Note: If this post looks familiar, then thank you for being one of my dedicated readers who has followed me from the very beginning. Yes, this is a repost of my very first blog post and this week marks the first anniversary of Science & Sensibility! Happy blog-aversary to us! Later this week, I'll share a few of my favorite posts and other milestones over the past year. And if you missed this one a year ago, here's an 'understanding research' lesson all wrapped up in a package of very good news (ok, not-so-new news) about midwives!

Often, the closer I look at a study, the less confident I become about the results. I've learned that you can't always trust a paper's title or abstract, and media coverage of new studies can be woefully misleading, even when it is not carefully orchestrated by those with a vested interest (which it often is). Being advocates for 'evidence-based care' means not just knowing that a study has been published, but knowing whether that study is any good, and in what circumstances (if any) the results are relevant and reliable. It also means having our guard up against deeply flawed studies that shape policy and practice despite their limitations. (Henci Goer has done a fantastic job deconstructing some of these influential studies in her series, When Research is Flawed.)

A systematic review synthesizes all of the literature on a given topic, using rigorous criteria for which studies will be included. For instance, Cochrane systematic reviews are typically confined to randomized, controlled trials in which there is no evidence that the randomization process has been intentionally subverted. For this reason, Cochrane reviews are considered the 'gold standard' of evidence.

But what happens when the trials that make up a systematic review themselves have flaws or limitations? We end up with Cochrane reviews that can mask problems in the literature, and we can inadvertently put the evidence-based 'stamp of approval' on a practice that still needs to be studied further. This is referred to as the 'garbage in, garbage out' phenomenon, and we see plenty of it in the obstetric literature.

One kind of garbage that Cochrane reviews rarely address is crossover. This is when some of the participants randomized to the 'control group' (e.g., no intervention) end up getting the intervention that is being tested. This problem is rampant in trials of induction, pain relief, and episiotomy, among others. Some women randomized to 'expectant management' end up getting induced; some women randomized to 'non-epidural pain management' end up getting epidurals; some women randomized to 'conservative use of episiotomy' end up getting episiotomies, and so on. This makes it much more difficult to use our statistical toolbox to discover differences between the two groups, and as a result we see smaller differences, or even no difference. The 'evidence-based' conclusion then becomes 'there's no difference is unwanted outcomes, so the intervention is harmless.' But 'no difference' can also mean 'this study wasn't big enough to find a difference' or, in this case, 'there was too much crossover to detect a true difference.'

The crossover problem usually drives me crazy because it often serves to perpetuate medical-model bias and medical-model practices. But I had an 'ah ha' moment when I discovered a crossover-of-sorts problem in the 2008 Cochrane systematic review of midwife-led care. This review was released to fanfare within the birth community. Finally, the enormous body of literature on midwifery had been synthesized by Cochrane reviewers and the conclusions were firmly in favor of midwife-led care! The results were, indeed, unusually impressive. While the conclusions of many Cochrane reviews are couched in tentative language and call for more research, the reviewers here concluded decisively, 'Midwife-led care confers benefits and shows no adverse outcomes. It should be the norm for women classified at low and high risk of complications' (p. 17). Still, when I looked a little closer, I was perplexed that some of the differences the Cochrane reviewers found were small or even non-existent. What? No difference in c-sections? Only a small difference in episiotomy? What's going on here?

A kind of crossover is the culprit, and this time it means that the good news just gets better! It turns out, of the 11 trials comprising over 12,000 women, in all but 1 of these trials (with only 318 participants), some or all of the women in the control groups were actually cared for by midwives. The difference was that that these midwives were supervised by physicians, or they shared their client caseload with physicians. The Cochrane reviewers were not interested in comparing midwives versus doctors. They took it for granted that midwifery care itself is safe, effective, and satisfying. This is, after all, a global consensus, to which the United States remains in stubborn and lonely opposition. Working from the assumption that midwives are an important part of the maternity care system, the question becomes how should we organize that system? Who should coordinate the care of childbearing women - midwives or doctors? Midwife-led care means that women receive their primary maternity care with the midwife, and the midwife engages an obstetrician or other consultant when some aspect of the woman's or baby's care falls outside of the scope of independent midwifery practice. This stands in stark contrast to the typical arrangement in the United States, when midwives are supervised by obstetricians or employed by hospitals, and obstetric protocols and productivity standards drive midwifery practice.

Women often believe that going to an obstetrician practice that employs midwives is getting 'the best of both worlds.' The Cochrane review of midwife-led care in fact tells us that such arrangements are ineffective, inefficient, and may be hazardous to the health of women and babies.

Citation: Hatem M, Sandall J, Devane D, Soltani H, Gates S. Midwife-led versus other models of care for childbearing women. Cochrane Database of Systematic Reviews 2008, Issue 4. Art. No.: CD004667. DOI: 10.1002/14651858.CD004667.pub2.

Tags