How to understand medical decisions

Blogs & discussion

Welcome to my blog

 

Here you can add some text to explain what your blog is about and a bit about you.

How is the use of a diagnosis justified?

By HUW LLEWELYN, May 10 2021 09:13PM

It is important to remember that it is not possible to identify all those and only those with a disease. The best that can be done is to assume or postulate that the disease is present based on some agreed criterion - this is what a diagnosis means. A failure to make this distinction between disease and diagnosis and to agree a diagnostic criteria will lead to endless confusion about diagnostic tests (as in the case of Covid-19 when trying to assess the sensitivity and specificity of a test result without agreed diagnostic criteria). It is only if we have clear diagnostic criteria that we can estimate probabilities of 'diagnoses' from other symptoms, signs and test results.


Sufficient and necessary criteria


A ‘sufficient criterion’ is a finding or combination of findings that justifies adopting a diagnosis as a hypothesis and acting on it so that it becomes a working diagnosis. There may be many such sufficient criteria but they may not account for all patients with the disease (as in Covid-19). From biological theories, they may be more likely to cover the severe forms of the disease or at least when symptoms are more prominent. The very early or mild forms cause more difficulty. New sufficient diagnostic criteria may be created all the time to cover more patients with a disease but they may never cover them all.


A ‘necessary’ criterion is a finding that occurs in all patients in whom using a diagnosis is justified but may also happen in many other diagnoses too. The absence of such a criterion means that the use of the diagnosis is ruled out. (Other diagnoses that the necessary finding covers may be ruled out too). However, if it considered ‘necessary’ for at least one sufficient diagnostic criterion established already to be present then the absence of all the known sufficient criteria rules out use of the diagnosis. This rule would then become a definitive diagnostic criterion because it identifies all those and only those in whom use of the diagnosis is justified. (However it is not a definitive disease criterion that rules out hidden disease that can spread to others as in Covid-19).


Arriving at diagnostic criteria


Diagnostic criteria can be arrived at from theoretical reasoning, shared experience, or agreements between experts. The entry criterion for a randomized controlled trial (RCT) can be arrived in the same way. If a RCT suggests benefit, its entry criterion can be adopted as another suffcient diagnostic criterion. A test or scoring system that predicts an important outcome can also be incorporated as a sufficient criterion. These would be the criteria that predict benefit more accurately. In order that patients are considered for such interventions, the set of patients shown to benefit in a RCT should be a subset of those with one of the suffcient diagnostic criteria for a diagnosis at least.


Simplifying diagnostic criteria


Diagnostic criteria that share findings can be combined to form broad criteria that will include patients who benefit for different reasons. in other words, many suffcient criteria (e.g. Type 1 Diabetes Mellitus, Type 2 Diabetes Mellitus, etc.) can become subsets of a broader sufficient diagnostic criterion (e.g. Diabetes Mellitus). If a particular treatment or demanding test is suggested by such a broad criterion, the decision to go ahead may also have to be supported by the original evidence of benefit for that particular intervention.


Oxford Handbook of Clinical Diagnosis


The Oxford Handbook of Clinical Diagnosis (access Chapter 1 from this link) outlines many of the broad suffcient diagnostic criteria. They are presented under the headings ‘Confirmed by’ (or ‘Affirmed by’).The book explains how diagnoses are arrived at by navigating between possible diagnostic criteria when a patient seeks help. This will involve estimating the probability with which various diagnostic criteria will be satisfied by using limited information.

Add a comment
* Required
RSS Feed

Web feed