Predictors of Chronic Fatigue Syndrome and Mood Disturbance After Acute Infection, 2022, Sandler, Lloyd et al

Andy

Andy

Senior Member (Voting rights)
Prospective cohort studies following individuals from acute infections have documented a prevalent post-infective fatigue state meeting diagnostic criteria for chronic fatigue syndrome (CFS) – that is, a post-infective fatigue syndrome (PIFS). The Dubbo Infection Outcomes Study (DIOS) was a prospective cohort following individuals from acute infection with Epstein-Barr virus (EBV), Ross River virus (RRV), or Q fever through to assessment of caseness for CFS designated by physician and psychiatrist assessments at 6 months. Previous studies in DIOS have revealed that functional genetic polymorphisms in both immunological (pro- and anti-inflammatory cytokines) and neurological (the purinergic receptor, P2X7) genes are associated with both the severity of the acute infection and subsequent prolonged illness.

Principal components analysis was applied to self-report data from DIOS to describe the severity and course of both the overall illness and concurrent mood disturbance. Associations between demographics and acute infection characteristics, with prolonged illness course as well as the PIFS outcome were examined using multivariable statistics. Genetic haplotype-driven functional variations in the neuropeptide Y (NPY) gene previously shown to be associated with brain responses to stress, and to trait anxiety were also examined as predictors. The sample included 484 subjects (51% female, median age 32, IQR 19–44), of whom 90 (19%) met diagnostic criteria for CFS at 6 months.

Participants with greater overall illness severity and concurrent mood disturbance in the acute illness had a more prolonged illness severity (HR = 0.39, 95% CI: 0.34–0.46, p < 0.001) and mood disturbance (HR = 0.36, 95% CI: 0.30–0.42, p < 0.001), respectively. Baseline illness severity and RRV infection were associated with delayed recovery. Female gender and mood disturbance in the acute illness were associated with prolonged mood disturbance. Logistic regression showed that the odds of an individual being diagnosed with PIFS increased with greater baseline illness severity (OR = 2.24, 95% CI: 1.71–2.94, p < 0.001). There was no association between the NPY haplotypes with overall illness severity or mood disturbance either during the acute illness phase or with prolonged illness (p > 0.05). Severe acute infective illnesses predicted prolonged illness, prolonged mood disturbance and PIFS. These factors may facilitate early intervention to manage both PIFS and mood disturbances.

Open access, https://www.frontiersin.org/articles/10.3389/fneur.2022.935442/full
 
"Subjects were classified as a “case,” that is, given a diagnosis of post-infective fatigue syndrome (PIFS), if they met the diagnostic criteria for chronic fatigue syndrome (CFS) (27) at 6 months after the onset of symptoms, following comprehensive medical assessment including laboratory investigations by a specialist physician, and structured mental health assessment by a psychiatrist."

Fukuda was used.

"For this analysis, data from the Somatic and Psychological Health Report (SPHERE) questionnaire (24), the Brief Disability Questionnaire (BDQ) (25), and the Profile of Mood States (POMS) (26) were used."
"The BDQ and POMS were used to capture functional impairment and mood disturbance, respectively."

POMS is not available as open access, despite apparently being created in 1972. Wikipedia describes how
"The POMS measures six different dimensions of mood swings over a period of time. These include: Tension or Anxiety, Anger or Hostility, Vigor or Activity, Fatigue or Inertia, Depression or Dejection, Confusion or Bewilderment. A five-point scale ranging from "not at all" to "extremely" is administered by experimenters to patients to assess their mood states."
source: https://en.wikipedia.org/wiki/Profile_of_mood_states
 
I haven't read the links but I can say that vestibular neuritis (possible Covid or EBV reactivation) for over 2 years can cause temporary transient functional impairment and mood disorder. It's miserable.

I've never had mood disorder during onset of ME/CFS in 31 years.
 
Lloyd said:
Participants with greater overall illness severity and concurrent mood disturbance
Click to expand...
Geez, what an awful study. Just a mix of odd things thrown together. Looks like they're actually holding on to believing it must be related to acute illness severity. Did not pay attention one bit. But they're actually talking about mood disturbance in very ill people? Seriously? People aren't even allowed to be ill anymore? Who the hell is a happy camper when severely ill? Do they simply have no coherent concept of what illness even means?
 
Interesting that the cohort was 51% female and the PIFS/CFS group was only 46% female (OR 0.72 (0.42, 1.25)), but the cohort was too small to draw any strong conclusions from this.

Both Ross River virus and Q fever were more likely to result in PIFS/CFS than EBV (but again, huge confidence intervals).
 
Abstract said:
Female gender and mood disturbance in the acute illness were associated with prolonged mood disturbance.
Click to expand...

Andy said:
The BDQ and POMS were used to capture functional impairment and mood disturbance, respectively.
Click to expand...
POMS is not available as open access, despite apparently being created in 1972. Wikipedia describes how
"The POMS measures six different dimensions of mood swings over a period of time. These include: Tension or Anxiety, Anger or Hostility, Vigor or Activity, Fatigue or Inertia, Depression or Dejection, Confusion or Bewilderment. A five-point scale ranging from "not at all" to "extremely" is administered by experimenters to patients to assess their mood states."
source: https://en.wikipedia.org/wiki/Profile_of_mood_states
Click to expand...

If POMS is measuring fatigue, vigor and worry, then someone who is unable to do the activities they did before becoming ill and are struggling or unable to get to work is likely to score higher on POMS than the average person without a post-infectious fatigue syndrome.

Like other abstracts we see from BPS researchers, I suspect this one has been carefully crafted to give the impression of things that aren't actually borne out by the data. I think the fact that they only report one odds ratio for being diagnosed with PIFS (initial illness severity) is probably significant (as in, other factors had little effect on being diagnosed with PIFS).
Abstract said:
Participants with greater overall illness severity and concurrent mood disturbance in the acute illness had a more prolonged illness severity (HR = 0.39, 95% CI: 0.34–0.46, p < 0.001) and mood disturbance (HR = 0.36, 95% CI: 0.30–0.42, p < 0.001), respectively. Baseline illness severity and RRV infection were associated with delayed recovery. Female gender and mood disturbance in the acute illness were associated with prolonged mood disturbance. Logistic regression showed that the odds of an individual being diagnosed with PIFS increased with greater baseline illness severity (OR = 2.24, 95% CI: 1.71–2.94, p < 0.001).
Click to expand...

Some poking around in the data is warranted.
 
Hutan said:
If POMS is measuring fatigue, vigor and worry, then someone who is unable to do the activities they did before becoming ill and are struggling or unable to get to work is likely to score higher on POMS than the average person without a post-infectious fatigue syndrome.
Click to expand...

In the PC Derivation and Validation section, they say they used depression and anxiety questions from the SPHERE questionnaire to measure mood disturbance and then validated those against the relevant parts of POMS.

But this seems pretty obvious and uncontroversial to me, people who report more depression and anxiety in their initial illness are likely to report the same at a later time point — of course! You could take the illness out of it and find the same by just giving questionnaires to people at two time points. No surprise either to see more for female gender as higher levels of reported anxiety and depression is a pretty established finding.

Their Table 3 is very weird. They claim to show hazard ratios for recovery (so lower is worse), but then some of the data is clearly in the opposite direction (like the different virus risks for PIFS). I think something went wrong there.
 
Commentary: a (fruitless) search for biopsychosocial gold?​

In a nutshell: this appears to be an unsuccessful attempt to find biopsychosocial gold by looking to see if neuropeptide Y (which has links to anxiety and stress) is linked to developing CFS after an infection. There is no link, and this new analysis simply confirms the previous findings that the severity of the initial infection increases the risk of CFS at six months.

Introduction: the Dubbo study (severity of initial illness predicts CFS)
The Dubbo study, published in 2006 (with more papers published later), was important because it followed patients with three different pathogens that can all trigger CFS: EBV (glandular fever), Ross River Virus (RRV) and the bacteria that causes Q fever. Regardless of the type of infection, people experienced broadly similar symptoms. The authors argued (with good reason) that this is because the symptoms are mainly driven by the body's sickness response. This causes symptoms (including fatigue, brain fog and low mood) that make us rest up while the body powers up the immune system to fight the infection. (Lying low also reduces the risk of being eaten when we are vulnerable.)

The study found that the best predictor of developing CFS at six months was the severity of the initial illness (measured by days of bedrest and days out of role during the acute illness).

Sadly, but unsurprisingly in this field, there have been no replications of this important work. Lenny Jason did something that came close (and, in many ways, was better), but the published analysis published of that study is uninterpretable.
Click to expand...

Neuropeptide Y

The neurotransmitter neuropeptide Y has been connected with ME/CFS before. It's the most abundant peptide (short protein) in the mammalian body and has a large number of roles, mostly concerned with eating and energy metabolism. The authors highlighted that variants of the peptide are linked both to being more anxious in general, and the response to stress and the new analysis looked to see if neuropeptide Y genetic variants affected the risk of CFS at 6 months.

The authors don't explain any more about their hypotheses that neuropeptide Y plays a role, but the idea would fit nicely with psychosocial models of CFS (of which lead authors Ian Hickie and Andrew Lloyd are enthusiastic proponents). People who are more anxious might worry more about symptoms, and if they are less resilient, they might cope less well with the illness. As a result, it might be argued that people would get into the claimed vicious cycle and become trapped in the illness by their own beliefs and behaviour.

But they found no link between neuropeptide Y and subsequent illness or recovery. They didn't even find a connection between the peptide and mood disturbance, either at baseline or later.

I believe the connection between the severity of initial illness (for three different pathogens and an "unconfirmed" fourth group, likely to include many other pathogens) remains one of the most interesting clues we have as to what causes ME/CFS.

The mood disturbance angle (a doomed approach)
What does the analysis of mood disturbance tell us? In short, nothing. Apart from the obvious point that people get cranky and depressed when they are very sick, it was always going to be impossible to find a link.

The study used a technique called Principal Component Analysis to identify core 'components', groups of symptoms from the SPHERE symptom questionnaire they used. Their analysis found an "illness severity" component and a "mood disturbance" component.

However, there was heavy overlap between the two, with items like "waking up tired?" and "feeling irritable or cranky?" in both components. Illness severity and mood disturbance were highly correlated (pearson correlation of 0.84 on a 0-1.0 scale). This is unsurprising, since, as the authors acknowledge, both illness severity and low mood are probably driven by the same thing: the sickness response.

In any event, as things turned out, they didn't even find a connection between neuropeptide Y (which has a genetic link to depression) and mood disturbance.
Click to expand...

I imagine the results were disappointing for the authors. If they had found a connection with neuropeptide Y, it would probably it would have been hailed as the 'bio' of biopsychosocial.

But I think the authors deserve credit for publishing the null results, even if they appear to have been rather shy when it comes to explaining the motivation behind the study.
 
Last edited:
LarsSG said:
Their Table 3 is very weird. They claim to show hazard ratios for recovery (so lower is worse), but then some of the data is clearly in the opposite direction (like the different virus risks for PIFS). I think something went wrong there.
Click to expand...
the different figures are because they needed to use different statistical approaches for different sets of data. The hazard ratios for illness severity and mood disturbance are hazard ratios of "recovery" (I know, it's odd) and the data is the number of days for recovery for each patient. They used Cox regression analysis to do this. As the hazard ratio is for "recovery", the numbers are < 1 for factors that increase the risk of non-recovery – effectively, they indicate recovery takes longer.

The final column is for post-infectious fatigue syndrome (PIFSs, which is whether or not a patient met the Fukuda case definition at six months). This is a binary outcome (yes/no) and was analysed by logistic regression which, as you probably know, spits out odds ratios for the outcome (PIFS). So these are >1.

ME/CFS Skeptic said:
I'm a bit confused: is this a new statistical analysis of the data of the Dubbo study that has been published 15 years ago?
Click to expand...
it is basically a reheat, with the addition of data on the neuropeptide Y genotype of each patient.
 
Last edited:
Looking at the reference given for the validation of the SPHERE questionnaire, it seems, despite a good sample size, only to have been validated in a selected population of patients already diagnosed with psychiatric problems with their GPs’ (or specialists’) conclusions being one of the independent variables used in the valuation.

Given the GPs were involved in patient selection and the evaluation of the questionnaire was validated against their conclusions, this runs the risk of becoming a circular process. However as there does not seem to have been any ‘normal’ controls or trialling with a ‘physically but not mentally ill’ population surely this questionnaire is only validated for use in an already diagnosed mentally ill population.
 
Sean said:
But they have already declared us mentally ill.
Click to expand...
Worth bearing in mind that the original analysis (backed up by this new one) found that the only significant predictor of CFS at 6 months was the severity of the initial infection and not psychological factors. I think all the data in the original study came from analysing SPHERE data.

Curiously, Peter White's (actually pretty decent) study of mono also found that days of bed rest during the acute infection similarly predicted the outcome. Though he chose to interpret bed rest as a proxy for excessive resting. I know.

One reason for the lack of follow-up of such interesting findings might be who made those findings and how the results sat with their view of the illness and those who stayed sick.
 
Simon M said:
the original analysis (backed up by this new one) found that the only significant predictor of CFS at 6 months was the severity of the initial infection
Click to expand...
There are good odds that this is the same problem others are having right now with Long Covid. It may be an individual predictor but most will have had a mild acute illness. There is so much wildness in how things like risk factors are interpreted, as if increased risks in a specific population meant it's the only population that counts. Very weird and obviously biased.

This is especially problematic considering how much statistical massaging is going on with this kind of data, that there is an issue with interpreting statistics that misunderstands that a lower % from a far larger population is a more significant problem overall than a slightly increased % in a much smaller population. We are definitely not at a stage where individual care is significant enough to try and pin down any cause to any individual.

I wouldn't put much thought into those findings, especially in anyone who needs to try and pin it to psychology. They make so many basic mistakes, mostly in false assumptions.
 
One of big problems is that doctors are more willing to diagnose Long Covid or ME/CFS if its preceeded with a hospitalisations from an infection. They have the clear cause and effect and the diagnosis is thus much easier. Those seriously ill at home can't be seen, can't get diagnosis and that alone will bias the data enormously. Fact is the vast majority of CFS and Long Covid sufferers do not currently have a diagnosis from a doctor but the bias in the diagnosis will directly appear in research around it as well.
 
You must log in or register to reply here.
Back
Top Bottom