PACE trial TSC and TMG minutes released

Touché @Sasha

 

Actually, that was just laziness on my part!

 

That's because you were too busy talking about cardigans!

 

Re: Step test and other 'objective' measures
I need to get all my bits together on this. I realised that I had a convo with @Graham (and Bob) about this 2 years ago that I had forgotten about (I then got distracted by the CFQ). @Tom Kindlon made comments on the Protocol in BMC Neurology that will also be very useful.
I expect Keith Geraghty has been looking at this too (that's where this thread started: https://www.s4me.info/threads/pace-trial-tsc-and-tmg-minutes-released.3150/page-2#post-56258).
If anyone knows of anywhere else this has come up, let me know.

In summary though, PACE auths reference Petrella et al, but don't appear to have used their equations. The equation they do cite in the Supplementary material in the Chalder Mediation paper doesn't then match the results they give, or even collected. So it's difficult to know what they did do.

If the test is self-paced, it won't necessarily give reliable results. And as far as I can tell from the Petrella paper, if the test is done in unhealthy individuals, it definitely won't give reliable results.

Passive actigraphy would have been better, but the auths already knew that it wouldn't show any improvement - because previous studies show that participants restrict their activity so they can get through the exercises, so overall activity doesn't change much.

 

Would it be helpful to be able to extract the numerical values from that graph from the mediation paper?

 

I think Graham's already done that. But I'm not sure it helps without knowing what they did.

 

Okay, I'm only just waking up, have no idea what I'm doing here, and inserted numbers into their equation that were plucked randomly from google, but I came up with a result of 142146.9. Rather more than 1. I'm guessing that shows I've messed up something.... sorry for just asking and hoping I can get someone to explain it to me instead of trying to work it out myself (my excuse is that others might be able to learn from my struggles).

From their appendix: "A measure of fitness was calculated as [Body Mass (in kilograms) x 9.81 x total step height (in metres) x 20] / time (in seconds) / %HRR, where %HRR = [(highest measured HR - resting HR) / (predicted max HR -resting HR)] x 100 (11) ."

84kg is the average weight of a Welsh man.

Google said: "In most step tests, the duration is variable (90 s-10 min), but the step height (23.0-50.8 cm) and stepping rate (22.5-35.0 steps/min) remain constant "

Random assumptions: 5 minutes of 23cm steps and 22.5 steps per minute = total step height (in meters) of 22.5 x 0.23 x 5 = 25.875. Time in seconds: 300.

Predicted maximum hr? Google suggests "The quick method is 220 – your age." ? Lets assume that their predicted heart rate is bang on. So just divide by one.

84[kg] x 9.81 x 25.875 [total height in meters] x 20 / 300 [seconds] / 1 x 100 = 142146.9

I've never understood what the numbers on the side of their fitness graph were meant to be for!

 

An A for effort there @Esther12 !

However:

- Total step height will be 0.4m (because there are two steps of 20cm each to alight).

- The "x 20" refers to the fact that you are supposed to do this exercise 20 times (up the 2 steps and down again).

- Petrella estimated that the time will be about 125s - which agrees with the alternative name of the test used by Chalder & co. (The "2-minute step test").

- It's self-paced, so although pt shld try to maintain a constant rate, they can go as fast or slow as they like (not so good though, coz Petrella found that the estimate worked best when the pace was moderate or fast, but that was with their equation and not the James one).

 

Thanks @Lucibee - D for attainment though. That's what I get for trying to blindly follow a formula without having any understanding of what some of it refers to (where does 9.81 come from?!).

I'm still getting a really high figure: 84[kg] x 9.81 x 0.4 [total height in meters] x 20 / 125 [seconds] / 1 x 100 = 5273.856

 

That's coz you're forgetting to divide by the %HRR

 

What figures would be right for the %HRR? Google failed me there, and I thought it looked like it was possible 'highest measured HR' would equal 'predicted max HR', and therefore we could just divide by 1.

 

I think you're not allowing for brackets. I get 0.5273.

 

Thanks both, yeah I was multiplying it all by 100 at the end. Got it now.

Does make their graph seem extra odd.

 

We need to stop trying to get the data to fit and actually find out what they did!

 

I'm brain dead at the moment, and almost catching up, but here are the notes I made at the time. There were two different ways of analysing the results of the step test: one was more objective, the other subjective. The big problem, of course, was that they were self-paced, and most of the potential comparison reference studies were not.​

In case anyone is interested, here's how far I have got. There are graphs for each, and both are flat and around the 20% mark. In both, the SMC group is marginally best. There are two calculations based on the step test.

The self-paced step test of fitness involves timing participants while they do 20 step-ups and step-downs (of 2 steps each), as well as gathering resting and post-exercise heart rates.

A measure of fitness was calculated as [Body Mass (in kg) x 9.81 x total step height (in metres) x 20] / time (in seconds)

(That part converts body mass into the force of gravity, which the patient is working against, in Newtons, multiplied by total lifting height to give work done in Joules, divided by time in seconds, to give power used in watts.)

then it is divided by %HHR=(highest measured HR - resting HR)/(predicted max HR - resting HR) x 100

In other words it is a measure of a patients actual power output in comparison with how much their heart had to work at it. Now what I can't find is any other set of referenced figures apart from a study by Petrella on older folk, but behind a paywall. I can't see any other way of working out what a healthy score would be.

The second is a measure of perceived exertion, which uses the Borg scale from 6 (very, very light exertion - almost resting) to 20 (utter exhaustion, where 19 on the scale is extremely strenuous, probably the most strenuous most people ever experience). They adjust it to give a score for physiological work done, by dividing it by the post-exercise heart rate as a percentage of the maximum predicted heart rate (which is 220–age for men, and 206–(0.88xage)) for women, then multiplying by 100 to give it as a percentage.

The Borg scale runs from 6 to 20 to give a rough idea of heartbeats per minute divided by 10 for a healthy adult at that level of work (i.e. 60 to 200). If a patient aged 40 pushed himself hard (perception of say 16) and had a final heart rate of 144, he would score 20% (if my maths is right). So it measures how hard the person thinks he is going, in comparison with how hard his heart is pumping. A high score would suggest that his perception was worse than his heart rate suggested, a low score would mean that his heart was working much harder than his perception of the task. By not subtracting the resting heart rate, it is much more a measure of heart rate than of increased heart rate, so it is affected by people having a high resting heart rate.

I thought I had found a good reference with Anderson but it was only for 17 patients (with bronchial problems). There also was a reference to Karloh, again only with a small group, but of healthy and bronchial patients. But it was for the Chester step test which is different.

 

Thanks @Graham - do you know whether they combined those two calculations in some way to obtain the score they used for the graph?
The ranges are very specific, and don't match anything we've tried so far.

I have a pdf of the Petrella paper...

 

Oh Esther, my faith in your invincibility has been shattered! It's the acceleration due to gravity, to change a weight in kg into a force in Newtons.

The values that I worked out from a large version of the fitness graph were

mean (calculated s.d. from displayed confidence interval)

SMC 1.83 (1.13) baseline and 1.99 (1.73) at 52 weeks
CBT 1.76 (1.13) baseline and 1.84 (1.36) at 52 weeks

Effectively, all "fixed fitness" unknowns are mass and age, where age gives theoretical max heart rate.

The two variables for each participant are the time taken and the increase in heart rate. Both of these are divisors, and for a fit person would be smaller, so the fitter someone is, the higher the score.

The only value I could find was in a study by Petrella, with a healthy 72 year old scoring 1.2 (the mean age of participants at the end of the PACE trial was 40).

 

There are two graphs - one marked Fitness and one Borg. They didn't combine the results. There is a description of all the tests in the appendix to Chalder et al's paper on rehabilitative therapies for cfs ...

They did think of using lifting weights for the Borg assessment, but decided resistance was futile (I'm baldy going).