How do we communicate risk?

A. Herxheimer

UK Cochrane Centre, UK

We do it badly, for two quite different reasons: (1) the concept of probability of harm is difficult to convey and explain clearly, particularly in a brief consultation, and (2) we use the word ‘risk’ as a euphemism for ‘harm’, though strictly it denotes only ‘probability’.

We must communicate that all interventions in health care can cause harms as well as benefits, and that an intervention is only worth using when the benefits clearly outweigh the harms. Both the benefits and the harms of an intervention have four distinct dimensions1 – (a) the kind of benefit or harm; (b) their intensity and duration; (c) the probability that they will occur; (d) if they occur, their possible effects on the person’s life.

We want people if possible to understand the important benefits and harms that we believe they should consider before choosing or accepting a treatment, and to help them compare these with the benefits and harms of other treatments, and of no treatment. Clinicians cannot communicate these complex matters without thorough preparation and help; patients/clients need time to take them in and opportunities to discuss them and ask questions. This means that both written and spoken communication is needed, often on more than one occasion. A written summary enables all the professionals in a community to use the same information as the basis for their conversations with patients, who should then get consistent messages from whichever doctors, nurses, pharmacists, etc, they encounter. That can help them to understand the concepts, to minimise misconceptions and misunderstandings, and so to save time and regrets.

A complication that can rarely be considered, let alone explained, is that the benefits and harms of medicines vary greatly with dosage and duration of use, as well as with the users’ individual characteristics; unfortunately rather little is known about these important variations.

1. Herxheimer A. Benefit, risk and harm. Australian Prescriber 2001; 24:8.