Women who have had a previous preterm birth

In a double-blind randomised controlled trial, 241 women who had previously had a spontaneous preterm birth before 34 weeks' gestation and who were planning further pregnancies, were randomised to oral azithromycin (two 1g doses, 4 days apart) plus sustained-release metronidazole (750mg daily for 7 days), or matched placebo.13 Treatment was repeated every 4 months until conception. In all, 124 women became pregnant and were available for study (which was fewer than the calculated sample size required). Antibacterial therapy had no significant effect on the rate of preterm birth occurring before 37 weeks' gestation (the primary outcome measure: 52% vs. 46% with placebo, p=0.568).

Pregnant women with asymptomatic bacteriuria

A Cochrane systematic review examined the effect of antibacterial therapy on women with asymptomatic bacteriuria.14 The review included 14 randomised trials (which overall were considered to be of poor quality) involving a total of 2,302 women. Meta-analysis of data from the studies indicated that, compared to placebo or no treatment, antibacterial therapy (various regimens) did not reduce the likelihood of preterm birth (defined as delivery before 38 weeks gestation) but did reduce the incidence of low birthweight babies (8.5% vs. 13%, risk ratio [RR] 0.66 95% CI 0.49 to 0.89). Bacteriuria persisted in fewer women who received antibacterial therapy (17.4% vs. 65.8%, RR 0.25, 95% CI 0.14 to 0.48). It was also effective in reducing the incidence of pyelonephritis (5.5% vs. 20.5%, RR 0.23, 95% CI 0.13 to 0.41). The reviewers calculated that seven pregnant women (95% CI 6 to 8) would need to be treated with antibacterials to prevent one episode of pyelonephritis. A more recent meta-analysis (involving six randomised or quasi-randomised controlled trials) which was part of a UK health technology assessment, found that antibacterial therapy was effective in helping prevent preterm birth at less than 37 weeks gestation (9.8% vs. 18.9% in controls, RR 0.47, 95% CI 0.33 to 0.69), as well as in reducing the incidence of pyelonephritis.10

Pregnant women with bacterial vaginosis

The efficacy of antibacterials in the prevention of preterm birth in women with asymptomatic vaginal flora abnormalities has been assessed in several meta-analyses, which have provided conflicting results.1,10,15–17 One of these analyses showed that, in women with bacterial vaginosis (15 trials involving a total of 5,888 participants), antibacterial therapy did not reduce the likelihood of preterm birth or of preterm prelabour rupture of membranes.15 In women with bacterial vaginosis and a previous preterm birth (five trials in a total of 622 women), antibacterial therapy was similarly ineffective in preventing preterm birth. However, it appeared to decrease the likelihood of preterm prelabour rupture of membranes (3% vs. 28%, odds ratio [OR] 0.14, 95% CI 0.05 to 0.38) and of low birthweight (13% vs. 34%, OR 0.31, 95% CI 0.13 to 0.75), in two trials involving a total of 114 women.15 By contrast, in a small subgroup of women with abnormal vaginal flora (“intermediate flora” or bacterial vaginosis, two trials involving a total of 894 women), antibacterials were apparently effective in helping to prevent preterm birth (6% vs. 11%, OR 0.51, 95% CI 0.32 to 0.81). The only other subgroup in whom antibacterials appeared effective was women treated before 20 weeks' gestation, who had a reduced likelihood of preterm birth (five trials involving a total of 2,287 women, 9% vs. 11%, OR 0.72, 95% CI 0.55 to 0.95).15 Only one meta-analysis (including 10 randomised controlled trials involving a total of 3,969 women) reported that in a subgroup of women who had had a previous preterm delivery, and who had oral antibacterial treatment for at least 7 days, there was a significant decrease in the likelihood of preterm delivery (OR 0.42, 95% CI 0.27 to 0.67).17

What NICE says

The National Institute for Health and Clinical Excellence (NICE) recommends that pregnant women should be offered routine screening for asymptomatic bacteriuria early in pregnancy because identification and treatment of the condition reduces the risk of pyelonephritis.18 NICE recommends that pregnant women should not be offered routine screening for bacterial vaginosis because of the evidence suggesting that identifying and treating those with the condition does not lower their risk for preterm birth and other adverse reproductive outcomes.18

Women with spontaneous preterm labour and intact membranes

A systematic review collated evidence on antibacterials used with the aim of inhibiting preterm labour in women thought to be in preterm labour with intact fetal membranes (11 randomised controlled trials involving a total of 7,428 participants).19 The reviewers found no clear evidence of benefit to either mother or baby from such treatment compared to placebo or no treatment. They concluded that prophylactic antibacterials could not be recommended in the routine management of women in preterm labour with intact membranes.

Babies born to women in the largest trial included in the systematic review (ORACLE II trial involving 6,295 women) were followed up until the age of 7 years.20 The primary outcome of this follow-up study was the presence of any level of functional impairment, classified according to the mark III Multi-Attribute Health Status (a classification system that assesses vision, hearing, speech, ambulation, dexterity, emotion, cognition and pain). Also, educational outcomes were assessed using results from national curriculum tests. Outcomes were determined for over 3,000 children (71% of the initial study cohort). Perhaps surprisingly, given the lack of difference in short-term outcomes between the study groups, children whose mothers had taken erythromycin (with or without co-amoxiclav) had a higher rate of any functional impairment at age 7 years than did children whose mothers had received no erythromycin (42.3% vs. 38.3%, OR 1.18, 95% CI 1.02 to 1.37). By contrast, use of co-amoxiclav (with or without erythromycin) had no effect on the proportion of children with any functional impairment, compared to that among children whose mothers did not receive co-amoxiclav. However, babies of those women given either erythromycin or co-amoxiclav had a higher incidence of cerebral palsy than babies of women who had received no erythromycin or no co-amoxiclav, respectively (erythromycin 3.3% vs. 1.7% with placebo, OR 1.93, 95% CI 1.21 to 3.09; co-amoxiclav 3.2% vs. 1.9% with placebo, OR 1.69, 95% CI 1.07 to 2.67).

Overall, there is strong evidence that co-amoxiclav and erythromycin given in the scenario of threatened preterm labour with intact fetal membranes and with no obvious signs of infection, are of no benefit to the baby in the short term and harmful in the long term. It follows that antibacterials should not be given routinely to women in threatened preterm labour with intact membranes. Despite the clear evidence from these studies, it is worth noting that 16% of UK maternity units surveyed were still giving antibacterials routinely to women in spontaneous preterm labour with intact membranes in 2009.21

Women with preterm membrane rupture

A systematic review of evidence on the use of antibacterials in women with preterm rupture of membranes included 22 randomised controlled trials involving a total of 6,800 women, and their babies.22 Most of the women recruited to the studies were not in active labour (i.e. they had preterm prelabour membrane rupture). Antibacterial therapy was associated with a reduction in the likelihood of chorioamnionitis (16.4% vs. 24.8%, average risk ratio 0.66, 95% CI 0.46 to 0.96) and with fewer babies being born within 48 hours (31.4% vs. 39.9%, average RR 0.71, 95% CI 0.58 to 0.87) and within 7 days, of randomisation (57.6% vs. 67.1%, average RR 0.79, 95% CI 0.71 to 0.89). There were also reductions in the following markers of neonatal morbidity: neonatal infection, use of surfactant, oxygen therapy and abnormal cerebral ultrasound scan before discharge from hospital. Necrotising enterocolitis was less likely with erythromycin than with co-amoxiclav (one trial involving 2,395 women: 0.92% vs. 1.99%, RR 0.46, 95% CI 0.23 to 0.94).

Babies born to women in the largest trial (ORACLE I involving 4,826 women) included in the systematic review were followed up until the age of 7 years.23 This follow-up study found that antibacterials seemed to have little effect on the health of the children as assessed by the mark III Multi-Attribute Health Status classification system and educational outcomes.

Specialist guidance

The UK Royal College of Obstetricians and Gynaecologists recommends that women presenting with symptoms and signs of preterm labour with premature membrane rupture should be prescribed erythromycin for 10 days.24 The guideline states that the use of co-amoxiclav is best avoided because it carries an increased risk of neonatal necrotising enterocolitis.

Potential clinical dilemmas

In women at risk of preterm delivery, the clinical distinction between those who have rupture of membranes and those who have intact membranes is not always clear. There are no sufficiently robust objective tests, so the diagnosis depends on identification of a pool of fluid in the vagina. This will be absent in women who have very little amniotic fluid left in the uterus, or in whom the fluid has drained from the vagina (giving the test a false-negative result); also fluid could be present in women with a watery vaginal discharge for other reasons (giving a false-positive result). Bedside ultrasonography is commonly used to attempt to clarify the situation, with a low ‘amniotic fluid index’ (AFI) supporting but not confirming the diagnosis of membrane rupture; a normal AFI makes membrane rupture less likely but does not exclude it.

Incorrect diagnosis of a woman with intact fetal membranes as having premature membrane rupture would result in a woman being given antibacterials and increasing the long-term risks of adverse outcomes (including cerebral palsy) for the baby. In contrast, failure to give antibacterials to a woman correctly identified as having premature membrane rupture would result in some short-term morbidity without the antibacterial-related increase in adverse outcomes in the long term. Therefore, it could be argued that the best long-term outcome would result from a policy of not giving antibacterials to a woman with symptoms and signs of preterm labour, unless there are obvious signs of infection or clinical certainty that membrane rupture has occurred. In reality, however, as all the trials were clinically based and involved clinicians judging whether a woman had intact or ruptured membranes using the clinical tools available to them, it follows that antibacterials should be used if preterm membrane rupture is believed to have occurred.