Article Text

Download PDFPDF

Communicating risks in health care
Flexible rather than standardised approaches to communicating risks in health care
  1. A Edwards
  1. Correspondence to:
 Dr A Edwards
 Department of Primary Care, Swansea Clinical School, University of Wales Swansea, Singleton Park, Swansea SA2 8PP, UK;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

A sufficiently flexible approach to risk communication is needed to accommodate a wide range of patient interpretation and preferences for information

Informed medical decision making depends at least partly on understanding the benefits and harms of different treatment options. This requires clear and relevant risk communication. In practice this means that information should include both harms and benefits where relevant, and be presented honestly—not hiding information thought to be less desirable. Attention should be paid to the potential pitfalls of “framing”—how different formats such as relative and absolute risk can manipulate decisions made.1 Certain types of risk information such as “natural frequencies” (for example, 1 in 10) are generally more consistently and accurately interpreted than percentages (for example, 10%). In percentages, the reference class (population or group to which the figure applies) is often not clearly specified.2 For any data, though, there are uncertainties and these should be shared where evident.3 There is support for having a range of information formats available (a “toolbox”) so that professionals can use the most appropriate one to aid discussions with individual patients.4 Such formats may be descriptive, numerical, or graphical. They may include patient narratives of their experiences to convey the pros and cons of decisions made in certain situations. Patient preferences for different formats and levels of detail will vary, and the materials should be there to accommodate this range.

However, the appropriate language for effective risk communication with patients is less clear. Both patients and professionals have difficulty in interpreting, incorporating, and remembering risk information, especially the statistical aspects.5 There are wide variations in the interpretation of probabilistic terms with regard to the frequencies they convey.6 Few studies of preferences for or understanding of risk information in practice have been done with real patients. Most current literature describes conceptual development or hypothetical scenarios, but people interpret information differently in hypothetical and real settings.1 There is a clear need for more research to identify the effects of different formats of risk information and effective ways of communicating such information to patients having to make a relevant health care choice.

Pending the availability of such findings, some authors have proposed the use of standardised “languages” for risk communication.7,8 These aim to encourage professionals to use agreed terms such as “common”, “sometimes”, or “rare” for risks of certain frequencies. The hope is that, through continued usage, the terms would become increasingly accepted and consistently interpreted. However, there are several potential pitfalls to this approach. People vary in their preferred style of risk information—verbal or descriptive versus more numerical approaches. Standardised terms may also not be sufficiently flexible for the range and complexity of risks encountered in clinical practice. A range of primary care professionals indicated that standardised approaches would be unlikely to help discussions with patients, although there was potential to improve communication between professionals.9 The data in the paper by Knapp et al10 published in this issue of QSHC indicate a further difficulty. These authors showed that the severity of the condition adds a further dimension to the variation in interpretation of probabilistic information. The frequency attributed to a “rare” serious side effect was different from a “rare” less serious side effect. Language is also dynamic, an ever-changing phenomenon, and thus perhaps inherently resistant to control.11 All these difficulties undermine efforts to standardise the language of risk, a strategy which could be viewed as inherently “top down” or paternalistic in a field which is trying to be patient centred.

Given this variability of interpretation, changing characteristics of language, and flexibility required for risk communication approaches with individuals, we need broader strategies for enhancing risk communication, whether for individuals or large groups of people. Some lessons can be illustrated by the example of the scare over third generation oral contraceptives in 1995. Poorly communicated risk information led to patient anxiety, loss of confidence in the treatments and the authorities providing the information and, crucially, the discontinuation of medication and unplanned pregnancies. This example informs us about the need for effective, balanced communication of risk information. This would include absolute and possibly relative risk formats, but not relative risk in isolation, and attention to the wider potential pitfalls of framing manipulations such as positive or negative framing (chances of survival or death).1 All probabilistic information requires adequate explanation appropriate to the individual. For information that must be disseminated in a uniform way, such as in patient package inserts, this requires a sufficient range of formats—descriptive, numerical, visual—to be likely to meet the preferences of a large proportion of potential readers.12 This is challenging, but the contraceptive scare story illustrates the huge price to pay when getting it wrong. When discussing information with individuals, professionals also need to explore the meaning and significance of the risk, and the chances to the individual of beneficial or adverse outcomes from different treatment options. They need to explore people’s understanding and reactions to the information, and how it integrates with other influences on health, health care, and decision making such as past experiences, family pressures, and so on.

Risk information is just one of the contributors towards decision making. Efforts to improve it should not bypass or undermine other major contributors such as support from others (family friends, professionals) and the trust that is placed in these people.3 Good risk communication is, however, important if people are to be adequately informed and enabled to make decisions about their treatment or care. For this they need sufficient information, tailored and clearly presented in a format or style and level of detail with which they are comfortable to aid their decision making. It should be relevant to their situation, contextualised for prior experiences, and seek to achieve sufficient understanding of the condition(s) and risks concerned. There are a number of issues that cause deep seated difficulties for attempts to standardise the language of risk communication. Our approaches to risk communication need to be sufficiently flexible to accommodate the range of patient interpretation and preferences for information.

A sufficiently flexible approach to risk communication is needed to accommodate a wide range of patient interpretation and preferences for information


Linked Articles