SEARCH

SEARCH BY CITATION

Recurrent stillbirth

  1. Top of page
  2. Recurrent stillbirth
  3. Knowing more about what happens to the endometrium during menstruation
  4. Emergency contraception
  5. Complementary medicine
  6. The psychology of successful medical care

In last month’s Editor’s Choice, I discussed the way in which science advances our understanding of the physical world, emphasising the ‘conditionality’ of knowledge (all hypotheses are potentially disprovable). This month we are publishing a paper by Bhattacharya et al. on page 1243 which both updates and substantially modifies the conclusions of a study by the same group published in BJOG as recently as 2008 (BJOG 2008;115:269–74). In the first study they looked at the outcomes of pregnancy following a stillbirth in 360 women in the Grampian region of Scotland and concluded that the odds ratio for recurrent stillbirth (following adjustment for confounding factors such as pre-eclampsia, placental abruption, prematurity and low birthweight) was 1.2, with wide 95% CI of 0.4–3.4 (the time period studied was 1976–2006). They concluded that ‘after adjusting for confounding factors, there was no significant increase in the odds of recurrent stillbirth.’ In this month’s issue, on page 1243, their much larger study of 2677 pregnancies following stillbirth (from the whole of Scotland) now shows a much larger odds ratio of 1.94, with 99% CI of 1.29–2.92 (the time period studied was 1981–2000). Why the difference, and was the previous study wrong? And if so, were we right to have published it? The authors suggest a possible type II error in their first study (the statistics only showed that a large difference was unlikely, and the numbers were relatively small). We need to remember that although we tend to concentrate on the possibility of type I errors (e.g. with P = 0.05 there is a one in 20 likelihood that a positive difference is due to chance), type II errors (not detecting a difference when it actually exists) can be just as important. Curiously, we tend to put less weight on this latter possibility. For example, in prospective studies the ‘power’ to avoid missing a real difference is commonly set at 80% (equivalent to P = 0.2 for a type I error) rather than 95% (equivalent to 0.05). Another possibility that Bhattacharya et al. suggest for the different conclusion is an inability to adjust for body mass index in their latest study, although it seems unlikely that obesity would have such a large effect.

The papers present a further interesting statistical issue, in that the larger data set included the data reported in 2008. Bearing in mind that complex statistics can be difficult for the general reader to understand, we commissioned a commentary from one of our statistical advisers, Amy Herring of the Department of Biostatistics at the University of North Carolina (USA). Her very clear and informative commentary on page 1173 comments on the use of a Bayesian approach. Thomas Bayes (1702–1761) was a Presbyterian minister whose interest in logic led him to study probability. In essence, his work led to the view that the results of observations should, where possible, be interpreted in the context of prior knowledge. For example, a positive mammogram with a 5% false-positive rate is more likely to be correct in a woman carrying the BRCA1 gene who has a prior lifetime probability of developing breast cancer of 50% than in the average woman where the lifetime risk is 5%. We can interpret a study based entirely on the results in that study alone (the ‘frequentist’ or ‘traditional’ approach), or we can interpret it in the context of prior probabilities (the ‘Bayesian’ approach). So in the case of pregnancy following previous stillbirth, we should look at the results in the context of previous studies (incorporating them in a form of ongoing meta-analysis) rather than the results of the individual study alone. This emphasises that knowledge is ‘conditional’, and that we should be regularly modifying our views on a particular topic as new information becomes available. I only wish that journalists would do this, they often report ‘the shock findings’ of the latest study without putting them in context. After reading these two papers, you will be in the position of being able to advise journalists on the proper Bayesian approach!

Knowing more about what happens to the endometrium during menstruation

  1. Top of page
  2. Recurrent stillbirth
  3. Knowing more about what happens to the endometrium during menstruation
  4. Emergency contraception
  5. Complementary medicine
  6. The psychology of successful medical care

Over the years, studies published in BJOG have tended to change from straightforward clinical observations to large and complex multicentre prospective randomised trials. However, as I often emphasise to trainees, there is still a lot of publication mileage in simple, straightforward, detailed and accurate observations. It might be thought that we already know what happens to the endometrium during menstruation. However the paper by Garry et al. on page 1175 illustrates that the endometrium does not regenerate following menstruation by mitotic cell division of already differentiated cells, but that new glands are formed by the differentiation of stem cells from the surrounding stroma. This important conclusion was derived from the study of 13 women undergoing a hysterectomy or curettage for benign indications. Their superb illustrations derive not only from standard haematoxylin & eosin staining, but also from immunostaining and scanning electron microscopy. I suspect that not only their new theory, but also their illustrations, will soon be finding their way into the textbooks.

Emergency contraception

  1. Top of page
  2. Recurrent stillbirth
  3. Knowing more about what happens to the endometrium during menstruation
  4. Emergency contraception
  5. Complementary medicine
  6. The psychology of successful medical care

Work continues to define the optimum regimen for medical termination of first-trimester pregnancy (see for example the paper by von Hertzen et al. on page 1186, which suggests that 800 μg misoprostol administered 24 hours after 200 mg oral mifepristone has the best balance between efficacy and side effects), although it would be ideal if all unwanted pregnancies could be prevented. However, contraception requires careful planning and there will always be occasions (whether coercive or voluntary) when unwanted conception is a risk. Emergency contraception is either hormonal or uses the insertion of an intrauterine device. In this month’s issue we have important papers about both of these approaches. On page 1197, Gaudineau et al. report their study of the use of the emergency contraceptive pill by 15-year-old girls in 11 European countries in 2006. The proportion of girls reporting themselves as having had sexual intercourse varied from 17.2% in Greece to 40.6% in Wales, and of these, the proportion who reported the use of the emergency contraceptive pill ranged from 1.7% in Hungary to 17.8% in France. This latter high figure may reflect the fact that in France, the emergency contraceptive pill is available free of charge for women under the age of 18 ‘over-the-counter’, and even from school nurses. Moreover, French girls had one of the higher rates of using a condoms/birth control pills, suggesting that the ready availability of the emergency pill does not encourage riskier behaviour.

On page 1205, Wu et al. report their study of almost 2000 Chinese women requesting emergency contraception within 120 hours of unprotected intercourse. Not only does the copper intrauterine device have a lower failure rate than the emergency contraceptive pill, it also continues to be effective when placed up to 5 days after unprotected intercourse, and moreover continues to provide contraception when left in place. The downside is that it requires a trained professional to insert it. However insertion was straightforward in 98.5% of women, and there were no pregnancies diagnosed before or at the first follow-up visit (the authors estimate that 35 pregnancies were prevented). Ninety-four percent of women continued to have the intrauterine device (the copper T380A) in situ at the 12-month follow up.

Complementary medicine

  1. Top of page
  2. Recurrent stillbirth
  3. Knowing more about what happens to the endometrium during menstruation
  4. Emergency contraception
  5. Complementary medicine
  6. The psychology of successful medical care

As the credit crunch starts to bite, the pressure to consider the cost-effectiveness of medical interventions is growing. In the UK, complementary (sometimes called alternative) medicine has been under a particular spotlight. Although there has long been doubt about the efficacy of techniques such as homoeopathy and acupuncture, their proponents have always said ‘but at least they don’t do any harm,’” and it is said that they make the recipients ‘feel better’. But does feeling better justify spending large sums of taxpayers, money on something that may not be cost-effective, if it works at all? For something to even register on the cost-effectiveness scale, it has to do something measurable against which the cost can be set (the main element in the cost usually being the practitioner’s time). In this month’s issue, on page 1255 we publish a paper from Denmark by Modlock et al. reporting the results of a randomised controlled trial of acupuncture for the induction of labour. In the 24 hours after treatment, seven women (12%) in the acupuncture group and eight women (14%) in the group given sham acupuncture had gone into labour or delivered. Perhaps it was the ‘wrong sort of acupuncture’. Why not read their paper and decide…

The psychology of successful medical care

  1. Top of page
  2. Recurrent stillbirth
  3. Knowing more about what happens to the endometrium during menstruation
  4. Emergency contraception
  5. Complementary medicine
  6. The psychology of successful medical care

There is growing appreciation that a major contributor to poor outcomes is a failure to apply appropriate medical care. It is estimated that as many as one in ten people admitted to hospital will die or become disabled as a result of medical error. Even when there is evidence-based effective treatment, it is often not carried out correctly or at all, leading to the increased use of tools for monitoring performance, such as the CUmulative SUM (CUSUM) charts reported by Boulkedid et al. on page 1225. On page 1278, Deneux-Tharaux et al. report a cluster randomised controlled trial of trying to reduce the incidence of severe postpartum haemorrhage by encouraging the use of a standardised protocol. As they say, it is one thing to prepare evidence-based guidelines, but ‘the harder job is ensuring the actual translation of these guidelines into clinical practice.’ They found that while some elements of care improved, there was no observable difference in postpartum haemorrhage rates between the units where the use of the protocol was encouraged, compared with no intervention. However, rates of postpartum haemorrhage decreased in both groups. An important component in effective medical care is the ability of professionals to work together in teams. In a paper on page 1262, Siassakos et al. report that measuring the knowledge, manual skills and attitudes of the individuals in a team did not correlate with the team’s clinical efficiency in the management of simulated eclampsia. We clearly need to know more about what makes a successful team. I suspect that ‘leadership’ may be a key element. In terms of individual performance, failure correctly to identify fetal heart rate abnormalities in labour, or to respond to them in a timely fashion, has been implicated in more than half of preventable fetal intrapartum death or damage. Work has been going on for over 25 years to produce a computerised system to assist in the interpretation of fetal heart rate patterns, and on page 1288 Costa et al. describe the Omniview-SisPorto system, which was developed for this purpose. They report a retrospective study of using the system in conjunction with clinical judgement by three experienced clinicians, to predict umbilical artery pH at birth. The system facilitated the prediction of pH to within a 0.1 margin in 70% of 104 births, compared with only 46% in 100 controls where it was not used. But will clinicians use it in practice? The system is now being put to the test in a randomised clinical trial (http://www.controlled-trials.com/ISRCTN42314164). It has taken a long time, but the study of how to encourage people to do the right thing is now being seen as just as important as knowing what the right thing is.