Making a diagnosis is the key initial activity of all medical interactions. A successful medical consultation is predicated on arriving at an accurate diagnosis, whilst a misdiagnosis is the harbinger of poor patient outcomes. Whilst all doctors are taught the symptoms and signs of diseases, and their appropriate investigations and treatments, the process of making a diagnosis itself is often overlooked in medical education. It is therefore pertinent to review the components of, and the principles behind, the diagnostic decision-making process.

A very helpful approach to understanding the process of diagnosis was provided by David Eddy and Charles Clanton in an article titled The art of diagnosis-solving the clinicopathological exercise. They described six stages in the diagnostic process which are:

• • Aggregation of elementary findings to reduce the size of the problem
  • Selection of a pivot which is the pathognomonic finding
  • Generation of a cause list
  • Pruning of cause list by fitting pattern to disease
  • Selection of diagnosis by excluding unlikely causes
  • Validation of the diagnosis

The first stage in diagnostic decision-making is to make the information collected manageable, and to make sense of the aggregated data. In this regard, chunking – the grouping similar information together – is a handy devise that helps to manage far-flung information and complex clinical presentations. Explaining the concept in their paper titled Chunking mechanisms in human learning, Fernand Gobet and colleagues show that chunking is an important mechanism in perception, learning and cognition; by bringing together ‘elements having strong associations with one another’,  they said it overcomes the natural limits of peoples’ cognitive capacity and increases their ability to deal with large amounts of information. Mirko Thalman and colleagues also explored the concept of chunking in their paper titled How does chunking help working memory where they emphasised that it also helps to keep information within working memory.

It is clear that many medical diagnoses do not require much effort after the information-gathering stage. This is especially the case where the clinical presentation is straight-forward, and where the doctor is experienced. In such cases, the diagnosis is often made based on pattern recognition, a decision-making tool which Arthur Elstein and Alan Schwarz describe as the product of ‘mental models, abstractions, or prototypes‘ acquired with experience. Writing in their paper titled Clinical problem solving and diagnostic decision making: selective review of the cognitive literature, Elstein and Schwartz argue that those who are best at pattern recognition are those who ‘have constructed more diversified and abstract sets of semantic relations, a network of links between clinical features and diagnostic categories’. They therefore emphasise that pattern recognition ‘is highly dependent on the clinician’s mastery of the particular domain‘.

One of the most insightful explorations of pattern recognition is by Gary Klein in his book The Power of Intuition. Klein puts pattern recognition at the heart of his recognition primed decision-making (RPD) model, arguing that the intuition that drives pattern recognition is not a random phenomenon, but the result of experience. He portrayed pattern recognition as an effective decision-making strategy when it used by experts, but it is limited and risky when used by novices, or when the ‘task is complex and uncertain’. Klein further delineated the best settings for intuitive decision-making in another book, Sources of Power. Referring to pattern recognition there as naturalistic decision making‘, Klein pointed out that it works best when it is used in ‘field settings‘ where the stakes are high, and where there is ‘time pressure, inadequate information, ill-defined goals and poorly defined procedures‘. Outside of these situations therefore, pattern recognition is a risky diagnostic strategy.

In effect, whilst Gary Klein extols the virtues of expert pattern recognition, he advocates analytical decision making for the vast majority of routine cases where the problem is novel or complex; where the practitioner is a novice; or where there is sufficient time. Similarly, whilst Elstein and Schwartz show that ‘easy cases can be solved by pattern recognition’, they stress that more complex cases ‘need systematic generation and testing of hypotheses’. It is in these situations therefore that analytical thinking becomes critical especially in generating and pruning the cause list or differential diagnosis, and in establishing the correct diagnosis. The first step in this strategy is the selection of the pivot symptom or sign, a crucial activity because the pivot focuses attention on the most important clues when approaching complex clinical cases. The pivot helps to narrow down the list of possible differential diagnoses, and it facilitates the making the correct diagnoses.