The diagnosis can be confirmed or excluded with a single finding or a combination of findings. A finding is a symptom (what the patient reported) a physical sign (what the doctor noted) or a test result (e.g. what a laboratory or x-ray department discovered).
A diagnosis can be confirmed or excluded by using agreed rules so that doctors can work with each other more easily. A diagnosis is usually regarded as certain because one or more definitive features have been shown to be present. For example the presence of two 'high' blood sugars in the morning after fasting overnight by convention confirms diabetes mellitus. There may be many other such key features that confirm the diagnosis (e.g. an extremely high blood sugar at any time, even after a meal or an HbA1c measurement). Each of these features is called a ‘sufficient’ diagnostic criterion.
Ruling out or excluding a diagnosis
The process of ruling out a diagnosis can be simplified by using a feature that is present in all patients with a diagnosis so that its absence rules it out (e.g. absence of a raised blood sugar at any time rules out diabetes). This is called a ‘necessary’ criterion (e.g. at least a single random high blood sugar at any time is necessary to diagnose diabetes). However, the presence of such a simple ‘necessary’ criterion does not confirm a diagnosis (e.g. a single high blood sugar alone does not confirm diabetes mellitus).
Diagnostic criteria are a matter of convention and they change from time to time. For example, the cut-off point for a ‘high’ blood sugar early in the morning was lowered in 1998 from 7.8mmol/l to 7mmol/l because some patients with blood sugars between 7.0 and 7.8mmol already had eye complications of diabetes and may have benefited if they had been treated for diabetes earlier. In January 2011 the diagnostic criteria were changed again by adding another 'sufficient' criterion - an HbA1c test result of at least 6.5%.
This illustrates that there should always be an element of caution with diagnostic criteria. Also, it is possible that a test was not done correctly or that it was done on someone else’s blood by mistake. It is also possible that some other new diagnosis is discovered in future that can produce the same features but needs to be treated differently. This is why the philosopher Karl Popper pointed out that a scientific hypothesis could never be confirmed in its entirety but only 'falsified' by an incompatible finding. If a patient’s illness fails to progress as expected from the diagnosis or treatment, then one has to keep an open mind.
If a diagnosis is ruled out, then a patient will not be offered treatments for that diagnosis because it is assumed that no treatment directed at that diagnosis will help. However, if the diagnosis is confirmed or showed to be probable, the patient is not automatically given a treatment. The doctor should first try to determine whether the treatment is indicated, i.e. whether there is a reasonable chance that the patient will benefit from the treatment.
Randomised controlled trials
If the test’s result is a numerical value, then its possible values can be divided into ranges and the result of a randomised control trial analysed on the patients in each range to see if the treatment is the best option within each range. The treatment is usually compared to a 'control' or sham treatment as some patients get better more often if they think that they are being treated – called the placebo effect.
Sometimes, a treatment is no better than placebo when the test result is only a little different from the results in the healthy population (i.e. when they are 'borderline'). The body is designed to repair itself and will do so more frequently when a disease is mild. The treatment can also be ineffective if the test result is extreme, because the process may be too advanced and it is too late. In some cases, the treatment may only perform better than a placebo in a few patients with a narrow range of test results.
'Normal' and 'abnormal' ranges
It is commonly assumed that if the patient has a test result that is in the top 2.5% or bottom 2.5% of the population then result is 'abnormal' and some treatment should be given. This is a fallacy. The ‘top or bottom 2.5% assumption’ should be tested to see how effective the treatment is on populations of patients with different ranges of test results as explained above. However, even if more patients with particular test result values benefit from treatment than placebo, the probability of benefit, and the patient’s perception of the value of that benefit, will have to be taken into account. This will also have to be balanced against the risk of adverse effects and the patient’s perception of how unpleasant these adverse effects will be.
© Huw Llewelyn 2016