I don’t mean that to be provocative, I mean it literally. The number any given hospital revenue cycle team reports as its denial rate is almost certainly not its denial rate.
Upstream Revenue Loss
Most health systems measure denials through their payer remittance files, the 835 transactions that come back after a claim is adjudicated. If a payer sends a denial CARC code, it gets counted. If it doesn’t, it doesn’t exist. The problem is that a significant and growing portion of what should count as a denial (or any sort of adverse payment outcome) never produces a CARC code at all.
We see this constantly in our client data. Clinical denials, including level-of-care scrutiny, medical necessity disputes, and DRG downgrades submitted via takebacks after the claim was initially paid (!), are increasingly handled before the claim is ever submitted. Across Sift’s clients in 2025, over half of medical necessity denials never generated a CARC code. They just disappeared into the data as if they never happened, but the revenue impact was very real.
This isn’t an accident. Payers have restructured how they communicate adverse decisions in ways that are harder to measure and harder to appeal, moving denial activity outside of the ‘traditional realm of 835 and claim adjudication’ into concurrent and takeback scenarios, further shifting the operational burden onto providers.
Payment Reshaping
There’s a second blind spot. Payers are also moving away from discrete denials toward what I’d call payment reshaping, approving inpatient stays but reimbursing them at a lower severity or classification than the care delivered warranted. The claim wasn’t denied. It was paid, just not at the expected amount. Revenue erodes without a denial record to trace it to, and health systems’ net collection rate declines while the denial rate looks fine.
Neither of these patterns shows up in standard denial reporting. To get the longitudinal view needed to fully understand payer reimbursement, you have to layer in clinical data, including UR activity, level-of-care correspondence, post-payment review letters, and connect that payer correspondence with the clinical context to financial outcomes. Most health systems have these data sources. They just live in separate systems, managed by separate teams, with separate workflows, and no mechanism to connect them.
It’s no longer feasible to treat denials management as an operations problem owned by revenue cycle, throwing more staff and workflows at it, and measuring progress with metrics that tell only part of the story.
You can’t fix what you can’t see, and you can’t see it if the measurement infrastructure was designed for a world where payers played by the rules that providers expect. They haven’t been playing by those rules for a while now.
The first step in addressing this mismeasurement isn’t technology. It’s redefining what counts as a denial. If a payer communicates a decision that results in lower reimbursement than the care delivered, whether through a CARC code, a concurrent review letter, a portal notification, or a post-payment takeback, that is a denial (we call them adverse payment outcomes), and it should be counted as one.
Until health systems measure the real problem, they’ll keep solving the visible one.
