CARC and RARC Code Interpretation for Infusion Remittances

CARC and RARC codes describe why a payment changed and what to do about it, and on paper, infusion remittances use the exact same code sets as every other specialty. The article develops how to read CARC/RARC combinations not as abstract code definitions, but as actionable signals tied to the specific failure modes that appear on infusion ERAs.

How CARC and RARC codes signal claim issues

Every line on an ERA carries three layers, and each one answers a different question. The group code answers who's financially responsible. The CARC answers why the payment changed. The RARC answers what specifically happened, or what needs to happen next.

Start with the group code, because it determines what's even possible before anything else. CO means the provider absorbs the adjustment under contract: appeal it or correct it, but never bill the patient. PR means the patient owes the balance, which moves the claim into patient billing rather than the denial workflow. OA sits in between, neither strictly contractual nor patient-owed, and what happens next depends entirely on the specific reason code attached. PI marks a payer-initiated reduction, which is often worth contesting rather than accepting.

CARCs are numeric, and there are 358 of them in the current list SPRY — CARC and RARC Codes Guide Mira Health. They tell you the category of failure, full stop. RARCs are alphanumeric, usually starting with M, MA, or N, and there are 1,185 of them. RARCs split into two functional types: supplemental codes that clarify the exact rule behind an adjustment, and informational codes prefixed "Alert," which convey process information without touching a dollar figure.

CO-16 is the cleanest illustration of why both layers matter. The CARC alone says almost nothing actionable, it just flags a missing element somewhere in the claim. The RARC paired with it names the missing diagnosis code, NDC, or documentation attachment. Skip that RARC and you're guessing.

Reading order matters too. Group code first, so you know who absorbs the hit. CARC second, so you know the category. RARC third, so you know the fix. Skip a step and the response is often wrong, not just slow. That's exactly why the group code has to come first.

Both code sets get updated three times a year, roughly March 1, July 1, and November 1. As of September 17, 2026, the CARC list in effect was dated November 1, 2025, and the RARC list was dated July 1, 2026. Internal denial-management tables that don't track those refresh dates will silently misroute claims, because a code's meaning or a required pairing can shift between cycles without anyone touching the billing software.

Infusion remittances and their harder version of the same codes

The difference lies in what generated the code, and therefore what actually fixes it.

Start with CPT hierarchy Fierce Healthcare — RCM Leaders Cite Payer Behaviors as Major Risks. Hydration can't be billed as the primary service if a drug infusion happened in the same encounter. The chemotherapy code stays primary regardless of the order infusions were actually administered in. Anything under 15 minutes has to bill as a push, using 96374 or 96375, not as an infusion. Get any of that wrong and the claim throws CO-97 or CO-11 in volume because the claim's own internal structure was invalid before it ever left the building.

J-code billing adds another layer of precision most specialties never have to think about. Drug identity, unit count, NDC, and dose all have to line up exactly. A unit mismatch throws a CO-16 that looks, on the remittance, identical to a routine missing-modifier CO-16. It isn't. It needs a completely different fix, one rooted in drug administration records rather than coding conventions.

Then there's authorization. Nearly every specialty infusion drug needs prior authorization before the first treatment, and that PA has to specify drug, dose, frequency, and duration. When the regimen changes or the authorization lapses mid-cycle, the result is CO-15 or CO-197, and in most cases retro-authorization simply isn't an option. That's a very different problem than a missing modifier.

The dollar stakes are different too. Buy-and-bill means the practice has already dispensed the drug before the claim resolves, so a denied biologic can represent tens of thousands of dollars in cost that's already gone out the door. A CO-50 on an infusion claim isn't the same animal as a CO-50 on an imaging claim, even though it's the same code, because the documentation required, the appeal pathway, and the financial consequence are entirely different.

Site-neutral payment policy, extended to certain drug infusions in 2026, adds a wrinkle that used to be cosmetic. Place-of-service modifiers now directly determine reimbursement rather than just describing where care happened. A missing or wrong modifier now produces a CO-45 or CO-16 with consequences that go well past the standard contractual write-off.

A lot of this is deliberate rather than accidental error. Roughly 54% of denials industry-wide trace back to deliberate payer policy shifts rather than submission mistakes Flobotics — RCM Statistics Overview. The code on the remittance looks the same as it did last quarter. The rule that produced it didn't.

The CO-50 + N115 combination and medical necessity denials on high-cost biologics

CO-50 means the payer doesn't consider the service medically necessary based on what was submitted. Straightforward enough on its own. That's not a documentation gap in the ordinary sense, it's a policy-driven denial, and it needs to be treated that way.

The fix starts with pulling the exact LCD or medical policy version in effect on the date of service, not the general coverage category it falls under. From there, every clinical criterion named in that policy has to map to something specific in the chart: step therapy completion, diagnosis specificity, lab values, documented failure of a prior treatment. For biologics specifically, the gap is usually step therapy and formulary exhaustion documentation. The drug choice was clinically sound. The file just never proved the prior drugs failed first.

Appeals built around this combination work better than most people expect, but only under a condition. Medical necessity denials carry roughly a 50% overturn rate on appeal, and that number holds when the appeal names each payer criterion by section rather than arguing in generalities DataRovers — Denial Management Complete Guide 2026. Vague appeals underperform. Specific ones, citing the policy's own language back at it, do considerably better.

For high-cost biologics, peer-to-peer review is frequently the stronger move over a written appeal, since the clinical reasoning behind a complex regimen rarely fits cleanly onto a page. Without N115, the picture changes: a CO-50 with no LCD reference is a commercial "not medically necessary" denial, and there's no LCD to anchor an appeal to, so the argument has to cite the payer's own published medical policy instead. Before any appeal goes out, the infusion team needs the treating physician's clinical notes, an ICD-10 code with real specificity, lab results that support the indication, prior treatment history, and any relevant published clinical guidelines backing the biologic that was chosen.

CO-197 and CO-15: authorization denials across the infusion treatment cycle

CO-197 and CO-15 land in the same spot on a remittance and get treated as interchangeable by a lot of billing teams. They shouldn't be. CO-197 means there's no prior authorization on file. CO-15 means an authorization exists but it's invalid or doesn't match what was billed. Same symptom, different disease.

Infusion authorization is a moving target by design. It has to be renewed at intervals the payer sets, often every three to six months, and a renewal that slips mid-cycle produces CO-197 on sessions that already happened. Any regimen change, a dose adjustment, a frequency shift, a switch in drug, requires a brand new PA, and billing under the old one throws CO-15. Even a change in the treating physician can trigger the need for reauthorization under most payer policies.

The retro-authorization gap is where a lot of revenue quietly disappears. Some payers allow retroactive authorization requests, but most don't for specialty infusion drugs, which means CO-197 on a drug that's already been infused can be permanently unrecoverable.

The scale of this problem is worth sitting with. Medicare Advantage insurers processed close to 53 million prior authorization requests in 2024, up from 49.8 million the year before, and 4.1 million of those were denied in whole or in part Fierce Healthcare — Rethinking RCM: Era Payer Complexity and AI. A 2025 physician survey found doctors completing an average of 40 PA requests a week, with three-quarters reporting that denial rates have climbed over the past five years Fierce Healthcare — Rethinking RCM: Era Payer Complexity and AI. This isn't a shrinking administrative burden, it's a growing one.

Timelines matter operationally too. Electronic PA submissions can clear in days; manual ones often take one to three weeks, and appeals or peer-to-peer reviews stretch that further. The resolution paths split from there: for a supplemental code, clarify the specific rule behind the adjustment; for an informational code, prefixed "Alert:", note that it isn't tied to a dollar adjustment. For CO-15, check whether the auth number, drug code, and approved units actually match what was billed. If it's a transcription error, correct and resubmit. If the auth was valid and the payer misapplied it, appeal. For CO-197, retro-auth may or may not be on the table. If it isn't, the path runs through escalation to the physician and the payer, with documentation of clinical urgency if the care was delivered emergently.

By the time CO-197 appears on a remittance, the money is usually already at risk. The actual failure happened weeks earlier, at scheduling or at the renewal date, not at the moment someone hit submit on the claim.

CO-97 and J-code bundling: when infusion hierarchy errors produce the wrong CARC

CO-97 tells you the payer already paid for this service inside another code. Simple statement, but it hides two genuinely different causes in infusion billing, and confusing them leads to the wrong fix.

One cause is a straightforward NCCI bundling edit, where CMS rules bundle two services together. The other cause is a hierarchy violation baked into the claim before it ever left the building: hydration billed as primary when a drug infusion happened in the same encounter. That one doesn't get fixed with a modifier. It needs a corrected claim with the code hierarchy rebuilt properly.

CO-236 appears in related territory, flagging a procedure or procedure/modifier combination that's incompatible with something else billed the same day under NCCI rules. It can stem from an unsupported modifier, a missing one, or codes that are mutually exclusive by definition, and the fix requires pinning down the specific edit pair involved.

Slapping a modifier onto a claim to make a CO-97 go away is only legitimate when the underlying services were actually distinct and documented as such. Doing it without that support isn't a denial fix, it's an audit exposure that becomes visible later.

Drug waste coding deserves its own mention. JW and JZ modifiers are under increasing payer scrutiny, and documentation requirements around them have tightened. A missing or incorrect waste modifier throws a CO-16 that reads like a routine information error but is actually a drug-unit reconciliation problem hiding under a generic code.

When a CO-97 shows up, read the RARC before doing anything else. If it points to a specific NCCI edit, the coder needs to confirm the modifier genuinely applies. If there's no RARC, or it points to claim-level bundling instead, check the code hierarchy first before assuming unbundling is the answer. As a practical rule, CO-97 on any drug infusion line means pulling the original claim to verify hierarchy before doing anything else, because the correct path splits completely at that first check.

CO-16 in infusion remittances: reading the RARC to find what the CARC hides

CO-16 is one of the highest-frequency information-deficiency denials across specialties, but in infusion, what's actually missing rarely matches the generic assumption.

A few RARC pairings occur constantly and each one means something distinct. M119, missing or invalid NDC, means the National Drug Code was absent, malformed, or didn't match the J-code billed, and the fix is reconciling the NDC against the drug administration record, a drug-specific task rather than a coding one. M51, missing or invalid procedure code, means the J-code or CPT was wrong or absent, and it needs to be checked against the current HCPCS year, since J-codes update annually Fierce Healthcare — RCM Leaders Cite Payer Behaviors as Major Risks.

NDC requirements on drug lines are becoming standard rather than optional, and a CO-16 paired with M119 is a failure mode that's genuinely specific to infusion, one that rarely occurs in general billing contexts at all.

Payers are required to attach a RARC to every CO-16.

There's a bigger pattern worth watching for here. CO-16 volume in infusion billing is a leading indicator of front-end problems, not back-end ones. If the same RARC keeps appearing across multiple claims, the actual fix lives upstream, in documentation or charge capture, not in another round of denial rework. N290 or a similar authorization-related RARC indicates the auth number was missing from the claim, which differs from CO-197 in that the auth exists but wasn't populated on the claim, so a corrected resubmission resolves it. Resolution routing proceeds by RARC. A missing NDC calls for a corrected claim with reconciled drug data. An invalid J-code calls for verifying the current HCPCS year and drug match, followed by a corrected claim.

Catching underpayments hidden in CO-45 remittances with infusion line-level review

CO-45 means the charge exceeds the fee schedule or contracted rate, and most practices treat it as an automatic write-off. Most of the time, that's the right call. Infusion claims are the exception, and they warrant a closer look line by line rather than batch processing.

Drug reimbursement runs on formulas that vary by payer and contract: ASP plus a percentage, WAC-based rates, carve-out pricing arrangements. The allowed amount on any given claim has to be checked against the correct formula for that specific drug and that specific payer, because a payer applying the wrong fee schedule year, the wrong drug tier, or the wrong site-of-care rate produces a CO-45 that looks perfectly contractual while actually being a payer-side error.

Site-neutral policy changes taking effect in 2026 make this worse in a specific way: the same drug, administered at the same physical location, can now reimburse differently depending on which place-of-service modifier gets attached. A CO-45 landing at an unexpected dollar amount may not be a routine contractual adjustment at all, it may be a misapplied site-neutral rate that nobody would catch without checking the formula directly.

CO-45 is the highest-frequency code across every specialty, which is exactly why it gets waved through without a second look. The discipline infusion billing demands is separating a legitimate contractual write-off from a payer calculation error dressed up in the same three digits, and that separation only happens by checking the drug, the formula, and the modifier against each other, line by line, rather than trusting the code to mean what it usually means. Infusion CO-45 lines warrant individual review rather than batch write-off.