Top Denial Reason Codes in Medical Billing and How to Fix Them
Understanding denial codes as fixable problems, not verdicts, recovers billions annually.
A denial code is a payer telling a practice, in a compressed and often unhelpful shorthand, what went wrong with a claim. It is the payer telling a practice, in a compressed and often unhelpful shorthand, what went wrong with a claim. It doubles as a repair instruction if you know how to read it. The trouble is that most billing teams treat denial codes as a verdict rather than a diagnosis, and that habit is what turns a fixable error into a permanent write-off.
The scale of the problem justifies the attention. Roughly 9% of claims submitted by U.S. hospitals get denied on first submission, and that 9% works out to about $262 billion in initially denied claims every year https://www.sprypt.com/denial-codes/what-are-denial-codes. Commercial payers run their own denial rates, somewhere between 10% and 16% depending on the payer, so the exposure spans every segment of the industry https://www.sprypt.com/denial-codes/what-are-denial-codes. None of this is cheap to fix, either. The average cost to resolve a single denial hit $57.23 in 2025, up from $43.84 the year before, and that trajectory compounds fast when a practice lets denials pile up unworked https://www.sprypt.com/denial-codes/what-are-denial-codes. A practice sitting on a stack of unresolved denials isn't just losing the claim value; it's paying more each year just to chase the same dollar.
Two categories of denial appear throughout this landscape, and the distinction changes what a billing team should do next. Hard denials mean the payer won't pay without a formal appeal: coverage limitations, medical necessity determinations, authorization gaps. Soft denials mean the payer just wants more information before it pays, and those are correctable and resubmittable if someone catches them quickly. The same denial code can sometimes fall into either bucket depending on circumstance, so the operational response has to match the type stamped on the remittance.
A coding system underlies both categories, and understanding it is necessary before any of the specific codes make sense. A CARC, or Claim Adjustment Reason Code, is the primary "why payment differs" code, and it's the type of code this guide is built around. A RARC, or Remittance Advice Remark Code, travels alongside the CARC and supplies the texture the CARC leaves out: what specifically is wrong, or what to do about it. Then there's the prefix. CO, for Contractual Obligation, means the adjustment is a write-off under the contract between provider and payer, and it isn't something a practice can turn around and bill the patient for. PR, for Patient Responsibility, means the opposite: the patient owes that balance, and billing should go to them, not back to the payer. Reading the CARC and RARC together, rather than treating either in isolation, gets a billing team to the correct fix faster and with fewer wrong guesses. That pairing is the whole method behind everything that follows.
CO-16: the claim is missing or wrong on something the payer won't specify without prompting
CO-16 means the claim is missing required information or contains some kind of submission error, and the payer's system stops processing it right there and kicks it back. It is, by most measures, the single most common denial code in medical billing, and also one of the most frustrating, because the CARC by itself refuses to say what the actual problem is. That information lives in the RARC riding alongside it. Billing teams that skip the RARC and start guessing waste a resubmission attempt on a fix that might not even be the real issue, and every wasted attempt pushes the claim closer to its timely filing deadline.
The root causes tend to cluster at a handful of predictable points in the workflow. A wrong date of birth, a misspelled name, or an incorrect insurance ID entered at registration will trigger it, and so will a missing or invalid NPI on the claim form. Incorrect modifier usage, a missing prior authorization number on the claim face, or a diagnosis code that isn't properly linked to its procedure line all produce the same CO-16 response from the payer's side, even though the underlying error is completely different in each case.
The fix follows a fixed sequence, and skipping steps is what causes repeat denials. Reading the RARC is step one, because it pinpoints the exact field or data element causing the rejection. Step two, calling the payer, only happens if the RARC still leaves the error ambiguous after that reading, and the point of the call is to confirm the exact problem before anyone touches the claim. Step three is correcting that specific field, checking that the surrounding data is clean, and resubmitting. Resubmitting on a guess is the expensive mistake here: a second CO-16 on the same claim doesn't just cost more time, it can close the window for appeal entirely.
CO-18: the duplicate claim trap that wastes time and delays payment on a claim that may already be in queue
CO-18 means the payer believes it has already received this claim, or something close enough to it, for the same patient, same provider, same date of service, same service line. It sounds like a simple clerical hiccup. In practice it splits into two scenarios that call for entirely different responses, and confusing one for the other is where billing teams lose time.
The first scenario is a true duplicate: the billing team submitted the identical claim twice, whether through a batching error or a manual resend. The response there is straightforward: cancel the second submission and confirm the first is moving through adjudication normally. The second scenario is the one that does real damage. A false duplicate happens when the original claim was denied and then resubmitted incorrectly as a brand-new claim instead of a corrected one, so the payer's system sees two submissions for the same service and flags the second as duplicate activity. The original denial, meanwhile, sits unresolved. It just sits there, unworked, while the clock on timely filing keeps running underneath it. Without solid visibility into claim status before anyone resubmits, a billing team can genuinely believe a claim was corrected when it never was, and that blind spot is usually a systems problem: multiple billing platforms generating the same claim from different entry points, or a manual resubmission process with no duplicate check built in anywhere along the way.
CO-29: timely filing denials, where the money is usually gone before anyone notices
CO-29 means the claim showed up after the payer's filing deadline had already passed. Most payers give a year from the date of service, but that's a generalization, not a rule; some commercial payers cut that window down considerably, and treating every payer as if it runs on the Medicare standard is how practices get caught out.
What sets CO-29 apart from nearly every other denial on this list is how final it is. Most denial codes describe a fixable problem. CO-29 usually describes money that is already gone, because appeals only succeed if the practice can prove the payer itself caused the delay, and that's a hard bar to clear in most cases. Timing makes this worse in a specific, measurable way. Claims sitting unresolved past 90 days have only a 50% to 60% chance of ever getting collected, and once a claim passes 120 days that probability drops below 30% https://www.sprypt.com/denial-codes/what-are-denial-codes. CO-29 denials tend to occur right around, or past, those exact thresholds, so the claim was often already a poor bet before the denial notice even arrived.
The causes are less about coding errors and more about workflow discipline. Claims sit unsubmitted on someone's desk waiting on a provider note that never comes quickly enough. Payer-specific deadlines go untracked, so a commercial plan running a 90-day filing window gets treated the same as a Medicare claim on its 12-month window, right up until it's too late. Denied claims get resubmitted as fresh new claims without anyone checking whether the original filing deadline has already closed, and system processing errors delay transmission without ever alerting the billing team that anything went wrong.
The fix is mostly procedural, not clinical. Every payer's deadline needs to live in one centralized reference that the whole team can see, rather than in one biller's memory. Claims should go out within 24 to 72 hours of the date of service as a fixed operational rule. Aging reports need to run weekly, and any claim closing in on 60 days with no adjudication signal should get flagged for immediate follow-up, not left to age further. When a CO-29 does land despite all that, the response is narrow: confirm the actual deadline that applied, pull the clearinghouse confirmation proving timely submission, and appeal only when that evidence shows the practice was on time and the payer lost or mishandled the claim. Absent that evidence, the appeal wastes the same time that caused the problem.
CO-22: coordination of benefits denials that stem from front-end eligibility gaps
CO-22 means the claim went to the wrong insurer entirely, because another payer was actually primary and should have been billed first. It's a sequencing error, not a data error, which is part of why it slips past people who are checking claims for accuracy rather than checking claims for order.
Coordination of benefits, or COB, is the system that decides which insurer pays first when a patient carries more than one active plan. When a patient has dual coverage, a COB order sets one plan as primary and the other as secondary, with the secondary picking up whatever the primary leaves on the table. Bill the secondary payer first, even with flawless information on both plans, and the claim comes back as CO-22 simply because it hit the payer system out of sequence. The common triggers are almost always front-desk gaps: a patient hands over just one insurance card, nobody asks about a second plan, or a record never gets updated after a coverage change.
Fixing it means identifying the actual primary payer first, which often takes a direct conversation with the patient or a real-time COB check run through the clearinghouse or the payer's own portal. Submit to that primary payer, wait for the EOB, and only then bill the secondary with that EOB attached. COB order isn't fixed once and done, either. Employment changes, Medicare's birthday rule, and open enrollment all shift which plan is primary, so an assumption that held true before can be wrong now. Prevention comes down to collecting complete insurance information at every visit rather than trusting whatever's on file from intake, and using real-time eligibility verification that actually surfaces COB status instead of a simple active-or-inactive flag. Patients with known dual coverage deserve a COB confirmation step before every single encounter.
CO-197: prior authorization denials and their growing frequency
CO-197 means a required prior authorization, precertification, or referral was never obtained before the service happened, or what was obtained doesn't line up with what actually got billed. It reads like a pure workflow failure, and often it is, but the environment around prior authorization has also gotten measurably harder to work in, independent of anything a practice does right or wrong.
A KFF analysis of the first-ever publicly reported prior authorization data found that insurers denied at least 1 in 8 standard requests across Medicare Advantage, Medicaid managed care, and ACA marketplace plans in 2025. That average hides enormous variation. Among Medicare Advantage plans specifically, denial rates ranged from as low as 2.3% at Alignment Healthcare up to 32.7% at UCare, against a program-wide average of 9.0% across roughly 590 reporting contracts. A patient's odds of getting an authorization denied, in other words, depend heavily on which plan they happen to carry, not just on the clinical facts of their case. And when practices do push back, the appeals data cuts against the payers: two-thirds of appealed authorization denials in Medicare Advantage got overturned, so the initial denial was simply wrong, with the entire burden of proving that falling on the practice.
At the practice level, the causes are concrete and mostly preventable. The service was scheduled before authorization ever got requested. Authorization was approved for a different provider, facility, or procedure than what ultimately got billed. Authorization expired somewhere between approval and the date of service, which occurs constantly in therapy plans, imaging series, and any procedure that gets rescheduled once or twice. Emergency services get delivered without any documentation showing an attempt to secure authorization was even made.
The fix starts with speed. The moment a CO-197 comes in, check whether the payer allows retroactive authorization for that circumstance, because some do and most don't, and the window to act closes fast. Where retroactive authorization isn't available, the appeal has to lean on clinical necessity documentation and, where it exists, evidence that the authorization process was started and failed because of a payer-side error, not a practice-side one. Every authorization number, approval date, valid-through date, approved CPT code, and approved provider or facility belongs in the EHR, tracked as structured data, not scattered across paper files or old email threads. High-risk categories like imaging, multi-visit therapy approvals, and anything a payer has recently added to its authorization-required list deserve a reconfirmation step before service delivery, every time, regardless of what was approved weeks earlier.
CO-50 and CO-96: medical necessity and non-covered service denials
CO-50 appears when the payer decides a service wasn't medically necessary. A provider orders an advanced diagnostic test, the claim goes out without documentation explaining why that specific test suited that specific patient's condition, and the payer denies it. The payer, in that moment, is saying the record submitted alongside the claim doesn't justify paying for it. It's saying the record submitted alongside the claim doesn't justify paying for it, which is a documentation failure, not a medical one.
CO-96 covers non-covered charges, and it gets treated as interchangeable with CO-50 when the two aren't. CO-96 can mean a flat plan exclusion, where the service is never covered under that plan no matter how well it's documented, or it can mean a medical necessity failure, where the service is covered in principle but the documentation on file doesn't support it for this particular patient. The two situations call for opposite responses. A medical necessity failure routes to appeal, and the only lever available in that appeal is stronger documentation.
Both codes tend to trace back to the same handful of root causes. A diagnosis code that's too generic to justify the procedure attached to it. A provider note that states what was ordered without ever explaining why: "ordered MRI" sitting alone in the chart, with no clinical rationale attached. An ICD-10 code that, on paper, doesn't match the payer's specific coverage criteria for that CPT code, even when the clinical picture actually supports it. And, often invisibly, a payer's medical necessity policy that got updated without the practice ever finding out, so a service that was reimbursable last quarter isn't this one. None of these causes are exotic. They're documentation gaps, and documentation gaps are the one category of denial a practice has full control over closing, well before a claim ever reaches a payer's desk. Each denied claim costs an average of $43.84 to resolve https://www.sprypt.com/denial-codes/what-are-denial-codes. High performers stay at or below 25 days in AR https://www.sprypt.com/denial-codes/what-are-denial-codes. The healthy benchmark sits in the 30–40 day range for days in AR https://www.sprypt.com/denial-codes/what-are-denial-codes. A typical Net Collection Rate is 90–95% https://www.sprypt.com/denial-codes/what-are-denial-codes.
