Thanks
I just checked with Dr Eliana Lacerda from the biobank who confirmed that while most cases came from GPs, some (non-severe) cases case from clinics including the Norfolk and Sulfolk ME/CFS service, the Department of Clinical immunology of the Royal Free Hospital and the CFS clinic from the Royal London Hospital of Integrated Medicine (RLHIM). Separately, the biobank have stated that severe cases were recruited via Greater London patient support groups.
I guess that illustrates just how hard it is to get a large population based sample - which is what we need.
Thanks for getting a response from Chris Ponting. And thanks to Chris for that. In an ideal world I sympathise with the points made. However, since I am trying to think this through as hard as I can I would like to raise some of potential difficulties. I am going to be very up front and that may feel uncomfortable for some although I suspect we are all aware of the reality.
For conditions like rheumatoid arthritis and multiple sclerosis there is about one person in 200/500 who has the disease. There is also something like one person in 200/500 (maybe more) who thinks they have the disease but does not. Telling them apart is not hard because there are objective diagnostic signs and tests. For ME/CFS there is no way to tell. Anyone who thinks they have ME is likely to read up on the symptoms and to think they have all of them.
When I talk to colleagues who look after people with ME/CFS many of them say that some patients obviously have an unexplained physical illness and some have 'other issues'. In ordinary parlance this second group are hypochondriacs, or people with inappropriate illness beliefs. Other colleagues who look after ME patients seem to assume that all patients have 'other issues'. Thus although they carefully avoid saying it they think that CFS or ME is hypochondria, even if it was triggered by a virus at some time. That is presumably why patients get sent for CBT. Such physicians will be more than happy sending hypochondriacs to MEGA because they will think that is the idea.
If we consider that we almost certainly have two groups of patients, the question is which one will be most regularly attending CFS outpatient clinics. I think the answer is group 2, because hypochondriacs like seeing doctors. So it may be that the psychiatrists are right - the people they see may be 80% hypochondriacs. Clearly it is not that simple but it is a major issue. What further worries me is that a large proportion of ME research in the past, and probably now, is done by people who do not understand, or underestimate, problems of selection bias.
So although I would agree that for something like RA or diabetes or MS a larger dirtier cohort may be OK, I do not think one can assume that for ME. I think there is a high chance of discovering genetic markers for hypochondria. And even for RA a large dirty cohort with a smaller clean cohort inside it must have some statistical disadvantage because relative risk indices will be blunted.
One might argue that one would do no better with a population based cohort. However, my instinct is that it would be significantly richer in people
without inappropriate illness beliefs and at least it would reduce the risk of referral bias.
Even with RA it became clear to me that with large multi centre studies methodology goes to pot very rapidly if one is relying on an assortment of clinical colleagues. Blunting of results can be by 50%. In other diseases GWAS studies may be reasonably immune to the problems but in ME I don't think one can assume that.