Two documents sit side by side, and at first glance they contradict each other.
One is Project Report No. 160 from AIHTA, Austria’s health technology assessment agency, published in Vienna in November 2024. Its conclusion: the strongest evidence is for multi-day and multi-week training courses.
The other is the meta-analysis by Jayawickrama and colleagues, which we covered in August. Its finding: anti-bias training changes what students say about fat patients, not what shows up when you measure below the level of self-report.
Put those two sentences next to each other and you get the familiar shrug: studies disagree, pick the one you like. That is not what is happening here. The two do not measure the same thing, they do not count in the same way, and they were not even written at the same time. Sort that out and no contradiction is left. Something less comfortable is.
What the Vienna report is
AIHTA is neither an advocacy organisation nor a treatment provider. It is Austria’s HTA agency, the body that prepares evidence for decisions inside the health system. For the German-speaking world that matters. Most of the sources this magazine relies on come either from the movement itself or from organisations with a commercial interest in treatment. This is the first state-backed evidence report on the subject in German.
The authors are Sarah Wolf and Julia Kern. The systematic search ran from 21 to 23 May 2024 across five databases. For the first research question, which strategies are recommended, they included 13 guideline papers and 13 reviews. For the second, what actually works, 30 primary studies.
The report is unusually clear about its own limits, and those belong in any citation of it. The effectiveness analysis explicitly claims no completeness: only studies published from 2014 onward, and only from Europe, North America, Australia or New Zealand. The trustworthiness of the guidelines used was not systematically assessed. Nine of the 13 reviews are non-systematic, and most of those give no information about their methods, their literature search, or what they included.
Above all, the evidence base is overwhelmingly American. Nine of 13 guideline papers, six of 13 reviews and 19 of 30 effectiveness studies come from the United States. AIHTA writes in its own critical reflection that transferability to Austria is therefore questionable, and it does not blame culture alone. In the US there is an established debate and correspondingly many tested measures; in Austria the focus has so far been on the medical perspective. The same caveat applies to Germany.
What it says, and what it does not
In the summary, one word about health professionals should not be skimmed past. For multi-day seminars or courses, the effectiveness analysis reports “controversial” results, based on three studies. Beliefs about fat patients improved. There were no significant changes in external weight bias, in attitudes toward patients, in communication with them, in their quality of life, or in their health status.
For students the picture is somewhat better. Long-term seminars showed significant improvements; one-day courses were inconsistent. Training improved students’ ability to recognise discrimination and strengthened empathy.
From this the discussion section concludes that multi-day training has the most evidence with the strongest indications of positive effects, and the conclusion states that the strongest evidence currently exists for multi-day to multi-week training. Both are relative statements. “Most evidence” here means three studies for staff, a few more for students, and effects that the conclusion itself describes as “mostly only short-term.”
One phrasing that circulates in coverage of the report is not in it: AIHTA nowhere calls multi-day training “most promising.” It says that is where the most evidence sits. Those are different claims.
Why this is not a contradiction
Three reasons, in ascending order of importance.
First, chronology. The AIHTA search closed on 23 May 2024. The Jayawickrama meta-analysis went online on 8 October 2024. The Vienna report could not have included it. These are not two camps. They are two points in time.
Second, method. AIHTA ran no meta-analysis. It summarises narratively and counts how many studies showed a significant change. Jayawickrama pools effect sizes across 35 studies and reports confidence intervals. A counted “three of three studies improved” and a pooled g are not competing answers to one question. They are two different questions.
Third, and this is the actual point, the endpoint. The AIHTA report lists the measurement instruments used across its included studies: twelve for explicit weight bias, six for internalised bias, four for empathy, and exactly two for implicit weight bias, both of them versions of the Implicit Association Test. The evidence Vienna draws its statement from is therefore almost entirely self-report. And self-report is precisely where Jayawickrama also finds an effect: explicit g = -0.31, confidence interval -0.43 to -0.19. What does not move is measured implicit attitude: g = -0.12, confidence interval -0.26 to +0.02, not significant.
So both reports say the same thing. What people say about fat patients can be shifted by training. What sits underneath was barely measured in Vienna, and was measured in the meta-analysis and did not shift.
Two further details point the same way. AIHTA notes that follow-up periods in most included studies were too short to show long-term effects. And it warns explicitly that training which is not sensitive and well structured can unintentionally reinforce existing bias. That is the same content as Jayawickrama’s prediction interval, which runs from -0.93 to +0.31: the next training can also go the wrong way.
Anyone citing either source as “training works” or “training does nothing” is citing it wrong. Neither sentence appears in either document.
The part they share that nobody quotes
The AIHTA report is not only an effectiveness analysis. Its first half collects recommendations, and there sits a checklist more concrete than anything in the training debate. Sixteen sources, eight guideline papers and eight reviews, recommend structural measures:
- step-free, accessible entrances
- wide seating without armrests in waiting rooms
- accessible toilets with grab rails
- a private place to be weighed, and the option of being weighed without having to see the number
- blood pressure cuffs, tape measures, examination instruments and gowns in appropriate sizes
- scales with a measuring range of 150 to 300 kilograms
- examination tables, chairs and hospital beds of adequate width and load capacity
- no stigmatising imagery on walls, in magazines, on the website or on social media, meaning no isolated body parts, no headless bodies, no stereotype scenes
Alongside those come the policy recommendations from ten guideline papers and seven reviews. One of them is new for the German-speaking record: discrimination on the basis of body weight should be formally recognised as a legitimate social concern and written into anti-discrimination law, which already prohibits discrimination on other personal characteristics.
We know that demand. We described it in our piece on Germany’s federal anti-discrimination survey, where it comes from an advocacy association and from the researchers commissioned by the government agency. What is new is who else is making it now: a state-backed HTA agency, in an evidence report, addressed to decision-makers inside the health system.
The point
And here is the sentence the whole debate hangs on. The report’s conclusion states that interventions for children and adolescents, and structural measures, were not examined in any of the identified studies. The discussion says it again at greater length: the effectiveness of recommended strategies such as communication-based interventions, changes to the environment, and patient empowerment strategies has not yet been studied.
There is not one effectiveness study on whether a waiting room with chairs that fit and a scale with an adequate range changes anything about care. Not because it was tested and failed. Because nobody tested it.
The training debate is loud because training is the only thing that gets evaluated regularly. It is cheap to demand, easy to measure, it produces certificates of attendance, and it leaves nobody with an invoice for new examination tables. The measures you could verify in an afternoon, by walking into a waiting room and looking at a scale, have no evidence base because they were never made a research subject.
That is not a defence of training and not proof for equipment. It is a statement about what has been considered worth measuring for the past decade.
Placing the source
AIHTA is part of the health system and argues from inside its frame. The report treats overweight and obesity as medical categories, lists treatment recommendations, and draws in part on guidelines from professional societies with an interest in treatment. Readers should know that.
What it recommends is nonetheless notable. The Health at Every Size model is named repeatedly as a central approach, on the grounds that it places the focus on health and wellbeing rather than weight reduction. Re-evaluating interventions and guidelines built on a weight-centred approach, meaning BMI as an indicator of health, also appears as a recommendation. That is not a position one expects from an HTA agency.
External review was carried out by Friedrich Schorb of the University of Bremen and Martina Hermann-Thurner of Austria. The report states explicitly that the external reviewers are not co-authors and do not necessarily agree with its content. The authors declare no conflicts of interest.
One small internal inconsistency, for completeness: the results section lists twelve instruments for explicit and two for implicit weight bias, the discussion section says ten and one. It does not change the order of magnitude.
What follows
Anyone in Germany or Austria who wants to do something about weight stigma in healthcare has, after these two documents, two options. The first is well studied and works in a limited, short-term way on what people say. The second has not been studied at all, but can be checked without a study.
We will keep citing both sources. Neither one as evidence that training works or does not.
Sources
- Wolf, S. and Kern, J.: Strategien zur Reduktion von Gewichtsstigmatisierung bei Personen mit Übergewicht oder Adipositas im Gesundheitswesen. AIHTA Project Report No. 160, Vienna, November 2024. Full text: https://eprints.aihta.at/1547/
- Jayawickrama, R. S. et al.: Interventions to reduce weight stigma in healthcare students. Obesity Reviews 26(2):e13847, online 8 October 2024. DOI 10.1111/obr.13847
Image: original editorial graphic.