We were delighted to receive such a positive response from a leading group of researchers emphasising the need for a large randomised controlled trial to evaluate the effectiveness of a screening and treatment programme for the prevention of early onset neonatal sepsis with Group B Streptococcus. As there is currently no systematic screening in operation in the UK, this would seem an ideal opportunity and place to conduct such a trial. The results will have relevance throughout the world.
We propose that such a trial should use a cluster randomised design with the maternity hospital as the unit of randomisation. Units would be randomised to a ‘screen’ arm or a ‘no-screen’ arm. Hospitals in the ‘screen’ arm would carry out a policy of swab-based screening1 where universal screening for vaginal and rectal colonisation of all pregnant women takes place at 35–37 weeks of gestation. Intrapartum antibiotic prophylaxis would be given to women who screen positive as well as those at high risk of Group B Streptococcus transmission. Those maternity units randomised to the ‘no-screen’ arm would continue to practice as recommended by the RCOG guidelines.2 The main outcome of such a trial should be confirmed and presumed neonatal sepsis. To use Group B Streptococcus culture proven sepsis would be inappropriate for two reasons: first, most neonatal sepsis is ‘presumed’, that is, in less than half of all cases of neonatal sepsis is the infecting organism isolated on culture from a normally sterile site (usually blood or CSF). Therefore, measuring only culture proven Group B Streptococcus sepsis would under-estimate the incidence of all sepsis due to Group B Streptococcus,3 just as it would under-estimate the incidence of any specific cause of sepsis such as Escherichia coli. Second, if a women is given penicillin during labour and her baby then becomes ill with sepsis, it is likely that Group B Streptococcus would not be isolated because of the antibiotic contained within the blood4 even though Group B Streptococcus may be causing the baby's illness. Therefore, the primary outcome should be overall neonatal sepsis (confirmed and probable) up to seven days after birth. Sample size calculations would require the inclusion of approximately 540,000 births which is equivalent to approximately 60 maternity units participating over a three-year period. Such a trial is feasible and as noted in our paper there is interest in collaboration from maternity units. The research agenda with regard to Group B streptococcus is currently the subject of discussion by a working group of the UK National Screening Committee. We await the results with optimism.