Controlling antibiotic prescribing for lower respiratory tract infections

We are prescribing antibiotics too often, especially for acute respiratory infections for which we know they make little difference. General practitioners everywhere have heard this message many times, but they struggle to take the necessary remedial action. One influential report from the UK sketches a worrying picture of a “post-antibiotic” world of widespread antibiotic resistance in which more people will die of incurable infections than of cancer. The report predicts that by 2050 incurable infections will be the second biggest killer,1 while currently routine high technology medical interventions (cardiac catheterisation, chemotherapy and radiotherapy for cancer, and major surgery) will become too dangerous without reliable antibiotic cover. The human and economic consequences would be catastrophic.1

Antibiotic prescribing in primary care is in the spotlight for two reasons: it is where the greatest raw tonnage of antibiotics is prescribed to humans (about three quarters of all prescriptions are issued in primary care); and for about half the indications (the acute respiratory infections) antibiotics are not very effective. Additionally, not using antibiotics results in a return to effectiveness again: resistance fades with time in the absence of antibiotics.2 Clearly this is where society can make some important antibiotic savings.

Yet prescribing by general practitioners around the world is changing slowly, if at all. Why? There are two main clusters of reasons. Doctors are unwilling to risk the adversarial relationship with their patients that “rationing” antibiotics for the greater good might cause. This might be related to unrealistic beliefs in their effectiveness held by both patients34 and clinicians,5 along with clinicians’ misperceptions about patients actually wanting antibiotics.6 The second cluster of reasons relates to clinical uncertainty and safety. Doctors know that distinguishing between a common self limiting acute respiratory infection and the early manifestations of dangerous diseases such as meningitis, mastoiditis, community acquired pneumonia, and peritonsillar abscess, which can declare themselves with frightening speed, may be impossible at one visit.

The study by Little and colleagues (doi:10.1136/bmj.j2148) is important for this second reason. In a large prospective, observational study of 28 779 adults presenting with lower respiratory tract infections, the authors compared adverse outcomes associated with three prescribing strategies: no antibiotics, delayed antibiotics, and immediate antibiotics.7 They looked particularly at rates of reconsultation, and death or admission to hospital, both within 30 days and adjusted for many known potential confounders. The data were 76% complete. Patients with serious illness were excluded to enable generalisation to the typical patient presenting with common cough.

There were four main findings. Firstly, antibiotics were prescribed often (>60% of illness episodes) for adults with acute cough—far more often than might be expected from good evidence of poor effectiveness. In a Cochrane review of six trials in which patients with acute bronchitis were randomised to antibiotics or placebo, antibiotics reduced the duration of cough by less than half of one day, from a mean duration of 10 days among controls (0.46 day, 95% confidence interval 0.87 to 0.04).8 Little and colleagues confirm the concerning over-prescription of antibiotics.

Secondly, and more encouragingly, doctors in this sample used delayed prescribing quite often (13% of illness episodes). This is a well studied strategy for helping patients (and clinicians) feel safe. The prescription is written but is accompanied by advice not to use it until some specific future event, such as failure to improve after a certain date or worsening symptoms. This strategy has been shown to reduce prescribing by 62% (95% confidence interval 34% to 75%).9 Thirdly, reconsultation, usually for unsatisfactory resolution of symptoms, was also common, at 22% overall. The reconsultation rate was no lower among participants given immediate antibiotics. The resolution of symptoms of acute cough often takes longer than many patients think, with some coughing for many weeks (a third for more than three).10 Better communication of the likely prognosis for acute cough is required.

The most important findings were that serious adverse events such as death or hospital admission are rare in people with acute cough and that an immediate antibiotic prescription is not associated with a significantly reduced risk.

This study adds to others showing that not offering an immediate prescription for antibiotics to people with common uncomplicated acute respiratory infections is a low risk strategy.11 Together, these studies provide good evidence that not prescribing antibiotics at the first visit for these common infections is a safe option and should enable general practitioners to improve communication with concerned patients, perhaps using delayed prescribing among other techniques to help achieve the best possible outcome for everyone.