The Sanitation Hygiene Infant Nutrition Efficacy (SHINE) Trial: Protocol for school-age follow-up

Background: There is a need for follow-up of early-life stunting intervention trials into childhood to determine their long-term impact. A holistic school-age assessment of health, growth, physical and cognitive function will help to comprehensively characterise the sustained effects of early-life interventions. Methods: The Sanitation Hygiene Infant Nutrition Efficacy (SHINE) trial in rural Zimbabwe assessed the effects of improved infant and young child feeding (IYCF) and/or improved water, sanitation and hygiene (WASH) on stunting and anaemia at 18 months. Among children enrolled to SHINE, 1,275 have been followed up at 7-8 years of age (1,000 children who have not been exposed to HIV, 268 exposed to HIV antenatally who remain HIV negative and 7 HIV positive children). Children were assessed using the School-Age Health, Activity, Resilience, Anthropometry and Neurocognitive (SAHARAN) toolbox, to measure their growth, body composition, cognitive and physical function. In parallel, a caregiver questionnaire assessed household demographics, socioeconomic status, adversity, nurturing, caregiver support, food and water insecurity. A monthly morbidity questionnaire is currently being administered by community health workers to evaluate school-age rates of infection and healthcare-seeking. The impact of the SHINE IYCF and WASH interventions, the early-life ‘exposome’, maternal HIV, and contemporary exposures on each school-age outcome will be assessed. We will also undertake an exploratory factor analysis to generate new, simpler metrics for assessment of cognition (COG-SAHARAN), growth (GROW-SAHARAN) and combined growth, cognitive and physical function (SUB-SAHARAN). The SUB-SAHARAN toolbox will be used to conduct annual assessments within the SHINE cohort from ages 8–12 years. Ethics and dissemination: Approval was obtained from Medical Research Council of Zimbabwe (08/02/21) and registered with Pan-African Clinical Trials Registry (PACTR202201828512110, 24/01/22). Primary caregivers provided written informed consent and children written assent. Findings will be disseminated through community sensitisation, peer-reviewed journals and stakeholders including the Zimbabwean Ministry of Health and Child Care.


•
This protocol describes the follow-up of school-age children after receiving early-life IYCF and WASH interventions in rural Zimbabwe.
• A comprehensive measurement of the child's health, growth, body composition, physical and cognitive function has been performed.
• Analysis of the 'exposome', including maternal HIV status, will identify the early-life factors that shape school-age physical and cognitive function.
• Capitalizing on the randomized trial design will allow the long-term effects of IYCF and WASH on child health outcomes to be estimated.

Introduction
Children are defined as "stunted" when their height-for-age Z-score (HAZ) is more than two standard deviations below the World Health Organization (WHO) reference standard, but linear growth faltering also affects many children who have not yet fallen below this cutoff 1 .Stunting is associated with increased mortality, poorer school performance, lower adult earnings and long-term chronic disease 2 .Stunting affects 22% (149 million) of children under 5 years old 3 , while up to 250 million children are at risk of not reaching their developmental potential 4 .This makes stunting a global health priority, although its causes, longer term impact and response to interventions remain poorly understood.
Given the ongoing global consequences of stunting, long-term follow-up studies reflecting contemporary conditions, geography and interventions for stunting are urgently needed to inform robust, cost-effectiveness analyses and advocacy of IYCF and WASH policies and programming (Figure 1).
The lack of long-term data evaluating combined neurodevelopment, physical fitness and growth has led to a call for further studies that measure a range of outcomes 5,6 .A holistic measurement is vital to understand school-age trajectories across different functional domains and the impact of risk and protective factors and potential interventions (Figure 2).At school-age, Children receiving additional nutrition by age 2 years initially had only modest effects on neurodevelopmental scores; however, in long-term follow-up, those who had received the nutrition intervention had higher IQ scores, greater work capacity and earnings (among men) and greater schooling (among women) 4 .The INCAP study was conducted at a time when global stunting prevalence was much higher (1969-1977): 50% of the study population had HAZ<-3.0,while currently worldwide 22% children have HAZ <-2.0.Furthermore, the impact of the intervention on linear growth was much greater than that seen in trials of complementary feeding interventions over the past 20 years (+0.62 HAZ compared to +0.11 HAZ).Thus, although the Guatemala trial suggests that complementary feeding interventions can have substantial long-term physical and neurodevelopmental benefits, it does not reflect today's situation in which Africa has the highest stunting prevalence, severe stunting is relatively rare but moderate stunting is 20-40%, and the average impact of interventions on HAZ is only 0.1-0.it becomes easier to undertake more detailed measures of cognitive development, including school performance, executive function and socio-emotional behaviour.

Aims and objectives
This study is undertaking long-term follow-up of a cohort of children enrolled in the Sanitation Hygiene Infant Nutrition Efficacy (SHINE) trial in rural Zimbabwe.SHINE was a 2x2 factorial cluster-randomized trial across two contiguous districts in rural Zimbabwe between 2012-17, designed to test the independent and combined effects of improved infant and young child feeding (IYCF) and improved water, sanitation and hygiene (WASH) on child height-for-age Z score (HAZ) and haemoglobin at 18 months of age.All outcomes were stratified by maternal HIV status.Detail of the trial design 7 and primary results [8][9][10][11] are published.In brief, SHINE found that the IYCF intervention modestly improved linear growth, reducing stunting prevalence by 20%, while the WASH intervention had no significant effect.Children who have been exposed to HIV and remain HIV-negative (HIV-exposed and uninfected, CHEU) had a higher risk of stunting and impaired neurodevelopment than children who have not been exposed to HIV (HIVunexposed, CHU), but also intriguing evidence of increased responsiveness to early-life interventions 10,11 .
Long-term follow-up of this cohort now enables us to address: i) the effects of the randomised IYCF and WASH interventions on school-age growth, body composition, cognitive and physical function; ii) the impacts of early-life stunting and a range of other exposures (the 'exposome') on school-age growth, body composition, cognitive and physical function; and iii) the differences in school-age growth, body composition, cognitive and physical function between CHEU and CHU.Therefore, the objectives are summarised as follows: 1) Evaluate the impact of the IYCF intervention on school-age growth, health, physical and cognitive function (both for children who have been exposed to HIV and those who have not been exposed).
2) Evaluate the impact of the WASH intervention on school-age growth, health, physical and cognitive function (both for children who have been exposed to HIV and those who have not been exposed).
3) Evaluate the relationship between the exposome during the first

Methods
SHINE enrolled pregnant women from 210 clusters 7,8 , each representing the catchment area of one to four community health workers (CHW) and covering a geographic area of 8,232 km 2 .For this long-term follow-up study, 1,275 children born to these women from 105 clusters in Shurugwi district were enrolled.A total of 1,000 children who originally participated in SHINE and were aged between 7-8 years during the period of enrolment and have not been exposed to HIV (CHU) were randomly selected by a computer from clusters by the data team.For those who had moved away or were otherwise lost to follow-up, randomised replacements within the SHINE cohort were selected to ensure 1,000 children.All children exposed to HIV were invited to enrol, with 268 CHEU enrolled in total; seven children found to be living with HIV were excluded from the analysis, and will form the basis of a separate case report, given their low number.Children no longer resident in Shurugwi, with unknown maternal pregnancy HIV status, or outside the age window were ineligible.There are no new interventions in the long-term follow-up study.

Participant and public involvement
School-age participants and their caregivers were not involved in designing the protocol or selecting outcomes.However, they were involved in design and implementation of the questionnaires and tools.All tests to measure school-age child growth, physical and cognitive function were discussed with the District Health Executive and community leaders.All questionnaires underwent cognitive interviewing with community members who suggested alterations and feedback.Tools were then pre-tested with local families to ensure acceptability, with feedback sought from children and caregivers.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 had received information and expressed an interest in joining, a mutually convenient date was booked for an assessment.
Screening, consent, assent and assessments Assessments were conducted by data collectors (DCs), who were primary care nurses extensively trained in study procedures.On the day of the assessment, the CHW introduced a pair of DCs (DC1 and DC2) to the household.Screening was undertaken to ensure the original SHINE child and primary caregiver were present.The verification for the primary caregiver (usually the mother) and child was mainly through the birth certificate, identity card and other papers such as the child's Ministry of Health baby health card, as well as noting any evidence of previous recruitment into the SHINE trial (e.g., family showing previous consent documents or a SHINE latrine at the household).Once both DCs verified that this was the correct child, one to two tents were pitched in or close to the household.Households were usually single-family dwellings surrounded by subsistence farmland and the mean distance between houses in SHINE was 82.6 m 8 , ensuring the tents were sufficiently isolated for confidentiality during informed consent and data collection.Written informed consent was obtained from the primary caregiver, and assent from the child, following a demonstration of the tools used to conduct the measurements.The primary caregiver was usually the mother, but it was the child born within the SHINE trial who is the focus of this follow-up study.Hence if the mother was not available (for example the mother had died or moved away), the primary caregiver was defined as the adult most responsible for the child's care and completed the caregiver questionnaire.
DC1 performed the caregiver questionnaire in a pre-specified order, while DC2 conducted the cognitive, physical and anthropometry assessments in a separate tent to minimize distractions, after building rapport with the child.The child was fed before the start of the tests (either by the family or the DC's who provided an additional snack of biscuits and Maheu, a traditional maize-based drink).The child was then given further regular breaks and snacks (also biscuits and Maheu) to reduce tiredness.All tests had standardized explanations and demonstrations to ensure appropriate understanding, and were translated into Shona or Ndebele-the local languages spoken in the district.The DCs remained blinded to the original trial allocation and the maternal HIV status throughout the visit.At the end of the visit, DC1 opened a sealed envelope to check the maternal HIV status in the original trial, which then determined whether questions were asked on HIV treatment or if HIV testing was appropriate.At the end of the visit, a small gift was given to the child and caregiver to thank them for their time comprising: a pencil, two pens, exercise book, story book, bar of soap, jar of peanut butter and two litre bottle of cooking oil.The start and end time and temperature at the beginning and end of the assessments was also recorded.

HIV testing and referrals
All primary caregivers were offered HIV testing at the end of the visit unless they were already known to be living with HIV or had a documented negative result in the previous three months.If the mother did not have HIV, the child was not tested.
If the mother was living with HIV, declined testing or was not available, the child was offered HIV testing with age-appropriate assent using role-plays.The Determine HIV-1/2 rapid test (Abbott) was used for initial testing; positive results were repeated using the HIV 1/2 Stat-Pak rapid test (Chembio).The option of dried blood spot testing was also available for participants who preferred not to be tested in real-time in the homestead.Referrals were made to local clinics for mothers or children testing positive for HIV or for other conditions found on visiting which necessitated a referral, as defined by a SOP.

Outcome measures
All assessment tools were integrated into a single assessment lasting 4-5 hours, using the School Age Health, Activity, Resilience, Anthropometry and Neurocognitive (SAHARAN) toolbox, which was developed, refined and piloted by our team 12 .Briefly, SAHARAN includes a caregiver questionnaire, child questionnaire and direct tests undertaken with the child, focused on three domains: cognitive function, growth and body composition, and physical function.The primary outcome of the long-term follow-up study is cognitive function, assessed by the mental processing index (MPI)-the total score from eight subtests of the Kaufman Assessment Battery for Children 2 nd edition (KABC-II) 13 .The subtests measure four domains of cognitive processing across learning, planning, simultaneous and sequential memory.All other outcomes are secondary outcomes and shown in Table 1-Table 3.

Support and supervision.
Ongoing support and monitoring were provided by a senior research nurse and a doctor, with 6-monthly standardisation exercises.Additional training was provided for any DCs 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) 14 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 (with a visit window of 3 to 5 weeks) 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.

Sample size
The primary outcome is the total KABC-II from all subtests, called the Mental processing index (MPI).The raw subtest scores are adjusted by age into 4-month blocks.A total of 1,000 children (500 IYCF vs. 500 non-IYCF) were assessed, providing 86% power to detect a 0.2 standard deviation difference in MPI between IYCF and non-IYCF arms with alpha 0.05, assuming intra-cluster correlation of 0.05 and sampling from 100 clusters.This will allow the exploration of the difference in IQ scores observed at 3-7 years of age among children followed-up in the Institute of Nutrition of Central America and Panama (INCAP) study, and is also the approximate magnitude of socio-emotional difference recently shown with a similar small-quantity lipid-based nutrient supplement trial 15 .Since data collection was undertaken directly after consent and enrolment, no adjustment for lost to follow-up was required.
The SHINE 2x2 factorial design also enables 500 WASH versus 500 non-WASH children to be assessed as a secondary outcome.For CHEU, the sample size is 268, which allows exploration of the association between the treatment arms (IYCF vs. non-IYCF, and WASH vs. non-WASH) for secondary outcomes.

Data management
DCs electronically collected the SAHARAN questionnaire and observation data onto password-protected Android tablets (Samsung Galaxy Tab A) using the Open Data Kit (ODK) platform (https://opendatakit.org/).The ODK forms were programmed with skip patterns, expected data ranges and free text sections to report any issues with data collection.Back-up paper forms were also provided in case of tablet failure.Data were manually checked by the Project Lead and Field Data Officer and then uploaded onto an ODK Aggregate Server, and stored on an SQL Server database at the end of each working day.All data were collected using a unique participant identifier (PID) to maintain confidentiality.A CC-BY license will be applied.Given this was a long-term follow-up of a cluster randomised trial (with the interventions previously completed), a data monitoring committee or interim analysis was not needed.

Statistical analysis
A Statistical Analysis Plan (SAP) designed to give transparency to possible influential statistics decisions for the study was finalised prior to unblinding and initiation of analyses and is available as Extended data 16 .
Reporting of results will follow the guidelines established in the extended CONSORT guidance for cluster-randomized trials 17 .
We used the SPIRIT checklist 16 when writing our report given that the SHINE cluster randomised trial primary outcome at 18 months has been completed.The analyses are described below according to the objectives 18 .

Effect of IYCF and WASH on outcomes (Objectives 1 & 2).
We will capitalize on the 2x2 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 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 Table 1 to Table 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 enrolment 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.

Evaluating early-life growth and exposures on school-age function (Objectives 3a, b, c).
As secondary outcomes, we will explore the effects of early stunting (by 1 month of age) and late stunting (by 18 months of age) on each outcome at 7 years of age, and the effect of LAZ as a continuous variable at 1 and 18 months.We will also perform a principal components analysis on the 7 year outcomes to reduce the number of outcome variables into a smaller set of functional biological domains that best represent the most informative combination of outcomes.Early-life factors can be grouped into environmental (e.g., socioeconomic status, food and water insecurity, household adversities), maternal (e.g., maternal education and depression) and nurturing factors (caregiving) (Figure 3).For the early-life exposome, we will use least absolute shrinkage and selection operator (LASSO) for Generalised Estimating Equations (GEE) to quantify the effects of pregnancy and baseline exposures on school-age outcomes 19 , exploiting the rich SHINE dataset to control for confounding by pregnancy-related factors (e.g., wealth, maternal education, haemoglobin, maternal capabilities), the trial interventions, and the time-varying nature of exposures experienced between birth and 18 months of age.

CHEU and CHU comparison (Objective 3d
). Baseline characteristics between CHEU and CHU groups will be compared using multinomial and ordinal regression models and Somers' D for medians, while handling within-cluster correlation with robust variance estimation.We will use GEE models to compare each functional outcome, assessed using the SAHARAN Toolbox, between CHEU and CHU groups.A secondary adjusted analysis will be performed including possible confounders as defined in the SAP.Poisson regression will be used to compare the cumulative incidence of illness episodes, clinic visits, and hospitalizations between groups during prospective surveillance measured by the CHW illness questionnaire.

Comparison between aspects of school-age growth and function and contemporary conditions (Objective 4).
Internal consistency of outcomes across cognitive domains will be compared, including the primary outcome (KABC-II total score) and other cognitive measurements.Similarly, consistency of physical function measurements and their associations with growth will also be measured, as previously described 12 .Univariable analysis will explore associations between school-age growth, cognitive and physical function (Table 1 to Table 3) and current environmental, schooling, nurturing and caregiving exposures (Table 4).We will use LASSO for GEE to explore the effect of contemporary covariates on 7-year outcomes.We will also apply hierarchical clustering of principal component scores to group children according to their school-age outcomes.We will explore univariable distributions (central tendencies, variance, prevalence) of school-age outcome variables within each identified cluster of children to gain further insights into trade-offs and prioritisation in growth and function 20 .

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 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 previously published metrics 21,22 .
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 novel metrics will be investigated within a recognised conceptual framework that evaluates child function 23 (Figure 3).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 1,000 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.

Dissemination
Results of the follow-up study will be disseminated through community gatherings, building on recent screenings of a locally made film about the SHINE trial.We will also leverage the district Community Engagement Advisory Board, which comprises key stakeholders from the community.Findings will be disseminated to the academic community through open-access peer-reviewed journals and conference presentations.In addition, a short animation video describing the school-age assessment metrics will be created and published

Study status
The study has currently finished recruiting with 1,275 children enrolled (100 HIV unexposed, 275 exposed to HIV, of whom seven were HIV positive).Data cleaning and analysis is currently ongoing.In addition, the monthly morbidity questionnaires and the annual Sub-SAHARAN monitoring visits are currently being performed.

Discussion
The

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 age 28 , while a study of neurocognitive outcomes across five African countries 29 found no differences between CHEU and CHU groups, but did not compare language, which may be most predictive of future function 29 .Other cohorts have reported poorer mathematic abilities and reduced IQ among CHEU in south-east Asia 30 , 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 and vulnerability climate change.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.The paper adequately examines the longterm effect of the SHINE trial.Such analysis is important to ascertain the sustainability and/or sustaining effect of nutrition interventions over a period of time.I recommend the paper to be approved with minor revisions.The authors need to include the number of district the SHINE trail was implemented in.This will then make sense of the limitation of selecting only one of the district for this study.
The first paragraph of the discussion seems out of place for me.It discusses more of the rationale of the research rather than beginning straight with the contribution of the paper to current scholarship.More of the information should be in the paragraphs information needs to be in the introduction.

2.
Is the rationale for, and objectives of, the study clearly described?Yes

Figure 1 .
Figure 1.Proportion of children that are stunted (UNICEF, WHO and World Bank, kindly reproduced from 'Our World in Data').Inset details describe the landmark INCAP study and the proposed SHINE study, both of which provide long-term follow-up of a nutrition intervention.Data from the landmark INCAP study in Guatemala 50 years ago indicate that early-life improvements in nutrition can confer long-term benefits for cognition.Children receiving additional nutrition by age 2 years initially had only modest effects on neurodevelopmental scores; however, in long-term follow-up, those who had received the nutrition intervention had higher IQ scores, greater work capacity and earnings (among men) and greater schooling (among women)4 .The INCAP study was conducted at a time when global stunting prevalence was much higher(1969-1977): 50% of the study population had HAZ<-3.0,while currently worldwide 22% children have HAZ <-2.0.Furthermore, the impact of the intervention on linear growth was much greater than that seen in trials of complementary feeding interventions over the past 20 years (+0.62 HAZ compared to +0.11 HAZ).Thus, although the Guatemala trial suggests that complementary feeding interventions can have substantial long-term physical and neurodevelopmental benefits, it does not reflect today's situation in which Africa has the highest stunting prevalence, severe stunting is relatively rare but moderate stunting is 20-40%, and the average impact of interventions on HAZ is only 0.1-0.2.UNICEF, United Nations Children's Fund; WHO, World Health Organization; INCAP, Institute of Nutrition of Central America and Panama; SHINE, Sanitation Hygiene Infant Nutrition Efficacy; HAZ, height-for-age Z-score; Ppln, population; Rnd, randomisation technique; Intn, Intervention; WASH, water, sanitation and hygiene.
Figure 1.Proportion of children that are stunted (UNICEF, WHO and World Bank, kindly reproduced from 'Our World in Data').Inset details describe the landmark INCAP study and the proposed SHINE study, both of which provide long-term follow-up of a nutrition intervention.Data from the landmark INCAP study in Guatemala 50 years ago indicate that early-life improvements in nutrition can confer long-term benefits for cognition.Children receiving additional nutrition by age 2 years initially had only modest effects on neurodevelopmental scores; however, in long-term follow-up, those who had received the nutrition intervention had higher IQ scores, greater work capacity and earnings (among men) and greater schooling (among women)4 .The INCAP study was conducted at a time when global stunting prevalence was much higher(1969-1977): 50% of the study population had HAZ<-3.0,while currently worldwide 22% children have HAZ <-2.0.Furthermore, the impact of the intervention on linear growth was much greater than that seen in trials of complementary feeding interventions over the past 20 years (+0.62 HAZ compared to +0.11 HAZ).Thus, although the Guatemala trial suggests that complementary feeding interventions can have substantial long-term physical and neurodevelopmental benefits, it does not reflect today's situation in which Africa has the highest stunting prevalence, severe stunting is relatively rare but moderate stunting is 20-40%, and the average impact of interventions on HAZ is only 0.1-0.2.UNICEF, United Nations Children's Fund; WHO, World Health Organization; INCAP, Institute of Nutrition of Central America and Panama; SHINE, Sanitation Hygiene Infant Nutrition Efficacy; HAZ, height-for-age Z-score; Ppln, population; Rnd, randomisation technique; Intn, Intervention; WASH, water, sanitation and hygiene.

Figure 2 .
Figure 2. The life course approach to child growth and development (reproduced from Child Health for all, 6 th Edition with permission from Oxford University Press).Healthy (blue) and stunted (red) trajectories are shown, illustrating that lifelong health and function are particularly affected by early-life conditions, as well as risk and protective factors throughout life.This applies to linear growth, physical function and health as well as cognitive function and mental health.IYCF, infant and young child feeding.

Figure 3 .
Figure 3. 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.SAHARAN, School-Age Health, Activity, Resilience, Anthropometry and Neurocognitive; IYCF, infant and young child feeding; WASH, water, sanitation and hygiene.

Table 1 . Cognitive outcomes to be measured, including the primary outcome
(MPI) and secondary outcomes.MPI, mental processing index; KABC-II, Kaufman Assessment Battery for Children 2 nd edition.

Table 4 .
Caregiver questionnaire contemporary exposures.MICS, Multi-indicator cluster survey from UNICEF; HFIAS, Household Food Insecurity Assessment Scale; HDDS, Household Dietary Diversity Score; FCS, Food Consumption Score; HWISE, Household Water Insecurity Experiences Scale; ARV, Antiretroviral.(Note that WG & SDQ sections are included in cognitive outcomes).
online, as well as workshops with key local and national partners including Ministry of Health and Child Care, United Nations Children's Fund (UNICEF), Food and Agriculture Organisation, World Food Program, Ag2Nut and the Agriculture, Nutrition and Health (ANH) academy.An earlier version of this article can be found on medRxiv (doi: https://doi.org/10.1101/2022.08.17.22278247).Statistical code and dataset will be available from the data repository on request, https://clinepidb.org/ce/app.A creative commons licence CC-BY 4.0 will be applied to the dataset.
66issing middle' of school-age child health and functionThe 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 ages 5-14 years, mortality is concentrated in sub-Saharan Africa 24 , but measures of child growth and development have frequently been overlooked25.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 conditions26, 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.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 children who have not been exposed to HIV is randomised to ensure balance between the SHINE intervention arms.However, we will try to follow all children exposed to HIV, which may lead to an imbalance in treatment arms for CHEU.Children living with HIV are too few in number to provide meaningful inferences, so they will form a separate case report.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.
Follow-up of the SHINE trial provides a unique opportunity to comprehensively characterise the long-term response to early-life nutrition and WASH interventions, for both children exposed to HIV and those who have not been exposed to HIV.In this context, use of the SAHARAN toolbox provides a holistic outcome measurement, which is crucial to understand school-age trajectories across different functional domains and the long-term impact of exposures including interventions, risk and protective factors.Our study will inform future policy and programming approaches to stunting since demonstration of long-term functional benefits would motivate scale-up of IYCF in areas of high stunting or HIV exposure.Conversely, a lack of school-age benefits suggests that alternative approaches to stunting, poor neurodevelopment and their subsequent mitigating strategies are urgently needed.The metrics themselves also represent paradigm shifts in evaluation of nutrition-sensitive interventions in a broader context; they will provide assessment tools for school-age growth and noncommunicable disease risk, with functional measures of physical health and human capital.It is only by generating both the measurement tools and long-term data that evidence can be translated into action.Mutasa contributed study design, supported field operations, data collection and reviewed the draft of the paper.Melissa J Gladstone contributed study design, choice and application of cognition tools, data interpretation and write-up.Elizabeth Allen contributed statistical design, feedback and also reviewed the draft of the paper.Jonathan C Wells contributed study design, choice and application of body composition tools, data analysis, interpretation and write-up.Melanie Smuk, Bernard Chasekwa and Robert Ntozini provided statistical planning, analysis, randomisation, support and write-up.All authors contributed to the manuscript, including revisions as required and approved the final manuscript.

the study design appropriate for the research question? Yes Are sufficient details of the methods provided to allow replication by others? Yes Are the datasets clearly presented in a useable and accessible format? Yes Competing Interests:
No competing interests were disclosed.Limitation -Current Food security situation is not assessed, which is one of the key factors for school outcomes, better to add it.Birth weight is one of the key variable of stunting and school outcome, is it included in the analysis?If not better to include under design or limitations.Details on data tools -It is not fully available in this protocol, however, Previous data tools are available with the previous study publication.New component could not found.
Reviewer Expertise: Maternal and Child nutrition, Population Health I confirm that I have read this submission and believe that I have an appropriate level of ○ ○ Is

the rationale for, and objectives of, the study clearly described? Yes Is the study design appropriate for the research question? Yes Are sufficient details of the methods provided to allow replication by others? Yes Are the datasets clearly presented in a useable and accessible format?
Reviewer Expertise: Public health nutrition, clinical nutrition, public health and migrant healthI confirm that I have read this submission and believe that I have an appropriate level of