More PACE trial data released

@Trish - the more I reflect on this, the more it doesn't make any sense. The whole point of using Borg is to make sure that pts (participants) are putting reasonable effort into the test, otherwise it is not going to give a good estimation of fitness. We don't know how it was being used here, because it doesn't then get included in any calculation of fitness. In Petrella's paper, any score less than 8 indicates that the pt is not putting enough effort into the test for it to be valid anyway.

One would hope that the researchers realised that it wasn't really telling them anything useful, and that's why they ditched it - but I wish they'd just say that.

If they'd done it properly, it could have been a useful measure of VO2max. If it was then done on consecutive days, it could have been used to indicate PEM. But they didn't. It just seems pointless.

 

Yes, they ignored it ... because it did not agree with the findings from subjective outcomes.

 

Are we sure a higher Borg score necessarily corresponds to an improvement? If I climb a set of steps today and it it feels like the effort it took me was a 10, and I then climb the same set of steps tomorrow and I rate the effort involved as feeling like 15, that suggests I struggled more the second time around, rather than improved?

 

So it's completely irrelevant, then. Nothing of value can be concluded outside of the ideological (and basically teleological) framing of Wessely's "the only predictor of illness severity is strength of belief", and even then the value is essentially in giving free rein to conclude whatever the researchers wanted to conclude 30 years ago when they began promoting this nonsense.

I'm not a scientist, but I find issue with the idea of researchers applying entirely useless tests as part of a controversial trial. It really speaks to how unserious the whole thing is. There simply aren't any circumstances in which this would give useful information, meaning it was a conscious and deliberate choice to spend time and an enormous budget on useless busywork.

Ultimately the only power PACE has is its price tag. This is the only reason it has outsized influence. And clearly most of it was simply wasted, spending the large funds they received in order to fulfil the only convincing feature of the trial: sunk cost.

 

And which does not take into account the wildly fluctuating nature of the disease. But then again this was universal through the trial so at least it's consistent in not caring one bit about being rigorous.

This puts to shame the way Volkswagen cheated on emissions testing. The level of cheating here is simply incredible to believe, likely the main reason why the facts are rejected, it's simply impossible to believe anyone who would such things, even be given full approval for it. And yet here we are.

 

Surely an 'improvement' would be a reduced rate of perceived exertion/effort (RPE) for the same or greater distance?

You could presumably convert RPE to a number between 0 and 1, and then divide the distance by that RPE to get a comparative score of before and after?

 

I only looked at @Lucibee's post very quickly and may have jumped the gun. Is it that lower Borg scores do correspond with improvements?

 

Good point. I meant only "improvement in Borg score" - any implication of an improvement anywhere else is purely in your own mind.

For ease of intuition, positive numbers on my "improvement" graphs indicate that the test seemed easier at 52 weeks and negative numbers indicate that it got harder.

And it's a lower Borg score that indicates an improvement here - because the test seems easier. Although you're right - who's to say that being able to exercise harder is actually more of an improvement than finding the test easier to perform...

[sorry - lots of edits]

 

RPE (Borg) was only done for the step test, and not the 6-minute walking test (the distance I presume you are talking about). It would be useful to have the comparative data from the step tests, because on their own, as @Snow Leopard points out, they are pretty much uncomparable.

Ideally, you'd want to know how long it took to do the 20 steps, baseline and final HR, and RPE. But then, doesn't ME do strange things to HR under exertion...?

 

Thank you. All really good info.

 

Actually, in the mediation analysis paper (Chalder et al, 2015 https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(14)00069-8/fulltext) they adjusted the Borg scale by dividing by the post exercise heart rate.

The appendix to that paper (available as a pdf from the link above) says:

This is how the Borg score is presented in Figure 2 of that paper.

 

It could be an indirect measure of motivation during exercise testing, I wonder if an increase (eg more effort from baseline) in Borg score at 52 weeks is associated with an increase in 6MWD at 52 weeks?

 

Sort of, but I think it's impossible to make any inferences about what is happening here [+ve Borg diff = test is easier at 52 wks vs baseline; -ve Borg diff = test is harder at 52 wks vs baseline]:



And here's PF diff vs Borg diff - again, there's definitely an association, but the data are clear as mud:


I'll edit to add that any sort of association between Borg and the walking test is dangerous because Borg was done with the step test, not the walking test. The two tests were done on different days at the start of the trial, but the protocol suggests that they were done on the same day later on in the trial. There are just too many unknown variables to take account of here.

 

The key thing about the Petrella paper for me is this graph:

Because any kind of physical disability completely messes with the predictability of the step test. Their main sample was *healthy* older adults. When they included those with other conditions - hypertension, post hip fracture - they get mush for data.

 

I can sort of see some logic in this. The Borg score alone is very nebulous, being simply a person's self-rated perception of how hard they found a physical exercise. Even for a given fit person such a rating will vary from day to day, situation to situation, on top of the world or down in the dumps. And it probably varies a lot from person to person.

So I see the above as an attempt to calibrate someone's Borg rating against something more objective, and so gauge how reliably their self-reporting of physical stress relates to their body's physical indications of physical stress. High normalised HR with low Borg score probably suggests someone highly tolerant to discomfort. Low normalised HR with high Borg score maybe suggests someone with poor tolerance to discomfort. And other combinations appropriately in between.

But I suspect there will be a bucket load of complicating factors here, especially if a given person's rating accuracy can maybe vary at different times.

 

My following comment very tentative, but ...

The general trend for weighting in the top-right quadrant: This suggests people who did achieve more physically, but self-reported (Borg) that they found it harder. To me this strongly suggests it likely the consequence of being driven to strive harder, putting more effort in but not necessarily improved actual physical function. i.e. The result of CBT, GET, etc. (What trial arms are represented in your charts @Lucibee?).

If they had experienced real physical improvement, then there would probably have been more of a tendency for greater weighting in the bottom right quadrant.

 

@Lucibee: Do those different quadrants have any broad association with anything in particular? Trial arms for instance?

The differences, what differences are these? Baseline vs weeks? Arm vs SMC?

 

+ve numbers for Borg difference indicate that it became *easier* not harder (RPE score goes down over time). I reversed the polarity because I thought it would make more intuitive sense (+ve indicating improvement, test seeming easier, participant getting fitter) - but clearly it doesn't. Oh well.

I haven't split by group.

Diff is difference between baseline and 52 weeks.

 

Inspired by @Lucibee I've done some charts, including breakdown by trial arm.

Note that my charts are "upside down" compared to Lucibee's, as I've not inverted the Borg values.

Any row with a zero value for Borg was clearly invalid (minimum valid value is 6), so omitted those rows.

Also omitted any rows with zero values for 6mwt, as presuming them invalid also.

Differences are 52 weeks minus baseline.

The thing that strikes me is that all arms show they same broad pattern: weighted to the right, suggesting a majority walked further, with a slight tendency for a majority finding it slightly easier. Given the key outcome metrics are relative, compared to the SMC arm, my broad impression is that there is probably not much difference. But my Excel skills fail me beyond this point (at least at this time of the day anyway), to pin down the hard detail of this further.