Study aims and designs
We undertook a cluster randomised trial to determine if the use of point-of-care intrapartum rapid test for maternal GBS colonisation, implemented at a maternity unit level, can reduce maternal and neonatal antibiotic exposure, compared with usual care where antibiotics are offered based on maternal risk factors. Testing of mothers without risk factors for early-onset infection in their babies was outside the commissioned scope of the trial.
We also assessed the real-time accuracy of a rapid nucleic acid amplification test to detect maternal GBS colonisation in women presenting to a labour ward with risk factors for neonatal early-onset GBS infection against a reference standard of selective enrichment culture in a nested test accuracy study.
In a subset of mother-child pairs, we determined the antibiotic resistance profile of any GBS isolated, and the carriage rate of other antibiotic-resistant bacteria and resistance genes in the maternal rectovaginal samples and compared the findings with the offspring’s faecal sample at 6–12 weeks of age.
Study setting and participants
Twenty UK maternity units were clusters. The units were eligible to participate if they were prepared to accept a policy of rapid test-directed intrapartum antibiotic prophylaxis to prevent neonatal early-onset GBS infection and had access to microbiology facilities for selective enrichment bacteriological culture of GBS. Clinical midwives and doctors identified potential participants and screened for eligibility in various locations including the delivery suite, the maternity triage unit or the induction ward. No research-specific consent was obtained for the cluster trial and diagnostic study, although it was for the microbiology sub-study. Women in the rapid test units received information about the test and provided verbal assent to have the vaginal-rectal swab. Pregnant women were eligible for inclusion in the trial if they had one or more of the following risk factors: a previous baby with early- or late-onset GBS disease; GBS bacteriuria during the current pregnancy (irrespective of whether the GBS bacteriuria was treated at the time of diagnosis with antibiotics); GBS maternal colonisation of the vagina and/or rectum in the current pregnancy; suspected, diagnosed or established preterm labour (less than 37 weeks’ gestation); and maternal pyrexia (≥ 38 °C). Women were ineligible if they were under 16 years of age, less than 24 weeks’ gestation, were in the second stage of labour at admission or considered likely to give birth to their baby imminently, having a planned elective caesarean birth or their baby was known to have died in utero or had a congenital anomaly incompatible with survival at birth.
Cluster randomised trial
Randomisation and masking
Randomisation of clusters was performed at the Birmingham Clinical Trials Unit using a minimisation algorithm incorporating the following factors: region (the Midlands, London and South East England), pre-trial intrapartum antibiotic usage rate (above or below the median of all sites) and the number of vaginal or emergency Caesarean births (above or below the median). Due to the differences in the strategies for testing women and for directing intrapartum antibiotic prophylaxis, it was not possible to blind women or their care team to the randomised allocation of their maternity unit.
Procedures in rapid test and usual care units
Maternity units randomised to the rapid test received a GeneXpert® Dx IV GBS rapid testing system (Cepheid, Sunnyvale, USA) and a supply of XpertGBS test cartridges. Trained clinical midwives obtained vaginal and rectal maternal samples using a double-headed swab. One swab was used immediately for the rapid test according to the manufacturer’s instructions, and a result was obtained in less than 55 min. If the rapid test had not been initiated on the GeneXpert machine within 15 min of taking the swab, the test was considered invalid. The other swab was used for the diagnostic test accuracy study.
The rapid test units were advised to go against national guidelines, and only women who tested positive for GBS with the rapid test or for whom a test result was not available were to be offered intrapartum antibiotic prophylaxis, unless there was a clinical reason for prescribing antibiotics, or if the woman requested antibiotics.
If the woman had not given birth 48 h after the test result was available, the test result was regarded as invalid, and it was advised that the woman should be re-swabbed and retested for GBS colonisation.
Usual care units followed their standard risk-based screening strategy where intrapartum antibiotic prophylaxis was offered to all women with risk factors.
The recommended antibiotic regimen for preventing early-onset neonatal GBS infection in both types of units in the study was in line with national recommendations, where benzyl penicillin is the first-choice antibiotic for GBS prophylaxsis . Subsequent clinical management of mother and baby was based on local guidance .
The primary outcome was the proportion of women with risk factors who received intrapartum antibiotic prophylaxis to prevent neonatal early-onset GBS infection. The secondary maternal outcomes were intrapartum maternal antibiotic administration for any indication, indications other than Caesarean section, any postpartum maternal antibiotic use and exposure to antibiotics for greater than 2 or 4 h before delivery. The neonatal outcomes were the proportion of newborns who receive antibiotics for any indication, with suspected or diagnosed early-onset sepsis requiring antibiotics and neonatal mortality at any time until discharge from the hospital. We also reported serious adverse events in the mother or newborn.
The proportion of women with risk factors for early-onset GBS infection in their newborns receiving intrapartum antibiotic prophylaxis was expected to be between 50 and 75%, from previous estimates and expected improvements in adherence to guidelines since then . With a sample size per unit of 83 women and a minimum of 20 units, we expected the trial to have 90% power to detect a reduction to 63% in rapid test units (for a comparative control of 75%), assuming an intracluster coefficient of 0.01 . This equated to a total sample size of approximately 664 per strategy group, rounded up to 1340 in total, for the cluster randomised trial.
All trial analyses were conducted by intention-to-treat analyses according to the randomised allocation of the maternity unit, excluding any units who withdrew before data collection started and participants later found to be ineligible. For participant and cluster characteristics, we summarised the categorical data by frequencies and percentages. We summarised the continuous data by the number of responses, mean and standard deviation if deemed to be normally distributed and by the number of responses, median and interquartile range if the data appeared skewed. For the primary analysis in the cluster randomised trial, we used a mixed effects binomial regression with a log-link to estimate the relative risk, and a binomial model with an identity link to estimate the risk difference. Both models allowed for clustering by maternity unit as a random effect and adjusted for minimisation variables as fixed effects. If the binomial model with the identity link did not converge, we only reported the relative risk. In the case of non-convergence of the binomial model with a log-link, a Poisson model with robust standard errors was fitted. We used Kenward and Roger method to correct the potential inflation of the type I error rate due to the small number of clusters . We used GLIMMIX in SAS to estimate model parameters, using a restricted pseudo-likelihood approach based on a marginal expansion which can be viewed as a generalised form of REML (“RMPL” option in GLIMMIX). A post hoc analysis tested sensitivity to the estimation procedure and small sample correction method by comparing results obtained from an adaptive quadrature method (with between-within small sample correction) and using a between-within correction with a restricted pseudo-likelihood marginal expansion approach (again with a between-within small sample correction) . Overall inferences did not change. Where covariate adjustment was not practical, unadjusted estimates were produced and explained (e.g. not possible due to low event rate, lack of model convergence or poor recording accuracy of covariates).
We pre-specified subgroup analyses for the effects of the rapid test on the primary outcome in each of the categories based on the following maternal risk factors: maternal temperature of 38 °C or above observed whilst in labour, previous baby with GBS disease, GBS detected in current pregnancy and preterm labour (< 37 weeks’ gestation). We summarised the treatment effects within each sub-group separately and performed an interaction test between each subgroup variable and the test strategy allocation.
Diagnostic test accuracy study
We compared the rapid test findings against the reference standard of selective enrichment culture in a cohort study nested within the randomised trial. The second vaginal-rectal swab collected for the trial was returned to the transport tube and sent to the local microbiology laboratory for selective enrichment culture to detect GBS according to the recommended methods . For eligible women, a single swab taken from her baby’s ear canal was also processed in the local microbiology laboratory by selective enrichment culture to detect the presence of GBS. The rapid test results preceded those of the culture test, which was interpreted blindly to the rapid test. The main test accuracy outcomes were the sensitivity and specificity of the rapid test, but we also estimated neonatal and mother-to-baby GBS transmission rates.
For the test to be proven useful, it should detect a higher proportion of maternal GBS colonisations than other tests, but not at the cost of low specify or overdiagnosis. In the GBS1 study, intrapartum antibiotic prophylaxis directed by screening with enriched culture at 35–37 weeks’ gestation was considered to be the most acceptable cost-effective strategy , where the sensitivity of the antenatal screening test was 75.8% (95% CI 47.2 to 91.5%) . If the sensitivity of the proposed rapid test was higher than 90%, which was the approximate upper limit of the enriched culture test sensitivity, we expected the rapid test performance to be acceptable for use in clinical practice. A sample of 676 women would have 90% power to show that the estimated sensitivity of the rapid test was greater than a fixed value of 90%, based on 167 cases of maternal GBS colonisation in the units randomised to use the rapid test and 10% failed tests.
We estimated the diagnostic accuracy of the rapid test through the standard calculations of sensitivity and specificity. Point estimates were presented with 95% confidence intervals that were calculated using binomial exact methods . We also undertook a binomial proportion test to compare the observed sensitivity with a hypothesised minimal performance value of 90%.
Microbiological study of bacterial antibiotic resistance
In a subset of women from sites in London and South East England that were randomised to the rapid test strategy, we obtained individual consent to test additional vaginal-rectal maternal swabs and the faecal samples of their babies from 6 to 12 weeks of age for maternal antibiotic resistance profile of GBS, maternal colonisation by other antibiotic-resistant bacteria and the carriage of those specific bacteria or resistance elements by the infant. The swabs underwent selective enrichment culture for GBS, methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE) and extended-spectrum β-lactamase producing (ESβL) Enterobacteriaceae. The presence of any bacteria of interest was profiled by a variety of techniques (including, but not necessarily limited to, antibiotic resistance, molecular/genetic characterisation and matrix-assisted laser desorption ionisation time of flight mass spectrometry). Antibiotic resistance was tested using the EUCAST methods and break points . Enterobacterales were tested for sensitivity to ampicillin, piperacillin and tazobactam, amoxicillin and clavulanate, cefpodoxime, gentamicin, cefuroxime, amikacin, co-trimoxazole, temocillin, ceftazidime, ertapenem and ciprofloxacin on Muller Hinton agar. Molecular testing for Gram-negative antibiotic resistance genes was performed using the GSL EasyScreen™ Sample Processing Kit (SP006, Genetic Signatures Ltd., Newtown, Australia) designed to rapidly isolate nucleic acids (DNA and RNA) from clinical samples via an automated purification system. Fisher’s exact test was used to compare the proportions. We reported the relative risk of carriage of resistant E. coli or resistance genes in infants born to mothers with or without carriage of strains with specific characteristics using a binomial regression model with a log-link.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.