Quantifying long-term health and economic outcomes for survivors of group B Streptococcus invasive disease in infancy: protocol of a multi-country study in Argentina, India, Kenya, Mozambique and South Africa

Research objectives

Our objectives in designing this study were:

Objective 1 - Long-term neurodevelopmental impairment:

Objective 2 – Long-term mortality: To assess mortality beyond initial hospital-discharge among children who had iGBS.

Objective 3 – Long-term economic costs and health-related quality of life:

Objective 4 – Short-term economic consequences: To estimate the costs to the healthcare system and households during acute episodes of sepsis and meningitis (irrespective of etiology) in neonates and young infants.

Study design

Long-term outcomes after iGBS disease in infancy (Objectives 1, 2, 3). We will use a matched cohort study design to collect data on NDI, socioemotional behavioral, and economic outcomes for survivors of iGBS in early infancy. Children with a history of infant iGBS (henceforth iGBS survivors), will be identified via hospital records in study sites, Health and Demographic Surveillance Systems (HDSS), or among participants of previous epidemiological studies. Children with no history of iGBS (henceforth, the non-iGBS comparison group) will be identified and matched to iGBS survivors based on sex and birth month and year.

Acute costs of neonatal sepsis and meningitis study (Objective 4). In addition to the main study measuring long-term economic outcomes, we will undertake a separate study to quantify the acute costs to the healthcare system and household linked to neonatal sepsis and meningitis. This study will involve a different study population: prospectively identified neonates admitted for clinically suspected sepsis or meningitis irrespective of the underlying etiology. Data on the costs associated with the period of acute hospitalization will be collected following discharge.

Study settings and teams

Since the major data gap on the long-term outcomes of iGBS survivors is in LMICs, this study was designed to collect data in these settings, including at least one country per GBS high-burden region (Africa, Asia, Latin America). For this collaborative work, we shared a call for data through multiple channels in 2018, including targeting previous collaborators, experts and known GBS researchers, scientific conferences and meetings, as well as sending direct requests from WHO headquarters to country offices and placing posts on social media platforms to reach the widest number of people. With these various approaches we aimed to ensure better geographical representation than currently seen in the literature. Among those who responded, potential study sites were identified based on the following criteria: (a) sites that had at least 10 post-discharge surviving iGBS cases that could be enrolled; (b) sites that had neurodevelopmental follow-up data or the ability to collect this type of data in children aged at least 3 years; (c) sites where the expected loss to follow-up was <20%. From those who expressed an interest in joining this project and fulfilled the above criteria, research teams from Argentina, India, Kenya, Mozambique and South Africa agreed participate in this work and lead investigations locally (Table 1).

In Argentina, the local study will be performed by the research organization Fundación Infant, in Buenos Aires; in Southeast India, the research activities are led by the Christian Medical College in Vellore; in Kilifi, Kenya, the work is being undertaken by KEMRI-Wellcome Trust Research Programme; in Mozambique, children are being recruited at the Manhiça Health Research Centre, in collaboration with the Barcelona Institute for Global Health; and in South Africa, the project is being led by the South African Medical Research Council Vaccines and Infectious Diseases Analytical Research Unit (VIDA).

Long-term health outcomes after GBS invasive disease in infancy (Objectives 1, 2 & 3)

Study populations. In the Kenya and Mozambique sites, which are also HDSS sites, parents of potential study participants, based on the HDSS database, will be contacted through standard recruitment procedures. Hospital records, which are linked to the HDSS database, will be used to identify all children who have been admitted with iGBS based on the case definition. The HDSS database will also be used to select matched non-iGBS children from the community.

In Argentina and India, hospital-based databases will be used to identify potential iGBS survivors and non-iGBS children using standard practices established by each site. Hospital-based databases will be used to identify children who have been admitted with iGBS based on the case definition. Hospital-based birth registries will be used to select matched non-iGBS children.

In South Africa, the same cohort of iGBS survivors and controls from three epidemiological studies that were conducted between 2012 and 2015 will be contacted for re-enrolment. These participants are expected to be 5–7 years old and originally consented to be followed until the age of 5. Using the study database, parents or primary caregivers of these children will be contacted by phone for interest and be given information about participating in the new study.

Invasive GBS disease case definition. Children with a previous diagnosis of either GBS meningitis or GBS sepsis in the first 90 days of life (days 0 - 89) will be recruited in these local epidemiological studies. Enrolment of children with a history of GBS sepsis is important to increase the, currently limited, number of studies with data on long-term disability post-GBS sepsis⁷. Table 2 below summarizes the case definition and the clinical and microbiological eligibility criteria used by each study site for identification of exposed children in this study.

Sample size and power calculation. The number of iGBS survivors included in the matched cohort study of long-term outcomes was based on the maximum number of cases expected to be identified, accounting for 20% being unreachable, ineligible or who refuse participation. The expected number of iGBS survivors for each site is summarized in Table 3. Based on anticipated recruitment of 200 iGBS survivors and a 1:3 ratio of matched non-iGBS children, and assuming a prevalence of our primary outcome (moderate/severe NDI) of 26% in iGBS survivors and 10% in the non-iGBS comparison group (based on a study of meningococcal serogroup B survivors¹⁶), a pooled analysis would be able to detect this difference using a two-sided test of binomial proportions with 99% power at a 5% significance level. Mild developmental impairment is likely to be more prevalent among iGBS survivors, including those with sepsis. Our power to detect a difference in the risk of overall NDI (including mild NDI) would be 78%, assuming detection of 32% and 20% NDI in iGBS survivors⁷ and non-iGBS children¹⁷, respectively.

Study procedures and data collection. Trained fieldworkers will contact the parents/primary caregivers of these potential participants about the study by phone (Argentina, India, South Africa) or in-person (India if phone contact information is not available, Kenya, Mozambique) and those contacted will be asked to make a one-time visit to the health facility with their child. Reasons for non-participation, such as migration, refusal or death will be recorded.

Children enrolled in the study and their main caregiver will receive an in-person assessment visit. Written informed consent will be obtained in-person either at the time of the initial house visit or before the in-person assessment visit. Only if appropriate consent/assent is obtained, will the child be enrolled in the study.

At the in-person assessment visits, the following information will be collected:

Data will be collected on paper forms or using a customized app (developed in collaboration with icddr,b, Bangladesh). The customized Android tablet-based app includes questionnaires and neurodevelopment assessment tools, translated into local language where relevant (Figure 1).

Figure 1. Screenshots from the customized data capture app.

Assessment of neurodevelopmental impairment (Objective 1). In this multi-country study, we will use several tools to diagnose NDI and socioemotional and behavioral outcomes in children over a wide age range from 18 months to 17 years old. The inclusion of children older than those enrolled in the majority of the previous epidemiological studies enables us to use more complex developmental assessments designed for older ages to detect specific mild NDI and other developmental delays. By including a wider age range, we can also better understand the developmental trajectory of infants who have survived iGBS. The developmental domains of interest are motor, vision, hearing, cognitive, language, and socioemotional; their definitions, by severity, are described in the Data analysis section. We will also explore growth outcomes and epilepsy.

The diagnostic tools used to identify NDI and other developmental measurements will be administered by experienced assessors, clinical psychologists and pediatricians, who will also perform clinical exams to identify impairment in hearing, motor and vision domains. Each local research team decided which neurodevelopmental assessment tools are appropriate for their setting, based on child’s age and cultural appropriateness or validation of the instrument and technical capacity of each site. There will be 26 different assessment tools and tests being used across the 6 neurodevelopmental domains, as well as anthropometric measurements for growth and an epilepsy screening questionnaire. Table 4 shows the matrix of assessments for each developmental domain, by age category and study site. When a need for further assessment and clinical management is identified, children will be appropriately referred into each site’s existing referral systems.

Mortality outcome (Objective 2). Whenever feasible, data on the cause of death will be captured through a variety of methods, including by reviewing medical records, verbal autopsy reports and interviews with parents, for iGBS survivors who died after the acute episode and before enrolment and for matched non-iGBS group who died. In Kenya, the list of iGBS survivors who could potentially be enrolled in the study only included children alive at the time of enrolment, therefore data on early mortality post-iGBS disease will not be collected in these sites.

Long-term economic costs and health-related quality of life outcomes (Objective 3). There are only limited data available on the economic consequences of iGBS, which cover only healthcare costs in the first two years of life¹⁴. In this study, we will collect information on variables that will allow comparisons of economic outcomes in families of iGBS survivors versus families of the non-iGBS comparison group. These data will also be used to inform future economic analyses that will be performed to assess the value of maternal vaccines against iGBS.

A summary of key economic variables is shown in Table 5. Information collected will include details of the monthly household income and expenditure, participating children’s healthcare utilization, out-of-pocket payments, and any expenditure on social care or special education in the 12 months preceding study enrolment. Additionally, information will be collected on time spent by the main caregiver providing informal care to the participant, as well as information on the costs of coping strategies, such as borrowing and asset sales. Information on the HRQoL of both the participant and the main caregiver will be collected using an EQ-5D-3L questionnaire in three countries where country-approved translations are available: Argentina (Spanish), India (English, Telugu, Tamil) and South Africa (English, Zulu).

Acute costs of neonatal sepsis and meningitis (Objective 4)

Case definition for neonatal sepsis and meningitis. To be able to collect data on acute costs, both cases of severe neonatal infection linked to GBS and cases of severe neonatal infection due to other bacteria will be enrolled, as the number of confirmed iGBS cases per hospital is anticipated to be small over the duration of our study. Participants will be babies admitted with a diagnosis of clinically suspected neonatal infection (sepsis or meningitis) combined with isolation of a pathogenic microbiological agent by culture or detection by polymerase chain reaction (PCR) in a normally sterile site (blood/CSF) on day 0 – 89 of an infant’s life. Babies born at <32 weeks of gestational age, born with severe congenital abnormalities, or with culture positive results only for organisms considered to be contaminants or skin commensals will be excluded (Table 6). In Mozambique the cases will be defined based on clinically suspected sepsis or meningitis because the number of bacteriologically confirmed cases is anticipated to be low due to the size of the hospital.

Sample size. At least 20 participants will be recruited in each of the four sites. The sample size of 20 per site was set as a practical minimum, considering both available resources and also a consensus by local research teams that this would be a feasible number to capture given the anticipated number of neonatal infections within the timeframe of data collection activities.

Study procedures. Participants will be identified prospectively either on admission or using clinical databases. Additionally, in India recent cases (within three months before the start of the study) will also be identified retrospectively from clinical records. Details on hospital resource use will be collected from medical records, including information on length-of-stay, type of hospital bed, and details of any drugs, diagnostic tests and surgical procedures. To capture the wider impacts of a participant's hospitalization, a questionnaire will be administered to the main caregiver, either at time of discharge, or by follow-up as soon as possible after discharge, to collect details on household demographics and economic impact. This will include any out-of-pocket payments, costs related to travel, accommodation, and caregiver time. Data will be gathered on the main caregiver’s HRQoL using an EQ-5D-3L questionnaire and the main caregiver will be asked to estimate their child’s HRQoL during their time in hospital using a Visual Analog Scale (VAS). The main caregiver will be encouraged to accompany the child to the assessment visit, but in the cases where they do not then these sections of the questionnaire will not be completed. No questionnaires will be administered in the case that a participating child dies in hospital, but data will still be collected from hospital records.

Data management

Data will be stored on secure servers locally after the end of the study, and anonymized data will be transferred to the team at the London School of Hygiene & Tropical Medicine, where data from different countries will be pooled. Analyses will be conducted jointly by all study partners.

Analysis plan

Objective 1 - Long-term neurodevelopmental impairment. To allow comparison between the different neurodevelopmental assessments being used in each of the five sites, we will undertake a mapping activity across all 26 tools by age bands (1–4, 5–9, 10+). The age bands are constructed based on key periods of development¹⁵. We will map similar constructs across the different assessments; e.g., gross motor measurements from all relevant tools will be mapped against each other allowing us to compare gross motor development across sites. We will do this for the following domains; gross motor, fine motor, cognitive and language. We will also do a similar mapping activity between the preschool and school-aged CBCL for the socioemotional and behavioral outcomes (e.g., anxiety, ADHD, and autism). Definitions domain-specific neurodevelopmental impairment and severity are described in Table 7.

In India, Kenya, Mozambique, and South Africa, motor and cognitive scores will be normalized using standard reference populations by assessment and site. In Argentina, where motor impairment is being assessed through a clinical exam, description of functional impact will be used.

Vision impairment will be defined using WHO categories of mild (visual acuity in best eye ≤6/12), moderate (visual acuity in best eye ≤6/18 and >6/60), severe (visual acuity in best eye ≤6/60 and >3/60), and blindness (visual acuity in best eye ≤3/60)^18,19.

Any hearing impairment will be defined as an unaided hearing threshold in the best ear of >26 decibels and further categorized into mild (audiometric hearing threshold level 26–30 decibels), moderate (threshold level 31–60 decibel), and severe/deafness (threshold level >60 decibel)^18,20. In South Africa and Mozambique, screening tests will be used first to identify any individual with any potential hearing impairment. Results from further diagnostic tests will be used to classify into impairment severity as categorized above.

Socioemotional behavior measures will be defined in all sites using the CBCL assessment. The main scoring is based on a principal components analysis that grouped sets of behaviors into different syndrome scales: (1) internalizing problem scales, which include anxious/depressed, withdrawn-depressed, and somatic complaints scores; and (2) externalizing problem scales, which includes rule-breaking and aggressive behavior. There is also a total problem score which is the sum of all the items. Each syndrome, internalizing and externalizing problem score, and total score can be categorized into normal (<93rd percentile), borderline (93rd-97th percentile), or clinical behavior (>97th percentile) based on the same normative samples to create standard scores based on sex and age for all sites.

Primary outcomes

There are 2 primary outcomes in this study, moderate/severe NDI and moderate/severe behavioral outcomes.

Moderate/severe NDI will be defined as:

Moderate/severe behavioral outcomes will be defined by scores within clinical ranges of ≥1 domain(s) of the problem scales.

Additional outcomes

Mild NDI (including socioemotional behavior outcomes) will be defined as:

We will further assign individuals into the following multi-domain impairment categories based on severity (adapted from²¹):

The distribution of mild and moderate/severe neurodevelopmental outcomes will be summarized for children with history of iGBS disease and the non-iGBS comparison group and further stratified by clinical syndrome (sepsis and meningitis). We will test the association between history of iGBS disease in early-infancy and moderate/severe NDI in a pooled analysis using a logistic regression accounting for matching factors of age and sex. As gestational age is likely to be an important confounder, we will adjust for this, if the data allows.

Objective 2 - Mortality. For iGBS survivors, and their matched non-iGBS comparison group, that were reachable (i.e. for whom we have information), we will describe the proportion of children who died before enrolment in each site. Where available, we will also describe the causes of death in both iGBS and non-iGBS groups.

Objective 3 - Long-term economic and health-related quality of life. We will assess the impact of iGBS on economic outcomes including healthcare utilization and costs, household out-of-pocket payments, household income and social care payments, and time spent by the main caregiver on informal care. The cost of hospital stays and attending outpatient clinics will be estimated using published unit costs (e.g. WHO-CHOICE)²². We will compare these outcomes between the iGBS and non-iGBS groups in each study site. If data allow, we will also analyze differences in costs and healthcare utilization linked to NDI. Information from EQ-5D-3L questionnaires will be used to estimate differences in QALYs of both children and their caregivers associated with a history of iGBS disease.

Objective 4 – Acute costs of neonatal sepsis and meningitis. Data from the sub-study on acute costs will be used to estimate the average length-of-stay, use of supportive care, drugs and diagnostics, during hospitalization for the acute neonatal sepsis/meningitis episode. These data will also be used to calculate the overall cost per episode. Other variables will be presented descriptively to characterize the impact of severe neonatal infection.

Ethics

Written informed consent will be obtained from parents or guardians. Whenever appropriate, based on local guidelines, assent will also be obtained from children participating in the study. The overarching protocol for this multi-country observational study was granted ethical approval at the London School of Hygiene & Tropical Medicine (approval number 16246). Institutional review boards in each of the operating countries granted ethics approval (Argentina approval number Protocol EGB-1, India approval numbers 11723 (CMC Vellore), 2019–7034 (ICMR); Kenya approval number SERU/CGMR-C/164/3882; Mozambique approval numbers 98/CNBS/2019; South Africa approval number M190241), as well as the institutional review board of the World Health Organization (approval number ERC.0003169).