Growth has flattened — behavioral health

Census is flat while admissions hold

PRESENTATION — WHAT THE OPERATOR SEES

Admissions are steady, marketing spend is up, and average daily census has not moved in three quarters. The bed board looks full most weeks, the waitlist is real, and nobody can say whether the constraint is demand, length of stay, or the building itself. Before diagnosing anything: establish which business is presenting, because multi-site behavioral health marketing fails on taxonomy before it fails on spend. Residential, PHP/IOP and outpatient share a name and almost nothing else — and the sharpest line in the federal data is not residential versus outpatient, it is OTP versus non-OTP.

DIFFERENTIAL — LIKELY CAUSES, MOST LIKELY FIRST

01
The building is already full — census is capacity-bound, not demand-bound

Once designated-bed utilization sits in the 90s, an additional admission converts to a waitlist entry, a denial, or a referral out. It does not convert to census. Demand purchased into a full unit changes cost and nothing else.

US substance use residential (non-hospital) facilities ran 85.1% utilization on 91,875 designated beds at 29 March 2024, 32 designated beds per facility. Of 2,852 facilities reporting beds, 878 ran 91–100% and 253 ran above 100%. Private for-profit ran 83.4%; private non-profit 85.5%. Mental health inpatient ran 97.0%.

SAMHSA, N-SUMHSS 2024 Annual Detailed Tables, Tables SU15 and MH6/MH13

NOT THIS IF — Utilization is at or below 50% — 482 residential facilities (16.9%) sit there, and for them the constraint is upstream of the building, at intake.
02
Census is moving with length of stay, not with admissions

Average daily census is admissions multiplied by average length of stay, divided by days. A half-day change in ALOS moves census further than a quarter of media does — and ALOS is set by utilization review, authorization throughput and discharge planning, not by acquisition.

Acadia Healthcare FY2025 same-facility: admissions +2.3%, patient days +2.1%, average length of stay −0.2%, revenue per patient day +2.8%, revenue +4.9%. Acadia's own definition, verbatim: "Average length of stay is defined as patient days divided by admissions." UHS reported behavioral health ALOS of 13.7 days in 2025 and 13.6 in 2024, with occupancy flat at 73% on average available beds in both years and average daily census of 17,575.5 on 23,987 available beds.

Acadia Healthcare FY2025 Form 10-K; Universal Health Services FY2025 Form 10-K, Behavioral Health Care Services operating statistics

NOT THIS IF — ALOS sits at or above the TEDS median for your level of care and has been stable four quarters running, while occupancy sits well below the national band.
03
The continuum leaks — completed episodes step down out of your census

A quarter of all SUD discharges transfer to further treatment. If you do not own the next level of care, every successful residential episode hands its census, and its alumni, to another operator. The step-down is this vertical's actual retention mechanism.

TEDS 2023 discharges by reason, of 1,474,025 with a reported reason: completed treatment 42.6% (n=627,897), transferred to further treatment 25.1% (n=370,294), dropped out 22.0% (n=324,461), terminated 4.1%, incarcerated 1.1%. Separately, only 544 dedicated partial hospitalization / day treatment facilities exist in the entire country — PHP is overwhelmingly a program inside a larger facility, not a standalone business.

SAMHSA TEDS 2023; SAMHSA N-SUMHSS 2024 Annual Detailed Tables, Table MH11 (facility-count column only)

NOT THIS IF — You run detox through outpatient under one licence and already report the internal transfer rate between your own levels of care.
04
You are at the top of your segment's distribution and calling it a plateau

Residential SUD has almost no top. Growth past a certain census is not a campaign problem — it is a second site, a second level of care, or a licence application.

Facility size by clients on census, 29 March 2024: residential SUD (n=3,333) — under 15 clients 50.4%, 15–29 26.0%, 30–59 14.9%, 60–119 6.6%, 120 or more 2.2% (73 facilities nationally). Hospital inpatient SUD (n=1,138) — under 15 clients 63.0%, 120 or more 2.1% (24 facilities). Outpatient SUD (n=12,091) is barbelled: 27.3% under 15 and 27.7% at 120 or more.

SAMHSA, N-SUMHSS 2024 Annual Detailed Tables, Tables SU14a / SU14b / SU14c (row percentages verified to sum to 100.0)

NOT THIS IF — You operate an opioid treatment program. Median clients per SUD facility with an OTP is 185 against 30 without one — a different business with a different ceiling, and that split matters more than residential versus outpatient.
05
Supply grew alongside the client pool in your catchment

Flat census against a growing category is a share problem. Share problems are solved with positioning and referral relationships, not with a bigger budget.

Clients in substance use treatment on the reference date, against facilities surveyed: 1,515,007 across 14,010 facilities (2021), 1,623,647 across 14,854 (2022), 1,592,193 across 14,620 (2023), 1,660,240 across 15,953 (2024). Facility count rose by 1,943 over the period.

SAMHSA, N-SUMHSS 2024 Annual Detailed Tables, Table SU45a

NOT THIS IF — No new licensed capacity has opened at your level of care in your catchment — check the SAMHSA National Directory before assuming it has.

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 · The building is already full — census is capacity-bound, not demand-bound

Designated-bed utilization, built the way SAMHSA builds it: clients on census divided by designated beds, on one reference day, computed separately per unit and per level of care. Pick licensed, available or designated beds and stay in that basis — UHS reports occupancy on two bases and they differ.

CONFIRMS IF

You sit in the 91–100% band or above with a live waitlist. 878 of 2,852 residential facilities sit in that band and 253 above 100%. There is no census left to buy.

EXCLUDES IF

You sit at or below 50%, where 482 residential facilities (16.9%) sit. Empty beds mean the loss is happening before anyone reaches the door.

02 · Census is moving with length of stay, not with admissions

Decompose census: patient days divided by admissions equals ALOS, per level of care, against TEDS 2023 medians — hospital detox 4 days, free-standing residential detox 5, residential/rehab in a hospital 7, short-term residential 21, MAOT residential 25, long-term residential 36, IOP 40, non-intensive outpatient 53. Then multiply the ALOS deficit by admissions and compare to the census gap.

CONFIRMS IF

ALOS deficit × admissions ÷ days accounts for most of the census gap. The lever is utilization review, authorization throughput and unused days — not media.

EXCLUDES IF

ALOS meets or beats the TEDS median for the level of care and the arithmetic leaves the gap unexplained. Note TEDS means vastly exceed medians in every ambulatory category; always test against the median, and always name the level of care.

03 · The continuum leaks — completed episodes step down out of your census

Internal transfer rate: of episodes completing your highest level of care in the last 12 months, the share admitted into your own next level of care within 30 days — and the share of those who left with no scheduled step-down at all.

CONFIRMS IF

Materially below the 25.1% national transfer share, and completions mostly exit the organization. Every one of those was a future admission you already paid for.

EXCLUDES IF

Your internal step-down rate is at or above the national transfer share and the discharges that exit are documented completions with a scheduled external handoff.

04 · You are at the top of your segment's distribution and calling it a plateau

Count clients on census on a single reference day and place the number in the SAMHSA size bands for your setting.

CONFIRMS IF

You land in the 60–119 band (6.6% of residential facilities) or the 120+ band (2.2%, 73 facilities in the country). The next unit of growth is a second site or a second level of care, and no campaign substitutes for either.

EXCLUDES IF

You are under 15 clients, where 50.4% of residential facilities sit. There is real headroom inside the licence you already hold.

05 · Supply grew alongside the client pool in your catchment

Count licensed facilities at your level of care in your catchment in the current SAMHSA National Directory, and against the prior edition.

CONFIRMS IF

New licensed capacity opened at your level of care while your census stayed flat — a share loss, not a demand loss.

EXCLUDES IF

Capacity in the catchment is unchanged or down while your census is flat. Then the problem is inside your own front door.

WHAT RESOLVES EACH

What resolves this, and how you will know it resolved

The building is already full — census is capacity-bound, not demand-bound	Rx 03 · patient conversion →	Same-day capacity matching and a disciplined bed board. What does not work, and we would not sell it: more spend. At 91–100% utilization the marginal inquiry buys a waitlist entry. The paying work is the hold protocol, the discharge-day handoff and the readmission path.
Census is moving with length of stay, not with admissions	Rx 04 · marketing attribution →	Length of stay is a clinical and utilization-review outcome. ADMEN does not set it and no marketing programme should claim to. What we instrument is ALOS, authorized-versus-billed bed days (unused days) and AMA rate alongside admissions, so census movement is attributable to the right lever.
The continuum leaks — completed episodes step down out of your census	Rx 03 · patient conversion →	Routing between your own levels of care, scheduled before discharge rather than after. Where the next level of care does not exist in-house, the honest answer is a referral agreement, not a campaign.
You are at the top of your segment's distribution and calling it a plateau	Rx 04 · marketing attribution →	First job is knowing where you sit in the national distribution. If the answer is the top band, this is a capital and licensing decision — de novo, acquisition or a new level of care — and marketing is not the constraint. Saying so is cheaper than a year of spend.
Supply grew alongside the client pool in your catchment	Rx 01 · patient acquisition →	Demand mapping by catchment and level of care, paired with Rx 02 brand-and-site so the continuum you actually run is legible to a family comparing three programmes in one evening.

WHAT "RESOLVED" LOOKS LIKE — Designated-bed utilization, SUD residential (non-hospital), on the SAMHSA construction

MEDIAN

85.1% national; private for-profit 83.4%, private non-profit 85.5%

TOP DECILE

253 of 2,852 facilities reporting beds (8.9%) run above 100% designated-bed utilization — possible only because clients also occupy non-designated beds, per SAMHSA's own footnote. A further 878 run 91–100%.

TARGET

Sustained utilization in the 91–100% band with ALOS at or above the TEDS median for your level of care. That is a distribution position, not a promise, and above 100% is not a target — it is a bed-count problem. For mental health inpatient the comparable national figure is 97.0% and the band is far tighter.

SAMHSA N-SUMHSS 2024, Tables SU15, MH6/MH13; SAMHSA TEDS 2023, Table E-4

HOW THIS DIFFERS BY SCALE

How this differs by scale

Single site	One site, one level of care: the ceiling is the bed count, and LOS plus discharge planning move census before demand does. A census of 28 puts you in the 15–29 band with 50.4% of residential facilities below you — that is not a plateau, it is the shape of the segment.
Group	Three to ten sites: the plateau is usually one or two sites dragging the group mean. Report utilization per site and per level of care before touching group spend — SAMHSA's own distribution shows 16.9% of residential facilities at or below 50% utilization sitting in the same national market as the 8.9% above 100%.
Platform	Platform scale: the public comparables show what flat looks like at size — UHS held 73% occupancy in both 2025 and 2024 with ALOS at 13.7 and 13.6 days, while Acadia grew same-facility patient days 2.1% and revenue per patient day 2.8%. At this scale census growth comes from beds added and levels of care added: Acadia added 1,089 beds in 2025, 778 of them in six new facilities.

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 →