Unexplained is a body question
New, unexplained or out of character does not start a behavioural conversation. It starts a physical one, and the record enforces the order rather than recommending it.
Diagnostic overshadowing — reading a bodily illness as part of someone’s behaviour or disability — is the standing finding of the UK mortality reviews. It is not a knowledge problem: the staff who miss it know that pain causes distress. It is an order-of-operations problem. So WellDash puts the order into the structure of the record rather than into a training slide.
A policy that says “consider physical causes” is not a safeguard. These four properties are what turn the intention into something the record actually enforces.
New, unexplained or out of character does not start a behavioural conversation. It starts a physical one, and the record enforces the order rather than recommending it.
Each first-pass prompt maps onto authored typed causes. A prompt with no authored mapping is an error in the system, never a quiet guess about the person.
What happened now, what is different from this person’s usual, and what is known about their life are shown side by side for a human to read together. The machine does not blend them into a verdict.
Treat, then resolve, dated: the dechallenge is the strongest thing a care record can hold, because it is the one piece of evidence that was tested rather than argued.
The override is simple to state and hard to hold: when a behaviour is new,
unexplained, or out of character for this person, no behavioural reading is served
until physical causes have been considered. In the record it is a strategy in its own
right — body_first — not a caveat attached to something else.
The clinical shorthand the field uses is constipation until proven otherwise, and it is a good summary of the posture: start from the most common, most silent, most treatable causes, and make the team prove their absence rather than assume it. NICE NG11 puts physical causes first for exactly this reason.
Structuring it as an override rather than a reminder matters because the failure it prevents is a failure under pressure. Nobody forgets that pain matters on a calm day. The record has to hold the order on the day when the behaviour is loud and the shift is short.
Seven first-pass prompts sit at capture — pain, infection, nutrition, constipation, hydration, medication, environment — the learning-disability nursing checklist most UK teams already know. Each prompt is bound by an authored mapping onto the precise typed causes it covers, and behind them sits a differential of twenty-four physical causes as a typed vocabulary rather than free text.
When something is off, three different kinds of knowledge matter at once, and blending them destroys all three. WellDash shows them as lanes:
They are laid side by side for a human to read together. The machine does not merge them into a single verdict, because the merge is exactly where a clinician’s judgement belongs and exactly where a confident-sounding system would do the most harm.
A synthetic example, of the kind the record holds every day: the tapping starts, the bowel clock reads five days, a laxative is given, the bowels open, the tapping stops. Each of those is a dated row, and together they are a dechallenge — the physical cause was treated and the behaviour resolved.
The reason this outranks any interpretation is that it was a test. And when the same chain has happened before, the record shows it as this person’s own repeated result rather than a hunch — which is what changes the conversation with a GP, a nurse or a panel from impression to evidence.
A clinician-authored graph of symptom → consideration → disposition triples sits behind the body-first work. It is used in exactly two ways, and no others: as typed vocabulary, and as authored disposition edges that a human activates. Every row carries its provenance, and every render carries the line that the human decides.
The Confidential Inquiry into Premature Deaths of People with Learning Disabilities (CIPOLD, 2013) established the scale of avoidable death in this population, and the NHS LeDeR programme has been reviewing lives and deaths since, publishing the recurring themes. Those themes are unglamorous and consistent: constipation, aspiration, sepsis, untreated pain, medication harm, and communication failures around all of them.
NICE NG11 translates this into guidance — physical causes first, before a behavioural formulation is treated as the explanation. The clocks and the differential in this system watch for what those reviews keep finding.
None of these are written from scratch at the moment they are needed. They print from what was already recorded — and a field nobody recorded prints as a gap, never as a guess.
The body-first override outranks every behavioural reading. The system surfaces the co-occurrence evidence — the bowel, sleep and intake clocks, the typed rule-outs, any prior dated dechallenge — so the human call is an informed one. Constipation until proven otherwise, with the person’s own record attached.
Never. Escalation is an authored, human-activated step armed with the evidence; the machine never routes an emergency call on its own.
As the safety lens it is. LeDeR’s themes shape what the clocks and rule-outs watch for, and published reviews were used to stress-test the machinery against known failure patterns. Its limits define ours: epidemiology shapes watches — it never becomes per-person prediction.
Treat the suspected physical cause and see whether the behaviour resolves — recorded, and dated. It is the strongest evidence a care record can hold about this person because it is a test rather than an interpretation, and when the same chain has happened before, the record shows it as this person’s own repeated result.
No. It is a first-pass prompt set — pain, infection, nutrition, constipation, hydration, medication, environment — that opens onto a typed differential of physical causes so the consideration is recorded as considered. What it produces is evidence for a human decision, never a conclusion.
Inside the same record and under the same rule. STOMP is the standing warrant that a chemical response is never the answer to a body problem, and a medication given as a response is recorded as an intervention with a before and an after like any other — so it can be examined rather than assumed.
The demonstration shows capture, the typed grammar, the body-first lens and the ranked responses on synthetic records.