The Land of Protocols

American ER Doc Gone Walkabout Episode 031

Welcome to the Land of Protocols, home of the Clinical Pathway (closely related to the Yellow Brick Road). (I’m not trying to make any direct connection to the Wiz, but…).

I’m not sure if the Wizard of Protocols has made it yet to Oz, but here in the US he’s already manipulating the levers behind the curtain, and we’ve seen quite an expansion of Protocols and Pathways (personal communication, 26th street Pub, blood alcohol level >100 mg/dl).

What’s not to like about a pathway: If I know a lot about a topic, the pathway helps to streamline my well-informed decision-making. If I don’t know much about a topic, the pathway helps me make reasonable decisions even when I don’t know the underlying data. If finalization of a clinical process requires multiple time-consuming steps, I can hand off those steps to a mid-level practitioner (physician assistant, or nurse practitioner) to follow a protocol, while I return to more acute and complex tasks. The pathway assures that I don’t miss a step in a process, even if I’m a forgetful old fart or an inexperienced young fart. And, it gives me a lever with recalcitrant consultants (“But, Doctor, the protocol – agreed jointly by our two clinical departments – calls for admissions of these cases, even when we both know that his problem is alcoholic stupor, and poor hygiene, and that he most certainly doesn’t really have…….).

The whole idea of simplifying operations isn’t new: “Civilization advances by extending the number of important operations which we can perform without thinking about them.” — Alfred North Whitehead, 1911

Thank goodness, there are, and can be, no unexpected consequences of the growing reliance upon, and use of, protocols and pathways in the ER.

Well, on further consideration, maybe a little, or a few minor problems. I’ll think about this a bit, and come back later.

OK, I’m back! After a few more personal communications at a local pub – as well as a few shifts following protocols and pathways. On further reflection, perhaps we should be a bit cautious here.

There’s a tension between the goals of patient satisfaction, efficiency, throughput and the goals of being thorough in considering less common causes of a symptom, more sophisticated plans of treatment, and in the needs of our other stakeholders – the trainees. Nicholas Carr puts it like this: “Abstract concerns about the fate of human talent can’t compete with the allure of saving time and money.”

Nicholas Carr’s article in The Atlantic, All Can Be Lost: The Risk of Putting Our Knowledge in the Hands of Machines, discusses some adverse effects of high tech “help” on human performance. He starts with the examples of commercial airline pilots who have become highly skilled computer operators but who have forgotten how to fly an airplane, such that when a computer becomes confused and disconnects – for example when the airspeed sensors ice up – the pilots don’t make the right moves and crash the plane. Admittedly this is a more complex technology, but isn’t too far afield from the medical trainee who has become so dependent upon pathways, that he forgets, or never learns, to be a physician: “Chest Pain, just plug it into the pathway – 2 EKGs, 2 troponins, if normal a CTCA or treadmill, and home.” Nothing to it. (While forgetting that, several days before the onset of pain, he had an ablation – anybody ever hear of pericardial effusions? Real case. I kid you not. WTF?)

But, although there are some prominent commentators who seem to be wholly unable to make a clinical decision that truly atypical chest pain is non-cardiac, there really is a substantial difference between someone with 2 weeks of chest pain that he can put his finger on and which began when he was struck by a drive while playing short leg (for the Yanks, this is a cricket position approximately equivalent to having your short stop playing 10 feet in front of the batter – the guy playing it, unless already brain damaged, usually wears a helmet and face mask that he borrows from a Canadian hockey player). There really is a difference between that and the 60 year old with a month of worsening but transient chest pain with exertion, who’s now had an hour of central chest pain, radiating to his neck and is soaked in sweat. The first should never be entered in the low risk chest pain protocol, and neither should the second – even with a normal EKG, he’s not low risk. For the pilot bred on computers: the sensors are iced over and we should push the nose down to gain speed, not pull back on the yoke. But, over-dependence on the protocols seems to make people forget to make clinical decisions.

When we forget to be physicians and make judgments, we also seem to have difficulty deciding which protocols to use: it makes my head want to explode when a handover involves some variant on the plan to finish the 6 hours ACS protocol, and if that is negative, then do the PE protocol, and while we’re at it make sure that the CT is timed so that we get a good look at the aorta. All the while forgetting the pertinent parts of the history that suggest a 90% likelihood that the patient should really be going home with a PPI.

Are we in danger of becoming dependent upon protocols to the point of becoming like the recent cases of Apple maps routing automobiles across an airport runway in Alaska. And people following the instructions. Or, the Inuit hunters who, relying on GPS directions, drive a snowmobile onto thin ice.

A corollary of the inability to make a judgment of who should be entered into a clinical pathway, is the attempt to shoehorn complex patients into pathways designed for simple patients. A chest pain pathway designed for a basically healthy person with perhaps a few risk factors, is not well designed to address the patient with multiple comorbidities, but we periodically see, sitting back in our observation area, the person with not only diabetes and hypertension, but with an autoimmune disease, multiple prior stents, HIV with a low CD4, asthma, a productive cough, and a recent bite by a black spider with a red spot on its abdomen. Although we could design a pathway that addressed every potential cause of chest pain, the order set would add more and more of those maddening checkboxes – with the correct box becoming harder and harder to find, and it would be too cumbersome to use. The role of the physician is to identify the cases that will truly fit into a pathway, and this takes skill and art, and not to allow paradigm creep to push you to fit inappropriate patients into the pathway. I would propose that one must learn to be a physician first, then use protocols to finish off the uncomplex parts of the job.

Even simplistic protocols can have glaring omissions: pregnancy and travel in the Wells’ PE score, or contact lens use in a recent conjunctivitis pathway. The Wells’ score was a research tool that explicitly removed pregnancy. The conjunctivitis pathway that ignores contact lens use presumably assumes that an experienced ophthalmologist would immediately recognize that a contact lens wearer is unlikely to be simple conjunctivitis. The eye protocol also faces a conundrum in applicability to different settings: the protocol that may be appropriate for a novice is too simplistic for the experienced, and the pathway that, unless the eye is itching, leads only to consultant referral – that may be fine if you are well supplied with ophthalmologic consultants, but would be impossible in a rural clinic.

A protocol, even a good one, can also supply some silly advice – think the, usually useful, autocorrect on your iPad that will sometimes (stealthily) replace your chosen word with something that is incomprehensible or misleading or obscene. In my dependence on autocorrect, I often miss these substitutions. Like I miss the redback/black widow envenomation when evaluating chest pain (perhaps a bad example, since I miss these whether using a protocol or not). Maybe the protocol should have a “hard stop:” if you have spent more than $20,000 evaluating chest and abdominal pain, stop and ask about spiders.

There is subtle pressure on the residents to ignore the H&P: If we are being encouraged to just follow the protocol, and most patients, even with strong evidence of non-cardiac source of pain, will eventually enter the pathway, why even bother with the parts of the history and physical that would push the patient out of the pathway – either because of higher risk or lower risk. The limits seem to gradually extend further and further away from the designated group – in both directions. There’s a financial incentive, too – nobody pays more for a detailed history and physical, but if we order (and interpret) additional tests, check more boxes, and enter a patient into “observation status,” the payment goes up – at least in the US.

An article in ACEP Now provides a particularly pernicious set of recommendations that explicitly and implicitly would lead to avoidance of much of the physical exam (only 10% yield in diagnoses – ignoring the much lower yields in ACS rule-outs and CT PE studies in low yield patients), deceiving our patients as we knowingly performing useless bedside maneuvers for show, and fabricating medical records for billing purposes. It’s one of the most disturbing articles I’ve read recently.

I would also submit that, although uniformity may provide a certain lowest common denominator that is reassuring in certain circumstances (think McDonald’s), it also crushes creativity. Now, creativity may not always work (think of that brilliant sounding menu item that tasted truly awful), but it sometimes advances our knowledge and progress (you probably won’t find Moroccan orange-apricot chicken at McDonald’s anytime soon – nor sticky date pudding – damn, just thinking of sticky date pudding makes me want to hop on Quantas and fly to Oz for dinner).

One of the ways of both advancing, and controlling the advance in medicine, is the contrast between the “early adopters” and the “late adopters.” A new technique (Captain Morgan hip reduction) comes along. The earlies try it. If it works well, the word gradually spreads and more and more try it. If it fails, the word spreads – perhaps more rapidly – and not everyone has to fail at it before it is abandoned. If a protocol were to dictate: “All hips will be reduced by the Stimson Stetson hang from the ceiling method,” nobody would try something newer and perhaps less traumatic and more successful. Or, if the protocol writer were to somehow decide that the new technique was the best since sliced bred, and write a protocol for it, there would fail to be that period of transition where the early adopters and late adopters have the opportunity to compare and contrast their successes and failures.

The Atlantic article provides the example of experimental subjects who learn to play a game with guidance of a computer. The control group learns to play the game by their own means. Not too surprisingly, the computer guided group learns the game more quickly. However, when the two groups are given the opportunity to play the game again, several months later, this time without guidance, the performance of the self-taught group exceeds that of the computer-guided group. The self taught seemingly internalize the game in a way that those who are “ guided” don’t. Part of the learning process seems to be enhanced by having to work it out yourself. That process may be short-circuited (IMHO “may” is far too diplomatic a word) by providing our trainee physicians with protocols and pathways that they may follow without actually learning the evidence base that supports, or fails to support, those protocols.

The recent flurry of interest in TTM suggests another issue with protocolized medicine: the inability to respond rapidly to changes. Even if many of us were to decide, based on the new (only a little – but no fever allowed) cooling study, that we should change the protocols, we would have many who hadn’t read (or even heard of) the articles. We would need them to catch up to speed, then achieve a consensus (between the early and late adopters), bring the other departments – neuro and cardiology and critical care – on board, and then write and rewrite, and ultimately publish a final draft. However, if we were to all agree that a protocol is a guideline, and a baseline below which care will not drift, we would be able to leave in place the “guideline”. But the earlier adopters would be free to move toward TTM while the lates continued with cooling to 33 Celsius. And, we’d compare notes, and perhaps more would move to a 36 target, or perhaps the earlies would drift back to 33. Yes, we know that the evidence requires large studies, sophisticated blinding, and sophisticated statistics to demonstrate or to disprove the value of a management plan. But, lots of medicine has no firm evidence base, much that is statistically sound evidence turns out to be impractical, and some seemingly sound evidence turns out to be tainted or flat out wrong. And the early adopter/late adopter conversation is IMHO a crucial part of the gradual advance of both the science and art. And, excessive reliance on protocols may hamper that.

Language really does matter. If we take a series of rules of thumbs/heuristics/shortcuts and call it a protocol or pathway, there is (at least in American English) a sense that we really shouldn’t deviate. Whereas when we call the same set of heuristics a “guideline”, there a different sense of generality that allows creativity and thought. Think: follow the concrete path 1275 meters, turn 60 degrees left onto the tarmac path, etc. vs “Walk generally east avoiding cliffs, waterfalls, and large trees until you come to….”.

So pathways are good: they make the inexperienced smarter, they make the smart faster, they streamline communications.

Pathways are bad: they make you stop thinking; they make you less smart; they subtly encourage you to shoehorn inappropriate square patients into round holes; they may be too simple to cover a reasonable range of clinical settings – or may be too complex to be useful without taking inordinate amounts of time; they may encourage over dependence on testing and imaging; the shortcuts in learning may, over the longer term, impair the retention of learning to deal with a clinical issue. Perhaps most critical to our improvement as physicians, over dependence on tight adherence to a pathway, the pathway may impair that tension between tradition and new knowledge that allows progress in our approach to clinical problem solving.

See you all in Oz, on the Yellow Brick Pathway.