Part of my job is to read scientific journals. This is mostly a joy, as a lot of them are much more than just a collection of bone-dry articles created by women and men in white coats, but a joyous celebration of science. Nevertheless, sometimes it’s quite a drag to work yourself through 4 weekly and 4 monthly journals on a saturday morning, because some articles are just not very entertaining. For me, the perfect paper should be statistically sound, enlightening and revolutionizing the way I look at science. This, unfortunately seldom happens. Rarely, papers are just plain boring. But then there are some where you ask yourself:
I was recently made aware by a friend of the paper “Feasibility of an exercise intervention for women with postnatal depression” by Daley et al (1) that was featured in the March edition of the British Journal of General Practice, a journal for general practitioners (family physician, for our American friends) that unfortunately often has the entertainment value of a dried raisin (some people describe its scientific content as similar, although I wouldn’t go that far).
Nevertheless, postnatal depression is a terrible, crippling problem affecting women in immensely vulnerable periods of their life, and exercise is a well established and successful intervention in affective disorders (and so much more), so I thought: this sounds really interesting, let’s have a go. The only thing that made me suspicious was the title: Why ‘feasibility’? Why not ‘the effects of exercise on postnatal depression’? I was soon to be informed.
The paper starts quite promising, making a valid point about the importance of the issue and the lack of evidence, apart from two small trials from Perth in Australia. As those, the author sniffs, were ‘conducted in the Gold Coast’, the warm weather conditions were likely to have a positive influence on the patients willingness to participate in exercise (what if it’s far too hot? according to the Perth weather centre the average maximus are between 30 and 18 degree celsius. I wouldn’t like to jog in 30 degrees celsius).
So the trial was held in Birmingham. Where the weather is of course much more suitable.
Next Daley et al described how they planned to set the trial: they recruited patients who were identified as depressed by either their GPs, health visitors or psychiatrists. Only patients who self-referred themselves had to have a score of >12 on the Edinburgh Postnatal Depression Scale. That, I thought, would make for a pretty heterogenous sample, as I am sure that these three groups of health professionals approach the diagnosis of depression differently. But on the other hand, looking at the problems the authors had finding any participants at all, heterogenity of the sample was the least of their problems. In one aspect at least there was homogenicity: as there were no ‘interpretation services’ everybody had to speak English. Of all places in Birmingham, where 16.5 percent of humans were, according to the 2001 census, not born in the UK, and 29.5 percent did not consider themselves white. So the representability of the sample in this particular area was, er, low (although in the end 26.3% of the sample were from non-white background). But that was only the beginning of the problems Daley and her co-authors had to grapple with, as their were terribly let down by their colleagues in primary care: of the 262 GP practices they wrote to, only 56 (21%) agreed to take part in the stdy. The rest didn’t consent to take part (26%) or didn’t bother to answer at all.
So much for the much the appetite for research in the West Midlands’ primary care community.
But it gets better: of the 56 practices that agreed to participate, 3 withdrew, 18 didn’t provide patient lists and 10 reported no eligible patients (riiiiight). To sum it up: they recruited 57 ladies, of whom 6 were found to be ineligible and nine did not start or finish the trial. In the end, they ended up with 31 ladies who actually completed the full study, including the 12 week follow up, divided into two randomised groups of 15 and 16 patients.
I think I would have thrown the towel by that stage. How can you produce meaningful data if your colleagues in the community have no interest in your research? According to Birmingham City Council’s website there were 15800 births in 2004. A conservative estimate gives the incidence of postnatal depression of about one in ten. Now have a look how many women were able to be recruited for an important study like this. Exercise is a cheap, non-pharmacological intervention that of course has so much more benefits than postnatal depression alone, but thanks to the small numbers in this paper there is still only limited data available for this particular implementation.
Initially I asked myself why this paper was published at all. I would have been interested to see a large sample of women respond well to exercise intervention, but that was not achievable with the difficulties the authors had during the recruitment process. So what does it tell me? Is it feasible to recruit women to an intervention like this? Yes, but it’s tricky. Does it work? Probably. But that’s pretty much all I can gather from it. I think in the end this paper had to be published to demonstrate that good science is not achievable without ‘a little help from your friends’. Willingly ignoring research in general practice won’t help to improve GPs and their chosen field’s reputation and will deprive their patients from the benefits of good science.
…and if I ever meet Dr Daley I’ll buy her a stiff drink for all those moments when she probably wanted to tear her hair out.
(1) Feasibility of an exercise intervention for women with postnatal depression: a pilot randomised controlled trial; Daley, Amanda J et al; British Journal of General Practice, Volume 58, Number 548, March 2008 , pp. 178-183(6)