The Impact of a Commercial Health Plan Price Transparency Tool on Patient Choices

To engage patients and to enable value-based decision-making, calls for price transparency in health care are increasing.  Yet there is very little evidence of the impact of health care price transparency tools on patient choices.  We evaluate the impact of Aetna’s Member Payment Estimator (MPE), a sophisticated, online price transparency tool comparable to those commonly available via private health plans and other third-party vendors, on provider choices and costs for a non-elderly, insured adult population.  The MPE provides real-time, personalized, episode level price estimates, including both total paid costs and patient out-of-pocket costs, at specific providers within a specified geographic area.

We estimate the impact of tool availability on patients’ provider choice and costs for ten medical procedures, including MRI/CT scans, Colonoscopy, Mammogram, Cataract, and Hernia repair that had price estimates available in 2011-12 and were among the most commonly searched (“commonly searched”) and Total Hip Replacement, Total Knee Replacement, Sleep Studies, Echocardiogram, and Carpal Tunnel Release that were introduced in 2012 (“2012 procedures”).

Data include query-level MPE utilization data linked to 2010-2012 medical claims for 318,140 enrollees who ever used the MPE to search for one of these procedures (“searchers”), 2011-2012 medical claims for a comparison sample of 71,468 enrollees who had one of the procedures, had access, but never used the tool (“non-searchers”), and 2011-2012 claims for a sample of 29,397 patients who had one of the procedures but did not have access to the MPE (“control group”). No access was due to market-level barriers (e.g., long-standing contractual arrangements as to how claims were adjudicated in those markets) that were unrelated to prices or demand for price information.

First, we exploit variation in the tool’s introduction in different markets and implement a pre-post difference-in-difference approach to isolate the effect of the MPE on spending for the “2012 procedures.”  We estimate linear regression models that control for patient characteristics (age, presence of a major medical comorbidity) for both the population overall (i.e., intent-to-treat) and for searchers (i.e., treatment-on-the-treated), the latter through an instrumental variables approach and through a propensity-weighted approach.  Analyses are weighted to adjust for sampling.

The cost estimates from the MPE are not binding, as such patient choice of provider with lower quoted prices may not be reflected in actual spending. For the “commonly searched” procedures, we examine whether searchers shifted to providers with lower expected total and out-of-pocket costs (e.g., lower quoted prices on the MPE) in 2011-2012 in comparison to the mix of providers they visited in 2010 before the MPE was available.

We find varying rates of tool use and impact by procedure.  However, searching for price information had no effect on actual spending or patient deductible spending in the intent-to-treat analyses. This is driven by the fact that a low proportion of patients (range 3% - 5% across procedure) used the MPE before their procedure. These findings provide important evidence that can inform policymaker and payer efforts to inject price information into health care so it is most effective.