Online usability and patients with long-term conditions: a mixed-methods approach

To improve the information position of health care consumers and to facilitate decision-making behavior in health the Dutch ministry of Health commissioned the National Institute for Public Health and the Environment to develop, host and manage a public national health and care portal (www.kiesbeter.nl) on the Internet. The portal is used by over 4 million visitors in 2010. Among them, an increasing amount of patients that use the portal for information and decision making on medical issues, healthy living, health care providers and other topics.

First objective is to examine what usability aspects of the portal kiesBeter.nl matter for chronic patients and their informal carers with regard to information seeking, self-management, decision making, on line health information and other variables. Second objective is to make evidence-based practical recommendations for usability improvement.

An innovative combination of techniques (semi-structured interviews; eHealth Literacy scale; scenario-based study using think-aloud protocol and screen capture software; focus group) is used to study usability and on line information seeking behavior in a non random judgment sample of three groups of patients (N=21) with long-term medical conditions (arthritis, asthma and diabetes).

The search strategy mostly used (65%) by the relatively well-educated subjects is 'orienteering'. Users with long-term conditions and their carers expect tailored support from a national health portal, to help them navigate, search and find the detailed information they need. They encounter serious problems with these usability issues some of which are disease-specific. Patients indicate a need for personalized information. They report low impact on self-management and decision making. Overall judgment of usability is rated 7 on a Likert type 0-10 scale. Based on the outcomes recommendations could be formulated. These have led to major adaptations to improve usability.

A non-representative composition of a small non random judgment sample does not permit generalization to other populations and cognitive bias cannot be quantified. However if mixed methods are applied valid conclusions can be drawn with regard to usability issues.