OIG: UnitedHealthcare Benefits of Texas (H0609) audit flags C78.00 mis-keys, $24.4M estimate

On September 23, 2026, the HHS Office of Inspector General issued report A-07-24-01215, a Medicare Advantage compliance audit of specific diagnosis codes that UnitedHealthcare Benefits of Texas, Inc. (contract H0609) submitted to CMS. OIG posted it September 28. According to OIG, medical records didn't support the reviewed codes for 172 of 230 sampled enrollee-years, and OIG estimated at least $24.4 million in overpayments for payment years 2020 and 2021.

This is the third report in OIG's high-risk diagnosis series we've covered since mid-September, after UnitedHealthcare of Wisconsin (H5253) and HumanaChoice (H5216). Same audit series, different contract, different numbers. What's new for the desk in the Texas report is a full table of potential mis-keys and a specific gap in how an acute-stroke edit was scoped.

What OIG audited in contract H0609

OIG says that as of December 2021, United covered 339,291 enrollees under H0609, and CMS paid United about $8.8 billion for payment years 2020 and 2021. The report notes H0609's official name then was PacifiCare of Colorado, Inc., which merged into UnitedHealthcare Benefits of Texas, Inc. in 2024.

OIG found 15,340 enrollee-years tied to its 11 high-risk groups (service years 2019 and 2020) and limited review to $33,255,917 in associated payments. It sampled 230 enrollee-years: 200 at random across 10 groups (20 each) and 30 nonstatistically from the mis-keyed group. Risk scores in that period ran on the V22 and V24 CMS-HCC models, not V28, so treat every HCC label below as an audit-period label.

Sample results by high-risk group

The pattern OIG describes is familiar. Across the stroke, AMI, embolism and five cancer groups, 96 of the 142 errors were records that showed a past history of the condition, not active disease at the physician visit. All 18 breast cancer errors were history-of. For a refresher on that line, see our explainer on acute vs history-of coding for HCCs.

OIG also rejected a few records outright as non-acceptable sources: echocardiogram results, a pulmonary function test, and a nurse note signed by a registered nurse who wasn't on the acceptable provider list. OIG said those records didn't meet CMS's acceptable data source requirements, so the HCCs couldn't be validated.

The mis-key table coders can actually use

OIG's mis-keyed group flags enrollees who had many codes for one condition and just one code for an unrelated condition that looks like a keying slip. Appendix E lists every pair OIG found unsupported. Here are five of them, including the largest:

Notice the shape: in each of these pairs, the digits match and only the leading letter differs. In one C78.00 case OIG describes, the enrollee had 17 E78.00 codes and a single C78.00, and OIG's reviewer couldn't find support for the lung code. OIG recommended that United review the 26 mis-keyed enrollee-years it didn't sample.

Why the acute stroke edit missed

Acute stroke codes failed for 19 of 20 sampled enrollee-years. For 13 of them, the record showed a prior stroke but nothing justifying an acute stroke code at the physician service.

The useful detail is in OIG's controls section. United officials told OIG the company had a data filter meant to stop inaccurate acute stroke codes, but that it applied only to in-home health and wellness visits and chart reviews. It didn't apply to diagnoses on claims from other sources, including those submitted directly by providers. OIG also noted United required 156 hours of training for new coders and a 95-percent accuracy rate, and still concluded its procedures for high-risk codes could be improved. If your own edits only cover some intake channels, that's the gap to close. Our acute stroke vs history of stroke guide covers the documentation side.

What OIG recommended and how United responded

OIG made four recommendations. United should:

• refund the $24,465,369 in estimated overpayments
• review the 26 mis-keyed enrollee-years OIG didn't sample and refund any resulting overpayments
• find similar errors after the audit period and refund them
• keep improving its compliance procedures for high-risk diagnosis codes

United disagreed with some findings and asked OIG to withdraw all four. It contested 25 sampled enrollee-years from the draft. OIG's independent medical review contractor reversed 9 of them and kept 16, which cut the error count from 181 to 172 and lowered the dollar figure. United also argued an FFS adjuster was required, that OIG ignored underpayments, that the sample was too small, and that OIG didn't follow CMS's RADV rules. OIG kept its recommendations, with the reduced refund amount.

Keep the legal status straight. The $24.4 million is the lower limit of a two-sided 90-percent confidence interval from a sample. OIG's own footnote says its recommendations aren't final determinations: CMS decides whether an overpayment exists, and the plan can appeal any disallowance through the RADV appeals process. This isn't a fraud finding or an FCA case.

What this means for your risk-adjustment queue

Use the Texas report as a QA list, not a headline.

• Run a mis-key screen. Flag any enrollee with one HCC-driving code that matches a repeated non-HCC code except for the first character (C78.00 vs E78.00, I62.9 vs G62.9, E43 vs I43). Pull the record before that code stays in the file.
• Make acute edits channel-blind. If an acute stroke or AMI edit only fires on chart reviews or in-home assessments, extend it to provider-submitted encounters.
• Check active cancer on a single DOS. Confirm treatment or active management before keeping an active cancer code, and use history-of when that's what the record says. The MEAT Criteria Guide is the bar.
• Reject non-encounter sources. Test results and notes from non-acceptable provider types don't support a risk-adjustment code on their own.

The ICD-10 Encoder helps you confirm the history-of alternative and billable status when a code doesn't hold up.