Home/Case Studies/Therapy Authorization Turnaround
22-clinic PT/OT/SLP group · Midwest · 12 months

Authorization leakage cut from $809K to $85K a year.

Visit-limit tracking that fires before the visit happens, 8-minute rule unit integrity, plan-of-care recerts on a calendar instead of a memory, ABN discipline against the CY 2026 thresholds, and Medicare Advantage prior auth held to the regulatory clock. $724K recovered in 12 months.

Get the same audit → See the leak funnel
Specialty
Outpatient PT/OT/SLP
Clinics
22 locations
Therapists
90 (58 PT · 19 OT · 13 SLP)
Volume
9,800 visits/mo
Geography
Midwest · 4 states
Recovered
$724K in 12 mo

The headline numberPrior auth on the payer's habit. Then on the clock.

BEFORE · BASELINE
12.3d
Median Medicare Advantage prior-auth turnaround · evals held, plans of care idling, episodes abandoned
AFTER · STEADY STATE
4.1d
Decision-ready packets on day zero · escalation on the CMS-0057-F clock · care starts the day approval posts
Recovered
$724K
12-month recapture of a $809K annualized leak
Auth denial rate
0.4%
Of 117,600 annual visits, down from 3.2%
Units per visit
2.95
Documented and compliant, up from 2.89
Recerts on time
99.2%
Plan-of-care recerts signed inside the 90-day window

The situationFull schedules. Leaking remits. Nobody watching the clock.

TAKEAWAY
The therapists were productive. The leakage lived entirely in the administrative layer between the schedule and the claim: expired auths, rounded-down units, lapsed recerts.
CONTEXT 01
117,600
visits per year · avg $96 net per visit
CONTEXT 02
31% + 24%
Medicare + Medicare Advantage payer mix
CONTEXT 03
5
leak buckets, all authorization-adjacent
Failure 01 · Visit-limit blind spot

The auth expired. The schedule did not know.

Authorizations lived in a spreadsheet updated weekly. The scheduling system booked visit 13 against a 12-visit auth, and the front desk found out from the denial six weeks later. 3.2% of all visits were delivered against an expired or exhausted authorization: 3,760 visits a year, $361K at $96 per visit.

Failure 02 · Per-code rounding

The EMR rounded each CPT separately.

Medicare's 8-minute rule counts total timed minutes per discipline per day. The EMR's charge logic rounded code by code, so mixed-remainder visits quietly dropped a unit. A 2,400-visit audit sample found 6.1% of visits lost one billable unit: 7,170 visits a year, roughly $32 a unit, $229K.

Failure 03 · Recert drift

Day 90 arrived. No signature.

Medicare plan-of-care certifications cover up to 90 calendar days. Recert requests went out when a biller noticed, and only 71% came back signed in time. Visits delivered after the lapse denied. Add ABN gaps above the KX threshold and Medicare Advantage prior auths idling 12.3 days, and the ledger reached $809K a year.

The leak funnelFive gates between a scheduled visit and a paid claim.

TAKEAWAY
Every dollar below is annualized baseline, measured from twelve months of 835 data. The five side-leaks sum to $809K. That is the number the 90-day plan attacked.
117,600 SCHEDULED VISITS / YR $11.29M gross expected at $96/visit GATE 1 · AUTH VALID + VISITS REMAINING Expired window or exhausted count = denial LEAK $361K 3,760 visits × $96 GATE 2 · 8-MINUTE RULE UNITS CORRECT Per-code rounding drops the mixed-remainder unit LEAK $229K 7,170 units × $32 GATE 3 · PLAN-OF-CARE CERT CURRENT Visit after day-90 lapse denies on the spot LEAK $49K 510 visits × $96 GATE 4 · ABN / KX DISCIPLINE No ABN above threshold = provider write-off LEAK $33K 380 claims × $88 GATE 5 · MA PRIOR AUTH ON TIME Slow approval = shortened or abandoned episodes LEAK $137K 340 episodes × 4.2 × $96 TOTAL LEAK $809K PER YEAR · BASELINE

Sum check: $361K + $229K + $49K + $33K + $137K = $809K annualized. Post-fix residual runs $85K a year, which is where the $724K twelve-month recovery comes from.

Exhibit · unit mathThe 8-minute rule, done as arithmetic.

TAKEAWAY
Medicare counts total timed minutes per discipline per day, then allocates units by largest remainder. EMRs that round each CPT code separately lose the mixed-remainder unit. Here 6.1% of visits lost one.
Total timed minutesBillable 15-min unitsValue at $32/unit
8 to 22 minutes1 unit$32
23 to 37 minutes2 units$64
38 to 52 minutes3 units$96
53 to 67 minutes4 units$128
68 to 82 minutes5 units$160
WRONG · PER-CODE ROUNDING 97110 · 18 min rounds to 1 unit 97112 · 20 min rounds to 1 unit 2 units $64 billed RIGHT · TOTAL TIMED MINUTES 18 + 20 = 38 min 38 sits in 38 to 52 97112 × 2 · 97110 × 1 largest remainder wins 3 units $96 billed DELTA PER VISIT +$32 × 7,170 VISITS = $229K/YR

Unit boundaries per the Medicare timed-code policy referenced on the CMS Therapy Services page (cms.gov, Medicare Claims Processing Manual chapter 5). The fix was a charge-scrub rule comparing billed units against total timed minutes on every visit before claim drop, plus a therapist-facing minutes widget in the EMR.

Exhibit · 2026 thresholdsEvery Medicare patient tracked against $2,480 and $3,000.

TAKEAWAY
For CY 2026, CMS set the KX modifier threshold at $2,480 for PT and SLP combined and $2,480 for OT, with the targeted medical review threshold at $3,000 (source: CMS Therapy Services page, cms.gov, CR 14252). Claims above the KX line without the modifier deny. The tracker makes both lines visible per patient, per discipline.
Medicare therapy threshold tracker · PT+SLP bucket · CY 2026 gold line = $2,480 KX · red line = $3,000 MR
Patient A · PT · knee TKA rehabYTD accrued $1,910 · 77% of KX threshold
$1,910 KX NOT YET REQUIRED
Patient B · PT+SLP · post-CVAYTD accrued $2,608 · above $2,480 · medical necessity documented
$2,608 KX APPENDED
Patient C · PT · chronic LBP maintenanceYTD accrued $2,451 · necessity review flagged before crossing
$2,451 ABN SIGNED · ON FILE
$2,480 KX modifier threshold (CY 2026) $3,000 targeted medical review threshold Track width scaled to $3,000

Baseline state: threshold accrual lived in a month-end report, so ABN conversations happened after the fact and 380 claims a year, averaging $88, became write-offs. The fix moved accrual to claim-time, fired a worklist at 80% of the KX line, and forced one of two exits before crossing: KX with documented necessity, or a signed ABN. ABN-on-file when required went from 46% to 98%.

Exhibit · recert timingNinety days is a calendar problem, not a memory problem.

TAKEAWAY
Medicare plan-of-care certifications cover up to 90 calendar days (Medicare Benefit Policy Manual, chapter 15, section 220). The fix is three dated triggers per episode, none of which depend on anyone remembering anything.
DAY 0 Eval + POC cert signed DAY 60 Recert drafted auto-generated from notes DAY 75 Physician chase fax + portal + call cadence DAY 85 Hard stop unsigned = schedule lock DAY 90 Cert expires visits beyond = denial LINEAR DAY SCALE · DAY 0 AT X=60, DAY 90 AT X=1040 · 10.889 UNITS PER DAY

Under the old workflow the chase started when a biller noticed, which averaged day 82. Under the calendar, the recert is drafted at day 60 from the therapist's own progress notes, the physician cadence starts at day 75, and an unsigned recert locks the schedule at day 85 so no visit can be delivered into a lapse. On-time recerts moved from 71% to 99.2%, and the $49K lapse bucket fell to $5K.

Exhibit · MA prior authHold the payer to the regulatory clock.

Medicare Advantage referrals sat a median 12.3 days between eval order and approved plan of care. Patients cooled off, slots went unfilled, and 340 episodes a year were shortened or abandoned, an average of 4.2 lost visits each: $137K.

Regulatory anchor · CMS-0057-F
The CMS Interoperability and Prior Authorization final rule (CMS-0057-F, cms.gov) requires impacted payers, including Medicare Advantage plans, to send prior-authorization decisions within 72 hours for expedited requests and 7 calendar days for standard requests, with compliance beginning January 1, 2026, and to give a specific reason for any denial.

The operational change: submission packets built decision-ready on day zero (eval, standardized test scores, POC, HEP evidence), expedited criteria applied wherever clinically supported, and an escalation ladder that files on day 8 for any standard request still pending, citing the rule. Median TAT settled at 4.1 days and episode abandonment fell by 90%.

MA prior-auth TAT · by plan · median days baseline vs steady state
MA Plan 1 · HMObaseline
14.2d
14.2
MA Plan 1 · HMOsteady state
4.6d
4.6
MA Plan 2 · PPObaseline
11.8d
11.8
MA Plan 2 · PPOsteady state
3.9d
3.9
MA Plan 3 · regionalbaseline
10.9d
10.9
MA Plan 3 · regionalsteady state
3.4d
3.4
portfolio median 12.3d → 4.1d · bars scaled to 15 days

The ledgerFive buckets. One tie-out.

TAKEAWAY
Annualized baseline leak $809K. Residual after fix $85K. Recovered $724K. Every row reconciles to the 835 and the visit log.
Leak bucketMechanismBaseline / yrResidual / yrRecovered
Expired / exhausted auths3,760 visits at $96 (3.2% of visits) → 470 visits (0.4%)$361K$45K$316K
8-minute rule unit loss7,170 dropped units at $32 (6.1% of visits) → 590 units (0.5%)$229K$19K$210K
POC recert lapses510 post-lapse visits at $96 → 52 visits$49K$5K$44K
ABN / KX write-offs380 claims at $88 avg → 34 claims$33K$3K$30K
MA prior-auth episode churn340 episodes × 4.2 visits × $96 → 32 episodes$137K$13K$124K
Total · 22 clinics117,600 visits/yr · $96 avg net$809K$85K$724K

Rounding: bucket values rounded to the nearest $1K from visit-level detail (for example 3,760 × $96 = $360,960 shown as $361K). Column totals computed on the rounded values shown: 361 + 229 + 49 + 33 + 137 = 809, and 45 + 19 + 5 + 3 + 13 = 85.

We thought we had a denial problem. We had a clock problem. Auth windows, cert windows, threshold lines, payer decision deadlines. Once every clock had an owner and an alarm, the denials mostly stopped happening at all.

VP of revenue cycle · 22-clinic outpatient therapy group

What the dashboard showsEvery active auth. Every clock.

VIEWERS
COO + clinic directors
one shared view, 22 clinics
GREEN
Auth healthy
3+ visits and 14+ days remaining
GOLD
Renew now
2 visits or 10 days left, request filed
RED
Schedule locked
no valid auth, booking blocked
Active authorizations · visits used vs approved refreshed 12s ago
Patient 4482 · PT · rotator cuffClinic 07 · commercial · expires in 31d
97110+
6/12
Patient 5107 · SLP · dysphagiaClinic 12 · MA plan 2 · expires in 24d
92526
3/8
Patient 3961 · OT · CTS post-opClinic 03 · MA plan 1 · renewal filed d-9
97530
10/12
Patient 6230 · PT · TKA rehabClinic 19 · Medicare · KX at 77%
97110+
9/20
Patient 5568 · PT · LBPClinic 09 · commercial · 2 visits left
97140
8/10
Patient 4815 · OT · stroke UEClinic 14 · MA plan 3 · auth lapsed, locked
97530
12/12
Patient 7093 · SLP · pediatric articClinic 21 · Medicaid · expires in 40d
92507
4/16
1,847 active auths · 61 gold · 3 red · zero unauthorized bookings this month
Engagement ScorecardQuarterly partner report · excerpt
CONFIDENTIAL

KPI movement baseline vs current.

Auth-related denial rate (% of visits)
3.2%
0.4%
MA prior-auth median TAT
12.3 d
4.1 d
Billable units per visit (compliant)
2.89
2.95
POC recerts signed on time
71%
99.2%
ABN on file when required
46%
98%
Annualized authorization leak
$809K
$85K
Q2 · 2026 ASP-RCM Senior Partner

ImplementationThe 90-day plan. Four phases, dated artifacts.

TAKEAWAY
Nothing here needs new software licenses. It needs the five clocks wired to the schedule, and a person on the hook for each one. The COO verifies each phase's artifact before the next opens.
01
Days 1-15 · Baseline

Leak ledger from 835s

Twelve months of remits coded into the five buckets. 2,400-visit unit audit sample. Auth log reconciled to the schedule. The $809K number signed off by the CFO.

02
Days 16-45 · Gates 1-2

Auth tracker + unit scrub

Authorization counts wired into scheduling with green/gold/red states and a booking lock. Charge-scrub rule compares billed units to total timed minutes on every visit before claim drop.

03
Days 46-70 · Gates 3-4

Recert calendar + ABN workflow

Day-60 draft, day-75 physician cadence, day-85 hard stop per episode. Claim-time threshold accrual with the 80% worklist against the CY 2026 $2,480 KX line.

04
Days 71-90 · Gate 5

MA prior-auth turnaround

Decision-ready packet template per plan, expedited criteria matrix, day-8 escalation ladder citing CMS-0057-F. Steady-state dashboards handed to clinic directors.

OutcomesBefore. After. In numbers.

Pre-engagement · baseline
Auth-related denial rate
3.2%
MA prior-auth median TAT
12.3d
Recerts signed on time
71%
ABN on file when required
46%
Annualized leak $809K
Steady-state · month 6+
Auth-related denial rate
0.4%
MA prior-auth median TAT
4.1d
Recerts signed on time
99.2%
ABN on file when required
98%
12-month recovery $724K

Common questionsFrequently asked: therapy authorization.

What counts as authorization-driven leakage in outpatient therapy?
Five buckets, all measurable from the 835 and the schedule: visits delivered after an authorization expired or the visit count exhausted; timed units lost to per-code 8-minute rule rounding; visits delivered after the Medicare plan-of-care certification lapsed; Medicare balances written off because no ABN was on file when services ran past the KX threshold without supporting medical necessity; and Medicare Advantage episodes shortened or abandoned while the prior authorization sat in a payer queue. For this 22-clinic group those five buckets totaled $809K on an annualized baseline.
How does the 8-minute rule drop billable units?
Medicare pays timed therapy codes in 15-minute units using total timed minutes per discipline per day: 8 to 22 minutes is 1 unit, 23 to 37 is 2, 38 to 52 is 3, 53 to 67 is 4. The common failure is EMR logic that rounds each CPT code separately. Example: 18 minutes of 97110 plus 20 minutes of 97112 is 38 total timed minutes, which supports 3 units. Per-code rounding bills 1 plus 1, so a unit worth roughly $32 vanishes. At this group that pattern touched 6.1% of visits.
What are the 2026 Medicare therapy threshold amounts?
For CY 2026, CMS set the KX modifier threshold at $2,480 for PT and SLP services combined, and a separate $2,480 for OT services. The targeted medical review threshold remains $3,000 through 2028. Both figures are published on the CMS Therapy Services page at cms.gov (CR 14252). Claims above the KX threshold without the KX modifier are denied, so the tracker watches every Medicare patient's accrual against both lines.
How did Medicare Advantage prior-auth turnaround improve?
Two levers. First, the CMS Interoperability and Prior Authorization final rule (CMS-0057-F) requires MA plans to decide standard requests within 7 calendar days and expedited requests within 72 hours beginning January 1, 2026. We built the submission packet so it is decision-ready on day zero and escalate on the regulatory clock, not the payer's habit. Second, evaluations are scheduled with a conditional hold so the plan of care starts the day the approval posts. Median TAT fell from 12.3 days to 4.1 days.
Why anonymize the client?
The master service agreement includes reciprocal confidentiality. The clinic count, therapist mix, visit volume, and dollar figures are real. A senior partner can walk you through methodology and host a reference call under NDA once both sides agree. The same authorization discipline runs inside our physical therapy billing, occupational therapy billing, and speech and audiology billing programs.
What does the free therapy authorization audit look like?
Send 12 months of 835 remittance data, your authorization log or EMR auth report, the plan-of-care recert report, and your Medicare threshold tracking export if one exists. Inside 30 days you receive a 4-page written audit covering auth-expiry denials by payer, 8-minute rule unit variance from a 300-visit sample, recert lapse exposure, ABN and KX gaps against the CY 2026 thresholds, MA prior-auth TAT by plan, and a 90-day fix plan. Yours to keep.

Want the same leak ledger built for your clinics?

A free 30-day audit. Send 12 months of 835s, your auth log, and your recert report. We return a 4-page written audit: the five leak buckets in dollars, unit variance from a 300-visit sample, ABN and KX gaps against the CY 2026 thresholds, MA prior-auth TAT by plan, and a 90-day fix plan. Yours to keep. No SDR follow-up.

Get my free audit All case studies