Bayes rule is probably one of the most brilliant inventions ever because it helps people to correctly determine the probability of an event happening. This is also true in clinical scenarios where healthcare practitioners find using probabilities to determine how likely a patient may be having a certain condition. This probability is mostly used to recommend medical tests and making referrals to specialists. For example, a general doctor may determine that a patient may be exhibiting signs of depression and refer them to a psychiatrist for further review. However, the psychiatrist needs to actually determine the probability that the referred patient suffers from depression or not. To do that, the specialist needs to take into account the sensitivity and specificity of the depression test(s) used and the statistics on prevalence of the condition on certain populations.

According to Tiemens et al. (2020), clinicians need to accurately determine the posterior probability of a patient having a certain condition by considering all the three aspects of the Bayesian Theorem: the pre-test/prior probability (prevalence), test results, and psychometric properties of the test (sensitivity and specificity). Clinicians will mostly likely not include the pre-test probability  and this has disastrous results in terms of misdiagnosis. In my experience, I have found that physicians have a higher chance of misdiagnosing a recurring condition and this has been explained by Bayesian Theorem. Most likely, I would believe that they find using the Bayesian Formula too complicated to determine the correct probability.


Tiemens, B., Wagenvoorde, R., & Witteman, C. (2020). Why every clinician should know Bayes’ rule. Health Professions Education, 6(3), 320-324. https://doi.org/10.1016/j.hpe.2020.05.002


Hi Allen,

Like you, I was also fascinated by the Bayesian Theorem article and it became more clear how of a particular even or condition probabilities could change based on the probability of prior condition or event and the correctness of a measure. During Covid-19, applications of Bayes Theorem in clinical testing of Covid -19 became not only critical, but a matter of life and death. Without the knowledge of Bayes rule, a person who tested positive for covid-19 would believe that the probability the test result was erroneous would be P(tested positive / do not have covid-19) which is the false positive rate. However, the actual probability would be  P(tested positive / do not have covid -19) * P(do not have covid-19) / P(tested positive). When you input actual figures, you will realize the resulting values would be very different, and based on the sensitivity and specificity of the tests, that could have meant a lot if the person was less or more likely to have the disease. Overall, knowledge of Bayes Theorem is critical and its application in clinical practice is very beneficial.

Hi Patel,

I liked your discussion this week mainly because it highlighted how lack of proper information sharing or flows could compromise the health of the patients. Although physicians rely on probabilities to determine the conditions that the patients could have, the way these probabilities are arrived at is also critical. For example, a key component of Bayesian Theorem is that you have to include the prevalence or pre-test probability of the patient having the condition based on statistics unique to the population group that the patient fits. I have found in my experience as a patient that doctors may only rely on post-test probability and the psychometric properties of the test used. In the case you have given, it is also critical that doctors make considerations based on information or suggestions from other experts to recommend further tests based on underlying conditions. I believe that doing so would result in fewer cases of misdiagnosis.