Intended for healthcare professionals


Antibiotics for acute respiratory tract infections in primary care

BMJ 2016; 354 doi: (Published 05 July 2016) Cite this as: BMJ 2016;354:i3482

Chinese translation


  1. Chris Del Mar, professor of public health
  1. Centre for Research in Evidence Based Practice, Bond University, Gold Coast, Australia
  1. cdelmar{at}

Fresh reassurance that reducing prescribing is safe

Primary care is increasingly becoming the focus of antimicrobial stewardship in the fight against the global crisis of antibiotic resistance. This is where the greatest tonnage of antibiotics is prescribed—perhaps three quarters of all used in medical care. Respiratory tract infections are the group of infections most commonly treated with antibiotics, principally made up of otitis media, sore throat, cough (“acute bronchitis”), rhinosinusitis, and undifferentiated infections.

Among this group, however, the antibiotic trade-off between benefits and costs—counting resistance among the latter—is least favourable. Respiratory tract infections are spontaneously remitting illnesses often caused by viruses, although it is difficult or even impossible to differentiate viral from bacterial cases in primary care. Strong evidence from Cochrane reviews shows that antibiotics on average reduce the duration of illness by less than a day.1 2 3 4 Accordingly we should conserve antibiotics in primary care so that resistance—which is directly amplified by antibiotic use—is minimised, and even reversed.5 An apparently simple solution would be to explain clearly to all interested parties that the benefits of antibiotics for patients with respiratory tract infections are lack lustre at best.

Solutions are rarely that simple however. There are many reasons why primary care doctors prescribe antibiotics for people with respiratory tract infections. Apart from unrealistic misconceptions among both patients and practitioners about the effectiveness of antibiotics, another important reason centres on safety. Primary care doctors worry about missing a serious infection masquerading as a common self limiting one, perhaps early in its evolution. Meningitis and other life threatening illnesses are notorious for sometimes presenting as undifferentiated respiratory tract infections.6 This worry may well manifest itself in the form of an antibiotic prescription “just in case.” Antibiotic prescribing rates might depend on primary care doctors’ tolerance of risk.

We have known for nearly half a century that primary care doctors prescribe antibiotics at widely different rates.7 This invites the question of whether the patients of low prescribing doctors fare less well than those of high prescribers. We can answer that (“not much”) with some precision in relation to illness symptoms and common complications using systematic reviews and meta-analyses of randomised controlled trials.1 2 3 4

But what about rarer, more dangerous complications of respiratory tract infections, that are too infrequent to be captured by trials? This was investigated in a linked paper by Gulliford and colleagues (doi:10.1136/bmj.i3410) using an enormous database of electronic health records, capturing 7% of all primary care consultations in the United Kingdom, roughly 45 million patient years of information.8 They divided the practices into fourths by antibiotic prescribing rates and compared the incidence of rare serious infections between the highest and lowest fourths. The authors included new cases of pneumonia, quinsy (peritonsillar abscess), mastoiditis, empyema, meningitis, intracranial abscess, and Lemierre’s syndrome (a rare anaerobic septicaemia arising from the throat).9 They tested the hypothesis that a low antibiotic prescribing rate might be associated with a higher incidence of serious infections.

The analysis found an association between low antibiotic prescribing and incidence of pneumonia and quinsy, but not the other serious infections. The size of the effect is important. The authors model this as one more case of pneumonia in each practice each year and one more case of quinsy in each practice every decade for every 10% drop in antibiotic prescribing. An average practice in the UK has 7000 patients.

Looking at this from an individual doctor’s point of view, in the UK where there is one primary care doctor for roughly every 1500 people,10 reducing antibiotic prescribing by 10% would mean risking one extra case of pneumonia every four or five years and one extra case of quinsy every 40 to 50 years. There was no association between prescribing rates at the practice level and incidence of mastoiditis or the really frightening conditions that can initially present as apparently innocuous respiratory tract infections—namely, meningitis, empyema, intracranial abscess, and Lemierre’s syndrome. The study’s large size means it is unlikely to be underpowered, and this demonstration of the small level of risk may help to convince doctors and patients that reducing antibiotic prescribing in primary care is a safe option. Alone, these findings are unlikely to be a major factor in changing practice. There are many other influences on antibiotic prescribing,11 such as the urge for patients to want, and primary care doctors to offer, something12 when there are few other treatment options; the notion among many doctors that the ongoing crisis in antibiotic resistance is not their problem,13 the inertia built in to old prescribing habits, time pressures, the affordability of antibiotics, and so on. We urgently need multiple, sustainable, cost effective interventions to address the problem of overprescribing, including delayed prescriptions14 and shared decision making.15 The findings of Gulliford and colleagues can support these interventions by providing specific and quantified reassurance for everyone about the safety of reducing antibiotic prescriptions for acute respiratory tract infections.


  • Research, doi: 10.1136/bmj.i3410
  • Competing interests: I have read and understood the BMJ policy on declaration of interests and declare the following: none.

  • Provenance and peer review: Commissioned; not externally peer reviewed.


View Abstract