The Sanitation Hygiene Infant Nutrition Efficacy (SHINE) Trial: Protocol for School-Age Follow-up

Community sensitisation and pre-screening

At the start of the follow-up study, sensitisation of the local leadership and community members was undertaken by the study team, together with the screening of a community-made film about the SHINE trial which explored their previous involvement and experiences of research. CHWs performed role-plays to explain the purpose, consent procedures and practicalities of the visit to participating families and the community. Once the family has received information and expressed an interest in joining, a mutually convenient date is booked for an assessment.

Screening, consent, assent and assessments

Assessments are conducted by data collectors (DCs), who are primary care nurses extensively trained in study procedures. On the day of the assessment, the CHW introduces a pair of DCs (DC1 and DC2) to the household. Screening is undertaken to ensure the original SHINE child and primary caregiver are present; once both DCs verify that this is the correct child, 1-2 tents are pitched in or close to the household (Figure 3). Households are usually single-family dwellings surrounded by subsistence farmland and the mean distance between houses in SHINE was 82.6m⁸, ensuring the tents are sufficiently isolated for confidentiality during informed consent and data collection. Written informed consent is obtained from the primary caregiver, and assent from the child, following a demonstration of the tools used to conduct the measurements. The primary caregiver is usually the mother, but it is the child born within the SHINE trial who is the focus of this follow-up study. Hence if the mother is not available (for example the mother has died, divorced or moved away) the primary caregiver is defined as the adult most responsible for the child’s care, and completes the caregiver questionnaire.

DC1 performs the caregiver questionnaires in a pre-specified order, while DC2 conducts the cognitive, physical and anthropometry assessments in a separate tent to minimize distractions, after building rapport with the child. The child is fed before the start of the tests (either by the family or the DC’s who provide additional snacks). The child is then given further regular breaks and snacks to reduce tiredness. All tests have standardized explanations and demonstrations to ensure appropriate understanding, and are translated into Shona or Ndebele – the local languages spoken in the district. The DCs remain blinded to the original trial allocation and the maternal HIV status throughout the visit. At the end of the visit, DC1 opens a sealed envelope to check the maternal HIV status in the original trial, which then determines whether questions are asked on HIV treatment or if HIV testing is appropriate. At the end of the visit, a small gift is given to the child and caregiver to thank them for their time.

HIV testing and referrals

All primary caregivers are offered HIV testing at the end of the visit, unless they are already known to be living with HIV or had a documented negative result in the previous 3 months. If the mother is HIV-negative, the child is not tested. If the mother is HIV-positive, declines testing or is not available, the child is offered HIV testing with age-appropriate assent using role-plays. The Determine HIV-1/2 rapid test (Abbott) is used for initial testing; positive results are repeated using the HIV 1/2 Stat-Pak rapid test (Chembio). The option of dried blood spot testing is also available for participants who prefer not to be tested in real-time in the homestead. Referrals are made to local clinics for mothers or children testing HIV-positive.

Support and supervision

Ongoing support and monitoring are provided by a senior research nurse and a doctor, with 6-monthly standardisation exercises. Additional training is provided for any DCs with coefficients of reliability (‘R’) below a set threshold in anthropometry or cognitive measurement, derived from the intra- and inter-observer technical error of measurement for continuous measurements (e.g. anthropometry)¹³ and intraclass correlation coefficient (ICC) for discrete measurements (e.g. cognitive test scores).

Measurement of illness and healthcare-seeking behaviour

The incidence of common infections is measured by prospective surveillance. The CHW visits families 4-weekly to conduct a 7-day illness recall questionnaire to capture episodes of fever, cough, diarrhoea, skin, ear or mouth infections or other illnesses, and associated health-seeking behaviours.

Effect of IYCF and WASH on outcomes (Objectives 1 & 2)

We will capitalize on the 2×2 factorial trial design to evaluate the IYCF and WASH interventions as two trials run in the same population, stratified by maternal HIV status. For the analysis of IYCF as the primary outcome, we will therefore combine the two IYCF-containing trial arms (IYCF alone, and IYCF+WASH) and compare them to the two non-IYCF arms (WASH alone, and standard-of-care). For the analysis of WASH as a secondary outcome, we will combine the two WASH-containing trial arms (WASH alone, and IYCF+WASH) and compare them to the two non-WASH arms (IYCF alone, and standard-of-care). The effect of IYCF and WASH on each outcome (as defined in Tables 1 to 3) will be evaluated using generalized estimating equations (GEE) with an exchangeable working correlation structure to account for within-cluster correlation, assuming no interactions between randomised interventions. All analyses will account for the nature of the distribution of the outcome and results will be presented as appropriate effects sizes (mean difference between groups and risk ratios) with a measure of precision (95% confidence intervals). Analyses will be by intention-to-treat at the child level, according to the mother’s assigned study arm based on her residence at the time of her enrollment into SHINE, regardless of subsequent moving or adherence to the interventions. The hypothesis test will be 2-sided and a significance level of 5% will be used.

The primary analysis will be unadjusted. A secondary adjusted analysis will account for original trial stratification factors and other possible confounders as defined in the SAP. Since we are performing a large number of tests, we will base our conclusions on associations observed in outcome groupings, and not on p-values. The generalisability of the sample will be assessed in comparison to the original SHINE cohort. If evidence of a large difference is seen, then weights will be applied as a sensitivity analysis to explore inference effect.

Development of novel outcome metrics (Objective 5)

Novel outcome metrics will be developed based on SAHARAN toolbox measurements conducted in 250 HUU children in the standard-of-care arm. We will use these data to undertake a factor analysis of the primary and secondary cognitive SAHARAN variables to select the most informative sparse combination of tests for development of a cognition metric (COG-SAHARAN). We will undertake reliability and agreement analyses to compare the performance of the shortened, open-access tools with the KABC-II tool, which is a gold-standard assessment, but is costly, time-consuming, complex and proprietary. A similar factor and reliability analysis will inform the generation of the growth metric (GROW-SAHARAN) from all growth and physical function outcomes. GROW-SAHARAN will measure the child’s growth, body composition and physical function to holistically assess school-age children’s nutritional status. Further analyses could tailor the GROW-SAHARAN to focus on key indicators within specific areas such as physical function, body composition or chronic disease risk. This is a sufficient sample size to derive the COG-SAHARAN and GROW-SAHARAN metrics, based on previous published metrics^18,19. As outlined above, we will employ factor analysis as a data reduction step to i) identify the underlying constructs that we are measuring, ii) evaluate which of the multiple sub-tests are driving the variability measured, and iii) determine whether all tests are required.

Standardisation exercises with the same children will allow us to test the reliability of the proposed measures using intra-cluster correlation coefficients and to measure Cronbach’s alpha. Finally, the relationship between these novel metrics will be investigated within a recognised conceptual framework that evaluates child function²⁰ (Figure 4). A combined child metric (“SUB-SAHARAN”) will also be developed after derivation of COG- and GROW-SAHARAN, by exploring the relationships between the two metrics. All metrics will then be operationalised in the 1000 CHU randomized to early-life IYCF or WASH interventions, and the performance of each metric compared to the more detailed analysis outlined in previous sections.

• Download figure
• Open in new tab

Figure 4:

Conceptual framework of objectives with exposure and outcome variables. Exposures are split into environmental, schooling, maternal and nurturing domains. Early-life exposures are described in black and contemporary exposures in blue text. Those exposures that were measured in both early-life and contemporary are written in black and blue text. Outcomes are based on the SAHARAN toolbox to provide school-age child growth, health and function. These outcomes will also be analysed in the standard of care arm to provide new cognitive (COG-SAHARAN), Growth (GRO-SAHARAN) and overall (SUB-SAHARAN) metrics.

The ‘missing middle’ of school-age child health and function

The SHINE follow-up study aims to characterize school-age health and function and provide important insights into the ‘missing middle’ of childhood. It is known that between 5-14 years, mortality is concentrated in sub-Saharan Africa²¹, but measures of child growth and development have frequently been overlooked²². Nevertheless, this life-stage remains crucial for growth to consolidate positive gains from early childhood. Height and BMI monitoring shows that children can experience negative trajectories of increasing BMI in response to adverse environmental conditions²³, increasing future non-communicable disease risk. Examining growth and development together is urgently needed to understand the longer-term impact of early-life exposures and interventions in relation to school-age growth and function⁶. Few studies have confirmed whether early growth gains translate into long-term improvements in cognitive and physical function and, conversely, whether interventions showing little or no obvious effect at early ages may confer meaningful health and functional benefits in later childhood. It is also important to understand the contemporary risk and protective factors that undermine or accelerate progress, respectively (Figure 2). School-age presents an opportunity to mitigate early disadvantages and consolidate early gains, thus providing sustainable benefits in development and longer-term health.

Strengths and Limitations of this study

Follow-up of the SHINE cohort at 7 years of age will determine whether nutrition interventions in the first 1000 days that modestly increase linear growth can restore long-term function, and whether WASH interventions in the first 1000 days have any impact on school-age outcomes, despite no effects on early-life growth. This will be achieved using the SAHARAN toolbox to measure growth, neurodevelopmental and physical health outcomes of school-age children, which has previously been extensively piloted. The study efficiently capitalizes on the factorial trial design to evaluate both interventions in the same children. The feasibility of a similar detailed battery in children has previously been demonstrated before in a cohort study of survivors of severe acute malnutrition²⁴. Limitations include the selection of only one of the two districts of the SHINE trial, as well as the inability to follow-up children who have moved out of Shurugwi district. The selection of HIV-unexposed children is randomised to ensure balance between the SHINE intervention arms. However, we will try to follow all HIV-exposed children, which may lead to an imbalance in treatment arms for CHEU. HIV-positive children are excluded from the follow-up because there will be too few children to provide meaningful inferences. Finally, all families received a latrine following the 18-month primary endpoint visit, so the focus on interventions is restricted to early-life, whilst recognising the long period between early-life randomized interventions and school-age outcome assessment.

Effects of HIV exposure on school-age outcomes

There remains an urgent need to clarify whether the health gap between CHEU and CHU in sub-Saharan Africa persists at school age, and to identify the biological, social and environmental exposures in early life that underpin these differences. There is a lack of long-term data: one study in Zambia found that the gap between CHEU and CHU growth outcomes had widened by 7.5 years of age²⁵, while a study of neurocognitive outcomes across five African countries²⁶ found no differences between CHEU and CHU groups, but did not compare language, which may be most predictive of future function²⁶. Other cohorts have reported poorer mathematic abilities and reduced IQ among CHEU in south-east Asia²⁷, but studies have focused on a limited range of outcomes. Therefore, applying the SAHARAN toolbox to provide concurrent assessments of growth, health, physical capacity and cognition will allow a holistic insight into the long-term effects of HIV, ART and other pregnancy exposures on health and human capital.

Derivation and application of metrics

Rural Zimbabwe represents a setting of subsistence farming, food and water insecurity, gender inequity, climate change and political/economic fragility. Measuring these environmental exposures, as well as other household and individual factors, enables their effects on school-age growth and function to be investigated. The SAHARAN toolbox takes approximately 4 hours to complete in the field, which is not feasible for widespread applicability. However, using these data to generate readily accessible and applicable shorter metrics will provide future tools that rapidly assess either school-age cognitive function (COG-SAHARAN), physical function and growth (GROW-SAHARAN), or both (SUB-SAHARAN). These metrics will be derived from indicators and tests previously used in other LMIC and hence should be widely applicable. We will provide a toolkit for operationalisation of the metrics in other contexts, allowing other research groups and programmes to adapt the tools prior to implementation.