Editorial Policy and the Claim Ladder

Most of this page is one thing: the ladder every efficacy claim on this site is graded against. It is the mechanism the whole site runs on, so it comes first and the housekeeping comes after.

The reason it exists is specific to this subject. Arts in health is a field where an unsourced percentage from a conference slide and a Cochrane review with a confidence interval get quoted in the same paragraph, at the same volume, often by people with no interest in misleading anybody. A reader cannot be expected to sort that out unaided. So every claim about whether something works carries a label, in the article, next to the claim, saying what kind of evidence is behind it.

The three tiers

Tier 1: supported

There is at least one systematic review, or an adequately powered randomised controlled trial, showing an effect on a stated outcome. The article must give you the numbers to check that judgement yourself: the effect size, the confidence interval, the certainty rating, the sample size, and the year. A Tier 1 label without those five things beside it is a failure of this policy, not a shortcut.

Tier 1 is narrower than it sounds. It applies to one outcome at a time. Rhythmic auditory stimulation has Tier 1 evidence for gait velocity after stroke; that says nothing at all about mood, and the article will not let it.

Tier 2: promising but limited

There is controlled evidence, and it is not strong enough to lean on. In practice a claim lands here for one of four reasons, and the article names which one:

• the trials are small, or there are very few of them
• risk of bias is high, or the certainty rating attached to the finding is low or very low
• the effect appears in a subgroup rather than in the whole sample
• the effect appears at an early timepoint and is gone by the later one

The last two matter more here than people expect, because a subgroup result at week 6 that has dissolved by week 10 can be reported perfectly accurately and still leave a reader with a completely false impression of what was found. On this site, if the whole sample result was null, the whole sample result is stated first.

Tier 3: practice and anecdote

Either nobody has run an adequately powered controlled study of it, or the controlled evidence that exists is null. This covers a great deal of what happens in care homes, on wards and in community halls, including work that people find valuable and that experienced practitioners believe in.

What a Tier 3 label is not

A Tier 3 label is not a statement that something does not work. This needs saying loudly, because it is the error that runs in the opposite direction to the one this site usually corrects, and it is just as damaging.

“No adequately powered controlled evidence” is a statement about the research literature. “It does not work” is a statement about the world. They are not the same, and the first does not imply the second. A great deal of arts in health has never been properly studied for reasons that have nothing to do with whether it helps: the interventions are hard to blind, the funding is short term, the outcomes people care about are difficult to measure, and the sector’s own money goes into delivery rather than trials.

So Tier 3 means the question is open. It means: nobody can honestly promise you this will help, and nobody can tell you it will not, and you are entitled to weigh it up knowing that. Where the controlled evidence is not merely absent but actually null, the article says so in those words, because that is a different and stronger statement than an empty file.

The tiers are per claim, not per activity

Nothing on this site is tiered as a whole. Music therapy is not “a Tier 1 thing”. A specific outcome, in a specific population, from a specific body of evidence, gets a tier. The same modality routinely carries a Tier 1 claim on one page and a Tier 3 claim on another, and that is the point of grading claims rather than grading subjects.

The rules that support the ladder

Every number carries its source, its year and its design. Not the source alone. A figure attributed to a named organisation with no indication of what kind of study produced it is exactly the sort of citation this site was built to stop repeating.

A before and after evaluation is never called a trial. An uncontrolled evaluation that measures a group before an activity and again afterwards cannot separate the activity from time passing, regression to the mean, or who chose to take part. Those evaluations are useful and this site describes plenty of them. It calls them evaluations, every time, including when the report quoting them calls them something grander.

Figures are quoted in the units the source used. No converting a result into a livelier looking number, no rounding in the flattering direction, no collapsing nested counts (studies inside reviews inside a publication total) into two large numbers side by side.

Whole sample first, subgroup second. Always in that order, whatever the abstract does.

Every article that makes an efficacy claim links to the methods explainer, so that the reasoning behind the label is one click away from the label itself.

Where a widely repeated claim does not survive its source, the article says so and shows the trail. Several pages here exist for no other reason. That is uncomfortable in a sector that campaigns for its own funding, and it is the most useful thing this site does.

Who checks what

Two reviewers, with genuinely different jobs, because the two failure modes are different.

Dr Rhian Vaughan, MA, PhD, music therapist registered with the Health and Care Professions Council, reviews everything describing what a registered arts therapist does: the boundaries between modalities, what a session actually involves, training and registration, and the line between clinical therapy and community activity. She does not comment on named individuals.

Dr Anna Bergström, PhD, MSc Epidemiology, a public health researcher, reviews the evidence. She owns the ladder above, checks every effect size, interval, certainty rating and sample size against its source, and signs off the pages where the evidence is weakest or most disputed.

Reviewed pages show the reviewer’s name and the date of review. A review confirms that a page describes its sources accurately. It is not clinical advice to any reader, and it does not make either reviewer anyone’s therapist.

Sourcing

Sources used here are systematic reviews and the Cochrane Library, peer reviewed primary studies, statutory regulators and professional bodies, and national and international health organisations. Sector reports, policy documents and programme evaluations are used and named, but they are treated as descriptions of what is happening rather than as evidence that it works, and where such a report quotes a figure this site goes to the underlying study before repeating it.

External organisations are cited, never claimed. No body named on this site is a partner, an endorser or an affiliate of it, and no logo of any such body appears anywhere here.

The voice this site uses

Articles are written in the first person by a named practitioner and reviewed by named reviewers. Where this site speaks as an organisation, as it is doing on this page, it speaks about its own editorial process only. It does not use an institutional voice to make claims about the world. There is no “we find” here, no “our research shows”, and no recommendation issued in the name of the Institute, because none of those would be true.

Corrections

If a figure here is wrong, or is right but described in a way that misleads, please say so through the contact page and quote the page and the sentence. Corrections to a factual error are made in the page itself and noted with the date. Where a claim’s tier changes because the evidence has changed, the change is recorded on the page rather than made silently.

Keeping it current

Two of the sources this site relies on most changed materially within eighteen months, in ways that altered what could honestly be claimed. That is normal for this field, not unusual. The locked figures are reviewed before each wave of new writing, and any page whose tier depends on a single review is checked when that review is updated. A page that has drifted out of date is corrected or taken down, not left up with a fresh date on it.

Independence

Nobody pays to be described here, to be graded here, or to be linked to from here. There are no sponsored pages, no paid placements, and no commercial relationships behind any listing on the find a practitioner page, which lists registers and associations rather than practitioners for exactly that reason. If that ever changes, it will be disclosed on the page it affects, not in a line at the bottom of this one.