This site is intended for health professionals only

At the heart of general practice since 1960

pul jul aug2020 cover 80x101px
Read the latest issue online

Independents' Day

Discussing pros and cons of treatment with patients

Professor Shaun Treweek outlines some advice on discussing preventative treatments with patients 

Remember to ask your patient how much he or she wants to know about the benefits and harms of treatment options. Most people want to know as much as possible but not everyone does.

Some people prefer words, some prefer numbers; giving both leads to greater understanding. [1]

Patients like narrative descriptions (eg, for antibiotics for acute otitis media in children: ‘Antibiotics increase the number of children who have diarrhoea, vomiting, or rash compared to no antibiotic treatment’) but people interpret words like ‘increase’, ‘reduce’ ‘may improve’ etc differently, and can get a false impression of the size of benefit or harm unless the narrative is supplemented with numerical information.

Although there is no sure-fire way of presenting numerical information to the public (including to health professionals), it is generally best to use natural frequencies, in other words, with the otitis media example, something like: ‘With antibiotics 27 children out of 100 would have diarrhoea, vomiting, or rash and 73 would not. For children taking a placebo, 20 out of 100 would have diarrhoea, vomiting, or rash and 80 would not.’ This approach is preferable to discussing percentages, relative risk, odds ratios, or NNTs. [2–4]

Simply saying that taking antibiotics leads to an increase in cases of diarrhoea, vomiting, or rash may in the minds of some patients lead to an overestimate of the size of potential harm, whereas this is less likely once the numbers are given and the difference of seven cases per 100 is made explicit.

Further examples of how risk information can be presented are available on this tool, the interactive Summary of Findings (iSoF) table:

It is useful to talk about uncertainty, which is also covered in the iSoF example. For that difference of seven cases, you could add ‘Due to the play of chance, there is uncertainty about this difference meaning that the difference could be as low as 3 extra cases or as high as 11 extra cases’. This is the 95% confidence interval, which can be shared with your patient without needing to use the ‘95% confidence interval’ label.

Graphics can be helpful. The iSoF example above shows some and NICE also have decision aids too (eg, NHS Choices also has a set of resources at

Remember the patient’s own baseline risk may be different to the baseline risk of the people included in trials, systematic reviews or guidelines. You might need to explain to the patient that the benefit (or harm) he or she experiences may be higher or lower than that of other patients.

Professor Shaun Treweek is professor of health services research, University of Aberdeen


1. Glenton C et al. Presenting the results of Cochrane Systematic Reviews to a consumer audience: a qualitative study. Med Decis Making 2010; 30: 566–77.

2. Knapp P et al. Perceived risk of medicine side effects in users of a patient information website: a study of the use of verbal de-scriptors, percentages and natural frequencies. Br J Health Psychol 2009; 14: 579-594

3. Santesso N et al. A summary to communicate evidence from systematic reviews to the public improved understanding and accessibility of information: a randomized controlled trial. J Clin Epidemiol 2015: 182–190

4. Büchter RB et al. Words or numbers? Communicating risk of adverse effects in written consumer health information: a systematic review and meta-analysis. BMC Informatics and Medical Decision Making 2014; 14: 76

Rate this article  (3.33 average user rating)

Click to rate

  • 1 star out of 5
  • 2 stars out of 5
  • 3 stars out of 5
  • 4 stars out of 5
  • 5 stars out of 5

0 out of 5 stars

Readers' comments (1)

  • The above example of a clear explanation seems rather mealy mouthed and I think would be more difficult for a patient to reliably grasp - in playing down patient understanding of NNT you've downplayed an important tool (see Pulses articles today on NNT) that I often find is actually well understood and much more easily conveyed to a patient. An 'overestimate' of harm usually isn't what we're usually worried about anyway, we're usually using these figures as a means to talk patients out of treatment, or advise why we aren't prescribing anything for self limiting illness to prevent a public underestimate of the harm of antibiotics, hypnotics etc.

    None of the cited trials appear to have occurred in real GP settings, time limitations and all. Realistically, I would imagine that like myself, most doctors (especially generalists dealing with broad knowledge) would struggle to retain accurate incidence/prevalence/confidence interval figures for every condition that they deal with. I usually just reference NNT/NNH, using the NNT website or diagrams, and convey it to what appears to be good effect to the patients, who are usually well versed in odds and probabilities.

    Unsuitable or offensive? Report this comment

Have your say