The effects of price transparency and debt collection policies on intentions to consume recommended health care: A randomized vignette experiment

Abstract

New laws promote price transparency in health care, though effects on patient decision-making are not known. Price disclosure may increase the salience of cost and cause lower-income patients to decline recommended care, worsening inequities in health outcomes. Whether patients perceive a disclosed cost as higher or lower than their expectations may also affect care decisions, but has not been studied. Scholars and policymakers have paid much less attention to the question of whether patients will have to pay the prices charged (whether disclosed or not), and how expectations regarding collections may also affect healthcare consumption. Some hospitals aggressively collect on unpaid medical bills. Others hospitals do not. Actively disclosing collection policies (whether aggressive or protective) could magnify or counteract effects of price disclosures, especially for low-income patients. To test the effect of price disclosure and debt-collection disclosures on willingness to obtain recommended care, we recruited a nationally representative sample (N = 2997) and deployed a full factorial, controlled experiment in a standardized clinical vignette model. We find that disclosing a higher-than-anticipated price increases the probability of declining recommended care (odds ratio = 1.900), with larger effects for low-income individuals. Even more, disclosing aggressive collections increases the risk of declining care (odds ratio = 4.493), at higher rates for low-income patients. Where patients fear collections, but do not know prices, they are most likely to decline care. Disclosure of an aggressive collections policy makes patients feel less informed, harms patient trust in providers, makes them feel that they were not treated fairly, and undermines their confidence in the value of their care. Mediation analysis shows that about half of the effect of collections risk is via these attitudinal variables.