Comments on the Transparency in Coverage proposed rule

The Department of the Treasury, the Department of Labor, and the Department of Health and Human Services have proposed revisions to the Transparency in Coverage (TiC) regulations. Matthew Fiedler and Yihan Shi submitted a comment letter responding to the proposed rule. In their analysis, Fiedler and Shi argue that adopting an appropriate standardized file format, such as Apache Parquet, is the single biggest step that the agencies could take to make the TiC data more usable. The key advantage of a standardized file format is that it would relieve data users of the need to write custom code for each payer’s files.

However, the choice of standardized format matters greatly. The JSON format that most insurers currently use is cumbersome for data users because it stores data inefficiently and is computationally costly to parse. CSV, the other format that the agencies discuss in the proposed rule, would likely outperform JSON, especially if the TiC schema’s nested structure were replaced with a multi-table relational structure. However, Apache Parquet would likely be a much better choice than either CSV or JSON. Parquet is explicitly designed to facilitate efficient reading, writing, and storage of large datasets, features that would allow for much lower file sizes and processing times. Parquet can also represent nested data structures (although it would likely still be advisable to move the TiC schema partway toward a relational structure alongside the change in file format). Importantly, Parquet also enjoys broad programming language support.

Fiedler and Shi also offer comments on two of the other proposals in the proposed rule:

• Including enrollment and utilization information in the TiC data could facilitate various types of analyses that are currently impossible. However, maximizing the utility of these data would require some modifications to the agencies’ proposals.
• Excluding service codes from the TiC data based on providers’ taxonomy codes could reduce file sizes and increase usability, but it is unclear how much. There are also steps that the agencies could take to mitigate the risk of inappropriately excluding service codes.

Read the full comment letter here