Donald (Raj) Woolever, MD, shows his usual profound understanding of medical heuristic thinking-in “The Art and Science of Clinical Decision Making” [May 2008]. I would like to take his probabilistic thinking one step further.
The real issue isn't the probability of a disease being present but the probability of significant harm from a disease. For example, a middle-aged male with chest pain may be at low risk (say, 5 percent to 10 percent pretest probability) for ischemic heart disease, but the consequences of missing ischemia are severe, thus the pretest probability of harm from ischemic heart disease is significant. The likelihood of costochondritis and reflux may both be moderate (say, 40 percent), but the consequences of initially missing these are minimal. We subconsciously integrate the probability of the disease being present with the probability of harm if we misdiagnose.
This process guides us as we sort through our individualized evaluation strategy, appropriately excluding some low-probability diagnoses while leaving the more likely diagnosis on the “back-burner.” If we fail to incorporate this probabilistic thinking, we will either jump to conclusions with potentially devastating results or risk causing harm and wasting resources by chasing every rabbit trail in the differential.
The final element is recursively reevaluating after the diagnosis is made. Do the treatments work? Is the patient following the expected pattern of recovery or deterioration? If not, we get to modify our probabilities and start over. Is it a failure of treatment or a failure of diagnosis? This is the power of follow up that we as family doctors excel at. We don't have to exclude every low-probability item on the differential because the final and most powerful test we run is the test of time.
I challenge clinicians to start thinking in terms of probabilities – high, medium, low and very low – and apply the probability of harm when making diagnostic decisions. Try it for a week; it may help sharpen your intuition as a diagnostician and help to get out of the trap of knee-jerk responses to presenting symptoms.