Cost per new patient is climbing — behavioral health

Cost per admission is climbing

PRESENTATION — WHAT THE OPERATOR SEES

Inquiry volume is holding or up, VOBs are getting run, and cost per admission has climbed for three quarters straight. The admissions team says the calls are worse. Marketing says the calls are the same. Nobody has the funnel from inquiry to VOB to authorization to admit in one place, so both are arguing from memory. Say this plainly before anything else: no institution in behavioral health publishes cost per admission, and none publishes marketing spend as a share of net revenue. Any industry average you have been shown is not from a source.

DIFFERENTIAL — LIKELY CAUSES, MOST LIKELY FIRST

01
You are paying for inquiries that arrive on days you cannot admit, and you have nothing to hold them with

Cost per admission rises mechanically with occupancy unless there is a structured way to keep a person engaged until a bed or a slot opens. Without one, every inquiry on a full day is a paid inquiry discarded — and the fuller the building, the more of them there are.

Of 15,953 US substance use treatment facilities in 2024, only 8,245 (51.7%) offer "interim services for clients when immediate admission is not possible." Among private for-profit facilities it falls to 2,678 facilities, 39.8% of that class, against 58.3% of private non-profits, 69.8% of local/county/community government facilities and 74.4% of tribal facilities. By contrast 99.4% offer assessment and pre-treatment services — the capability gap is specific to holding someone you cannot admit today.

SAMHSA, N-SUMHSS 2024 Annual Detailed Tables, Table SU17a

NOT THIS IF — You run a documented interim protocol with a contact cadence and can produce a hold-to-admit rate. Then the loss is elsewhere in the funnel.
02
The cost is being incurred at intake, not in media

Media buys inquiries. Intake buys admissions. VOB, authorization and admit are three separate conversions, each with its own failure mode, and when none of them is measured the whole loss gets attributed to the channel that is measured.

Kipu Health — which runs intake software for 3,600+ behavioral health facilities covering 3.8M+ patient episodes as stated 26 January 2026 — puts it directly: "Admissions is the financial nerve center — high-performing organizations are treating the 'front door' as a critical driver of revenue integrity, recognizing that the majority of revenue cycle errors originate at intake." Kipu's own executive KPI set names cost per admission alongside reimbursement rates and claims performance.

Kipu Health, Elevate 2026 findings; Kipu Health installed-base figures per PRNewswire release dated 2026-01-26

NOT THIS IF — You already report inquiry → VOB → authorization → admit weekly by referral source and the steepest drop is at the top of the funnel rather than inside it.
03
You are bidding for the one referral source that was always going to find you, and leaving the others unworked

Self-referral is the largest single source of admissions and the only one paid media competes directly for — which means it is also the one where cost rises fastest as competitors enter. Professional referral sources are admissions demand outside the auction: relationship work with a fixed cost, and most facilities do not staff them.

TEDS 2023 admissions by treatment referral source: self or individual 49.7% (n=642,483), criminal justice/DUI 24.5% (n=315,967), substance use care provider 9.1%, other community referral 8.3%, other health care provider 7.2%, employer/EAP 0.7% (n=9,595), school 0.5%. Separately, only 5,273 of 15,953 SUD facilities (33.1%) report a professional interventionist or educational consultant relationship, and 11,330 (71.0%) do any community outreach at all.

SAMHSA TEDS 2023, Table D-1; SAMHSA N-SUMHSS 2024 Annual Detailed Tables, Table SU17a

NOT THIS IF — You are private-pay residential. TEDS is reported by Single State Agencies and skews to publicly funded treatment — the 24.5% criminal-justice share is far higher and the 0.7% EAP share far lower than a cash-pay programme sees. No source publishes the private-pay referral mix; only your own attribution data can.
04
Readmission is the market and you are buying strangers

Most people entering treatment have been in treatment before. Alumni are the cheapest admission in the vertical and the largest single pool of future ones, and alumni management is an operating function, not sentiment.

62% of participants in the NAATP/FoRSE outcomes cohort had received treatment for a substance use disorder prior to admission; median time since prior treatment was 10 months, mean 2 years 5 months. Payment mix in the same cohort: 46% insurance, 9% self-pay, 45% a combination.

NAATP / FoRSE, Addiction Treatment Providers Outcomes Pilot Program Final Report (2016–2019), 748 participants at 8 member organizations

NOT THIS IF — You already run post-discharge contact on a schedule and can report the share of last year's admissions that came from alumni or alumni referral.

HOW TO TELL THEM APART

How to tell these apart in your own numbers

Each of these is a measurement you can run yourself, without us.

01 · You are paying for inquiries that arrive on days you cannot admit, and you have nothing to hold them with

Count inquiries in one month where the answer was "not today" — no bed, no slot, no authorization yet. Then count how many of those received a scheduled contact within 72 hours, and how many were admitted within 30 days.

CONFIRMS IF

The second number is near zero. You are inside the 60.2% of private for-profit SUD facilities with no interim service at all, and your cost per admission is carrying every one of those discarded inquiries.

EXCLUDES IF

A majority get a scheduled follow-up and a measurable share admit inside 30 days. The interim path exists and is working; look further down the funnel.

02 · The cost is being incurred at intake, not in media

Build the funnel and hold it weekly: inquiries → VOB completed → authorization obtained → admitted, split by referral source and by level of care. There is no external benchmark for VOB-to-admit — no Tier A, B or C source publishes it — so the comparison is your own trailing 12 weeks and the shape of the drop.

CONFIRMS IF

The steepest single drop sits at VOB→authorization or authorization→admit rather than at inquiry→VOB. That is an operations and utilization-review problem wearing a marketing complaint.

EXCLUDES IF

The funnel converts at a stable rate at every internal step and the only thing that moved is inquiry volume or inquiry quality.

03 · You are bidding for the one referral source that was always going to find you, and leaving the others unworked

Attribute every admission in the last 12 months to a named referral source — alumni, interventionist, hospital, EAP, court, self. Then list the sources that produced admissions last year and zero this year, and the ones nobody is assigned to.

CONFIRMS IF

Nearly all admissions trace to self-referral and paid channels, professional sources are unstaffed, and named referrers have gone quiet without anyone noticing.

EXCLUDES IF

Referral sources are named, owned, and contributing, and the cost increase is confined to one paid channel — then it is an auction problem, not a mix problem.

04 · Readmission is the market and you are buying strangers

Of admissions in the last 12 months, the share who had previously been treated by you, plus the share referred by someone who had. Then the share of discharges from 12 months ago with any logged contact since.

CONFIRMS IF

Prior-treatment share sits far below the 62% observed in the NAATP/FoRSE cohort and post-discharge contact is undocumented. The cheapest admissions available to you are unworked.

EXCLUDES IF

Alumni contact runs on a schedule and produces a measurable, attributed share of admissions.

WHAT RESOLVES EACH

What resolves this, and how you will know it resolved

You are paying for inquiries that arrive on days you cannot admit, and you have nothing to hold them with	Rx 03 · patient conversion →	A structured hold for the person who calls on a day you cannot admit: same-day capacity matching, a contact cadence, and a route to a lower level of care or a partner rather than a dropped call. This is the single highest-yield intake fix in the vertical and it is federally documented as absent in six of ten for-profit facilities.
The cost is being incurred at intake, not in media	Rx 04 · marketing attribution →	Instrument the front door before spending against it: inquiry, VOB, authorization, admit, by source and level of care, reconciled weekly. What does not work is optimising media against an admissions number nobody can decompose.
You are bidding for the one referral source that was always going to find you, and leaving the others unworked	Rx 01 · patient acquisition →	Paid channels are correct for self-referral — half of admissions nationally — but they should be priced against the professional sources they are substituting for. Referral-source attribution and field-visit logging are operations work, not media work, and they are where cost per admission actually falls.
Readmission is the market and you are buying strangers	Rx 03 · patient conversion →	Post-discharge contact on a schedule, with readmission and alumni referral tracked as named sources. Beware the claims that circulate here: a widely repeated "over 80% regress within the first year" figure is stated by a vendor with no citation and should not be used to justify the programme. The 62% prior-treatment rate from NAATP/FoRSE is real and is enough.

WHAT "RESOLVED" LOOKS LIKE — There is no published cost-per-admission benchmark in behavioral health. No Tier A, B or C source publishes it and none publishes marketing spend as a share of net revenue, so the honest target is a trend against your own baseline plus one capability that is federally measured: the share of inquiries that receive a structured interim contact when immediate admission is not possible.

MEDIAN

51.7% of US SUD facilities offer interim services at all; 39.8% of private for-profit facilities do

TOP DECILE

No distribution is published. The highest-coverage operator classes are tribal at 74.4% and local/county/community government at 69.8% — that is the ceiling visible in the federal data, not a percentile.

TARGET

Carry the capability at all, above the 51.7% national norm and well above the 39.8% for-profit norm, with a measured hold-to-admit rate. On cost per admission itself the only defensible target is a downward trend against your own measured baseline, decomposed by referral source. The one absolute anchor that exists anywhere is Kipu's observed 71.2 admissions per facility across 174 treatment centers in 2019–2020 — an observation, not a benchmark, and the uplift claim attached to it is vendor marketing.

SAMHSA N-SUMHSS 2024, Table SU17a; Kipu Health admissions study (174 treatment centers), PRNewswire 11 August 2020

HOW THIS DIFFERS BY SCALE

How this differs by scale

Single site	One site: cost per admission is dominated by whether one person owns the call log. The fix is usually a protocol and a calendar, not a budget — and at a median residential census of 14 clients, three recovered inquiries a month is a material swing, which is why cost per admission by level of care is worth reporting even at one site.
Group	Small group: the cost gap between sites is usually the interim protocol and the referral-source ownership, not the media plan. Compare cost per admission only within the same level of care — an OTP at a median 185 clients and a residential programme at 30 are not comparable denominators.
Platform	Platform: cost per admission has to reconcile to the income statement, and the labour line will dominate it. UHS reported salaries, wages and benefits at 54.0% of net revenues in 2025 against 53.6% in 2024, with net revenue per adjusted admission up 7.5%. At platform scale an admissions problem that is really a staffing problem is expensive to misdiagnose.

OTHER PRESENTATIONS — BEHAVIORAL HEALTH

A differential narrows the field. It does not replace the examination — that is what the six weeks are for. Every figure above is an industry reference range, not a client's numbers; those stay sealed. Sources are set out at /sources.

APPLY — 6 / QUARTER →