Enhancing early detection of neurological and developmental disorders and provision of intervention in low-resource settings in Uttar Pradesh, India: study protocol of the G.A.N.E.S.H. programme

Introduction Around 9% of India’s children under six are diagnosed with neurodevelopmental disorders. Low-resource, rural communities often lack programmes for early identification and intervention. The Prechtl General Movement Assessment (GMA) is regarded as the best clinical tool to predict cerebral palsy in infants <5 months. In addition, children with developmental delay, intellectual disabilities, late detected genetic disorders or autism spectrum disorder show abnormal general movements (GMs) during infancy. General Movement Assessment in Neonates for Early Identification and Intervention, Social Support and Health Awareness (G.A.N.E.S.H.) aims to (1) provide evidence as to whether community health workers can support the identification of infants at high-risk for neurological and developmental disorders and disabilities, (2) monitor further development in those infants and (3) initiate early and targeted intervention procedures. Methods This 3-year observational cohort study will comprise at least 2000 infants born across four districts of Uttar Pradesh, India. Community health workers, certified for GMA, video record and assess the infants’ GMs twice, that is, within 2 months after birth and at 3–5 months. In case of abnormal GMs and/or reduced MOSs, infants are further examined by a paediatrician and a neurologist. If necessary, early intervention strategies (treatment as usual) are introduced. After paediatric and neurodevelopmental assessments at 12–24 months, outcomes are categorised as normal or neurological/developmental disorders. Research objective (1): to relate the GMA to the outcome at 12–24 months. Research objective (2): to investigate the impact of predefined exposures. Research objective (3): to evaluate the interscorer agreement of GMA. Ethics and dissemination G.A.N.E.S.H. received ethics approval from the Indian Government Chief Medical Officers of Varanasi and Mirzapur and from the Ramakrishna Mission Home of Service in Varanasi. GMA is a worldwide used diagnostic tool, approved by the Ethics Committee of the Medical University of Graz, Austria (27-388 ex 14/15). Apart from peer-reviewed publications, we are planning to deploy G.A.N.E.S.H. in other vulnerable settings.

overall aims of the project. These limitations, notwithstanding, there are numerous strengths to this project, including: a) it is population-based; b) the use of community workers to conduct the screening exams; c) the project starts early in life (1-5 months); d) the implementation plan seems sound; e) there appears to be an adequate number of workers; f) this approach may in particular be sensitive to CP and epilepsy

REVIEWER
DR. KAI JIN Usher Institute, University of Edinburgh REVIEW RETURNED 24-Jun-2020

GENERAL COMMENTS
This study aimed to detect the neurodevelopmental disorders (NDD) in infants in India. A prospective observational cohort study is used, which is appropriate design to examine the outcome interest (NDD) and exposure variables before the development of the disease. Such design is usually used to answer the question about the association between "risk factors" and disease outcome (neurodevelopmental disorders) and have the strength of temporal relationship between an exposure and an outcome.
However, this paper requires for significant revision in research questions, methods and analysis. Lack of clarity of the research questions has impact on the research design and data analysis. I strongly recommend the authors follow the STROBE protocol (Strengthening the Reporting of Observational Studies in Epidemiology) particularly in the method section.
Research questions 1. Objectives have been repetitively stated several times over the paper without the consistency: 1) Page 6 /26 , line 24 to 33; 2) Page 7/26 , line 3 to 30; 3) Page 13 -14; 4) aims were not stated in the abstract; 2. Lack of clarity: 1) Primary objective: "Our primary objective is to relate the quantity and quality of movement and postural patterns" What relationship is to be examined with the quantity and quality of movement and postural patterns?

2)
Secondary objective "Our secondary objective is to relate the results to a number of predefined confounders. Specific questions will be whether (i) movements and postures at 1 to 5 months, (ii) the 12-to 24-month-outcome, and (iii) the correlation between (i) and (ii) are associated with (a) the access to and frequency of antenatal care, (b) a maternal micronutrient deficit during pregnancy, (c) the safety of child birth,(d) a low birth weight (< 2500g) and undernourishment of the infant, and (e) a delayed cry at birth." What are "the results" to be examined? Are these cofounders or exposures? Which research questions is this analysis for?
3. "To determine whether two or more independent samples had the same distribution in MOS and/or HINE" What is "two or more independent sample", how is this relevant to the research questions?
4. What is the attempts within analysis plan to address for potential confounder? 5. Describe any methods used to examine subgroups and interaction 6. If there is loss to follow up, how to address the missing data 7. Any plan for sensitive analysis

Reviewer 1
In high-income countries there is a long history of attempting to predict neurodevelopmental disorders in early childhood from measures obtained in infancy; this dates back to the development of the Apgar score and in the 1950s and 60s, in the United States, the large National Collaborative Perinatal Project that was particularly focused on predicting cerebral palsy. Admittedly the approach adopted in a high-income country might not generalize to a low-income setting, such as rural India. Nevertheless, it is not apparent to me that the authors are fully aware of this long history of research.

REPLY
Thank you very much for the valuable comments to help amending our manuscript. We are very grateful for the critical remarks on our study protocol and the understanding of the need to adapt clinical and study procedures to the situation in Uttar Pradesh, India.
We are well aware of the long history of attempting to predict neurodevelopmental disorders in early childhood but refrained from reviewing the various attempts from Virginia Apgar to brain imaging in infants born preterm in high-resource settings in this protocol paper. The corresponding author (C.E.) has been working on this topic since the early 1990s.
We do agree with the Reviewer, that the approach adopted in a high-income country cannot be easily generalized to a low-income setting, such as the region of our focus in this project, the rural Uttar Pradesh. Uttar Pradesh is the most populous state in India with 59 million people living below the poverty line, with a disability rate of 15.5%.
In 1997, The Lancet announced our easy to apply, non-invasive and nonintrusive General Movements Assessment (GMA) as the most promising diagnostic tool to identify infants at risks for neurological/developmental disorders and disabilities already at preterm, term, and early post-term age 16 (Manuscript) . Since then, more than 150 studies worldwide in high-and low-resource settings and several meta-analyses have repeatedly demonstrated the validity of this highly reliable and efficient predictive tool. In 2017, the American Medical Association published the Guidelines for early identification of an increased risk for cerebral palsy 18 (Manuscript) (with C.E. as coauthor), stating "the 3 tools with best predictive validity for detecting cerebral palsy before 5 months' Another 12% of the children were diagnosed with intellectual disabilities (including Down syndrome, which does not require the GMA for diagnosis). 7% of the children were diagnosed with hearing or visual impairment, and 4% with epilepsy. Only 2% were diagnosed with psychiatric disorders (ASD, ADHD, and psychosis). Although these numbers refer to the year 2019, the distribution remained similar during the last seven years. Highlighting ASD in the abstract and outlet might have been misleading; we have accordingly changed that in the manuscript.

Reviewer 1
Additional concerns about this project is whether there are other tools that can be used in this setting that the authors are not considering (for example, see work coming out of the INS in Dhaka, Bangladesh); that the statistical approach is rather simplistic; and that no mention is made of how to account for missing data.

REPLY
We appreciate the MRI studies carried out in Dhaka under the supervision of the Reviewer. e.g.44, 45 We were impressed to read that the resting state networks appear to reflect differences between infants from the impoverished families and more affluent families. 45 We recently found that young infants with aberrant general movements exhibited decreased restingstate functional connectivity between the basal ganglia and regions in parietal and frontotemporal lobes. 46 This study was conducted at the University of Chicago Comer Children's Hospital; there is currently no chance to implement a similar study in Varanasi, India.
Both organizations involved in our G.A.N.E.S.H. project are non-profit, nongovernmental organizations. Their activities are depending on funds and donations. MRI is not yet affordable for the general screening process. In case that GMA reveals abnormalities, and subsequent detailed neurological examination by a child neurologist (first author; having been working in the region since 15 years) confirms the necessity of a brain imaging, the infant will be referred to a private Diagnostic Centre (cooperating with our project initiators), or to the Banaras Hindu University, Department of Radio Diagnosis & Imaging, to perform a 1.5 Tesla MRI diagnostic scan.
We were able to acquire extra funding for additional statistical consultancy and have discussed our protocol with clinical advisors and clinical psychologists. We are convinced that stateof-the-art approaches will be applied once data collection is completed. To ensure statistical and mathematical correctness we will include the institutional statistic board at the University Medical Center Göttingen, Germany. free access to all therapies including 1-to 2-week camps for parent's training.
We have currently collected data of 1,300 families; 17 (1.3%) refused to participate in followup assessments and services; one infant with spinal muscular atrophy died. Hence, loss to followup is hardly present, and we do not expect any significant bias on the results.
Missing data mainly relate to birth weight. Especially after home-deliveries not all neonates are weighed. However, we take the anthropometry (Z-scores) in all infants at the appointment for the general movement video recording, which is at least once between postnatal day 4 to 120.

Reviewer 1
There is one additional issue I feel compelled to raise which has to do with the ethics of screening at a population level. Specifically, I understand that children can be referred out for a more formal evaluation (e.g., by a child neurologist) but will services be readily available for children who receive a diagnosis of a disability/disorder? How long will the family have to wait for services? These issues should be considered in the context of the overall aims of the project.

REPLY
Thank you very much for raising this important question. The screening (recording and assessment of general movements) is done by a community health worker certified for GMA. Assessment is done within the first three days after video recording, mostly on the same day. In case of (a) abnormal general movements, (b) low weight, (c) diarrhoea, (d) fever, (e) exanthema or any other health conditions, and most of all, in case of (f) parental concerns that mainly relate to their observation of suspected fits, an appointment with the child neurologist or the paediatrician is scheduled within maximally 1 week. If indicated, EEG will be performed at the Hospital of the Ramakrishna Mission Home of Service. MRI is usually performed within 2 weeks after the prescription. In case the family cannot afford the transportation into the city of Varanasi, a driver from Kiran Society or Ramakrishna Mission will take care of the transportation. Community health workers are also assisting if the family members are not comfortable to bring the child to the expert for further evaluation.

Reviewer 1
These limitations, notwithstanding, there are numerous strengths to this project, including: a) it is population-based; b) the use of community workers to conduct the screening exams; c) the project starts early in life (1-5 months); d) the implementation plan seems sound; e) there appears to be an adequate number of workers; f) this approach may in particular be sensitive to CP and epilepsy.

REPLY
We are most grateful to the Reviewer for recognizing the strengths of our project. Indeed, the majority of infants already screened are diagnosed with motor impairments, often cerebral palsy, and epilepsy. Especially in case of fits observed by parents, an immediate referral to EEG (for which G.A.N.E.S.H. covers the costs) and the start of antiepileptic medication is imperative.
Thank you very much, indeed, for the comments and suggestions to amend our protocol paper. We hope to have addressed all issues properly.

Reviewer 2
This study aimed to detect the neurodevelopmental disorders (NDD) in infants in India. A prospective observational cohort study is used, which is appropriate design to examine the outcome interest (NDD) and exposure variables before the development of the disease. Such design is usually used to answer the question about the association between "risk factors" and disease outcome (neurodevelopmental disorders) and have the strength of temporal relationship between an exposure and an outcome.
However, this paper requires for significant revision in research questions, methods and analysis.
Lack of clarity of the research questions has impact on the research design and data analysis. I strongly recommend the authors follow the STROBE protocol (Strengthening the Reporting of Observational Studies in Epidemiology) particularly in the method section.

REPLY
We are very grateful for the valuable comments and critical remarks on our study protocol. These were extremely helpful to revise and amend our manuscript. As suggested by the reviewer, the STROBE protocol is now implemented in this version, particularly in the Method section.

Reviewer 2
Research questions 1. Objectives have been repetitively stated several times over the paper without the consistency: 1) Page 6 /26 , line 24 to 33; 2) Page 7/26 , line 3 to 30; 3) Page 13 -14; 4) aims were not stated in the abstract; REPLY Thank you very much for pointing this out. Rewriting the manuscript according to the STROBE protocol we also double checked the manuscript for consistency (1-3 above; Abstract and end of Introduction section). Aims (4 above) have been added to the Abstract.

Reviewer 2
2. Lack of clarity: 1) Primary objective: "Our primary objective is to relate the quantity and quality of movement and postural patterns". What relationship is to be examined with the quantity and quality of movement and postural patterns?

REPLY
The "quantity and quality of movement and postural patterns" refers to standard assessments of newborn neuromotor functions. To clarify, the relevant sentence reads as follows: "Our primary objective [now: research objective 1] is to relate the quantity and quality of movement and postural patterns assessed by means of the Prechtl GMA (at 1 to 5 months) and MOS (at 3 to 5 months) to the outcome at 12 to 24 months, which is assessed in paediatric and neurological examinations as well as developmental tests." We have checked similar aspects in a number of papers that associated assessments in infancy with outcome measures and found comparable phrasing in these publications. Indeed, there is a huge body of knowledge from all over the world e.g.17,25 (Manuscript) that GMA (assessing movements and postures during early infancy as in the proposed protocol) has high predictive values for outcome assessments performed during toddlerhood, e.g.29 (Manuscript) preschool age, e.g.47,48 school age, e.g.49,50 and even puberty 51 . However, none of these studies are carried out in low-resource settings.

Reviewer 2
2) Secondary objective "Our secondary objective is to relate the results to a number of predefined confounders. Specific questions will be whether (i) movements and postures at 1 to 5 months, (ii) the 12-to 24-month-outcome, and (iii) the correlation between (i) and (ii) are associated with (a) the access to and frequency of antenatal care, (b) a maternal micronutrient deficit during pregnancy, (c) the safety of child birth, (d) a low birth weight (< 2500g) and undernourishment of the infant, and (e) a delayed cry at birth. "What are "the results" to be examined? Are these cofounders or exposures?

REPLY
We were indeed wrong to use the term "predefined confounders" instead of "exposure variables" for the factors listed from (a) to (e). Thank you for helping us to clarify this issue.

Reviewer 2
Methods 1. Study design: a. The schematic of study design would be very helpful to understand this design and strongly recommended to provide;

REPLY
We are grateful for this suggestion and have added Figure 1 illustrating the study design.

Study population:
a. Give the eligibility criteria to include and exclude criteria;

REPLY
The eligibility criteria are now given in the Participants topic. We include infants younger than 5 months of age irrespective of gender, family background, medical history, and current health status.
Infants older than completed 5 months are not included in the study.

Reviewer 2
b. Describe the sampling methods: e.g. convenient sampling;

REPLY
Our study is a population based study aiming to answer the research objectives for all infants younger than 5 months of age and living in a defined area, namely in 29 (cluster of) villages in ten blocks of four districts of Uttar Pradesh as described in the Setting section. We have defined a study interval of 2-years for the first assessments (1 to 5 months), which now needs to be expanded because of COVID-19 pandemic regulations.
Please see also reply below.

Reviewer 2
c. Describe method of follow-up. What is the efforts to improve the future loss to follow up;

REPLY
The G.A.N.E.S.H. project is perceived by the communities as part of the ongoing services, which have been provided by the Kiran Society and the Ramakrishna Mission for decades. The first two authors (a child neurologist and a paediatrician) have both been working more than 15 years in the target area and are well known to the community. Approximately 90% of newborns and young infants are brought to the services or, alternatively, the community health workers are welcomed at the family home. The remaining 10% have moved to the mother's home (in another district) after having given birth. Though this is part of the Indian culture, the number of moves to other regions is small as the majority of maternal grandparents live in the same region. Altogether, the number of families not willing to participate in the G.A.N.E.S.H. project is negligibly small. Since many years, all infants receive neurological and paediatric assessment and follow-up examination for free. In addition, we provide micronutrient and food supply if needed; in case of need of further evaluations affordable or free X-ray, MRI, surgical corrections (cleft palate, clubfoot), or genetic testing; free access to all therapies including 1-to 2-week camps for parent's training.
We have currently collected data of 1,300 families; only 17 (1.3%) refused to participate in follow-up assessments and services; one infant with spinal muscular atrophy died. Hence, loss to follow-up is hardly present, and we do not expect any bias on the results.
But, of course, again the regulations due to COVID-19 pandemic might increase loss to follow-up and can hardly be foreseen now. Overall, the established child and family care through the involved institutions and the personal involvement of a sustainable health-care provider system guarantees that the loss to follow-up is reduced to a minimum.

Reviewer 2
3. Page 12 intervention program: not clear, please clarify: a. Do the infants with abnormal repertoire GM at 3-5months receive the intervention?

REPLY
We have now added that infants with abnormal poor repertoire GMs during the first two months, will be re-checked concerning their GMs at 3 to 5 months and the result will determine the further procedure. As described in tops (b) and (c) of the Intervention topic in the main text, intervention is provided if the infant does not develop fidgety movements or if his/her fidgety movements are exaggerated, abnormal, and/or if the MOS <20.

Reviewer 2
b. How long is the intervention; REPLY As we are not studying the effect of intervention, and we are subsuming allopathic treatments (e.g. antiepileptic drugs), micronutrient supply, education in hygiene and sanitation, physio-and occupational therapy etc. as "intervention", treatment is subject to personalized medicine to meet the needs of the individual infant and his/her family.

Reviewer 2
c. Are outcome of interest assessed at 12 -24 month after intervention?

REPLY
In this study we are not evaluating the effectiveness of interventions (please see also answer below). As in numerous studies in high-resource settings, we are studying GMA as a predictor for outcome and thus related to diagnosis made.

Reviewer 2
d. Any intention to compare the effectiveness of the intervention in this program?

REPLY
Infants and families receive individual treatments tailored to their specific needs. We do not apply any special type of intervention which we could compare to the treatment as usual approach.
Please keep in mind the very low resource setting and restricted possibilities in this respect. It is not an intervention study but rather aims at providing the best available treatment at the earliest possible time. It would certainly be of interest to evaluate different interventional approaches to this population but this is beyond the scope of the project and far beyond feasibility in such a vulnerable setting.

REPLY
We have now included in the Variables topic that a potential confounder might be undernutrition, especially after weaning from breastfeeding after 6 months of age. It can be directly attributed to inadequate dietary intake or infection or disease that affects the child. Lack of sanitation and hygiene, inadequate care and maternal mental health, economic deprivation, and food insecurity might be contributory factors. We will be able to control the confounder bias (undernutrition), as we will document any recognizable external disadvantages hampering thriving, and compare the Zvalues of the anthropometric measurements at birth, 1 to 5 months, and 12 to 24 months.

Reviewer 2
5. Describe power and sample calculation and rationale

REPLY
We have now included a Study size topic stating the following: Taken an alpha cut-off of 5% and a beta cut-off of 20%, the sample shall involve 1,261 children (i.e. with a pooled incidence of 9.2% neurological/developmental disorders in Indian children younger than 6 years, and an expected incidence of 7% in the study group, with a birth rate in the target area varying from 19.9 to 28.7 per 1,000 we expect to see 1,000 to 1,500 newborns per year, resulting in about 70 to 105 children with neurological/developmental disorders and disabilities) to achieve a predictive power of .95 and an effect-size of .15 using Cohen's f2. According to our eligibility criteria, we aimed to assess a cohort of at least 2,000 infants born in the first and second year of the study. Until March 15, 2020 (13 months after launching the study, before data collection was deferred due to the COVID-19 pandemic), we have successfully recruited 1,300 infants.

Reviewer 2
Statistical methods: In principle, statistical analysis plan should reflect the research design and research questions, and emphasis should be on which analyses, comparisons, and statistical tests have been planned given the objective of the study.

REPLY
With the assistance of a biostatistician, we have rewritten the Statistic methods topic according to the STROBE guidelines.

Reviewer 2
1. Statistical analysis Page 14: "First, we evaluate descriptive parameters (medians and interquartile ranges or percentages) to summarise sample characteristics" What variables are to be summarised? How this analysis is relevant to the objectives of this study?

REPLY
The reviewer is right; we did not summarise variables. What we meant is to describe the clinical characteristics of all infants involved in the study. We have rephrased the relevant sentences.

Reviewer 2
2. Page 15; "Odds ratios and 95% confidence intervals are calculated for (a) preand perinatal data that reveal a significant difference between normal and abnormal GMs, and (b) normal and abnormal GMs that reveal a significant difference between the categorical outcome variables assessed at 12 to 24 months." Which research questions is this analysis for?

REPLY
With the assistance of a biostatistician, we have rewritten the Statistic methods topic according to the STROBE guidelines.

Reviewer 2
3. "To determine whether two or more independent samples had the same distribution in MOS and/or HINE". What is "two or more independent sample", how is this relevant to the research questions?

REPLY
With the assistance of a biostatistician, we have rewritten the Statistic methods topic according to the STROBE guidelines.

Reviewer 2
4. What is the attempts within analysis plan to address for potential confounder?

REPLY
Whether indeed undernutrition will be a confounder can be determined by comparing the Zvalues of the anthropometric measurements carried out at three time points: birth, 1 to 5 months, and 12 to 24 months. The potential confounder(s) will be taken as covariates in our mixed ANOVA analyses.

Reviewer 2
5. Describe any methods used to examine subgroups and interaction REPLY As we are not conducting a clinical trial we did not describe subgroup analysis and interaction tests. In addition, previous studies e.g. 16,17,20,29 (Manuscript) revealed that there is no subgroup (demographics, i.e. gender, recording age group) effect on the GMA identifying infants at high-risk for developmental disorders and disabilities. Following the advice of the reviewer, we shall certainly evaluate for any subgroup modification of the predictive power of the outcome. Our statistician shall conduct a test for interaction using GLM models to evaluate subgroup differences.

Reviewer 2
6. If there is loss to follow up, how to address the missing data REPLY