Department of Medicine

Ambulatory Care Course (MS3) - clinical reasoning

When you hear hoof beats, it must be a zebra.

Before you start.

  • Avoid competing demands on time and emotions
  • Have quick accesss to good knowledge by choosing good resources and using them often enough to be facile with them

Medicalization of the hypothetico-deductive method of scientific discovery (Popper 1959. ISBN: 0415278449)

Steps (Eddy 1982. PMID: 7070446) Type of
logical reasoning

Organizing information

  1. Aggregation of groups of findings into patterns in order to induce hypotheses. Alternatively, use rules, algorithms, causalisty
    1. make a one to two sentence summary (PMIDs: 9795669, 1930535, 12191059)
    2. Biases to avoid when doing this (Leape 1994. PMID: 7503827)
      1. Biased memory - overuse of dramatic anecdotes
      2. Availability heuristic - using first idea that comes to mind
      3. Confirmation bias - ignoring new data that is contradictory
 

Building up ideas

  1. Selection of a "pivot" or key finding
    1. highly specific or unusual
  2. Building up ideas - generation of a cause list (hypothetical)
    1. Avoid premature closure (PMID 3736379 and PMID 15857490)
    2. Tools to help:
      1. Google (PMID: 17098763)
      2. PubMed. Use handheld version of clinical.uthscsa.edu (BMJ)
      3. DxPlain. Use clinical.uthscsa.edu (PMID: 8190157)
 Induction and abduction

Reducing and pruning ideas

In all three of these steps, avoid base rate neglect and conservatism (Anderson 2000. ISBN 0716736780. p337 and PMID 3947184)

  1. Pruning ideas - reducing the list of possible diagnoses
    1. Be careful with your logic and void over-generalizing statements that you read. (specifically, converses may not be true)
    2. Avoid anchoring biases (when "occurs when a prior belief has a stronger than expected
      influence on the way new information is processed and new beliefs are formed" - http://dx.doi.org/10.1197/jamia.M255):
      1. Biased memory - overuse of dramatic anecdotes
      2. Availability heuristic - using first idea that comes to mind
      3. Confirmation bias - ignoring new data that is contradictory
      4. Overconfidence
  2. Selection of a diagnosis(s)
    1. Consider using Occam's razor
    2. Conciously review the reasons for and against an idea (http://dx.doi.org/10.1002/acp.1068, http://dx.doi.org/10.1197/jamia.M2557). DxPlain can help with this.
  3. Validation of the diagnosis
    1. Consider using Sutton's law
    2. Avoid competing heuristic bias (PMID 3385753)
 Deduction