The impact of a charge transparency tool on outpatient provider markets.

Last registered on April 08, 2021


Trial Information

General Information

The impact of a charge transparency tool on outpatient provider markets.
Initial registration date
April 07, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
April 08, 2021, 6:16 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
April 08, 2021, 10:16 AM EDT

Last updated is the most recent time when changes to the trial's registration were published.


Primary Investigator

New York University

Other Primary Investigator(s)

PI Affiliation
PI Affiliation
New York University

Additional Trial Information

Start date
End date
Secondary IDs
Executive Summary:

While there is great interest in the potential for price transparency to improve the functioning of health care markets, there is conflicting evidence on the effects of transparency tools on consumer behavior and very limited evidence on the effects of these tools on provider pricing decisions. Prior research has largely focused on insurer-specific tools, making it difficult to assess market-level effects; and has mainly used weak study designs. The roll-out of the New York Healthcare Online Shopping Tool (NY HOST) offers a unique opportunity to systematically and rigorously assess the impact of transparency – here focused on provider charges -- on consumer and provider behavior.
FAIR Health will randomize the implementation of NY HOST so that while charge information will be released everywhere in the state, there will be variation in the specific procedures and provider types for which information is available across the state. The variation induced by this design, combined with the comprehensive charge information available from FAIR Health, will allow our team to conduct an exceptionally strong evaluation of the effects of transparency tools at both the consumer and market levels, including information on how consumer shopping behavior responds to the availability of the tool, whether providers respond to charge disclosure by lowering or raising charges, whether the tool reduces charge dispersion, and how allowed amounts respond to changes in disclosure of charges. We will also assess the impact of hospital charge disclosure by comparing the four hospital markets where NY HOST will be releasing charge information to four matched markets where data will not be disclosed.

This project will evaluate the impact of the NY HOST on consumer and provider behavior in New York’s outpatient healthcare and hospital markets. Supported by a grant from the New York State Health Foundation, FAIR Health has launched the NY HOST transparency initiative. The NY HOST website will build on the existing FAIR Health consumer tools by adding information on quality and prices. FAIR Health is an independent not-for-profit that collects private insurance claims data from approximately 60 claims administrators and private insurance companies nationwide. To contribute data to the FAIR Health database, a contributor must submit all its claims to FAIR Health. Currently, the FAIR Health database contains claims representing approximately 75% of the privately insured population of New York State, as well as Medicare Part A, B and D claims and sample of Medicare Advantage claims.

FAIR Health will use its claims data as the source of pricing information for NY HOST. Claims information in the data that will be licensed by NYU includes actual non-discounted billed fees for services, procedure codes, place of service and zip codes for the place where the service was rendered. These data also include limited, HIPAA-compliant demographic information on patients (age and gender). The data available for analysis will also include a de-identified indicator for each insurer and an indicator, when applicable, for type of insurance plan. The data will also include information on the mean, mode, median, 25th, 50th, 60th, 70th, 80th percentile levels from the imputed allowed amounts reported in the FH Allowed Medical Module for the relevant time period for each procedure in each NYS geozip (and zip code within Manhattan).

We will link these data to the Medicare Physician Compare database and the National Plan and Provider Enumeration System as the source of information on physician board certifications, year of medical school graduation, medical education, residencies, hospital affiliations and other physician attributes reported on the site.

In the outpatient care portion of the site, NY HOST will include specific billed charge data for identifiable providers associated with 100 common procedures (under randomization, information will be available for 75 procedures in each geozip as described below), de-identified aggregated billed charge data for each of the remaining CPT procedures in every geozip in New York, de-identified aggregated data reflecting the median imputed “allowed” amount for each CPT procedure at the geozip level, and estimated de-identified bundled median “charge” and “allowed” amounts for 25 procedures/conditions based on the risk-adjusted Prometheus Payment Methodology established by HCI3 (the Health Care Incentives Improvement Institute).

NY HOST will include 20 physicians (and other healthcare professionals—e.g., physical therapists, acupuncturists—as appropriate based on the procedures selected) on a landing page when the consumer selects a specific procedure in a specific geozip. The 20 professionals will be randomly ordered on the landing page, from among those who meet a pre-specified volume threshold. In addition to the 20 professionals featured on the landing page, NY HOST will also include a feature that will allow the consumer to click a button in order to expand their search of healthcare professionals. Provided other professionals in the geozip meet the relevant selection criteria (e.g., frequency threshold), they will be listed on secondary pages offered on the site. Healthcare professionals on both the landing page and secondary pages will also be able to be sorted by various criteria (e.g., gender, EMR, etc.)

In the hospital phase of the site, NY HOST will offer pricing information for all hospitals with more than 50 beds in each of 4 upstate communities: Buffalo, Syracuse, Albany, and North Westchester. Pricing information will be updated periodically.

Providers and hospitals will be permitted to opt out of charge disclosure.

As part of the NY HOST project, FAIR Health will market the new pricing product extensively through diverse channels including digital/online ads and social media ads as well as more traditional forms of outreach. These efforts will be targeted to consumers statewide, with an emphasis on consumer audiences with chronic conditions, bilingual Hispanic communities, the uninsured and newly insured. FAIR Health will survey site users and analyze non-identifying information about site visitors (e.g., Google analytics). FAIR Health will share geozip level aggregated information about site use with the study team.
External Link(s)

Registration Citation

Allcott, Hunt, Sherry Glied and Grace Kim. 2021. "The impact of a charge transparency tool on outpatient provider markets.." AEA RCT Registry. April 08.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


The intervention is a randomized controlled trial (RCT) embedded within the roll-out of a market-wide price transparency tool. The RCT was designed to have specific provider-level charge information released for a select set of randomized procedure and 3-digit geozip combinations across the state. For one set of procedures, provider-level charge information was released across the state. For another set of procedures, selected because they were both common and had a high rate of out-of-network use, provider-level charge information was released only in randomized procedure and geozip combination across the state, while those procedure-geozip combinations randomized to the control group had only aggregated median charge information posted on the website. The experiment ran from September 12, 2017 through August 30, 2019.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Our key outcome variables of interest include: utilization patterns for procedures that have been randomized to exhibit charge data in an area relative to those that have not; which physicians (i.e., low price or high price) opt in or out of charge disclosure; and changes in providrs' charges in areas randomized to charge disclosure relative to those in other areas.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
From a select set of 107 procedure codes, encompassing 30 categories, 11 categories were assigned to be randomized, and the remaining 19 categories were assignd to have all 50 procedures have provider-level charge information release across the state. Of the 57 procedures in the randomized categories, there were 7 that were new (as of 2017) and 3 that were discontinued. Since there are 31 geozips across the state, there were 31*57=1767 geozip-by-procedure observations to be randomized, in 341 geozip-by-category groups. The randomization algorithm assigned a random number to each category and, within each geozip, assigned categories with progressively larger random numbers until the T group has 25 or more procedures. The randomization algorithm chooses the max (over 1000 trials) of min pvalues in t-tests of equality of mean covariates between T and C. The randomization checks at the end of that the randomization has been carried out correctly, including that there are no missing data and that the min-pvalue is sufficiently large and indeed equals the max achieved in the trials.
Experimental Design Details
Randomization Method
Randomization done in office by a computer (Stata, StataCorp, LLC).
Randomization Unit
1767 geozip-by-procedure observations to be randomized, in 341 geozip-by-category groups.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
1767 geozip-by-procedure observations to be randomized, in 341 geozip-by-category groups.
Sample size: planned number of observations
1767 geozip-by-procedure observations to be randomized, in 341 geozip-by-category groups.
Sample size (or number of clusters) by treatment arms
948 in treatment arm; 819 in control arm.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials