This story has been discussed on Mel's blog. Thanks to Alexicographer, I have a copy of it, which I want to comment on briefly.
Alexicographer suggested (via the comments to the first link, above) that the study would cost about 4 hours' worth of a post-doc's time plus a mac-book (which I assume can be substituted for any kind of equivalent computing device). All I can say to this is, obviously I am not yet a post-doc. I am pretty sure it would take me four hours to work out whether I was supposed to be running a Cox regression or (let's be honest) any other form of statistical analysis you might care to name. As it happens, I'm in the middle of studying up on this as we... well, I should be doing it right now, but this counts, right? In other words, if you are a post-doc, and you have feedback or corrections, please leave them - I would be grateful!
My first comment is that, as suspected, everyone should start using the firefox add-on which redirects every Daily Mail link to pictures of tea and kittens, because honestly - you are not reporting science to the public so much as blatantly trolling. In fact, if your (I'm talking to you again now, not the Daily Mail) Aunt Jane tells you to adopt as a result of having read the Daily Mail article, my recommended response is to give her a sympathetic smile, and gently explain to her that she has been trolled.
The Daily Mail [that is not a link to the Daily Mail, it's a link to Dan and Dan's song about the Daily Mail - I refuse to link to The Daily Mail directly and I thoroughly recommend the song] reports that,
"Scientists say the study throws new light on the age-old question of whether life fulfillment provided by children can actually extend your years. The answer appears to be yes – but only compared with people who want children and are unable to have them. In these circumstances, adoption may reduce the risk of early death, according to Danish scientists."
Bollocks. Wow. They really had to put their best spin-doctors on the job to draw that message out of it. First of all, we don't know whether the differences in mortality are due to "life-fulfillment" - a possibility I would list as number seventy-hundred-and-eighty-seven, under such things as "leading a less adventurous life-style during your forties" or "less likely to have an undiagnosed clotting disorder which prevents successful IVF treatment and at the same time increases the chances of death". As it happens, the scientists specifically state that the study does not provide evidence that having children, even after infertility, extends your life years.
They specifically state that. I guess The Daily Mail reporter got sidetracked looking up big words like "exogenous" and never made it through the whole article.
With regards to the second sentence of the quoted paragraph, it is true that, in this study, adoption was found to be associated with a reduced risk of death amongst those for whom IVF didn't work. However, the authors feel that this is not because adoption "reduces the risk of early death" but because being able and eligible to adopt is associated with a range of other factors which, together, reduce the risk of early death.
Again, they specifically state that, and it is an even stronger statement than the one I described above, which merely tells us that the study is not designed to prove a causal link (although there is a lot of speculation in the statement).
Correlation, you must remember, is not causation. (This reminder is given three times in exactly those terms - once in the abstract, once in the box-summary, and once in the discussion of the results. I can understand how the Daily Mail missed the latter, but honestly... they couldn't read a box summary before writing a newspaper report? Or they did, but they decided it would be more fun to forgo scientific reporting in favour of trolling?)
Mel complained about the division of parents vs non-parents, stating that the idea of being "childless until proven parenting" is noxious. And I don't disagree. The way society divides parents and and non-parents based on whether they are raising or have finished raising living children ignores a wide range of circumstances. This was always a beef of mine (and not just mine) when Mother's Day rolled around (and it still is). Society is yet slightly divided on the question of whether an expectant woman should celebrate or not, but those in the trenches of infertility tend to be more or less excluded - which doesn't seem right, when you consider how much more parenting some infertile men and women have done (both in a physical and emotional sense), when compared to their expectant or newly-minted counterparts.
However, the scientists make clear that their distinction is between those who registered a birth after IVF treatment and those who did not, during a followup period of three to fourteen years after the start of treatment. They openly admit not only the possibility but the probability of "unobserved comorbidity*" given the available data (which was taken from publicly-available registries) whilst pointing out that adjusting for the factors they could observe had not really altered the results. They do consider number of IVF treatments in their analysis. One could still argue that they should have used different terminology, whilst admitting, to be fair, that the debate over the alternative remains unresolved, even within the community. Possibly, to be accurate, it should have been an acronym such at PRB (parents, or patients, or people registering a birth) vs PNRB. In any case, you can be sure the Daily Mail would still have come out with "mothers" and "childless".
A few things must be brought to and taken away from studies like this. Firstly, note that the outcome under observation is rare. The Daily Mail will tell your Aunt Jane that you are FOUR TIMES MORE LIKELY TO DIE if you don't have kids, at least through adoption, but actually, almost everyone they followed - regardless of category - survived. Put it this way: if one person dies, on average, from cause X in one year and in 2012 you happen to get an extra person, that is a DOUBLING OF DEATHS FROM CAUSE X THIS YEAR according to the Daily Mail, or a single extra death to everyone with an ounce of common sense**. And if your unlucky statistic from 2011 happened to hang on til the first of January 2012, giving no deaths in 2011 and two in 2012, then it is still one death per year according to those with common sense, or an AN INFINITE INCREASE IN DEATHS FROM CAUSE X COMPARED TO LAST YEAR AND POSSIBLY THE BEGINNING OF THE APOCALYPSE according to the Daily Mail. So tell your Aunt Jane that either way, things will probably turn out just fine.
Secondly, these data tell a story about a population. They can't tell you if you are going to die soon (which you are probably not - see above). Statistics, as everyone who has ever sat down at a fertility clinic already realises, is not a crystal ball.
Thirdly, just because science tells you X, doesn't mean the right choice is necessarily Y. We make our decisions against a background not just of medical facts, but of values and circumstances. Context is everything when it comes to forging our onward path. This doesn't devalue scientific research or the input it provides, but we must remember to limit its use to its proper purpose.
This study should be a welcome report to infertility patients - especially those who for whom treatments fail. It may turn out that there are concrete steps which can be taken to improve your chances of surviving into old age - medicating with aspirin, perhaps, or increased screening for certain cancers. However, it should be recognised that this is piece, maybe, twelve of a one-thousand-piece scientific puzzle - and the authors are not only aware of this, but are at pains to emphasise it to their readers. We don't yet know why IVF patients who never end up registering births or adopting have an increased rate of death in the immediate followup period after starting IVF, compared with those who do. The authors claim that "because [their] study is based on a natural experiment, the results are less likely to be due to reverse causation" - an assertion which doesn't impress me much, under the circumstances, based on my limited knowledge of statistics, epidemiology, and reproductive medicine. "Less likely" seems (to me, here) a far cry from "impossible". We - and especially your Aunt Jane - must remain open to the idea that death coinciding with infertility treatment may be a cause of childlessness, rather than the other way around***.
In conclusion: this is an interesting, if somewhat limited study, which has been over-reported in what must have been either a slow news week or a please-distract-them-from-the-real-news week. The authors themselves deserve to be thanked for their efforts and encouraged to dig further. The Daily Mail deserves to be forever transformed into pictures of tea and kittens***.
Part two of our trip to Cambodia is coming.
*I'm not sure exactly what sort of "comorbidities" the authors had in mind here - my assertion is the fairly vague one that they knew they didn't have all the info.
**There are valid statistical techniques for handling rare outcomes. What you learned about small numbers leading to unreliable results is an oversimplification which borders on an outright lie. Note that over twenty-one thousand couples were assessed.
***The authors do raise this point when talking about the group of adoptive parents who, they speculate, may have a decreased death rate owing to "survivor bias" - given that Danish laws do not permit parents to concurrently pursue adoption and IVF treatment. On the other hand, it must be noted that they have adjusted for the number of treatment cycles.
****I had to repeat both links again, just to encourage everyone to follow them.