Nearly 15 years ago when I first presented the results of our systematic review on antibiotics for acute otitis media, one paediatrician snarled, “You’re making it too complicated. It’s simple: otitis media is an infection; the treatment of infection is antibiotics.” So that was that. The art of therapeutics could be boiled down to a simple reflex: right diagnostic label -> right treatment. As a young GP researcher I felt slightly bruised by the comment, but I’m now thankful for the stimulus to consider why we need more nuanced approaches to treatment. Diagnosis is an important, though not always essential, first step. Individual treatment involves knowing more than the label though. Tolstoy noted that, “No two unhappy families are unhappy in the same way,” which is true not only for mental illness, but nearly all illness. Typical cases are the rarity.
To help patients rather than fighting diseases means considering consequences not labels. Those consequences can be current or future symptoms, which will vary considerably based on the extent of the condition and the person with the condition. The “label=treatment” reflex commonly leads to over treatment: “otitis media = antibiotics;” “depression = SSRI;” “diabetes = oral hypoglycaemic;” “asthma = inhaled corticosteroid;” etc. For each of those conditions, some patients will benefit from the “reflex” therapy, but that will not account for the enormous range of severity and future risk within any disease category. For someone labelled as “hypertensive” the impact of a antihypertensive drug might be a relative risk reduction of around 25% but the number needed to treat (NNT) varies from treating 13 people for five years to prevent one CVD event, to treating 80 for five years, depending on age, blood pressure level, and other risk factors. That NNT of 80 means 79 of the folk treated for 5 years did not benefit from it. Further, the recent Cochrane review of 12 randomized trials (8,900 patients) of antihypertensive drug therapy for adults with mild hypertension (systolic blood pressure (BP) 140-159 mmHg and/or diastolic BP 90-99 mmHg) and without cardiovascular disease found no overall effect: an NNT of infinity.
So where should we draw the line? Many patients might be interested in where we do, suggesting we might involve them in such value-laden decisions. As Glyn Elwyn recently tweeted: evidence based medicine + shared decision = smart health decisions.
The “label=treatment” reflex has a further danger: when the definition of a disease is expanded—by changing the blood pressure, HbA1c or other threshold – the newly defined are at high risk of overtreatment. A new series in the BMJ is examining this problem for a range of conditions. For example, CT angiography has almost doubled the number of patients who appear to have pulmonary emboli but this has not changed the mortality from pulmonary embolism. So should we revolt against the tyranny of disease labels? Probably not—they do serve as a useful initial guide to thinking about management. But we need to be more sceptical about the arbitrariness of many disease definitions, and more cautious about our therapeutic reflexes. We must diligently remember to treat the patient not the label.
Paul Glasziou is professor of evidence based medicine at Bond University and a part time general practitioner.