Neurodevelopmental Impairment in Children After Group B Streptococcal Disease Worldwide: Systematic Review and Meta-analyses

Group B Streptococcus (GBS) is a leading cause of invasive disease in infants, causing mortality and neurodevelopmental impairment (NDI) in survivors. This article estimates the percentage of survivors of infant GBS disease with NDI.

in survivors of infant infection. NDI may follow invasive infections including pneumonia, sepsis, and meningitis [7]. The neuroimaging findings, mechanisms of brain injury, and subsequent NDI after invasive infectious disease are outlined in Box 1. Data are insufficient to estimate the risk after invasive neonatal disease other than meningitis, after which around 23% (95% confidence interval [CI], 19%-26%) of survivors are estimated to have moderate to severe NDI [7]. Intrauterine and neonatal insults, including infections, are estimated to cause NDI in 39% (interquartile range [IQR], 20%-55%) of survivors [8].
Group B Streptococcus (GBS; Streptococcus agalactiae) is a leading cause of infant sepsis and meningitis. Its contribution to NDI and associated disability has not, however, previously been assessed by a comprehensive systematic review or pooled estimates. The long-term consequences of infant GBS disease are important to understand, to estimate the full burden of GBS disease, including in terms of standard measures used to assess public health priorities, such as disability-adjusted life-years (DALYs).
We aimed to estimate the percentage of survivors of infant GBS disease with NDI ( Figure 1) as part of a supplement estimating the burden of GBS disease in pregnant women, stillbirths, and infants, which is important in terms of public health policy, particularly vaccine development, as outlined elsewhere in this supplement [17]. The supplement includes systematic reviews and meta-analyses on GBS colonization and adverse outcomes associated with GBS around birth [16,[18][19][20][21][22][23][24]. These are reported individually according to international guidelines [25,26] and are used for estimates of the burden of GBS worldwide [27] ( Figure 1). OBJECTIVES 1. To provide a comprehensive, systematic literature review and meta-analyses to assess the following parameters: (i) percentage of survivors with any NDI after infant GBS disease; (ii) percentage of survivors with moderate to severe NDI after infant GBS disease. 2. To assess these data for input to estimate the burden of GBS in pregnancy, stillbirth, and infants. 3. To evaluate the data gaps and make recommendations to improve the data on NDI after infant GBS disease.

METHODS
This article is part of a study entitled "Systematic estimates of the global burden of GBS worldwide in pregnant women, stillbirths and infants. " It was submitted for ethical approval to the London School of Hygiene & Tropical Medicine (reference number 11966) and approved on 30 November 2016.

Exposure
Case definitions for invasive infant GBS disease include GBS meningitis (clinical signs of possible serious bacterial infection and cerebrospinal fluid (CSF) culture /polymerase chain reaction (PCR)/latex agglutination positive for GBS or blood culture/PCR/latex agglutination positive for GBS with CSF leukocyte count of >20 × 10 6 /L), GBS sepsis (clinical signs of possible serious bacterial infection and blood culture/PCR/latex agglutination positive for GBS), and GBS pneumonia. These definitions are fully described in Supplementary Table 1a.

Outcome
The outcome of interest is NDI. Impairment is defined as a problem in body function and structure such as significant deviation or loss (Supplementary Table 2) [28]. For this analysis of NDI, we combined the specific definitions used by the Global Burden of Disease study 2013 (GBD2013) [29] and others [30] to be consistent with the literature. Therefore, NDIs are categorized as intellectual and/or motor, vision, or hearing impairment and severity classified as mild, moderate, or severe. To maintain consistency with GBD2013 definitions, we did not include social, language, and behavioral NDI in the quantitative analysis for this review. Disability weights for each impairment domain and severity used in GBD2013 are also listed in Supplementary Table 1b.

Box 1. Neuroimaging, Mechanisms of Brain Injury, and Neurodevelop mental Impairment After Infant Meningitis and Sepsis
Meningitis and sepsis can cause brain injury in term and preterm infants. Magnetic resonance imaging (MRI) findings consistently show cerebrovascular involvement, and abnormal findings on neonatal MRI have been clearly associated with poor neurodevelopmental outcome at 2 years [9]. A retrospective cohort study in France found that all 9 cases of infant GBS meningitis in term babies had abnormal findings on MRI, of which 56% showed ischemic infarction (neonatal stroke). Furthermore, a case series of 8 term infants with GBS meningitis and ischemic stroke on MRI found 2 recognizable patterns of injury, deep perforator arterial infarction, and more superficial cortical injury [10,11]. Case reports of neuroimaging in term infants with GBS meningitis have also found severe global cerebral vasculopathy and transverse myelitis [12,13]. In preterm infants, inflammatory cytokines associated with infection increase the permeability of the blood-brain barrier and have an adverse effect on myelin and myelin-producing cells, resulting in periventricular leukomalacia, a condition strongly associated with neurodisability [14,15]. In addition, sepsis causes brain injury indirectly through disseminated intravascular coagulopathy and hypotension; this is important to consider in cases of neonatal sepsis without meningitis [15]. Furthermore, bloodstream infection can have a sensitizing effect in the development of hypoxic-ischemic encephalopathy, which is also associated with neurodevelopmental impairment [16].

Data Searches and Inputs
We identified data through systematic review of the published literature and through development of an investigator group surveying clinicians, researchers, and relevant professional institutions worldwide. For this paper, we did systematic literature searches of PubMed/Medline, Embase, the World Health Organization Library Information System (WHOLIS), Literature in the Health Sciences in Latin America and the Caribbean (LILACS), and Scopus until January 2017. We searched databases with variants of terms related to "child," "disability, " "impairment, " "GBS, " "morbidity," and "mortality. " Medical subject heading (MeSH) terms were used where possible (see Supplementary Table 3 for the full list of search terms). There were no date and/or language restrictions applied and we translated texts to English when published in other languages. We used snowball searches of article reference lists to identify additional studies. We requested unpublished data from an international network of investigators. Where reporting of child neurodevelopmental outcomes was not clear, or outcomes were not reported separately for GBS, we contacted authors for clarification of terms and definitions used, and assessment of survivors of infant GBS disease. If authors were uncontactable, 2 clinicians (M. K. L., N. R.) discussed and came to a consensus to classify impairment. We matched NDI outcomes to GBD2013 and International Classification of Diseases [31] definitions to ensure optimal comparability between studies. The full search strategy is illustrated in Figure 2 [26]. Two independent investigators (M. K. L., N. R.) performed the database searches, screened titles for duplicates and for eligibility, and screened abstracts to assess their suitability for inclusion, and both reviewers extracted data. Where there was discrepancy between the 2 reviewers, a third investigator (A. S.) made the final decision.

Inclusion and Exclusion Criteria
We included studies of infants with GBS disease in the first 90 days after birth (see case definitions), which reported study characteristics and assessed child neurodevelopmental outcomes at a median of ≥6 months of age. Studies with nonrepresentative samples of cases (eg, with a selection bias of preterm births) and unsuitable article types were excluded (Supplementary Table  4). Only cases of moderate or severe NDI were included in the main meta-analysis as there was lack of consistency in definition and proportions of survivors with mild or profound NDI.
The quality of the studies and risks of bias were assessed using inclusion and exclusion criteria (see Supplementary Table 4) on validity of study methods, interpretation of results, and applicability of the results.

Data Collection
Data on study characteristics and results were extracted into prespecified Excel abstraction forms, and then imported to Stata 14 software (StataCorp, La Jolla, California) for meta-analyses.
Meta-analyses were done to assess risk of NDI by severity, length of follow-up, and neonatal mortality rate (NMR) (per 1000 live births) context.
We used random-effects meta-analyses to estimate risk of NDI using the DerSimonian and Laird method [32]. The primary meta-analysis was of GBS meningitis survivors followed up for a median of ≥18 months with moderate to severe NDI. We decided to focus on moderate to severe impairment as more consistent results are seen with moderate to severe NDI [33] and there was lack of consistency in the definition and proportions of survivors with mild NDI. We chose a median follow-up of ≥18 months to include as many studies as possible while demonstrating the most accurate possible estimate of NDI, and cases of NDI are likely to manifest with longer-term follow-up. A threshold NMR of 5 per 1000 live births was used for the sensitivity analyses to reflect newborn care context as high-income countries have an average NMR of 4 per 1000 [34]. Given lack of variation with varying NMR settings (see below), we did not report the primary meta-analysis by NMR setting.
The following sensitivity meta-analyses were done to assess the effect of length of follow-up, severity of NDI, and NMR setting: (1) GBS meningitis survivors followed up for a median of ≥18 months with any NDI; (2) GBS meningitis survivors followed up for a median of ≥18 months with moderate to severe NDI where NMR was ≥5 per 1000 live births; (3) GBS meningitis survivors followed up for a median of ≥18 months with moderate to severe NDI where NMR was <5 per 1000; (4) GBS meningitis survivors followed up for a median of ≥6 months with any NDI; (5) GBS meningitis survivors followed up for a median of ≥6 months with moderate to severe NDI; (6) GBS meningitis survivors followed up for a median of ≥6 months with moderate to severe NDI where NMR was ≥5 per 1000; (7) GBS meningitis survivors followed up for a median of ≥6 months with moderate to severe NDI where NMR was <5 per 1000.

Study Characteristics
All 18 studies followed up survivors of infant GBS meningitis; 5 of these studies also followed up survivors of infant GBS sepsis ( Table 1). The numbers of GBS sepsis cases and survivors assessed were too few to pool in a meta-analysis (see Supplementary  Table 5 for summary of results); therefore, we have meta-analyzed NDI outcomes after infant GBS meningitis only.
Ten of the 18 articles included were published between 1973 and 1992. The remaining 8 studies were from 2000 onward. One article [38] contained datasets from 2 periods (1976 and 1983) and were entered into the meta-analyses separately. Years of data collection ranged from 1970 to 2017, with the median years of data collection ranging from 1971 to 2016. Most (8/18) studies were done in the United States, 2 in the United Kingdom, 2 in Sweden, and 1 each in Germany, Denmark, Singapore, China, Tunisia, and South Africa, the latter 3 being the only studies conducted in middle-income contexts (Figure 3). The majority (15/18) of studies were carried out in hospitals, while 3 of 18 studies used national surveillance data. The median length of follow-up of infant GBS meningitis survivors ranged from 6 months to 10.5 years. Neonatal mortality rates at studies' median point of data collection ranged from 1-15 per 1000 live births [51][52][53][54]. Further detailed characteristics of studies included after systematic review for NDI following infant GBS meningitis are given in Table 2.
A wide range of neurodevelopment assessment methods were applied. In total, 44 different methods were used (Supplementary Table 6); 31 methods were used to assess child cognitive, motor, language, and socioemotional or behavioral development; 8 methods to assess vision; and 9 methods to assess hearing. Not all studies covered all domains or described the methods (Supplementary Table 6). The number of child development assessments done for each child ranged from 1 to 19 (Supplementary Table 6). The primary meta-analysis was for risk of moderate to severe NDI in survivors of infant GBS meningitis at a median ≥18 months of follow-up.
When followed up over a median of 18 months, 18% (95% CI, 13%-22%) (Figure 4) of GBS meningitis survivors had moderate to severe NDI. There were insufficient data for estimates by United Nations subregions, other than for developed countries, for which the estimate was also 18% (95% CI, 13%-23%), reflecting the fact that this was where most data were from.  Sensitivity meta-analyses were for risk of NDI in survivors of infant GBS meningitis by severity, length of follow-up, and NMR (per 1000 live births) context.
A series of 7 meta-analyses (meta-analyses 1-7) acted as sensitivity analyses to assess the variation of risk of NDI with as long follow-up as possible in different NMR settings. Supplementary  Figure 1 demonstrates the number of studies included in each meta-analysis.
1. GBS meningitis survivors followed up for median of ≥18 months with any NDI: 32% (95% CI, 25%-38%) (Supplementary Figure 2) Figure 4), the percentage of GBS meningitis survivors with moderate to severe impairment where NMR is <5 per 1000 was very similar to results in settings where NMR is ≥5 per 1000 live births. 4. GBS meningitis survivors followed up for median of ≥6 months with any NDI: When followed up for a median of at least 6 months, 27% (95% CI, 20%-34%) GBS meningitis survivors were reported to have any NDI (Supplementary Figure 5). 5. GBS meningitis survivors followed up for median of ≥6 months with moderate to severe NDI: 15% (95% CI, 11%-20%) of GBS meningitis survivors were reported to have moderate to severe NDI at a median of at least 6 months of follow-up (Supplementary Figure 6).
6. GBS meningitis survivors followed up for median of ≥6 months with moderate to severe NDI where NMR is ≥5 per 1000: The percentage of GBS meningitis survivors with moderate to severe NDI was 15% (95% CI, 10%-20%) (Supplementary Figure 7) when followed up for a median of at least 6 months where NMR is ≥5 per 1000 live births. 7. GBS meningitis survivors followed up for median of ≥6 months with moderate to severe NDI where NMR is <5 per 1000: A slightly greater percentage of GBS meningitis survivors followed up for a median of at least 6 months in settings where NMR is <5 per 1000 had moderate to severe NDI, 18% (95% CI, 8%-28%) (Supplementary Figure 8), compared to settings where NMR is ≥5 per 1000. Table 2 summarizes the results of the above meta-analyses, with more details in Supplementary Figures 2-8.

DISCUSSION
GBS is an important contributor to NDI after infant meningitis, of which it is a leading cause. However, data are insufficient to assess its contribution to NDI after all infant GBS disease, which is important as there are many more cases of infant GBS sepsis compared to infant meningitis [27]. Our estimates of moderate to severe NDI following GBS meningitis in 18% (95% CI, 13%-22%) of survivors is consistent with the estimate of NDI after meningitis of all infectious etiologies, which is 23% (95% CI, 19%-26%) [7]. The slightly lower point estimate may be due to differences in the geographies covered. In our analysis, there were no data from low-income contexts, and data from only 3 middle-income contexts (China, Tunisia, and South Africa) and a higher proportion of data from developed countries compared to the previous work, which limits the generalizability of the NMR sensitivity analyses. Improving the data are critical to direct public health interventions. Currently, and mainly in high-income contexts, early-onset cases (the first week after birth) of invasive GBS disease are reduced through intrapartum antibiotic chemoprophylaxis, but maternal vaccination may offer an alternative strategy in the future, with the potential to reduce the morbidity, as well as the mortality burden from GBS disease, and be more feasible in settings where a large proportion of deliveries occur at home. There were insufficient data to determine whether NDI in GBS meningitis survivors varies between different mortality contexts. For settings with an NMR <5 per 1000 live births and ≥5 per 1000 live births, the risk of moderate to severe NDI was 18% (8%-28%) and 18% (12%-24%), respectively, at median follow-up of at least 18 months. Whether there is a difference in very high-mortality settings (NMR >15/1000 live births) is unknown. However, as >90% of the world's births are in low-or middle-income contexts, the number of cases are likely to be highest in these contexts [2].
Overall it is likely that we are underestimating NDI after invasive infant GBS disease, due to challenges in detection, particularly at the youngest ages. As expected, the prevalence of NDI increased with follow-up to 18 months compared to 6 months (18% [95% CI, 13%-22%] vs 15% [95% CI, 11%-20%]). While we acknowledge that these confidence intervals overlap considerably, the upward trend suggests that longer follow-up is needed for accurate case ascertainment. The range of assessment tools used for NDI diagnosis limits comparability, and demonstrates the need for standardization of methodologies, as well as validation of tools in different contexts. There were 44 different methods used in the studies in this review, which results in heterogeneity and lack of true comparability. This review is further limited by use of developmental screening tests by some studies, rather than diagnostic developmental assessments. While there is no universal gold standard for high-risk newborn follow-up, studies should ensure comprehensive neurodevelopment assessment of all domains, and of vision and hearing, and should align methods of assessment with contemporary studies. Neurodevelopment assessment tools, where not developed locally, should be translated, culturally adapted, and validated in the setting. NDI in GBS-associated preterm and neonatal encephalopathy survivors is also important to consider in estimating the total burden of GBS disease but was not investigated in this review.
To improve the data, ideally, we should systematically investigate infants with signs of possible serious bacterial infection to ascertain a bacterial infectious etiology, and follow-up for longer periods of time to determine NDI outcomes, using standardized timing of neurodevelopmental assessment with tools validated in the context. To investigate the long-term impact of infant infection, particularly in terms of NDI, longer-term cohort studies are required ( Figure 5). To avoid recruitment bias, study participants should be well described, including data on, for example, comorbidities, and gestational age and birthweight, which (if low) can contribute to NDI. Furthermore, while we have focused on physical NDI, a comprehensive assessment of all child development outcomes, including socioemotional and behavioral outcomes, would also improve understanding of the total burden of NDI after infant GBS disease [55,56].

CONCLUSIONS
GBS is a leading cause of infant meningitis, and almost one-fifth of GBS meningitis survivors experience moderate or severe NDI. There is an additional, as yet unquantified burden associated with other invasive infant disease, such as GBS sepsis. It is critical to look toward improving the health and well-being of survivors of infant GBS disease, and supporting their families, for whom there are financial, social, psychological, and emotional impacts. Prevention strategies (intrapartum antibiotic chemoprophylaxis) for early-onset invasive infant GBS disease are currently limited to developed countries, and only around the time of birth. Maternal GBS vaccination may be able to reduce the burden of GBS disease further, particularly GBS meningitis, which mainly presents as late-onset disease beyond 7 days of life (Table 3). . has served as a member of the Presidential Advisory Committee for Seqirus Inc and of the CureVac Inc Scientific Advisory Committee, as well as undertaken consultancy work for Pfizer Inc. C. C. has received institutional compensation from Novartis for conducting GBS studies. P. T. H. has been a consultant to Novartis and Pfizer on GBS vaccines but received no funding for these activities. M. I. has undertaken sponsored research from Pfizer on pneumococcal disease in adults and from Belpharma Eumedica (Belgium) on temocillin antimicrobial susceptibility in Enterobacteriaceae. K. L. D. has received funding by the Bill & Melinda Gates Foundation to work on research on GBS serocorrelates of protection to inform vaccine trials, and travel expenses from Pfizer to attend a meeting on an investigator-led project on GBS. S. A. M. has collaborated on GBS grants funded by GlaxoSmithKline and by Pfizer and received personal fees for being member of its advisory committee; he has also collaborated on a GBS grant funded by Minervax. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Table 3. Key Findings and Implications
What's new about this? • The first systematic review of NDI findings after GBS and, despite 3 decades of focus on GBS disease in infants, data were insufficient to assess NDI outcomes after GBS sepsis. Only 18 studies met inclusion criteria for follow-up after GBS meningitis.
What was the main finding?
• Eighteen percent (95% CI, 13%-22%) of survivors of infant GBS meningitis have moderate or severe NDI at a median follow-up time of >18 months.
How can the data be improved?
• Cohort studies with adequate follow-up (>18 months) using standardized assessment tools, validated locally, and times of clinical review, particularly in resource-poor settings.
What does it mean for policy and programs?
• Lack of data limits our assessment of the burden of NDI after invasive infant GBS disease, which may be a considerable burden and increases with improvements in survival.