The trial tested if metronidazole, a broad-spectrum antibiotic, could cure suspected infection during pregnancy and prevent premature delivery.

Previous research showed that levels of the protein fibronectin were an accurate predictor of premature labour - if fibronectin is detected in a vaginal swab, a mother is more likely to have an early labour. Fibronectin is a protein which binds the maternal and fetal membranes together and when these membranes start to detach (before labour) fibronectin is released. An infection in the womb is thought to trigger this membrane separation (thus fibronectin release) and the hypothesis was that treatment with antibiotics could delay the onset of labour. The trial treated some women with antibiotic and some women with a placebo.

In this study, 117 women were identified as fibronectin-positive (i.e. fibronectin was detected in a cervical swab) and 98 of these were randomised to receive either the antibiotic metronidazole or a placebo. 33 women in the treatment group delivered before 37 weeks' gestation compared with only 18 in the placebo group (p = 0.06). Analysis showed a difference in delivery before 37 weeks of 62% (treatment group) compared to 41% (placebo group) (p = 0.051).

On the basis of this trial, treatment of women at risk of preterm labour with metronidazole is not recommended. In fact it may adversely affect the pregnancy outcome.

Although it is known that infection does play a role in preterm labour, its specific role is now less convincing and there are perhaps more factors and interactions that have not yet been fully elucidated.