Making a correct diagnosis is the central cognitive endeavor of every clinician, since an accurate diagnosis usually leads to appropriate treatment. As PAs and NPs, we dread missing a diagnosis: jaw pain that turns out to be angina or back pain that ends up being an aortic aneurysm. And then of course, there are preventable infections and medication errors.
Several studies in the medical literature indicate that misdiagnosis occurs in 15% to 20% of all cases; in half of these, the patient is harmed. The vast majority of misdiagnoses, about 80%, are due to cognitive errors—in other words, errors in thinking!
This was recently brought home to me through an online course required by my malpractice insurance carrier. The focus was cognitive errors in diagnosis. Prior to starting, I expected it to be a no brainer. After all, every clinician knows what malpractice entails and how best to avoid it, right? We have attended CME courses and read plenty of articles on the topic.
Well, this course was different and had a rather sobering effect on me. As a result, I started to ponder the process we go through to establish a diagnosis.
For the most part, formulating a diagnosis is largely subconscious, and our ability to do it increases with experience.1 The normative model is Bayes’ theorem, an application of conditional probabilities. Clinicians use information on the prevalence of various clinical features in different disease entities to determine the probability that a particular condition is present. The theorem is considered a milestone in logical reasoning and a conquest of statistical inference, although it is still treated with suspicion by most clinicians.
What makes this method potentially impractical is the complexity of the calculations and the fact that not all required information may be readily available. Would you agree that it is impossible to search for and consider all required information or evaluate all possible hypotheses? Therefore, the search for the correct clinical diagnosis is limited to satisfactory explanations within the constraints of the clinical environment.2
Another model for clinical diagnosis is the hypothetico-deductive model. Clinicians achieve a diagnosis by generating multiple competing hypotheses from initial patient cues and collecting data to confirm or refute each. This model has been validated through empirical studies.3 Most clinicians, in my experience, use a combination of intuitive, reflective, and analytical problem solving, with some approaches given more emphasis than others.
On the next page: Categories of diagnostic errors >>