Psychometric properties of the PROMIS® Fatigue Short Form 7a among adults with [ME/CFS], 2019, Yang et al

Paywall, https://link.springer.com/article/10.1007/s11136-019-02289-4
Sci hub, https://sci-hub.se/10.1007/s11136-019-02289-4

 

Where can we find this questioninnaire?

 

I think this is it, https://www.archives-pmr.org/article/S0003-9993(11)00680-0/fulltext#sec1.4

 

Here it is: http://www.healthmeasures.net/search-view-measures?task=Search.search (scroll down to 'view measure').

 

The questionnaire's 'homepage' - http://www.healthmeasures.net/index.php?option=com_instruments&view=measure&id=160&Itemid=992

 

This seems somewhat useful for mild idiopathic fatigue. And even at that, only somewhat. It doesn't actually give more information than simply asking for a single 0-10 estimate, so what's with the extra steps that add nothing?

Otherwise useless for ME, the obsession with fatigue is seriously getting pathological. I don't see how this questionnaire informs about anything. It's guesstimate on top of guesstimate on top of guesstimate. About as much precision as measuring something by squinting at the distance between two fingers and estimating size in eggs. Which size of eggs? Your choice, doesn't matter. Chicken? Ostrich? Whatever.

Still better than the Chalder scale, which isn't that hard but that's a very low bar.

 

I have not read the article yet so these comments are preliminary...

Jin-Mann S. Lin, one of the authors, was on the Common Data Elements Initiative subgroup on fatigue. The other CDE subgroups ("domains") included e.g. PEM, Sleep, pain, cognitive, neuroimaging, immune, biomarkers, QOL, etc

From the CDE discussion on the fatigue subgroup, I understood that PROMIS [edited to clarify - the PROMIS fatigue instrument] was intended as a generic instrument that could be used to compare the severity of fatigue across all instruments. But it had not been evaluated against the experience of fatigue in ME and could not be recommended at that time for that purpose. To be used in this way in ME, I understand the intent was for a generic instrument that did not "overload" the assessment of fatigue with aspects of ME that should be evaluated separately - e.g. post-exertional and cognitive impacts of ME. Historically, it seems that some of the "fatigue" instruments and fatigue descriptions have done that.

The CDE fatigue subgroup produced this summary document that gives additional context on the discussion and recommendation.
https://www.commondataelements.nind.../files/Doc/MECFS/Fatigue_Subgroup_Summary.pdf
This document also discusses the cross-linking that could be done between PROMIS and other instruments to improve cross study comparability.

Note that the fatigue group did not include the Chalder tool in its recommendations.

If I remember correctly from work a few years ago, PROMIS had some ceiling/floor effects when applied to ME. I'll be interested to see if they have addressed that in this document.

 

For that purpose it seems OK.

But I don't see what this adds beyond what a simple 0-10 scale would do. It's just a bit too generic to do more than that. Actually I'd say moving to a simple 0-10 scale but doing multiple ratings per day for a time period would probably be more useful when it comes to fluctuating illness.

This can even be adjusted to ask for more pertinent questions like rating PEM, pain, brain fog or overall sickness/malaise, that is terms more appropriate for the disease (in adapting it to ME anyway). More questions frankly don't change the quality of what is measured, their choice is too arbitrary and usually biased from the researchers not being able to experience it in the first place and instead making assumptions, might as well go with quantity instead.

 

I cannot se any point in these composite score systems unless it is specified exactly what they are to be used for. Are they to judge a current state, or make diagnosis, or measure a change with time or treatment? You want different measures for each, and for treatments very likely different for different types of treatment. In comparison to what we have had in rheumatology for 25 years this just seems wooly-headed.

 

Sadly there is very little that is not woolly headed with regard to this illness. It is a deep repository for grey wooliness

 

The HealthMeasures site cited above has more info about the intent of PROMIS.
http://www.healthmeasures.net/explore-measurement-systems/promis
This site states the tools are intended to be relevant across all conditions for the assessment of symptoms and functions. besides for cross-disease comparisons, I think the intent is to be able to use it for outcome assessment.
It also notes that NIH has been funding this effort.

For ME specifically, the issue of ceiling/floor effects is still an open question for me - that would impact its usability for ME. Cant speak to other diseases.

Edited to add this link just as an FYI- it shows the other PROMIS instruments/areas
http://www.healthmeasures.net/explore-measurement-systems/promis/intro-to-promis

 

It's this one: https://www.rheumatoidarthritis.org/treatment/acr-score/

 

The results of the this study indicate that this is indeed a problem. In my view, the authors downplay this soo much, that it becomes misleading.

They only assessed ceiling effects by looking at what % of patients scored the maximum (greatest fatigue score) on all of the 7 items. Because this was the case for only 1,8% of participants the authors argue there's no problem here. But by looking at the maximal score for all of the 7 questions, you only look at one of the many positions where there could be ceiling effects. Things become clearer if you look at the % of patients that scores the maximum for each individual item.

Table 2 shows that for three outcomes more than half of the patients already had the maximum score. That means that if they get worse they cannot indicate it on this part of the questionnaire. In contrast, the % of patients who had the lowest score was very low for all items. That means that if patients improve, they can indicate this on pretty much all items of the questionnaires.

Now suppose the fatigue scores are randomly redistributed: some patients get worse and an equally large group of patients gets better. On the items with ceiling effects, the improvements would look much bigger than the deterioration even though they are similar in size. Some patients cannot indicate their deterioration, while practically all the improvements are recorded on the questionnaire.

Now, the authors seem to argue that this doesn't matter much because this is only a problem on some items. There's almost always another item on the questionnaire where patients can indicate their deterioration. I, however, suspect there will be types of deterioration that only show on some on items and not on others. For some patients, their fatigue might get significantly worse but not enough to affect their ability to take a bath or perhaps it's a type of fatigue increase that is unrelated to thinking clearly etc.

The questionnaire wouldn't be able to record these forms of deterioration. And that's not simply a lack of sensitivity to these changes, because the questionnaire will record improvements of a similar type and size. So if fatigue scores are randomly redistributed it will only show the improvements. Treatments will seem more effective than they are and trials will be insensitive to patients getting worse.

I think that shows in their results. The questionnaire was able to differentiate the improvers from those that did not improve. But they noted an "inability to diferentiate between 'stable' and 'worsened' for a large proportion of participants."

For this reason, I think this questionnaire should not be used to measure treatment effects or track status over time. It might be helpful to compare fatigue scores across different illnesses, but for randomized trials, we would need a questionnaire that doesn't have ceiling effects and that is equally likely to record deterioration versus improvement.

 

It's unfortunate that table 5 only shows the comparison between those who improved and those who didn't improve. This was unlikely to be a problem in the first place. The suspected problem was ceiling effects; patients already scoring the maximum on many of the questionnaire items, making it difficult to record deterioration. And that is exactly what they found: "no significant difference was found between the “stable” and “worsened” groups." Yet na data is given for this comparison, there's no table due to "space limitations"...

 

I initially misread this as:

"Analyses were conducted using data from the Multi-Shite Clinical Assessment of ME/CFS study"

 

Thanks for this, @Michiel Tack. Very helpful

The other issue to consider is the CDC Multicam study only included patients that could get to the offices of the clinicians involved in that study. Those who are more severely ill and are bedbound are unlikely to be in that study and even some homebound might not be.

If this is not considering what the magnitude of the ceiling effects for PROMIS Fatigue would be in this more severe population, then the underestimate of impact would be even greater.

Unless I missed it, they did not specifically mention this as a limitation.

 

Completely agree. And is assessed in patients across the full range of severity