Pull your clearinghouse denial report for the last 90 days and look at the age column, not the dollar column. In most short-staffed cardiology groups the dollars look manageable at a glance; it's the aging that tells the real story. A stack of denials sitting at 40, 55, 70 days old is not a billing inconvenience. It is money that is quietly crossing the line from recoverable to written off, and the reason it's sitting there is almost never the payer. It's that the one person who could rework those claims spent the day answering the phone. For a three-provider group, the claim denial rate cost per provider small practice managers actually pay is not the denial itself — it's the denials nobody had time to touch before the clock ran out.
This is written for the practice manager who signs off on the deposit and knows the collections number is soft but can't quite point at where the softness lives. It lives in the denial queue, and it got there through a staffing decision that looked reasonable on paper.
Reading the denial report your PM system won't flag
Your practice management system is excellent at telling you what you submitted and what got paid. It is close to silent on what got denied and then abandoned. Charges submitted, payments posted, AR aging by patient — those reports run themselves. But there is no standard report titled "denials that will hit timely-filing before anyone works them," and that is precisely the number that determines your real collection rate.
Cardiology makes this worse than most specialties because of coding density. A single visit can carry an E/M code, a stress test, an echo with interpretation, a Holter reading, and a nuclear study, each with its own modifier rules and medical-necessity requirements. Payers deny cardiology claims for a long and specific list of reasons: missing prior auth on advanced imaging, LCD medical-necessity mismatches, modifier 26 versus TC professional-technical splits, bundling edits that collapse a stress echo into the office visit. Every one of those is a winnable denial if a knowledgeable person works it. Every one is a permanent loss if that person is on hold with a patient about a parking question.
The denials don't announce themselves. They land in a worklist inside the clearinghouse, and worklists have no lobby, no ringing phone, no patient tapping the glass. They wait patiently, which is exactly the problem. In a well-staffed billing office they get worked in daily batches. In a group where billing shares a body with the front desk, they wait until the wait becomes fatal.
The denial-rework clock nobody is watching
Denials run on two clocks at once, and a short-staffed office loses both. The first is the appeal-effort clock: a denial is dramatically easier to overturn in the first two weeks, while the visit is fresh, the documentation is close at hand, and the payer's own systems still have the context loaded. Push it out to 45 or 60 days and the same appeal takes twice the effort for the same dollars.
The second clock is the hard wall: timely filing. Most commercial payers give you 90 to 180 days from date of service to get a clean claim in, and many give a separate, shorter window — sometimes just 60 to 90 days from the denial — to file an appeal or a corrected claim. Blow through it and the payer owes you nothing, regardless of whether the original denial was their error. There is no appeal to "we filed late." A denial that was 100% winnable on day three becomes an uncollectable write-off on day 91, and nothing about the medicine changed.
flowchart TD
A[Cardiology claim submitted] --> B{Payer adjudicates}
B -->|Paid| C[Payment posted]
B -->|Denied| D[Denial lands in work queue]
D --> E{Biller has open time today}
E -->|Yes| F[Reworked inside 14 days<br/>65 to 70 percent overturned]
F --> C
E -->|No, on phones| G[Denial ages in backlog]
G --> H{Timely filing window}
H -->|Still open| E
H -->|Expired| I[Permanent write off<br/>zero recovery]Notice where the diagram forks. Every denial passes through the same gate: does the biller have open time today? In a group where that biller is also the single receptionist answering phones and doing check-in, the honest answer most days is no. The claim doesn't get denied to death by the payer; it dies in your own queue while the clock runs.
Putting real dollars on the per-provider leak
Let's make this concrete for a three-cardiologist group, because "denials are a problem" moves nobody and a number on a page moves budgets.
Assume the group collects around $2.4M a year — roughly $800K per provider, which is unremarkable for cardiology carrying imaging and testing. Industry denial rates for the specialty commonly land in the 10-15% range on first submission; call it 12%. That puts roughly $288K in denied charges in play every year across the group, or about $96K per provider.
Now the part that separates a healthy billing operation from a leaking one: the rework rate. A staffed billing office works nearly all of its denials and overturns 65-70% of them, so the permanent loss is small. A short-staffed office simply never touches a chunk of the queue. If 30% of denials go unworked past their deadline — a conservative figure when one person is splitting time three ways — that's about $86K a year across the group, near $29K per provider, gone. Not disputed and lost. Never worked at all.
Here's the uncomfortable framing. That $86K is not a soft loss like an underpayment you might negotiate. It's the highest-margin money in the practice, because the visit already happened, the cardiologist already spent the time, the overhead was already absorbed. You are not paying to earn it a second time — you're just failing to collect what you already produced. There is no marketing spend, no new patient, no additional chair time that returns margin like a reworked denial does. And there is no line on any report that says you lost it.
One body, three jobs, and the queue that always loses
The root cause is a staffing structure that looks like thrift and functions like a leak. In a lean cardiology group the billing role and the front-desk role are frequently the same person, or a two-person desk where whoever is free grabs the phone. It feels efficient. It is not, because the three jobs compete on wildly different time horizons and the phone always wins.
A ringing phone is an interrupt with a five-second SLA. A patient at the window is physically present and cannot be deferred. A denial in a work queue has a 45-day soft deadline and a 90-day hard one. Human attention resolves that competition the only way it can: it services the loudest, nearest demand and defers the silent, distant one. So the phone gets answered, the window gets cleared, and the denial queue grows by every claim that bounced today. Do that for two weeks and the oldest denials cross out of the easy-rework window. Do it for a quarter and some cross timely filing entirely.
flowchart LR
P[Phones ringing<br/>5 second demand] --> B[One billing person]
W[Patients at window<br/>present demand] --> B
D[Denial queue<br/>45 day deadline] --> B
B --> R{Attention goes to<br/>loudest nearest task}
R -->|Phones and window win| Q[Denials keep aging]
R -->|Denials rarely reached| Q
Q --> L[Write offs at timely filing]The fix is not to work the biller harder or to lecture about prioritization. It's to remove one of the three competing demands entirely, and the phone is the one that both interrupts the most and least requires a human. This is where CallSphere Health's AI front desk changes the staffing math directly: it answers 100% of calls 24/7, books appointments into the live schedule, handles reminders and waitlist refills, and does it all without ever tapping your biller on the shoulder. The features that matter here aren't glamorous — they're the ones that give a skilled biller back the two to three hours a day the phone was eating.
Point those recovered hours at a denial queue where roughly two-thirds of items are winnable, and the arithmetic reverses. The same person who was letting $86K age into write-offs now has uninterrupted blocks to actually work the queue inside the appeal window. You didn't hire a second FTE; you stopped spending your billing FTE on a task a machine handles better. Against that recovered $80K-plus a year, the pricing on AI phone coverage stops being a cost line and starts being the cheapest denial-recovery tool in the building.
Turning recovered hours into recovered claims
Freeing the time is step one; the payoff comes from what you do with it. A few concrete moves turn recovered billing hours into recovered dollars:
- Work denials youngest-to-oldest by deadline, not by dollar. The instinct is to chase the biggest claims first. The better rule is to protect claims nearest their timely-filing wall, because a $180 denial you can still file beats a $900 denial that expired yesterday.
- Batch by denial reason, not by patient. Ten prior-auth denials on nuclear studies work faster together than ten unrelated claims, because the fix and the payer contact are the same. Cardiology's coding density makes this batching especially productive.
- Set a standing daily denial block. Even 90 uninterrupted minutes a day, now possible because the phone isn't pulling the biller out of it, clears more of the queue than a scattered full day of interrupted attempts.
- Watch the aging column weekly. The metric that predicts your write-offs is not the denial rate; it's the share of the denial queue older than 30 days. If that number is falling, your recovered hours are working.
None of this requires new billing talent. It requires the billing talent you already pay for to spend its day on billing. A three-provider cardiology group doesn't have a denial problem so much as an attention problem wearing a denial costume — the claims are winnable, the person is capable, and the only thing standing between them is a phone that never stops ringing. Take the phone off the person, and the queue that always lost finally gets worked.