About this scale

The framework

Items are grouped using the 2020 Five Domains Model: nutrition, physical environment, health, behavioural interactions, and the mental state that arises from the first four. The model is the standard structure in animal welfare science for reasoning about what an animal is actually experiencing, and it is published open access under CC BY 4.0, so it can be built on freely with attribution.

Each item is a five rung ladder of observable behaviour, rated for today (the sleep item asks about last night). Both choices are deliberate and follow how human serial instruments are built. The stated window matches the intended one-to-three-day scoring cadence, so one bad day enters the record the day it happens; an item with no window invites the owner to rate a vague recent pattern, which one bad day never moves, and the record smooths in the owner's head before the maths ever see it. All smoothing in this tool lives in the trend logic, none in the questions.

Why behavioural rungs rather than "as usual". The first version of this instrument anchored every item to the animal's own baseline with "as usual" wording, and that carried a known weakness this page described at the time: an owner living through a slow decline recalibrates, so "as usual" quietly comes to mean "usual lately" and the curve flattens. That comparator lived in the owner's memory. A behavioural rung does not: "needing a hand to stand" is true or false today regardless of what has quietly become normal, so slow decline is caught as rung crossings that no amount of recalibration can hide. The comparison to the animal's own normal has not been lost; it has moved to a recorded baseline, described below, where it cannot drift. The drinking item is the one designed exception: more and less are both bad, so it is a deviation ladder read against the recorded normal, which is what stops a stable, chronically thirsty patient (a managed diabetic, say) from carrying a standing penalty for a state that is its normal.

The starting point is the median of the first three scores, not the first score on its own. It was the first score until July 2026, and that left the comparison asymmetric: the recent end of it was a three-entry average, chosen so that one bad day could not flip the reading, while the end it was measured against was a single unsmoothed day that could be any bad day at all. An owner whose first score landed on a bad day carried a depressed reference point from then on, and a real decline afterwards could read as holding steady. That is not an unusual case, because people reach for a tool like this when today was bad. The middle of the three is used rather than their average, because an average absorbs part of a decline that begins on the third day and would quietly raise the threshold above the five points stated below.

Response format

Every item carries its own five rung ladder, scored 0 to 4 and always presented worst first. The first version of this instrument shared one anchor set across twelve items, and format consistency was where its speed came from; the ladders trade some of that first-pass speed for responsiveness and drift resistance, since each answer is an observable state rather than a frequency judged against memory. Behavioural rungs are memorable, so repeated use recovers most of the speed. Worst-first ordering is primacy, kept from the first version: in a self-administered scale the option shown first pulls answers toward it, and of the two errors, an owner who calls the vet when they did not need to is the better one.

The full ladders are printed with the items below. Alongside the twelve scored items sit two recorded, unscored questions: the felt-of-change question asked with every check, and the good-days item asked weekly. Both are described in their own sections.

The items: dog

The items: cat

Six items carry across close to unchanged. Six are reworded, because grooming, hiding and jumping carry far more signal in cats, and because panting means something quite different in a cat than in a dog.

Domains and weighting

Comfort deliberately carries three items, the heaviest weighting in the instrument. A 2022 scoping review of the nine generic canine and feline quality of life tools in the literature found that only five of them included even one item relevant to pain, in a category of tool built almost entirely for animals with chronic and terminal disease. That gap is the most common design failure here, and over-weighting comfort is the correction.

Scoring and output

All twelve items must be answered before a result renders. There is no partial scoring and no skip logic, which avoids a misleadingly high total from an incomplete check. The felt-of-change question and the weekly items are recorded alongside the result and are never added to it.

The result leads with a domain profile, sorted most-in-need first as a proportion of each domain's own reference, and names the first domain with a fixed sentence about what usually sits behind it. The total is shown but deliberately subordinate in the visual hierarchy, because it is a tracking figure rather than a finding.

There are no interpretive bands and no thresholds. That is a considered response to the standing criticism of scored quality of life instruments: they weight physical deterioration heavily and affective state lightly, so an animal can score poorly while remaining content and engaged. A banded total invites the owner to read a number as a verdict. A domain profile invites them to ask which part of the day is under strain, which is a question that usually has an answer.

The recorded baseline

Where the assessment is used repeatedly with storage behind it (the Wattle Pet app, or the paper sheets), the animal's normal is recorded once, on the same ladders, answered for how the animal was before it became unwell or for its steady state on treatment. Everything afterwards can then be read against a reference that is written down rather than remembered, which is what makes recalibration drift structurally impossible rather than merely discouraged. This follows the patient-specific instruments in human outcome measurement (PSFS, MYMOP) and client-specific outcome measures in veterinary arthritis research.

The baseline changes interpretation only: which domain is called out first, what counts as strain, and a reference line on the chart. The total is the raw sum with or without one, so scores remain comparable across surfaces, and the trend logic never reads the baseline at all: a constant offset cancels in every subtraction it performs. A domain whose recorded normal is zero is never called the area needing attention, because it has no distance to fall: a three-legged dog is not permanently failing at mobility.

A recorded baseline is edited in exactly two situations: the description was wrong, or something about the animal has permanently changed, such as a new diagnosis or a recovery to a new steady state. It is never updated on a schedule, never prompted, and no check score ever feeds it. A baseline that drifted down with decline would rebuild the exact failure it exists to prevent.

On this page there is no baseline flow, because nothing entered here is stored. The assessment scores identically without one.

The felt-of-change question

Every check ends with one retrospective change question, "Compared with a few days ago, how does your pet seem overall?", answered much worse, a little worse, about the same, a little better, much better. It is recorded with the entry, shown to the vet beside the score series, and never summed into anything.

The exclusion is the point, and it follows the human literature. Retrospective change ratings correlate more with current state than with measured change, sequential deltas do not sum, and a slow decline changes less per day than an observer can notice, so a change-question instrument reads "about the same" all the way down. The SF-36 carries a health transition item and excludes it from all eight of its scales for the same reasons. What a change question is good for is what it does here: a second signal whose blind spots do not overlap with the items'. The ladders catch the slow change an owner adapts to; the owner's felt sense catches what twelve items never ask about.

When the two disagree, the disagreement itself is read. A run of "worse" against steady scores usually means the owner is seeing something outside the item set, and the copy says so and asks them to note it down and tell their vet; falling scores against a steady felt sense is the record catching what day-to-day adaptation hides. The owner's observation is never overruled by the number.

The weekly review

The good-days item ("Thinking about the whole of the past week, how many good days has your pet had?") and the carer section are asked about every 7 days, not with every check. The reason is arithmetic: a question whose recall window is seven days, asked every two days, has five days of overlap with its own previous answer and cannot move. Its window is its cadence, which is also why the daily items ask about today. The good-days answer is tracked as its own weekly series rather than inside the total, which is why the total is out of 48.

Trend logic

Day to day noise on a scale like this runs to roughly 3 or 4 points, so the tool reads a trailing three entry mean rather than any single day, and needs at least three dated entries before it will say anything at all. The thresholds below are carried over from the first version and are provisional: the ladders are sharper than the items they replaced and rated for a single day, so legitimate day to day movement will be larger, and the noise floor will be re-estimated once real series exist on the new items.

What fires is absolute; only the wording is personalised. The floor reads the raw rolling average and never the distance below a recorded baseline, because an animal at the bottom of these ladders is in a grave state whatever was normal for them. Where a baseline exists and the recorded scores have borne it out, the wording stops implying a change that has not happened and splits the advice by whether a vet is already involved. It still says the score is low, and a pet that has fallen 5 or more below their own normal while under the floor gets the full message regardless. A baseline is treated as a claim until the series agrees with it, because people reach for a tool like this on a bad day and may describe that day as normal.

The rise and the fall run the same test with the sign flipped, so a lift has to hold for the same week before it is named, for the same reason one good Tuesday is not a recovery. Until 2026-07-29 there was no rising state at all: the classifier read a single directional subtraction, so an improving pet fell through to "holding steady" and was told its scores were stable.

The framing throughout is that a fall which holds means the current plan is losing control of the condition, and is a prompt to revisit analgesia, medication, appetite, nursing support or comfort care. It is never framed as a euthanasia trigger. Single low days are explicitly normalised in the copy, on the grounds that an instrument which alarms owners over one bad afternoon is worse than no instrument. Equally, a stable score is not read as good news on its own: whether it is depends on the level it has settled at and on what the owner and vet are aiming for, and the copy says so rather than congratulating the plan.

The carer section

Seven optional items, unscored, reported separately, and never combined with the animal's total. They cover the recognised dimensions of veterinary caregiver burden.

Items 1 to 6 share one anchor set: No, not at all, Rarely, Some days, Mostly, Yes, comfortably. Item 7 is the exception, mirroring item 12 on the animal's side, and uses Not coping, Struggling, Up and down, Well enough, Very well. Both run 4 down to 0, in the same direction as the animal's items, so a higher answer always means things are going better.

Those six carried the animal's anchor set until 2026-07-29, which put "Yes, as always" on questions about the owner's own sleep and finances. "As always" exists to pin a judgement to the animal's own baseline, and there is no baseline to pin to here.

Why this is in here at all. Spitznagel's work on veterinary client caregiver burden found that once the animal's own quality of life is controlled for, caregiver burden and income remain significant predictors of whether an owner begins considering euthanasia. Burden, anticipatory grief and animal quality of life behave as distinct constructs, and decision related guilt is repeatedly flagged as an under-recognised source of distress. Separately, the 2022 review found that no generic pet quality of life tool assesses carer impact at all.

It is presented to the owner as optional and as a weekly rather than a daily exercise, since carer strain shifts more slowly than the patient does. The output is three tiers of fixed supportive copy keyed on how many answers sit at the bottom two anchors, with Australian support services listed underneath. Deliberately no number: scoring how well someone is coping with a dying animal is exactly the false precision the rest of the tool avoids.

Why we wrote our own

The obvious question is why this is not simply one of the established scales. Several good instruments exist, and we looked at all of them. The practical position is that the well-known ones are variously protected by copyright, licensed commercially, or published without a scoring method that can be implemented faithfully. Those are entirely reasonable choices by the people who made them, and none of it is a criticism.

It does mean that a free tool intended to be handed out, printed, and used by any clinic that wants it has to be written from scratch. So it was.

Copyright protects expression, not ideas, systems or methods. Domain structure, a 0 to 4 response format and trend tracking are all free to build on. Specific item wording is not. Every item on this page was written originally for this assessment, and the framework it hangs on is openly licensed. No item here is taken from, adapted from, or paraphrased from any other quality of life scale. The same applies to the carer section: the constructs of caregiver burden are shared and well described in the literature, and none of the wording is borrowed.

What has not been validated

This instrument has not undergone formal psychometric validation, and we do not claim that it has. No factor analysis, no test-retest reliability, no discrimination testing. Treat it as a structured way of paying attention, grounded in published welfare science, and as a means of generating a trend that an owner can bring to a consultation.

For context rather than excuse: this is close to the state of the whole field. The 2022 review of nine generic tools found none had been evaluated for criterion validity or inter-rater reliability. The 2025 review of 41 disease-specific canine instruments concluded that none had been thoroughly evaluated across all necessary psychometric properties, and that a third of the published instruments were not fully available to read. Several of the most familiar scales in daily use have no published statistical validation behind them either.

If this tool accumulates enough consistent use to make a validation study worthwhile, we will run one and publish it.

Using it in your practice

You are welcome to use this with your own clients. The printed sheets are the part worth handing out: the score sheet carries every question with its five answers, so an owner can score at the kitchen table with no phone and no website, and the tracking sheet holds a month of totals.

There is no account and no data collection. Nothing an owner enters leaves their browser, which also means nothing syncs, so the paper sheet is the record. A client arriving with two weeks of scores changes the shape of a palliative consultation considerably.

Wattle Clinic will eventually let a practice send a check to a client and have the result land on the patient record with a trend. That is not built yet.

References

• Mellor DJ, Beausoleil NJ, Littlewood KE, McLean AN, McGreevy PD, Jones B, Wilkins C. The 2020 Five Domains Model: Including Human and Animal Interactions in Assessments of Animal Welfare. Animals 2020;10(10):1870. Open access, CC BY 4.0
• Quality of Life Measurement in Dogs and Cats: A Scoping Review of Generic Tools. Animals 2022. Open access
• Instruments to Assess Disease-Specific Quality of Life in Dogs: A Scoping Review. Animals 2025. Open access
• Spitznagel MB, Jacobson DM, Cox MD, Carlson MD. Caregiver burden in owners of a sick companion animal: a cross-sectional observational study. Veterinary Record 2017;181(12):321. PubMed
• Spitznagel MB, Carlson MD, et al. Validation of an abbreviated instrument to assess veterinary client caregiver burden. J Vet Intern Med 2019;33(3):1251. Open access
• RCVS Knowledge. Quality of life assessment tools. Curated index