Claims Denial Investigation Release

Our Article

Today, we are releasing an article that analyzes publicly available U.S. health insurance claims denial data, including some new data we acquired via a public records request.

This constitutes the first of what we hope to be a series of in depth analytical investigations into problems plaguing U.S. healthcare.

The Problem

All indications from well-studied data suggest claims denials in U.S. health insurance are pervasive and problematic, but only a highly limited set of data has been made publicly available and studied in detail in the literature. We set out to dive deeper on existing public data to understand the extent to which well studied phenomena occur consistently across different market segments, and generally to improve understanding of the entire landscape of claims denial data.

We were limited in this endeavour by the fact that publicly available claims denial data is sparse, and what data exists is inconsistent.

Our Commitment To Reproducibility

To foster feedback from the community and continued reporting, analyses, and advocacy for more transparent access to a broader class of data, we are open sourcing all of the information necessary to completely reproduce our analyses end to end.

Open Article Sources

You can read our full, interactive article on the web here, or read, edit or recompile our full, static article on the web or in print by accessing this pdf, or this source latex.

Open Analysis Sources

You can also reproduce all of the data, static and interactive figures, numerical claims and analyses, and plots appearing in our articles by running the code in our open source investigations repository.

Each data source considered in our article is provided and served to the public by a government agency, or by us (in the case of public records we obtained), and the source code referenced above includes the entire process of passing from the raw records provided to the public (e.g. pdfs that are difficult to parse), to producing the numbers, plots, etc., quoted in our article.

Our hope is that by providing transparent and reproducible findings, we will empower the community by allowing them to validate our results, easily contribute incremental adjustments or improvements to our analyses, and help us find mistakes in our analyses and thereby ensure they become more accurate over time. We also hope to ultimately build trust in our outputs as a result.

Open Data Sources

Finally, we provide direct access, or guidance on how to acquire such access, to all data analyzed in our report. All data sources we consider except one are and have been made available to the public by government sources. We additionally provide direct access to the PA public records data we consider in the article that we obtained via public records request.