Intended for healthcare professionals


The antibiotic course has had its day

BMJ 2017; 358 doi: (Published 26 July 2017) Cite this as: BMJ 2017;358:j3418
  1. Martin J Llewelyn, professor of infectious diseases1 2,
  2. Jennifer M Fitzpatrick, specialist registrar in infection2,
  3. Elizabeth Darwin, project manager3,
  4. SarahTonkin-Crine, health psychologist4,
  5. Cliff Gorton, retired building surveyor5,
  6. John Paul, consultant in microbiology6,
  7. Tim E A Peto, professor of infectious diseases7,
  8. Lucy Yardley, professor of health psychology8,
  9. Susan Hopkins, consultant in infectious diseases and microbiology9,
  10. Ann Sarah Walker, professor of medical statistics and epidemiology3
  1. 1Department of Global Health and Infection, Brighton and Sussex Medical School, Falmer, BN1 9PS, UK
  2. 2Department of Microbiology and Infection, Brighton and Sussex University Hospitals NHS Trust, Brighton, UK
  3. 3Nuffield Department of Medicine, University of Oxford, UK
  4. 4Nuffield Department of Primary Care Health Sciences, Oxford, UK
  5. 5Oxford, UK
  6. 6Public Health England, Royal Sussex County Hospital, Brighton, UK
  7. 7Oxford Biomedical Research Centre, Oxford, UK
  8. 8Faculty of Human and Social Sciences, University of Southampton, UK
  9. 9Royal Free London NHS Foundation Trust, London, UK
  10. Correspondence to: M Llewelyn M.J.Llewelyn{at}

With little evidence that failing to complete a prescribed antibiotic course contributes to antibiotic resistance, it’s time for policy makers, educators, and doctors to drop this message, argue Martin Llewelyn and colleagues

Antibiotics are vital to modern medicine and antibiotic resistance is a global, urgent threat to human health. The relation between antibiotic exposure and antibiotic resistance is unambiguous both at the population level1 and in individual patients.2 Reducing unnecessary antibiotic use is therefore essential to mitigate antibiotic resistance.

Avoiding overuse requires healthcare professionals and the public to be well informed about antibiotic treatment, as set out in the first objective of the World Health Organization Global Action Plan.3 Public communication about antibiotics often emphasises that patients who fail to complete prescribed antibiotic courses put themselves and others at risk of antibiotic resistance. For example, in materials supporting Antibiotic Awareness Week 2016 WHO advised patients to “always complete the full prescription, even if you feel better, because stopping treatment early promotes the growth of drug-resistant bacteria.”4 Similar advice appears in national campaigns in Australia,5 Canada,6 the United States,7and Europe.8 And in the United Kingdom it is included as fact in the curriculum for secondary school children.9

However, the idea that stopping antibiotic treatment early encourages antibiotic resistance is not supported by evidence, while taking antibiotics for longer than necessary increases the risk of resistance. Without explicitly contradicting previous advice, current public information materials from the US Centers for Disease Control and Prevention (CDC) and Public Health England have replaced “complete the course” with messages advocating taking antibiotics “exactly as prescribed.”1011 We explore the evidence for antibiotic duration, clinical effectiveness, and resistance, and encourage policy makers, educators, and doctors to stop advocating “complete the course” when communicating with the public. Further, they should publicly and actively state that this was not evidence-based and is incorrect.

Origins of the idea

Concern that giving too little antibiotic treatment could select for antibiotic resistance can be traced back to the dawn of the antibiotic era. When Howard Florey’s team treated Albert Alexander’s staphylococcal sepsis with penicillin in 1941 they eked out all the penicillin they had (around 4 g, less than one day’s worth with modern dosing) over four days by repeatedly recovering the drug from his urine. When the drug ran out, the clinical improvement they had noted reversed and he subsequently succumbed to his infection.12 There was no evidence that this was because of resistance, but the experience may have planted the idea that prolonged therapy was needed to avoid treatment failure.

Fleming’s early work showed that sensitive bacteria could be “acclimatised” to penicillin in the laboratory.13 In his 1945 Nobel prize acceptance speech, Fleming painted a vivid clinical vignette in which an imagined patient with a streptococcal throat infection who takes insufficient penicillin, transmits the infection—now in resistant form—to his wife, and is thus responsible for her subsequent death from antibiotic resistant disease.14 Fleming advised, “If you use penicillin, use enough!” Ironically, Streptococcus pyogenes has never developed resistance to penicillin, and we now know that for most forms of antibiotic resistance that currently threaten patients, selection of resistance in the bacteria being treated is of limited importance.

Antibiotic treatment drives resistance

The scenario envisaged by Fleming was of target selected resistance (box 1). Infections typically begin when a small population of microorganisms gain access to the host and replicate. Genetic mutations conferring antibiotic resistance may arise spontaneously during replication and be selected for during treatment. Target selected resistance can occur with inadequate antimicrobial dosing or with monotherapy for infections for which spontaneous resistant mutations arise on treatment, such as tuberculosis, gonorrhoea, and HIV.

Box 1: Selection of antibiotic resistance

Target selection—For certain “professional” pathogens, such as Mycobacterium tuberculosis, spontaneous resistance conferring mutants may be selected during treatment, can be transmitted before cure is achieved, or can re-emerge after treatment failure. Other professional pathogens where this may apply include HIV, malaria, gonorrhoea, and Salmonella typhi

Collateral selection—Many bacterial species that live harmlessly in the gut, on our skin and mucus membranes, or in the environment can also cause disease as opportunist pathogens. For such organisms, resistance selection occurs predominantly during antibiotic treatment of other infections. Resistance in opportunists may be passed easily to other strains of the same species of bacteria or to different species. Key examples include methicillin resistance in Staphylococcus aureus, extended spectrum β-lactamase producing Enterobacteriaceae and carbapenem resistance in Klebsiella pneumoniae

Early trials of tuberculosis treatment showed resistance emerging during monotherapy15 and underpin the need for combination therapy for this disease. Transmission of such pathogens during or following inadequate treatment may allow resistant strains to spread from person to person.

However, most of the bacterial species now posing the greatest problems do not develop resistance through target selection. The clinical threat comes mainly from species such as Escherichia coli and the so called ESKAPE organisms (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter spp, Pseudomonas spp, Enterobacter spp), which are all found harmlessly in us, on us, or in our environment. They can also act as “opportunistic” pathogens.

When a patient takes antibiotics for any reason, antibiotic sensitive species and strains present among commensal flora on their skin or gut or in the environment are replaced by resistant species and strains ready to cause infection in the future.16 This collateral selection (box 1) is the predominant driver of the important forms of antibiotic resistance affecting patients today. The longer the antibiotic exposure these opportunist bacteria are subjected to, the greater the pressure to select for antibiotic resistance.217

Importantly for these opportunistic pathogens, resistant strains are transmitted between asymptomatic carriers rather than people with disease. Furthermore, many resistance conferring genes can pass easily between bacterial strains or species. Thus antibiotic selection may drive outbreaks of resistant infections independently of transmission of a specific strain or species.18

From fear of undertreatment to harm from overtreatment

Traditionally, antibiotics are prescribed for recommended durations or courses. Fundamental to the concept of an antibiotic course is the notion that shorter treatment will be inferior. There is, however, little evidence that currently recommended durations are minimums, below which patients will be at increased risk of treatment failure.

Historically, antibiotic courses were set by precedent, driven by fear of undertreatment, with less concern about overuse. For many indications, recommended durations have decreased as evidence of similar clinical outcomes with shorter courses has been generated (table 1). However, the picture is patchy and complicated by comparisons of new and established agents that may have different pharmacological properties (eg, long acting macrolides versus short acting penicillins).

Table 1

Indications for which duration of antibiotic treatment has been evaluated by randomised controlled trial

View this table:

For most indications, studies to identify the minimum effective treatment duration simply have not been performed.28 For example, pyelonephritis has historically been treated for two weeks. Trials have shown that shorter courses of quinolones are effective (seven days for ciprofloxacin23 and five days for levofloxacin24), but no such data exist for β-lactams, which are the main antibiotic class used. Current international guidelines recommend 10-14 days’ treatment with β-lactams, based purely on absence of data for shorter courses.29

Shorter duration of treatment has been shown to reduce clinical efficacy in a few cases. A notable example is otitis media, where five days’ treatment is associated with a lower clinical cure rate (66%) than 10 days (84%) in children under 2 years.19 Even in this situation though, differences relate to prolongation of symptoms not treatment failure, disease recurrence, or selection for resistant pathogens.

For the opportunist pathogens for which antimicrobial resistance poses the greatest threat, no clinical trials have shown increased risk of resistance among patients taking shorter treatments.

The key argument for changing how we discuss antibiotic courses with patients is that shorter treatment is clearly better for individual patients. Not only does an individual patient’s risk of resistant infection depend on their previous antibiotic exposure217 but reducing that exposure by shorter treatment is associated with reduced risk of resistant infection and better clinical outcome. In hospital acquired pneumonia, for example, randomised controlled trial data indicate that short treatment strategies have equivalent clinical outcomes to longer courses and are associated with lower rates of infection recurrence and antibiotic resistance.2526

Is the concept of an antibiotic course still valid?

The concept of an antibiotic course ignores the fact that patients may respond differently to the same antibiotic, depending on diverse patient and disease factors. Currently, we largely ignore this fact and instead make indication specific recommendations for antibiotic duration that are based on poor evidence. This situation is changing in hospital practice, where biomarkers of treatment response such as procalcitonin can guide when to stop antibiotic treatment.30 Outside hospital, where repeated testing may not be feasible, patients might be best advised to stop treatment when they feel better, in direct contradiction of WHO advice.4 Of note, a recent clinical trial found that using fever resolution to guide stopping antibiotics in community acquired pneumonia halved the average duration of antibiotic treatment without affecting clinical success.21 Further similar studies are needed.

“Complete the course”: a barrier to antibiotic conservation

The fallacious belief that antibiotic courses should always be completed to minimise resistance is likely to be an important barrier to reducing unnecessary antibiotic use in clinical practice and to developing evidence to guide optimal antibiotic use. The idea is deeply embedded, and both doctors and patients currently regard failure to complete a course of antibiotics as irresponsible behaviour.3132

In primary care, strategies have been developed to avoid unnecessary antibiotic courses being started—for example, through enhanced communication training, point-of-care tests, and use of delayed prescriptions.333435 However in secondary care, strategies to reduce overuse aim to change, or ideally stop, antibiotics 48-72 hours after they are started, but these are challenging to implement.36 Reasons for this include diagnostic uncertainty and team behaviour, but patients’ and healthcare professionals’ concerns about the risks of incomplete treatment are likely to contribute. Designing trials of antibiotic sparing treatment is notoriously difficult,37 particularly if participants are invited to consent to receiving shortened antibiotic treatment on the basis that this could reduce their risk of antibiotic resistance, when they have been taught from school that it increases this risk.

What should we advise patients?

The “complete the course” message has persisted despite not being supported by evidence and previous arguments that it should be replaced.1838 One reason it may be so resilient is that it is simple and unambiguous, and the behaviour it advocates is clearly defined and easy to carry out. Nevertheless, there is evidence that, in many situations, stopping antibiotics sooner is a safe and effective way to reduce antibiotic overuse. Daily review of the continued need for antibiotics is a cornerstone of antibiotic stewardship in hospitals,39 but in primary care, where 85% of antibiotic prescriptions are written, no such ongoing assessment is attempted.

There are reasons to be optimistic that the public will accept that completing the course to prevent resistance is wrong if the medical profession openly acknowledges that this is so, rather than simply substituting subtle alternatives such as “exactly as prescribed.” Completing the course goes against one of the most fundamental and widespread medication beliefs people have, which is that we should take as little medication as necessary.40 Concerted and consistent efforts have successfully educated the public that antibiotics do not treat viral infections, for example.

Research is needed to determine the most appropriate simple alternative messages, such as stop when you feel better. Until then, public education about antibiotics should highlight the fact that antibiotic resistance is primarily the result of antibiotic overuse and is not prevented by completing a course. The public should be encouraged to recognise that antibiotics are a precious and finite natural resource that should be conserved. This will allow patient centred decision making about antibiotic treatment, where patients and doctors can balance confidence that a complete and lasting cure will be achieved against a desire to minimise antibiotic exposure unimpeded by the spurious concern that shorter treatment will cause antibiotic resistance.

Key messages

  • Patients are put at unnecessary risk from antibiotic resistance when treatment is given for longer than necessary, not when it is stopped early

  • For common bacterial infections no evidence exists that stopping antibiotic treatment early increases a patient’s risk of resistant infection

  • Antibiotics are a precious and finite natural resource which should be conserved by tailoring treatment duration for individual patients

  • Clinical trials are required to determine the most effective strategies for optimising duration of antibiotic treatment


  • Contributors and sources: LY, MJL, TEAP, SH, and ASW are investigators on an on an NIHR Programme Grant for Applied Research (PGfAR) called ARK-hospital, which aims to reduce antibiotic overuse in hospitals through clinical review of antibiotic prescriptions [RP-PG-0514-20015]. ASW, STC, TEAP, and SH are investigators in the NIHR Health Protection Research Unit (NIHR HPRU) in healthcare associated infections and antimicrobial resistance at Oxford University in partnership with Public Health England (PHE) [grant HPRU-2012-10041]. CG is a retired building surveyor. ED is a project manager for ARK and the HPRU. JP is Public Health England regional microbiologist for the south east. The article is based on published good quality randomised clinical trials and observational cohort studies. All authors have contributed to this paper and concur on its content. MJL is guarantor. The views expressed in this publication are those of the authors and not necessarily those of their employers.

  • Competing interests: We have read and understood BMJ policy on declaration of interests and have no relevant interests to declare.


View Abstract