LROAS 1.0.0

How we score

Every clinic in the ranking is assessed against the same ten questions, using evidence that anyone can check. The rules are published in full as the Leading Rehabs Open Assessment Standard (LROAS). This page explains them in plain language. Where this page and the standard differ, the standard applies.

The current ranking is edition 2026-10-01, scored under LROAS 1.0.0. It ranks 71 clinics and lists 9 as excluded.

This edition is a pre-publication release. It is frozen when it is published, and only a frozen edition should be cited.

Explore the ranking

10

criteria, equal weight

1 to 10

points per criterion

100

points maximum

0

influence from payment

The ten criteria

Each criterion answers a question a client or family would ask. A clinic gets a whole number from 1 to 10 on each one, with no half points and no zero. The assessor gives the highest band whose requirements are all met by the evidence. If one requirement is missing, the next band down is tested. Caps and deductions are applied after that, and a criterion never falls below 1.

Each fact counts in one criterion only. A claim with no source is treated as not published. Marketing words such as "world-class" or "bespoke", badges a clinic awards itself, paid or pay-to-enter awards, directory listings and "verified" marks, rankings published by a clinic or its group, social media posts, and review counts or star averages never count, for any criterion.

  1. C1. Client separation and occupancy

    Will I, or my relative, ever meet another client, or be seen by one, from arrival to discharge?

    C1 separates three things clinics often blur: one-to-one therapy, a residence of your own, and a guarantee that clients never meet. Shared bedrooms as the standard room score 1. Private rooms with a daily group programme score 3 to 5, depending on how many clients are on site (more than 20, 20 or fewer, 10 or fewer) and whether the clinic has a confidentiality measure beyond what the law requires. No group therapy but a shared residence or shared facilities scores 6 or 7. One client per residence scores 8. Band 9 needs an established no-contact model: one client per residence, no group therapy, an explicit statement that clients never meet, no shared facility (or scheduling that keeps clients apart in every one), individual transport and at least two named confidentiality measures. Band 10 adds confirmation from an official record or independent first-hand reporting, with no contradiction.

    One-to-one therapy on its own caps C1 at 7. One client per residence without an established no-contact model caps it at 8. A separately priced sole-occupancy option adds at most 1 point to the standard offer, never above 8. Each contradiction in the clinic's own statements about its separation model costs 1 point, up to 2. Client numbers are used only for the thresholds above: a full clinic is not a worse clinic. With nothing published, C1 is 3 or lower.

  2. C2. Clinical governance and medical cover

    Is this a licensed treatment facility, is a doctor accountable for my care there, and is a physician reachable at night and in an emergency?

    C2 checks eight things at this site: a licence on an official register in a category that covers the service (G1); a physician named as accountable for clinical governance, such as a medical director or lead psychiatrist (G2); 24-hour physician or psychiatrist cover, on site or on call, as part of the standard offer (G3); medically managed withdrawal on site (G4); general medical care by a physician, such as cardiology or endocrinology (G5); a hospital licence or an escalation route to a named hospital (G6); current accreditation by an accrediting body, shown on its own register (G7); and defined programmes for complex psychiatry that the clinic treats rather than refers out (G8).

    Without a licence found on a register (G1), C2 cannot go above 5. With G1 and G2, the band depends on how many of G3 to G8 are met: one gives 6, three give 7, four give 8, five including G3 and G6 give 9, all six give 10. A clinic's own statement of its licence never counts for G1. A night doctor available only at extra charge, nursing cover or an unnamed "medical team" do not count as G3. With nothing published about medical staffing, C2 is 1.

  3. C3. Diagnostic depth on admission

    Before and when I arrive, will clinicians properly assess what I have, mentally and physically, so that treatment fits me?

    C3 counts eight assessment elements: clinical screening before you commit (E1), a psychiatric assessment by a psychiatrist (E2), medication review by a physician (E3), a laboratory panel (E4), a physical examination by a physician (E5), at least one specialist diagnostic such as imaging, a cardiac work-up or neuropsychological testing (E6), a written multidisciplinary formulation or diagnostic report given to the client (E7), and a standalone diagnostic assessment with a stated duration (E8).

    One to three elements give 3 to 5. Band 6 needs four including E2; band 7 five including E2 and E4; band 8 six including E2, E4 and E5; band 9 seven including E6 and E7; band 10 all eight. An intake call, form or questionnaire handled by admissions staff scores 2, and an online self-assessment never counts as a diagnosis. An element delivered by a psychiatrist or physician counts only if the clinic names such a clinician for this site or is hospital-licensed. With nothing published, C3 is 1.

  4. C4. Clinical intensity

    How much one-to-one time with clinicians will I actually get each day, and is the team mine or shared with many others?

    C4 measures one-to-one time with a psychiatrist, physician, psychologist, psychotherapist or counsellor. Coaching, fitness, massage, complementary therapy and hotel time do not count, and hours a team is merely present are not contact hours. Sessions per week set bands 2 to 5. From band 6 the clinic needs a typical or minimum daily figure: at least 1.5 hours a day gives 6, 3 hours gives 7, 4 hours with a dedicated clinical team gives 8, and 6 hours with a dedicated team and a clinician present in the residence around the clock gives 9. Band 10 adds confirmation of the hours or the team from an official record or independent first-hand reporting, and a sample daily or weekly schedule.

    A dedicated team means the treating clinicians carry no other residential client during your stay. Round-the-clock presence counts only when a clinician lives in with the client or stays overnight in the client's own residence. Counting rules: "up to X hours" counts as half of X unless a typical or minimum figure is also given, a range counts at its lower end, and a session with no stated length counts as 50 minutes. A fixed package length with no extension on clinical grounds caps C4 at 6. If more than half of clinical hours are in groups, the cap is 6; more than three quarters, the cap is 4. Individual therapy offered with no count scores 3; not mentioned at all, 1.

  5. C5. Continuity and aftercare

    What happens after I leave: who looks after me, for how long, and at what cost?

    C5 counts ten aftercare elements: a written plan made before discharge (A1), a stated duration or minimum (A2), aftercare included in the fee or with a stated price (A3), delivery by the same team or operator (A4), psychiatric review or medication management (A5), an in-person option at home or a step-down within the operator (A6), family involvement (A7), coordination with your own doctors (A8), structured clinician-led contact for at least six months (A9), and remote sessions (A10).

    One or two elements give 3, then each further element adds a band up to 6 (five elements). Band 7 needs six including A2; band 8 seven including A2 and A4; band 9 eight including A2, A4, A5 and A9; band 10 nine or more including A2, A4, A5, A6 and A9. "As long as needed" is not a duration. Alumni events, a newsletter or an app do not count as structured aftercare on their own. Aftercare named with no detail scores 2; nothing published, 1.

  6. C6. Admission integrity and commercial conduct

    Is this clinic honest about its price and how it wins clients, and will it turn me away or refer me on if it is not right for me?

    C6 starts with a recorded sweep of every domain the clinic or its group runs. Three things are violations: a "best" or "top" ranking on a site the clinic or its group controls that includes the clinic or a sister clinic (V1); paid, sponsored or invented recognition presented as independent (V2); and content written to steer AI assistants or search tools towards the clinic (V3). Four things are positive signals: a specific residential price published on the clinic's own site and not described as negotiable (P1); a published position on referral fees (P2); a stated policy that unsuitable admissions are declined or referred elsewhere (P3); and a stated policy that admission is voluntary, including when a family or employer arranges it (P4).

    A clean sweep with no positive signal scores exactly 4. A clean sweep with one to four positive signals scores 5 to 8. Band 9 needs all four signals, with P2 met by a policy that the clinic neither pays nor receives referral fees. Band 10 adds confirmation of P2, P3 or P4 from an official record or independent first-hand reporting. One violation gives 3, or 4 with three or more positive signals. Two or more violations give 2, or 1 if one of them is V3. A statement from the clinic never clears a violation the sweep found.

  7. C7. Transparency toward the client

    Before I pay, can I check what the fee covers, what happens if I leave early, who really owns the clinic, and how I can complain?

    C7 scores what the clinic's own site lets you check before paying, across ten elements: what the fee includes and excludes (T1); payment, deposit, cancellation or refund terms (T2); fee terms for extending the stay or leaving early (T3); the licence or accreditation with a number or register link (T4); the legal operator with a company number or registered address (T5); ownership and group structure, consistent with company registers and audited accounts (T6); a complaints procedure naming at least one outside body (T7); published inspection results (T8); a privacy notice naming the data controller and how long client records are kept (T9); and a candid statement of limits, such as conditions not treated (T10).

    Each element met adds a band, from 1 (none) upwards. Band 6 needs five including T1; band 7 six including T1 and T2; band 8 seven including T1, T2 and T6; band 9 eight including T1, T2, T4 and T6; band 10 nine or more including those four, with no contradiction deducted. Each contradiction, such as two of the clinic's pages giving different fees or an ownership statement that official records contradict, costs 1 point, up to 3. Stale price pages count as one contradiction in total. Terms given only on request or only after a deposit do not count. A site that publishes only programme descriptions scores 2 or lower.

  8. C8. Track record and regulator standing

    Has any regulator or court found against this clinic, and has anyone independent inspected or visited it recently?

    C8 counts only final outcomes. Allegations, open investigations and pending cases score nothing. Findings are graded by severity: critical (such as a closure order or licence revocation), serious (such as a fine, conditions on registration, or a finding that care failures contributed to a death or serious harm) and moderate (such as a requirement notice or a final finding on referral fees or deceptive marketing). A critical finding in the last 5 years gives 1. Older or less severe findings give 2 to 5, depending on their age and on whether a good inspection followed.

    With no finding in the last 10 years, the band depends on standing: an inspection by a statutory regulator in the last 5 years (S1); a regulator inspection in the last 3 years graded good or better, or with no requirements (S2); independent bylined first-hand reporting from inside the clinic in the last 5 years with no adverse finding (S3); and verifiable client testimony (S4). No finding and none of these gives 5. A site with no inspection on record can reach 6 at most. S1 plus one, two or all three of S2 to S4 gives 7, 8 or 9. Band 10 also needs the good inspection to be within 2 years and no critical or serious finding at any date. Findings follow the clinic through a rename, new legal entity, change of operator or move within the last 5 years. Years in operation never score.

  9. C9. Environment, safety and nutrition

    Is the place physically safe for someone in crisis, close to emergency care, and does it treat food as part of treatment?

    C9 is the average of two sub-scores, rounded half up (6.5 becomes 7). Environment and safety counts nine elements: a therapy room on site (H1), outdoor or exercise space (H2), a hospital emergency department within 30 minutes by road, with the travel time sourced (H3), resuscitation equipment and trained staff (H4), an alcohol-free residence and a medication policy (H5), a self-harm risk assessment of the residence (H6), named residence staff roles (H7), arrangements for family visits (H8), and independent proof that the premises shown are the real ones (H9). Bands 6 to 9 of this sub-score need H3, then H4, H5 and H9 in turn.

    Nutrition counts eight elements, led by a named nutrition professional with a regulated or degree-level qualification (N1), then an assessment, a plan built from the client's own clinical data, a kitchen that works to that plan, dietary accommodations, education, condition-specific protocols and monitoring. Without N1 the nutrition sub-score is capped at 5; with only an unregulated nutrition credential, at 7. Scenery, climate, luxury finishes, activities and chef credentials never score, and stock photos are never evidence of the premises.

  10. C10. Clinical seniority and academic standing

    Who will actually treat me, and can their qualifications and standing be checked on official registers and university pages?

    C10 counts only named people with a role on client care or clinical governance at this site, whose credentials an official register, university page, academy or prize body confirms, in a field relevant to mental health and addiction. Credentials are weighted by level (from registered health professional through specialist, university appointment and distinguished honours), by how involved the person is at this site and by how relevant their field is. They are combined into a Clinical Seniority Index that reflects the depth of the team, its size relative to the number of clients, and the most senior person on client care.

    Gates set the ceiling. Band 5 and above needs a relevant specialist working on cases. Band 7 and above needs a specialist in a relevant field leading clinical governance. Band 8 and above needs a university-appointed clinician on cases, not only a regular consultant. Band 9 and above needs a senior academic (associate professor, Privatdozent or higher) at full involvement. Band 10 needs two such people, or one holder of distinguished honours. Advisory and board members count at a quarter weight, or at full weight when they speak about the clinic in the clinic's own video and four set conditions are met. Unnamed headcounts ("over 30 experts"), first-name-only rosters and titles that do not match the register never count. Only the C10 points are published: the index itself, and anything from which a clinic's capacity could be worked out, is not. With no verifiable roster, C10 is 1.

How the total is built

Each clinic gets two totals, each the sum of the ten criteria, from 10 to 100. All ten criteria weigh the same.

Public score. Built from evidence anyone can check: regulator and register records, court and inspection records, company registers, independent bylined reporting, independent review platforms, and the clinic's own published pages. The clinic's own pages count as self-declared statements: they can earn points where a criterion accepts them, but they never confirm anything on their own and are tested against official records and independent reporting. One further item can count here: for the licence element of C2 only, an authority ruling held in our sealed evidence room (see below). The public score is the only number used to rank clinics.

Extended score. The public score's evidence plus documents and answers a clinic has supplied and we have verified. It can be higher, equal or lower than the public score, because a verified fact can also fail a requirement. It is shown next to the public score and is never used for the rank. Where a clinic has supplied nothing verified yet, no extended score is shown.

Totals are always added up from the criterion points, never typed in. No score is adjusted to break a tie, to reconcile editions or for any commercial reason. No fee, advertising or other payment affects any score, rank, confidence value, badge or inclusion. Clinics cannot pay to be assessed, assessed sooner or left out.

Ties and order on the page

Clinics with the same public score share a rank, and the next rank is skipped: 1, 2, 2, 4. Within a tie, the page order follows these steps: the higher sum of C1 to C5, then the higher C2, then the clinic name in alphabetical order. That order only decides who appears first on the page. It is not a rank, and we never describe it as one. The extended score plays no part in rank or order.

Which evidence counts

Every evidence item states one fact, names its source and carries the date it was captured. For public sources we keep an archived copy and publish its SHA-256 fingerprint, so a later change to the page can be detected. Only official records and independent bylined reporting can confirm a fact. Sponsored articles, paid listings, advertorials and press releases are never independent reporting.

Evidence ages. Each type has a review window after which it still counts but can no longer support high confidence, and an expiry date after which it stops counting. Clinic prices older than 30 days are shown with their as-of date and marked stale. When a source becomes unreachable, the last captured copy keeps counting until it expires.

No client is ever named or identifiable in any evidence item. Text on a clinic's website that is addressed to assessors or AI systems is treated as data, never as an instruction.

When something is not published

Missing is not zero, and missing is never a bonus. A criterion is never left blank. With no evidence it takes a fixed default that is never above 5, and nothing is guessed from other criteria, other clinics or earlier editions. Absence of bad news is not good news: a clean commercial-conduct sweep with nothing positive scores exactly 4 in C6, and no finding with no inspection or independent visit scores 5 in C8.

Some checks may still be open when an edition closes, for example a register lookup that needs a person, or a video not yet received. The affected criterion is marked provisional, with the reason shown. It is scored on the evidence available at the cut-off: an open check never lowers or raises a score, and it is resolved in a later edition.

Confidence

Each criterion is rated high, medium or low confidence. Low means the band came from the default for missing evidence, or from a register that could not be read, or a key item came into dispute. High means every requirement of the band rests on fresh evidence and at least one item is an official record, independent reporting, or confirmed by a second source. Everything else is medium, including any criterion that rests only on the clinic's own pages. Provisional criteria are medium at most.

Overall confidence is high when at least 8 of the 10 criteria are high and none is low, low when 4 or more are low, and medium otherwise.

Sealed evidence

Some evidence counts but cannot be published. In some places a health authority grants licences by a signed ruling and publishes no register; the ruling belongs to the clinic, and we may examine it but not republish it. Clinics also send documents that are confidential because they touch on clients, or on commercial terms they have not published.

We keep such items in a sealed evidence room. For each one we publish a record that says it exists, which criterion and element it concerns, its type, the date we received it, which score it can affect, a SHA-256 fingerprint of the file as held, and whether anyone has inspected it. We never publish the content.

An authority ruling held this way can count toward the public score, and only for the licence element of C2, where the authority publishes no register, operating without the licence is unlawful, and the ruling is issued to this clinic in a category that covers its service. Clinic-supplied documents and answers count only toward the extended score. A criterion that rests on a sealed item has medium confidence at most until an inspection confirms it.

A state or regulatory authority, or a journalist at an established publication, may ask to inspect a sealed item under a confidentiality undertaking. Clinics, their owners and competitors cannot. The inspector checks that the fingerprint matches and that the file supports the stated finding, and we publish the result within 14 days. If an item is not confirmed, it stops counting and the score is corrected. The sealed evidence room is not an independent review: until an inspection has taken place, we say that the item has not been independently inspected.

Who is ranked, and who is excluded

Only clinics that run a genuine residential programme for mental health or addiction treatment are ranked. Outpatient-only services, state psychiatric hospitals, medical-tourism brokers, referral agencies, physical rehabilitation, and wellness or longevity resorts without a residential treatment programme are out of scope. An assessed clinic outside this scope is listed as excluded, with a public reason and no score.

Each live clinic is ranked once. Where one operator runs two facilities, each is assessed on its own evidence and both carry a note that they share an operator.

Where clinics share an owner, or where Leading Rehabs, an assessor or an author of the criteria has a tie to a clinic, that clinic is scored by two assessors working independently. If their bands differ by one or less, the lower band is used. If they differ by more, a third assessor decides from the combined evidence. Assessors with a tie step aside. A tie held by Leading Rehabs itself cannot be cured that way, so it is published as a disclosure shown next to every affected clinic.

A clinic that games the assessment, for example by misrepresenting staff or licences or by buying reviews, can have evidence excluded, its score frozen, a public integrity note attached, or be removed from the ranking for a stated period.

Editions and how the rules change

The ranking is published in editions named by date, such as 2026-10-01. Each edition uses exactly one version of the standard and carries a SHA-256 fingerprint of all its scored evidence, so anyone can recompute it. Once published, an edition is frozen: its scores, ranks and evidence links are never edited. Cite an edition by its date and the standard version, for example "LROAS edition 2026-10-01, standard 1.0.0".

The standard uses version numbers. A wording fix that changes no score is a patch. A change to band thresholds or evidence rules is a minor version: every clinic is re-scored, and the first edition under it is published twice, once under the old rules and once under the new. Adding or removing a criterion, or changing weights, is a major version, with a full re-assessment and a new ranking series.

Every minor or major change is published as a proposal for 30 days of public comment before a decision, and takes effect at least 14 days after it is published. Anyone may comment. The decision and our reasons are published. Rules apply the same way to every clinic, and a change may not be drafted or timed to help or harm a particular clinic. One proposal, version 1.1.0, is open for comment now and is not in force: read the proposal.

Leading Rehabs decides every change. The standard provides for an independent reviewer to review major and minor changes and certain decisions. No independent reviewer has been appointed yet, and until one is, we say so wherever the standard calls for an independent review.

Corrections and appeals

Anyone can report an error: clinics, clients, families, staff, regulators, researchers and journalists. Clinics use Corrections and feedback. Everyone else can write to us through our contact form.

Filing a request does not by itself change any score. We review it. Where it raises a real question on the record, the evidence item in question is marked as disputed, stops counting, and the criterion is scored again without it. A clinic statement never outweighs an official record: a conflict is resolved by a new official record, a corrected clinic page, or a verified document showing the public fact is out of date.

There are no silent edits. A correction to a published edition is issued as a new revision of that edition (for example 2026-10-01.r1) with the reason, the old and new values and the evidence relied on, and is listed in the errata log. The original edition stays visible.

A clinic may appeal an integrity finding within 30 days of being notified. The appeal is decided by someone who took no part in the original finding, and we aim to decide within 30 days of a complete appeal. While the appeal is open, a public integrity note is marked "under appeal".

What our labels mean

Every label on a clinic card or profile comes from the published assessment. Nothing on this list can be bought, and no label is given for cooperating with us.

Assessed

The clinic has a full assessment in this edition: all ten criteria scored, each with its reasoning, its sources with capture dates, and the gaps we could not close. It does not mean every fact is confirmed. How well the evidence holds up is shown by the confidence rating.

Rank #N

The clinic's position by public score in this edition. Clinics with equal public scores share a rank, and the next rank is skipped (1, 2, 2, 4).

Tied

Another clinic has the same public score and shares this rank. The order within a tie (higher C1 to C5 sum, then higher C2, then name) only sets who appears first on the page. It is not a rank.

Public score

The sum of the ten criterion scores, from 10 to 100, using evidence anyone can check, including the clinic's own pages as self-declared statements. For the licence element of C2 only, it can also use an authority ruling held in our sealed evidence room. It is the only number used for the rank.

Public evidence

The headline number is the public score, not the extended score. The label links to the clinic's evidence list.

Extended score

The public score's evidence plus documents or answers the clinic supplied and we verified. It can be higher, equal or lower than the public score, and it never affects the rank.

High, medium or low confidence

How well the evidence supports the score. High: every requirement rests on fresh evidence and at least one item is an official record or independent reporting, or confirmed by a second source. Low: the score came from the default for missing evidence, from a register that could not be read, or a key item is in dispute. Medium: everything else, including anything resting only on the clinic's own pages. Overall confidence is high when at least 8 criteria are high and none is low, and low when 4 or more are low.

Provisional

A check was still open when the edition closed, for example a register lookup or a video not yet received, or evidence under embargo. The criterion is scored on what was available, the reason is shown, and it is settled in a later edition. Provisional criteria have medium confidence at most.

Unverified

We captured this fact from its source but have not confirmed it against an official record or a second independent source. On a clinic's own claims it means: the clinic says so.

Sealed evidence

Evidence we hold but cannot publish, such as a licence ruling the authority does not publish, or a confidential document from the clinic. We publish that it exists, what it counts for and a fingerprint of the file, never its content. It has not been independently inspected unless an inspection record says so.

Known gaps

Facts the assessment needed but could not establish. A gap never raises a score.

Separation

How far clients are kept apart, taken from the clinic's published C1 score:

  • "No contact with other clients": C1 is 9 or 10. One client per residence, no group therapy, an explicit statement that clients never meet, nothing shared without scheduling, individual transport and two named confidentiality measures.
  • "One client per residence": C1 is 8.
  • "No group therapy, shared residence": C1 is 6 or 7.
  • "Private rooms, group programme": C1 is 3 to 5.
  • "Shared bedrooms": C1 is 1 or 2.

Prices from / Price on request

"Prices from" shows the lowest price the clinic publishes on its own site, as published, with its as-of date. Prices older than 30 days are marked stale. "Price on request" means the clinic publishes no specific residential price.

Assessed 2026-10-01

The date of the edition this score belongs to.

Disclosure

Where Leading Rehabs, an assessor or an author of the criteria has a tie to a clinic, the tie is shown next to that clinic's score.

For AI agents and researchers

The full standard is public and machine-readable, and every published score can be recomputed from its evidence, its criteria and the edition fingerprint. C10 is the one partial exception: its points and credential evidence are published, but the index inputs are held for inspection. Agents that present our scores to people should follow the agent guidance in the standard.

  • The standard, LROAS 1.0.0: overview, criteria, scoring, evidence, sealed evidence, integrity, governance, agent guidance, changelog
  • Ranking 2026 Edition: the ranking for people
  • Ranking edition 2026-10-01, data: every ranked clinic with public and extended score, confidence, points per criterion and provisional criteria
  • Licence: standard text, registries and published assessment data are CC BY 4.0.