pjclinch wrote:So they state their selection criteria. And...?
Does that mean they don't suffer from the problems of self-selecting cohorts who have their behaviour modified by the intervention being looked at, and by the study itself? Again I don't really see why that is necessarily the case.
May I suggest looking at the selection criteria used in a few of the systematic reviews and seeing if that's a fair description of what they did?
I'm not thrilled...
The first one I dropped in to wasn't (and wasn't suggesting it was) really a systematic review (their ref 6) and doesn't particularly address the issue.
The next one looks more promising, (their ref 21) is indeed a systematic review, and notes the inclusion of randomised studies with "Non-randomised studies were assessed for the presence of potential confounders and classified as being at low, medium, or high risk of bias", but then goes on to point out that, "58 papers of 59 studies were included. The quality of the studies was poor for all four randomised controlled trials and most cluster randomised controlled trials" so they're saying up front that the stuff you might want to look at behaviour modification, as opposed to therapeutic effect, is not that numerous and not done very well. I only skimmed it, but there is no obvious mention of risk compensation.
I then dropped in to a random pick of one of the 6 listed studies (their ref 45) and the authors point out a clear caution that "Adherence to the interventions was low", which really isn't a great start. And then I notice it's in Hong Kong, where (along with much of the far east) the culture of mask wearing is completely different to the UK (it wouldn't be unusual to see oriental tourists in face masks in Western cities at any time in the last decade). And the case there is what to do in case of flu infection of a family member, and I think it's a bit of a stretch to compare behaviour with a known infection of someone you are living with to the situation with Covid 19
And of the original paper I wonder where they got the idea that risk compensation is being put forward as a reason not to promote masks? I've been following the news for the last few months on Covid 19 and it's a new one on me. I don't doubt that some eejits have used it as a personal rationalisation of the reality that they don't like the look of them with their MAGA cap and ALL LIVES MATTER T-shirt, but that's hardly the same as a serious suggestion for wider public health policy.
So while my initial assumption was indeed off the mark and you were right to get me looking at the primary material, the reality isn't, at first stab, really that much better. I haven't exactly been thorough, but experience suggests to me that I have little to gain by being that and I've got a lot to do today now the clock has come past the hour I'd given over to this private research project.
I've already declared my reservations on the work suggesting cycle helmets are not prone to risk compensation in the wild, because the work broadly tries to assume that there are two kinds of cyclists, those in helmets and those who aren't, and unless you're looking at quite limited outcomes (such as might one expect more vehicles to pass closer if you're wearing one) that's not a terribly accurate model of the different cultural groups out on the road who will be wearing lids (or not) for quite different balances
of reasons (e.g., "everyone else has got one on", "nobody else has got one on", "I want to show that I take cycling seriously", "I want to show cycling is normal", "I've seen a shattered helmet and have somehow got the idea my skull is just as weak as EN1078", "I think Egan Bernal looks good in these", "I have heard they make children safer and want to set a good example", "I do it to keep my partner/parents/teachers from moaning at me", "I paid £30 for this hairdo", "it's a faff", "I thought you had to wear one" etc. etc.).
It strikes me that risk compensation is often seen in an overtly simplistic manner. In practice people will do what on balance they want to do. The on balance
is important there, because there are lots of behavioural factors involved, and reaction to perceived existential danger is only one of them. The perception of the dangers of Covid 19 vary enormously between different groups, and thus so will their reaction. The sort of person that, for example, bothers to do a behaviour study over several weeks would, I suggest, be rather more likely to wash their hands and wear a mask too, than someone who thinks being right next to a dozen strangers with no face covering on a very crowded beach is no problem because they've been stuck inside for a while.
That risk compensation is generalised is openly apparent in my work place rules, set by NHS Tayside and NHS Scotland who I'll assume have got a bit of Applied Clue between them. In my bit of the hospital (literally an outlier) masks are not required, but if I go in to the main clinical areas they are. You might say "but that's risk management
", but risk compensation is a form of risk management, balancing danger with other stuff. That risk compensation is subject to cultural variation can be seen by moving from the Lab Block (masks not compulsory) to the main hospital (masks compulsory for staff and visitors). In the lab block, in masks or not, staff stick to the Keep Left system and create space around one another in the corridor space. In the main hospital, less educated visiting public tend not to. And so on.
[edited for typos]