clinical decision making

see also:

• neuropsychology of decision making
• clinical decision support systems (CDSS)
• Clinical Decision Making in Emergency Medicine - Prof. Patrick Croskerry - ICEM 2012 webcast

• before we can develop computerised CDSS's we should first look at the cognitive processes that clinicians use to come to a decision whether it be on provisional diagnosis, an investigation or a management option.
• not only does this process require background knowledge, but also the ability to deal with imprecise data - rarely is the information clearly black and white.
• acquiring data is a major skill in itself requiring experienced communication and rapport skills, knowing which questions to ask, when to ask and how to ask them to minimise patients giving incorrect data, insufficient or irrelevant data whether intentional or not - patients do lie, and they lie frequently - that is human nature but there is an art to minimising this or at least detecting it. Whilst computers can store far more data elements than human minds, it is unlikely a computer will adequately replace an experienced clinician in acquiring and analysing such complex data, much less being able to consider pathophysiologic aspects rather than just pattern recognition or probabilistic aspects - hence the danger of internet-based self-diagnosis programs for patients.
• whilst a medical student is necessarily taught to approach this in a very systematic but lengthy process, in a busy ED, the experienced clinician will use their experience and develop their own efficient iterative approach of gaining data, interpreting it, looking for more data and validation of your cognition or pattern recognition process.

• basic demographic data such as age, sex, race, occupation, culture
• past history of illnesses or allergies that may be relevant to the current condition or its management
• current symptoms
• current signs of clinical disease or absence of signs - requires experience in clinical examination and their interpretation.
• available investigation results
• presence or absence of disease risk factors
• patient expectations or preferences - does the patient have a terminal illness and aggressive management would be futile?

• the ability to rapidly extract the important data from the background noise of irrelevant or non-discriminatory data is paramount.
• critically evaluate the relationships of different pieces of information based upon:
  • pathophysiologic knowledge
    • eg. time course of the pain
      • sudden onset suggests a mechanical cause such as torsion or rupture
      • gradual, increasing onset suggests inflammatory cause such as appendicitis or a growing mass causing pressure
  • pattern recognition
    • the classic “spot diagnosis” scenario
    • unfortunately, uncommon presentations of common illnesses are often more common than less common illnesses.
  • probability
    • physicians must deal with substantial degrees of uncertainty
    • consider patterns of disease in the current community - seasonal conditions such as influenza, croup, etc need to be taken into account
  • intuitive thinking
    • subconscious cognition which presumably draws upon elements of the above
  • diagnosis by aphorism
    • simplified mechanisms to memorise complex problems
  • red flags
    • clinicians often use a cognitive concept of red flags - clinical features that act as warnings that all may not be as anticipated.
• listing of findings in order of importance;
• selection of one or preferably two to three central findings
  • an error here may be critical as it may drive the clinician - and their team erroneously along an incorrect care pathway with potentially disastrous results.
  • whilst it is important to consider a finding as a Red Herring, there is a danger that it may not be!
• listing of illnesses in which these central findings can be found;
  • BUT you can't diagnose an illness if you don't think about it or are unaware of it.
  • this is where CDSS that give differential diagnoses such as Isabel and even this wiki can assist the clinician - not by providing the final diagnosis but by jogging their memory so they don't forget to include potential diagnoses.
• the data then needs to be fit into known possible disease patterns and ranked according to likelihood based on the clinician's concept of their prevalence in this scenario whilst considering options of more rare conditions but ones that are potentially life threatening and need further exclusion.
• experienced clinicians are often able to achieve this rapidly through pattern recognition as a result of years of exposure to similar though processes, although significant anecdotal experiences may adversely color their perception of risk and probabilities.
• junior clinicians usually must resort to a more logical cognitive process.

• double check that the cognitive process is valid - perhaps by looking for confirmatory findings or investigations.
• searching for final diagnosis by selection of an illness that better explains the findings of the patient;
• revision of all positive data, in order not to leave any finding considered important without an explanation.
• a danger is the premature conclusion of a definitive diagnosis or syndrome, without the existence of unquestionable data to establish it
• failure to consider reasonable alternatives is a major source of error.

• see Diagnostic Failure: A Cognitive and Affective Approach. Pat Croskerry
• see also:
  • neuropsychology of decision making
  • regret and regret avoidance and decision making