Group B streptococcus infection during pregnancy and infancy: estimates of regional and global burden

Background: Group B streptococcus (GBS) colonisation during pregnancy can lead to invasive GBS disease (iGBS) in infants, including meningitis or sepsis, with a high mortality risk. Other outcomes include stillbirths, maternal infections, and prematurity. There are data gaps, notably regarding neurodevelopmental impairment (NDI), especially after iGBS sepsis, which have limited previous global estimates. In this study, we aimed to address this gap using newly available multicountry datasets. Methods: We collated and meta-analysed summary data, primarily identified in a series of systematic reviews published in 2017 but also from recent studies on NDI and stillbirths, using Bayesian hierarchical models, and estimated the burden for 183 countries in 2020 regarding: maternal GBS colonisation, iGBS cases and deaths in infants younger than 3 months, children surviving iGBS affected by NDI, and maternal iGBS cases. We analysed the proportion of stillbirths with GBS and applied this to the UN-estimated stillbirth risk per country. Excess preterm births associated with maternal GBS colonisation were calculated using meta-analysis and national preterm birth rates. Findings: Data from the seven systematic reviews, published in 2017, that informed the previous burden estimation (a total of 515 data points) were combined with new data (17 data points) from large multicountry studies on neurodevelopmental impairment (two studies) and stillbirths (one study). A posterior median of 19·7 million (95% posterior interval 17·9–21·9) pregnant women were estimated to have rectovaginal colonisation with GBS in 2020. 231 800 (114 100–455 000) early-onset and 162 200 (70 200–394 400) late-onset infant iGBS cases were estimated to have occurred. In an analysis assuming a higher case fatality rate in the absence of a skilled birth attendant, 91 900 (44 800–187 800) iGBS infant deaths were estimated; in an analysis without this assumption, 58 300 (26 500–125 800) infant deaths from iGBS were estimated. 37 100 children who recovered from iGBS (14 600–96 200) were predicted to develop moderate or severe NDI. 40 500 (21 500–66 200) maternal iGBS cases and 46 200 (20 300–111 300) GBS stillbirths were predicted in 2020. GBS colonisation was also estimated to be potentially associated with considerable numbers of preterm births. Interpretation: Our analysis provides a comprehensive assessment of the pregnancy-related GBS burden. The Bayesian approach enabled coherent propagation of uncertainty, which is considerable, notably regarding GBS-associated preterm births. Our findings on both the acute and long-term consequences of iGBS have public health implications for understanding the value of investment in maternal GBS immunisation and other preventive strategies. Funding: Bill & Melinda Gates Foundation. We collated aggregated data for all outcomes (cases of maternal colonisation, iGBS cases in the first 3 months of life, deaths among infants younger than 3 months with iGBS, patients who recovered from iGBS with NDI, stillbirths caused by GBS, cases of maternal iGBS, and excess preterm births associated with maternal GBS colonisation). Most studies included were identified in a series of systematic reviews published in 2017,3, 7, 8, 9, 10, 14, 19, 20, 21 and we included new data on stillbirths and NDI. Note that the search strategy, selection criteria, data extraction, and assessment of study quality were described in the original publications of these systematic reviews. Bayesian hierarchical models22 were used to analyse the evidence. We estimated the absolute burden for 2020 on the basis of the country-specific numbers of births in the UN World Population Prospects 2019 for 183 countries included in the dataset based on population size (only countries with 90 000 or more inhabitants were listed). All relevant outcomes were studied as shown in figure 1 and panel 1, with predefined case definitions. Bayesian models allowed a coherent propagation of uncertainty in the estimation of the burden and accounted for the multilevel structure of the data (eg, for several outcomes, multiple studies were identified per region). Posterior medians and intervals (95% posterior intervals) were reported that account for the uncertainty in the parameter estimation; relevant sensitivity analyses were also reported. Parameters were estimated either at the country (eg, maternal colonisation), regional (eg, case fatality rate), or global (eg, risk of NDI after GBS meningitis) levels, depending on data availability (appendix pp 4–22). Details on the statistical models, including assumptions and prior probability distributions, are in the appendix (pp 4–22). Our estimation process followed the GATHER statement.18 Results are shown by Sustainable Development Goals regions. Overview of GBS colonisation in pregnancy and relevant outcomes During pregnancy, GBS can cause stillbirths, be associated with prematurity, and lead to maternal infection or, more rarely, death. Some newborn babies born to mothers who are GBS colonised develop invasive disease during the first week of life. Infants can also develop GBS disease between days 7 and 89. The fatality risk is high, and depends on adequate access to care. Of those children who survive the acute episode, some will be diagnosed with neurodevelopmental impairment. Neurodevelopmental impairment risks after GBS meningitis and GBS sepsis differ and were modelled separately. This figure was adapted from a study by Lawn and colleagues, with permission.3 GBS=group B streptococcus. GBS colonisation of the maternal genitourinary tract is associated with early-onset iGBS (EOGBS)24, 25 and quantification of its frequency is a key step in the estimation process. We included data from a systematic review,23 and developed a hierarchical model to regress country-level maternal GBS colonisation prevalence on country-level covariates.26 Using this approach, we estimated the colonisation prevalence in pregnant women for countries with and without GBS colonisation data, adjusting for study differences (appendix pp 4–5) in terms of the swab site and laboratory diagnostics, and accounting for parameter uncertainty. iGBS in infants might present as EOGBS (0–6 days) or LOGBS (7–89 days), and clinical syndromes include sepsis or meningitis (figure 1, panel 1). Given that routine surveillance data might underestimate EOGBS in most settings, as argued previously by Edmond and colleagues6 and Dagnew and colleagues,27 here we estimated the risk of EOGBS in neonates born to mothers who were GBS colonised. For modelling EOGBS risk, it was essential to consider intrapartum antibiotic prophylaxis use; hence we developed a model to regress study-level EOGBS risk in babies of mothers who were GBS colonised who received intrapartum antibiotic prophylaxis.2, 19 We used this regression in combination with national coverage data on intrapartum antibiotic prophylaxis, based primarily on a review with data for 92 countries,20 plus additional assumptions for those countries without data (appendix pp 6–7): we assumed 0% intrapartum antibiotic prophylaxis coverage in low-income countries and 80% coverage in high-income countries with no coverage data (appendix pp 6–7). In a secondary analysis, we used a Bayesian evidence synthesis model26 to combine data from studies directly assessing EOGBS incidence and studies estimating the risk of EOGBS among infants of mothers who were GBS colonised. Similar to EOGBS, cases of LOGBS are likely to be under-reported in surveillance systems. Hence, we used an approach similar to a previous analysis21 and estimated region-specific proportions of LOGBS compared with EOGBS, which provided a multiplication factor to apply to EOGBS estimates. Full details are in the appendix (pp 6–9). Since most EOGBS cases (68%) occur within 24 h of birth,21 we assumed a high case fatality rate of 90%, using similar assumptions to Seale and colleagues,7 for EOGBS cases without skilled birth attendance (a marker of low access to any health care, including antibiotic treatment). A dataset from UNICEF and WHO on country-level skilled birth attendance was used.28 For cases of patients with EOGBS with skilled birth attendance and all cases of LOGBS, we applied case fatality rates estimated by region on the basis of published literature (appendix pp 10–12). Assuming a fixed high mortality rate (of 90%) for cases of EOGBS without skilled birth attendance means that uncertainty in this parameter, which is key in the estimation of number of deaths, is not incorporated in the estimates. We also undertook a sensitivity analysis that considered the scenario that EOGBS without skilled birth attendance could have the same regional case fatality rate as EOGBS with skilled birth attendance. iGBS can lead to NDI in those who recover. In this analysis, we estimated the risk of moderate and severe impairments, which is more likely to be consistent across settings and study designs compared with milder NDI types; in the appendix (pp 15–16), we present the results of analyses that also considered mild NDIs. For meningitis, we used studies reviewed in 2017,14 novel data from a national cohort in Denmark (which included children with outcome data available at the age of 10 years),15 and reanalysis of data from a multicentre observational study29 performed in Argentina, India, Mozambique, Kenya, and South Africa including all meningitis cases, irrespective of age (4–17 participants in each centre, with a total of 38 participants); that study was performed as part of this research project. For GBS sepsis, we used data from the same study in Denmark (805 children with data at the age of 10 years) and four smaller studies (4–61 participants in each study, with a total of 109 participants), also reviewed in a systematic review by Kohli-Lynch and colleagues,14 for high-income countries; for low-income and middle-income countries, data from the aforementioned multicentre study were used (22–31 participants in each centre, with a total of 108 participants) in a separate meta-analysis. Note that data from the Argentinian site in the multicentre study were not included because the low proportion of identified patients who recovered from iGBS who were assessed might have been linked to selection bias. Definitions of NDI, including of moderate and severe presentations, are described in detail in these previous studies;14, 15, 30 both the Danish study and the multicentre study, done as part of this project, used multidomain definitions consistent with the Global Burden of Disease, Injuries, and Risk Factors Study.31 Since most of these studies did not include comparator groups, our analysis does not estimate the association between iGBS and NDI, but rather the risk of moderate and severe NDI in patients who recovered from iGBS. Hence, to estimate the number of NDI cases attributable to iGBS, we used the risk of moderate or severe NDI in children without a history of iGBS in the Danish study (2·1%) as a conservative estimate of the comparative risk. Although the Danish cohort is the largest to date, and included information into the second decade of life, it is probable that the counterfactual risk varies considerably within and between countries. Similar to the EOGBS case fatality rate in children with no access to skilled birth attendance, this counterfactual risk was assumed to be fixed; hence estimates do not incorporate uncertainty in this parameter. Another assumption is that the risk of NDI estimated in these analyses represents a lifelong risk, which is probably an underestimation, because most studies recruited children younger than 10 years. Data inputs included a literature review on the proportion of stillbirths with evidence of GBS infection, with six studies published after 2000, one of which also included data collected before that year.9 We updated the searches and identified no additional published studies, but were able to add unpublished data from the Child Health and Mortality Prevention Surveillance (CHAMPS) network32 including seven study sites in Africa and Asia, corresponding to a total of 509 stillbirths. In the CHAMPS network, a panel of local experts evaluated all data available for stillbirths enrolled with a post-mortem investigation including clinical information and collection and testing of tissue and non-tissue specimens to establish whether GBS was involved in the causal chain leading to each stillbirth. We applied a hierarchical model to combine data within regions. For regions with no studies, we used the model to predict the proportion. Region-specific proportions of stillbirths caused by GBS were then applied to country-specific stillbirth risk from the Global Health Observatory data repository. Sensitivity analyses with different prior assumptions were included in the appendix (p 42). GBS also causes disease in pregnant women. The little research on this, all from high-income countries, was reviewed by Hall and colleagues.8 We used data reported in that review, together with those from a study in England,33 to estimate the risk of GBS-related morbidity during pregnancy or post partum (up to 42 days after delivery). Of note, the inclusion of the 2020 study in England was not the result of a systematic search but was suggested by the project’s Scientific Advisory Group. Since these studies, which primarily reported cases where GBS was cultured from blood or cerebrospinal fluid, or both, did not estimate risk given maternal GBS carriage, we applied our estimates directly to country-specific number of births. Using previously reported data,10 we estimated the association between maternal GBS colonisation and preterm births. To incorporate all available evidence,34 we performed a meta-analysis on case-control studies and used the posterior distribution of the coefficient as a prior distribution in a meta-analysis model of cohort and cross-sectional studies. The overall odds ratio was used, together with country-level frequencies of preterm births,35 to calculate the excess number of preterm births associated with maternal GBS colonisation. Two different approaches were used for this calculation, which also required estimated country-level prevalence of maternal GBS colonisation; in the appendix (pp 21–22), we present the results of the meta-analyses and describe these approaches. All analyses were done using the Hamiltonian Monte Carlo algorithm in PyStan (version 2.19), the interface for the Stan libraries in Python (version 3.7).36 Details on the models are presented in the appendix (pp 4–22, 43), including additional assumptions in our analyses and probable limitations. For the primary data collection in high-income, low-income, and middle-income countries, the overarching protocol for the observational study was granted ethics approval at the London School of Hygiene & Tropical Medicine (approval number 16246). Institutional review boards in each of the countries granted ethics approval (Argentina approval number protocol EGB-1; India approval numbers 11723 [Christian Medical College Vellore], 2019–7034 [Indian Council of Medical Research]; Kenya approval number SERU/CGMR-C/164/3882; Mozambique approval number 98/CNBS/2019; and South Africa approval number M190241), as well as the institutional review board of WHO (approval number ERC.0003169). The Danish electronic cohort study was approved by the Danish Data Protection Agency (record number 2015–57–0002). In the Netherlands, the study protocol (EPI-408) was submitted to the Centre for Clinical Expertise at the National Institute for Public Health and the Environment. The study protocol was exempted from further approval by an ethics research committee, according to Dutch law for medical research involving human patients. This study was reviewed by the Centers for Disease Control and Prevention and was conducted consistent with applicable federal law and Centers for Disease Control and Prevention policy. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.