Intended for healthcare professionals


Antibiotics for acute cough in primary care

BMJ 2009; 338 doi: (Published 05 May 2009) Cite this as: BMJ 2009;338:b834
  1. Alastair D Hay, consultant senior lecturer in primary health care,
  2. Katy V Jüttner, academic foundation year 2 doctor
  1. 1Academic Unit of Primary Health Care, NIHR National School for Primary Care Research, Department of Community Based Medicine, University of Bristol, Bristol BS2 8AA
  1. alastair.hay{at}

    Prescribing can be reduced by improving communication and measuring C reactive protein

    Acute cough is the most common symptom managed by healthcare services, and it is associated with large personal, societal, and healthcare costs. In the linked randomised cluster controlled trial (doi:10.1136/bmj.b1374), Cals and colleagues assess the effects of communication skills training and the use of C reactive protein (CRP) testing on the prescription of antibiotics to patients with lower respiratory tract infection in primary care.1

    The participants were older than 18 years, they had an acute (no more than four weeks’ duration) cough considered by the general practitioner to be caused by lower respiratory tract infection, and they had at least one chest symptom or sign and at least one systemic symptom or sign.2 The investigators tried to recruit patients with what most clinicians would recognise as acute bronchitis, influenza, acute exacerbations of asthma, and acute exacerbations of chronic pulmonary obstructive disease. They excluded patients who currently or recently needed antibiotics or hospital admission.

    The trial used a factorial design, and the units of randomisation were individual general practitioners’ practices. Although cluster randomisation aims to prevent the control group from also receiving a group level intervention, such as communication skills training, it is not the preferred design for individual level interventions, such as tests (like CRP) or drugs, because it reduces trial power and introduces the risk of post randomisation selection bias. This occurs when study clinicians’ invitations for patients to participate are influenced by the clinicians’ knowledge of group allocation. Patients in the intervention and control groups were similar, however, which suggests that selection bias did not occur in this study.

    Communication skills training consisted of a two hour training seminar before and after consultations with simulated patients in routine surgeries.2 General practitioners were encouraged to: elicit patients’ worries and expectations (including the expectation for antibiotics); have a balanced discussion about the benefits and harms of antibiotics; and explain that acute bronchitis is a self limiting condition, that the body’s immune system will clear it with time, that its natural course is often longer than four weeks, that adequate fluid and food intake is beneficial, and that activity should be limited. Finally, general practitioners were encouraged to summarise and elicit the patient’s understanding of information and discuss alarm symptoms that should prompt the patient to consult them again. Three quarters of patients in the “communication skills practices” remembered the general practitioner using these skills.

    Point of care testing for CRP in primary care varies widely between countries. It is used in around 40% of consultations for respiratory tract infection in Sweden,3 but it is barely used in the United Kingdom. A CRP cut-off point of ≥40 mg/l has been shown to be 70% sensitive and 90% specific for community acquired pneumonia in primary care,4 but a recent systematic review concluded that CRP could help rule out community acquired pneumonia only if the probability of the patient having this condition is >10% (such as in emergency departments).5 In Cals and colleagues’ trial, the general practitioners were given information about the likelihood of acute bronchitis or community acquired pneumonia with different concentrations of CRP. All patients in the “CRP practices” had their CRP concentrations measured.

    About 50% of patients in the control practices were prescribed antibiotics compared with about 25% in the communication skills training practices and the CRP practices. These reductions were maintained when measured after 28 days. Severity and duration of illness and patient satisfaction were similar in all groups. Although the study probably lacks enough power to be certain, the effects seemed to be no greater in patients who received both interventions than in those who received one. The internal validity of the study is high, and the 20 practices (40 general practitioners) were broadly representative of practices and general practitioners in the Netherlands. Also, the reductions in prescribing were greater than those reported in a previous randomised controlled trial where patients were given written information only.6

    Questions that remain include whether these reductions can be reproduced, whether the reductions are safe, whether CRP testing increases the long term likelihood of patients consulting for similar illnesses in the future (the so called “medicalisation of self limiting illness”), how the interventions affected patients’ antibiotic consumption, and whether the interventions are cost effective.

    The most important and difficult question to answer is the one about safety. Weak observational data suggest that the welcome trend towards lower rates of antibiotic prescription are associated with an increase in some septic complications, particularly in the very young and very old.7 8 Research to predict which patients are at risk of complications of respiratory tract infection is under way (, and in the absence of other evidence, the National Institute for Health and Clinical Excellence has advocated the use of no (or delayed) prescription of antibiotics for respiratory tract infections in all but the highest risk patients.9

    In the meantime, what should clinicians do? We believe there are three aspects to consider. Firstly, clinician’s preferences for the intervention are likely to determine uptake. Secondly, differences exist in the starting points of different healthcare systems. Clinicians who already have point of care CRP testing may think that this study exonerates its use in patients with lower respiratory tract infection. For those who do not have access to this test, an economic evaluation would be helpful. Thirdly, CRP testing and enhanced communication skills have different strengths and weaknesses. CRP testing requires blood sampling, which is particularly problematic in children, and although it reduced prescription rates without affecting the severity and duration of illness in patients with acute cough, it may not be effective in other infections. In contrast, good communication skills are a cornerstone of high quality patient centred care for all conditions. They are required by regulatory bodies,10 are promoted to improve adherence to drugs,11 improve patient outcomes,12 and are incentivised financially (in the United States with reduced insurance premiums and in the UK through the quality and outcomes framework). Although most of us do not (yet) have access to the training used in the trial, we can all try to improve our communication skills by reflecting on the issues that it covered. After all, what we say and how we say it seems to matter.


    Cite this as: BMJ 2009;338:b834



    View Abstract