Presenting risk information to people with diabetes: Evaluating effects and preferences for different formats by a web-based randomised controlled trial
Introduction
There has been much attention to how risk information should be communicated [1], [2], [3], [4], but relatively little to how it should be represented. Informed decision making depends on awareness and understanding of the options and the benefits and harms relating to different treatment options. Existing evidence indicates that bias may be introduced and decisions manipulated by presenting risk in different ways [5]. For example, data about treatment effectiveness can be presented in absolute risk (e.g. a 10% absolute risk reduction in disease from 20% to 10%) or relative risk (50% relative risk reduction for the same figures) formats. These formats affect understanding and decisions, such as about cholesterol-lowering drugs or for hypertension [6], [7]. A wide range of web-based information and decision support resources for patients is already available. Many of these are in the form of ‘decision aids’ and ‘interactive health communication applications’ [8], [9]. Information must be communicated effectively to support patients in understanding their conditions and promote involvement in treatment decision making. There have been few randomised trials of risk representation formats, such as graphs, numerical and narrative descriptions, for patients with real conditions [10].
We therefore sought to evaluate an online information resource for actual patients, and in the context of a common condition—diabetes. The resource included different formats of representing the harms and benefits relating to either ‘tight control’ or ‘usual treatment’ approaches to managing diabetes, defined principally by the Diabetes Control and Complications Trial [11] and the UK Prospective Diabetes Study [12], [13] for types 1 and 2 diabetes, respectively. A decision about whether tight control or usual treatment is most appropriate for an individual person with diabetes involves a trade-off between the potential benefits and harms. In essence, tight control reduces the incidence of micro-vascular complications (retinopathy, neuropathy and nephropathy), but with the adverse effects of more frequent hypoglycaemic episodes and greater weight gain. People need information about a number of elements if they are to be able to make informed decisions about which regime is most appropriate for them personally. These include information about the nature of the harms and benefits, the probabilities of their occurrence with either the tight control or usual treatment regime, and what these regimes entail in terms of commitment and impact on daily living. This ‘decision’ is likely to be dynamic, changing according to circumstances, and people's confidence about which regime is most appropriate is unlikely to be static.
The information resources in this study provided people with information about the regimes and the nature of complications, hypoglycaemic episodes and other aspects. These focused on the issues of tight control and usual treatment and did not address the full range of issues that matter to patients [14]—this is similar to the nature of other widely available online resources in the UK [15]. The resources then contained different formats of representing the chances of the harms and benefits associated with the different regimes.
After reviewing the literature we identified different representation formats that could be introduced for evaluation. These included using more detailed numerical information such as absolute risk, relative risk, and number-needed-to-treat formats. Absolute risk information can be in either percentage terms or ‘natural frequency’ (e.g. 15 in every 100). Gigerenzer suggests that natural frequencies are much less open to misinterpretation because they almost automatically identify the ‘reference class’ – the group of people to which they refer – and which may be omitted frequently when using percentages [16]. Another format was the use of ‘anchoring’ information—that is relating the harm or benefit data to more everyday or events that the user would be expected to be familiar with, such as the risk of a road accident or of winning a lottery. It has been suggested that this puts the harm or benefit information into perspective and easier to understand or use in decision making. However, this anchoring approach is usually advocated by practitioners and not from data from patients [17], [18]. A number of visual representation formats have been evaluated in the literature: line graphs, bar graphs, pie charts, crowd (or ‘smiley face’) figures [19], [20], [21], [22]. People's preferences for different formats varies, usually with bar charts found to be preferred most [23], [24], [25]. However, preferences for receiving information in a particular format have not been found to be linked to improved understanding [19], [26]. Different representations can lead to different interpretation and different decisions about treatments. It is not clear which format is ‘better’ for most people, if this is to make the most informed decisions [27], and it is likely that individual, demographic and age characteristics will affect which is optimal for individuals [28].
The premise for this study was to provide information in a range of formats, from which people could select the format most useful to their decision making. People would be directed only to look at the information they desired and were not ‘obliged’ to read all of it. Within this, however, a balance of effective information provision versus information overload would be anticipated, and not all formats could be presented: within the graphical formats we selected bar graphs, a thermometer scale as equivalent to a line graph but with interactivity to show changes, and crowd figures. We sought to evaluate the effects of information provision, including a range of representation formats and which users could select for themselves, on users’ decision making in relation to tight control usual treatment of diabetes. We also structured the trial design to examine the effects of individual components of the information (numerical, anchoring, or graphical) and sought data that would provide insight into why the intervention might or might not have been effective.
Section snippets
Setting and participants
The study was conducted online. Recruitment publicity was disseminated via Diabetes UK (website and printed newsletter versions (regional and national), and annual conference) to reach people with diabetes and/or their carers, inviting them to participate via the url: www.diabetesonlinetrial.co.uk
Participants were asked before randomisation about their last home blood glucose monitoring result or last HBA1c result to assess both familiarity with these readings and with UK measurement scales.
Design
A
Participants
The study ran from March to July 2004 and 710 people visited the website, completed demographic data and were randomised to control or one of the intervention groups (see Fig. 2). Of these 508 (71.5%) completed the questionnaire survey for quantitative outcome data.
Most people (77.5%) had been alerted to the study via the Diabetes UK website. The characteristics of the sample are shown in Table 1. Age, gender and type of diabetes in the sample were similar to those of the Diabetes UK
Discussion and conclusion
Different risk representations did not affect decision conflict (i.e. uncertainty about the best option). Qualitative analysis of comments indicated that participants strongly supported clear and effective information provision such as made available in this study. However, there was evidence that the information provided often did not align with individual patient information needs, and that it would need to address a broader range of lifestyle issues. Regarding the risk representation itself,
Acknowledgements
We gratefully acknowledge input from several other contributors to the project: Hayley Hutchings, Lecturer in Statistics, University of Wales Swansea; Ivy Cheung, Senior Lecturer in Statistics, University of Wales Swansea; Professor Ian Russell, Professor of Public Health, University of Wales Bangor; Dr Alan Sykes, formerly Senior Lecturer in Statistics, University of Wales Swansea; Jonathan Peterson, Technical Systems Manager, British Medical Journal Knowledge; Luisa Dillner, Deputy Head
References (52)
- et al.
Challenges for designing and implementing decision aids
Pat Educ Counseling
(2004) - et al.
Tailored risk notification for women with a family history of breast cancer
Prev Med
(1999) - et al.
Comparison of tailored interventions to increase mammography screening in non-adherent older women
Prev Med
(2003) - et al.
The development of COMRADE—a patient-based outcome measure to evaluate the effectiveness of risk communication and treatment decision-making in consultations
Pat Educ Counseling
(2003) - et al.
Genetic counselling: information given, recall and satisfaction
Pat Educ Counseling
(1997) - et al.
Health promotion in primary care: physician–patient communication and decision making about prescription medications
Soc Sci Med
(1995) Communicating risks: means that patients too have to learn to live with uncertainty
Br Med J
(2003)- et al.
Risk communication in clinical practice: Putting cancer in context
J Nat Cancer Inst Monogr
(1999) - et al.
Risk communication strategies: state of the art and effectiveness in the context of cancer genetic services
Eur J Hum Gen
(2003) - et al.
The effectiveness of one-to-one risk communication interventions in health care: a systematic review
Medical Decision Making
(2000)
Presenting risk information—a review of the effects of ‘framing’ and other manipulations on patient outcomes
J Health Commun
Communicating the risk reduction achieved by cholesterol reducing drugs
Br Med J
Patients’ responses to risk information about the benefits of treating hypertension
Br J Gen Practice
Communicating evidence for participatory decision making
J Am Med Assoc
The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus
N Engl J Med
Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33)
Lancet
Effects of Intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34)
Lancet
Diabetes service provision: a qualitative study of newly diagnosed type 2 diabetes patients: experiences and views
Diab Med
Simple tools for understanding risks: from innumeracy to insight
Br Med J
Risk perception and communication: recent developments and implications for anaesthesia
Anaesthesia
Communication about risk: diversity among primary care professionals
Fam Practice
Different formats for communicating surgical risks to patients and the effect on choice of treatment
Pat Educ Counseling
Interpretation of graphic data by patients in a general medical clinic
J Gen Int Med
Cited by (0)
- 1
Polly Brown is employed by British Medical Journal Knowledge, the producers of the Best Treatments resources, whose content provided the basis for the content evaluated (in the control group) in this study.