Contribution of individual psychological and psychosocial factors, Dubbo Infection Outcomes Study, 2019, Cvejic, Hickie et al

Uhhhh. Seriously? Of course. I mean, absolutely of course.

 

Ah. Yes. Good old optional requirements. Those are common nowadays.

It's required. But don't bother with it. Nobody cares anyway, right?

 

I completely agree, it has all potential to be a massive scandal (but many things may have).

It shows a lack of any real responsibility, which is more than just only doing and letting what is commonly monitored (so to say).

One could go an with consideration, say, would such writers commit any crime if it only would be safe enough. But most outstanding is the irony that exactly people that are not able to take a real responsibility (and thinking logically) are constantly adressing others not to be able to really care for themselves (and therefore being sick).

This is a logical possibility, especially when the illness is utterly complex anyway. I would expect that researcher are able to figure this out, but no, this is commonly too difficult to consider.

If the "neuroticsm" is directly caused by the illness there won´t be many a chance, and you can interpret whatever you want anyway. I think btw, that the reasonable possibilities @Invisible Woman mentioned are not important (if you are not in a really bad situation), at least this is my impression from forums and in my own experience.

Testing a hypothesis they probably themselves do not really believe in ... ah no, it´s of course only "too clear".

 

It's always a bit suspicious when authors quote p values, but not effect sizes in the abstract. Usually the intent is to mislead naive readers (medical doctors who only read abstracts) into thinking there is a large effect, when there is barely an effect at all.

 

https://www.students4bestevidence.net/blog/2016/04/05/tutorial-hazard-ratios/

 

Sorry, @Hutan. Completely missed this. Terrific analysis!

@dave30th, any chance you could take a look at this one?

 

Yes, collaboration would be great! Suffering from a spoon shortage here, so perhaps between the two of us we could get enough brain cells together!

 

@Hutan I did try to look up the NHMRC funding. The grant numbers are really old, so I suspect that the NHMRC grants they list might be for the original Dubbo study. Given how long ago that is, I’m not sure how much benefit there is in writing to NHMRC (especially given the results of the TCR have just been announced, and we could argue that their funding decisions have improved considerably).

I haven’t looked at the Mason Foundation yet, but I’m planning to follow up to see how old that funding is too. Though I’d be more inclined to still write to them even if it was, because some of their recent funding decisions have been questionable.

 

It's not possible to untie a Gordian knot in a methodical way. In the legend, Alexander thought out of the box and slashed it with a sword. I think Lloyd has found he tied one and simply declared his undone. Easy peasy. Who will be next I wonder

 

The two papers are looking at two different things. The original BMJ paper he's referring to in the interview found that neuroticism and other psychiatric scales didn't predict ONSET of postviral fatigue syndrome. The Cvejic analysis of follow-up data found that higher neuroticism score predicts lower likelihood of recovery.

 

Worth putting a copy of the questions in the neuroticism part of the scale up here. They relate to things like 'worry a lot' and self-consciousness of what people think of you.

I'm only looking at the abstract and reading into it that they only did these questionnaires after people had become ill. With a range of different illnesses!

Someone with say a nasty virus that is particularly bad in its acute form - and which then becomes significantly worse and has work, friends and family getting worried/frustrated you are still not feeling any better would be a direct reason why someone would answer these questions higher - cart before horse.

To claim this is neuroticism seems a very dodgy conclusion. It isn't personality which is the only differentiator - the two people have different illnesses. With different prognoses (even if only based on gut instinct of how ill you feel).

For all the claims of stability in personality scales when they are done in the Big 5 methodology (but with limitations made clear on accuracy) scrapping bits out and administering them as if you test it on its own and it isn't something quite different out of context isn't right. It would hardly look like a personality survey without the other sections.

There must be validity issues to use the questionnaire tools that they did - none of them are that stable, then using them out of context and misrepresenting them must make them next to useless for the task of 'personality'. They also have no proof of it being stable or any measures from before they got the illness to make their assertion this was even present before other than some assertion their questionnaire and measure is stable and valid when used in that way.

How on earth was this funded? I find it very strange anyone would do this - what could it possibly achieve?

 

Do we know the exact wording on the locus of control questions? and where it sat ie was it after 3 questions about their mental health and 'primed' in a certain direction?
I imagine it to be insightful as to whether it is asking to 'describe their situation' or 'describe how they feel'. Good books on chronic stress (in its real form rather than when it is mixed up with anxiety) note locus of control as a major objective way to cause stress in a human being - no subjective in 'how you perceive it' tosh but removing options for people to step themselves away from harm and have some elements that are not unpredictable.

Like in a retirement home examples are giving someone a regular timeslot they get visits rather than sporadic people popping in all the time then not coming for a month etc. Or workload changing randomly with no control over deadlines or how you can do said work to make achieving them possible. In the old days they might have called that playing psych-games on someone it is so obvious (now we've got CBT nonsense justifying it is the experiencer 'being sensitive'). I don't know the context of what services these people might have been under.

Either way when you think about how many factors/much heterogeinity there is built in here with different diseases of different severities, different socioeconomics vs not that big a sample and all these various measures ... so many variations in explanations for individual answers being hidden under this one. I'd be very intrigued if severity didn't play a part if it was this latter definition, if it wasn't likely these other bits muddying waters on this particular one too.

Intriguing..

 

One caveat is that the variables aren't standardised z-scores. So the hazard ratio's are representative of a one unit increase in the specified variable rather than 1 standard deviation increase. I don't know why researchers do this because it makes it hard to interpret the effect size and compare the influence of different factors.

In table 1 they report a standard deviation of approximately 5 points. So with an increase in one standard deviation the chance of recovery would decrease with 3*5 = 15%. Which still seems much smaller than the effect of initial severity, sex and serology.