“What are you looking for?” I asked my supervising attending physician. I had been the intern on the bone marrow transplant service for less than a week. “Which numbers are important?” I wanted to know. “You listen every day as I rattle off hundreds of numbers in a few minutes, with no emphasis or pause, on one patient after another. You don’t write a thing down, yet tomorrow you will remember what the bilirubin level was today. I doubt you’ll remember what all the labs I said were, so which ones do you care about?”
I was lost in the data. On my other services I could put the pieces together. While rattling off the numbers of my critically ill patient I could link the fever with the low blood pressure and the high white blood cell count and arrive at a suspicion for septic shock. I would ignore the other hundreds of data and hone in on the very few – less than five if our cognitive psychologist colleagues are correct, that fit into our short-term memory. I could put those five together successfully. Even more importantly, I could figure out the right five to put together.
But on the bone marrow transplant service I was lost. These were sick people who developed a variety of complications that were unknown in other fields: Graft versus Host Disease, which is when the healthy bone marrow cells from the donor (graft) attack the cells of the leukemic recipient (host). This disease can be present only in patients on this service. The massive immunosuppression with the conditioning regimens (lethal doses of chemotherapy and/or radiation) allows a variety of unusual infections to prosper. The radiation and chemotherapy themselves can result in inflammatory responses by the body that lead to organ injury, and of course, all the “regular” afflictions that are seen in Intensive Care Units across the country do not make exceptions for the bone marrow transplant patients.
And this is the thing. For although doctors are generally smart, how else could we get through medical school, we’re not that smart. We are still human first, and our short-term memory is limited to only five objects. I knew my attending didn’t have a better memory than me. Or perhaps he did, but not that much better. What he had was a context for hanging together all these variables, a context he developed over years of becoming an expert.
This principle of short-term memory was illustrated with chess players, and applies to medicine, as it does to all other disciplines. The researchers showed chess players of varying abilities a position from an actual chess game. They then asked them to recall the exact position of the more than 20 pieces on the board. The more expert the chess players were, the greater the number of piece positions they could recall exactly. But when they showed the players a board of randomly arranged pieces, the expert’s memory was shown to be no better than the novice’s. The expert chess players didn’t have better memories than the novice players. What they had was a context for remembering the piece positions together.
And so, too, my attending, the expert, could recall far more variables about a patient than I, the novice, could. He knew that some diseases presented with a constellation of symptoms, laboratory and imaging abnormalities. These hung together for him, but seemed to me a random assortment of abnormalities. He knew which diseases were likely to occur in this population of patients, and so he focused on those particular data. If there was a piece of data that should’ve hung together in this illness and I didn’t provide it in my barrage, he asked for it. The hundreds of other data he ignored.
In the modern era of medicine, we are so overwhelmed by data, that the challenge of picking out which five elements to focus on is enormous. Often diagnoses are delayed or missed because an additional clinical study, already done and in the chart, but the Nth piece of information, is ignored or simply “not seen.” In many ways reading a patient’s chart is akin to asking the patient questions. We cannot just read the notes in the chart and review the studies they had done. The electronic charts are full of information, most of it irrelevant, occasionally some of it wrong. So instead, we do a “chart biopsy”: we interrogate the chart. We search for the studies and tests we would like to order to clarify the diagnosis. Did the patient have a chest X-ray already? A CT scan? Any evaluation for blood clots? Any heart studies?
We, doctors and patients alike, are trained to think that physicians consider all the data, and then carefully integrate what we consider to be the main pieces of data, and make a decision based on that interpretation. Indeed, this is what doctors aspire to. But the reality is that the five factors on which we base our decisions can vary significantly. They can vary among different physicians and they can vary within the same physician on different days, indeed sometimes on the same day! This variation in healthcare decisions has been increasingly illuminated recently and accounts for at least some of the variability of medical care. It seems reasonable that faced with the same information, clinicians ought to come to the same conclusions with some consistency. But we do not, and at least some of the problem seems to be the huge influx of information we are meant to process, and that we process it inconsistently.
* Please note identifying patient details, such as name, have been altered or omitted to protect patient privacy.