Evidence check · July 2026Current rules and proposals, kept separate. View sources
WELLDASH

Support that plans its own exit.

Independence is not built by leaps. It is built by delivering support at the least intrusive level that works and then fading it, one rung at a time, on evidence. That only becomes real if the record can say which rung was used, who actually did the task, and what the hands were doing — every time. This note is how WellDash records the fade.

Two rails, scored togetherWho did it, every timeOne axis per revisionCitations attached

What most records lose, and this one keeps.

Each of these is a difference the record either holds or destroys. A blended support score destroys three of the four before anyone gets to the analysis.

Distinction 1

Two rails, not one score

How much physical help, and how much prompting, are different questions with different answers. They are recorded concurrently, never blended into a single support number.

Distinction 2

Doing-for is not a rung

Who performed the task is its own field. Doing it for the person is a different kind of record — the model reablement is defined against — never the top of the assistance ladder.

Distinction 3

Bearing is not guiding

Carrying the weight while the person still moves preserves their motor learning; steering the movement takes it. Most real hands-on care is both at once, and the record says both.

Distinction 4

A floor is not a failure

Support that rests at a level authored for safety is a legitimate outcome, not an unfinished fade. The record distinguishes a floor from a plateau nobody noticed.

The two ladders NICE NG74

Physical help and prompting, scored at the same time

Every response in a WellDash plan step is a configuration — four dials recorded together: strategy, cue, assistance, manner. Independence work moves two of them.

  • The assistance rail, faded downwards: hand_over_handpartial_physicalsupervisionnone. supervision is not an absence of support; it is the standing-by-ready posture, present and attentive and not intervening.
  • The cue rail, faded independently: none · environmental · gestural · modelled · verbal_enabling · verbal_direct. modelled is the demonstration rung — show the person first, hands off entirely. A demonstration is a teaching dose, never doing-for.

Recording them concurrently is the whole point. A person can need no hands at all and still need to be shown; another needs a steady arm and no words at all. A single blended support score erases both facts, and with them the ability to say which change caused the improvement.

The reablement line SCIE

Who did the doing — the question a support level cannot answer

performed_by rides every execution: person, shared, or carer_did_for. It answers “who did it?” separately from “how much help was given?”, and the distinction is not academic.

  • Doing the task for someone is the model reablement is defined against, not a heavier rung of helping them do it. SCIE draws exactly this line: traditional home care uses a doing-for approach with little attempt to develop self-care skills; reablement uses a doing-with approach.
  • The doing-for → shared → person curve is the reablement line, dated, computed from the same rows the carer already recorded. Nobody fills in a second form to produce it.
  • It also catches the failure mode the research names: the 2010 SPRU evaluation warns that conventional care following reablement can undo it by doing things for people again — and its service-user data shows conventional services leaving some people more dependent, not merely failing to improve them. Without a who-did-it field, that regression is invisible in the record.
Hands on

What the hands did: bearing, guiding, or both

When hands are on, touch_kind records what they were doing — bearing (taking the load while the person still moves) and guiding (directing the movement). It is deliberately not an either-or: most real hands-on care is both at once, and the record says both.

The distinction matters because the two do opposite things to learning. Weight-bearing leaves the person’s own movement intact — they are still doing the moving. Steering replaces it. Two shifts can log identical assistance rungs while one preserved the person’s motor learning and the other quietly took it over. Reablement reporting that cannot tell those apart will overstate progress in one direction and physical contact in the other.

Consent. Consent rides every touch. A recorded touch without a recorded basis for it is a gap in the record, not a detail.
Working rules

The rules that keep a fade attributable

  • One axis per revision. Change the physical rung or the cue, never both, so the record can attribute what worked. A revision that moves both dials produces a result nobody can read.
  • Struggle steps support up now; success steps it down next time. Help arrives in the moment it is needed and is withdrawn on the next occasion, not in the middle of a task the person is currently failing.
  • Acquisition runs most-to-least; maintenance runs least-to-most. Learning something new starts with enough support to succeed; holding onto something learned starts with the lightest touch that might do.
  • Never fade past the authored floor. Where safety requires a level, that level is the resting place — and the record shows it as a floor, not as a fade that stalled.
  • Graduation is a sustained window plus generalisation probes. One good day is not independence; the same thing done unsupported across time, and in a second setting, is.
Published science SPRU 2010

Where the shape comes from

Scoring physical assistance and cognitive prompting concurrently, as two rails, is warranted by the performance-assessment literature: PCPT-ALF (Bowen 2017) scores physical ability and cognitive support at the same time with high internal consistency, and PASS (Chisholm 2014) grades assistance through a nine-level hierarchy rather than a single present-or-absent flag.

The fade mechanics themselves are classical: graduated guidance fades touch (Foxx & Azrin 1973); time delay fades the cue (Touchette 1971); support shifts contingently with success and struggle (Wood & Middleton 1975); and fading direction differs between acquisition and maintenance (Demchak 1990; Libby et al. 2008). Over-helping is not neutral — prompt dependence and learned non-use are the recorded harms of helping too much.

UK reablement carries the same posture at guideline level. NICE NG74 recommendation 1.1.3 tells practitioners to “learn to observe and guide and not automatically intervene, even when the person is struggling to perform an activity”. The DH-funded SPRU evaluation names that practice “standing and watching” and is explicit that it is an important job in itself rather than passivity — and it records supervision of hazardous tasks as a named element of extended care plans, which is where the permanent-floor doctrine comes from.

Boundary. The components carry their citations; the token set itself — the rung names, supervision’s position on a four-step assistance axis, and the concurrent matrix with the cue rail — is WellDash’s own, stated plainly as ours-to-test. No published instrument grades a supervision rung in this position. It is validated continuously against each person’s accumulated record rather than by a one-off norming study.
For the report

What this puts in front of a practitioner or a panel

The point of typing the fade is not tidiness. It is that the independence case writes itself out of work already done, in the form the person asking actually needs.

  • A support-over-time view that shows this person against their own usual range — the fade as a dated line, not an assertion in a review meeting.
  • Assistance grouped by the function the behaviour was serving, so “he needs more help” becomes “he needs more help in these situations, and less in these”.
  • Who actually did the task, across dated rows — which separates the person’s progress from the team’s habits, and is the difference between a service that reables and one that quietly takes over.
  • A floor that reads as a floor. Support resting at an authored level is evidence of a considered plan, and it is recorded as such rather than as an unexplained plateau.

None of it is re-keyed. The printable documents assemble from the same captured substrate the app already holds — and a field nobody recorded prints as a gap, never as a guess.

FAQ

Questions, answered the way we answer them

How do we record fading support?

Each execution logs its cue rung and its assistance rung, plus performed_by and, when hands were on, touch_kind. The fade is the rung falling across dated rows, and the doing-for → shared → person curve computes from the same rows.

What is the difference between hands-on guidance and weight-bearing?

Guiding steers the movement; bearing carries the load while the person still moves. Most real hands-on care is both at once, and the record says so — which keeps reablement reporting honest about physical contact.

Why record prompting and physical help separately?

Because they are different kinds of support and they fade in different directions. A person can need no hands at all and still need a demonstration; another needs a steady arm and no words. Scored on one blended number, both facts vanish. Scored on two rails at once, the record can attribute what actually changed.

Is doing the task for someone the top rung of the assistance ladder?

No. Doing it for the person is a different kind of record, not a heavier rung — it is the model reablement is defined against. WellDash records it as performed_by carer_did_for beside the rung, so a shift that did the task for someone can never read as a shift that supported them heavily.

Is your prompt hierarchy validated?

Every component carries its primary citation — the prompt-fading literature, PASS’s assist hierarchy, PCPT-ALF’s concurrent scoring, UK reablement guidance. The exact token set is WellDash’s own, stated plainly as ours-to-test, and it is validated continuously against each person’s accumulated record rather than by a one-off norming study.

Can support legitimately stay where it is?

Yes. A permanent floor authored for safety is a legitimate resting level, not a failure of the plan — UK reablement research records supervision of hazardous tasks as a named element of extended care plans. WellDash never fades past the authored floor.

See a fade recorded on one synthetic person.

The demonstration shows capture, the typed grammar and the ranked responses on synthetic records.