When diagnosing patients, Occam’s razor sometimes fails

I don’t like to brag, but if there is one area of my skills as a doctor about which I am proud, it would be my skill as a diagnostician.  I like to play Sherlock Holmes and figure out what’s going on with people, and I think I’m pretty good at it.

So I lied.  I do like to brag … a little.

In most people’s mind’s eye, the role of diagnostician is this:

Doctor: “So Mrs. Smith, what brings you in today?”

Patient:”I feel like I am dying.  I have trouble catching my breath, I am running fevers of 108 every day for the past two months, my fingers are turning black, I pass out at least two times every hour, and I’ve been vomiting up blood.”

Doctor (puts his hand in beard in a thoughtful expression): “Hmm … sounds serious.  Are there any other symptoms you have been having over the past two months?”

Patient: “Well, yes, now that you mention it, I’ve got this strange rash on my feet and they’ve really started to smell bad.”

Doctor (turns to nurse with knowing expression): “Just as I suspected.  That last bit of information was crucial in tying this all together.  There is a rare foot fungus that causes all of your symptoms.  That’s why I always need you to tell me everything so we can find the right diagnosis.”  (Doctor pulls out a vial of oil from his white coat).  “Let me put this oil (which comes from the moss of a tree that only grows in Tasmania) on your feet.”

The rash vanishes and the patient’s color returns to a healthy pink glow.

Patient: “I am healed!  Thank you doctor!  How did you know that was the problem?”

Doctor: “Give me all the facts and I can figure out what’s wrong.  Never forget the wisdom of Occam’s razor: ‘The simplest explanation is usually the right one.’  I look for the one explanation that ties all of the symptoms together and that is usually the answer.”

As a clinician, I fantasize about being the heroic detective who notices those obscure facts that others would miss, coming up with the life saving  diagnosis when all others had failed.  This, unfortunately, is not how it usually works when dealing with real human patients, and my desire to find a single diagnosis to explain what is going on can actually distract me from finding the answers my patients need.

Here’s how the real interaction often goes:

Doctor: “So, Mrs. Smith, what brings you in today?”

Patient: “For the past 6 months I’ve gotten more and more tired.  I just have no energy at all.”

Doctor: “Are there any other symptoms?”

Patient: “Yes, now that you mention it, I’ve been losing a lot of my hair, I am gaining weight, I’m constipated, and my skin is real dry.”

Raise your hand if you think you know the diagnosis?  If you said “low thyroid,” you fell into my clever trap.  Even though these symptoms are classic for hypothyroidism, I have some information you don’t: Mrs. Smith just had a battery of blood work that was completely normal, which included thyroid testing.  I also know some other facts about Mrs. Smith:

  • She’s married to an alcoholic.
  • Her mother recently died suddenly.
  • She had a hysterectomy 3 months ago.
  • She has a history of bad environmental allergies.

It turns out that Mrs. Smith isn’t sleeping well at all (related to her marital situation and loss of her mother), which explains her fatigue.  Inexplicably, a large percentage of my patients who don’t sleep well fail to mention this fact, instead focusing on their extreme fatigue.  I point out that there is a well-established link between lack of sleep and fatigue, and that fixing sleep will go a long way in improving fatigue.

The hair loss is related to her recent surgery and the loss of her mother.  There’s a condition known as telogen effluvium where a person can lose up to a third of their hair following a particularly stressful event (such as surgery or a large psychological trauma).  It accounts for the vast majority of acute hair loss in my office.

Her dry skin is related to allergies, which everyone in my town seems to have, and the constipation is irritable bowel syndrome she’s having related to stress in her life.

This is not the solution Hugh Laurie would’ve uncovered on an episode of House,  nor is it the glamorous deduction Benedict Cumberbatch would’ve made on Sherlock.  This would make really dull television, to be sure, but it is by far the rule as to the answers I uncover as a diagnostician.

Occam’s razor be damned.

This is frustrating.  It frustrates patients who have thoroughly researched their symptoms and have come up with the “one diagnosis to rule them all” which explains (and fixes) everything.  It frustrates doctors in training who get excited when they hear the patient say “all the right things” that point to a particular diagnosis, only to be turned back by negative lab tests.  Finally, it frustrates experienced doctors like me when we have patients for which only one clinical diagnosis makes sense but the data rule out the only explanation we’ve got.  Yet this is reality, and we must always bow our knee to the facts before us.

Here’s how I approach diagnostic problems in the real world (you may call this “Rob’s razor” if you want):

Listen to the story.  Patients will usually tell you what is wrong with them.  Pay attention to the entire history, and don’t make theories until you’ve heard everything.

Don’t assume you’ve heard everything.  Even after you’ve heard everything, you are inevitably missing important information.  This may be chapter 1 of a the patient story, and simply the passage of time will make a confusing story begin to make sense.

First focus on the things that pose the largest risk.  Make sure chest pain is not the heart, fever and cough is not pneumonia, and abdominal pain is not appendicitis.  This can be done simply by getting a clearer history, or it may require further testing.

Then address problems that are common.  Common problems presenting in uncommon ways are more common than weird stuff.  I look for patterns: episodic abdominal pain suggests gallbladder.  Constant chest pain lasting for two days is never ischemic heart pain.  Weird chest pain in a 50-year-old diabetic smoker is more worrisome than classic pain in a 20-year-old female.

The older people get, the less likely you find a single diagnosis.  Pediatrics is usually simple, as kids are usually sick with one thing.  Adults, on the other hand, often have multiple problems at once.  You will usually be wrong if you assume all symptoms are related in an adult.

When in doubt, blame medications.  I had a person recently with itching in the ear that would not stop.  We tried multiple things to relieve this, but couldn’t get it better.  She was taking a blood pressure pill (ACE inhibitor) which sometimes causes a relentless cough, and I remembered that chronic cough could also be caused by irritation of the ear canal.  So we stopped the medication and the symptom went away.  To be certain, I had her restart it, and her symptoms quickly returned.  The more medications a person takes, the more likely they are having side effects.

Be willing to wait for an answer.  Stories develop, and sometimes you hear things differently when you’ve heard it the fifth time.  Be patient.

Accept little victories.  While I like to put oil on a patient and cure their symptoms, I usually don’t hit the home run.  It’s often better to aim for a 10% improvement, or improvement of a single symptom, than to fix them all at once.  Over time, a bunch of 10% improvements can make a big difference.

Remember: Some problems go away on their own.  Some things need Father Time, not Dr. Rob to get better.

This all gets back to my role as a physician as a helper, not a healer.  I like to be the medical magician who pulls a diagnostic rabbit out of the hat, but more often I’m the hand that helps people up when they are down, making the most out of a tough situation.  It’s not glamorous, but it’s the way things usually work.  Accept this fact and be pleasantly surprised on the occasion when Occam is actually right.

Rob Lamberts is an internal medicine-pediatrics physician who blogs at Musings of a Distractible Mind.

View 5 Comments >

Most Popular