A first look at CMS's own Removed and Deleted Records file, and what the correction surge means for your next audit.
Every program year, the Centers for Medicare & Medicaid Services (CMS) publishes a Removed and Deleted Records file alongside the main Open Payments dataset. Across the full 2023 and 2024 files, the number of General Payment records that reporting entities corrected or deleted after publishing them rose from 896 to 160,183, a jump of almost 179x in a single year.
That's not an estimate or a sample. It's CMS's own correction log, covering every General Payment record removed or deleted industry-wide. And it's one of the least-discussed pieces of Sunshine Act disclosure data, even though it may be the single most CMS-native audit-risk signal in the entire dataset.
CMS defines the two categories precisely, and the difference matters for how you read the number:
Both categories describe the same underlying event from a compliance standpoint: a manufacturer published a payment record, then walked it back. Whether the correction was a data-entry fix, a reclassification, or something more serious isn't visible in the aggregate file. What is visible is how often it happened, and how fast that changed.
| Program year | Removed/deleted records | Share of that year's volume |
|---|---|---|
| 2023 | 896 | 0.0061% |
| 2024 | 160,183 | 1.0335% |
Put in plain terms: roughly 1 in 16,000 General Payment records was corrected after publication in 2023. In 2024, it was roughly 1 in 97. Almost all of that increase, 99.7% of the 2024 total, was General Payments rather than Research Payments.
CMS's own audit guidance, Open Payments: Audit Best Practices 2026, lays out what draws attention under issue-based selection: inaccuracies in amount, an incorrectly reported recipient, records that should have been reported but weren't, entity size, and the relative rate of disputed payments.
A self-initiated deletion sits inside that same category, arguably a more serious version of it. It's an entity admitting a published figure was wrong enough to pull. A shift this large, across the industry, in a single year, is exactly the kind of pattern CMS's own criteria point at.
You don't need CMS's full dataset to run a version of this check on your own reporting entity:
This is one of three signals qordata found by processing the full three-year Open Payments dataset directly, with no sampling. The other two look at where payment spikes concentrate and how often research and general payment relationships overlap for the same physician.
It's a file CMS publishes every program year alongside the main Open Payments dataset. It lists every record that was DELETED (pulled from the system by the reporting entity after publication) or REMOVED (dropped from current publication because an update made it ineligible).
Removed and deleted General Payment records rose from 896 in 2023 to 160,183 in 2024, a roughly 179x increase. That moved the correction rate from about 0.0061% of records to 1.0335%.
Not by itself. It means a reporting entity corrected data after publishing it. But CMS's own audit criteria list records that should have been reported but weren't as a factor in issue-based selection, and a self-initiated deletion sits in that same category.
Compare your reporting entity's correction rate for the most recent program year against the 2024 national benchmark of 1.03%. qordata's CMS Audit Risk Report includes the full methodology and benchmark, and qordata's CARE team can run this check against your own reporting data directly.
