Sore throats and antibioticsBMJ 2000; 320 doi: https://doi.org/10.1136/bmj.320.7228.130 (Published 15 January 2000) Cite this as: BMJ 2000;320:130
Applying evidence on small effects is hard; variations are probably inevitable
- Chris Del Mar (), professor of general practice
General practice p 150
The liberal use of antibiotics for sore throats is increasingly frowned on.1 There are three reasons why a clinician might use antibiotics for sore throat: to reduce the risk of complications, to shorten (or reduce the severity of) symptoms, and because of factors related to the consultation (perceived patient demand, ways of terminating the consultation, and so on). Nearly 30 years ago Howie showed a huge variation in different general practitioners' use of antibiotics for sore throat.2 Have general practitioners been thirsting for information on which to base their management? The notion that summarising evidence about an area of care will result in a sort of regulation of doctors' management appears to be naive. A Cochrane review summarising the advantages of antibiotics for sore throat has been available for several years.3 But there is nothing to suggest that there is less variation in practice. General practitioners do not access evidence based information well.4
Perhaps this evidence based information is inappropriate or unhelpful? What does it show? Antibiotics reduce the incidence of both suppurative and non-suppurative complications of sore throat. A new study from Holland, published in this week's issue (p 150), has confirmed that antibiotics protect against quinsy.5 This apparently well conducted study also confirms the previously established benefits of antibiotics in reducing the duration of symptoms. However, the trial shows something new: that putting patients on seven—rather than three—days of treatment is more effective at reducing the duration of symptoms. The data are puzzling because this became apparent before day 3—when the treatments were identical. (They could be reanalysed to see how the combined penicillin treatments differ from the control up to day 3.) There was a greater effect than in previous studies. Perhaps these Dutch general practitioners focused on more severe cases than did those in other countries in the past—they have a general reputation for parsimony with antibiotics.6 New data may arrive in time to help define subgroups of people with sore throat who would derive greater benefit from antibiotics.
But this is fine detail. Broadly, the evidence does more than simply establish a statistical benefit for antibiotics in complications and symptom control: it also gives an estimate of the size of the effects. The benefit is so modest that one can dispute its clinical importance. This is because the size of the effect is small (however statistically significant) or because the chance of suffering complications is so tiny that even a reasonable relative reduction conferred by antibiotics yields a similarly unimportant absolute benefit.
This brings the result of the evidence into an area of decision making that is complicated. There is no single course of action that will suit all—or even most—patients. The evidence must be applied in different ways according to different local conditions. These will include environmental factors (such as places in the world where acute rheumatic fever is so common as to be a central consideration), history (for example, previous middle ear infections), and social factors. General practitioners put as much weight on social factors as on the physical examination in deciding whether or not to use antibiotics.7 Both patients and their doctors dance delicately around the complicated negotiation of antibiotics for upper respiratory infections, each aware of the others' sensibilities.8
If that sounds inexcusably non-objective and chaotic, consider this. At some point the benefits and harms resulting from treatments (together with doctors' additional worries to mix into the decision, including emerging antibiotic resistance1 and costs to society) are so finely balanced that patients and their doctors must decide on a choice that is likely to be tipped one way by personal preference alone.
To expect a one line answer from the evidence (a guideline, for example) is to ask too much. Nor is there any suggestion that doctors want to abdicate that responsibility. Instead they must skilfully meld the information into the consultation. Information is now becoming available in a variety of different forms, including a handy compendium of evidence.9 Perhaps we should also start thinking about how patients can gain access to good quality information themselves, quite separately from their doctors.
We know that the way of presenting evidence based information can greatly influence the decisions people make.10 The challenge is for us to learn more about how to apply this science. In the meantime broad brush efforts (such as deciding on a practice policy of presenting evidence based information to patients with upper respiratory infections) can indeed yield reduced use of antibiotics.11