|Home | About | Journals | Submit | Contact Us | Français|
Radcliffe: Oxford. 2006 PB, 152 pages, £21.95, 1 84619 055 X.
The book is well written: even the more technical sections explaining what obstacles researchers meet and how they tackle them are expressed in a straightforward language that a nonresearcher like me is able, on the whole, to understand and sympathise with. Occasionally, informality slips into lack of precision but, in general, the style is clear, flowing and engaging.
My reservations are an indication of the extent to which the book made me think. Take the title. I guess Dr Summerton had problems with it because his definition is more than 70 words and, even then, confines itself to ‘the context of a known abnormality’, whereas a lot of primary care concerns symptoms which may not signify any abnormality: the topic of medically unexplained symptoms is given only two pages in the whole book. Elsewhere Dr Summerton writes that: ‘Patient-centred diagnosis should be based on the best available scientific evidence … clinical experience is also a critical component …’. Compare this with the Cochrane website: ‘The practice of evidence based medicine means integrating individual clinical expertise with the best available external clinical evidence from systematic research’.1 Dr Summerton's diagnoses might be patient-specific but they do not have the patient-centred ‘shared understanding’ important to McWhinney.2
Let us look at some examples from the book. For cranial arteritis likelihood ratios for clinical features and ESR relate to the outcome not of blindness, which would be patient centred, but of abnormal histology. Presumably the patient-centred outcome has not been researched, but this is not acknowledged or discussed. For migraine Dr Summerton refers to McWhinney's suggestion that the name matters less than whether or not the condition is sumatriptan-sensitive — diagnosis as ‘management naming’ rather than ‘disease naming’3 — but again this idea is not developed. Elsewhere Dr Summerton writes: ‘obviously, it is always necessary to exclude organic disease when presented with a symptom of possible organic significance, such as unexplained weight loss, chest pain or palpitations’. I think being patient centred means that it is often unnecessary — and inappropriate — to do this. So a lot of the book, despite its title, has a rather biomechanical approach.
My other main reservation is that I am not sure where to start the proposed Bayesian journeys or whether I like the routes. Take cranial arteritis and migraine again. We are given the population prevalence of cranial arteritis and the prior probability of migraine in patients presenting to primary care clinicians with new onset headache. These are based on different denominators. Furthermore, negative likelihood ratios for a specific diagnosis, such as cranial arteritis, may help me reassure the patient but reassurance about cranial arteritis is of little value if the patient turns out to have migraine or trigeminal neuralgia. Likelihood ratios for ‘persisting and disabling’ or ‘needing referral’ might well be more useful: they would lead to the higher posterior probabilities in primary care that Dr Summerton hopes for. I would have liked some discussion of this, especially in the sections advising how to judge the validity of research results and where research needs to be directed in the future.
In the meantime I have learned some useful things. I had already come across the distinction between (volunteered) symptoms and (elicited) semeions.4 Dr Summerton introduces me to another useful distinction: iatrotropic symptoms are those ‘that cause a patient to consult’ and non-iatrotropic symptoms ‘those that are elicited during the course of the medical interview’. Surely they will have different likelihood ratios and I now have words — however awkward — to use when discussing this notion with checklist-laden medical students and junior doctors.