The basic premise of the book is that our brains actually function via two systems, which he terms System 1 and System 2. System 1 is instinctive, utilizing heuristics to make decisions quickly. A heuristic is a 'rule of thumb', a device to boil down complex information to a simple choice. In other words, it acts as an approximation, that is usually 'good enough' for the situation at hand. System 2 is calculating and rational; it functions more like a computer, coolly processing inputs and trying to generate the optimal decision as an output.
Kahneman's work points out the biases inherent to both of these systems. His most famous result may be prospect theory, which basically shows that humans do not value incremental changes in probability equally, and that they are susceptible to whether those changes are framed as gains or losses. Another big point that Kahneman makes is that our "selves" really consist of two separate entities: our experiencing selves, and our remembered selves. For example, think about a strenous hike or other physical activity you did. In the moment, your experiencing self may be in agony from the stress and pain; however, your remembering self may view it as a very cherished memory because of the overall payoff.
What does this have to do with medicine? Everything! Both patients and doctors suffer from these biases. While the biases are part of being human, that does not mean we should simply accept making poorer decisions because of them. As a physician, here are some things you can do to mitigate these biases from how you decide.
Gathering A History
When you first meet a patient, you will necessarily have to make an initial assessment. That 'subjective' portion, the S in SOAP note, is the lynch pin for the rest of your encounter. If you have bad information, you will order the wrong exams, make the wrong diagnosis, and implement the wrong plan.
Garbage in, garbage out.
Therefore, getting a precise history is critical. However, to do this, you must combat both your own biases and the patient's. The big one to combat here is recall bias. For example, patients with a history of cancer who present with abdominal are more likely to over-emphasize or recall factors that may suggest a mass as the underlying cause of their pain. Do not disregard this history, but do try to correlate it with objective data, such as a CBC to check for anemia, or imaging.
The explosion in objective data one can acquire on a patient has been a major boon to healthcare. However, there is a downside to such tests. This bias is on the provider's side. Whether it is an anchoring/recall bias (the last patient with shortness of breath had a pulmonary embolism, so now every future patient with SOB has a PE) or satisfaction of search (one abnormal test result precluding ordering other relevant tests), these biases can lead to either the under- or over-utilization of appropriate testing modalities.
One way to guard against this is to understand treatment algorithms at great length. While there is a tendency to avoid 'algorithmic thinking' in medicine, the notion is often misapplied. One can go beyond algorithms only after they have fully mastered them, and can confidently assess that the current patient does not fit the algorithm. This is a wholly different matter from simply ignoring an algorithm altogether.
Making a Plan
Finally, once all the history and data have been gathered, and an assessment, it is time to come up with a treatment plan. As Kahneman's book title implies, this is a good time to think slowly. Have all the patient's problems been addressed? Does every abnormal lab value have either an explanation or a plan to address it? A good treatment plan will also include some way to assess its own effectiveness, whether that is by scheduling a follow up exam or test, or some other objective measure. A plan without follow up is simply wishful thinking.
Ultimately, the practice of medicine is a human practice, subject to human biases. However, as our understanding of these biases advances, it behooves us to mitigate these biases to the best of our abilities. If we are to do no harm, we must ensure that we unblind ourselves to the harm our biases may cause.