A 9% Denial Rate Is the Least Useful Number in the Report That Published It

On July 31, 2026, AMS Solutions — a physician-founded revenue cycle firm — published its State of Medical Billing 2026 benchmark report, covered the same day by HIT Consultant. The number that will travel is the headline one. The industry-wide claim denial rate has risen to 9 percent, up from 7.5 percent in 2023, while days in Accounts Receivable have stretched to an average of 42, up from 38. Two clean figures, easy to quote, and both of them close to useless for the practice owner reading them.

They are close to useless because no practice bills at 9 percent. The same report breaks that figure apart by specialty, and what comes out is not a tidy distribution around an average. It is four separate problems that happen to produce similar-looking percentages. Neurology leads the major specialties at a 14 percent initial denial rate. Home health sits at 13 percent with 52 days in Accounts Receivable. Physical therapy runs 11 percent at 45 days. Family practice, by contrast, does not have a denial problem at all. It has a revenue-never-billed problem.

What makes the report worth an hour of your time is that it does not stop at the percentages. For each specialty it names the mechanism, and in every case the mechanism is far smaller and more specific than the word "documentation" implies. Not a culture problem. Not a training problem. A modifier. A filing window. An unsigned page. A fee schedule nobody updated.

That distinction is the entire argument. A denial rate is not a performance score. It is a symptom, and in this report every symptom has a CPT code and a two-character modifier attached to it. The practice that treats its denial rate as a number to be lowered is chasing an average. The practice that treats it as a diagnosis knows which four claims to fix on Monday morning.

Every Specialty in That Report Fails in a Different, Nameable Way

Start with neurology, the worst performer in the set. The single largest denial driver is not clinical judgment or thin notes. It is professional-technical split errors — the -26, -TC and global modifiers on EMG, nerve conduction studies and EEG. One study, two separately billable halves: a technical component for running it and a professional component for reading it. Bill the global code when the equipment belongs to somebody else, or bill only the technical when you performed both, and the claim comes back. AMS Solutions puts the annual cost of that one mechanic at $30,000 to $80,000 for a mid-sized practice.

Sitting next to it is a quieter loss that will never appear in a denial report at all. CPT 64615, chemodenervation for chronic migraine, carries a work RVU update, and the report puts the national average professional component near $173 under the Medicare Physician Fee Schedule. Practices still billing against an outdated commercial fee schedule are leaving 10 to 18 percent of every one of those claims uncollected. That claim is paid. It is simply paid at the wrong number, and a Clean Claim Rate of 100 percent will never surface it.

The other three specialties break on the same pattern with different parts:

• A Calendar, Not a Chart: In home health, the report attributes reimbursement variances of $200 to $800 per 30-day payment period under the Patient-Driven Groupings Model to OASIS-E scoring inaccuracies, missing face-to-face encounter detail, and Notice of Admission filings that missed the five-day window. Two of those three are administrative deadlines rather than clinical failures, which means the care was delivered correctly and the money moved anyway.
• A Threshold Nobody Is Watching: Physical therapy's 11 percent traces to 8-minute rule unit miscalculations, missing KX modifiers once a patient crosses the 2026 Medicare threshold of $2,330, and plans of care that were never signed — the CO-228 rejection. A practice can deliver a full course of medically necessary therapy and lose the back half of it to a modifier that is triggered by a running dollar total, not by a diagnosis.
• Revenue That Was Never Billed: Family practice is the inverse case. National Chronic Care Management enrollment sits under 15 percent of eligible Medicare panels, while practices reaching 40 percent capture generate more than $385,000 in annualized incremental revenue at $30 to $50 per patient per month. Nothing was denied here. The service was never enrolled, never coded and never submitted, so it never entered the revenue cycle to be measured in the first place.

Four specialties, four mechanics, one average sitting on top of them. Anyone whose response to the 9 percent figure is that the practice should work its denials harder has already missed what the report found.

The VOSKPO Approach: Specialty Depth Where a Denial Rate Cannot Reach

Here is the uncomfortable part for a practice that has already outsourced its medical billing services. Every failure above is knowable in advance, and every one of them needs somebody who has billed that specialty before. A biller covering eight specialties across a hundred accounts is not going to know that EEG splits differently from EMG, or that the KX modifier attaches to a running dollar threshold rather than to a condition. That knowledge is not general. It lives down in the codes, and depth of coverage is what decides whether anyone in your revenue cycle actually holds it.

This is what we mean at VOSKPO when we say we do not simply process claims — we engineer revenue. At VOSKPO LLC the edit logic is built per specialty, and the people who work your account work that specialty rather than whatever landed in the queue that morning.

• Specialty Rule Sets, Not a Generic Scrubber: We build the pre-submission edit set around the mechanics that actually deny in your specialty — split-billing logic for -26, -TC and global on neurodiagnostic studies, KX tracking against the current Medicare therapy threshold, Notice of Admission calendars that count in days rather than in good intentions. A generic scrubber catches a field left blank. A specialty rule set catches a modifier that is present, valid and wrong.
• Fee Schedule Reconciliation, Not Only Denial Management: We reconcile paid amounts against your Payer Contracts and the current year's published RVU values, so a claim paid at last year's rate is caught as an underpayment instead of filed as a success. Denial Management by definition only ever sees the claims that were refused. The money lost on the claims that were accepted needs a separate check, and most practices do not run one.
• Automation Aimed at Charge Capture, Not at the Appeal: Our AI medical billing tooling runs at the front end, where a correction costs minutes, rather than at the appeal, where it costs staff hours and a share of the work never gets done at all. Revenue cycle automation pointed at the back end only makes a practice faster at recovering money it should not have lost, which is a smaller prize than it sounds.

Build the Denial Map Your Denial Rate Is Hiding

None of this requires you to accept our reading of the report. It requires you to look at your own distribution instead of your own average, and the data to do that is already sitting in your practice management system.

• Sort Your Denials by Code and Modifier, Not by Payer: Pull the last twelve months of denials and group them by CPT code and modifier combination rather than by insurer. Most practices find that a small handful of code-and-modifier pairs account for the majority of the volume, which turns an abstract percentage into a short, fixable list with names on it.
• Reconcile Paid Claims Against the Current Fee Schedule: Take ten of your highest-volume paid procedures and compare what was actually deposited against the current published rates and your contracted terms. Payer reimbursement at an outdated rate looks identical to correct payment on every report you already run, so it will not be found unless somebody deliberately goes looking for it.
• Name the Service You Are Eligible to Bill and Do Not: Every specialty in the report had one. For family practice it was Chronic Care Management, sitting under 15 percent nationally against a best-in-class 40 percent. Identify yours, check the ICD-10 coding and eligibility requirements that gate it, and decide whether the gap is a clinical choice or simply nobody's job.

An industry average tells you the weather. It does not tell you whether your own roof leaks, and 2026 is not a year to be managing healthcare RCM by weather report. If you would like a second set of eyes on which code-and-modifier pairs are quietly costing your practice the most, VOSKPO works these distributions for a living. Visit voskpo.com to request your Free Revenue Review.