An evaluation of strategies commonly used by health advocate programs

Abstract

Many non-urgent and routine medical procedures, such as MRI and CT Scans, are performed under standardized industry protocols with minimal process variations. However, large price variations exist among providers offering these services within the same geographic region. Many insurers and independent vendors have implemented health advocate/concierge programs to steer beneficiaries to high-quality and low-cost providers. These programs also aim to reduce costs. The Blue Cross Blue Shield (BCBS) of Texas’ Benefits-Value-Advisor (BVA) program is one of the most cited examples. It uses three key strategies: recommendation, unconditional monetary rewards, and persuasion. However, their effectiveness in influencing beneficiaries’ choices, particularly for routine medical procedures, has not been tested. This study fills the gap through a behavioral experiment, which provides insights into how beneficiaries may respond to these strategies and providing guidance for health plan managers. A full-factorial-between-subjects experiment was conducted with 500 subjects recruited through Amazon Mechanical Turk. The experiment design included three treatments: recommendation, copay waiver, and persuasion, each with two levels (Yes or No), resulting in eight distinct scenarios. Subjects were randomly assigned to one of eight scenarios. The survey data were analyzed using a logit regression model. We found that subjects who received recommendations were 32.7% more likely to select the lowest-cost provider and had a 27.3% greater chance of choosing lower-cost providers. However, the effectiveness of recommendations diminished when subjects mistrusted their insurance companies. Neither copay waiver nor persuasion had a significant impact on provider choice. We acknowledge that these findings are derived from a hypothetical experimental setting using an online sample, and that actual beneficiary behavior in clinical settings may differ due to additional factors such as physician referrals, appointment availability, urgency of medical needs and convenience. Our results should be interpreted as providing suggestive evidence rather than directly generalizable predictions of real-world behavior.