Price Points is a publication that creates novel, quantitative, public research using Turquoise Health price data and other healthcare datasets. This repository contains replication code for all Price Points projects, as well as general research and exploratory analyses related to healthcare prices in the United States.
We're making this repository open-source because we want to:
- Publish our methods for scrutiny and replication by industry experts, academics, etc.
- Build a community of practice around price transparency data research
- Provide examples of working with/analyzing Turquoise Health data
- Support developers and others who want to perform their own analyses
For a full list of completed and ongoing research projects, see the Projects README.
Important
All projects are released on an informational basis only and are not official Turquoise Health products. Please use the dedicated email below for any questions about Price Points projects.
Price Points primarily uses Turquoise Health hospital and payer rates data for its research.
Secondary data includes Census/ACS data, TIGER/Line shapefiles, travel times, Dartmouth Atlas data, etc.
Whenever possible, Price Points will publish replication data along with each analysis. Such data is typically a subset or aggregated version of the underlying Turquoise Health rates data.
In cases where publishing replication data isn't possible (due to size or licensing restrictions), we'll try to make the data available via other means e.g. research/licensing agreements, the Turquoise Community tier, etc.
If you're a researcher and want to use Turquoise Health data, you currently have three options. In order, from least to most access:
- Request access to Turquoise research datasets, which contain a limited subset of hospital and payer negotiated rate data.
- Request (free) researcher access to the Turquoise backend through the Community Tier. This provides nearly full access to the main Turquoise Health rates tables, but limited customer support.
- Contact the Turquoise Health sales team for full access to the underlying data (including historical rates).
Any research related to healthcare is in-scope, as long as the data exists to support the analysis. That said, we tend to choose projects that most benefit from the scale of Turquoise rates and other national datasets. That means analyzing things at the national or state level, rather than at the individual payer/provider level.
We try to show our work, fact-check/review rigorously, and incorporate solid domain knowledge, but healthcare is complicated and we won't always get things right. If you've found an error, bad assumptions, or missing context in one of our research projects, please create a GitHub issue or reach out directly.
See the Contact section below.
I’m Dan Snow, a data scientist and policy wonk currently living in the Bay Area. I'm employed by Turquoise Health, which provides the data, domain knowledge, and funding needed to do this work.
This repository contains three sections:
- Projects - Contains the code and data for each project. See the dedicated README for more info.
- Analyses - Contains exploratory and one-off research code unrelated to any specific project.
- Packages - Dedicated Python package for this repository. Contains helper functions, DB connections, etc.
Code in this repository uses the MIT license. Datasets and other linked assets may use different licenses.
Please cite Price Points work where appropriate. See the CITATION file or the GitHub sidebar for APA/BibTeX citation templates.
You can reach out to Dan directly at dan.snow@turquoise.health.