Physical function and psychosocial outcomes after a 6-month self-paced aquatic exercise program for individuals with [ME/CFS], 2025, Broadbent+

Have you seen this from the "Resources for Long COVID Researchers" from the Patient-led Research Collaborative.
https://patientresearchcovid19.com/resources-for-long-covid-researchers/

 

That raises an interesting question in my mind, one that I think has an easy answer but is very troubling: could it be that this has encouraged the mass publication of misleading conclusions in the pragmatic rehabilitation industry?

After all, most of their results are null, so there would be very little interest in publishing them, in an industry where publications numbers are everything. In a very real sense, an influencer industry, where publications and citations replace clicks and eyeballs. So they publish fake positive findings, whereas if negative findings were just as sought after, maybe we wouldn't have seen decades of tweaking and tuning turning negatives into false positives.

Of course there are easy incentives here in encouraging misleading conclusions to be published, lots of money and biases involved, but if the alternative is to not get published at all, it explicitly punishes a field that is trying to find itself, and instead steers it into building a giant fantasy instead.

 

I have seen this, but thank you for bringing it up again. I was hoping there might be something more directly ME/CFS related, though this LC resource is a great starting place

 

I don't really see a problem. To me it is fairly obvious that this community has the resources to advise researchers. It does not take very long to hang around and read some threads to get an idea of how things work. I have dipped into forums over a period 25 years and never found it that hard to engage with the people who are seriously interested in dialogue.

Recipes are a mirage in this area. Each study calls for different entry criteria and outcome measures. I have designed multimillion dollar drug studies and you take nothing as fixed. You sit down, sometimes day after day, thrashing out all the pros and cons every option you can think of and generally there are tricky problems that require creative solutions to get the best answer from.

The usual problem is not that researchers come here not knowing agreed procedures. The problem tends to be that they feel they have to stick to agreed procedures because it looks respectable.

 

I seriously hope that nobody uses that. Asking about 209 symptoms is not a good place to start.

 

Unfortunately I think your speculation is actually just spot on. And I think this is a problem that extends way beyond the pragmatic rehabilitation industry.

I can’t share details about this but just as a general indicator, I know at least two grad students at different institutions who are both forced to do work pertaining to different cell subtypes which just…don’t seem to actually exist.

Both projects were something that seemed plausible based on prior data. That speculation won multiple grants and now their PIs are just in too deep to acknowledge that nothing can be replicated and there’s an alternative explanation that is less flashy. They’re basically spending years convincing these poor grad students that the reason they can’t find more evidence of the celltype is because they’re not good enough scientists.

Most researchers aren’t even motivated by personal financial gain and prestige—it’s really just that getting continued funding requires proof that your previous grants resulted in exactly the data you promised and churned out publications in good journals.

You simply can’t show that you “wasted” thousands of dollars on a negative result if you want to keep your lab funded. It’s awful but many researchers see it as a necessity.

I’m being a little bit doomer-y here, there is wiggle room for negative findings and this isn’t an inevitability. But I’ve seen it enough times in my short research career so far to get disillusioned pretty quickly.

 

I was interested to know how severe the patients in this study were. So I emailed Dr Broadbent with a few questions about SF36 physical function scores, which were not reported in this paper. She explained that they will be reported in another paper which they expect to publish later this year. With her permission, I am sharing her answers to my questions below. Her answers are bolded.

Dr Broadbent clarified that baseline SF36 physical function scores: “for the CON group were 47.1(21.1) and for the INT group, 50.6(27.6).”

To put these values in context, here are the mean baseline SF36 PF scores (standard deviation) for the exercise group in a few other studies that will be familiar:

Wearden 2010 (FINE): 30 (18)
White 2011 (PACE): 37 (15)
Clark 2017 (GETSET): 47 (22)
Broadbent 2025: 51 (28)

There are studies with higher baseline SF36 PF scores e.g. CBT studies like Tummers 2012, Knoop 2008, Wiborg 2015, Janse 2018. Baselines ranged from 39-54 in the different arms of Jason et al.’s 2007 RCT.

Van Campen et al. 2020 found that 95% of their cohort who were deemed to have moderate ME/CFS using International Consensus Criteria had SF36 PF scores in the range 30-59. So Broadbent et al.’s average patient had a baseline SF36 PF score that would place them in the moderate severity category.

Dr Broadbent also explained that there was a within-group difference in SF36 PF scores in the INT group comparing pre- and post-intervention scores, but not in the CON group. She emphasized that “this is one finding only and the data is not officially published”.

Edited to bold Dr Broadbent's answers so that it is clear what came from Dr Broadbent and what I have written.

 

Thank you for asking and reporting it here!

Did she give any reasons for why the scores were not included in this publication?

 

You're welcome. No, she just explained that they would be written up separately.

 

Since there are no/almost no between-group differences, I was interested in what the participants might have valued in the intervention. In a qualitative study of 11 participants in the pilot study of aquatic exercise, Broadbent et al. 2020 report:

 

Is this info from her or did you dig this up?

If van Campen 2020 is this study, I’m very skeptical of any numbers they provided. These are the number of steps for each category:
The mean (SD) number of steps per day was 8235 (1004) for mild, 5195 (1231) for moderate and 2031 (824) for severe disease.

Anyone with ME/CFS will tell you that those numbers are completely unrealistic. I believe the average for the entire population is about 5,000.

 

That makes no sense to me, but I guess they want more publication points?

 

That info is not from Dr Broadbent, it is from table 3 of this study https://pmc.ncbi.nlm.nih.gov/articles/PMC7551321/

I have no reason to be sceptical of the numbers. Are there other studies that found lower steps per day for each severity? You could have a look at Rekeland et al.'s data, though the sample is very small.

 

That’s the one I linked.

As I said, the average daily step count for an adult is 5,000. Something is off if the average person with mild ME/CFS has a daily step count of 8,000, and 5,000 for the average moderate person.

 

In Fluge et al.'s rituximab trial they report mean steps per day in the low-mid 3000s for a cohort with mean SF36 PF scores in the low-mid 30s. For the group who got worse, they had a mean SF36PF of 27 and mean steps around 2500/day. That would all be consistent with van Campen's numbers. From the 2019 Fluge et al. paper https://pubmed.ncbi.nlm.nih.gov/30934066/:

 

Apologies for all of the questions.

Did Dr Broadbent provide you with the reference to van Campen 2020 or was that something you found yourself? What about the other studies in your post about their reply?

PS. In an ideal world, she would sign up here and let us ask questions without a patient as a go-between.

 

The influencer industry model doesn't really allow for that. Such a journal would have a very low impact factor, making it unappealing. Sometimes our psychobehavioral overlords publish in their own self-managed Journal of psychosomatic medicine, which isn't much different from the owner of a newspaper publishing an editorial they wrote, but they do get lots of publications in reputable journals like the Lancet and, especially, the BMJ.

This is basically the difference between being published on a random blog and a major newspaper. PACE got a lot of weight mainly because it was so expensive, but also because it was published by the Lancet. And the same for the awful MMR paper, it would have been thoroughly ignored if it had been published on a random blog, or even in the Journal of Romanian experiential medicine, but it was defended by the editor-in-chief of a major academic journal for 12 years, and that's why it did so much harm.

Academia really has to be thought of as an influencer industry, where being published in a major journal is like going 'viral' (except nor organically, it's more like the old industry model where a few producers make all the decisions about who gets prime billing) and getting big funding is like getting a big sponsor. Exactly like TikTok. It's not about substance, it's about shelf presence, like in a grocery store. Companies selling products pay grocery stores to have their products prominently displayed where consumers will see them first.

No one wants to be on the lower shelf in the middle aisle. Such a journal would be the equivalent of being stocked in the back room, it would be a mark of shame simply because of how the influencer industry is built on display prominence first, and substance last.