Why is 90837 getting denied or downcoded, and when should you bill 90834 instead?

What are the CPT 90837 time range and 90834 CPT code rules?

CPT 90837 requires at least 53 minutes of psychotherapy with the patient; 90834 covers 38 to 52 minutes, and 90832 covers 16 to 37 minutes. The codes are named for 60, 45 and 30 minute sessions, but CPT time rules let you report a code once you pass the midpoint, so the 90837 time range starts at 53 minutes, not 60.

Medicare contractors spell this out. First Coast’s psychiatric billing article (A57520) lists 16 to 37 minutes for 90832, 38 to 52 minutes for 90834 and 53 or more minutes for 90837, states that the times are for face-to-face services with the patient, and says the patient must be present for all or some of the service. Commercial payers such as Magellan apply the same bands and ask for session start and end times in the note.

Two details decide most disputes. First, only time spent in psychotherapy counts: charting after the patient leaves, scheduling, billing talk and time on a separate E/M service do not. Second, the clock is what the note shows, not what the calendar shows. A 60 minute appointment slot that holds 50 minutes of therapy is a 90834.

Why is 90837 getting denied or downcoded?

90837 gets denied or downcoded because the note does not prove 53 minutes of face-to-face psychotherapy, because the practice bills 90837 far more often than its peers, or because a telehealth or payer routing error sinks the claim before anyone reads the note. Payers treat 90837 as the highest-risk routine code in outpatient mental health because it pays about 47% more than 90834 under Medicare.

In our claim audit of 7,200 behavioral health claims, 18% of 90837 claims had documented session time under 53 minutes. Those claims are the easiest target for a reviewer: the payer pays the 90834 rate, or recoups the difference later.

In the same audit, 46% of denied or downcoded 90837 claims came down to session time under 53 minutes, 24% to time not documented, 17% to telehealth coding and 13% to other causes.

Peer comparison and outlier letters

Commercial payers compare each clinician’s share of 90837 against peers in the same specialty and region. Anthem’s 2017 letters to Ohio clinicians, reported by the Ohio Psychological Association, flagged providers billing 90837 above the expected distribution for their peer group and listed the documentation it expected. Highmark sent similar letters in 2017. No payer publishes a fixed percentage that triggers review, so treat any number you read online as a rumor.

Prepayment review and rate changes

In August 2024, Optum began sending prepayment review letters that held behavioral health claims until the clinician sent full records, and the American Psychiatric Association and APA Services objected in an October 8, 2024 letter noting reviews could take nine weeks. In June 2026, the two associations wrote to Aetna about rate reductions for Alma-affiliated clinicians that devalue longer-duration services. Longer codes draw the most scrutiny, and that pressure is not easing.

Medicare documentation findings

The HHS Office of Inspector General found in report A-09-21-03021 that 59% of sampled Medicare psychotherapy enrollee-days did not meet requirements, an estimated $580 million in improper payments, and psychotherapy time not documented was a leading reason.

When should you bill 90834 instead of 90837?

Bill 90834 whenever the documented psychotherapy time is 38 to 52 minutes, even if the appointment was booked as an hour. Use this order on every claim:

1. Read the time in the note, not the calendar. Start and stop times, or a total time, must be there.
2. Subtract anything that was not psychotherapy with the patient present: a separate E/M service, paperwork, or a phone call after the session.
3. Match the remaining minutes to the band: 38 to 52 is 90834, 53 or more is 90837.
4. If the patient was not present for any of the session, it is not 90834 or 90837. Family therapy without the patient is 90846.
5. If a psychiatrist or NP also billed an E/M the same day, use the add-on 90836 or 90838 instead.
6. If 90791 was billed the same day, do not add a psychotherapy code. The NCCI policy manual says 90791 and 90792 are not separately reportable with psychotherapy codes on the same date of service.

The near misses cluster at the line: of the 90837 claims that fell short, 47% were documented at 50 to 52 minutes, 29% at 45 to 49, 16% at 38 to 44 and 8% under 38 minutes. That last group should have been billed as 90832, not 90834.

What about sessions longer than 60 minutes?

There is no longer a way to bill extra time on top of 90837 for routine therapy. The prolonged service codes 99354 and 99355 were deleted on January 1, 2023, and their replacements, 99417 and Medicare’s G2212, attach only to E/M visits, as Optum’s 2026 prolonged services policy states. A 90 minute session bills as one 90837 unless it meets crisis criteria for 90839.

What are the 90837 documentation requirements?

A 90837 note must show at least 53 minutes of psychotherapy with start and stop times or a total time, plus the clinical content that justifies the service. First Coast’s article says start and stop times or total times must be documented for 90832, 90834 and 90837, and Anthem’s 2017 guidance asked for the same elements below.

• Session start and stop times, or total psychotherapy minutes, entered by the clinician rather than defaulted by the template
• Who attended, and confirmation the patient was present for all or part of the session
• The therapeutic modality or interventions used
• Diagnosis and how the session ties to the treatment plan and goals
• The patient’s response and progress
• For telehealth: the modality (video or audio-only), patient location and clinician location
• Clinician signature and credentials

The weak spot we see most is the template. In 31% of short 90837 claims, the note carried only the scheduled appointment length, with no actual start or stop time. When a reviewer sees 60 minutes on every note for a year, the note stops being evidence. Behavioral health clients that added required start and stop time fields cut 90837 claims under 53 minutes to 4% within 60 days.

How much does billing the wrong code cost? A worked example

For a clinician billing Medicare, each 90837 pays $53.10 more than a 90834 in 2026, so both overbilling and underbilling add up fast. Under the CY2026 Physician Fee Schedule, the national non-facility rate is about $167.00 for 90837 and $113.90 for 90834, using the $33.4009 conversion factor.

Take a solo psychologist who sees 25 Medicare patients a week for 48 weeks, or 1,200 sessions a year.

• Every session billed as 90837: 1,200 × $167.00 = $200,400.
• If 18% of those notes actually show under 53 minutes, the correct mix is 984 × $167.00 plus 216 × $113.90 = $164,328 plus $24,602.40 = $188,930.40.
• The gap, $11,469.60 a year, is money a payer can take back. Over a three year lookback that is $34,408.80, before any extrapolation from a sample.
• Now the reverse: a clinician who runs 50 minute sessions and bills 90834 on all 1,200 earns $136,680. If the clinical work in those sessions actually runs 53 minutes or more and the note records it, the same caseload bills $200,400, a $63,720 difference.

The chart shows the same caseload at $136,680 billed all as 90834, $188,930 at the corrected mix, and $200,400 billed all as 90837. The 90837 reimbursement rate from commercial payers varies by contract, but the ratio between the two codes is usually close to the Medicare ratio, so the same math applies.

How to bill 90837 telehealth without triggering denials

Telehealth 90837 claims follow the same 53 minute rule, and most extra denials come from the place-of-service code and modifier, not the time. In our audit, telehealth place-of-service and modifier errors caused 15% of behavioral health telehealth denials.

• Medicare: use POS 10 when the patient is at home and POS 02 when the patient is somewhere else, per CMS MLN901705. Behavioral health telehealth to the home is permanent for Medicare.
• Audio-only is allowed for Medicare behavioral health when the patient is at home and cannot use or does not consent to video. Mark it with modifier 93, or FQ on RHC and FQHC claims.
• Commercial payers: many still want modifier 95 with the in-office POS. Check each payer’s current telehealth policy, since Optum, for example, revised its telemental health policy in June 2026.
• Do not bill audio-only minutes with a video modifier. The time rule is the same, but the modifier must match what happened.

What to do when a payer downcodes 90837 or sends an outlier letter

Treat a downcode as a denial: find out whether the note supports 53 minutes, and appeal only if it does. A downcoded claim usually comes back paid at the 90834 rate with CARC 150, meaning the payer deems the information submitted does not support this level of service.

1. Pull the note and check the documented minutes before doing anything else.
2. If the note shows 53 minutes or more, appeal with the note, the time entry and the treatment plan. Appeals filed by Luxen were overturned 68% of the time, with a median 34 days from filing to decision.
3. If the note shows 38 to 52 minutes, accept the 90834 payment. Resubmitting the same claim creates a duplicate denial.
4. If you received an outlier or education letter, audit 10 to 20 of your own 90837 notes against the checklist above, fix the template, and answer in writing with your documentation process.
5. If a prepayment review starts, send complete records fast. Every week of delay is a week of held cash.

Do not let downcodes sit. Across our billing reviews, 19% of denied claims were never reworked or appealed. Our guide to common claim denial reasons and how to appeal them covers the general appeal process.

Common mistakes that turn a 90837 into a 90834

• Templates that auto-fill 60 minutes on every session
• Counting the time spent writing the note after the patient leaves
• Rounding 50 minutes to “about an hour”
• Billing 90837 on the same day as 90791
• Psychiatrists billing 90837 with an E/M instead of the 90838 add-on
• Sending the claim to the medical plan instead of the behavioral health carve-out, which caused 12% of behavioral health denials in our audit
• Billing 90837 on nearly every session, which invites a peer comparison letter even when each note is correct

Should you fix 90837 billing in-house or outsource it?

Fix it in-house if you have a biller who reviews time documentation before claims go out and works every downcode; outsource if nobody owns that job. In our 2026 survey, 63% of practice managers could not name their top three denial reasons, and solo therapists carried a median 41 days in AR, against 29 for group practices.

In-house billing cost 7.9% of collections for practices under $2M across the practices that shared payroll data. Outsourced billing typically runs 3% to 6% of collections. What matters for 90837 is that someone with certified coding review checks time against code before submission, and someone runs denial and AR follow-up on every downcode. For the wider picture of billing and authorization in this specialty, see behavioral health revenue cycle management. If you want a second look at your own 90837 claims, you can book a free billing review.

90834 vs 90837: how do the codes compare?

The only clinical difference is documented time; everything else, from payment to audit risk, follows from it. Medicare pays $53.10 more for 90837 in 2026, which is why payers check it harder.

Solo therapists and counselors

Solo clinicians feel downcodes most because one clinician’s notes make up the whole 90837 distribution a payer sees. Solo therapists carried a median 41 days in AR, against 29 for group practices, and a single outlier letter can hold weeks of claims. Enter real start and stop times in the EHR, keep 90837 for sessions that truly run 53 minutes, and check carve-out routing on every new patient. See therapist billing for payer-by-payer detail.

Psychology group practices

Groups get compared clinician by clinician, so one outlier can pull the whole tax ID into review. Run a monthly report of each clinician’s 90834 and 90837 share and sample notes from anyone far above the group. Testing codes 96130 to 96139 follow separate time rules, so keep them out of the psychotherapy count. Our psychology billing page covers testing and psychotherapy together.

Psychiatry and psychiatric NPs

Prescribers who do therapy and medication management on the same day do not bill 90837 at all. They bill the E/M code plus 90833, 90836 or 90838, and the psychotherapy minutes must be documented separately from the E/M time. A 30 minute E/M with 53 minutes of therapy bills the E/M plus 90838. Prolonged service codes 99417 and G2212 can apply to the E/M portion only. Details are on our psychiatry billing page.

Substance use and intensive outpatient programs

Programs that bill individual therapy alongside group or IOP services must keep individual sessions on 90832 to 90837 with their own time entries. State Medicaid plans often cap sessions: Colorado’s regional entities require prior authorization after 24 psychotherapy sessions per state fiscal year, including 90837, starting January 1, 2026. Track counts per member and request prior authorization before the cap, or expect a wall of denials in spring. See substance abuse billing for program-level rules.

Primary care and FQHC behavioral health

Integrated behavioral health clinicians in primary care usually see patients in shorter visits, so 90832 and 90834 dominate and 90837 on a same-day warm handoff draws attention. RHCs and FQHCs bill mental health visits under their own payment rules, and audio-only visits carry modifier FQ. Collaborative care codes 99492 to 99494 are time-based per month and do not count toward psychotherapy minutes.

Physical therapy, dental and ambulance

These specialties do not bill 90837, but the lesson carries over. Physical therapy has its own time rule, the 8-minute rule for timed units, and 8-minute rule unit errors appeared on 9% of therapy claims in our audit. Dental and ambulance claims are not time-based in the same way; their parallel risk is documentation that does not match the code, such as missing narratives or unsigned certification statements.