The course of the illness for ME patients in Norway, 2021, Schei and Angelsen

Yes I think the method of data acquisition is pretty well bound to give a shape dominated by artefact but I don't know what that artefact will consist of. If we had a poll of S4ME members and got either the same or a different shape it might become clear what the artefact is. It might even be interesting to see how the curve evolved with time as members contributed - indicating perhaps that a particular age and duration cohort might be quick to respond and another slower. If repeated calls were made to encourage even the most causal members to answer we might get some clear clues.

 

I do not find it surprising that the number of cases increased over time. Underrepresentation of older people and increase of awareness of ME/CFS seem like plausible explanations for this.

The decline since 2012 is probably because of a delay in diagnosis and maybe it takes a while to come into contact with patient organisations. Still, it is a bit surprising that the decline already starts in 2012.

The outliers in 2000 and 2009 are interesting. Is there a way to test statistically how likely it is to have such outliers based on the shape of the data, for example using bootstrapping?

 

Yeah. I'm never sure what date to use for my onset. Simplified it went something like this:

1977 virus -> mild ME -> 95% remission

1997 virus -> mild-moderate ME -> 85% remission

2012 virus -> moderate ME -> deteriorating to severe ME

2016 diagnosis (not before time!)

So do I count from way back in 1977? Or from 2012 when my current relapse started? From 2016 when I was finally diagnosed?

Do I count the remission periods as not having ME? I was pretty fit especially during the first remission but even then... I remember going out dancing on Saturdays. Next day I'd have a horrendous "hangover" which would last into Monday or even Tuesday. Thing is I only ever had one alcoholic drink. Or none at all. Still got a "hangover". Very nearly lost friends over it because I became convinced somebody had to have spiked my drink. So, 95% fit but still there was PEM.

How would that work if one of the causes for the much lower reported incidence in the earlier years is that doctors weren't diagnosing ME/CFS due to lack of knowledge then? There'd be low numbers of diagnoses reported in the Norwegian Patient Register data but we still wouldn't know if that's due to actual low incidence or due to lack of diagnosis/misdiagnosis.

And presumably the patient register would only have date of diagnosis, not date of onset?

 

@Simon M not necessarily existing registries or data but thinking specifically to the Decode ME study. 20000+ are about to receive a questionnaire which will include validation questions for enrollment. This is a huge sample base that could collect some other epidemiological data.

Clearly onset date is not cut and dried for some. How to ask that question would require some thought. I can pin the month down that my daily functioning declined abruptly to below 50% resulting in end of career but in retrospect GF at 18 and Chicken Pox whilst a teacher may have been precipitating factors.

I'm reminded by Ron Davis' comment that in order to make a hypothesis you first need data. Even if an unrefined attempt to gather data is done now with DecodeME it could be a springboard to reframing questions better in the future. It could throw up some interesting results.

 

I’m not sure if this is relevant, but there is presumably an artefact due to average age of onset? I think PWME should have a cluster around the over 40’s (if I’m remembering rightly).

If that is so then we know that there are spikes in the birth rate and country populations in general. I don’t know about Norway, but in the U.K. this is definitely a variable (so called baby boomers as one example, but there are others). I think my age group 50-54 is the largest group in terms of head count in the U.K. at the moment (offspring of the baby boomers)

I would be interested to see the graph adjusted for average population numbers. Not sure if that’s sound logic since I’m in a bit of a fog this morning but might that account for some of the hump?

https://www.closer.ac.uk/data/total-fertility-rate/



https://www.statista.com/statistics/1033637/fertility-rate-norway-1800-2020/

 

We'd probably be best starting from the first occurrence of recognisable ME, wouldn't we? It is thought to have a relapsing/remitting pattern in some people, and presumably those of us who've had lengthy remissions but then developed symptoms again still had the underlying vulnerability.

As some older patients weren't diagnosed for many years, the definition of 'recognisable ME' would probably have to be fairly rudimentary. Unexplained fatigue, taking much longer than previously to recover from activity, and frequent sore throats and 'phoney colds' is probably as good as we can get.

Many of us can recognise even mild ME in our young selves reliably enough, though, because the pattern is so distinctive. The first few times we develop 'flu-like symptoms after doing something as ordinary as dancing at a club or going out for a bike ride might be brushed aside, but there comes a point when we realise we're having to leave gaps between activities in a way that our friends just don't. Young women not on hormonal contraception eventually realise that the 'virus' which floors us completely every four weeks is so reliable that we can predict it to within a day or so.

 

I think DecodeME is going to be problematic in terms of epidemiology, just as the Norwegian poll is. It will be a 'non-probability sample' to quote Bragée. With luck that will not affect the genetics but I think it is likely to make any epidemiological information hard to interpret.

 

There's a video of a joint presentation by Chris Pointing, + someone from the UK biobank(?), which showed the results of a Genome-Wide Association Studies of Hypertension(?). The genetic markers were clearer with larger numbers - think they went to 500K. So maybe the more studies (more data) the better? Or would the noise simply continue to mask the pattern, even with larger numbers, so you need to clean the data (improve the sampling & control selection)?

 

My estimate is that the sampling should either be local population based as for the ME Biobank cohort (rather than through internet enquiry) or, if via internet, it should be as high a percentage of the total ME population as possible. Larger numbers help with the statistical power problem but his is a different issue.

If the number needed for the statistics is more than you can get through local population searching - which they have decided it is, then I would like to see samples being collected from about 50% of all UK ME cases. If there are 120,000 PWME in the UK then that means 60,000. Their target is 20,000 which I think leaves room for a significant false positive signal to come up due to sampling bias. 60,000 is probably impractical so there has to be a compromise, but I would favour collecting as many samples as they can and then testing a random sample.

Let us say that there is a gene allele A that for some irrelevant hum-drum reason that may be very indirect means that people with this allele are more likely to volunteer for GWAS projects. It might be that they have an unusual blood group and are sought out by blood banks so are used to contact with healthcare systems. Even if when they get ME they cannot give blood they might be readier to volunteer for research. Let us say that 1% of the population have this allele. It might be an allele that goes with choosing science subjects at school. It could be all sorts of things.

If there are 120,000 PWME and 20,000 volunteer then you might find that 800 subjects have A, instead of the expected 200. The number could be up to 1,200 but will not be that bad. If 60,000 volunteer, with the same degree of bias you still cannot get more than 1,200 and probably will not get more than 1,000, instead of the expected 600, because as more people volunteer the bias from 'keen volunteers' will drop off. For 20,000 subjects you get a fourfold skewing that would show up strongly on the genetic risk analysis. For 60,000 you might get a 1.67fold skewing that would show up much less.

It may be that this is a problem that would show up with other GWAS studies and turns out not to be a big issue. But I don't think one can be sure that there will be no bias with ME just because there is no bias for diabetes or hypertension.

 

Just to help my poor non-maths brain, are we talking about two different things? DecodeME is looking for genetic factors associated with ME, but the Norwegian study is looking at a specific aspect of the epidemiology (as in, the population and temporal distribution) of ME/ME onset?

I can understand why DecodeME won't necessarily give us reliable information on epidemiology. Participants in the study will have as accurate a diagnosis as we can currently achieve, whereas the comparator Biobank group is much less clearly defined; it's likely to include people with undiagnosed ME, as well as people who've been through a less rigorous diagnostic screening (including self-diagnosis).

If we're only asking a question about date of onset, though, is a 'non-probability sample' from DecodeME – i.e., it only includes people who're reasonably likely to have an accurate diagnosis of the condition, and also happen to have taken part in one specific study – not okay? Genuine question, I don't really understand the issues.

ETA: cross-posted with Jonathan, who might have answered some of this in his post!

 

What about if people take an interest in ME research only after they have had ME for about four years and got to hear about support groups and go on being interested in research for about ten years then lose interest? You would get a completely bogus graph of date of onset. That is what worries me about the Norwegian study.

 

Thank you for your answer.

I can see why you'd like to be in a position to collect more samples than you'll eventually test i.e. to reduce potential volunteer bias.

Are you concerned that, since we don't have a biomarker filter for ME, then that may make an ME GWAS study more challenging?

 

That is something to think about but it does not worry me that much. If there is a process to be studied that gets a bit diluted with other processes GWAS can cope with that. The problem with the sampling bias is that it could throw up gene alleles that have nothing to do with ME at all and people might spend years researching them for no benefit.

 

The three peaks in 2000, 2009 and 2012 coincide with economic crashes, recessions and/or sovereign debt crises. In every recession many people lose their jobs, and some of these job losses are permanent. Some of these people end up on disability.

 

looking at the graph, it's 2015 before we could be confident there is a decline, rather than simply a plateau, and the survey was done in 2019. The decline could be both a delay in getting diagnosis and also taking a while to get interested in research:

that's way above my pay grade . But I've seen quite a lot of large datasets and I think those spikes are pretty unusual – more likely to have specific explanations, than to be noise.

it could certainly collect useful data, though that would be separate from enrolment as the study wants to minimise the burden of joining.

While @Jonathan Edwards is right about sample bias and interpretation, it still like to give us a much better estimate of certain factors than we've ever had before. For instance, we could use the sample to track recovery, where large robust samples are needed. I think a lot of existing recovery studies are clinic-based, and these are likely to be more biased, for various reasons.

my original post points out that population growth since the 90s is only 27%. By comparison, the number of cases is seven times higher. By the early 2010s, than it is in the 90s, so I don't think this would be a significant factor.

 

I agree that the odd spikes look real. Whether they are due to epidemics or economics is another matter!