We have seen in the last post that uncertainty constantly dogs diagnostic decision-making, a fact that makes it incumbent on the doctor to keep an open mind throughout the process. The need for doctors to be prepared to change their minds about any diagnosis they make is a point emphasised by Kevin Eva and colleagues in their paper titled Swapping horses midstream: factors related to physicians’ changing their minds about a diagnosis. The authors pertinently pointed out that closing a diagnosis early opens the door to premature closure – one of the major causes of diagnostic error. It is in this context that they advocated the concept of diagnostic flexibility – an attitude that enables physicians to change their diagnosis. Based on their survey of more than 250 physicians, they found that older doctors were most unlikely to change their mind when they are exposed to additional information that contradicts their initial impressions, even though they were more likely to make the right diagnosis the first time. The authors therefore concluded that ‘early hypotheses are more likely to be accurate when one has more experience on which to draw’. Whilst diagnostic flexibility is a virtue for all doctors, it is clearly more important the less experienced the doctor.

Perhaps the best protection against premature closure, and therefore a key feature of diagnostic flexibility, is the generation of a list of differential diagnoses – particularly in complex cases. This point was stressed by Jason Maude in his paper titled Differential diagnosis: the key to reducing diagnosis error, measuring diagnosis and a mechanism to reduce healthcare costs. Maude argued that a list of differential diagnoses serves as ‘a vital trigger to stimulate thinking at the time of the consultation’,  and that it is ‘the most accurate indicator of diagnostic accuracy‘. He asserts that not compiling a differential diagnostic list, and compiling one that is not sufficiently broad, are major causes of diagnostic error. Apart from helping to ‘refute competing hypotheses‘, and to firm up the correct diagnosis, Maude adds that the differential diagnosis also guides the rational use of investigative tools.

Whilst most doctors are trained on the need to generate a list of differential diagnoses, many are not aware of the best way of creating and making use of this. A helpful approach to making differential diagnoses was outlined by Fredric Wolf and colleagues in their paper titled Differential diagnosis and the competing-hypotheses heuristic. A practical approach to judgment under uncertainty and Bayesian probability, an approach that is itself based on the method of David Eddy and Charles Clanton outlined in their paper titled The art of diagnosis-solving the clinicopathological exercise. The process entails the generation of ‘a cause list’ after identifying what the pivot feature is (this is discussed in a previous post titled The art and science of making a diagnosis) The physician then prunes down the differential diagnostic list by the technique of ‘pairwise heuristic‘ – serially assessing two items on the list against the pivot feature, starting with the two most likely diagnoses and then going down the list. The diagnosis that eventually emerges is then validated.

Another method of maintaining diagnostic flexibility is the time-honoured technique of follow-up. This approach was explored by Donald Redelmeier in his paper titled Improving patient care. The cognitive psychology of missed diagnoses, where he advocated the use of follow-up to overcome the ‘cognitive fallibilities‘ that shadow the diagnostic process. Redelmeier views follow-up as ‘a feasible strategy to prevent cognitive shortcuts from causing harm, since it allows clinicians to reconsider the entire picture from an alternative perspective‘. He argued that follow-up is an opportunity to overcome such biases as availability and anchoring which frequently influence the initial diagnosis. Follow-up, he maintains, counters the influence of these biases by ‘providing more distance from initial impressions‘, and by enabling verification of the diagnosis by consulting evidence-based sources. Redelmeier also viewed follow-up as a means to prevent premature closure‘ because it enables the doctor ‘to reconsider the case when he or she is less fatigued‘.

Perhaps the most important patient safety benefit of follow up that Redelmeier highlighted is the opportunity it offers ‘for corrective intervention’ if the initial diagnosis was wrong. Consequently, he advocated follow-up as an ‘opportunity to learn from mistakes for the benefit of future patients’. He however cautions that follow-up is only effective in this regard if it is done early, before any irreparable damage occurs. Other benefits of follow up the author explored, apart from inhibiting cognitive errors from impairing decision-making, were its value in reducing healthcare costs and improving patient satisfaction.

In the next post, we will take a look at the cognitive drivers of diagnostic failure.