Measures

Is the work actually helping?

A score on a cadence is the clearest answer anybody has to that, and it is the same evidence a payer asks for. PHQ-9 and GAD-7 are included; a practice can add its own.

Two numbers, because they disagree.

Since the last session and since the beginning answer different questions. A reading can be up on the fortnight and still be a long way down on the year, and a practice that only sees one of those is either alarmed by a bad week or blind to a slow slide.

PHQ-9 R. Delacroix · every four weeks

4 Aug 15 Moderately severe +3 since last −6 since baseline
7 Jul 12 Moderate −2 since last −9 since baseline
9 Jun 14 Moderate −7 since last −7 since baseline
12 May 21 Severe Baseline

August is three points worse than July and six points better than May. Both are true, and a clinician walking into that session needs to know both. Which direction counts as better is a property of the instrument, not an assumption — a scale where a higher score is the good news is read the right way round.

Severity bands

A number means nothing until it sits in a band.

These are the bands the instruments publish, not thresholds we chose. Switch between them: the ladders are different lengths because the scales are, and a practice adding its own defines exactly this — items, a maximum, and where the lines fall.

PHQ-9 GAD-7

The ninth question

Asking somebody about self-harm is a duty, not a field.

The PHQ-9 ends by asking whether a person has had thoughts that they would be better off dead. The moment a client can answer that alone, at home, at eleven at night, this product has opened a channel for disclosing suicidal ideation. If the answer lands in a row nobody reads until the next appointment, we have built something worse than paper: the appearance of being heard, without the substance of it.

Sending one home is off until a practice turns it on: A measure containing a self-harm item cannot go to the portal unless the practice has enabled it deliberately and written down what it actually does out of hours. The default is off. A practice that has not thought about this has not agreed to it, and silence reads to a client as a promise of monitoring that nobody is providing.
The crisis instructions show every time: Whatever the practice writes there is shown after every submission, not only after a worrying one. Showing it conditionally would tell somebody what they had just disclosed, from a screen, with nobody in the room.
The alert is written before anyone is told: The record is created first and the notification is best-effort on top of it. Somebody can disclose this at two in the morning on a Saturday, and that disclosure must not depend on an email succeeding.
It is acknowledged by a person, never cleared: An alert is closed by somebody saying they have seen it, and that is kept on its own audit trail. Who saw the flag and when is the record that matters if anything ever goes wrong.
Which item is risky is the item's own property: Detection reads a flag on the question rather than a hardcoded position in a known questionnaire. A practice's own scale, or a revised edition that moves its items, is handled without anybody changing code.

What ships

PHQ-9 and GAD-7, attributed

Both were released by their authors for use without permission, and the product carries that attribution with them rather than presenting them as ours.

Your own

A practice can add its scales

Items, maximum values, reverse scoring and severity bands are all yours to define, so an instrument a practice already uses on paper does not have to be abandoned.

Cadence

Assigned, then due by itself

A measure is assigned to a client on a rhythm — every four weeks, or once for a baseline — and comes due without anybody keeping a list.

Never interpreted

The number, not a verdict

A client is shown their score and the published band label. The product does not tell them what it means for them, and it does not generate advice. That reading is the clinician's work.

Answers to these are the most sensitive thing in the record.

Item-level responses are self-reported symptoms, which makes them among the most revealing information a practice holds. Each answer is encrypted in the database, scoped to the practice that owns it, and covered by a policy. They are read as part of the chart, and opening a chart writes an audit entry naming who opened it.

Scoring runs on the server, never in the browser. A score worked out on the client's own machine is a score that can be altered and cannot be evidenced afterwards.

How records are held

The rest of it

Start with the week you are already having.

Set up a practice, put a real week in the calendar, and see whether it fits before you move anything across.

Start free Read the documentation