The ops review was meant to be easy.
The billing vendor for your residential program reported a 6 percent denial rate for Q2. During the same period, your internal outpatient team reported 18%. Then, based on data from the EHR dashboard, your board packet displayed 11%. One organization, three "denial rates," three distinct narratives.
The CFO kept the rate constant in the budget for the following year. The revenue cycle director added two FTEs after noticing the increased figure. The board pressed to cut them when it noticed the lower vendor number. Since none of the three had the same denominator, numerator, or temporal basis, they were all "right."
This article's topics include:
Tiny worked examples of the primary methods used to compute the denial rate.
How the number is affected by claim-lifecycle (cohort) versus activity-based, claim versus line, count versus dollar, and initial versus total definitions.
Which definition should you use for staffing and daily operations, and which for your main KPI?
How to make your figures finally reconcile by standardising the denial rate across systems, vendors, and reports.
Why do denial rate reports conflict so often?
Fundamentally, the denial rate has to be straightforward: denials divided by a certain amount. In reality, practically every aspect of that statement is debatable:
What qualifies as a "denial."
whether you quantify lines or claims.
Whether you're counting dollars or events.
which date is the metric's anchor.
First-pass denials or any denials during the claim's duration are important to you.
Internal teams, billing providers, and various EHRs silently make different decisions. You only learn when:
Your EHR's "denial rate" appears to be half that of your vendor.
While your report utilised dollar-value, a benchmark you took from an HFMA webinar used claim-counts.
In this year's board deck, a "Q2 denial rate" was based on claim filing dates, whereas last year's was based on payment dates.
This is exacerbated for SUD and behavioural health by:
lengthy per diem claims with several lines.
Use various denial codes to separate payers.
frequent assessments of medical need and authorisation at the same time.
Denial rates will continue to lead to poor staffing, poor payer negotiations, and poor board discussions until you establish a single definition and direct all systems toward it. The glossary entry for denial rate on your intranet or in your BI tool exists because of a shared definition, which allows everyone to view the same information.
What are the main ways to calculate denial rate?
Denial rate definitions differ along five major axes:
Activity-based (remittance or incident) versus claim-lifecycle (cohort).
dollars-based versus count-based.
claim-level vs. line-level.
First-pass versus overall.
Date selection (submission, adjudication, and posting).
After focusing on the first four, we will discuss dates in the standardisation part.
What is claim-lifecycle (cohort) denial rate?
Definition: Track a group of claims filed over time. Ask yourself, "What share of this cohort was ever denied at least once at any point in its life?"
Tiny example:
100 claims were sent in July.
Over time, fifteen are denied eligibility, five are denied due to medical necessity, and three of those are eventually reversed on appeal.
July's cohort rejection rate is 20/100, or 20%.
Main points:
Even if a claim is rejected more than once, it can only be tallied once.
Until you are certain that remits have come and appeals have been settled, the figure is not final.
For root cause, you can divide by denial type, payer, program, or region.
What makes operators enjoy it:
It provides you with the actual "outcome" of your submission processes.
For end-to-end analysis, it works well with the clean claim rate and first-pass resolution rate.
Why it's difficult
It's slow. Particularly with Medicaid and appeals, denials on July submissions may take 60 to 120 days to complete.
Accurately building across several systems is more difficult.
What is activity-based (remittance / event) denial rate?
Definition: Regardless of when the claim was filed, examine every denial event that occurs during a time frame. Compare these with the lines or claims handled during that time.
A small example
Your team posts 400 claims totalling remits for all payers and service dates in July.
Fifty denial incidents are included in those remits (some are repetitions of the same assertion).
July's activity-based denial event rate is 50/400, or 12.5%.
This is what the majority of clearinghouse and EHR "denial dashboards" default to. Instead of looking at the original submission dates, they examine 835s that were uploaded or adjudication dates within a period.
What makes operators enjoy it:
Almost real-time. This week, a payer system disruption is visible.
This month's workqueue is experiencing denials, which indicates a staffing pressure.
Why it is distorted
It counts twice. A claim that is rejected, amended, and then rejected once more may be counted twice.
Cohorts are mixed. Submissions from April, May, and June may be the source of July's activity.
More difficult to compare using cohort-based benchmarks.
What is count-based vs dollar-based denial rate?
Count-based: counts of claims or lines make up the denominator and numerator. Dollar-based: Dollar quantities (charges or anticipated reimbursement) make up the denominator and numerator.
A small example
Rejected in July
Ten denials of claims. $10,000 is the total amount of refused charges.
The total amount of charges you filed for the time was $200,000.
Claim denial rate depending on counts:
10% is the ratio of 10 rejected claims to 100 submitted claims (sample numbers).
Denial rate in dollars:
10,000 dollars rejected out of a total of 200,000 dollars submitted is 5%.
Why the distinction matters
While paying the premium per-diem without any problems, a payer may reject several small group therapy codes. This results in a considerably lower dollar-based denial rate and a high count-based denial rate.
For medical necessity, a different payer may consistently reject expensive residential days. As a result, while counts remain stable, the dollar-based denial rate rises.
In reality, boards and CFOs are concerned with dollar-based. Front-line RCM teams perceive their workload to be count-based.
What is line-level vs claim-level denial rate?
Claim-level: The entire claim is deemed refused if any line on it is rejected. Line-level: the percentage of all service lines that were rejected is calculated by counting each line individually.
Small example (one assertion):
Five lines in one claim:
Four paid per-diem days.
Due to "no auth on file," one per diem day was rejected.
At claim level:
For metric purposes, this claim is "denied."
This small sample's claim-level count-based denial rate is 1.
100% is the ratio of denied claim to total claim.
Line-wise:
Twenty percent is one refused line out of five.
This distinction is crucial to behavioural health:
IOP and PHP claims sometimes combine several days or services.
Because authorisation spans for residential per-diem frequently vary, some days experience auth problems while others do not.
Claim-level is acceptable if all that matters is whether any part of the claim has to be reworked. Line-level data is more precise when it comes to payer behaviour or use.
What is initial (first-pass) vs total denial rate?
Initial (or first-pass) denial rate: was this claim or line initially denied by the payer?
Total denial rate: Was there ever a refusal of this claim or line, even after adjustments or appeals?
A small example
100 claims were filed in June.
Initial answers:
15 are rejected.
85 pay or transition to a paid/zero-pay position without being rejected.
Of the 15 rejected:
Ten are fixed, and the second submission is paid.
Five are rejected once more before being written off.
Initial rate of denial:
Fifteen percent (15 initial denials out of 100 submissions).
Total denial rate (depending on cohort):
Out of 100 submissions, 15 claims received at least one denial, or 15%.
If "total" were defined as "ever denied, including repeats," you could count 20 denials for each of those 15 statements. Double counting appears at that point.
First-pass denial rate and first-pass resolution rate are closely related. It provides information about front-end quality, including eligibility, demographics, authorisations, and codes. The total denial rate, which correlates with FTEs and burnout, indicates how much work affected your denial work queues.
How do different setups change the “right” denial rate definition?
Program type, payer mix, and billing methodology all play a major role in your optimum definition.
Clinics for outpatients with brief episodes
Features that
One or a few CPT codes for each visit.
brief incidents, frequently just one DOS per allegation.
Low average charge per claim due to high volume.
Repercussions
Because most claims contain one or two lines, claim-level and line-level are frequently comparable.
The count-based denial rate keeps a careful eye on how frequently your team handles a claim.
What's effective:
Use the initial, dollar-based, claim-level denial rate as your primary KPI. It is simple to convey to payers and leadership.
activity-based count of staffing denial occurrences, as denials are received and handled promptly.
Take caution:
Denials based on eligibility may increase by employer group or state program. For true insight, break down the denial rate by payer or product.
Carve-out plans for behavioural health may generate peculiar rejection codes that are difficult for your EHR to map. These must be normalised by your denial-code library.
Residential, PHP, and IOP with long per-diem episodes
Characteristics:
Claims are sometimes rolled across lengthy episodes, with multiple days per claim.
excessive dependence on concurrent and initial authorisation.
Some days are paid, while others are denied for "no auth," "not medically necessary," or "level of care not covered." These partial denials are frequent.
Repercussions
The story is drastically altered by line-level versus claim-level. A claim can be flagged for a single refused day.
Because rejected days at high ASAM levels can occasionally ruin profitability, dollar-based metrics are more important.
What's effective:
Initial, claim-level, dollar-based cohort denial rate by payer and care level should be your primary KPI. It clearly illustrates financial danger.
Line-level denial analysis by denial cause, including utilisation and medical necessity, is important for clinical and UR teams.
Because each refused day frequently results in interdisciplinary labour across UR, clinical, and billing, staffing should employ an activity-based count of denied lines.
Take caution:
An auth problem for days 8 through 10 appears to be the same as a complete denial for all days if you only report the claim-level denial rate. You lose important nuance when discussing auth policies.
The way that state Medicaid MCOs encode authorisation and medical-necessity denials varies greatly. That has to be handled state-by-state in your mapping.
Single-state vs multi-state behavioral health
Single-state:
Normalising denial codes and categories is made easier.
The behaviour of the Medicaid program is consistent.
Multi-state:
Medicaid and MCO editing regulations vary greatly.
Different combinations of CARC and RARC may be used for the same real-world denial in each state.
Significance:
For the metric, you want a single master definition of "denial," but your denial-code library should include adaptable payer-specific mappings.
You can see when a state's program modifies its regulations in the middle of the year by using activity-based indicators.
Conclusion:
Ensure that the KPI definition is consistent across states. The technique, date basis, numerator, and denominator are the same.
Dashboards should be divided by payer and state to analyse variations.
In your metric specification, note exceptions that are unique to a state.
In-house billing vs outsourced billing vendors
Internal groups:
If you take the time to standardise, you will have greater control over definitions.
Risk: Each of IT, finance, and billing may develop a denial metric of their own.
Outsourcing suppliers:
Most bring their own definition of denial rate, which is frequently activity-based and claim-count orientated.
Your finance system or EHR may not work with their gateway.
How to do it:
Make your standard definition public and mandate that vendors map to or comply with it.
When requesting contracts and RFPs, ask vendors for:
precisely which denominator and numerator they employ.
They may count funds, count lines, or make claims.
Ignoring this will put you back in the "6 percent vs. 18 percent" predicament, with no clear means to reconcile and no ability to see how well your vendors are performing.
Which denial rate should you manage to?
Five denial rates are not necessary for each board packet. One main "managed number," along with a few operational partners, are required.
For the majority of SUD and behavioural health organisations, I advise:
Primary KPI: Initial, claim-level, dollar-based, cohort denial rate by submission month.
Numerator: Total denied permitted dollars for claims filed throughout the period that receive a rejection as their initial adjudication answer.
Denominator: The total anticipated amount of money permitted on all claims filed at that time.
Cut by: Program, payer, care level, and occasionally geography.
Why
Dollar-based is closely linked to income.
Front-end quality (eligibility, auth, coding) is reflected in initial denials, which are genuinely fixable.
Budgets and quantities are scheduled on a cohort-by-submission-month basis.
- Operational companion: Activity-based denial event rate (count-based, claim-level) by posting month.
Denial events posted during the period are the numerator.
Denominator: The total number of claims during the time with remits posted.
Why
This aids with staff denial resolution and informs you of the number of things that enter the work queue.
It is almost instantaneous.
- Supplemental: Line-level denial analysis for residential, PHP, and IOP.
Counts and dollars by refusal reason, per-day, or per-service may be displayed instead of a single "rate."
drives discussions about authorisation and documentation in clinical and UR settings.
In every report, you should be clear:
"This is the initial cohort denial rate by submission month, dollar-based, and claim-level."This is based on activity, event count, denial rate at the claim level, and remit-posting month.
The fact that the two numbers are different becomes normal and not stressful when they are properly labelled.
How do you standardize denial rate across your organization?
Three components make up standardisation: one owner, one temporal basis, and one definition.
How do you nail the numerator and denominator?
Put this in a plain-language specification that your RCM and finance departments can use.
As an example:
Metric designation: "Initial denial rate (dollars, claim-level, cohort)."
Dividend:
any claim made throughout the time frame.
Don't include assertions that are only informational or free.
If you have that reasoning in place, use the estimated authorised amount (not full charges) as the dollar base.
Calculator:
subset of denominator in which at least one rejection code (CARC with denial status on the 835) that lowers payment is present in the initial adjudication response.
As long as at least one line is refused, include claims with partial payments.
Don't include "soft edits" that don't involve staff work, such as informational messages that don't affect payments.
Next, describe "denial" in terms of codes:
Consult the CARC and RARC codes associated with your centralised denial-code library that you believe to be true denials.
Choose whether to incorporate:
Rejections of medical necessity and use.
Denials of eligibility and benefits coordination.
Services not covered.
Denials of repeated claims.
Be specific. "We include CARC codes in categories X, Y, Z. We exclude duplicates and informational edits."
How do you fix the date basis?
The date fields that power your metric must be selected:
Use the 837 or your billing system's submission date for cohort metrics.
Use the adjudication date if you publish in real time or the remit-post date in your PM/EHR for activity metrics.
Reasons to care:
A "Q2 denial rate" may refer to distinct claim populations if one report truncates according to the submission date and another according to the payment date.
Processing delays for state Medicaid programs may be lengthy. Your time series won't be distorted by those delays thanks to a submission-based cohort.
Put this in your specification: "For this metric, claims are attributed to the month they were first submitted to the payer."
How do you document shared definitions?
Make a basic internal "metrics catalogue" with the following:
A brief description and the metric name.
Details about the denominator and numerator.
Count versus dollar-based; claim against line-level.
First versus overall.
time-based.
sources of data (data warehouse, clearinghouse, Supabill, billing platform, EHR).
Connect this to your lexicon:
Denial rate link.
Since the first-pass resolution rate and clean claim rate are located in the same area of the funnel, cross-reference them.
Instead of using their preferred BI tool, anyone gathering data for the board or a payer should begin there.
Who owns denial rate and what cadence works?
Identify a clear owner, typically the VP of Revenue Cycle or the RCM director.
Establish a reconciliation rhythm:
Owner evaluates the denial rate every month across:
PM or EHR system.
Warehouse/BI.
ports for vendors.
applies the standard definition to compare them.
records and addresses the reasons for variance:
date filters.
inclusion or exclusion of particular programs or payers.
Disparities between claims and lines.
Quarterly
Give the board and executive team a one-page "denial rate definition and sources" PowerPoint.
Verify the definitions of any external benchmarks that are utilised, or clearly identify any discrepancies.
How do you force vendors to align or map?
Although you might not be able to compel every external party to alter their internal reports, you have two choices at your disposal:
Preferred: The vendor creates a report that precisely fits your requirements.
Acceptable backup: You calculate the measure in your own environment using the raw data that the vendor provides (claim-level with denial flags, dollars, dates, and CARC/RARC codes).
Add this to scorecards and contracts:
"The vendor shall either supply comprehensive claim-level extracts that enable the client to calculate it or monthly denial reports that complies with the client's metric demand."
Demand a monthly side-by-side table that maps the vendor's measure to yours if they insist on quoting "their" denial rate.
How can AI help keep denial rate definitions consistent?
An agent may reconcile claim-by-claim, tag mismatches, and generate a single source-of-truth denial rate if your EHR, clearinghouse, and data warehouse don't agree.
Whether billing is done internally or externally, Supabill uses a single, uniform metric definition for all payers and programs, then presents it to clients in the same manner. In order to ensure that your internal and vendor numbers match, Supabill's agents monitor remits, tag rejections with your standardised reasons from the denial-code library, and compute both cohort and activity-based denial metrics.
One truthful limit:
Bad or missing source data cannot be fixed by AI. An AI agent still requires human judgement to create the appropriate business rules or to initiate manual review if authorisations are not recorded, if employees select arbitrary refusal reason codes, or if payers submit incomplete 835s. Agents do the counting, while humans maintain the definition and governance of metrics.
Top comments (0)