But it means the 5 year results are potentially biased - I predict those who did not improve are less likely to care about returning a questionnaire at 5 years.
... and those who didn't follow up were less likely to have depression. So what happened to the depressed patients? Were they re-diagnosed with depression and treated for that? We can assume that Collin and Crawley's version of 'CFS/ME is probably largely populated by people with primary clinical depression as well as other diseases to which they have misapplied their label (I have evidential proof, albeit anecdotal and small) that this can happen with the NOD.