A denial rate is a fraction, and practices compare fractions built from different inputs. One team reports claims denied on first submission. Another reports every denial event, so a claim worked twice counts twice. A third reports hospital data it found in a vendor article. Before you can say whether a rate is normal, you have to say what sits in the numerator and the denominator.
What this covers
- HFMA defines the initial denial rate as total initial denial claims divided by total claims submitted, with the denominator averaged over the three months before the reporting month. The same framework publishes a dollars version.
- MGMA’s practice data puts claims denied on first submission at 8% for single-specialty groups in its 2023 DataDive Practice Operations set. MGMA notes the rate sat at 8% in 2019 as well.
- Hospital denial rates run higher. Optum’s 2024 denials index reports a 12% average denial rate for 2023, up from 10% in 2020 and 9% in 2016. That figure describes hospital remits, not a physician practice.
- Claim volume is half the fraction. A smaller denominator raises the rate with nothing changing in how claims are handled.
- Payer mix moves the rate too. HHS OIG found Medicaid managed care denied 12.5% of prior authorization requests in 2019, against 5.7% in Medicare Advantage.
A denial rate is a fraction, and both halves of the fraction move. Change the payer mix or the claim volume and the rate moves with nothing changing in the work.why a rate needs its definition and its population attached
Most benchmark arguments are really definition arguments. Two billing managers can both report a denial rate near 10% and disagree about whether that is good, because one counts first-pass denials at the claim level and the other counts every denial event the payer returns. The number is not wrong in either case. It is answering a different question. This article sets out what the published figures actually count, gives the practice-level ranges that exist, and separates the two forces that move a rate without any change in performance: claim volume and payer mix.
What a denial rate actually measures
A denial rate is a fraction, and the work is deciding what goes in each half. HFMA’s Claim Integrity Task Force defines the initial denial rate as total initial denial claims divided by total claims submitted, with the denominator taken as the average of the three months before the reporting month. The same task force defines a dollars version: gross charges on initially denied claims divided by total submitted gross charges. The count catches frequency. The dollar figure catches exposure, and the two will not agree.
HFMA’s MAP Keys add a second measure with a different denominator. MAP Key AR-5, the remittance denial rate, divides total claims denied by total claims remitted, counted at the claim level from 835 remittances or paper remits. Only actionable denials count, meaning denials the practice can correct and rework. Patient-responsibility adjustments and duplicate-claim denials are excluded.
- Initial denial rate by volume: claims denied on first submission divided by claims submitted. A first-pass measure.
- Initial denial rate by dollars: gross charges on initially denied claims divided by submitted gross charges. A measure of financial exposure.
- Remittance denial rate, MAP Key AR-5: all actionable claims denied divided by claims remitted. Includes denials that show up after a resubmission.
- Denial write-offs, MAP Key AR-6: net dollars written off as denials divided by net patient service revenue. The final loss.
A rejection fails an edit at the clearinghouse or the payer’s front end and never reaches adjudication, so it produces no claim number and no remittance advice. MAP Key AR-5 counts only claims that produced a remit, which leaves rejections out of the rate entirely. Track them in a separate report, with a separate workflow, because a rejection points at a formatting, eligibility or coding error caught before the payer ever saw the claim.
One counting rule holds across the HFMA definitions: take only the first denial on a given claim. If a claim is denied, corrected and denied again, it stays one denied claim. A system that logs every denial event will show a rate climbing for a reason that has nothing to do with payment performance.
The published figures, and what each one counts
There is no single denial rate benchmark. There are several figures in circulation, each computed on a different population, and the population matters more than the number. Read what each one counts before you compare anything to it.
| Source | Figure | What it counts | Population |
|---|---|---|---|
| MGMA DataDive Practice Operations, 2023 | 8% | Claims denied on first submission | Single-specialty physician practices |
| MGMA Practice Operations report, 2022 | 7% to 8.14% | Claims denied on first submission across 2019 and 2021 | Primary care, nonsurgical and surgical practices |
| Optum 2024 Revenue Cycle Denials Index | 12% | Average denial rate on 2023 claims | Hospital and health system remits |
| KFF analysis of HealthCare.gov data, 2023 | 20% in network, 36% out of network | Claims denied, as reported by the insurer | ACA marketplace plans |
| HHS Office of Inspector General, 2023 | 12.5% vs 5.7% | Prior authorization requests denied in 2019 | Medicaid managed care vs Medicare Advantage |
| KFF prior authorization metrics, 2025 | 12% to 18% | Standard prior authorization requests denied | Medicare Advantage, Medicaid managed care, ACA marketplace |
| CMS Comprehensive Error Rate Testing, FY 2025 | 6.55% | Medicare FFS improper payment rate | Not a denial rate. A retrospective sample of paid claims. |
The practice-level figure quoted most often is MGMA’s 8%, the single-specialty aggregate for claims denied on first submission from the 2023 DataDive Practice Operations data set. MGMA notes the rate sat at the same 8% in 2019, so the figure has been stable even as denial volume grew. MGMA’s earlier report, built on 2021 data, put primary care at 8.00%, nonsurgical specialties at 8.14% and surgical specialties at 8.14%, after all three fell in 2020.
The hospital figure is a different population. Optum’s 2024 denials index reports an average denial rate of 12% for 2023, up from 10% in 2020 and 9% in 2016, based on roughly 124 million hospital claim remits. A hospital mix carries higher-dollar services, more inpatient stays and more payer scrutiny, so the rate runs above a physician practice’s. Quoting 12% at a five-provider practice compares that practice to a population it does not belong to.
A practice that benchmarks against a hospital denial rate can read its own 9% or 10% as healthy when its actual peer group sits near 8%. The gap then goes unexamined, and the dollars behind the gap stay in the write-off column.
Why the payer-side figures are a different measure
The insurer-reported figures describe what payers deny, not what a provider submits. KFF’s analysis of federal transparency data found that HealthCare.gov insurers denied 20% of in-network claims in 2023 and 36% of out-of-network claims. Across the 175 reporting insurers, the in-network rate ranged from 1% to 54%. No provider can use a range that wide as a target, because it measures insurer behavior rather than practice performance.
Prior authorization data is another population again. The HHS Office of Inspector General found that Medicaid managed care organizations denied 12.5% of prior authorization requests in 2019, more than twice the 5.7% rate in Medicare Advantage that year. KFF’s first read of the new federal transparency filings, covering 2025, found insurers denied 12% of standard prior authorization requests in Medicare Advantage, 14% in Medicaid managed care and 18% in the ACA marketplace. Those are authorization decisions, not adjudicated claims, so they do not sit in the same fraction as a claim denial rate.
Medicare’s improper payment figures belong in a fourth column. CMS reported a Medicare fee-for-service improper payment rate of 6.55% for fiscal year 2025 through its Comprehensive Error Rate Testing program, with Part B providers at 8.44%. That program samples paid claims, reviews the record behind each one, and estimates what share of dollars should not have been paid. A paid claim in that sample was never denied. The rate and a provider denial rate measure different stages of the same dollar.
Whatever the source, a denial rate is comparable only to another rate computed the same way on the same population. State the numerator, the denominator and the period next to every number you report. A figure without that context is a headline, not a benchmark.
Claim volume is half of the number
The denominator in the HFMA initial denial rate is an average of the three months before the reporting month, so it falls when claims fall. A practice that loses a provider, slows scheduling for a month or closes for a holiday week submits fewer claims. A fixed number of denials then divides into a smaller base, and the rate climbs while the billing team did nothing differently.
The reverse happens when volume grows. A new provider, a second location or a seasonal surge raises the denominator. If the denials added by that growth arrive at the usual rate, the reported rate stays flat or drops, even though the absolute denial count rose. A falling rate and a rising denial count are both possible in the same month, and neither one contradicts the other.
Two measurement choices compound the effect. Count at the claim level, not the denial-event level, or a claim denied three times counts three times. And count at the encounter level only when the practice defines it that way, because one encounter can produce more than one claim.
Use a rolling three-month denominator to match the HFMA definition, so a single slow month cannot distort the rate.
Group by claim or encounter number so a reworked claim stays one denied claim in the numerator.
The dollar rate shows exposure that a claim count hides, since a few high-charge denials can move dollars without moving volume.
A rejection never reached adjudication, so it belongs in its own report with its own cleanup workflow.
Payer mix moves the rate with nothing changing
Denial rates differ by payer, and a practice’s payer mix differs from its neighbor’s. A behavioral health practice with a large Medicaid managed care share is exposed to prior authorization scrutiny that a commercial-heavy practice largely avoids. The same clinical work, submitted the same way, produces a higher rate.
The payer data is clear on the spread. In KFF’s analysis of the 2025 filings, standard prior authorization denial rates ranged from 5% to 17% across Medicare Advantage insurers and from 2% to 23% across Medicaid managed care insurers. The averages hide that spread, and the spread is the part a practice feels.
| Payer segment | Average denial rate, 2025 | Range across insurers |
|---|---|---|
| Medicare Advantage | 12% | 5% to 17% |
| Medicaid managed care | 14% | 2% to 23% |
| ACA marketplace | 18% | 3% to 25% |
Two practices can therefore report the same 8% and be in different operational shape. One earns it on a commercial book. The other earns it on a Medicaid-heavy book where 8% means the team is outperforming the population it works in. The number alone does not tell you which practice you are looking at. That is the first thing medical billing services data reviews check, and it is why a single blended rate belongs in a footnote rather than at the top of the report.
We found no published denial rate benchmark scoped to behavioral health or ABA services. The figures in this article count all services, and the payer transparency data does not break out an ABA authorization for 97153 or any other behavioral health code. Until a behavioral health-specific benchmark exists, compare a behavioral health practice to its own trailing rate split by payer, not to a national number built on a different case mix.
Building a rate you can compare
The useful move is to publish a definition and hold it. Write down the numerator, the denominator, the period and the exclusions, then compute the same fraction every month for at least a year. The trend tells you more than any single figure, because you are comparing your practice to your practice.
- The numerator and the denominator written out in plain language.
- The period set as a rolling three months, to match the HFMA definition.
- Denials counted at the claim level, with only the first denial on a claim counted.
- Rejections excluded and reported in a separate report.
- Patient-responsibility adjustments and duplicate claims excluded.
- A dollar version reported beside the volume version.
- The rate split by payer and by service line, so a mix change shows up as a mix change.
That is the discipline denial management runs on. A rate no one defined cannot be improved, because no one can say what change would move it. A rate that shifts when the payer mix shifts is a report on the book of business, not on the billing team.
Denial rate benchmark questions
The closest peer figure is MGMA’s practice data, which puts claims denied on first submission at 8% for single-specialty groups in its 2023 DataDive Practice Operations set. That is a first-pass rate, not a final one. No published benchmark is banded by practice size, so treat the figure as a reference point and compare against your own trailing rate.
The initial denial rate counts claims denied on first submission. The final denial rate counts only the claims still denied after rework and appeals are exhausted. HFMA’s claim integrity framework measures the initial rate by volume and by dollars. The final rate is closer to the money actually lost, and it is always the smaller of the two.
Yes. The HFMA denominator averages the three months before the reporting month. When claims submitted fall, a fixed number of denials divides into a smaller base and the rate rises with no change in handling. A growing practice can see the opposite: more denials in absolute terms and a flat or falling rate.
Denial rates differ by payer, so a change in mix moves a practice rate without any process change. HHS OIG found Medicaid managed care denied 12.5% of prior authorization requests in 2019, against 5.7% in Medicare Advantage. A practice that adds Medicaid or Medicare Advantage volume carries more authorization scrutiny than a commercial-heavy one.
Hospital claims carry higher-dollar services, more inpatient stays and a wider service mix, all of which draw more payer review. Optum’s 2024 denials index reports a 12% average denial rate across more than 1,400 hospitals. Physician practice data from the same period sits near 8%. The two numbers describe different populations.
No. A rejection fails an edit before the claim reaches adjudication, so it produces no claim number and no remittance advice. HFMA’s remittance denial rate counts only claims that produced a remit. Report rejections separately, because they point at different causes and get fixed by a different workflow.
We found no published denial rate benchmark scoped to behavioral health or ABA services. The figures in this article count all services, and the payer transparency data does not break out ABA lines. Until a behavioral health benchmark exists, compare a behavioral health practice to its own trailing rate split by payer.
The bottom line
A denial rate is only useful when its definition travels with it. Write down the numerator, the denominator and the period, then compute the same fraction every month and split it by payer. The trend will tell you more than any single national figure, because you will be comparing your practice to your practice.
Which number are you actually benchmarking against?
Send us one month of remittance data. The MedFactor team will compute your initial denial rate by volume and by dollars, break it out by payer, and show you where the rate is being moved by mix rather than by handling.
Request a free denial benchmark reviewThis article describes general billing and benchmarking practice rather than legal or clinical advice, and published benchmark figures change with each data release, so confirm any figure against the source and the period it covers.
