In 2016, January 15, I asked this question or posted this challenge in one of my blogs:
Challenge: is there a systematic way of calculating subjective probability? Am looking for one.
Subjective probability is a probability derived from an individual’s personal judgment about whether a specific outcome is likely to occur.
I have been looking for published systematic way of calculating subjective probability of clinical diagnosis or probability estimation of clinical diagnosis. I cannot find one since 1985.
Thus, this thoughts, perceptions, opinions and recommendations (TPOR) on probability estimation of clinical diagnosis. Actually, I have been using the probability estimation of clinical diagnosis circa 1985 on my patients and in my teachings (under my Management of a Patient Process). However, I have deferred its formal publication as I wanted to look for published articles to support my practice.
When I came across the two methods of probability estimation, which included subjective probability, I felt more confident in publishing my way of calculating probability of clinical diagnosis (as subjective probability defined as a probability derived from an individual’s personal judgment).
A clinical diagnosis is based on processing of symptom and sign data obtained from a patient. The processes that I use are pattern recognition (primarily) and prevalence (secondarily).
If the clinical diagnosis that I formulated is based primarily on pattern recognition and supported by the prevalence data, then I consider the degree of certainty is sure or quite sure, probable or highly probable.
If the clinical diagnosis that I formulated is just based on prevalence data, then I consider the degree of certainty is not sure.
Other factors that will affect the certainty of the clinical diagnosis are:
- if it is based primarily on sign data and supported by symptom data, then certainty is quite certain;
- if it is based primarily on symptom data not supported by sign data, then certainly is not quite sure as a rule;
- degree of certainty of my physical examination findings;
- reliability of the symptom data from the interview.
At the end of the processing of the symptom and sign data, I will formulate two (2) differential diagnoses, one primary clinical diagnosis (which is the more likely diagnosis) and one secondary clinical diagnosis (which will be the diagnosis if the primary clinical diagnosis turns out not to be the one).
Then, I will estimate the degree of certainty or uncertainty of my differential diagnoses. I start quantifying the degree of certainty of my primary clinical diagnosis. If certain, quite certain and most certain, I assign 90%, 95% and 98% respectively. If uncertain, the range starts with 60% going up to 80%, depending on my estimation of uncertainty. After assigning probability percentage to the primary clinical diagnosis, the balance from 100% will be assigned to the secondary clinical diagnosis. For example: 98% to 2%; 95% to 5%; 90% to 10%; 80% to 20%; 70% to 30%; and 60% to 40% (primary clinical diagnosis – secondary clinical diagnosis).
This has been my way of calculating the probability of my clinical diagnosis, subjective but with a clear process.
Until I find a more systematic way of probability estimation, I will continue my current practice.
This probability estimation of clinical diagnosis is an important step in determining what to do next in the patient management.
The next step is to ask the question: do I need a paraclinical diagnostic procedure?
As a rule, if you are certain, quite certain, and most or very certain of your clinical diagnosis, there is no need for a paraclinical diagnostic procedure.
If you are not certain, as a rule, there is a need for a paraclinical diagnostic procedure.
There are exceptions to the rule. I will discuss them here. If interested, see the following:
Lastly, I say this probability estimation of clinical diagnosis is useful in preventing the indiscriminate use of diagnostic procedures.
Medicine is the science of uncertainty and art of probability.