Editor’s Choice

Systematic reviews

The number of review articles submitted to BJOG has risen six-fold since 2003. Perhaps even more importantly, the number of systematic (as opposed to narrative) reviews submitted has risen from two in 2003 to 50 in 2009, and they now make up more than half of all reviews submitted. Constructing a systematic review is a complex technical task, requiring appropriate methodology and well-trained reviewers (I use the plural advisedly, because they usually require a team). The York Centre for Reviews and Dissemination has an excellent summary of the methodology on their website (http://www.york.ac.uk/inst/crd/pdf/Systematic_Reviews.pdf). Requirements include careful framing of the review question and inclusion criteria, a comprehensive search strategy, quality assessment, and data extraction and synthesis. Information on how to write up the results can be found on the equator network website (http://www.equator-network.org/). The resulting papers should conform to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses, PLoS Med 2009 Jul 21;6(7):e1000097. Epub 2009 Jul 21), which replaces the QUOROM statement (QUality Of Reporting Of Meta-analyses). All of this indicates that undertaking a systematic review is not for the fainthearted, or the technically uneducated. But the work put into them improves the reliability of the conclusions, and hence their relevance to clinical practice. Analysis of our reader survey of July/August 2009 (and thank you to the hundreds of readers who participated) shows that such reviews are the most popular type of article with our readers, followed by narrative reviews, the editor’s choice (hurrah!) and main articles. They are also highly cited (good for our impact factor). I am therefore unrepentant that this month we publish no fewer than four systematic reviews.

Is period pain relieved by acupuncture?

The dominance in Western culture of so-called ‘allopathic medicine’ is based on the use of the scientific approach, which tests hypotheses. There are, however, other disciplines which have developed according to different philosophies, such as homoeopathy, osteopathy, chiropractic and Ayurveda (traditional Indian medicine). Acupuncture has been used for several millennia in China and its environs. It is therefore appropriate that the systematic review of its use for primary dysmenorrhoea by Cho and Hwang on page 509 comes from the Hospital of Korean Medicine in Seoul. Despite claims by some practitioners of ‘complementary medicine’ that the use of randomised controlled trials is inappropriate for disciplines not based on the scientific method, it seems likely that widespread acceptance of complementary medicine will only occur if it develops a scientific evidence base. Although Cho and Hwang identified 27 randomised controlled trials, only nine described clearly their methods of randomisation, and none described how they concealed group allocation. This is important because although the overall result suggested a reduction in period pain with acupuncture, a combination of the two trials that compared acupuncture with sham acupuncture did not find a significant difference, which raises the possibility of a placebo effect. The potential mechanisms by which acupuncture might work (assuming that we do not accept the idea that it redirects the flows of Qi [life energy] within the body) are discussed. The authors end with a plea that future researchers should follow the guidelines for clinical trials laid out in the CONSORT statement (http://www.consort-statement.org/) and the STRICTA recommendations for trials involving acupuncture (http://www.stricta.info/).

Reducing pain at hysteroscopy

All gynaecologists must be aware that pelvic examination is embarrassing for many women, uncomfortable for all, and painful for some. So approaches to mitigate pain are always valuable. Clark et al. (page 532) describe a technique for hysteroscopy that not only avoids the need to grasp and steady the cervix, it even avoids the need for a vaginal speculum. The vagina is distended with appropriate fluid, and the hysteroscope is steered into the cervical canal under visual control. This technique significantly reduced the amount of pain reported by the women, without a significant increase in the number of failed procedures. If you are not already using a vaginoscopic technique for hysteroscopy, you should read this paper and reconsider.

Protecting the vagina from radiotherapy damage

One of the great successes of radiotherapy from its earliest days has been the treatment of carcinoma of the cervix. It is particularly useful for avoiding surgery in frail women but there are serious adverse effects including stenosis, shortening and loss of elasticity of the vagina. In an attempt to minimise such adverse effects, numerous narrative reviews have recommended ‘dilatation three times weekly for an indefinite time period’, and this is standard practice in the UK. However, dilatation can be painful and embarrassing, so Johnson et al. thought they should check that this advice is evidence-based. Their systematic review on page 522 concludes that there is in fact no good evidence that routine vaginal dilatation during, or immediately after, pelvic radiotherapy is of benefit, and it may even be harmful. This may be another example of someone many years ago having an idea that seemed reasonable at the time but that was then perpetuated from one review to the next, becoming embedded both in the literature and in clinical practice, without ever being properly questioned. However, the data remain sparse, and there is a clear challenge to those who might wish to cling to their traditional advice to carry out more randomised controlled trials.

Depression after preterm birth

Systematic reviews are not easy to do, not just because they require a rigorous application of a demanding methodology, but sometimes because the data on which they are based are flawed. Vigod et al. On page 540 investigated the question of whether mothers of preterm babies are more likely to become depressed through the first year of their baby’s life than mothers of babies born at term. They had a number of problems regarding definitions. First, papers did not all use a single standardised measure of depression (such as the Edinburgh postnatal depression scale) and second, they sometimes failed to use the standard definition of preterm birth (<37 completed weeks of gestation). This made it difficult to compare studies in the usual quantitative way. Accordingly, the reviewers had to use qualitative techniques to combine varying measures of depression, and they also included studies defining preterm birth as <38 weeks of gestation, which would inevitably include a large number of babies considered to be ‘term’ by conventional standards. Because of the stress for parents associated with having a preterm baby needing intensive care, one would a priori expect a higher incidence of depression, and this was confirmed. Perhaps the most important point is that this was still present 1 year postnatally, and the authors recommend that clinicians are vigilant for signs of treatable depression in this vulnerable group.

Swine flu—pandemic or damp squib?

About 6 months ago, it was suggested by several editors that we should solicit a paper on ‘swine flu’—influenza A strain H1N1 (the name is something of a misnomer, because some viral gene segments are derived from human and avian strains, and not just from pigs). However, I had reservations, not just because we are a monthly journal with an inevitable lag time between submission and publication but because it was not clear at that time how dangerous the strain was going to be. In the event, my caution seems to have been justified, because death rates in the general population in developed countries seem, if anything, to have been less than with the usual seasonal strains, to the extent that drug companies have been accused of profiteering by exaggerating the risk. However, it is easy to be wise after the event. Moreover, pregnant women react differently to many infections, with lung involvement being particularly prominent (as for example with varicella-zoster virus and with previous influenza pandemics) and maternal deaths from H1N1 virus infection in the USA have been reported (Lancet 2009;374:451–8). Moreover, while on secondment to the maternity department at Groote Schuur Hospital in Cape Town, South Africa, in September/October 2009, I witnessed daily admissions of pregnant women with severe respiratory difficulty associated with H1N1 virus infection, and we have also had a steady trickle of such women through our own service in London from November to January, although luckily no deaths. The question remained—what is the usual severity? Accordingly, we are pleased to publish on page 551 the report by Lim et al. of 211 pregnant women with H1N1 virus infection seen at the KK Women’s and Children’s Hospital in Singapore between 26 May 2009 and 14 September 2009. Cough was the most prevalent symptom, occurring in 90.5%. Other recorded symptoms were: runny nose (62.1%), sore throat (58.8%), muscle ache (32.2%), headache (18%) and breathlessness (13.3%); 62.2% presented to hospital with fever. More than two-thirds of the women did not need to be admitted; there were two cases of pneumonia, one requiring admission to intensive care. Both recovered. The authors’ conclusion was that H1N1 virus generally causes a mild infection in pregnancy. This time we were lucky.

Pulmonary hypertension in pregnancy

I have long known that pulmonary hypertension and pregnancy are a dangerous mixture, and this knowledge was brought into sharp focus 11 years ago when a woman under my care died suddenly on the operating table, just minutes after seeing her newborn baby for the first time. At the subsequent inquest, the importance of giving as accurate a prognosis as possible to women with pulmonary hypertension who are thinking about becoming pregnant was emphasised. We have generally counselled that mortality is between 30 and 50%, but in the last 10 years, all the half-dozen women we have cared for in our service have survived their pregnancies. So does that mean that with modern treatment the risk is less? Our advice inevitably relies on small series of this relatively rare condition. We are therefore pleased to publish on page 565 the series of ten pregnancies in nine women cared for in two large UK referral hospitals by Kiely et al., there were no deaths directly attributable to the pregnancies. However, most needed delivery between 32 and 35 weeks, exposing their babies to the risks of prematurity. These encouraging results may represent a type II error—survival being the result of random good fortune—and also reporting bias (units with poor outcomes may be less likely to report their results). Ideally, a prospective international register of such cases should be established, to provide a more reliable estimate of risk.

And finally …

The hyperglycaemia and adverse outcome study (HAPO, N Engl J Med 2008;358:1991–2002) has added greatly to our understanding of the effects of hyperglycaemia on pregnancy outcome. On page 575, the writing group led by Boyd Metzger report that a high maternal body mass index is associated with an increased frequency of pregnancy complications, even allowing for variations in blood sugar. And my enthusiasm for individualising care and avoiding routine pregnancy supplements unless specifically indicated has been strengthened by the case reported by Aslam et al. On page 620. Routine iron supplementation was associated with acute postpartum jaundice in an Irish woman homozygous for the C282Y mutation of gene HFE on chromosome 6p21.3. Approximately one in five to one in ten people of Celtic origin are carriers of this mutation, which greatly enhances iron absorption, with one in 200 being homozygous and at risk of iron overload (in the most western Celts, the Irish, more than 1% are homozygous), although variable penetrance results in clinical haemochromatosis in only about 1 in 5000. Women are relatively protected because of blood loss at menstruation. C282Y is now recognised as one of the most common genetic mutations in Celtic peoples and its frequency may have increased to compensate for the infertility induced by gluten intolerance (coeliac disease), which is also common in Celts, as they converted from hunter-gatherers to wheat farmers over the last several thousand years (Whittington Medical Hypotheses 2006;66:769–72).