- Fiona Godlee, editor, BMJ
Two studies this week address clinically important questions in obstetrics. Firstly, can a single progesterone measurement predict whether a woman who develops pain and bleeding in early pregnancy will lose her baby? As Jorine Verhaegen and colleagues explain in their meta-analysis, these symptoms occur in 30% of pregnant women during the first trimester and are the commonest cause of consultation in early pregnancy (doi:10.1136/bmj.e6077). The authors analysed 26 cohort studies with data on more than 9000 women. They conclude that a single low progesterone level is highly predictive of a non-viable pregnancy in women with symptoms and an inconclusive ultrasound scan.
For those women who get through the first trimester, there is still the risk of preterm delivery, affecting nearly 15 million babies each year, more than 10% of all births. Efforts to delay delivery have resulted in a multiplicity of treatment options, but which is best? This is proving a hard question to answer given the lack of good direct evidence comparing each option with all others. So David Haas and colleagues have now done a network meta-analysis, combining direct evidence from head to head comparisons with indirect evidence against a common comparator (doi:10.1136/bmj.e6226). Figure 2 on bmj.com explains this better than words and makes clear the extent of the task these authors undertook.
They conclude that prostaglandin inhibitors are probably the best first line tocolytic agent, with calcium channel blockers a close second. But their paper reflects a great deal of uncertainty, not least about whether any of these agents is better than placebo in reducing neonatal morbidity and mortality. As Zarko Alfirevic emphasises in the linked editorial, delaying delivery makes sense to buy time for the mother to receive antenatal corticosteroids, but there is as yet no evidence that “forcing babies to stay in a potentially hostile uterine environment for another week, or even longer, is clinically beneficial” (doi:10.1136/bmj.e6531). Given this uncertainty, women need to be told that, although these drugs may prolong pregnancy, they may not make their babies healthier.
Obstetrics was famously awarded the wooden spoon by Archie Cochrane in 1979 for being the least evidence based medical specialty. It has done many things since then to redeem itself, including creating the first compendium of evidence, Effective Care in Pregnancy and Childbirth, which was the model for the BMJ Group’s Clinical Evidence (clinicalevidence.bmj.com). Systematically reviewing the evidence on effectiveness and safety of common treatments is a huge task and not one that should be undertaken lightly. So I don’t know whether to congratulate or question the efforts of the two Frenchmen whose book, Guide to 4000 Useful, Useless or Dangerous Medicines, has quickly become a best seller. In the wake of the Mediator scandal (doi:10.1136/bmj.c6882), it aims to stem France’s love affair with pharmaceuticals (doi:10.1136/bmj.e6996). Whether right or wrong about individual drugs, its success highlights a hunger for independent information, while France waits for its new NICE-like body, the National Agency for the Safety of Medicines and Health Products (ANSM) set up in May, to get running.
Cite this as: BMJ 2012;345:e7017