Last year, I celebrated my 30th year as a doctor and my son began his training as a (graduate entry) medical student. I have come to enjoy the intergenerational ‘grand rounds’ in which one of us describes a case in the time-honoured format—starting with a structured history, going on to the clinical examination and adding diagnostic tests that progress from the simple and non-invasive to all the wonders and dreads of modern technology—while the other tries to guess the diagnosis from as few clues as possible. Given that most medical knowledge now lies in the category ‘forgotten by the mother and not yet encountered by the son’, this book is likely to become well thumbed by both of us as we play our diagnostic game.
Much of this book reflects the fact that Huw Llewelyn is a mathematician and logician as well as a highly experienced physician. In many cases, diagnosis can and should be a process of deduction that begins with a ‘diagnostic lead’ (a single symptom or sign, such as ‘right iliac fossa pain’, that gets you started), the cause of which can be progressively narrowed and refined by incorporating factors such as age and gender; the timing and speed of onset; the pattern of associated symptoms, signs and pre-existing conditions; and the results of investigations. Frontal headache in a teenager who was well until yesterday is likely to have a different cause from frontal headache that has been present for many months in a 65-year-old with hypertension and depression. Evidence can often be collected in the history and clinical examination that is ‘suggestive’ or ‘confirmatory’ (use these terms with care—they are defined in the book) of particular diagnostic possibilities. More rarely, certain tests or combinations of tests can effectively ‘rule in’ or ‘rule out’ certain diagnostic options.
You probably knew all that already, so what will you learn from this book that goes beyond standard teaching on clinical diagnosis? For me, the added value was in the sophistication with which the principles of probability and decision science have been applied to the many and varied challenges of clinical practice. The book’s (mainly implicit) message is that if you take a logical and step-wise approach, using your experience, history-taking skills, and clinical acumen to select the best diagnostic leads and add granularity to your decision tree, you will often render costly and unpleasant diagnostic tests redundant. Less commonly, you will justify the expense and inconvenience of such tests in selected patients.
The skilled diagnostician is not the one who rattles off a long list of differential diagnoses for every symptom, applies algorithms mechanically, ticks all the boxes on a blood request form or scans the head of every patient with blurred vision. Rather, the skilled diagnostician is the one who combines thoughtful history-taking, focused clinical examination, and judicious investigation so that each successive step contributes to an emerging picture of the problem and informs the selection of the next step. As the authors say (p.[link]), ‘It is important to understand that clinical diagnosis is not a static classification system based on diagnostic criteria or their probable presence. It is a dynamic process.’
The bulk of the book is a treasure-trove of diagnostic puzzles from red throat to wasting of the small muscles of the hands, from which I predict hours of fun for students and seasoned clinicians alike. There are also sections on biochemical conundrums such as hyponatraemia, and radiological old chestnuts such as a round opacity on the chest X-ray. Reassuringly, theoretical sections such as ‘Grappling with Probabilities’ and ‘Bayes’ and other rules’ are relegated to a final chapter that can be safely omitted by those whose interests are more clinical than mathematical.
Despite its emphasis on deductive logic, this book is by no means an uncritical offering to the gods of decision science. Llewelyn and his coauthors are careful to point out (as Dave Sackett and colleagues did back in the 1970s) that many diagnoses are made intuitively—for example via the pattern recognition that allows us to look at a patient and instantly think ‘Down’s syndrome’ or ‘chicken-pox’. They also remind us that mild symptoms are often both non-specific and self-limiting (hence may need no more active management than advising the patient to return if not improving), and they warn us of the dangers of over-diagnosis and that increasingly common problem in modern diagnostics, the ‘incidentaloma’.
Like the birth of a third child, the publication of the third edition of a book is cause for much celebration: it tends to both reflect and build on significant success with earlier versions. Perhaps it is too early to encourage the authors of the Oxford Handbook of Clinical Diagnosis (3rd edition) to contemplate a companion volume to this magnum opus. But if they were open to such a suggestion, I would encourage them to team up with experts in public understanding of science and produce a version of the book aimed at patients and carers. After all, if your patients were reading the wisdom distilled in these pages, that would surely make for some interesting and productive conversations.
Trisha Greenhalgh OBE
Professor of Primary Health Care and Dean for Research Impact
Barts and the London School of Medicine and Dentistry
Queen Mary University of London