In one of the previous posts I discussed the use of expert-assessment in an attempt to characterize exposure without the need for expensive (both in terms of money but also man (or women)-hours). That seemed to work, but not always and not as good experts would like us to believe. So where does that leave us then? We still have no money to do an extensive study of personal exposure measurements to back up the assessment of occupational factors in relation to human health.
The toolkit of exposure assessment is limitless and drawing from the vast amount of elaborated, sophisticated and ingenious methods, an alternative method often used is…well…let’s just ask every participant in an epidemiological study him/herself what they are exposed to and then hope that makes any sense. The good thing here is that it is relatively cheap. Funders like cheap, and journalists couldn’t care less about such details – being busy and all. So why not, let’s give it a go!
Unfortunately, and this probably comes as no surprise, self-reported exposure assessment may not be as good as one may hope. There are several ways in which this could go wrong, for example, random people who are send a questionnaire may not in fact have a clue to what they are exposed, and since that has never prevented anyone from still having an opinion they will still give an answer (that is, if they bother to reply at all). If they do actually have some idea (sitting in front of you computer 9 hours a day may give you a hint when it comes to 50/60Hz electromagnetic field exposure for example), reporting of exposures may also be related to how much attention a certain exposure has received in the media, especially in relation to some nasty disease, or people may have their own motives to report being exposed to something, or not…
And of course, then there are men and women…
A recent paper (Eng and colleagues (link to abstract here)) investigated just this issue in New Zealand. They invited a random sample of men and women to take part in a telephone interview, which collected information on self-reported occupational exposure to specific dusts and chemicals, physical exposures and organisational factors (37% response rate).
Of course men and women or, just to assure a good balance, women and men, don’t always do the same jobs. In this particular population, there were higher proportions of females in the professionals, technicians and associate professionals, clerks, and service and sales workers whereas there were higher proportions of males employed in legislation, administration and managing, agricultural and fishery, trades, and plant and machine operation and assembly. To be honest, that is not very interesting and if you are interested in that just read the paper.
But what they also did, cleverly, is to create 1:1 matched (on current occupation) male/female pairs. Because they did this, we can now see whether within the same jobs women and men report similar occupational exposures. And of course they didn’t.
Within these matched pairs, thus for the same jobs, men reported more often than women that they were exposed to smoke/fume/gas, and oils and solvents. Men were also more likely to use vibrating tools and work irregular hours, and more often worked night shifts. Men were also more likely to report job stress compared with women in the same occupation; something that is most often argued, but also shown in for example the UK reporter network for occupational ill health (THOR) (Although not when psychiatrists do the reporting, interestingly enough) (link), to be the other way around. Compared with men with the same occupation, female workers were more likely to report carrying out repetitive tasks, working at very high speed, and working in awkward or tiring positions.
Still matching, but after more specific job-coding (using 5 digit coding, for those interested), men remained to report higher exposures to welding fumes, herbicides, wood dust and solvents.
So what can cause these differences?
The most straightforward explanation would be that there just are gender differences in reporting or perception of exposure. The authors however, dismiss this finding based on the results of a Swedish study in which no differences in the validity of reporting between men and women were reported. Personally, I don’t think this is a very good comparison given that the Swedish study was limited to physical risk factors and musculoskeletal disorders. So I am not that convinced this hypothesis should be dismissed so easily.
However, an alternative explanation may be differences in exposure duration or intensity between women and men in the same job. For example, the authors mention that female workers are more often employed part time and, therefore, more likely to experience shorter exposure duration that they may not report.
A third reason may be that women and men are actually very accurate in reporting there exposures, and the reported gender differences can be explained by the fact that men and women with the same occupation do not always carry out the same tasks. This could be due to the different physical capabilities of men and women or maybe because of socialized gender roles.
All this sounds plausible. But could it be something else as well? Something far more boring? Here, to the credit of the authors they also discuss the explanation for nerds. Let me entertain you…
To make these comparisons, all jobs from people were coded using a coding scheme. Unfortunately, these are fairly generic coding schemes not specifically designed to distinguish situations with different exposures from each other. So for example, being employed as a farmer can be cause for widely different types of exposures, while even a ‘sales person’ could have widely different occupational exposures depending on what they actually sell (clothes or car batteries) and where they sell it (where any products are also manufactured or maybe on the market square in the city centre).
So what does this tell us?
The paper was accompanied by a Commentary (link) which essentially went through a similar distillation process as I have done above. To my surprise, this commentary reached the following conclusion, based on this paper in combination with several others: “Gender bias has serious negative impacts on our health and our humanity.”
Something that may well be true, but in this context seems to be a commentary that was started by writing a conclusion with the data subsequently molded to fit the argument.
What I would make of this paper? Maybe I am going to stick with the authors and say that job code is not a very useful indication for occupational exposures in general population studies. That is a fairly safe bet. I would also say that self-reported exposure is likely to be biased. That is also a fairly safe bet; in fact I have been involved in some work several years ago showing this in the context of occupational exposures and asthma (link).
It is intriguing, since in the absence of measurement data in the described paper, we don’t know whether the men were right, or the women, or both,…or neither. So maybe the cheap options are not the holy grail after all, and depending how (un)lucky you are they may actually cost you more in terms of sample size (to try and make up for all the errors made by men, women and experts of either gender) than when other, more expensive methods were used.
I personally have a feeling I may touch upon those other methods in future columns…
Sense about Science
The Fun Police 
Posted on September 21, 2011
0