JMIR Publications

ABSTRACT

Background:

Two barriers to effective enrollment decisions are low health insurance literacy and lack of knowledge about how to choose a plan. To remedy these issues, digital decision aids have been used to increase knowledge of health insurance plans options and to guide the decision process. Previous research has shown that the way information is presented in a decision aid can impact consumer choice, and existing health insurance decision-aids vary in their design, content, and layout. Commercial Virtual Benefits Counselors (VBC) are digital decision aids that provide decision support by mimicking the guidance given by an in-person Human Resources (HR) counselor, while more traditional HR websites provide information that requires self-directed navigation though the system. Yet, little research has compared how decision processes are impacted by these different methods of providing information.

Objective:

The objective of this study is to study how individuals interact with two different types of health insurance decision aids (guided virtual benefits counsellors that mimic conversations with a human HR counselor and self-directed HR websites that provide a broad range of detailed information) to make employer provided health insurance decisions.

Methods:

Sixteen employees from a local state university were recruited to make mock employer provided health insurance decisions using one of the two systems (VBC vs HR website). Decision outcomes were examined using pre- and post-experiment measures of literacy and knowledge, as well as examining decision conflict and usability of the system. Participants also took part in a retrospective think-aloud interview, cued using eye tracking and gaze data, about their decision processes and interactions with the decision aid.

Results:

Both the VBC and HR website resulted in better post-experiment health insurance knowledge, Δ = 19.0%, Z=3.4, p<.001. However, differences in health insurance literacy and knowledge before and after interacting with the decision aid did not depend on the system they used. Furthermore, participants rated both systems as similarly useful, Δ=0.4, t(14.0)=0.25, P=.80, ns, and mentioned similar decision factors during their decision-making process; however, the VBC was rated as being easier to use, Δ=2.9, t(13.4)=2.1, P=.06. Finally, participant confidence in their health insurance enrollment decision did not depend on which system they used. However, individuals who had lower pre-experiment knowledge were more conflicted about their hypothetical enrollment choice, P=.03, and focused on different decision factors than those with higher pre-experiment knowledge.

Conclusions:

This study showed that health insurance decision aids help individuals improve their confidence in selecting and using health insurance plans. However, prior health insurance knowledge plays a large role in how users interact with and benefit from decision aids even when information is presented in different formats.