Good points, Trish.
I have never been terribly attracted to crossover trials and tried to think why. (Presumably they may give more data from fewer subjects - each giving outcome for two or more trial arms.)
I think the problem is in knowing whether being on LDN first arm and placebo second is the same as vice versa. What if 6 months of LDN sets you up for a good run for a year? I guess my objection to crossover is that it introduces more unknowns.
The point about pain is crucial. If we want to see if LDN reduces pain we need a strictly blinded trial with a pain scale outcome. Cut and dried. Pain could also be included as a secondary measure in a trial with actimetry as primary outcome measure of 'total impact on ME'. But the pain result would have to be repeated if the primary measure came out negative and the secondary measure positive (at least if only without Bonferroni). This is where being picky about power is critical. Planning for more than one trial may be the right approach.
What I had not said is that actimetry looks the best option if the trial is designed to test the impact of LDN on ME as a whole. There seems a good consensus that the primary aim for patients for an ME drug is being able to do more of actitivites that make life worth living. Whether through helping pain or cognition or whatever, that means that a validated actimetry index should reflect the aim. Not everyone may be able to do more, even if feeling better, but with trials you allow for that. If a drug is capable of reaching the primary aim for some it should show statistically. Outcome measures do not need to pick up every case of improvement. It is more important that they are simple and robust. Actimetry is criticised for missing things or being misleading but if, as you say, it is done over extended periods, and validated against diaries in a calibration study, then it ought to pick up a drug benefit. As we have discussed, scepticism about actimetry may partly reflect the fact that so far nothing has shown a useful benefit on it - simply because nothing so far has been good enough.
The bit about optimal dose of LDN being reached begs the question of how you do that without losing blinding - I think we have covered that. It makes the case for an objective measure.
When you put it in those terms I’d say crossovers need to be out definitively in ME because of the way that PEM accumulates - then hits after 6mnths of just pushing your boundaries to the limit (and getting away with it … then not .. boom)
you can kid yourself a regime of whatever is making you better because you get the ‘ability to do’ or walk the tightrope but that’s not really ‘tackling ME’ but ‘IS ME’ (and I think where much of the medical profession is still stuck - because even patients need years of living doing this same false hope mistake to realise ).
to me what would/should be really accurate is the plain ‘a year on’ comparison of groups - with yes good measures to that.
Just as I think the public don’t get that PwME would have walked through hot coals of hell with GETs hurt doesn’t equal harm for the promise of that having an endpoint to their disability. Or even just a ‘chance’ of that being worth the effort. BUT Anyone voluntarily signing up would have asked from a medical professional one question alone: will/could it make me worse?
And I almost have to remind myself of that too when looking at measures.
Last edited: