Intended for healthcare professionals


Infection and preterm delivery

BMJ 1999; 318 doi: (Published 27 February 1999) Cite this as: BMJ 1999;318:548

There is not yet enough evidence that antibiotics help

  1. Peter Brocklehurst, Clinical epidemiologist.
  1. National Perinatal Epidemiology Unit, Radcliffe Infirmary, Oxford OX2 6HE

    Preterm delivery is defined as delivery before the 37th completed week of pregnancy, and in 1996in Scotland 84% of neonatal deaths of normally formed infants were associated with preterm delivery.1 The aetiology of preterm delivery is poorly understood, though recent evidence suggests that infection may be implicated in a substantial proportion of cases.2 The part that infection plays in the development of preterm labour or preterm, pre-labour rupture of membranes leading to preterm delivery has been the focus of much research in recent years.

    One element of this work has been the finding of a strong association between the presence of bacterial vaginosis and preterm delivery. Bacterial vaginosis is a common infection of the female genital tract, caused by heavy concentrations of a mixed group of organisms, including Gardnerella vaginalis, Mycoplasma hominis, and anaerobes including curved rods and Mobiluncus species. Many of these organisms are present in small numbers in the vagina normally. Symptomatic infection is characterised by a grey vaginal discharge with a characteristic fishy odour. It is not associated with vaginal mucosal inflammation and rarely causes vulval itch.

    Bacterial vaginosis is often asymptomatic and is found in up to 20% of women during pregnancy depending on how often the population is screened.3 A substantial body of evidence now exists that associates bacterial vaginosis infection in pregnancy with a poor perinatal outcome, in particular an increased risk of preterm delivery.4 This strong association has led many researchers and clinicians to believe that bacterial vaginosis may be the cause of preterm delivery in these women. Therefore a series of treatment trials have been undertaken using antibiotics with known efficacy against bacterial vaginosis, in particular metronidazole and clindamycin.5 The results of these trials in preventing preterm delivery, however, have not been encouraging. 6 7

    There is some suggestion that treating women with a previous preterm delivery who have asymptomatic bacterial vaginosis in pregnancy may reduce the risk of subsequent preterm delivery. However, this has been observed in only two trials, one where it was found in a subgroup analysis of a small number of women6 and one which had substantial methodological problems.8 There is as yet no convincing evidence that screening all women antenatally for bacterial vaginosis and treating those affected will have any impact on the incidence of preterm delivery. The results of further continuing trials are awaited.

    Other investigators have concentrated on the observed association between spontaneous preterm delivery and subclinical or asymptomatic chorioamnionitis. In these women it is postulated that bacteria, from whatever source, set up an inflammatory reaction in the fetal membranes leading to the cascade of events culminating in preterm delivery. This hypothesis is currently being tested in a large randomised controlled trial, the ORACLE trial, which aims to determine whether antibiotics can improve neonatal outcome in women presenting with preterm labour or preterm pre-labour ruptured membranes.2 This trial aims to recruit 10000 women and should be completed in the year 2000.

    The ORACLE trial is likely to produce results directly relevant to clinical practice and policy. The disadvantage of many trials in this area is that their main outcome measure to assess the effectiveness of antibiotics is gestation at delivery. While this may superficially appear to be appropriate, increasing gestational age may not improve neonatal or maternal wellbeing. Firstly, antibiotics may benefit mother and baby without affecting the duration of pregnancy. Secondly, lengthening gestation may not confer any benefit to the fetus/neonate and may even result in harm, as suggested by overviews of trials of tocolysis in pregnancy.9

    Although advocates of the link between infection and preterm delivery may claim that antibiotics treat the cause of the condition rather than try to suppress the symptoms and are therefore fundamentally different from tocolysis, the problems of relying on gestation at delivery as an outcome measure remain.10 This has been highlighted by recent evidence about the well documented link between chorioamnionitis and cerebral palsy. Work in rabbits has shown that experimentally induced chorioamnionitis treated with antibiotics and delayed delivery results in white matter lesions in the fetal brain.11 Whether this damage is a consequence of the chorioamnionitis alone or the combination of chorioamnionitis and pregnancy prolongation is not yet known. But if infection in humans does lead to preterm labour or preterm pre-labour rupture of the membranes as a consequence of chorioamnionitis, either from the vagina or elsewhere, then the best management option may be delivery. Attempts to maintain the fetus in a hostile environment may result in more harm than benefit.

    Therefore any trial that evaluates the use of a treatment to prevent preterm delivery must show that the intervention benefits the baby and not just the obstetrician. Subclinical chorioamnionitis or bacterial vaginosis may well turn out to account for a substantial proportion of preterm deliveries. This has not yet been demonstrated, however, and until it has the use of antibiotics to prevent preterm delivery must continue to be seen as an experimental treatment which may result in more harm than good.