How to understand medical decisions

Theories and facts

It is important to clarify what is meant by a ‘fact’.  A ‘fact’ is an observation that others accept as if it had been their own.  To be accepted, a fact should at least include a time and date and enough details to allow it be checked by any one who wishes to do so.  The same priciple applies in law.  If the original event has be accessed by the observer (e.g. by looking at the original x-ray image), the fact will have been ‘corroborated’.  If a similar event is re-created (e.g. by repeating an x-ray and finding the same picture) the event will have been ‘replicated’.  


Replication is very important in science and medicine.  If the reader of a research paper thinks that the probability of replicating its findings is high, its content may be accepted as if the reader had made the observations.  If the probability of replication is not high and the ideas in the paper are important then the reader may repeat the study to see if the result can be replicated. If the result is replicated, then the probability of further replication becomes even greater.  


The concept of replication can be applied to an individual patient’s story and examination findings as reported by someone else.  If a listening doctor thinks that he or she would very probably find the same thing by repeating the assessment, then he or she would not repeat the assessment.  However, if there was some doubt (e.g. if the first assessor was an inexperienced student or doctor) then the assessment would have to be repeated to see if the findings could actually be replicated.  


A fact becomes ‘evidence’ when it is used to predict another fact or a number of facts.  For example, it may be a ‘fact’ that a patient has had severe central chest pain with specific changes on an electrocardiogram.  This combination of findings becomes ‘evidence’ if it is used to predict that the patient has probably had a coronary artery thrombosis.  The chest pain can be described as ‘particular’ fact or evidence obtained from a ‘particular’ patient.  This is the same terminology that is used to describe ‘particular’ propositions in logic.  


There are also ‘general’ propositions of fact.  An example of a general proposition is: “Most patients with severe central chest pain and characteristic changes on an electrocardiogram will also have a coronary artery thrombosis”.  This general proposition is an assertion but is also ‘a fact’ if it is a description of observations made on a group of patients at specified times and places.  It becomes ‘evidence’ if it is used to make a prediction.  For example, if a new patient appeared with severe central chest pain and with the same ECG findings, then we could use this fact and the facts about previous patients as ‘evidence’ to predict that the new patient will probably have a coronary artery thrombosis.  


When making a prediction based on evidence, we make use of ‘particular’ evidence from the new patient and ‘general’ evidence from the previous patients.  We conclude that the new patient with central chest pain and ECG changes will probably have coronary thrombosis from the ‘general fact’ that most previous patients with central chest pain and the ECG changes turned out eventually to have had a coronary artery thrombosis.  During this reasoning process based on facts we thus combine both types of evidence – the general and the particular.  


The ‘prediction’ in this case is that there will be a thrombosis (i.e. a blood clot) visible inside the coronary artery.  This might be ‘confirmed’ by injecting a radio-opaque dye into the artery and seeing the picture on a video screen.  However, it is often not possible to confirm a prediction and the prediction will then be forever ‘probable’.  It is also important to bear in mind that a diagnosis is the title to an array of images that represent predictions about the evolution of the disease process, its various underlying molecular and other mechanisms, its various causes and complications and the response of all these processes to treatment.  Because of this, the ‘evidence’ displayed by the patient is used to make a large variety of predictions that come under the umbrella term of the diagnosis.  


A prediction that is about to be tested is called a ‘hypothesis’, for example, a provisional diagnosis.  When a diagnosis is provisional, other possibilities have to be borne in mind.  In medical practice, these other hypotheses are called the ‘differential diagnoses’.  If there is no intention (or it is not possible) to check whether a prediction is correct, it is called a ‘theory’.  A diagnosis is therefore a theory that is applied to a particular patient (if no further information is to be sought to support it) or hypothesis if further information is to be sought.  


General theories are the same as diagnoses but in medicine, instead of being applied to individual patients, they usually apply to populations of patients.  (In other branches of science the predictions can also apply to particular events e.g. that there was a ‘big bang’ that took place a long time ago.)  General theories about populations predict currently unseen things that may be happening now, things that may have happened in the past and things that may happen in the future, which are based on past facts from research that are used as evidence.


Some hypotheses can be confirmed and refuted.  These are simple hypotheses such as “the chest pain will probably go in 10 to 20 minutes after injecting the morphine”.  If the pain does go within this time gap, then the hypothesis is confirmed; if not the hypothesis is refuted.  However, if the hypothesis is complex e.g. that “the patient has probably had a coronary artery thrombosis” then this is the title to a number of events, some of which can be confirmed (e.g. an image showing blockage of dye on an x-ray screen) and some of which can never be confirmed (e.g. the electrons involved in the chemical reaction that results in the blood clot).  If the theory is a mathematical formula that predicts a precise value (e.g. the result of a chemical reaction) then not all the results that the formula is capable of predicting will be confirmed.  In this sense complex theories and hypotheses cannot be confirmed n their entirety.  However, they can be refuted if one of the predictions is shown to be incorrect.  It was the philosopher Karl Popper who pointed this out.


Before a ‘fact’ is going to change a diagnosis or a theory, the ‘fact’ (whether it is about an individual patient or a group of patients) has to be reliable and likely to be confirmed by others.  This is done by either corroborating the evidence (i.e. seeing the original evidence) or replicating it (making a new observation).


© Huw Llewelyn 2016