that diagnosis must always be based on a broad constellation of consistent findings, not on any one red flag. We like red flags, we want red flags, because they’re easy, but it never works that way. The body is an interdependent system, and if a pathology is present, then it almost always has multiple effects detectable in multiple places.
This idea can be looked at differently by asking another question: is it possible to be severely, acutely sick without showing it? I don’t mean long-term problems like cancer; you can’t look at someone and detect that. But if someone’s dying in front of you, of a proximate cause like hypoxia, is it always obvious based on their presentation?
Generally the answer is yes. That’s why it’s wrongheaded to look at a healthy patient with pink skin, normal respiratory rate, calmly denying shortness of breath, but with a low oxygen saturation, and say, “Oh no — he’s hypoxic!” If your oximeter says 72%, what’s more likely — that the number is wrong, or that the patient is somehow hypoxic without any other evidence of it?
Call this the phenomenon of the Hidden Killer. Is it common? Is it real?
It is not common. But it is real. And that’s what’s not recognized when people say, “Treat the patient…”
Why do we take 12-lead ECGs on chest pain patients? Because a clinical assessment alone cannot reliably detect ST elevation, which (simplifying the issue!) is diagnostic for a heart attack.
Why do we take CT scans of blunt head injury patients? Because a clinical assessment alone cannot reliably detect intracranial hemorrhage.
Why do we perform abdominal ultrasounds in multi-system trauma patients? Because a clinical assessment alone cannot reliably detect abdominal bleeding.
Now, some critics will say that all of these will indeed present with obvious, frank clinical findings. The major STEMI patient will eventually enter cardiogenic shock. The head bleed will become comatose and present with Cushing’s Triad and herniation. The abdominal hemorrhage will have guarding, distension, and eventually outright shock.
All true enough. But we’d like to find them earlier than that. It’s true that severe and late pathologies are usually obvious, but our job is to find them when they can still be treated, not after their effects are permanent or lethal. Heck, we could also just provide no medical care and wait until everyone died to make a diagnosis, which would extremely easy to assess, but a little pointless. It is rare that big problems do not have a big assessment footprint, but “small” problems can still be a big deal.
Consider the much-maligned pulse ox. Surely it does not replace a full assessment. But when used appropriately and its role understood, it provides valuable information. A drop from 99% to 94% saturation may not be clinically obvious, but it is potentially significant and surely worth knowing about. What about the patient who is non-verbal at his baseline? Is he going to complain if he drops from 95% to 87%? Will it be frankly obvious from his skin and breathing? Maybe, maybe not. (How about if he’s on a mechanical ventilator at a fixed rate?) If not obvious, does that mean it’s no big deal?
Is the pulse ox always correct? No. But like all things except magic, it’s wrong in predictable ways, ways that can be accounted for, and when it is wrong, that can tell you something too. It requires adequate peripheral circulation, and poor perfusion will make it read low. How is the patient’s distal perfusion? Pink and warm? Good capillary refill? Then you’re probably okay. Carbon monoxide poisoning will make the sat read high. Has the patient been in enclosed spaces with heaters or open flames? Working around engines? Is there any potential source of CO in their history? If not, you’re probably okay. Alternately, does their sat read unusually high compared to their clinical presentation? You might then consider carbon monoxide — something you might not have otherwise have known without the oximeter. It didn’t give you a correct number, but by knowing how and when it fails, it gave us a useful answer.
Here’s a recent example. I picked up a patient with a blood pressure of 54/4. That is a ridiculous blood pressure; arguably, nobody should have it, on the theory that a pressure that low should be pretty close to unobtainable. But, there it was. We diverted to the nearest hospital and I was subsequently chewed out by the receiving nurse.
Do I think that patient truly had a central arterial pressure of 54/4? Nah. Although she wasn’t doing well, her skin was better than that, and although she was altered and combative, she wasn’t comatose. However, her pressure was undoubtedly low, and just how low? If I don’t go with this number, then I’ve got no guidance. The clinical picture was clouded. I couldn’t ask if she knew what day it was; I couldn’t ask what her complaints were; she was non-verbal. She was tachycardic and hypoxic and diaphoretic; she was certainly sick. So, treat the patient, or treat the number? The number may not have been right, but it was concerning enough that it couldn’t be ignored without an assessment that otherwise screamed “no problems here!”, which was not what we had.
Treat the patient? We always treat the patient. A hands-on physical and history is a vital, vital tool for assessment, but other tools are also useful. Some people lament the downfall of the traditional clinical assessment, from the days when doctors with fingers like pianists made diagnoses from findings like Ewart’s sign, and it is shame, but the reason that the high-tech tools like imaging and labs have become de rigueur is that they work well — they diagnose many problems with a speed, sensitivity, and reliability that is not otherwise possible. Nobody would ever say, “Treat the patient, not the unstable cervical spine fracture,” because we recognize that’s the sort of thing you may not otherwise notice until it’s too late. That’s why we spend big bucks on CT scanners.
It all matters. It’s all useful. We should neither cast aside our individual numbers nor ignore the bigger picture. Data is something that, like money and sex, you can never have too much of.