The $8 million lesson in Colorado's EVV audit

In July 2026, the HHS Office of Inspector General released its audit of Colorado's electronic visit verification system for Medicaid personal care services. The headline number was a recommendation that the state return roughly $8 million in federal funds, with a substantially larger amount set aside for CMS to evaluate.

What makes the report worth reading closely — even for agencies that don't bill personal care services — is the Colorado Department of Health Care Policy and Financing's response. HCPF disagreed with many of the conclusions, and its stated reason was that the majority of findings were, in its words, "isolated documentation issues," not evidence that services went undelivered or that caregivers were ineligible.

Read that again, because it is the most important sentence in Colorado HCBS compliance right now.

Why this matters to a host home agency

Host home providers don't use EVV the way personal care agencies do. The specific mechanics of the audit — location capture, visit records, attendant check-in — aren't your daily reality. The underlying logic absolutely is.

Every dollar in Colorado HCBS rests on a documentary claim: that a specific service was authorized, delivered, and described well enough to prove it. When an auditor arrives — CDPHE for compliance, HCPF or a federal reviewer for payment — they do not observe the care. They read the record of it. If the record is thin, generic, undated, unsigned, or inconsistent with the med log, the reviewer's finding is not "the care was poor." The finding is that the claim was not supported.

Unsupported is functionally identical to unearned when the money is recouped.

Three patterns the audit reinforces

1. Small, repeated documentation gaps extrapolate

HCPF's response noted that the audit's conclusions relied on extrapolation from a relatively small sample of findings. That objection is worth taking seriously as a methodological point — and it is also exactly why routine defects are dangerous. A reviewer does not need to find a problem in every note. They find it in a sample, calculate a rate, and apply that rate across the population. A defect that appears in five percent of your notes is not a five-note problem. It is a five-percent-of-revenue problem.

This is the single strongest argument for auditing every note rather than spot-checking. A sample-based internal review reproduces the auditor's math without reproducing the auditor's coverage.

2. "The service was delivered" is not a defense

Agencies consistently underestimate this. When a finding lands, the instinct is to explain what actually happened that day — the caregiver remembers the outing, the family confirms it, the person clearly received support. None of that is in evidence. The reviewer is evaluating a document created at the time, and testimony offered afterward carries little weight against a contemporaneous record that doesn't say what you now say it means.

The corollary is practical: the only reliable window to fix a note is while the caregiver still remembers the day. A defect caught the next morning takes two minutes to correct. The same defect surfaced eighteen months later during a post-payment review is unfixable, and the unit is simply lost.

3. Systems get audited, not just providers

Notice who was audited here. Not an individual agency — the state's system for ensuring that documentation happened. The findings concerned whether Colorado's procedures ensured that required verification information was collected at all.

Apply the same lens to your own agency. If a surveyor asked you to demonstrate not just that your notes are good, but that you have a system guaranteeing every note gets written, checked, and corrected — what would you show them? A binder of good examples is not a system. A documented, auditable process that catches the exceptions is.

What to do about it this quarter

• Audit every note, not a sample. If your internal QA reviews ten notes a month, you are measuring your documentation the same way an auditor does — badly, and after the fact. Coverage is the only defense against extrapolation.
• Shorten the correction window. Measure how long it takes from a note being written to a problem in it being identified. If the answer is longer than a day or two, most of your defects are already permanent.
• Check the note against the claim, not just against itself. A well-written note attached to a unit with a lapsed PAR is still an unbillable unit. Documentation review and billing review are the same review.
• Keep an evidence trail of your own oversight. When you catch and fix something, log it. A record of your agency finding and correcting its own defects is materially better in a review than a clean-looking file with no history.
• Run a self-survey before someone runs one for you. Walk the inspection points on your own schedule, score honestly, and fix what scores poorly while it's still cheap.

Sources: HHS OIG report A-07-24-03260 · HCPF response · The Colorado Sun. This article is general information for Colorado HCBS providers, not legal or billing advice.