Have you ever wondered how you, as a caregiver – doctor, nurse, therapist, or any other medical provider – come up with a diagnosis or determine what tests to order? As it turns out, there is an ever-growing body of research on our thought processes.

How do you determine a patient’s diagnosis?

You might think, “I gather the patient’s history and it sounds like that certain diagnosis with which I’ve had experience in the past.” In podiatry, the classic example of this is plantarmedial heel pain that has an onset on first weightbearing after periods of rest. This symptom, termed poststatic dyskinesia, is used to pathognomically diagnose plantar fasciitis. For those of you in the plantar fasciosis camp, let’s hold off on arguing the pathophysiology so we don’t get lost in the weeds.

---

“Clinical scripts allow experienced physicians to make what appear to trainees as miraculous diagnoses with a minimum of thought.”

---

This is definitely a common pathway and is, in fact, used most often by experienced clinicians. This process is called a clinical script. Over time, we build up a library of clinical scripts that explain the common pathologies with which we are presented. A patient with a great toe that deviates toward the second and complains of pain most likely has Hallux valgus with bunion deformity. This is true about plantar forefoot pain with a new onset second toe that is elevated – most likely predislocation syndrome. Clinical scripts allow experienced physicians to make what appears to trainees as miraculous diagnoses with a minimum of thought. In reality, there’s nothing magical or miraculous about it. We’re simply engaging in experienced pattern recognition, something human beings are evolutionarily adapted to do well.

Consider this. You’re living 650,000 years ago on the African savanna. High grasses abound and your family group lives in the cave just over the hill. You’ve been foraging for some delectable nuts when you see a wave of the grasses off to your left. You also hear a deep rumble near that same area of waving grass. What do you do?

The answer is intuitively obvious: You run like crazy!

The evolutionary explanation says the person who sits around to think about what the waving grass and growling mean will likely not be around long enough to reproduce. On the other hand you – the one whose ancestors survived – know that these two factors added together equals danger. It might have simply been a hedgehog, but you can’t afford to be wrong. When the stakes are high, split-second decisions are necessary for survival. As a result of this process over millions of years, we have become strong pattern recognizing machines, and this translates to our modern day utilization of clinical scripts.

The Outliers

But what happens during that small percentage of the time, when we encounter the relatively rare situation, in which the symptoms and examination do not match one of our scripts?

Let’s take a modified example of the plantarmedial heel pain script. Here’s the new clinical scenario. You are seeing a 70-year-old female patient of normal body habitus, who complains of unilateral left-sided plantar heel burning and tingling pain, that is present at all times but worsens at night. She swims for exercise daily, eats well, and is otherwise healthy with no significant medical history. Your physical examination is entirely negative for any local pain, strength is completely intact, and your neurological examination is negative.

This is a much more difficult diagnosis to make than the simple plantar fasciitis one. The patient’s history sounds suspiciously like a neurological problem, with a wider differential diagnosis. At this point, the differential might include any of the following: tarsal tunnel syndrome, Baxter’s neuritis, medial calcaneal neuritis, lumbosacral radiculopathy (with any of its potential causes such as spinal stenosis, herniated disc, or plexopathy), piriformis syndrome, other mononeuropathy, nerve tumor, or other much more rare diagnoses (such as multiple myeloma affecting the spine, or a neurological disorder such as Guillain Barre, or multiple sclerosis). Granted, the earlier diagnoses on the list are much more likely, but even so, we’re still looking at perhaps four differentials.

At this point, the clinician would have a couple of options. First, he might randomly test everything she can think of with a giant battery of tests. This is woefully expensive and inefficient, and puts the patient through a lot of unnecessary testing. She might alternatively ask a friend. On TV, they call it “phoning a friend.” In medicine we call it “referral.” Sounds so much more professional.

The third option, though, is to use rational thinking, and figure out what specific testing should be ordered based on the likelihood that one of these differential diagnoses is the correct one. That my friends, is when the Bayesian approach kicks in and where we leave our little intellectual cliffhanger for next week.