Preprint Large-scale psychometric assessment and validation of the Modified COVID-19 Yorkshire Rehabilitation Scale C19-YRSm…, 2025, Horton et al.

[SNT Gatchaman]

Large-scale psychometric assessment and validation of the Modified COVID-19 Yorkshire Rehabilitation Scale C19-YRSm patient-reported outcome measure for Long COVID or Post-COVID syndrome
Mike Horton; Adam Smith; Stephen Halpin; Ruairidh Milne; Darren Winch; Clare Rayner; Roman Rocha Lawrence; Darren Greenwood; Nawar Barkley; Rachael Evans; Joseph Kwon; helen dawes; Connor Wood; Paul Wiliams; Harsha Master; Mae Mansoubi; Mir Ghazala; Johannes De Kock; Jordan Mullard; Mike Ormerod; Stavros Petrou; Daryl O'Connor; Rory O'Connor; Manoj Sivan

BACKGROUND
The C19-YRS was the first condition-specific scale for Long COVID/Post-COVID syndrome. Although the original C19-YRS evolved to the modified version (C19-YRSm) based on psychometric evidence, clinical content relevance and feedback from patients and healthcare professionals, it has not been validated through Rasch analysis.

OBJECTIVES
To psychometrically assess and validate the C19-YRSm using newly collected data from a large-scale, multi-centre study (LOCOMOTION).

METHODS
1278 patients (67% Female; mean age = 48.6, SD 12.7) digitally completed the C19-YRSm. The psychometric properties of the C19-YRSm Symptom Severity (SS) and Functional Disability (FD) subscales were assessed using a Rasch Measurement Theory framework, assessing for individual item model fit, targeting, internal consistency reliability, unidimensionality, local dependency (LD), response category functioning and differential item functioning (DIF) by age group, sex and ethnicity.

RESULTS
Rasch analysis revealed robust psychometric properties of both subscales, with each demonstrating unidimensionality, appropriate response category structuring, no floor or ceiling effects, and minimal LD and DIF. Both subscales also displayed good targeting and reliability (SS: Person Separation Index (PSI)=0.81, Cronbach alpha=0.82; FD: PSI=0.76, Cronbach alpha=0.81).

CONCLUSION
Although some minor anomalies are apparent, the modifications to the original C19-YRS have strengthened its measurement characteristics, and its clinical and conceptual relevance.

Link | PDF | Preprint: MedRxiv [Open Access]

 

[Utsikt]

Why is this a «rehabilitation scale»?

CONCLUSION
Although some minor anomalies are apparent, the modifications to the original C19-YRS have strengthened its measurement characteristics, and its clinical and conceptual relevance.

And how has this paper proven that clinical and conceptual relevance?

 

[Utsikt]

The link in the Appendix doesn’t work, so here’s another one for the updated scale:

https://c19-yrs.com/wp-content/uploads/2024/07/Modified-C19YRS-Self-Report.pdfhttps://c19-yrs.com/wp-content/uploads/2024/07/Modified-C19YRS-Self-Report.pdf

 

[rvallee]

Although some minor anomalies are apparent, the modifications to the original C19-YRS have strengthened its measurement characteristics, and its clinical and conceptual relevance.

Sigh. No, you're not measuring anything. I know I'm the only one here bothered by this but these people are supposed to be scientists and scientific words are supposed to have accurate meaning.

Measurement is the quantification of attributes of an object or event

You're not quantifying shit. There is no quantity involved here, let alone as a real number, in both the mathematical and general sense. You're qualifying, using ratings. Those are not at all the same thing, and those do not amount to measurements by any definition of the word anymore than IMDB ratings are a measure of anything.

And without a ground truth to compare to, all of this is useless. This is not what validation means in a professional context. Good grief this profession needs to be put in a wood chipper. This would be bad enough if it wasn't for AI, and the whole issue of 'overfitting', wasn't so well-known. You can make any data you want fit any model you prefer, it doesn't matter: all models are wrong, some are useful, and this one only has limited use.

 

[rvallee]

Looking at the questionnaire, it's not bad in itself. It just doesn't add anything that dozens more before it haven't done, because it's about 90% identical as all the others. It certainly doesn't add anything that any prior version could have, because it's completely generic.

Pretending like this measures anything is where things go wrong. It's a rough guide, needs a lot of interpretation, where all the utility comes from asking about symptoms, which is the exact same thing as just having patients self-report. With the problems being symptoms, of course this has some use, but then what does this add that any other similar questionnaire asking the same questions? Nothing, of course.

People suffering from symptoms go see health care professionals, so of course using symptoms as a guide has relevance, but it leaves us no further than things were 50 years ago, or even centuries ago, when the same questions could be asked without pretending that they can add up to some arbitrary score.

FUNCAP is better than this in most aspects. But everyone wants to develop their own instruments based on their own models. And for what?! No reason at all. Just busywork creating the illusion that work is being done.

 

[Utsikt]

The granularity also makes to make no sense in a rehab environment. It’s a scale from 1 to 4.

The genius of FUNCAP is that the granularity increases with the severity, so it’s possible to differentiate between severe and very severe or worse. It’s kind of like a log scale.

 

[Sean]

I know I'm the only one here bothered by this

I don't think that is true.

Directly or indirectly, concerns about methodology (which measurement definitely falls under) is the basic underlying issue dealt with here about almost everything coming out of the research.

Particularly for the psycho-behavioural stuff, as they are by far the worst offenders.

I don't care what the correct explanation is about ME/CFS. I am and always have been completely agnostic on it. What I do care about is how it is determined to be the correct explanation.

What is it that researchers claim to be measuring, and how are they measuring it? For which we still have no satisfactory answers in any direction.