Category: Billing Insights

Your biller just handed you last week’s commercial remit batch. Half the claims paid at the contracted rate. A quarter paid short with a CO-45. The rest came back with a CO-197, a PR-204, and a couple of N657s in the remark column — and nobody on the team can tell you in plain English what happened or which ones are worth appealing. That is the daily reality for most behavioral health billing teams, and misreading these documents is one of the fastest ways to leave money on the table.

An EOB (Explanation of Benefits) or ERA (the 835 electronic remittance file) is the payer’s line-by-line accounting of what they did with each claim. Every adjustment is tagged with a Claim Adjustment Reason Code (CARC) and often a Remittance Advice Remark Code (RARC). Learn to read them and you stop guessing at denials — you know within 30 seconds whether the claim is appealable, a contractual write-off, or a patient balance transfer.

What you’ll take away from this

  • CARC codes tell you why money was adjusted; RARC codes give you the supporting detail. You need both to work a denial correctly.
  • Group codes (CO, PR, OA, PI) tell you who owes the balance — provider, patient, or nobody. Misreading the group code is how balances get billed to patients illegally.
  • Roughly half the denials behavioral health facilities see are procedural (auth, eligibility, timely filing) and reversible on appeal — if you catch them inside the payer’s window.
  • Ninety days of remits, sorted by CARC and payer, will show you exactly where your front-end process is broken.

What are CARC and RARC codes on a behavioral health EOB?

CARC stands for Claim Adjustment Reason Code. It’s a standardized code set every payer in the country uses on the 835 remittance file and the paper EOB. Each CARC explains one specific reason a claim line was adjusted — paid short, denied, reduced, or reassigned.

RARC stands for Remittance Advice Remark Code. RARCs supplement CARCs with additional context. A CARC of CO-16 (“Claim/service lacks information”) is nearly useless on its own — the RARC that follows it, like N657 (“This should be billed with the appropriate code for these services”) or MA130 (“Your claim contains incomplete and/or invalid information”), is what tells your biller what to actually fix.

Every CARC comes with a two-letter group code that assigns financial responsibility for the adjusted amount:

  • CO — Contractual Obligation. Provider write-off. You cannot bill the patient. Example: CO-45 (charge exceeds fee schedule).
  • PR — Patient Responsibility. Transfer to patient. Example: PR-1 (deductible), PR-2 (coinsurance), PR-3 (copay).
  • OA — Other Adjustment. Neither party owes it. Usually informational.
  • PI — Payer Initiated Reduction. Payer’s own policy adjustment, not part of your contract. Often appealable.

If your team is billing balances to patients based on CO-coded adjustments, you have a compliance problem, not just a revenue problem. That’s the first pattern to audit.

How do you read a behavioral health remittance advice line by line?

Every ERA has the same anatomy. Working top to bottom on a single claim, you’ll see:

1. Claim header

Patient name, member ID, claim number (the payer’s internal ICN), dates of service, billed amount, allowed amount, and paid amount. If the paid and allowed are both $0, the whole claim was denied. If the allowed is populated but the paid is short, you’re looking at a partial payment with a patient-responsibility transfer or a contractual adjustment.

2. Service line detail

Each CPT or HCPCS code billed appears on its own line — H0015 for intensive outpatient, H0018 for residential, 90837 for a 60-minute therapy session, 90853 for group therapy, and so on. Each line shows billed, allowed, adjustment amount, and paid. This is where CARC and RARC codes attach.

3. Adjustment segment

For every dollar of the difference between billed and paid, there must be a CARC (and often a RARC) explaining where it went. If billed was $1,200, allowed was $800, and paid was $640, you should see a CO-45 for $400 (contractual) and a PR-2 for $160 (coinsurance). Those numbers balance to the penny or the remit is wrong.

4. Provider-level adjustments (PLB)

At the bottom of the remit you’ll find provider-level balance adjustments — takebacks, forward balances from prior overpayments, interest, and offsets. Behavioral health facilities miss these constantly. A payer can recoup a $4,000 overpayment from a claim two months ago by silently reducing this week’s deposit, and it only shows up in the PLB. If your posted payments don’t tie to your bank deposit, the PLB is usually where the reconciliation broke.

5. Check/EFT trace number

The BPR and TRN segments tell you which deposit this remit belongs to. If you’re posting manually, this is how you match a remit to a bank deposit.

Which CARC and RARC codes matter most in behavioral health billing?

Behavioral health sees a narrower band of denial codes than general medical because most of the payment friction concentrates around authorization, level-of-care documentation, and medical necessity. These are the codes worth training every biller to recognize on sight:

Authorization and medical necessity

  • CO-197 — Precertification/authorization absent. Either UR never got auth, or the auth on file doesn’t match the dates or level of care billed. Fixable if you can produce the auth number or file a retro-auth request within the payer’s window.
  • CO-50 — Non-covered services because not deemed medically necessary. The classic UR fight — payer says the patient didn’t need residential, they needed PHP. Requires a clinical appeal with the medical record, not just a corrected claim.
  • CO-198 — Precertification exceeded. You billed 14 days of residential; auth covered 10. Appeal with continued-stay documentation.

Eligibility and coverage

  • CO-27 — Expenses incurred after coverage terminated. Member lost coverage mid-stay. Bill secondary if there is one, then patient.
  • CO-26 — Expenses incurred prior to coverage. Coverage started after admission. Check for a prior plan.
  • PR-204 — Service not covered under the patient’s current plan. Benefit exclusion. Usually not appealable unless you have a single-case agreement.

Coding and claim setup

  • CO-16 with RARC N657 or MA130 — Missing/invalid information. Fix and resubmit as a corrected claim, not an appeal.
  • CO-97 — Benefit included in another service already paid. Common when billing an E/M code alongside a therapy code on the same day. Modifier 25 often resolves it.
  • CO-4 — Procedure code inconsistent with the modifier used. HJ, HF, and level-of-care modifiers are frequent culprits in SUD billing.

Timely filing and duplicates

  • CO-29 — Timely filing limit expired. Occasionally reversible with proof of original submission (clearinghouse acceptance report). Otherwise a hard write-off.
  • CO-18 — Duplicate claim. Not always what it looks like. If two dates of service are truly distinct, appeal with documentation.

The pattern to internalize: CO-197, CO-50, and CO-198 are your clinical fights and belong in the utilization review workflow, not general billing. Sending a UR denial to a biller who only handles corrected claims is how facilities lose appealable dollars.

How should behavioral health facilities work denials from the EOB?

Reading the codes is step one. Turning that reading into recovered revenue requires a triage workflow. When a remit posts, denials should be sorted within 24 hours into three buckets:

Bucket 1 — Corrected claim. CO-16, CO-4, CO-97, most CO-18s. Data errors. Fix the claim, resubmit through your clearinghouse. Turnaround is typically 14–30 days. No formal appeal needed.

Bucket 2 — Clinical appeal. CO-50, CO-197, CO-198. These need the medical record, the UR notes, ASAM criteria documentation, and a written appeal letter from a clinician or UR specialist. Payer appeal windows range from 60 to 180 days from the remit date — miss the window and the denial is final.

Bucket 3 — Write-off or patient bill. PR codes go to patient statements. Legitimate CO-45 contractuals get adjusted off. CO-29 timely filing with no proof of prior submission — write off and figure out why it aged.

The mistake we see most often when auditing a new client’s AR: everything gets dumped into one “denials” worklist and worked in the order it arrived. A CO-197 that’s 45 days from its appeal deadline sits behind a CO-16 that just needs a taxonomy fix. Triage by code, then by deadline.

This is also why catching problems before submission matters more than working denials after the fact. A clean claim with verified benefits, confirmed authorization, and correct modifiers doesn’t generate a CO-197 or a CO-50 in the first place. Our team runs pre-submission scrubs against the CARC patterns we see repeat by payer — if a Medicaid MCO in your region is denying H0015 without a specific modifier combination, we know before your claim goes out, not after it comes back.

What does the EOB tell you about your revenue cycle overall?

A single remit is a transaction. Ninety days of remits is a diagnostic. Pull your denial data by CARC code and payer and you’ll see exactly where the leak is:

  • High CO-197 volume from one payer → auth process is broken for that payer specifically. Fix intake or the UR handoff.
  • High CO-27 across payers → verification of benefits is stale by the time you bill. Re-verify closer to service or at discharge.
  • High CO-16 with RARC around taxonomy or NPI → credentialing or claim setup issue. Check that every rendering provider is loaded correctly.
  • High CO-45 write-offs relative to billed → your contracted rates are underwater. Time to renegotiate.
  • Rising PLB takebacks → a payer is auditing prior payments. Get in front of it before the recoupment grows.

The remittance advice isn’t just proof of what got paid. It’s the most honest report card you’ll get on your front-end operations — VOB, authorization, credentialing, coding — because it shows you exactly which failures the payer caught. Read the codes, sort the patterns, and every remit becomes a roadmap for the next month’s process fixes.

If you want a second set of eyes on where your denials are actually coming from, our free 6-month billing audit pulls your remits by CARC and payer and quantifies the recoverable dollars before you commit to anything. Book a call here.

Frequently Asked Questions

What’s the difference between a CARC and a RARC code?

A CARC (Claim Adjustment Reason Code) explains why a specific dollar amount was adjusted on a claim — for example, CO-197 means authorization was missing. A RARC (Remittance Advice Remark Code) provides supplemental detail that clarifies or expands on the CARC. Many CARCs, especially CO-16, are too generic to act on without the accompanying RARC telling you what specifically was missing or invalid.

Can I bill a patient for a balance adjusted with a CO group code?

No. CO stands for Contractual Obligation, meaning the provider is contractually required to write off that amount under the payer agreement. Billing the patient for a CO-coded balance is a contract violation and, depending on the payer and state, can trigger compliance issues. Only PR (Patient Responsibility) group codes transfer to the patient statement.

How long do I have to appeal a behavioral health denial after receiving the EOB?

Appeal windows are set by the payer contract and vary widely — commercial payers typically allow 60 to 180 days from the remit date, Medicare Advantage plans generally 60 days, and Medicaid MCOs anywhere from 30 to 90 days depending on state. Always calculate the deadline from the remittance date, not the date of service, and build a tickler system so clinical appeals for CO-50 and CO-197 denials don’t age out.

Why does my bank deposit not match the total on my remittance advice?

Check the Provider Level Balance (PLB) segment at the end of the ERA. Payers use the PLB to recoup prior overpayments, apply interest, forward balances, and net out offsets. A silent takeback from a claim posted months ago will reduce your deposit without touching any current-week claim payment. Reconciling deposit to remit means reconciling both the claim payments and the PLB adjustments.

Which CARC codes are worth appealing versus writing off?

Procedural denials — CO-16, CO-4, CO-97, most CO-18s — are usually fixed with a corrected claim rather than a formal appeal. Clinical denials — CO-50 (medical necessity), CO-197 (no authorization), CO-198 (auth exceeded) — are worth full clinical appeals with medical records and UR documentation when the treatment was justified. Hard exclusions like PR-204 and expired timely filing (CO-29 without proof of prior submission) are typically write-offs unless you have specific documentation to challenge them.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You admit a self-pay client for residential detox. Thirty days later, the family calls demanding an itemized breakdown because the bill is 40% higher than what admissions verbally quoted. Now you’re in a dispute you can’t win — because you never issued a written good-faith estimate, and the No Surprises Act says the patient can challenge any bill that exceeds the estimate by more than $400.

Out-of-network behavioral health facilities carry more NSA exposure than most operators realize. The rules were written with emergency rooms and surprise anesthesia bills in mind, but they land squarely on residential SUD, PHP, and IOP programs that admit self-pay and out-of-network commercial patients. Getting the paperwork wrong doesn’t just risk a patient dispute — it puts your facility in front of a federal Independent Dispute Resolution process you didn’t budget for.

The short version

  • Good-faith estimates (GFEs) are required for every self-pay and uninsured patient at scheduling — and behavioral health admissions almost always count as “scheduled.”
  • Balance billing is prohibited for emergency services and certain non-emergency services at in-network facilities, but out-of-network residential and outpatient behavioral health has narrower protections — with big exceptions around consent.
  • The notice-and-consent waiver form lets patients waive balance-billing protections for some non-emergency services, but it must be delivered on specific timelines with specific disclosures.
  • Documentation is the whole game. If it’s not in the chart with a signature and a timestamp, it didn’t happen — and IDR arbitrators will side with the patient.

Does the No Surprises Act apply to my out-of-network SUD or mental health facility?

Yes. The NSA applies to any licensed healthcare facility and any licensed provider billing for services. That includes residential treatment centers, PHPs, IOPs, detox facilities, mental health hospitals, and the individual clinicians (LCSWs, LMFTs, MDs, psychiatric NPs) who bill under them. If you have an NPI and you send claims or invoices, you’re in scope.

The two obligations that hit behavioral health hardest are the good-faith estimate requirement for self-pay and uninsured patients, and the balance-billing prohibition for out-of-network providers delivering emergency or certain post-stabilization services.

Where operators get tripped up: many assume “we’re out-of-network, so NSA doesn’t apply to our commercial patients.” That’s backwards. Out-of-network is exactly when the balance-billing rules kick in. If a commercial patient’s plan pays partial and you try to bill them the difference without a valid notice-and-consent waiver, you’re in violation — and the patient can dispute through the federal IDR process. The plan’s out-of-network payment becomes the ceiling unless you’ve done the paperwork correctly.

The one place NSA gives you more room is pure self-pay. Self-pay and uninsured patients don’t have balance-billing protections in the traditional sense, but they do have the GFE right — and if your bill exceeds the estimate by $400 or more per provider, they can invoke Patient-Provider Dispute Resolution.

What has to be in a good-faith estimate for a residential or PHP admission?

The GFE must be delivered in writing (paper or electronic, patient’s choice), and it must include specific elements. Missing any of them makes the estimate defective:

  • Patient name and date of birth
  • Description of the primary item or service — for behavioral health, this typically reads as “residential substance use disorder treatment, estimated 30 days” or “partial hospitalization program, estimated 4 weeks”
  • Itemized list of items and services reasonably expected, grouped by provider or facility
  • Applicable diagnosis codes, service codes (CPT/HCPCS/Rev codes), and expected charges
  • Name, NPI, and TIN of each provider or facility expected to bill
  • State, service location, and dates of anticipated service
  • Disclaimer that the estimate is not a contract, that actual charges may differ, and that the patient has dispute rights if the bill exceeds the estimate by $400 or more per provider

Timing matters. If services are scheduled at least 10 business days in advance, the GFE must be delivered within 3 business days. If scheduled 3–9 business days in advance, it must be delivered within 1 business day. If the patient requests an estimate without scheduling, you have 3 business days to deliver.

For residential and PHP programs, the tricky part is estimating length of stay. Build tiered GFEs (14-day, 30-day, 45-day) tied to your typical medical necessity ranges, and reissue an updated GFE whenever the treatment plan changes. Any time expected charges change by more than the $400 threshold, a new GFE is required.

One operational note: co-treating providers — the psychiatrist doing med management, the primary therapist billing under their own NPI, the medical director — each need their own line items. This is where a lot of facilities fall down. Admissions is generating a facility estimate but forgetting the professional fees, and the patient’s total bill blows past the $400 threshold on the provider side alone.

When can I balance-bill an out-of-network commercial patient?

Under the NSA, balance billing is prohibited for:

  • Emergency services at any facility, in-network or out-of-network
  • Non-emergency services delivered by out-of-network providers at in-network facilities (the classic “surprise anesthesia” scenario)
  • Post-stabilization services following an emergency, until the patient can safely be transferred or gives informed consent

For out-of-network behavioral health facilities providing scheduled, non-emergency services — most residential admissions, PHP, IOP — balance billing is technically permitted if you obtain a valid notice-and-consent waiver using the federal Standard Notice and Consent form.

The waiver requirements are strict. The notice must be delivered at least 72 hours before the appointment for services scheduled more than 72 hours out, or on the day of service (before care begins) if scheduled with less notice. It must include a good-faith estimate of charges, a list of in-network alternatives, and clear language that the patient is not required to consent and can seek care elsewhere.

The waiver cannot be used for certain categories — including some ancillary services and emergency-related care. Several state parity and consumer-protection laws layer additional restrictions on top of the federal rules, particularly for mental health and SUD services. California, New York, Texas, and Florida each have wrinkles that override or supplement the federal standard.

If you’re admitting a patient in acute withdrawal or acute suicidal crisis, that’s an emergency service under most interpretations — and no waiver can override the balance-billing prohibition. Charging the patient the difference between billed charges and the plan’s out-of-network payment is a violation.

How do I build NSA compliance into admissions and billing workflows?

Compliance is a workflow problem, not a legal problem. The regulations are clear enough. What breaks down is the handoff between admissions, clinical, and billing — and that’s where facilities get caught.

A workable operational model:

At inquiry / pre-admission: Verification of benefits determines insurance status. If uninsured or self-pay, GFE generation starts immediately. If commercial and out-of-network, VOB confirms plan status and admissions determines whether notice-and-consent is appropriate or whether the service qualifies as emergency (in which case no waiver applies). A thorough verification of benefits process is the foundation — if you don’t know the patient’s network status and remaining benefits, you can’t build a defensible estimate or waiver.

Before admission: GFE or notice-and-consent delivered with required timing. Signature captured, timestamped, stored in the chart. Copy provided to patient in their preferred format.

During treatment: If length of stay, level of care, or provider mix changes, a revised GFE is issued. This is where utilization review and billing have to talk — every step-down, every extended stay authorization changes the estimate.

At discharge and billing: Final charges reconciled against the most recent GFE. If the delta is greater than $400 per provider, expect a potential dispute. Bills flagged for review before they go out. Documentation packaged and retained for the 6-year record requirement.

Fragmentation hurts here. If your VOB is done by one vendor, your billing by another, and admissions is generating GFEs from a template someone downloaded in 2022, the pieces don’t line up. Global AHS handles VOB, utilization review, billing, and appeals in-house — which matters because a compliant GFE has to reflect what UR is actually authorizing and what billing is actually going to submit. When those functions live under one roof, the estimate a patient signs at admission is the estimate their bill matches at discharge.

What happens if I get NSA compliance wrong?

Two enforcement paths matter. The first is Patient-Provider Dispute Resolution for self-pay patients — a federal process where a Selected Dispute Resolution Entity reviews the case and, if the bill exceeds the GFE by more than $400, typically caps the patient’s obligation at the estimated amount. You eat the difference. The second is federal IDR for commercial out-of-network disputes, where a certified arbitrator picks between the plan’s offer and your offer under a baseball-style arbitration model.

Beyond disputes, HHS can impose civil monetary penalties per violation. State enforcement varies but generally piggybacks on federal complaints. And commercial payers have used NSA violations as grounds for network termination and audit expansions — meaning a compliance failure with one patient can trigger a takeback review across your entire commercial book.

The cost of building a clean GFE and consent workflow is trivial compared to the cost of one IDR loss with an arbitrator who reads your documentation and sees gaps. Get the paperwork right up front, and the rest of the revenue cycle gets easier — cleaner claims, fewer patient balance disputes, faster cash.

If you want a second set of eyes on your current GFE templates, consent workflow, and VOB-to-billing handoff, Global AHS offers a free 6-month billing audit that includes an NSA compliance review. Request the audit here.

Frequently Asked Questions

Do we have to issue a good-faith estimate to insured patients who are staying in-network?

Not currently. The GFE requirement under the No Surprises Act applies to uninsured and self-pay patients. Regulations for insured patients (Advanced Explanations of Benefits) have been announced but enforcement has been repeatedly delayed. Best practice is to build the GFE workflow now so you’re ready when AEOB enforcement begins.

How does the $400 dispute threshold work for a 30-day residential stay?

The $400 threshold is applied per provider or facility, not per line item. If your facility’s final bill exceeds the good-faith estimate for facility charges by $400 or more, the patient can initiate Patient-Provider Dispute Resolution. Professional fees billed under separate NPIs (psychiatrist, primary therapist) each have their own $400 threshold against their own line items in the GFE.

Can we use the notice-and-consent waiver for a patient admitted in acute withdrawal?

No. Emergency services — including care for acute intoxication, withdrawal management, and acute psychiatric crisis under most interpretations — cannot be waived through the notice-and-consent process. Balance billing is prohibited for emergency services regardless of network status, and any consent form signed under emergency conditions would not hold up in a dispute.

What happens if length of stay changes and we don’t update the good-faith estimate?

If actual charges exceed the original GFE by more than $400 per provider, the patient can initiate dispute resolution and the arbitrator will typically cap the bill at the original estimate. Any material change — extended stay, level-of-care change, added services — should trigger a new GFE. Tie the update trigger to your utilization review workflow so it happens automatically when authorizations shift.

How long do we have to keep good-faith estimates and consent forms on file?

The federal record retention requirement is 6 years from the date the GFE or consent was issued. Store the signed documents in the patient’s chart or a dedicated compliance repository with clear timestamps. If a dispute is filed years later, the burden is on the facility to produce the documentation.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

A commercial payer sends a records request six months after discharge. They want notes for 42 dates of service across residential and PHP. Your clinical team pulls the chart, exports the PDFs, and someone in billing hits send. Three weeks later, a retro-denial letter asks for a refund on roughly a third of those days — not because care wasn’t provided, but because the notes don’t prove it was medically necessary at the level billed.

This is the audit outcome nobody prepares for, and it’s almost always preventable at the point of documentation. Progress notes aren’t a clinical formality. They’re the evidence file for every dollar you’ve already collected.

What auditors actually look for in behavioral health progress notes

  • UR reviewers and auditors read for medical necessity at the specific level of care billed — not just that the patient attended treatment.
  • Copy-forward notes, group attendance rosters passed off as individual notes, and vague symptom language (“patient did well today”) are the top clawback triggers.
  • Notes must tie back to the treatment plan, the assessed diagnosis, and current functional impairment — every session, every day.
  • The safest habit: write each note as if a stranger will read it in 18 months with a checklist and a refund request.

What do UR reviewers actually read for in a progress note?

Utilization review nurses and payer medical directors are not reading for narrative quality. They’re scanning for a small set of specific elements that justify the level of care on that specific date. When those elements aren’t there, the day gets denied — or worse, retro-denied after payment.

Here’s what a UR reviewer’s eye tracks to, in order:

1. Current symptoms with specificity. “Patient reports anxiety” is not documentation. “Patient reports panic episodes 3x this shift, HR elevated per nursing, unable to complete morning group, required 1:1 support for 40 minutes” is documentation. The reviewer needs to see that today’s symptoms still require the intensity of the level of care being billed.

2. Functional impairment. Can the patient perform ADLs? Are they sleeping? Eating? Regulating affect without staff intervention? Auditors look for evidence the patient cannot safely step down. If your PHP notes read like IOP notes, expect a downgrade.

3. Risk factors. SI/HI, cravings intensity, withdrawal symptoms, medical comorbidities, environmental risk on discharge. If risk isn’t documented, the reviewer assumes it isn’t there — and questions why the patient needs 24-hour care.

4. Response to interventions. Not “patient participated in CBT group.” Instead: “Patient engaged in cognitive restructuring targeting the relapse trigger identified in yesterday’s individual session; identified 2 of 3 distortions with prompting.” The reviewer wants active treatment, not attendance.

5. Continued need for this level of care. Every note should answer the question: why not a lower level of care today? If nothing distinguishes the note from an outpatient encounter, the auditor will make that distinction for you.

Which documentation patterns get claims clawed back?

Across the denial patterns we see in behavioral health billing, a handful of habits produce most retro-denials. If any of these describe your current chart flow, you have exposure.

Copy-forward notes. The single biggest audit trigger. When day 14 contains identical phrasing to day 3 — same symptom description, same intervention, same plan — the auditor concludes either the patient didn’t need 14 days or the clinician didn’t reassess. Both conclusions end the same way. Most EHR systems make copy-forward one click away; that convenience is a liability.

Group notes as individual notes. A group note that says “patient attended process group, participated appropriately” is not an individual progress note and won’t support an individual therapy CPT code. Auditors know the difference. They read for individualized content — what did this patient say, feel, work on, struggle with.

Missing treatment plan linkage. If the treatment plan lists three goals and the note doesn’t reference which one was worked on today, the auditor treats the session as unfocused. Notes should name the goal or objective addressed.

Time discrepancies. If the session is billed as 53 minutes (CPT 90837) but the note says “met with patient briefly,” that’s a clawback waiting to happen. Time-based codes need time-based documentation — start time, stop time, or at minimum total duration.

No update on prior symptoms. If yesterday’s note said the patient reported SI and today’s note doesn’t mention SI at all, the reviewer wonders whether it was reassessed. Continuity across notes is what proves ongoing clinical thinking.

Discharge criteria never referenced. Notes should periodically speak to progress toward discharge criteria — not just at week 4, but throughout. Otherwise the auditor asks how the clinical team decided when to step down, and whether that decision was defensible.

What does a defensible progress note look like at each level of care?

The elements are consistent; the intensity of documentation changes.

Detox and residential

Notes need to establish 24-hour need. That means vitals when clinically relevant, withdrawal scoring (CIWA, COWS) with actual numbers, safety checks, medication response, and functional status changes across the day. A residential note that could plausibly describe an IOP session is a residential note that will get downgraded.

PHP

The reviewer is looking for daytime structure that requires 20+ hours per week of active treatment, plus evidence the patient still needs that intensity. Symptom monitoring, medication management interactions, response to structured therapy, and the specific reasons the patient cannot yet function at IOP. “Continues to require PHP due to ongoing symptoms” is not sufficient — name the symptoms, name the functional deficit.

IOP

Individual and group content must be distinguished. Skills being taught must be named. Homework, between-session application, and family or environmental factors should show up. IOP notes that read like weekly check-ins get flagged fast.

Individual therapy sessions

CPT-specific documentation is non-negotiable. 90834 vs 90837 comes down to time, and time must be documented. Medical necessity for the frequency of individual therapy has to tie to the diagnosis and functional impairment — not “patient prefers individual work.”

How should treatment centers audit their own notes before the payer does?

The cheapest audit is the one you run on yourself. A basic internal chart review process catches roughly the same issues a payer would, and gives you time to build documentation coaching into clinical operations.

A workable internal audit rhythm looks like this:

  1. Weekly random sample. Pull 5–10 charts per week across levels of care. Read them the way a UR reviewer would — with a checklist for the elements above.
  2. Flag copy-forward patterns. Compare notes across consecutive days for each patient. Identical or near-identical language is a red flag, full stop.
  3. Cross-check CPT to note. Does the note support the code? Time-based codes especially — 90837, 90847, H0015 — need documentation that matches the billed unit.
  4. Track denial reasons back to notes. When a claim gets denied or a day is cut on concurrent review, pull the note. If the note was thin, that’s a coaching opportunity, not just a billing problem.
  5. Feedback loop with clinicians. Auditing without coaching just creates resentment. Share patterns, not individual call-outs. Show clinicians the phrasing that got a day approved on appeal.

This is the pre-submission work that separates facilities getting paid clean from facilities stuck in a permanent appeals cycle. Catching documentation gaps before a claim goes out is the entire point — it’s why our behavioral health billing and utilization review teams sit close to the chart, not downstream of it. When UR and billing are handled by the same people reading the notes daily, weak documentation gets flagged in real time instead of surfacing six months later in an audit letter.

What should you do when a payer requests records for audit?

The instinct is to send everything, fast. That’s usually the wrong move. A few operational rules:

Confirm the exact scope. Which dates of service, which patients, which levels of care. Don’t send more than requested — you’re inviting scrutiny on records they didn’t ask for.

Review before you send. Have someone read every note being submitted. If a note is weak, that’s the moment to know — not after the clawback letter. You cannot amend a note after an audit request lands, but you can prepare an appeal narrative in advance.

Include the treatment plan and assessments. Progress notes make sense in the context of the initial assessment and treatment plan. Sending notes alone forces the auditor to fill in gaps unfavorably.

Track the timeline. Payers have response windows, and so do you. Missing a records deadline can result in automatic denial regardless of whether the care was justified.

Loop in appeals early. If your team can see certain notes are thin, start drafting the medical necessity argument now. Appeals written under a deadline after a denial hits are always weaker than appeals prepared alongside the initial records production.

Documentation doesn’t need to be beautiful. It needs to be specific, individualized, and tied to why this patient needs this level of care today. If your notes clear that bar consistently, most audits become paperwork instead of clawbacks. If you want a second set of eyes on where your current notes are exposed, our free 6-month billing audit includes a documentation review — start there.

Frequently Asked Questions

How long do payers have to request records for an audit?

It varies by payer and contract, but commercial payers commonly reserve the right to audit for 12–24 months post-payment, and some contracts extend further. Medicaid MCOs and Medicare have their own lookback windows defined in program rules. The practical takeaway: assume any date of service could be reviewed for at least two years, and document accordingly.

Can we amend a progress note after a payer requests it?

No — not in a way that would hold up. Late amendments made after an audit request are generally viewed as tampering and can escalate the situation from clawback to fraud investigation. Legitimate late entries must be clearly dated as such and made before any audit notice. The only real fix for a weak note is a stronger appeal narrative built around the existing documentation.

What’s the single most common progress note issue that triggers denials?

Copy-forward or template-heavy notes that don’t show individualized clinical thinking day to day. When a reviewer sees identical language across multiple days, they conclude the patient’s condition didn’t warrant continued stay at that level of care — or that the clinician didn’t reassess. Either conclusion supports a denial.

Do UR reviewers and post-payment auditors read notes the same way?

Broadly yes, but concurrent UR reviewers are asking “does this patient still need this level of care today?” while post-payment auditors are asking “was every day billed medically necessary at the level billed?” The documentation that satisfies one generally satisfies the other, which is why note quality should be built for the stricter audience — the post-payment auditor.

How often should treatment centers audit their own charts?

A weekly random sample of 5–10 charts across levels of care is a reasonable baseline. The goal isn’t perfection — it’s catching patterns early enough to coach clinicians before those patterns show up in a payer audit. Facilities that don’t self-audit find out about documentation problems from a clawback letter, which is the most expensive way to learn.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You bill a 90837 for a 55-minute individual therapy session. Two weeks later it comes back denied for medical necessity, or downcoded to 90834, or flagged as a duplicate because the client also had a 90791 the same week. Multiply that across a dozen clinicians and a full schedule, and the AR aging report starts looking ugly fast.

Most behavioral health denials aren’t mysterious. They come from a short list of CPT and HCPCS coding mistakes that payers have been flagging the same way for years. Below are the errors that show up most often on 90791, 90834, 90837, and the H-codes — with the fix for each.

The short version

  • 90791 gets denied most often for frequency (billed more than once per episode without justification) and for being billed same-day as a therapy code by the same provider.
  • 90837 is the single most audited outpatient therapy code — commercial payers routinely downcode to 90834 when documented time doesn’t clearly support 53+ minutes.
  • H-codes (H0001, H0004, H0005, H0015, H2035, H2036) are Medicaid-specific and vary by state — using the wrong code or wrong modifier for the state’s fee schedule triggers automatic rejection.
  • The fix for almost all of it is pre-submission scrubbing tied to actual documentation, not just format checks.

Why does 90791 keep getting denied?

CPT 90791 is the psychiatric diagnostic evaluation — the intake. It’s a high-value code, and payers scrutinize it accordingly. Four patterns cause the majority of denials.

Billed too frequently. Most commercial payers allow one 90791 per provider per episode of care, and some limit it to one per 6 or 12 months regardless of provider. Billing a new 90791 when a client returns after a 60-day gap, or when they transition from PHP to IOP with the same provider, will get denied. If a re-evaluation is clinically warranted, documentation has to explicitly state why — a significant change in presentation, a new diagnosis, or a gap long enough to meet payer criteria.

Billed same-day as therapy by the same provider. 90791 and 90834/90837 are generally not payable on the same date of service by the same rendering provider. Different providers, same day, is usually fine with proper documentation. Same provider, same day, needs a modifier and a clinical reason — and even then, many payers will deny.

Wrong provider type. 90791 can be billed by licensed clinicians (LMFT, LCSW, LPC, psychologists, psychiatrists), but state and payer rules on who counts as “licensed” vary. Associate-level or registered interns often can’t bill 90791 directly — the claim needs to go under a supervising provider with the appropriate supervision modifier.

Missing or vague diagnosis. A 90791 claim with an unspecified F-code (like F19.20, “other psychoactive substance dependence, uncomplicated”) when the intake clearly supports a specific diagnosis invites denial or audit. Code to the highest specificity the documentation supports.

What’s the difference between 90834 and 90837 for billing purposes?

90834 is individual psychotherapy, 38–52 minutes. 90837 is individual psychotherapy, 53 minutes or longer. The reimbursement gap is meaningful — often 30 to 50 percent higher for 90837 — which is exactly why payers audit it.

Commercial payers have been running 90837 utilization reviews for years. Some send letters asking clinicians to justify “high utilization” of 90837. Others quietly downcode. A few require prior authorization for 90837 after a threshold number of sessions per year.

Three fixes prevent most 90837 denials and downcodes:

Document actual start and stop times. Not “session lasted approximately one hour.” Actual times — 2:03 PM to 3:01 PM. Payers auditing 90837 want to see 53 minutes of face-to-face therapeutic contact documented, not inferred.

Justify the length clinically. The note should make it obvious why the session ran long — trauma processing, crisis stabilization, family involvement, complex case conceptualization. “Continued to discuss coping skills” for 55 minutes reads like padding. “Client disclosed new trauma history; extended session to complete safety planning and grounding” reads like a 90837.

Don’t default to 90837. If a program’s billing pattern is 85% 90837 and 15% 90834, that’s a payer audit waiting to happen. Coding should reflect the actual mix of session lengths, not a facility policy of “bill the higher one when possible.”

Which H-codes cause the most Medicaid denials?

H-codes are HCPCS Level II codes used primarily by Medicaid programs for substance use and mental health services. They’re where state-by-state variation causes the most trouble. A code that pays cleanly in one state gets rejected in the next because the state’s Medicaid MCO uses a different code or requires a different modifier for the same service.

H0001 (alcohol and/or drug assessment). Frequently denied when billed same-day as 90791, since they overlap conceptually. States differ on which is preferred — some Medicaid MCOs want H0001 for SUD intakes and reserve 90791 for mental health intakes. Billing the wrong one is an automatic rejection.

H0004 (behavioral health counseling, per 15 minutes). Time-based, so unit calculations matter. Billing 4 units for a 55-minute session is defensible; billing 6 units for the same session is fraud territory. Some states require rounding rules that follow CMS’s 8-minute rule; others don’t.

H0005 (alcohol and/or drug services; group counseling). Denials usually come from group size violations (payer caps group at 12, facility ran 15) or from missing documentation of each individual client’s participation in the group note.

H0015 (intensive outpatient program). A per-diem code with strict hour requirements — typically 9+ hours per week of structured programming. Denials happen when the client attends fewer hours than the code requires and the facility bills H0015 anyway. Many states also require specific modifiers (HF for SUD, HE for mental health).

H2035 and H2036 (alcohol/drug treatment program, per hour and per diem). These get confused with H0015 and with each other. Using the per-diem code for a program that should bill per-hour, or vice versa, gets denied every time.

The fix is a state-specific Medicaid billing grid — the exact codes, modifiers, unit rules, and documentation requirements for every MCO the facility contracts with. Generalist billing teams often don’t build this, which is one reason behavioral health billing works better when the team only bills behavioral health.

Which modifiers get missed and cost the most money?

Modifier errors are boring, invisible, and expensive. The ones behavioral health facilities miss most often:

HF (substance abuse program) and HE (mental health program). Required by many Medicaid MCOs to distinguish SUD services from mental health services when the same base code is used for both. Missing modifier = denial.

HO, HN, HM (masters, bachelors, less than bachelors level provider). Medicaid uses these to pay at the correct rate for the provider’s credential level. Wrong modifier = wrong rate or denial.

95 or GT (telehealth). Post-pandemic rules keep shifting. Some payers want 95, some want GT, some want POS 02 or POS 10 instead of a modifier, some want both. Getting this wrong on a telehealth-heavy caseload creates a wall of denials fast.

59 or XE/XS/XP/XU (distinct procedural service). Needed when two services that would normally bundle are legitimately separate. Overuse triggers audits; underuse triggers denials on legitimately separate services.

How do you stop these denials before the claim goes out?

Every mistake above is catchable before submission. The question is whether anyone is actually catching them.

A pre-submission scrub that only checks claim format — valid NPI, valid diagnosis code, valid date — will pass all of the errors above. What’s needed is a clinical-plus-coding scrub that checks:

  • Does the documented session time support the CPT code billed?
  • Is the diagnosis specific enough to support the service?
  • Are frequency limits being respected across the episode of care?
  • Is the correct H-code and modifier combo being used for this specific payer in this specific state?
  • Is the rendering provider credentialed and enrolled with this payer for this code?

That’s the difference between billing that catches problems early and billing that just submits and hopes. At Global AHS, denial prevention starts at the coding review before claim submission — not at the appeal after the denial. The verification of benefits process feeds directly into how services get coded on the back end: authorization limits, allowed codes, and payer-specific rules get flagged up front, not discovered in a denial letter.

If you want to see how your current coding is actually performing, request a free six-month billing audit. The specific codes, denial reasons, and dollar amounts left on the table become visible when someone actually looks.

Frequently Asked Questions

Can 90791 and 90837 be billed on the same day?

Generally not by the same rendering provider. Most payers consider the intake and the therapy session overlapping when performed by the same clinician on the same day. Different providers on the same day is usually acceptable with clear documentation. Same provider requires a modifier, a clinical justification, and even then many payers will deny — so it should be avoided when possible.

Why is 90837 getting downcoded to 90834?

Commercial payers downcode 90837 when documentation doesn’t clearly support 53 or more minutes of face-to-face therapeutic contact. The fix is documenting actual start and stop times (not “approximately one hour”), and writing a clinical rationale for why the session ran long. If the note reads like a 45-minute session that got stretched, the payer will pay it as one.

How often can 90791 be billed for the same client?

Most commercial payers allow one 90791 per provider per episode of care. Some limit it to once per 6 or 12 months regardless of provider. A repeat 90791 needs explicit documentation of why re-evaluation is clinically necessary — significant change in presentation, new diagnosis, or a gap in care long enough to meet payer criteria.

What’s the difference between H0015 and H2036?

Both are used for intensive outpatient and structured outpatient programming, but they’re not interchangeable. H0015 is specifically an IOP per-diem code with a minimum hour requirement (typically 9+ hours per week). H2036 is a broader alcohol/drug treatment per-diem. Which one to use depends on the state Medicaid program and the payer’s fee schedule — using the wrong one causes automatic denials.

Do telehealth behavioral health sessions need a modifier?

It depends on the payer. Some require modifier 95, some require GT, some want the place of service code (02 or 10) instead of a modifier, and some want both a POS and a modifier. Post-pandemic rules keep shifting and vary by payer and by state Medicaid program. A telehealth-specific billing grid per payer is the only reliable way to keep denials down.

Should a billing team catch coding errors before submission?

Yes — and most format-only scrubs don’t. A useful pre-submission review checks whether documented session time supports the CPT code, whether frequency limits are being respected across the episode, whether the correct H-code and modifier combo matches the payer’s rules, and whether the rendering provider is credentialed for that code. Catching those issues before submission is far cheaper than appealing them after denial.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

A commercial payer sends a letter demanding $84,000 back on residential claims paid two years ago, citing a retrospective medical necessity review. You have 30 days to respond. Your billing team is already stretched, the clinical documentation is buried in an archived chart, and if you do nothing, the payer offsets that $84,000 against your next check — which was supposed to cover this month’s payroll.

This is a payer takeback, also called a recoupment or overpayment recovery. It’s one of the more disruptive things that hits a behavioral health revenue cycle, because the money already landed in your account, you already spent it on operations, and now the payer wants it back — often on claims that paid cleanly the first time.

The wrong response is to panic-refund or ignore the letter. The right response is structured: verify the request is valid, decide whether to refund or appeal, and hit the deadline every time.

What every operator should know about takebacks

  • Not every recoupment demand is valid — audit the request against your contract, state prompt-pay laws, and the payer’s own lookback limits before cutting a check.
  • You almost always have appeal rights, and requesting an appeal typically pauses the offset while the dispute is reviewed — but only if you ask for the stay in writing.
  • Response deadlines are short (usually 30–45 days) and non-negotiable. Miss them and you lose the right to dispute regardless of the merits.
  • The single biggest predictor of recoupment risk is documentation quality at the time of service — the fight starts in the UR and clinical charting workflow, not the appeal letter.

What is a payer takeback and why did I get one?

A takeback is a payer’s retroactive claim that they paid you too much — or shouldn’t have paid you at all — and want the money back. It shows up as a demand letter, a remittance code on a subsequent EOB, or a silent offset on your next payment where the check comes in smaller than expected with a note buried in the ERA.

The common triggers in behavioral health are retrospective medical necessity reviews (the payer decides after the fact that residential or PHP wasn’t warranted), coordination of benefits issues (they discovered another payer was primary), coding audits (they’re challenging the level of care billed, often H0010 vs. H0035 vs. H0018), duplicate payment claims, eligibility errors (the member wasn’t actually covered on those dates), and credentialing gaps (they’re claiming you weren’t in-network or properly credentialed when the service was rendered).

The reason you’re seeing more of them: commercial payers have expanded their Special Investigations Units and payment integrity vendors over the last several years, and behavioral health — particularly SUD residential and PHP — is a favorite target because average claim values are high and clinical documentation standards vary widely across facilities. When a payer sends one recoupment and gets paid without pushback, they send more. Treatment centers that fight back with organized appeals get sampled less often over time.

When should I refund vs. appeal a recoupment demand?

The instinct to just refund and move on is usually wrong. Before you write any check, run the request through a short checklist.

Refund when: The payer is right. If COB shows another plan was primary and paid, if the member truly wasn’t eligible on those dates, if you billed a duplicate — pay it back promptly and document the correction. Fighting a legitimate overpayment wastes time and damages the payer relationship you need for the next contracting cycle.

Appeal when: The demand is based on retrospective medical necessity denial and your clinical documentation supports the level of care. Appeal when the lookback period exceeds what your contract or state law allows — many states cap commercial recoupment at 12, 18, or 24 months, and Medicaid MCOs have their own rules. Appeal when the payer’s cited policy wasn’t in effect on the date of service. Appeal when they’re recouping on a claim you already appealed and won. Appeal when the coding challenge is subjective and your notes justify the level billed.

Negotiate when: Some of the claims in the batch are defensible and some aren’t. Payers will often accept a partial refund plus documentation on the rest rather than fight through 40 individual appeals. This is the most common outcome on large-batch recoupment demands, and it’s where an experienced RCM partner earns their fee — knowing which claims to concede and which to defend, and getting that agreement in writing before you cut the check.

Quick rule of thumb: if the recoupment letter cites “medical necessity” and your utilization review team got prior authorization for those dates, you appeal. The payer already approved the level of care in real time. Retrospectively pulling it back is a fight you can usually win if the UR notes and clinical documentation are in order — which is exactly why front-end utilization review is the first defense against back-end takebacks.

What are the response deadlines and what happens if I miss them?

Deadlines vary by payer, product, and state, but the ranges are narrower than most operators realize:

  • Commercial payers: Typically 30–45 days from the date of the recoupment notice to request an appeal or dispute. Some contracts specify 60. Read the actual letter — the deadline is always stated in it.
  • Medicaid MCOs: Usually 30–60 days, but state Medicaid rules override the MCO’s default. Some states require the plan to give providers 60–90 days.
  • Medicare: Overpayment appeals follow the standard five-level appeal process. First-level redetermination is 120 days from the demand letter, but if you want to stop recoupment during the appeal, you generally need to file within 30 days.

Missing the deadline is the single most expensive mistake in recoupment defense. Once the appeal window closes, the payer’s determination stands, the offset proceeds, and you’ve lost the right to contest even if your documentation is airtight. Observation from our own book: a meaningful share of recoupment losses at treatment centers aren’t because the appeal was weak — they’re because the letter sat in an inbox until it was too late.

Practical fix: any recoupment or overpayment letter should be logged the day it arrives, with the deadline calculated and calendared, and assigned to a named person. If mail goes to a facility address that gets checked weekly, you’re already burning a quarter of your response window.

How do I actually write and file a recoupment appeal?

The appeal itself is straightforward if the documentation exists. It should include:

A cover letter that references the specific recoupment ID, claim numbers, dates of service, and member IDs — and states clearly that you are formally appealing the recoupment and requesting that any offset be paused pending resolution. Cite the contract provision or state law that supports your position (prompt-pay statutes, contractual lookback limits, appeal rights language).

The clinical case for each claim: the assessment, level of care criteria met (ASAM dimensions, if applicable), UR authorization numbers and dates, progress notes for the disputed dates of service, and physician documentation. If the payer approved the level of care prospectively, lead with that — retrospective denial after prospective approval is an uphill fight for them.

Coding and billing documentation: the original claim, the ERA showing payment, and any prior correspondence. If they’re challenging the level of care code, include the crosswalk showing why the code you billed matches the services rendered.

Send it certified mail or through the payer’s provider portal with confirmation. Follow up in writing at 30 days if you haven’t heard back. Escalate to the payer’s provider relations rep, then to the state department of insurance or Medicaid oversight if the payer is non-responsive past their own stated timelines.

One note on offsets during appeal: most commercial contracts allow the payer to begin recouping via offset even while an appeal is pending, unless you specifically request a stay in your appeal letter. Ask for the stay in writing. Some payers grant it as a matter of policy; others don’t, but you won’t get it if you don’t ask.

How do I stop recoupments from happening in the first place?

Every recoupment is a signal that something upstream was weak. The fix is a tighter front end.

Pre-service: Real verification of benefits that captures COB status, actual level-of-care benefits, authorization requirements, and any medical necessity criteria the payer uses. If a payer requires an ASAM-based assessment for residential authorization, capture it before the admit, not after.

Concurrent: Utilization review that documents medical necessity in the payer’s language, not just clinical shorthand. If the reviewer at the payer needs to see specific ASAM dimensions or DSM criteria to approve continued stay, those need to be in the notes for every review call.

Claim submission: Clean claims that match the authorization, correct level-of-care codes, and complete documentation packaged and ready if requested. Global AHS works only in behavioral health and SUD billing, which means we know the specific payer patterns — which plans retrospectively review H0010, which ones challenge H0018 length of stay, which ones send COB requests six months after payment — and we build the front end to prevent the takeback, not just fight it after.

Post-payment monitoring: Watch the ERAs. Recoupment offsets often appear as small adjustment codes on unrelated claims. If you’re only reconciling at the check level, you’ll miss them until they add up.

Every treatment center will get recoupment letters. The ones that stay solvent treat those letters as a workflow, not a crisis — read the letter, calendar the deadline, decide refund vs. appeal, file cleanly, and fix whatever upstream gap let it happen.

If you want a second set of eyes on a recoupment letter that just hit your desk, or a look at where your front end is leaking, request a free 6-month billing audit — we’ll quantify the exposure before you commit to anything.

Frequently Asked Questions

How long does a payer have to request a recoupment on a claim they already paid?

It depends on the contract and state law. Many commercial contracts allow lookback periods of 12 to 24 months, and several states cap it by statute. Medicare and Medicaid have their own rules — Medicare generally allows a much longer overpayment recovery window. Always check both your contract and the applicable state prompt-pay or overpayment statute before assuming the request is valid.

Can a payer just take the money out of my next check without asking?

Often yes, if the contract permits offset. Most commercial and Medicaid MCO contracts include language allowing the payer to recoup overpayments by offsetting future payments after providing notice. You typically retain the right to appeal, and requesting a stay of offset in your appeal letter can pause the recovery while the dispute is reviewed — but you have to ask for it in writing.

Should I always appeal a recoupment demand?

No. If the payer is genuinely correct — a coordination of benefits issue, an ineligible member, a duplicate payment — refund it promptly and move on. Appeal when the demand is based on retrospective medical necessity denial and your documentation supports the level of care, when the lookback exceeds contract or state limits, or when the payer’s cited policy didn’t apply on the date of service.

What’s the deadline to appeal a payer takeback?

Typically 30 to 45 days from the notice date for commercial payers, 30 to 60 days for Medicaid MCOs (depending on state rules), and specific tiered timelines for Medicare. The exact deadline is always stated in the recoupment letter. Miss it and you generally lose the right to dispute regardless of the merits, so the letter should be calendared the day it arrives.

What documentation do I need to win a medical necessity recoupment appeal?

The original clinical assessment, ASAM or equivalent level-of-care criteria documentation, all UR authorization numbers and dates, daily progress notes and physician documentation for the disputed dates of service, and the original claim and ERA. If the payer prospectively authorized the level of care, lead with that — retrospective denial of a prospectively approved stay is a fight the payer often loses when documentation is complete.

How can I reduce how often I get recoupment letters?

Tighten the front end. Thorough verification of benefits that captures COB and authorization requirements, utilization review that documents medical necessity in the payer’s specific language, clean claim submission matched to the authorization, and post-payment ERA monitoring to catch small offsets before they compound. Most recoupments trace back to a weakness upstream of billing, not a billing error itself.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

Your biller has been out for two weeks. There’s a stack of aged claims nobody’s touched, a concurrent review got missed because the one person who knew the payer portal login is the one who’s out, and you’re doing napkin math on whether to hire a second biller, replace the first with a bigger team, or hand the whole function to an outside RCM shop. That’s the decision. Most operators make it on gut instead of numbers.

The answer isn’t universal. A 12-bed detox with one commercial contract has a different equation than a 60-bed residential with five MCO contracts and PHP/IOP step-downs. But the framework is the same. Below is how to run the math — and the operational questions the math doesn’t capture.

The short version

  • Fully-loaded cost of an in-house billing team is roughly 2x base salary once you add benefits, software, training, turnover, and management time.
  • The break-even isn’t just cost — it’s net collection rate. A 4–6 point difference in collections usually dwarfs any labor savings.
  • Facilities under ~$4M in annual net revenue almost always come out ahead outsourcing. Above ~$15M, a hybrid model often wins. The messy middle is where the framework matters.
  • The hidden costs of in-house — key-person risk, credentialing gaps, UR misses, appeal backlogs — show up on your AR aging report, not your billing salary line.

What does an in-house billing team actually cost?

Operators underprice this. Every time. Base salary for a behavioral health biller in most markets runs $55K–$75K, with a senior lead or manager pulling $85K–$110K. That’s the number on the offer letter. It’s not the number that hits your books.

Add employer taxes and benefits at 25–30% of salary. Add billing software licenses, clearinghouse fees, and the modules your EHR system charges extra for. Add training — a new biller doing behavioral health takes six months to be genuinely useful, and someone senior loses productive hours teaching them. Add turnover, which runs high in RCM roles. Every time a biller leaves, you eat 3–6 months of ramp on the replacement plus whatever fell through the cracks during the gap.

Then add the layers people forget: a manager to actually manage the biller (or your CFO doing it and getting pulled off finance), a credentialing person or contracted service, appeals handling, and someone competent to cover PTO and sick days. A one-person billing shop isn’t a billing department. It’s a single point of failure with a job title.

For a small facility, fully-loaded cost for one biller comes to $85K–$110K. A functional team — biller, UR coordinator, credentialing support, a lead — realistically runs $300K–$450K annually before software and the space they sit in.

How do you calculate the break-even against outsourcing?

Outsourced RCM in behavioral health typically runs 5–8% of net collections — sometimes higher for facilities with heavy UR demand, sometimes lower for high-volume commercial-heavy books. Surface-level math:

In-house cost / expected net collections = your effective billing overhead percentage.

If your fully-loaded team costs $350K and you collect $5M net, you’re at 7% — right in the outsourced range. If you collect $8M with the same team, you’re at 4.4%, and in-house looks cheaper on paper. If you collect $3M with that team, you’re at 11.6%, and outsourcing is a no-brainer on cost alone.

The surface math misses the biggest lever: net collection rate. If an outsourced team collects 94% of what’s collectable and your in-house team collects 88%, the 6-point gap on $5M is $300K of additional cash — which covers the outsourced fee and then some. This is where most in-house shops actually lose. Not because the biller is bad, but because they’re doing billing, UR follow-up, appeals, patient responsibility, and payer phone calls all at once, and appeals older than 45 days quietly get abandoned.

The honest way to run the comparison: what’s your current net collection rate, what are your days in AR, and what’s your first-pass denial rate? If you don’t know all three off the top of your head, that itself is an answer.

When does in-house billing actually make sense?

It’s not never. In-house works when a few conditions line up:

  • You’re above roughly $12–15M in net revenue and can support a real team with redundancy — not one biller and a prayer.
  • Your payer mix is stable and concentrated in a handful of contracts your team can master, rather than a shifting book of out-of-network commercial and multiple Medicaid MCOs.
  • You have leadership bandwidth — an experienced RCM director or CFO who’s been in behavioral health billing specifically, not a general accountant.
  • Your levels of care are relatively simple. If you’re running detox, residential, PHP, IOP, and OP with different auth cadences and different payers for each, operational complexity favors specialists.

Even facilities that meet all four often run a hybrid — in-house for day-to-day billing and posting, outsourced for UR, credentialing, contract negotiations, and complex appeals. That’s a legitimate model, and often the strongest at scale.

What operational risks does the cost math miss?

Cost per collected dollar is the right headline metric, but it hides four risks that eat facilities alive:

Key-person risk. One biller who knows your payer portals, your auth patterns, your appeal history — and takes it all with them when they leave. If your billing function collapses when one person goes on vacation, you don’t have a department. You have a liability.

UR and billing disconnect. When utilization review lives with clinical and billing lives with finance, denials for medical necessity and level-of-care downgrades fall in the gap between them. Nobody owns the appeal, and the days-to-appeal clock runs out. Handling billing and utilization review under one roof — internally or through a partner who does both — closes that gap.

Credentialing lag. A new clinician sits uncredentialed for four months, and every session they run in that window is either written off or billed under someone else’s NPI, which is its own problem. In-house credentialing done part-time by an admin is where a lot of revenue quietly disappears.

Contract staleness. Payer contracts signed three years ago at rates that made sense then don’t get renegotiated because nobody has the time or leverage. This is where a partner with active contract negotiation workflows pays for itself in a single renewal cycle.

None of these show up on the salary line. All of them show up in AR aging and net collection rate.

How do you actually make the decision?

Run this in order:

1. Pull your current numbers. Net collection rate, days in AR, first-pass denial rate, aged AR over 90 and 120 days, and total billing/UR/credentialing cost (salaries + software + benefits + your time). If you can’t pull these, that’s the first problem to solve.

2. Benchmark honestly. Healthy behavioral health facilities generally run days in AR in the 30–45 day range, first-pass denial rates under 10%, and net collection rates above 92%. If you’re materially outside those, in-house isn’t working — regardless of what it costs.

3. Model both scenarios at 12 and 24 months. Not just cost — cash collected. A partner that moves your net collection rate from 87% to 93% at a facility doing $6M generates roughly $360K in additional annual cash. That’s the number that matters.

4. Stress-test the in-house model for turnover. What happens if your lead biller leaves in month seven? What’s the backfill plan? What does 60 days of degraded performance cost you?

5. Get an outside audit before you decide. Global AHS runs a free six-month billing audit for facilities weighing the switch — not as a pitch, but because you can’t compare options without knowing what’s actually leaking. We look at denial patterns, aged AR, UR outcomes, and credentialing status and give you the number in dollars.

The decision is rarely “in-house is better” or “outsourcing is better.” It’s usually “here’s what’s leaking now, here’s what each model would cost to fix, and here’s the payback period.” Once you have that on paper, the answer picks itself.

If you want the audit before you decide, get in touch.

Frequently Asked Questions

At what revenue level does in-house billing start to make sense?

As a general rule, behavioral health facilities under about $4M in annual net revenue come out ahead outsourcing, because they can’t support a team with real redundancy. Above roughly $12–15M, in-house or hybrid models become viable — assuming stable payer mix and experienced RCM leadership. The range in between depends more on payer complexity and current collection performance than on revenue alone.

What’s the typical fee for outsourced behavioral health billing?

Most outsourced RCM in behavioral health runs 5–8% of net collections, with variation based on payer mix, level-of-care complexity, and whether utilization review and credentialing are included. The right comparison isn’t the fee percentage in isolation — it’s the fee against the net collection rate the partner actually delivers.

How do I calculate the fully-loaded cost of an in-house biller?

Take base salary, add 25–30% for taxes and benefits, then add software and clearinghouse costs, a share of management time, training and ramp cost for new hires, and expected turnover cost. A biller with a $65K salary typically has a fully-loaded cost of $90K–$110K annually. A functional team of biller, UR coordinator, and credentialing support usually runs $300K–$450K before software.

What’s the biggest hidden risk of running billing in-house?

Key-person risk. Most in-house shops are one or two people who hold all the payer knowledge, portal logins, and appeal history in their heads. When one leaves — and in RCM roles, turnover is high — collections drop for 3–6 months while a replacement ramps, and appeals with tight deadlines get abandoned. That gap rarely shows up on the P&L as a billing expense, but it shows up in aged AR.

Can I run a hybrid model with in-house billing and outsourced UR or credentialing?

Yes, and at scale it’s often the strongest setup. Larger facilities frequently keep day-to-day billing and posting in-house while outsourcing utilization review, credentialing, contract negotiations, and complex appeals to specialists. It keeps operational control internal while pulling in expertise on the functions where mistakes are most expensive.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You’re billing methadone administration daily, submitting HCPCS codes for buprenorphine, and running counseling claims through your behavioral health payer mix — and your denial rate on MAT services is still running higher than the rest of your book. Sometimes the payer says the counseling is bundled. Sometimes they say it isn’t. Sometimes they pay for six months and then claw it back after a records request.

Medication-assisted treatment sits at the intersection of medical, behavioral, and pharmacy billing rules. That’s exactly why payers deny it and auditors flag it. If your billing team learned MAT on the fly, you’re likely leaving 10–20% of legitimate revenue on the table and carrying audit exposure you don’t see yet.

The short version

  • Bundling rules vary by payer and level of care. Opioid Treatment Program (OTP) weekly bundles under Medicare (HCPCS G2067–G2080) work differently than commercial fee-for-service MAT, and Medicaid MCOs have their own carve-outs.
  • The three biggest denial drivers are missing or wrong modifiers, counseling billed separately when it’s bundled into the OTP weekly rate, and diagnosis codes that don’t support medical necessity for the medication administered.
  • MAT audits are aggressive and retrospective. Payers routinely request 12–24 months of records and recoup on documentation gaps — even when the clinical care was appropriate.
  • Clean MAT billing requires payer-specific rules, tight counseling documentation, and a UR process that anticipates medical-necessity questions before the claim goes out.

Is your MAT program billed as a bundle or fee-for-service?

This is where most billing errors start. The answer depends on your program type, your payer, and the setting.

Opioid Treatment Programs (OTPs) billing Medicare use the weekly bundle codes G2067 through G2080. These bundles include the medication (methadone, buprenorphine, or naltrexone), drug testing, individual and group counseling, and intake activities within a defined weekly episode. If you bill an office visit or counseling code on top of the bundle for the same week, expect a denial — or worse, a recoupment down the line.

Commercial payers are less consistent. Some mirror the Medicare OTP bundle structure. Some pay office-based opioid treatment (OBOT) as separate E/M visits (99202–99215) plus HCPCS codes for the medication and administration (J-codes for injectable naltrexone, Q-codes and J-codes for buprenorphine products). Some allow separate counseling codes (H0004, H0005, 90832–90837) when documentation supports a distinct encounter.

Medicaid MCOs are where it gets ugly. Rules differ by state, and often by plan within the state. One MCO will pay for MAT medication and counseling on the same date of service with a modifier 25. Another will bundle them. A third will pay both but require a separate authorization for each.

The practical answer: build a payer-specific billing grid for MAT before you submit a single claim. Every new contract you sign should trigger an update to that grid. If your billing team is guessing, denials are inevitable.

Which MAT denials show up most often?

After enough appeals work, the same denials show up over and over.

1. Bundled service billed separately. This is the OTP weekly-bundle problem — counseling, testing, or an E/M billed on top of G2067–G2080. Denial reason codes typically point to CO-97 (payment included in another service). These are rarely worth appealing on Medicare; on commercial, sometimes.

2. Missing or wrong modifier. Modifier 25 on an E/M performed the same day as a procedure. Modifier 59 to indicate a distinct counseling service from an administration. HG (opioid addiction treatment program) for state Medicaid claims that require it. When these modifiers are missing or misused, claims deny as duplicates or as bundled.

3. Diagnosis-code mismatch. Billing J0571 for buprenorphine with a primary diagnosis that isn’t F11.20 (opioid dependence, uncomplicated) or another accepted opioid use disorder code will draw a medical-necessity denial fast. Same for extended-release injectable naltrexone in an alcohol-use-disorder context — the diagnosis has to match the FDA indication the payer recognizes.

4. Missing authorization for injectable medications. Extended-release naltrexone and long-acting injectable buprenorphine both routinely require prior authorization on commercial and Medicaid plans. If your intake or clinical team administers the injection before authorization is on file, you’re eating the medication cost — which is not small.

5. Frequency and duration limits. Some payers cap counseling encounters per week. Some cap drug screens per month. Exceeding the cap without a documented clinical justification means the claim above the cap denies, even if the encounter happened.

Catching these before submission is the entire game. This is the part of the process where behavioral health billing and utilization review have to talk to each other — because a UR team that knows the payer’s medical-necessity language can flag a diagnosis or authorization gap before the biller ever sees the claim. Global AHS handles both functions in-house, which is why we catch problems early instead of appealing them later.

How do MAT audits actually work — and where do treatment centers get exposed?

MAT is a high audit-risk service line for three reasons: it involves controlled substances, it involves recurring high-dollar claims, and it involves counseling that has to be documented at a level payers accept.

Payer audits — commercial special investigations units (SIU), Medicaid integrity contractors, and Medicare RACs — typically pull 12 to 24 months of records and look for specific things:

Counseling documentation. If you bill counseling as part of an OTP bundle or as a separate H-code, the auditor wants to see a note that documents the encounter — date, duration, participants, clinical content, and a signature from a credentialed clinician. “Group check-in” is not enough. If the note doesn’t support the code billed, the payer recoups.

Medication administration records. For methadone, the daily dosing log has to match the days billed. Missed doses billed as administered is a fraud finding, not just a recoupment.

Medical necessity at the medication level. The physician’s note should establish the diagnosis, the rationale for the specific MAT medication chosen, and ongoing assessment. Templated notes that look identical week over week are a red flag.

Credentialing. Auditors will verify that the prescribing provider was properly enrolled with the payer on every date of service. Any gap — a lapsed enrollment, a provider added to your NPI group late — becomes a recoupment target. This is where clean credentialing and payor enrollment pays for itself: a missed enrollment date on a high-volume MAT prescriber can cost six figures on a look-back.

The retrospective piece is what surprises operators most. A payer can pay MAT claims cleanly for a year, then audit, and recoup on documentation issues going back the full look-back window. If your notes aren’t audit-ready today, you have a liability building silently on your balance sheet.

How do you structure MAT billing to reduce denials and audit risk?

A few practical fixes that move the needle:

Pre-bill scrubs specifically for MAT. Generic claim scrubbers catch obvious errors — invalid codes, missing NPIs. They don’t catch payer-specific bundling rules for MAT. Build custom edits: if payer X is billed G2067, block any counseling code on the same date of service. If payer Y is billed J0571, require modifier and diagnosis validation.

Tie authorizations to administration schedules. Injectable MAT medications should not be administered until the authorization is confirmed and documented. Build the check into the clinical workflow, not just the billing workflow.

Document counseling to the code, every time. If the code is 90837 (60-minute individual psychotherapy), the note has to support 53+ minutes of face-to-face time. If it’s H0005 (group counseling), the note has to identify the group, duration, and the participant’s engagement. Clinicians resist this — but the alternative is recoupment.

Reconcile MAT AR separately. Pull MAT service lines out of the general AR aging and watch them weekly. Days in AR on MAT should track close to the rest of your book. If MAT AR is aging faster, something in the front-end process is broken.

Review payer contracts for MAT-specific language. Some commercial contracts pay MAT services at a different rate than the standard behavioral health fee schedule. If you didn’t negotiate MAT rates specifically, you may be getting paid the default rate on high-cost medications. Contract renegotiation is the fix — and it’s worth doing before your next term.

When does outsourcing MAT billing pay for itself?

The straight answer: when the volume of MAT services justifies specialized attention, and when your in-house team is a generalist behavioral health billing team without deep MAT experience. Most operators cross that line without realizing it.

MAT billing is not a subset of standard SUD billing — it has its own coding logic, its own audit profile, and its own payer behavior. A billing team that handles 50 MAT claims a month using the same workflow as PHP and IOP claims will make predictable mistakes. Those mistakes compound. By the time a payer audit lands, the exposure is already built in.

A specialized team catches the payer-specific bundle rules, keeps the counseling documentation aligned with the codes billed, and reconciles authorizations against the administration schedule. That’s the difference between MAT as a clean revenue line and MAT as a chronic denial and recoupment problem.

Global AHS offers a free 6-month billing audit that quantifies exactly where your MAT claims are leaking — denials, underpayments, and audit exposure — before you commit to anything. Request the audit here.

Frequently Asked Questions

Can we bill counseling separately from the OTP weekly bundle codes G2067–G2080?

Under Medicare, no — individual and group counseling are included in the OTP weekly bundle. Billing counseling separately for the same week will deny as bundled (CO-97). Commercial and Medicaid MCO rules vary; some payers allow separately reportable counseling with modifier documentation, others mirror the Medicare bundle. Build a payer-specific grid before submitting.

Which MAT services most commonly require prior authorization?

Extended-release injectable naltrexone (J2315) and long-acting injectable buprenorphine products almost always require prior authorization on commercial and Medicaid plans. Oral buprenorphine may require PA depending on the payer. Administering these medications before authorization is confirmed is the fastest way to eat the medication cost yourself.

How far back can payers audit MAT claims?

Commercial payer look-back windows commonly run 12–24 months, though contracts vary. Medicare and Medicaid can go further under specific circumstances — up to several years for suspected fraud. The practical implication: documentation that isn’t audit-ready today is a liability building on your balance sheet, even if claims are paying cleanly right now.

What documentation do auditors look for on MAT counseling encounters?

Date of service, duration, participants (individual or group with group ID), clinical content specific to the encounter, and a signature from a credentialed clinician. Templated notes that repeat week over week are a red flag. If you bill 90837, the note must support 53+ minutes of face-to-face time. Weak notes get recouped even when the care was clinically appropriate.

Should MAT AR be tracked separately from the rest of our behavioral health AR?

Yes. MAT has different denial patterns, different authorization requirements, and different audit exposure than PHP, IOP, or outpatient behavioral health. Blending it into your general AR aging hides problems. If MAT days in AR run higher than the rest of your book, that’s a front-end signal — usually authorization or documentation — worth investigating immediately.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You billed a batch of telehealth IOP sessions with POS 02 like you always have, and half came back denied. Your biller swears nothing changed. The payer swears their policy is public. Meanwhile, your admissions team is telling patients telehealth is fine — and your AR is sliding past 45 days on services that should have paid in 21.

The post-PHE telehealth rules for behavioral health are inconsistent by design. Medicare loosened some rules and tightened others. Commercial payers each interpret parity laws differently. Medicaid MCOs run policies that contradict the state manuals they claim to follow. If your team is still billing telehealth the way they did in 2022, you’re losing money on underpayments and setting up recoupments 12–18 months out.

The short version

  • POS 10 (patient home) and POS 02 (other location) are not interchangeable — using the wrong one denies claims or pays them 15–20% short.
  • Modifier 95 vs. GT vs. 93 depends on payer, service type, and whether audio-only is permitted. Defaulting to 95 for everything will cost you.
  • State Medicaid programs and their MCOs set their own telehealth rules for SUD and mental health — including which levels of care can even be delivered virtually.
  • Audio-only coverage for behavioral health is the single biggest area of payer disagreement right now. Get the policy in writing per payer.

What POS code should you use for behavioral health telehealth?

Since January 2022, CMS split the old POS 02 into two codes: POS 02 (telehealth provided other than in the patient’s home) and POS 10 (telehealth provided in the patient’s home). For behavioral health, most sessions land in POS 10 — the client is at home on a phone or laptop. But this is where facilities get tripped up.

If your patient is in a sober living residence, is that “home”? If they’re physically at your PHP facility but joining a group virtually because the clinician is remote, that’s not POS 10 — that’s a facility-based service with a remote provider, which is a different billing scenario entirely. If the patient is in your outpatient office using a telehealth kiosk to see a psychiatrist across town, that’s POS 02.

Medicare pays POS 10 at the non-facility rate (higher) and POS 02 at the facility rate (lower). Commercial payers vary — some mirror CMS, some pay both at parity, some only pay one. Getting this wrong doesn’t always cause a denial; sometimes it underpays you by 15–20% and no one catches it for months.

The practical rule: match POS to where the patient physically is at the moment of service, not where the provider is. Document the patient location in the note. Payers are auditing this, and “the patient was at home” without a location statement in the record is starting to fail on post-payment review.

Which telehealth modifiers actually apply to behavioral health services?

Three modifiers do most of the work, and they are not interchangeable:

Modifier 95 — synchronous audio and video telehealth. This is the default for a real-time video therapy session, 90837 or 90834 delivered over a HIPAA-compliant video platform. Most commercial payers require 95 alongside POS 10.

Modifier GT — the older CMS modifier for interactive audio/video. Medicare largely dropped GT in favor of 95, but a handful of Medicaid programs still require GT on institutional claims. If you’re billing UB-04s for outpatient behavioral health, check the state manual before you assume 95.

Modifier 93 — audio-only synchronous. This is the modifier that separates the winners from the losers right now. Since 2022, CMS accepts 93 for behavioral health audio-only services with specific documentation requirements (patient must be unable or unwilling to use video, provider must be capable of video). Some commercial payers pay audio-only mental health with 93. Others require 95 with a special place-of-service code. Others don’t cover audio-only at all, or only cover it for established patients.

The consequence: if you bill 90837 + 95 + POS 10 for an audio-only session because your clinician wasn’t paying attention, and the payer audits the note, they can recoup the payment. Some payers are actively doing this — pulling notes, finding “phone session” language, and clawing back on modifier mismatch. Your clinical documentation and your billing modifier have to agree, session by session.

How do state Medicaid and MCO rules change telehealth billing?

This is where post-PHE billing gets genuinely unpredictable. State Medicaid agencies each set their own telehealth policies for behavioral health, and the MCOs administering those benefits often layer additional rules on top.

Some patterns that show up across states:

Level-of-care restrictions. Several states allow individual therapy via telehealth but exclude IOP or PHP group sessions from telehealth reimbursement entirely — or cap the percentage of a program’s hours that can be virtual. If you’re running a hybrid IOP and billing 100% of hours as telehealth, some Medicaid MCOs will deny the overage without warning.

Originating site rules. A few states still require the patient to be at a qualifying originating site (a clinic, an FQHC, a school) rather than at home for Medicaid telehealth. This is a pre-PHE holdover that some states never fully repealed, even after loosening temporarily.

Provider licensure and location. If your clinician is licensed in State A and the patient is physically sitting in State B when the session happens, most Medicaid programs will not reimburse — even if the provider is credentialed with that Medicaid MCO for in-state services. Compact licensure helps for some professions, not others.

Audio-only in Medicaid. States are all over the map. Some cover audio-only mental health permanently. Some sunset it in 2024. Some cover it only for specific CPT codes. This is a per-state, per-MCO answer, not a generalizable one.

The takeaway for operators: your billing team needs a matrix — payer by payer, state by state — that says which combinations of POS, modifier, and service type will actually pay. Building that matrix from denial data is one of the first things we do inside a behavioral health billing engagement, because it’s usually the fastest way to stop the bleeding.

What telehealth billing mistakes are causing the most denials right now?

From facility AR reviews, five mistakes drive most post-PHE telehealth denials:

1. POS 02 used for home-based sessions. The single most common. Payer either denies outright or downcodes to the facility rate.

2. Missing or wrong modifier. No 95 when required, or 95 attached to an audio-only session that should have carried 93.

3. Non-covered service billed as telehealth. Certain CPT codes — some testing codes, some group therapy codes, some SUD-specific H-codes — aren’t on the payer’s telehealth-eligible list. Billing them as telehealth denies them entirely; billing them as in-person when they were virtual is fraud.

4. Cross-state licensure violations. Provider licensed in one state, patient in another. Denied or clawed back.

5. Documentation that contradicts the claim. Note says “phone call,” claim says video. Note doesn’t mention patient location. Note doesn’t mention modality at all. On audit, all of these fail.

None of these are exotic. They’re the residue of billing workflows built during the PHE when everything paid and nothing was audited. That grace period is over. If you haven’t rebuilt your telehealth billing rules against current payer policies, you’re accumulating denials you don’t know about yet — some of which will surface as recoupments a year later.

How should treatment centers verify telehealth coverage before admission?

Your verification of benefits process needs telehealth-specific questions, not just general behavioral health coverage. At minimum, before you admit a patient to a program that will include virtual sessions:

Confirm whether the plan covers telehealth for the specific level of care (OP, IOP, PHP). Confirm whether audio-only is covered. Confirm whether there’s a limit on the number or percentage of virtual sessions. Confirm the patient’s home state matches the provider’s licensure. Get the rep’s name and a reference number.

This is tedious work. It’s also the difference between a clean claim and a denial 60 days later when the payer decides IOP group therapy wasn’t a covered telehealth service under that plan. A generalist RCM vendor won’t ask these questions — they’ll pull standard mental health benefits and move on. A behavioral-health-only shop knows to dig, because we’ve watched the denials come back on exactly this.

What documentation supports telehealth claims on audit?

Payers auditing telehealth claims want to see, in the note itself: the modality (video, audio-only, or in-person), the platform used in general terms (“HIPAA-compliant video platform” is enough), the patient’s physical location at the time of service, the provider’s location, the start and stop time, and the reason for telehealth if audio-only was used.

If your EHR template doesn’t prompt for these fields, they get skipped. If they get skipped, your appeal on the eventual denial is weaker. The fix is a template change, not a training memo — clinicians will comply with what the software forces them to fill in, not with an email from last quarter.

Next step

If your telehealth denials are climbing and you can’t tell whether it’s POS, modifier, or payer policy driving them, that’s the exact question a free 6-month billing audit answers. We’ll pull the denial data, map it against current payer rules, and tell you where the leaks are before you commit to anything.

Frequently Asked Questions

What is the correct POS code for behavioral health telehealth sessions delivered to a patient at home?

POS 10 is the correct code when the patient is physically at home during a synchronous telehealth session. POS 02 applies when the patient is at any other location, such as a clinic or telehealth kiosk. Using POS 02 for home-based sessions is one of the most common causes of telehealth denials and underpayments post-PHE.

When should modifier 93 be used instead of modifier 95 for behavioral health telehealth?

Modifier 93 is used for audio-only synchronous telehealth services when video is not used. Modifier 95 is for synchronous audio and video telehealth. If a therapy session was conducted by phone only, the claim should carry modifier 93 (if the payer covers audio-only), not 95. Billing 95 on an audio-only session risks recoupment on audit.

Does Medicaid cover audio-only behavioral health telehealth?

It depends on the state and the specific Medicaid MCO. Some states cover audio-only mental health services permanently, some sunset coverage after the PHE, and others limit coverage to specific CPT codes or patient circumstances. Confirm the current policy directly with each state Medicaid program and MCO before billing audio-only sessions.

Can behavioral health providers bill telehealth for patients located in another state?

Generally no, unless the provider is licensed in the state where the patient is physically located at the time of the session, or unless a licensure compact applies. Most Medicaid programs and many commercial payers will deny or recoup claims where the provider is not licensed in the patient’s state at the time of service.

What documentation is required to support a telehealth behavioral health claim on audit?

Notes should document the modality (video, audio-only, in-person), the platform type, the patient’s physical location during the session, the provider’s location, start and stop times, and the reason audio-only was used if applicable. Missing any of these elements weakens the record on payer audit and appeal.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You billed seven days of medically monitored detox. The payer paid three, then clawed back two more on retrospective review because the nursing notes on days five and six didn’t show active withdrawal symptoms. The patient was clinically appropriate for that stay. The billing was technically correct. But the documentation didn’t defend the level of care — and now you’re eating the loss or fighting an appeal you’ll probably lose.

This is the detox billing problem in one paragraph. The rules aren’t hidden. They’re just unforgiving, and they punish facilities that treat withdrawal management like a fixed protocol instead of a level of care that has to be justified every single shift.

What it takes to keep detox claims paid

  • ASAM level has to match the clinical picture in the chart — not the intake decision from three days ago.
  • Length-of-stay limits are soft on paper, hard in practice. Most commercial payers expect 3–5 days for uncomplicated withdrawal and require concurrent review to go longer.
  • Documentation triggers denials more than coding does. CIWA/COWS scores, vital sign trends, and medication administration times are what UR nurses look for.
  • Concurrent review timing is a common failure point. Miss a review window by a shift and the payer can deny the entire remaining stay.

Why does ASAM level of care matter for detox billing?

Withdrawal management sits across three ASAM levels: 3.2-WM (clinically managed residential), 3.7-WM (medically monitored inpatient), and 4-WM (medically managed intensive inpatient). Each has different staffing expectations, different medical acuity thresholds, and — critically for billing — different reimbursement rates and documentation demands.

The most common pitfall is billing 3.7-WM when the chart supports 3.2-WM. It’s an easy mistake because facilities often admit at 3.7 based on intake risk factors — history of seizures, benzo dependence, elevated vitals — but the patient stabilizes within 24 hours. If nursing continues charting routine assessments without documenting ongoing medical necessity for the higher level, the payer will downcode on review. You’ll get 3.2 rates for 3.7 work.

The opposite problem is worse: billing 3.2-WM when the patient actually needed 3.7. Facilities do this to avoid pre-auth friction, then face denials when the medical picture (IV fluids, PRN benzos every four hours, cardiac monitoring) obviously exceeds 3.2 scope. Payers read charts. If the treatment doesn’t fit the code, the claim doesn’t get paid.

The fix is boring and it works: match the ASAM criteria dimension-by-dimension in your admission note, and re-document alignment at every shift change. Dimension 1 (intoxication/withdrawal), Dimension 2 (biomedical), and Dimension 3 (emotional/behavioral) are the ones payers actually check. If your UR team isn’t looking at these three every day, you’re gambling.

How many days will insurance cover for detox?

There’s no universal LOS ceiling for withdrawal management, but there’s a de facto industry range that commercial payers work from: 3–5 days for alcohol, 5–7 for opioids with medication-assisted treatment, and 7–10 for benzodiazepines when tapering is medically supervised. Medicaid MCOs are often tighter. Anything beyond those windows requires clear, ongoing clinical justification.

Payers enforce LOS three ways. First, pre-authorization: most will approve an initial 2–3 days and require concurrent review for extensions. Second, concurrent review windows: miss the deadline (often every 24–72 hours depending on the payer) and the remaining stay is uncovered. Third, retrospective review: even after payment, they can pull the chart 60–90 days later and take money back if the documentation doesn’t defend the days billed.

The pitfall we see most often is treating the authorization as the finish line. UR gets a 5-day approval, and the clinical team relaxes. Then day 4 comes, the patient is stable, and no one has documented why day 5 is still medically necessary. The claim gets paid, then clawed back. This is why utilization review can’t be a once-a-day check — it has to be tied to shift documentation in real time.

The other LOS trap: continued stay based on psychiatric or behavioral factors after medical withdrawal has resolved. That’s not a detox day anymore. If the patient no longer meets 3.7-WM criteria but needs continued care, the level has to step down — usually to residential (3.5) or PHP — and the billing has to reflect that. Continuing to bill detox codes after withdrawal has resolved is one of the fastest ways to trigger a payer audit.

What documentation gaps cause detox denials?

Denials in detox billing almost always trace back to one of five documentation gaps:

1. Missing or inconsistent withdrawal scoring

CIWA-Ar for alcohol and COWS for opioids are the standard, and payers expect to see scores every 4–8 hours during active withdrawal. Scores that stay flat at zero across multiple shifts undermine the case for continued inpatient-level care. Scores that jump without a corresponding nursing narrative look fabricated. The pattern payers want to see: elevated scores at admission, response to intervention documented, gradual downtrend tied to medication and time.

2. Vital signs that don’t support the level of care

If you’re billing 3.7-WM, the vital sign record should show a reason for medical monitoring — tachycardia, elevated BP, temperature, or documented risk of deterioration. Three days of stable, normal vitals with no PRN interventions is a downcoding invitation.

3. Medication administration gaps

MAR (medication administration record) inconsistencies are a favorite target on retrospective review. If the physician order says librium 25mg q6h PRN CIWA >8, and the MAR shows doses given with CIWA scores of 4, the whole medical necessity case gets shaky. Every PRN dose needs a documented trigger.

4. Physician documentation that lags

Daily physician progress notes are non-negotiable for medically monitored and medically managed levels. Notes that are copy-pasted across days, or that don’t reference the current withdrawal status, get flagged. The physician has to document why today’s care is still at this level.

5. Discharge criteria not addressed

Payers want to see the discharge target from day one. If the chart never articulates what “ready for step-down” looks like for this patient, extended stays are hard to defend.

How often should concurrent review happen during detox?

Pre-authorization for detox is usually fast — commercial payers turn it around in a few hours — but the initial approval is short. Two or three days is common. From there, concurrent review keeps the authorization alive, and this is where facilities bleed revenue.

Common concurrent review failures: the UR nurse calls the payer with dimension summaries but no specific clinical detail; the review happens after the authorization window has closed, meaning any denied days are non-appealable; the reviewer references an ASAM level that doesn’t match what the chart actually supports; or — the most preventable one — no one on staff knows which payer requires review every 24 hours vs. every 72 hours, so reviews get missed on the tighter ones.

Every commercial payer and Medicaid MCO has its own concurrent review cadence and its own preferred format. Tracking that by hand across a caseload is how facilities miss windows. This is one of the areas where working with a billing partner that only handles behavioral health matters — a generalist RCM shop is learning your payer mix on your claims, and detox is the wrong place to be someone’s learning curve. Our team handles behavioral health billing exclusively, and we build the concurrent review calendar into the account setup so nothing gets missed.

What are the most common detox coding and claim submission errors?

Coding errors compound documentation errors. A few patterns worth naming:

Wrong revenue code paired with the right HCPCS. Detox is typically billed with H0008–H0014 or ASAM-aligned codes depending on the payer contract, matched to revenue codes in the 126x or 116x range. Mismatched pairings get rejected at the clearinghouse or denied on receipt. Payer-specific crosswalks matter.

Bundling errors. Some services performed during detox — lab draws, EKG monitoring, physician consults — may or may not be separately billable depending on the payer and the level of care. Billing them separately when they’re bundled generates denials; not billing them when they’re separately reimbursable leaves money on the table.

Missing modifiers. HG for MAT-related services, HF for substance abuse programs, and payer-specific modifiers can be the difference between clean pay and rework.

Late submission after concurrent review lapses. If a review was missed and days went unauthorized, some facilities still submit those days hoping for the best. This creates appeal work and, worse, patterns that flag your NPI for tighter scrutiny on future claims.

Discharge billing errors. The discharge day itself is handled differently by different payers — some pay it, some don’t. Getting this wrong across a full patient census adds up.

The through-line on all of this: detox billing isn’t a submit-and-hope process. Cleaner claims come from catching mismatches before submission — the ASAM level not matching the chart, the concurrent review window that’s about to close, the CIWA gap on day three. That’s the work.

Want to see where your detox claims are leaking revenue?

We offer a free 6-month billing audit for behavioral health facilities. Send us your detox claim data and we’ll show you where documentation, ASAM alignment, and concurrent review are costing you paid days. Start the conversation here.

Frequently Asked Questions

How many days will most commercial payers approve for alcohol detox?

Most commercial payers approve an initial 2–3 days and extend through concurrent review, typically capping uncomplicated alcohol withdrawal management at 3–5 total days. Longer stays require documented ongoing symptoms, complications, or comorbidities. Benzodiazepine and severe polysubstance cases can justify 7–10 days when tapering is medically supervised and documented shift by shift.

What’s the difference between ASAM 3.2-WM and 3.7-WM for billing purposes?

3.2-WM (clinically managed residential withdrawal) is for lower-acuity patients who need monitoring but not active medical intervention. 3.7-WM (medically monitored inpatient withdrawal) requires 24-hour nursing, physician oversight, and evidence of moderate-to-severe withdrawal risk. Billing 3.7 rates without chart evidence — PRN medications, elevated vitals, active symptoms — is the most common cause of downcoding on retrospective review.

Why do payers deny detox days even after they’ve been paid?

Retrospective review. Commercial payers and Medicaid MCOs can pull charts 60–90 days after payment and recoup funds if the documentation doesn’t support the days billed. The most common triggers are flat CIWA/COWS scores across multiple shifts, physician notes that don’t justify continued inpatient-level care, and MAR entries where PRN medications were given without documented symptom triggers.

How often should concurrent review happen during a detox stay?

It depends on the payer. Some commercial payers require review every 24 hours, others every 48 or 72. Medicaid MCOs vary further. Facilities that don’t track cadence by payer end up missing windows, and any days after a missed review are typically non-appealable. Building a payer-specific concurrent review calendar into the intake workflow is the fix.

Can we keep billing detox codes if the patient is stable but not ready for discharge?

No. Once medical withdrawal has resolved, the level of care needs to step down — usually to residential (ASAM 3.5) or PHP — and the billing has to reflect that. Continuing to bill withdrawal management codes for a stable patient is a fast path to audits, recoupments, and payer scrutiny of future claims from your NPI.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →

You open the AR aging report Monday morning and there it is again: a six-figure column sitting past 90 days, another chunk creeping toward 120, and a handful of claims about to hit timely filing. Your biller says they’re “working it.” The controller wants a number for the board. Somewhere in that pile is real money — money that funds payroll, not a write-off entry.

Aged AR in behavioral health is uniquely painful. Claims sit because a level-of-care authorization was short, because the payer wants a records request nobody saw, or because the original biller left and the trail went cold. This is a playbook for working those buckets before they die.

What it takes to clean up aged AR without leaking cash

  • Segment first, chase second. A single “90+” bucket is useless — split by payer, denial reason, and dollar value before touching the phones.
  • Timely filing and timely appeals are two different clocks. Most commercial payers give 90–180 days to file and a separate window (often 60–180 days from EOB) to appeal. Miss either and the claim is dead.
  • Work highest-dollar, highest-recoverability claims first — not the oldest. Age alone doesn’t predict payment.
  • Root-cause every write-off. If you don’t code why a claim died, the same denial will kill the next 40 claims behind it.

Why does behavioral health AR age faster than other specialties?

A residential or PHP claim is not a 99213. It’s a multi-day, multi-service, authorization-dependent bill with concurrent review requirements, level-of-care downgrades mid-stay, and a payer mix that includes out-of-network commercial plans and Medicaid MCOs with wildly different rules. Any one of those can stall a claim for weeks.

The common failure points that show up when auditing aged AR:

  • Authorization gaps. A patient stayed 14 days at residential; the payer authorized 10. Days 11–14 sit in AR waiting for a retro-auth that nobody submitted.
  • Level-of-care mismatches. The clinical record supports PHP but the claim went out as residential (or vice versa). The payer denies, and the correction never gets rebilled.
  • Missing documentation requests. The payer sent a records request to a fax line nobody monitors. Ninety days later, the claim is denied for non-response.
  • Coordination of benefits. Secondary payer needs the primary EOB. It never got attached. Claim sits.
  • Biller turnover. The person who knew what happened on that account is gone, and the notes say “following up.”

None of these are exotic. All of them are fixable — but only if you work the aging report with a system instead of a checklist.

How do you segment an aged AR bucket before you start working it?

Before anyone picks up a phone, run the aging report through four filters. The goal is to know exactly which claims are worth an hour of a biller’s time and which are already dead.

1. Segment by payer

Group 60/90/120+ day claims by payer category — commercial in-network, commercial out-of-network, Medicaid MCO, Medicare, and self-pay. Each group has different filing windows, appeal rights, and negotiation leverage. An out-of-network commercial claim at 100 days is a very different animal than a Medicaid MCO claim at 100 days.

2. Segment by denial reason

Pull the CARC (Claim Adjustment Reason Code) and RARC codes from the 835s. If 50 claims from one payer are denied with CO-197 (authorization missing), that’s a project — not 50 individual phone calls. Batch the fix.

3. Segment by dollar value

Sort descending. The top 20% of claims by dollar value almost always represent 60–70% of recoverable revenue in an aged bucket. Work those first. A $28,000 residential claim gets your senior biller; a $340 UA claim gets a batch appeal or a write-off decision.

4. Segment by timely filing status

Flag every claim within 30 days of its filing or appeal deadline. Those move to the top of the queue regardless of dollar value, because a claim that expires tomorrow is worth zero tomorrow.

Once the report is segmented, you have a real work plan instead of a wall of numbers.

How do you actually recover 90 and 120+ day claims?

Here’s the sequence that works. It’s phone calls, portals, and paper — but it’s how money comes back.

Step 1: Verify claim status directly with the payer

Don’t trust your billing system’s last status update. Log into the payer portal or call. You’re looking for one of four answers: paid (and posted incorrectly on your side), pending (with a specific reason), denied (with a specific code), or no record of the claim. Each answer routes to a different action.

Step 2: If “no record,” resubmit — but check timely filing first

Claims fall out of clearinghouses more often than people realize. If the payer has no record and you’re still inside the timely filing window, resubmit with proof of original submission (clearinghouse acknowledgment reports). If you’re outside the window, appeal with proof of timely submission — payers will honor this if you have the acknowledgment.

Step 3: If denied, work the appeal, not another rebill

Rebilling a denied claim without addressing the denial reason just resets the clock and often locks you out of appeal rights. Read the denial. If it’s authorization, request a retro-auth with clinical documentation. If it’s medical necessity, appeal with the clinical record and — where relevant — ASAM criteria documentation supporting the level of care.

Step 4: For records requests, respond with a cover letter

Don’t just fax the chart. Include a cover letter that references the claim number, DOS, patient, and exactly what documents are attached and why they support the billed level of care. Payers process organized submissions faster than data dumps.

Step 5: Escalate stuck claims

If a claim has been “in review” for 45+ days past normal turnaround, escalate. Ask for a supervisor. Reference the state prompt-pay statute if applicable (most states have one for commercial claims, typically 30–45 days). File a complaint with the state department of insurance if the payer is out of compliance. This works more often than operators expect.

Step 6: Negotiate on out-of-network aged claims

Some commercial payers will settle aged out-of-network claims at a discount rather than continue appeals. If the alternative is a write-off, a 60% settlement is real money. This is where contracting experience matters — knowing what a payer will actually agree to versus what they’ll stall on.

When should a behavioral health facility write a claim off?

Write-offs are a business decision, not a filing status. A claim gets written off when the cost of continuing to work it exceeds the probability-weighted recovery. Some defensible defaults:

  • Past timely filing with no proof of original submission: write off.
  • Past appeal deadline with no pending action: write off.
  • Under $200 with a denial that requires clinical review: usually write off — the labor cost exceeds the claim.
  • Denied for non-covered service with clear policy language: write off, and update your verification of benefits process so it doesn’t happen again.

Every write-off should be coded with a reason. If you’re writing off $80,000 a quarter and 40% of it is “authorization not obtained,” that’s a UR problem, not an AR problem — and it will keep bleeding until upstream gets fixed.

How do you keep aged AR from rebuilding after cleanup?

The cleanup is the easy part. Keeping the report clean is the hard part, because aged AR is a symptom of upstream breakdowns — VOB, authorizations, concurrent review, coding, and posting. If those don’t tighten, you’ll be doing another cleanup in six months.

The controls that actually hold:

  • A 30-day touch rule. Every unpaid claim gets a documented action every 30 days. No exceptions. If it hasn’t been touched in 30 days, it goes on a daily escalation list.
  • Weekly aging review with the biller and the clinical/UR lead in the same room. Most aged AR problems are cross-functional. Solving them in a billing silo doesn’t work.
  • Denial coding at the claim level. Not a summary. Every denial gets a reason code so trends show up.
  • A concurrent review process that survives staff turnover. If the only person who knows the auth status is out sick, you have a process problem.

This is where handling the full stack in-house pays off. A denial that traces back to a missed concurrent review isn’t going to get solved by a biller who doesn’t talk to the UR team. At Global AHS, utilization review and billing sit under one roof — the person appealing a medical-necessity denial talks directly to the person who did the concurrent review, not a ticket in a black box.

If you’re staring at an aging report and you’re not sure what’s recoverable, a structured audit will tell you. Global AHS offers a free six-month billing audit that quantifies which aged claims are still workable, which are dead, and where the upstream leaks are. No commitment required to get the number — book the audit here.

Frequently Asked Questions

What’s the timely filing window for most behavioral health claims?

It varies by payer. Commercial plans typically allow 90 to 180 days from date of service; Medicaid MCOs range from 95 to 365 days depending on state and plan; Medicare is 12 months. Appeal windows are separate and usually shorter — often 60 to 180 days from the EOB date. Always check the specific payer’s provider manual, because filing and appeal deadlines are the two clocks that kill aged claims most often.

Should we work the oldest claims first or the highest-dollar claims first?

Highest-dollar with the best recoverability signal, unless a claim is about to hit a filing or appeal deadline. Age alone doesn’t predict payment probability — a clean $22,000 residential claim at 110 days is often more recoverable than a $600 claim at 65 days with a medical-necessity denial. Sort by dollar value within each denial-reason segment and work top-down.

Can we still get paid on a claim past timely filing?

Sometimes. If you have clearinghouse acknowledgment reports showing the original submission was inside the window, most payers will honor it with a proof-of-timely-filing appeal. If the claim genuinely was never submitted in time, recovery is unlikely on commercial and Medicaid — though some payers have hardship or extenuating-circumstance exceptions worth pursuing on high-dollar claims.

How often should aged AR be reviewed?

Working AR is a daily activity; formal aging review should happen weekly with billing, UR, and clinical leadership in the same conversation. A monthly cadence is too slow — claims move across aging buckets every 30 days, and by the time you notice a trend at 90 days, another cohort is already piling up behind it.

What denial codes should we prioritize when working an aged bucket?

Prioritize denials that are correctable and appealable: CO-197 (authorization missing/invalid), CO-50 (non-covered, medical necessity), CO-16 (missing information), and CO-11 (diagnosis inconsistent with procedure). These typically have real recovery paths. Denials like CO-29 (past timely filing) or CO-27 (coverage terminated) usually mean write-off unless you have documentation that overrides the denial reason.

Not sure where your billing is leaking?

Global AHS will audit your last 6 months of billing for free. We pull denials, aged AR, timely filing misses, undercoded services, and underpaid claims, then hand you a written report showing the exact gaps and what they’re costing you. No commitment, no sales pressure — just your numbers, laid bare.

Request your free 6-month audit →