Discussion in 'Latest ME/CFS Research' started by Dolphin, Jan 21, 2011.
Well done to Angela for posting her comment, "Questions about 'CFS' and 'ICF' selection criteria of cohort" http://www.biomedcentral.com/1471-2180/11/2/comments
I think when they give the SF-36 physical function scores for the groups, they actually mean the Physical Composite Scores (PCS) because otherwise the unaffected twins are quite/very disabled.
So presuming they are PCS scores, the CFS/idiopathic chronic fatigue cohort is quite mildly affected. Unfortunately this is what happens with these population studies: they pick up lots of mild cases/people who may not have "proper" ME/CFS.
In the study Table 1 does pretty cleary quote Median 'SF-36 physical function' (Interquartile range) Affected twin: 41 (27-48); Unaffected twin: 48 (39-52). And from those figures the unaffected twins look quite disabled, which is very odd. guess this is one for the authors to clarify.
As for case definition, the devil is in the detail. They say they've used Fukuda, as modified by the 'Identification of ambiguities in the 1994 chronic fatigue syndrome research case definition and recommendations for resolution' 2003 paper and although this was first-authored by Reeves it was on behalf of the International CFS working group that included Lenny Jason, who is a co-author of the paper. This 2003 'Ambiguities' paper doesn't make such big changes. It's the later 2005 'CDC Empirical' paper, also by Reeves, that introduces the use of the SF36 Role Emotional subscale and in the process effectively changes the case defintion, and it appears this definition wasn't used.
That said, we don't know how this current study implemented Fukuda eg which scales were used and what thresholds esp for fatigue levels and activity [this is the sort of thing 2003 ambiguities paper said needed clarifying].
Yes, but they also say "mental function". But there is no such thing as a mental function subscale of the SF-36. Again, I think they meant Mental Composite Score (MCS). The scores of the unaffected twins would be close to population norms then. If you want to only use two scores rather than eight, that is the norm.
I sent in this comment. Whether they put it up, I don't know. If not, I might contact the authors. I'm fairly sure it's not the physical functioning subscale:
See what you mean and they must have something wrong in that table, probably what you say. In which case - why don't they use the SF36 PF subscale, it's much more relevant! Glad you submitted a comment, hopefully the authors will now clarify.
Thanks. The PCS uses data from various SF-36 subscales including the physical functioning subscale. But I much prefer when they give the individual eight subscale score, like was done in the Nunez paper.
Lenny Jason did some work comparing componenets of the SF36 with CFS status. As I recall, Physical Function, Role Physical and Vitality correlated best with CFS, while the Physical Composite Score correlated weakly at best so it's not a particularly relevant measure to use for comparing CFS with non-CFS.
Firstly, I'm quite excited by this general approach - i.e. a broad search for evidence of viral infection without any prior assumptions made on the actual identity of the virus, plus its an interesting, if disappointing, read.
Although the cohort made be an important issue I would appreciate someone with a science background reviewing the methodology used.
The following is a quote from the paper :
"There are several reasons why a chronic infection important to the etiology of chronic fatiguing illness could have escaped detection. For example, viral titers might be beneath the detection limit of our approach, the infection might be intermittently active and not during our sampling, and a salient infection might occur in body compartments or tissues where viral particles do not appear in blood. It is also possible that a salient infection occurred earlier in life, was cleared, but the infection sequelae are responsible for clinical state. Such infections, in the case of known viruses, can in many cases be detected via serology. Finally, it is possible that chronic fatiguing illness represents a similar clinical endpoint for multiple different disease etiologies (which may or may not be infectious in nature) and that etiological heterogeneity effectively lessens the probability of detection."
Plus, given that the current authors clearly state that the Lo/Alter virus was a different MLV from XMRV, whereas others state otherwise, it appears the phylogeny is more a statement of personal belief/philosophy than any clear cut scientific distinction. I'm pretty sure its too early to be able to state definitively whether they are or are not part of the same family or to fully describe/sequence this family.
That being the case, I wonder how well founded it is to discard identified BLAST sequences on the basis that they represent known human or non-human sequences. Wouldn't a retrovirus that had recently jumped species from mouse to human and perhaps represents a hybrid contain both human and no human sequences?
Can you let me know the study?
I'm guessing this might be in the context of diagnosing the condition i.e. using particular thresholds to diagnose.
However, that's not the only way scales can be used - they also give an idea of general health and functioning.
I'm appending a formula to calculated the PCS. The formula I think may vary from national group to national group so that the country has a norm of 50 and SD of 10.
But one can see that it largely uses five of the 8 SF-36 subscales with the most weighting on the physical functioning scale.
The PCS here (presuming that is what it is) has a median of 41, or 0.9 SDs from the population norm.
Here are some scores from other studies which show the values are generally a lot lower for CFS samples. Remember the population norm for the PCS is 50 with a SD of 10 (this is different from the eight subscales which are marked out of 100 and the population average is much higher).
My first thought is: needle in a haystack. This study should by no means be the last word in such studies.
GBV-C is certainly an interesting link and may well be a causative factor for some patients.
Measuring substantial reductions in functioning in patients with chronic fatigue syndrome
Yes, it was used in that context, but it seems similar to this context too. What's needed really is a measure of how severe the CFS case is relative to his/her perfect control - the unaffected twin. Lenny's comparison suggests that SF subscales like PF and vitality do that better than other SF measures, but from the data you quote above, the PCS still picks up substantial changes. If the table figures are indeed PCS as you suggest then the study has some very mild cases of CFS.
I'm looking at that and can't see the PCS mentioned at all. So I think you may be misremembering?
Often studies will eight use the 8 subscales or else the two composite scores (PCS, MCS) (need the eight subscale scores for that) but not both.
The twins have a median score of 48 (slightly before the norm) (vs 41 for the CFS/CF group). There is of course a statistically significant difference. But on the SF-36, people with CFS score very low. It may be the survey which was used to say that people with CFS are more severely affected than various other groups but I'm not sure.
The upper figure of the IQR for the CFS/CF group, 48, means that 25% have a PCS of 48 or higher. Making a diagnosis of CFS very questionable or disabling chronic fatigue very questionable.
This is really kind of an astonishing result. The WPI has reported finding evidence of many viruses......using their arrays and yet they found almost nothing. Dr. Chia reports he finds enteroviruses, Dr. Lerner regularly finds EBV. Dr. Chia finds much more evidence for enteroviruses in the gut than in the blood - but he can find them in the blood. And yet nothing!
I wish I knew how good their technology is....They believe it's quite good - The deep Roche 454 sequencing, combined with the efficient enrichment of virus particles, makes it likely that most viruses present in the serum of these individuals were detected. XMRV has shown how hard it is to interpret results.
I assume that the WPI is using different technology. I hope they will publish their pathogen array findings at some point.
This technology apparently only picks up evidence of active infection in the blood - so if you have past (inactive) infection (at least in the blood) it wouldn't show up.
The Dubbo studies suggested that it is the 'infection sequelae' that is responsible for CFS - not the original infection. They showed higher cytokine levels early in the infection of people who later came down with CFS compared to those who did not. So an alternate explanation is that those cytokine levels somehow disrupted normal functioning - resulting in this chronic illness state.
HHV6 Foundation helped fund a latter Dubbo study to intensively look at herpesviruses in those people and it same up empty.
This makes sense to me - lots of different types of people..ending up with similar symptoms caused by different things.
This figure shows BLAST sequences classified by the closest homologues - that come from about 25 different virus families with the Flavaviridae easily, easily being the most dominant.
Actually almost 10% of patients with confirmed GBC-V virus is not bad...it is not known to cause any disease, though.
This was from a major RO1 grant from the NIH; they don't come to CFS very easily.
The CDC looked for GB-C in CFS in 2005 and came up with a similar finding...
Another study with the same cohort
Another study with the same cohort
Full free text at: http://www.plosone.org/article/info:doi/10.1371/journal.pone.0005805
A problem with the Dubbo cohort is that 98-99% were better or didn't have CFS at 2 years. If one wants to see what causes CFS to stay, patients who get stuck with it/don't get better could be said to be more interesting. It's useful you highlight it, I'm just giving an alternative explanation.
Agreed.....I don't mind if they got better - I was better two years later - but I do recall most of them worked their way out of it....
I wonder how many of these twins had been to see a doctor? This was a random sampling survey like the CDC.....
This is disappointing but you know what they are not doing? They are not stressing them...they are not requiring them to ride on a bicycle or doing jumping jacks -- and several studies have shown that things really start showing only after you do that... I would consider the study design to be obsolete, really. Remember the Light study - same receptor levels at baseline - big difference after exercise.
If they are going to measure anything -pathogens, cytokines, gene expression - I think they should knock them on their butt first.
Note to self: Don't participate in any studies that Cort designs. :tongue:
When I dug out the paper I noticed that too! Oops. Though if you used the individual subscales to create the PCS it wouild almost certainly be less powerful than the most discriminating subscales (like Role Physical) since PCS includes the weaker items as well as the stronger ones. The point, though, is that the figures you quote from another study shows that PCS can do a good job of discrimnating between healthy and CFS, which is all that matters.
If those figures from the new study are indeed for PCS then the study falls since the health of the 'CFS' patients would be implausibly high, at least to be representative of the wider patient population.
You can also try a Google Site Search
Separate names with a comma.