Cognitive errors can lead to mis-diagnosis and incorrect treatment. In the latest issue of Medisch Contact (a Dutch medical journal), the issue of cognitive errors in diagnosis is reviewed and the use of diagnostic checklists is proposed.
Errors can be grouped into three categories:
- No-fault errors – Errors which cannot be reliably eliminated. Examples include a missed diagnosis of a newly-emerged disease that has no current diagnostic tests, or an atypical presentation.
- System errors – Errors caused by organizational problems such poor communication, technical errors or insufficient supervision.
- Cognitive errors – Errors in clinical reasoning.
Currently there is significant effort in most health systems to identify and reduce system-errors. However, attempts to reduce cognitive errors also warrant ongoing attention. Traditional medical training emphasizes a systematic approach to history taking and examination of patients, with the goal of avoiding initial biases and generating an appropriate differential. However, as clinicians become more experienced they often adopt a more heuristics-based diagnostic approach. Experienced practitioners use heuristics such as pattern-matching and learned likelihood ratios as a shortcut to more rapidly focus on a diagnosis and to limit the differential.
This approach can be very efficient and accurate. For example, while driving a car on a hot day it is not uncommon to see a mirage that may be interpreted as a puddle of water on the road ahead. Inexperienced drivers may consider the possibility of water on the road in their differential and change lanes to avoid the hazard. Most experienced drivers quickly and correctly recognize the mirage as a visual illusion. In this manner, the heuristic allows safer operation of the vehicle by limiting distraction and preventing an unnecessary lane change. Without the use of heuristic shortcuts in medicine it is likely that costs would rise considerably while many patients would undergo unnecessary investigations.
However, the study of cognitive psychology has identified common cognitive errors that can arise when using shortcuts to make a diagnosis. One of these is termed the “availability heuristic“, in which diagnoses that the physician is most familiar with or that they have most recently seen are considered the most likely. Another is the “anchoring heuristic” which occurs when an initial diagnosis fails to be re-evaluated on the basis of additional, contradicting evidence. A related error occurs with “premature closure“, when alternative diagnoses are not considered after the initial diagnosis is made. “Blind obedience” occurs when the clinician puts excessive faith in technical results or the opinion of an expert, thereby limiting personal consideration of alternative hypotheses and making an incorrect diagnosis. Finally “framing” occurs when the facts of the case are presented in such a way as to lead others to the same, incorrect conclusion.
The authors of the Medisch Contact article suggest that the use of a diagnostic checklist may help clinicians to reduce errors by raising awareness of potential cognitive errors. They suggest an approach to diagnosis, although admit that this is not specifically evidence-based and has not yet been tested or validated in practice.
Step 1: Review the medical process.
… completed a comprehensive history?
… completed a comprehensive physical exam?
… developed a differential?
… attempted to rule out my initial diagnosis?
Step 2: Diagnostic time-out
… considered my biases?
… considered the impact of my own emotions (empathy, frustration, etc.)?
… considered the impact of fatigue?
… planned appropriate follow-up?
… ensured that the most serious potential diagnosis has been considered?
A standardized approach to diagnosis that explicitly attempts to compensate for frequent cognitive-errors is a worthwhile endeavour. However, there is inherent conflict between the goal of reducing cognitive error and increasing efficiency. For instance, a multi-disciplinary team may be able to provide high quality care more efficiently. But this efficiency is realized, for a large part, through the delegation of tasks from high cost individuals to lower cost individuals. An attempt to reduce “framing” biases, resulting in each team member repeating the work of the others, would quickly wipe out efficiencies resulting from the team approach. An alternative is to have low-cost team members completing systematic and comprehensive investigations, followed by factual reporting and only then by a debate on the basis of each team member’s own diagnostic interpretations. This would almost certainly lead to improved results and possibly with only minimal loss of efficiency. In fact, it sounds a little like an idealized version of ward rounds. Maybe the decades of refining medical training have not been entirely off-base!
Future solutions should support the heuristic approach of expert decision-making by automatically compensating for frequently made or predictable cognitive errors. A decision support system that can suggest alternative diagnoses and appropriate additional investigative steps would be a very useful tool and should lead to higher quality care.
1. Klein J, van de Meerendonk H. Diagnostische denkfout is te voorkomen. Medisch Contact. 2012; 27: 1648-1651.
2. Redelmeier RA. The cognitive psychology of missed diagnoses. Ann Intern Med. 2005; 142: 115-20.