Clinical reasoning: Identifying errors and correcting

Yesterday I attended a presentation on clinical reasoning by Professors Vanessa Burch (University of Cape Town) and Juanita Bezuidenhout (University of Stellenbosch). Here are the notes I took during the presentation.

  1. How does CR work?
  2. How do errors occur?
  3. Do clinician educators contribute to errors?
  4. Can we identify students with CR difficulties?
  5. Can we improve CR skills?

How does CR work?
Graphical representation of the clinical reasoning process by Charlin et al. (2012).

Graphical representation of CR, from Charlin et al. (2012).
Click on image to enlarge.

High level CR appears to be intuitive but is really pattern recognition that happens as a result of lots of experience.

Students don’t have the illness scripts (i.e. patterns to recognise clinical presentations / clinical knowledge organised for action) and so they spend more time in System 2 reasoning, rather than system 1 reasoning (see Charlin et al, 2012). Side note: for additional detail on how pattern recognition actually works, see Stephen Pinker’s book, “How the mind works“.

Are we mindful of the complex thinking processes that make up CR, and do we expect students to be operating at the same level? Do we explicitly tell students about the CR process or expect them to “absorb it”?

We can act on illness scripts without acknowledging that they exist. This is why awareness of our behaviour (i.e. metacognition or mindfulness / reflection in action) is so important. System 2 processes act as a balance to prevent acting on patterns that are similar but not the same. This could be the basis for CR errors. See below the process from Lucchiari & Pravettoni’s cognitive balanced model that describes a conceptual scheme of diagnostic decision making.

Click on image to enlarge.
Click on image to enlarge.

It is also important to be aware that belief systems (i.e. cognitive biases and heuristics) exist, and that they can influence behaviour / decision making, which may lead to CR errors (Lucchiari & Pravettoni, 2012). See image below.


Novice practitioners tend to miss subtle differences in clinical presentations. Students must articulate their reasoning processes so that you can help them to link the facts (i.e. the clinical information) to the diagnosis. If the student missed the conceptual relationship between variables, they are prone to making mistakes.

Audétat et al (2012) use Fishbein’s integrative model of behaviour (and associated belief systems) to explain why managing clinical reasoning difficulties is so challenging (see below).

Click on the image to enlarge.
Click on the image to enlarge.

There is a tendency, in the clinical context, to emphasise service delivery above all else, with educational needs taking a distant second place. In other words, increase the students’ case load with little thought given to how this may impact on their learning (or the actual management of the patient). The clinical environment is therefore almost always not a very good educational environment that is conducive to learning.

Clues to identify students with CR difficulties:

  • Often not aware that we’re in System 2, while students are in System 1 → talking past each other because we’re in different spaces.

Clues at at the bedside:

  • Limited semantic transformation of patient interview. Student unable to do anything with the information at hand.
  • No logical clustering of complaints. The student can’t categorise like information in a clinically logical way.
  • No order of priority attributed to complaints. Students can’t decide what the most important problem is.
  • Key information not obtained during patient interview. Student doesn’t think to ask important questions → non-existent or faulty illness scripts (non-existent illness scripts are less dangerous than poorly configured ones because it’s easier to correct).
  • Physical examination excessively thorough or cursory. Student unable to make reasonable progress through the case.
  • Too many investigations ordered.
  • Inability to interpret results of investigations. Student unable to articulate a reasoning process, or they reason incorrectly, when confronted with a different set of variables e.g. X-ray, rather than a patient.

Strong beliefs in incorrect illness scripts can make novices see things that aren’t there e.g. seeing pneumonia on an X-ray that is clear. Belief systems are powerful drivers for behaviour.

CR errors are often left “unfixed” because trying to do it in the clinical context is too time consuming. These should be addressed later.

Other ways to see CR errors:

  • Discharge letters and case notes may be unstructured and lack clarity. Lack of illness scripts (or faulty ones) prevent students from linking concepts, which is evident in how they write narratives.
  • Too much / little time spent with the patient.
  • Emotional reaction to students: negative affect on the part of the patient (ask patients how they experienced the student’s management), or on the part of the clinician (there’s something about the student – that isn’t related to rudeness or some other inappropriate behaviour – that you find upsetting.

Can CR be taught?

Every clinician thinks differently.
There is no right or wrong way to think.
Diagnostic competence requires knowledge.

The challenge is to:

  • Organise accurate knowledge in a user-friendly way. This is about developing appropriate semantic networks / conceptual relationships.
  • Create rapid access routes to the knowledge. Create opportunities to access the semantic networks quickly.
  • Provide enough opportunities to use the pathways. Practice, practice, practice.

Avoid students thinking that they don’t know the diagnosis. Help them to move towards thinking or knowing the diagnosis.

The key to success is structured reflection. How do we get into their heads, and how do we show them what is in our heads?

Reflection must be structured because it doesn’t help for the student to keep thinking the wrong thing. It’s no good asking the student to “have another go” because they just gave it their best shot. When the student keeps guessing the wrong answer (or, even if they guess the right answer), it’s not useful.

How do we get students to “think again” (i.e. System 1 and 2 thinking) in a structured and explicit way?

  • Prioritise 3 possible diagnoses
  • Column 1: What fits the diagnosis (Yes)? This identifies if they have an illness script. Begin by removing the diagnoses that definitely don’t fit, so that they don’t continue with the faulty illness script.
  • Column 2: What doesn’t fit the diagnosis (No)?
  • Column 3: What do you still need to find out (Data needed)?

This process will  help students to articulate an illness script in a structured way. The steps require that you explicitly articulate your (i.e. the clinician’s) own thinking process. Students could also write a narrative explaining their reasoning process for the different columns.

Anxiety and loss of self-esteem will cause students to crash and be unable to take in anything that you say. You must first create an environment where they can take articulate their thinking process. It’s not about giving them the answers or the facts, it’s about taking them through a reasoning process.

We cannot help students think on a case by base basis. There are too many cases. We need to help them to work this out on their own.


  • Audétat, M.-C., Dory, V., Nendaz, M., Vanpee, D., Pestiaux, D., Junod Perron, N., & Charlin, B. (2012). What is so difficult about managing clinical reasoning difficulties? Medical education, 46(2), 216–27.
  • Lucchiari, C., & Pravettoni, G. (2012). Cognitive balanced model: a conceptual scheme of diagnostic decision making. Journal of evaluation in clinical practice, 18(1), 82–8.