Using conjoint analysis to model the preferences of different patient segments for attributes of patient-centered care

Abstract

Background: A consensus regarding the components of a patient-centered approach to healthcare does not exist. Although patient-centered care should be predicated on patient preferences, existing models provide little evidence regarding the relative importance of different care processes to patients themselves.

Objective: To involve patients in the design of a model of patient-centered care for a corporation of Canadian teaching hospitals.

Methods: Using themes from focus groups and interviews, a conjoint survey was developed comprising 14 four-level patient-centered care attributes. Sawtooth Software's Choice Based Conjoint module (version 2.6.7) was used to design the survey. Each participant completed 15 choice tasks, each task presenting a choice between three hospitals described by a different combination of patient-centered care attribute levels. Latent class analysis was used to identify segments of participants with similar patient-centered care choice patterns. Randomized First Choice simulations were used to predict the percentage of participants in each segment who would choose different approaches to improving patient-centered care.Representative hospital service users were recruited from a corporation of five Canadian teaching hospitals serving a regional population of 2.2 million.

Results: A total of 508 patients and family members of children completed a choice-based conjoint survey. Latent class analysis revealed two segments: an informed care segment and a convenient care segment. Participants in the informed care segment (71.3% of the sample) were more likely to have higher education, be non-immigrants, speak English as a first language, and be outpatients or family members.The information needed to understand health concerns, an opportunity to learn health improvement skills, teams that communicated effectively, short waiting times, and collaborative treatment planning were more important to the informed care segment than to the convenient care segment. Convenient settings, a welcoming environment, and ease of internal access exerted a greater influence on the choices made by the convenient care segment. Both segments preferred hospitals that provided health information and gave prompt feedback on patient progress.

Conclusions: This study suggests that many patients would exchange an increase in waiting times for prompt feedback, information, and the skills to improve their health.