Same information, different decisions: format counts

The function of information is to help us make better decisions.¹ The amount of clinical information, measured by journal articles, has doubled over two decades,² but thanks to evidence based approaches the content now seems more reliable. For example, the review article has changed from a vehicle to advance the author’s reputation³ to a balanced synthesis of evidence we can safely use to inform clinical and policy decisions.⁴ To justify this special position, much effort is expended on assembling reliable content—from comprehensive literature searches⁴ to peer review and the editorial process.⁵ However, just assembling the right words and data is not enough to ensure that better decisions will be taken. Elting’s study in this issue shows that doctors’ decisions can easily be manipulated by changing the format of words and data—for example, by presenting information in tables, graphs, or pictograms (p 1527).⁶ What is more, the formats preferred by doctors were not the formats which led to optimal decision making.

These disquieting results cannot be dismissed as isolated findings due to a quirky sample or the decisions studied. Considerable evidence exists that the format in which information is presented significantly influences clinical decisions, ranging from the accuracy of obstetric judgements⁷ to the speed of interpreting laboratory tests⁸ or intensive care data.⁹ As shown vividly by pharmaceutical advertisements, the potential for influencing doctors’ decisions by changing the format of information extends beyond the choice of graphics to the statistics used to describe trial results¹⁰ and how options are worded, such as “chances of survival” versus “chances of death.”⁶

Why should clinicians worry now about this problem, which previously troubled only psychologists?¹¹ The increased affordability, power, and connectivity of information technology are providing many institutions with an opportunity to revise the format and delivery of medical records,¹² directories of local services, drug formularies, and other sources of clinical knowledge.¹ However, before we redesign records and knowledge resources for electronic access we must fully understand how to format such information to make it easy to find and clear to interpret. Otherwise, we risk propagating information in formats which, by ignoring subtle design principles, will mislead on a large scale.⁷,¹²

Fortunately, the discipline of information design offers us a way forward. Information designers use their skills in psychology, graphical design, perception, and typography to make information, from railway timetables to electricity bills, more accessible and usable.¹³ To ensure that the format of clinical documents is as reliable as their content, those of us who design records, knowledge resources, and other clinical information systems need to tap this skill.¹⁴ We must also ensure that the formats we select lead to appropriate decisions by empirical study,¹⁵ not merely by asking clinicians to indicate a preference, since preferences can mislead.⁶ Marshall McLuhan perhaps overstated his point when he claimed that the medium is the message, but the results of Elting et al show that we must not only make the content of clinical documents evidence based, but also develop formats appropriate to electronic and paper media, and test the effects of these formats on clinical decisions. A recent series explored these issues in depth.¹⁶

Information in Practice p 1527