 |
 |
Dear Amy, Ash and Donald There are different issues involved here. I recognise two types of thought process – the first is intuitive and quick but not transparent. This is what basic medical training and clinical experience equips doctors to do. Kahneman calls this ‘fast thinking’ in his recent book ‘Thinking fast and slow’. There is also a second type thought process which I describe as ‘transparent’ (Kahneman calls it ‘slow’). The second type is transparent and logical. It is used when justifying diagnoses and decisions to other doctors in the team, in written opinions, when justifications are given in grand rounds and some professional exams. This process is described in the Oxford Handbook of Clinical Diagnosis. Few doctors would follow a plan that they could not justify logically in this way. However, many doctors (e.g. general practitioners) are not often expected to give their reasoning in writing. Unfortunately evidence is only collected currently for aspects of this transparent process, which is why I think that decisions are so variable and confusing. The attached Appendix of the Oxford Handbook of Clinical Diagnosis explains what is missing and what needs to be done. The first chapter also describes how to explain to patients which of their findings were used as ‘particular evidence’ for each diagnosis and consequently the treatments offered for each diagnosis. I think that if this was always done, the worst decisions would be avoided, because the decision making doctor would know that the patient could easily get a second opinion. EBM does not yet address the best findings for use in differential diagnosis by probable and logical elimination, the logic of necessary criteria to rule out diagnoses and all their possible treatments, the sufficient criteria to rule in diagnoses so that their treatments should be considered (treatments are also often considered if these criteria are only probably or possibly present). Having arrived at a probable or confirmed diagnosis, the next step is to consider the probability of benefit (e.g. the proportion who would improve on each possible treatment option minus the proportion who improve on placebo or no treatment at all.) This has to be done for each patient with that patient’s particular test results and particular degree of severity of condition. The next step is to consider the probability of adverse effects and then to discuss with the patient whether the risks are worth the potential benefit. In the absence of supportive evidence at each step (if possible in their own clinical settings), doctors have to guess, which is what causes problems (especially if there is unwitting bias). I explain how to go about collecting some of this evidence in the Appendix of the Oxford Handbook of Clinical Diagnosis (which I can only forward indpendently for those who have asked). With best wishes Huw