And the cognitive bias of the week is…..ZEBRA RETREAT!  In my previous blogs I’ve been exploring the importance of generating a broad differential list early in the diagnostic process.  This is particularly important to mitigate against biases such as premature closure and affective bias (‘gut-feel’).  If we have generated a good differential list, it should include some zebras or ‘rare’ diagnoses.Zebra retreat is where a rare diagnosis figures prominently in the differential diagnosis list, but the doctor is reluctant to, or retreats from, making the diagnosis.  This can result in a delayed diagnosis or more importantly, missing a diagnosis not to be missed that requires urgent attention.  The zebra retreat bias is on a spectrum with availability bias (tendency to diagnose based on examples that easily come to mind) and base rate neglect (failure to factor disease prevalence into diagnostic reasoning).  If we are not finding any zebras, we may be over-influenced by availability bias, however if we are finding too many zebras, it’s likely that we are neglecting the prevalence of the disease (resulting in  poor use of resources and overdiagnosis).


An example:

Aesha Anand, aged 3 years, has been booked as a ‘fit-in’ appointment as her mum Trish is worried about a bright red rash that has just appeared on Aesha’s face.   Trish tells you that Aesha has been unwell with a low-grade temperature for a few days, as well as a runny nose, dry cough and watering eyes.  There is also a faint fine rash on Aesha’s abdomen.  Aesha has no other significant medical history.  She attends a local daycare centre a few days a week.  You note that Aesha’s vaccinations are up to date.   Trish also mentions that the family has just returned from a one month stay with Aesha’s father’s relatives in Southern India.

On examination, Aesha is alert and interactive.   She has a temperature of 38.1°C, and her other vital signs are normal.   She is not dehydrated, and does indeed have a confluent red rash on her cheeks and a morbilliform rash on abdomen.   You are aware that there has been recent reports of Parvovirus B19 (‘Slapped Cheek’) at the daycare centre.  Other diagnoses such as measles and rubella cross your mind but you don’t think this fits the description of measles and it is not likely as Aesha is fully vaccinated.   In any case, you are not keen on putting Aesha through a blood test to exclude it (you are unfamiliar with other diagnostic methods) and the extra paperwork of contacting the Public Health Unit with a presumptive diagnosis.   You provide Trish with some information about ‘Slapped Cheek’ and ask her to return in 24 hours if Aesha’s condition has not improved or she has become more unwell.

The next morning, Trish and Aesha are on the doorstep when you arrive at the practice.  Aesha has had temperatures of 40°C all night and is quite miserable in herself.   The rash is now more prominent.  Luckily overnight you realised that your knowledge about measles was a little outdated, so you did some reading about the diagnostic tests.  You decide to order Measles PCR on a Nasopharyngeal swab and first-catch urine just to be sure and to ring the Public Health Unit to notify a presumptive diagnosis.

So why might we be reluctant to find the zebra?

  • we are keen to utilise resources effectively or avoid costly tests;
  • we don’t want to been seen as ‘unrealistic’ in our diagnoses;
  • we are under-confident about our skills;
  • the context or environment is perceived to prevent us from pursuing the zebra diagnosis;
  • peer-pressure may force us to retreat;
  • lack of access to specialists to assist with the diagnosis;
  • lack of knowledge to make the diagnosis confidently;
  • we are tired or distracted.

Let’s consider some strategies to address Zebra retreat:

  • Considering the worse case scenario;
  • Making a deliberate effort to address an outstanding issue, even when it’s not convenient to address it;
  • Ensuring your clinical knowledge is up to date;
  • Considering your diagnostic repertoire compared to your peers – do you diagnose rare things commonly or not much at all?  Is this because of your patient demographic, clinical knowledge, or perhaps due to biases such as availability bias and base-rate neglect.

Remember too, that Zebra retreat is influenced by anchoring bias – “If it looks like a duck, walks like a duck, and quacks like a duck, then it just may be a duck” – unless it is a Zebra impersonating a duck!    To put things into perspective however, there is a more dangerous cognitive error than Zebra retreat – that is not even considering a zebra might be there in the first instance.


Croskerry, P. (2015). 50 Cognitive and Affective Biases in Medicine. In. Dalhousie University: Critical Thinking Program.

Miller, C. S. Skin-deep diagnosis: affective bias and zebra retreat complicating the diagnosis of systemic sclerosis. (1538-2990 (Electronic)).

Morgenstern, J. (2015). Cognitive errors in medicine:  The common errors.

3 thoughts on “Retreat! It’s a zebra…”

  1. My long and complex case has been dogged by many cognitive biases, including Zebra Retreat. When I read Justin Morgenstern’s list of medical cognitive biases my eyes were out on stalks. I could tick so many of them. I have drawn many cartoons about my story and in them I am usually accompanied by my faithful Zebra. I have written up my story, it is long, personal, and I occasionally speculate, but when I do I emphasise that it is speculation. It has been read by Dr Eoin O”Sullivan, who uses the bare bones of it (and my cartoons) to teach his students about diagnostic errors. Dr Pat Croskerry has read it and described it as “erudite and powerful”. If anyone is interested in reading it please contact me and I can send a copy.

    1. Christian Jaehne

      I would like a copy of you work. Just as Rebecca said, I feel there are too few coal-face reports in this area. Thank you in advance.

Please feel free to comment.

Scroll to Top
%d bloggers like this: