Cardiology Revenue Cycle Management: What CPT 2026 Broke in Your Systems

The first cardiology practice that called me about CPT 2026 had already rebuilt their charge-capture templates. They had not touched their claim edit rules, and that is where the money was going.

It is an easy mistake to make, and it is the one I now expect. The templates are visible. A coder opens one, sees a deleted code, and fixes it. The edit rules are not visible. They sit inside a billing system or a clearinghouse configuration, they were written years ago by someone who has left, and they keep running against code numbers that stopped existing in January.

Current Procedural Terminology (CPT), the American Medical Association’s code set for reporting medical procedures, changed cardiovascular reporting more in 2026 than in any recent year. I want to walk through what actually changed, then show you the part almost nobody is discussing: what a code-set replacement of this size does to the systems underneath your revenue cycle. If you want the general shape of the cycle before the cardiology specifics, our revenue cycle management services page covers it, and the 13 steps of the revenue cycle is the plainer version.

What Changed In Cardiology CPT Codes for 2026

CPT 2026 deleted six percutaneous coronary intervention (PCI) branch codes, deleted two coronary thrombolysis codes, added two Category I codes for complex PCI and chronic total occlusion, and eliminated the entire lower-extremity revascularization series, replacing it with 46 territory-based codes.

That is the summary. Here it is properly, sourced to the American College of Cardiology’s Coding Corner (December 4, 2025) and, for the two new intervention codes, to the Society for Cardiovascular Angiography and Interventions. Not to a billing vendor’s blog:

• Deleted Codes: 92921, 92925, 92929, 92934, 92938, 92944 · What it means: The additional-branch PCI codes. Each remaining main code was revised to read “branch(es)”
• Deleted Codes: 92975, 92977 · What it means: Coronary thrombolysis
• New Codes: 92930 · What it means: Complex PCI involving bifurcation or multiple lesions (SCAI)
• New Codes: 92945 · What it means: Chronic total occlusion (CTO) revascularization (SCAI)
• New Codes: 75577 · What it means: Coronary plaque assessment, replacing deleted Category III codes 0623T to 0626T (ACC)
• Eliminated Codes: 37220 to 37235 · What it means: The full lower-extremity revascularization series
• Replacing it Codes: 37254 to 37299 · What it means: 46 territory-based codes

The AMA’s CPT 2026 code set also added new definitions and an illustration clarifying coronary lesions and segments, and updated the definitions for major coronary arteries, branches and bypass grafts to match the new structure.

Read those last two entries again. A 16-code series was retired and 46 codes took its place, organized on a different principle. The old series was structured around vessel and intervention type. The new one is structured around anatomical territory. That is not a renumbering.

Why 46 Codes Replacing 16 Is A Systems Problem, Not A Coding Problem

Your coders will learn the new codes. That part resolves itself within a quarter, because coders are good at this and the specialty societies published guidance before January.

The systems will not learn them.

Four things in a typical cardiology practice are keyed to specific code numbers, and all four are now carrying dead values:

1. Charge-capture templates. The visible one. Someone opens it, sees 37225, replaces it. This is the fix everyone does first and it is the smallest of the four.
2. Claim edit rules. The rules that catch a problem before submission: this code cannot go out without that modifier, this pair triggers a bundling denial, this combination needs documentation attached. Written against old code numbers, they now either fire on nothing or fail to fire at all. A rule that fails silently is worse than no rule, because the practice still believes it is protected.
3. Payer policy mappings. Every payer’s coverage policy, prior-authorization trigger list and frequency limit is stored against codes. When the codes change, the mapping has to be rebuilt payer by payer, and payers publish their updated policies on their own schedules rather than the AMA’s.
4. Denial analytics. Your denial dashboard groups by code. Twelve months of history is keyed to codes that no longer exist, so trend lines break at the January boundary. You lose the ability to answer “is this getting better” at exactly the moment you most need to.

Territory-based coding adds a fifth problem that renumbering would not have. The new lower-extremity codes require the claim to reflect anatomical territory, which means the mapping from what the physician documented to what gets billed is a different logic problem than it was. That logic lives somewhere. Usually it lives in a person’s head, which does not scale and does not survive turnover. A claims processing configuration that encodes it explicitly is the difference between a rule you can audit and a rule you have to re-derive every time someone leaves.

If you cannot see which of your denials trace to the transition, that is its own finding. A revenue cycle analytics view that can slice denials by code, by payer and by date range is what turns this from a suspicion into a number.

Where Cardiology Claims Actually Die

Cardiology carries a higher denial rate than most outpatient specialties, and the reasons are structural rather than sloppy. Six mechanisms account for most of it.

1. Bundling. The National Correct Coding Initiative (NCCI), the CMS edit system that defines which procedure pairs cannot be billed together, publishes procedure-to-procedure edits that hit cardiology hard because cardiology encounters genuinely do stack multiple procedures. The edits are public and they update quarterly. Whether your system checks against the current quarter’s version is a different question.
2. Frequency limits. Medically Unlikely Edits (MUEs) cap the units of a service billable per patient per day. Echocardiography and device monitoring run into these regularly.
3. Professional and technical component mismatch. Diagnostic studies split into a professional component (the interpretation) and a technical component (the equipment and staff). Bill the wrong one, bill both when the facility already billed the technical side, or omit the modifier, and the claim denies. This is the single most cardiology-specific denial mode on the list.
4. Modifier errors. A consequence of the point above, plus the same-day-service modifiers that a stacked cardiology encounter requires.
5. Medical necessity and documentation gaps. Nuclear imaging and stress testing carry coverage criteria that the documentation has to satisfy explicitly. An interpretation that a clinician would find complete can still miss the element the payer’s policy names.
6. Prior authorization. Concentrated in advanced imaging and device services.

The useful thing here is the shape rather than the size. In the practices I have looked at, denials concentrate: a small number of payer-and-code combinations produce most of the volume. That matters because it changes the fix from “improve coding accuracy,” which is not actionable, to “find your top five combinations and build a rule for each,” which is. Denial management built that way pays for itself faster than a general accuracy push, because it targets work you can actually measure.

One Encounter, Four Sets of Billing Rules

Here is the thing that makes cardiology different, and it is easier to see in a single visit than in a list of denial causes.

A patient comes in. The cardiologist performs an evaluation and management (E/M) visit, interprets an electrocardiogram, supervises and interprets a stress test, and checks an implanted device. One patient, one afternoon, four billable services.

Now count the rule sets that just engaged.

The E/M visit has its own documentation and level-selection logic. The electrocardiogram interpretation is a professional component, billable only if the interpretation is documented and only if nobody else has already billed it. The stress test splits into professional and technical components, and which of the two you bill depends on where the test was performed and who owns the equipment. The device check has its own timing rules, its own frequency limits, and documentation requirements that differ by device type.

Four services, four modifier decisions, four denial modes, one claim.

None of these is hard in isolation. A competent coder handles any one of them without thinking. What makes cardiology expensive is that they arrive together, dozens of times a day, and a single wrong modifier on the third service denies work that was clinically and administratively correct in every other respect.

This is also why generic revenue cycle tooling underperforms here. Most of it models a claim as one service with one set of rules. A cardiology encounter is four claims’ worth of rule evaluation wearing one claim’s clothing, and a system that does not model the components separately cannot check them separately.

What Your Systems Have To Know That A Generic Tool Does Not

Everything above describes knowledge. The question that decides whether a practice keeps losing money is where that knowledge lives.

In most practices it lives in people. A billing lead knows that this payer wants the modifier in a particular position, that this device check denies if it lands inside the global period, that the territory mapping for the new lower-extremity codes works one way for one payer and another way for the next. That knowledge is real and it is genuinely expert. It is also unversioned, unauditable and one resignation away from gone.

Moving it into systems means four specific things, and none of them is exotic:

Edit rules as maintained data, not code. The rules change quarterly. If updating them requires a developer, they will not get updated quarterly.

Payer policy variance modeled explicitly. Not one rule set with exceptions bolted on, but the recognition that payers genuinely differ and the differences are the product.

Frequency and timing checked before submission, against the patient’s own history, not after the denial arrives.

Write-back into the record the staff actually work in. A validation that lives in a separate dashboard gets checked when someone remembers. A validation that updates the claim’s status where the biller already works gets checked every time.

That last one is the piece that most often decides whether an automation is used or quietly abandoned, and it is also the hardest part of the engineering. Connecting to a major electronic health record system is well-trodden. The difficulty sits in the last mile: the payer-specific rules that do not generalize, the cases that fall outside the standard integration path, and getting the write-back to land in the right place in the right workflow. Anyone who tells you that part is routine has not done it.

Most Cardiology Denials Start Before Anyone Writes A Code

This is the finding that surprises practices most, and it is consistent enough that I now check it first.

Working with a multi-location specialty group, we traced their denial volume back through the workflow expecting to land on coding. We did not. The largest single contributor was coordination of benefits errors and wrong managed care organization selection, both captured at eligibility verification, at the front desk, before the patient was seen. The coding was fine. The claim was doomed at check-in.

Cardiology is exposed to this for a structural reason: the patient population skews older, which means Medicare Advantage plans, secondary coverage and coordination-of-benefits complexity are the norm rather than the exception. Every one of those is a place where the wrong plan gets selected in a dropdown and nobody finds out for 30 days.

The same is true one step further along. Prior authorization on advanced imaging and device services is where a clinically appropriate procedure becomes an unpaid one, and the burden is administrative rather than clinical. The federal Interoperability and Prior Authorization rule, CMS-0057-F, requires affected payers to run a Prior Authorization application programming interface by January 1, 2027, which will change the mechanics of this substantially. Our prior authorization services page covers what that integration work involves.

The practical implication is uncomfortable if you have just spent a quarter on coding accuracy: your denial reduction ceiling is set at the front desk, not in the coding queue.

Outsource, Buy, or Build the Layer

I will give you the honest version, because the pages that rank for this question are almost all owned by companies that sell one of the three answers.

1. Outsourcing is the right call for a lot of cardiology practices. If you are under roughly 10 providers, if your denial volume is steady rather than growing, and if you do not have anyone whose job includes owning billing technology, a good outsourced partner will outperform anything you build. Outsourcing revenue cycle management covers where that line usually falls, and our list of revenue cycle management companies is a starting point for the shortlist. If the distinction between billing and the full revenue cycle is still fuzzy, medical billing versus revenue cycle management sets it out plainly.
2. Where outsourcing moves the problem instead of solving it is when the issue is upstream of billing. If your denials originate at eligibility, an outsourced billing partner inherits bad data and works denials faster. Faster is worth something. It is not the same as fewer.
3. Building is worth it when three things are true at once: your denial volume is concentrated in patterns specific enough to encode, you have the volume to make the maintenance worth funding, and the fix has to live inside a workflow you control. Cardiology hits all three more often than most specialties, which is why this comes up here more than elsewhere.

The number people underestimate is not the build. It is the maintenance. Edit rules change quarterly, payer policies change continuously, and CPT 2026 just demonstrated that the code set itself can be restructured. Whoever owns this, in-house or outsourced, is signing up for ongoing work rather than a project. Budget for that or the thing you build will be accurate for one year and quietly wrong for the next three.

For a sense of how differently this plays out in another specialty with its own rule structure, our behavioral health revenue cycle guide walks the same argument through session-based billing and carve-out payers.

What CMS-1850-P Stacks On Top Of This

The code changes are not the only thing moving. The CY2027 outpatient proposed rule, CMS-1850-P, published July 7, 2026, proposes to remove 637 services from the Inpatient Only list as the second phase of a phase-out that eliminates the list entirely on January 1, 2029. It also proposes extending site-neutral payment to off-campus imaging without contrast.

For cardiology specifically, both matter. Procedures migrating out of the inpatient setting land in hospital outpatient departments and ambulatory surgical centers, and the billing profile differs by setting: different payer rules, different authorization requirements, different claim edits. A procedure your practice bills correctly today can deny next year for no reason other than where it was performed. Our explainer on what CMS-1850-P costs your revenue cycle goes through the rest of the rule, and organizations in rural or critical access settings have a different exposure profile covered in critical access hospital reimbursement.

The comment period closes August 31, 2026.