Introducing Price Points

Opening a treasure chest full of healthcare data together

Way back in 2021, the Centers for Medicare and Medicaid Services (CMS) mandated that U.S. hospitals start publicly posting their prices online. The goal was to “increase market competition, and ultimately drive down the cost of health care services.” CMS also explicitly wanted to give journalists, researchers, and policymakers some data to chew on, with the hope that they too would help create a more efficient healthcare system.

It’s now 2025, and compliance with the CMS mandate is… surprisingly good. Hospitals were initially slow to respond, but CMS stepped up fines + enforcement and created a standardized schema, so most hospitals (~90%) now post some kind of machine-readable file (MRF) containing their prices. But compliance ≠ results, and the actual impact of dumping 5,000+ MRFs onto the internet is a little less clear.

For instance, no one can currently say whether the CMS mandate actually increased competition. Prices are moving - descriptive statistics suggest they are converging - but no large studies to date have quantified a causal effect (studies of the entire U.S. healthcare system are difficult, to say the least).

The impact on research and policymaking has likewise been muted. Collating every hospital’s MRF is a Herculean task, and once combined, the data is massive and requires substantial care/domain knowledge to use. As such, the flurry of whitepapers and reports that CMS probably hoped for hasn’t materialized. Most studies that use price transparency data are small and focused on specific markets and/or procedures. Few have looked at the national picture or dug deep into healthcare pricing and contracting.

However, one result of the CMS mandate is clear. There’s now a huge amount of valuable new healthcare data to sort through, including negotiated rates for every common procedure and drug across almost every hospital and insurer in America. This data is a vast ocean of prices that few have deeply explored, so starting today, I’m going to dive into it.

Once every few weeks, I’ll publish novel, quantitative analyses using price transparency data from Turquoise Health. Turquoise is paying me to do this as an FTE, but doesn’t tell me what to publish – or what not to publish, other than insisting I remain relatively neutral. The goals here are simple: unpack the data, fill research gaps, and explore price transparency’s impact – and do it all in public to help others with the same goals.

To that end, I’ve created a public GitHub repo which will contain the code and methods for each analysis. I’ll also include any datasets necessary for replication, as long as licensing allows (not all data used by Turquoise is theirs to share, e.g., claims data). My hope is that sharing code, data, and domain knowledge will improve the price transparency ecosystem, make research easier, foster more collaboration, and ultimately drive down healthcare prices.

So with that, let’s dive in, starting with a question: