Study design

Observational cohort study which follows the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Study setting and profile

In 2019, G.A.N.E.S.H. was implemented in 29 (cluster of) villages in 10 blocks of four districts of Uttar Pradesh, where the burden of poverty and undernutrition is particularly high: the Varanasi district (roughly 3 700 000 inhabitants; 1535 km²; eight villages in the Kashi Vidyapeeth block, which is covered by KS); the Mirzapur district (roughly 2 500 000 inhabitants; 4521 km²; 12 villages in the Majhawan and Shikhar blocks, likewise covered by KS; seven villages or cluster of villages in the Chhanbey, Haliya, Jamalpur, Lalganj, Majhawan, Pahari and Patehara blocks, covered by RKM); the Sonbhadra district (roughly 1 800 000 inhabitants; 6788 km²; one cluster of villages in the Robertsganj block, covered by RKM) and the Azamgarh district (roughly 4 600 000 inhabitants; 4054 km²; one cluster of villages in the Lalganj block, covered by RKM).

Eastern Uttar Pradesh (particularly the districts of Mirzapur and Sonbhadra) has one of the lowest health indicators of India’s eight socioeconomically backward states, also referred to as Empowered Action Group states. Poor health is characterised by lack of nutrition, chronic diseases and a high prevalence of infectious diseases resulting both in high mortality and morbidity and also in an increase of poverty due to out-of-pocket expenses and lost earnings.14 The main sources of livelihood are agriculture, carpet weaving and pottery, but youth unemployment is high. According to the latest census from 2011,2 the literacy rate then was between 64% in the Sonbhadra district and 75% in the Varanasi district, with a 15%–20% lower literacy rate in women.

The target area is characterised by a lack of trained medical staff willing to work in rural areas and a lack of healthcare infrastructure. Most of the care is provided by the family. Families who seek treatment will often turn to non-allopathic providers such as practitioners of Indian traditional medicine and quacks (ie, unqualified and unskilled health practitioners pretending to have medical skills).14 The lack of neurodevelopmental health professionals, particularly in rural areas, places specialist neurodevelopmental care out of the reach of most people. The G.A.N.E.S.H. programme addresses this critical issue and seeks to provide healthcare to families in need.

The target population of G.A.N.E.S.H. comprises approximately 50 000 people. We aim to assess a cohort of at least 2000 infants born in the first and second years of the study (2019–2020) and collect their outcome data during the second and third years of the study (2020–2021). Due to the COVID-19 pandemic, data collection is deferred since 15 March 2020 and the timeline of the study will be rescheduled.

In order to discuss the procedure and aims of the G.A.N.E.S.H programme, community health workers of KS and RKM arranged meetings with ASHAs and ANMs in each block. ANMs are village-based female health workers establishing the first contact between the community and health services; in remote areas, ANMs also conduct home deliveries;15 usually each ANM is supported by four to five ASHAs. ASHAs are local women trained as health educators and appointed by the Ministry of Health and Family Welfare of India. Their tasks include trying to persuade women to give birth in hospitals, taking children to immunisation clinics, encouraging family planning, treating basic illnesses and injuries with first aid and improving village sanitation.15

Each community health worker involved in the project is linked with one ANM and will be informed by them about each live birth in her village. Every Wednesday and Saturday, the government offers free services in primary health centres in our target areas, where ANMs and ASHAs check the infants’ weight and growth, promote and assist with breastfeeding and other nutritional aspects, and vaccinate. Community health workers linked to G.A.N.E.S.H. can also recruit mothers with newborn and young infants there. The community health worker approaches the parent(s) within 4 weeks after delivery and explains the G.A.N.E.S.H. project in detail, including, for example, mutual commitments. He/she hands out leaflets with relevant information provided in words (Hindi) and pictures. Parents/caregivers are assured of confidentiality and anonymity in reports and publications generated as part of G.A.N.E.S.H. Participation is voluntary, and families are assured that refusing to participate will not affect their access to care. If parents agree to participate, their (written) consent is obtained. Illiterate parents are informed in the presence of a literate witness and consent with their finger-print.

Participants

The birth rate in the target area varies from 19.9 to 28.7 per 1000.2 With a population of 50 000, we expect to see 1000–1500 newborns per year. Apart from living in the target area, inclusion criteria comprise all infants younger than completed 5 months of age irrespective of gender, family background, medical history and current health status. Infants older than 5 months of age are excluded.

Community health workers will collect demographic (please see also the Variables section), including the household composition and living condition, data from the mothers. Besides the medical history during pregnancy and delivery, maternal age and weight will also be documented.

The G.A.N.E.S.H. project is perceived by the communities as part of the ongoing services, which have been provided by KS and RKM for decades. More than 90% of newborn infants are brought to the free services in the primary health centres mentioned above, or, alternatively, the community health workers are welcomed at the family home. As part of the Indian culture, after having given birth, young mothers may move to the grandmother’s home, possibly in another district. As the majority of the maternal grandparents also live in the same region, the number of moves to other regions is extremely small. Altogether, the number of families not willing to participate in the G.A.N.E.S.H. project is negligibly small.

Variables

After the recruitment of the infant, a community health worker collects the following data (ie, exposure variables): demographical data (infant’s gender, infant’s age in days, parents’ age, marital status, number of children, family member with disability, living status and household composition, religious affiliation and caste); maternal medical condition, especially during pregnancy and delivery, according to the documentation of antenatal care and/or the medical history collected during institutional delivery; maternal nutritional status (weight and height); access to and frequency of antenatal care; site of delivery (home vs institutional/hospital) indicating safety of child birth; complications during labour (documented in the medical history of the delivery) including intrapartum-related complications such as difficulties to breathe and/or delayed cry at birth; the infant’s gestational age at birth and birth weight. Frequency of antenatal care, age, weight, height, number of children, parental age and maternal weight are continuous data; all other data are categorical.

At 1–5 months, we assess the following variables: (1) GMs: from birth to the end of the second month post-term age, we assess whether GMs are normal, poor repertoire or cramped-synchronised11 (categorical data). From the beginning of the third month to the end of the fifth month post-term age, we assess fidgety movements as normal, abnormal or absent16 (categorical data) and (2) the motor optimality score17 (MOS; ordinal data) and its five subscores (ordinal data) as well as the number of normal and abnormal movement and postural patterns (continuous data); (3) clinical data include status of breastfeeding and the infant’s medical condition since birth such as fever, diarrhoea (categorical data); (4) weight (measured with a portable weighing scale), length (measured on an infantometer) and head circumference (measured with a tape measure extending from the middle of the forehead to the farthest part in the rear of the head) are given as Z-scores, weight-for-age Z-scores and weight-for-length Z-scores (continuous data).

The outcome at 12–24 months is either normal or classified as neurological and developmental disorders including cerebral palsy, epilepsy (orthopaedic) congenital malformations, muscular dystrophy and spinal muscular atrophy, developmental delay, genetic disorders, visual impairment, hearing impairment, or a high risk for autism spectrum disorder (without the list being exhaustive; categorical data). The scores of the Hammersmith Infant Neurological Examination (HINE)12 are ordinal data. In addition, each child’s nutritional status is assessed and categorised as normal, underweight, wasted (weight-for-age Z-scores that are more than 2 SD below the median of the WHO child growth reference standards) or stunted (with height-for-age Z-scores that are more than 2 SD below the median of the WHO child growth reference standards). Head circumference Z-scores (continuous data) will also be documented.

Data sources and measurement

All variables described in the first paragraph of the section ‘Variables’ are obtained by a community health worker abiding by a predefined interaction protocol and, if available, from the medical records of delivery.

Motor behaviour at 1–5 months is assessed by means of the Prechtl GMA,11 which is widely regarded as the best clinical tool to predict cerebral palsy in infants under 5 months18 and has proved invaluable for contributing to identify children at risk for developmental delay,11 cognitive dysfunction,18 intellectual disabilities,19 20 speech and language disorders21 and autism spectrum disorders.22 GMA is a diagnostic tool for an early assessment of the integrity of the young nervous system based on observation of young infants’ spontaneous movements. It is entirely non-intrusive and extremely time-efficient and cost-efficient. For a reliable assessment (kappa values ranging from 0.88 to 0.92),11 17 a 2–5-min video of the infant is recorded following a number of basic principles (infant lying in supine position, comfortably dressed, not crying).11 Performed by trained assessors, the summary estimates for sensitivity and specificity of GMA are 0.98 (95% CI: 0.74 to 1.00) and 0.91 (95% CI: 0.83 to 0.93), respectively.23 GMA is ideal for the neurological assessment of young infants, especially in low- and middle-income countries that struggle to provide expensive medical equipment for diagnoses.24 From birth to the end of the second month post-term age, normal general movements (GMs) include a fluctuating sequence of arm, neck, trunk and leg movements. They wax and wane with changing intensity, speed and range of motion. Their onset and ending is typically gradual and smooth. Changing of direction and rotations across the trunk and limb axes are fluent and elegant.11 From 3 to 5 months, GMs appear as fidgety movements with a very small amplitude and moderate speed of shoulders, wrists, hips and ankles in all directions and of variable acceleration.11 Most infants will develop normally if fidgety movements are present and normal, especially if they co-occur with other smooth and fluent age-specific movement patterns.11 16 18 The neural mechanisms of GMs are specific neural networks, central pattern generators (CPG), which are most probably located in the brain stem. We believe that the CPGs for GMs are constantly modulated by more rostral parts of the brain. The resulting variable movements in turn modulate the CPG-activity by variable sensory feedback.11 18 If a structural impairment of the young nervous system becomes functional, supraspinal projections are impaired and reduce the modulation of CPG activity, resulting in less variable GMs.18 Before 3 months of age, abnormal GMs are described as: (1) ‘poor repertoire’ (the sequence of movements and the intensity, speed and range of motion lack variability) or (2) ‘cramped-synchronised’ (the limb and trunk muscles appear rigid and contract almost simultaneously, then relax almost simultaneously).11 As a sign of functional impairment, fidgety movements can be (3) ‘abnormal’ (exaggerated in amplitude and speed), (4) sporadically present (not exceeding 3 seconds25) or (5) missing altogether at a post-term age of 9–16 weeks (6) and (7) are summarised as ‘absent fidgety movements’.11 16 26 Summary estimates of 0.98 for sensitivity and 0.91 for specificity make cramped-synchronised GMs followed by absent fidgety movements very reliable early markers of spastic cerebral palsy.16 17 23 25

The motor repertoire of infants aged 3–5 months consists of fidgety movements as well as other movement patterns such as swiping, wiggling-oscillating limb movements, kicking, arm and leg movements to the midline, fiddling, leg lifts or arching.17 27The detailed GMA for 3-month to 5-month-old infants comprises the following five subcategories: (1) temporal organisation and quality of fidgety movements; (2) quality of other observed movement patterns; (3) age-adequate movement repertoire; (4) postural patterns and (5) movement character. Adding the scores of the five subcategories reveals the MOS11 17 with a maximum of 28 (ie, best possible performance) and a minimum of 5. An MOS ranging from 25 to 28 is considered to be optimal; scores from 20 to 24 are mildly reduced and an MOS below 20 requires intervention. A score below 9 indicates a very high risk for neurodevelopmental disabilities, especially for non-ambulatory cerebral palsy.17 Various studies revealed that a reduced MOS was associated with later motor,17 28–30 cognitive29 31 and language21 29 impairments. A recent worldwide study on 468 children later diagnosed with cerebral palsy shows that such a detailed assessment of early movements and postures provides insight into a child’s later functional capabilities and impairment.17 Intraobserver reliability is high with intraclass correlation coefficients (ICC) ranging from 0.80 to 0.9817 25 and Cohen Kappa ranging from 0.87 to 0.91.30

In February and November 2018, KS and RKM staff involved in G.A.N.E.S.H. (medical doctors, physiotherapists, medical social workers, community-based rehabilitation field workers, special educators and one psychologist) received training from CE. They are now GM Trust-certified (basic and advanced level; www.general-movements-trust.info) to perform GMA. As a side note, for better readability, we have summarised the professions of medical social workers, community-based rehabilitation field workers and special educators under the term ‘community health workers’ throughout the manuscript.

After parental consent has been received, a GMA-trained KS/RKM staff member videotapes the infant’s GMs according to GMA standards (the infant shows 2–5 min of active wakefulness, must not be crying, fussy or sucking on a pacifier, and must be lying in supine position without manipulation).11 Video-recordings are preferably at two points in time: (1) during the first 2 months of life; and (2) during the fidgety movement period between the end of the second month and the end of the fifth month post-term age (corrected for preterm delivery). Infants are videotaped in the parents’ home, in health camps, or in a primary health centre. The initial evaluation is performed by the GMA-certified staff member who has recorded the video, the second evaluation is performed by the first author (MT) and a third evaluation by CE, who is not familiar with the medical history of the infant. CE supervises the GMA as licensed tutor of the GM Trust. In a pilot sampling approach, we recorded and assessed 400 infants revealing ICCs ranging from 0.85 to 0.92 between the primary and secondary assessors and the supervisor.

The neurological assessment at 12–24 months follows the protocol of the HINE.12 This method is easily performed and was designed for examining children between 2 and 24 months of age. It includes 26 items that assess cranial nerve function, posture, quality and quantity of movements, muscle tone and reflexes and reactions. Each item is scored from 0 (minimum) to 3 (maximum score) revealing an overall global score from 0 to 78 (best neurological performance).12 Both the sensitivity and the specificity are around 0.90.32

The TDSC13 is based on 17 test items from the Bayley Scales of Infant Development (Baroda Norms). It was validated both at hospital and community levels and has a sensitivity of 0.85 and a specificity of 0.91, which makes it a simple, reliable screening tool that can be applied by community health workers.33 The test-giver draws a vertical line through the chart indicating the actual age of the child. If the child can complete items that are to the left of the line, then there is no delay for that item. If an item lies to the left of the line and the child cannot complete this item, then an item delay is assumed. In this case, a trained assessor at KS will administrate the Bayley Scales of Infant and Toddler Development, Third Edition (cognition, language, motor, social–emotional and adaptive behaviour domains) with the child. Bayley-III scores exceeding 85 are considered to be normal. In addition, two psychologists (one at KS and one at RKM) were trained to identify sociocommunicative deficits which, along with repetitive and restricted behaviours, may indicate an increased likelihood of developing autism spectrum disorder.

A potential confounder especially after weaning from breastfeeding (usually after 6 months of age) might be undernutrition. It can be directly attributed to inadequate dietary intake, infection or disease that may affect the child. Lack of sanitation and hygiene, inadequate care, maternal mental health, economic deprivation and food insecurity might also be factors influencing the child’s development. Undernutrition and other external disadvantages hampering thriving will be revealed and documented at all assessment appointments.

Intervention programmes

The infant is referred to a targeted intervention programme if at least one of the following criteria is observed: (1) cramped-synchronised GMs during the first 2 months; in case of abnormal poor repertoire GMs during the first 2 months, GMA is repeated at 3–5 months, which will determine the further procedure; (2) absence of fidgety movements or exaggerated, abnormal fidgety movements at 3–5 months; (3) MOS <20; (4) asymmetries of any kind, especially asymmetric segmental wrist and/or finger movements;17 (5) body weight and/or head circumference at or below the third age-related percentile on the WHO charts; (6) facial dysmorphism and/or malformations such as congenital talipes equinovarus, cleft palate; (7) acute illness and (8) any other parental concern.

Infants at high-risk for neurological and developmental disorders and disabilities and infants who need medical examination are referred to a paediatrician at RKM (SV) and/or a neurologist at KS (MT), who have worked together closely for more than 15 years. A second paediatrician is volunteering to assess infants and children in paediatric camps. All three of them are GMA-trained. KS has a special Medical Fund to guarantee that infants in need receive further diagnostic examination such as EEG, X-ray, MRI or even genetic testing. If indicated, physio-therapist, occupational-therapist, speech/language therapists and special educators treat infants individually at home or at the Parents and Child Care Unit of KS. Since G.A.N.E.S.H. has been launched, every staff member involved in interventions has received regular workshops and practical demonstrations of early intervention techniques.

Trained RKM and KS personnel also offers education and counselling on basic health, nutrition and hygiene. Severely anaemic or malnourished mothers and underweight infants are given supplementary nutritious food, vitamins and micronutrients (especially calcium and vitamin D, iron and folic acid tablets) as well as medical care. Infants who need intensive medical treatment are referred to the Ramakrishna Hospital or the district hospitals of Varanasi and Mirzapur. Families in need are invited early on to attend paediatric or rehabilitation camps providing free services, examinations and, if necessary, additional medication.

Bias

Measurement biases may occur because blinding is not possible: it has taken the medical doctors and staff members 15 years to win the participating population’s trust, which is the basis for conducting this study. To avoid the interviewer bias, the community health worker will follow a predefined interaction protocol. In case the medical record of pregnancy and delivery is not available (lack of antenatal care or home delivery), a potential recall bias of the mother is hard to be ruled out. We will be able to control the confounders (eg, undernutrition), as we will document any recognisable external disadvantages hampering thriving, and compare the Z-values of the anthropometric measurements at birth, 1–5 months and 12–24 months.

Study size

Taken an alpha cut-off of 5% and a beta cut-off of 20%, the sample shall involve 1261 children (ie, with a pooled incidence of 9.2% neurodevelopmental disorders in Indian children younger than 6 years,4 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 1000,2 that is, we expect to see 1000–1500 newborns per year, resulting in about 70–105 children with neurological and developmental disorders and disabilities) to achieve a predictive power of 0.95 and an effect-size of 0.15 using Cohen’s f². According to our eligibility criteria, we aimed to assess a cohort of at least 2000 infants born in the first and second years of the study. Until 15 March 2020 (13 months after launching the study, before data collection was deferred due to the COVID-19 pandemic), we have successfully recruited 1300 infants.

Statistical methods

Data will be analysed according to the research objectives of the study using IBM SPSS Statistics in its updated version with significance set at p<0.05. A lead biostatistician of the University Medical Center Goettingen, Germany, will provide expert input for the analyses.

Initial descriptive analysis will explore the characteristics and distributions of the variables. Any outlying data will be identified and the missing data explored. Histograms, boxplots and scatter plots, tables of means and SD or medians, interquartile ranges and cross tabulations will be used as appropriate.

Sensitivity, specificity and positive and negative predictive values will be determined with 95% CIs. This will be based on the positive/negative ‘high risk of neurological and developmental disorders’ status on GMs and MOS (at 3–5 months) compared with confirmed or suspected neurological and developmental disorders/no neurological and developmental disorders diagnoses (at 12–24 months). The association between the MOS (at 3–5 months) and the HINE (at 12–24 months) will be examined using Spearman’s rank-order correlations. To test the predictive power of the MOS-subcategories (polytomous ordinal predictors) on the HINE cut-off outcomes (polytomous ordinal criterion), we will run ordinal regression analysis. The predictive power of abnormal movement and postural patterns (metric predictors) on HINE cut-off outcomes (polytomous ordinal criterion) will be again estimated by means of ordinal regression analysis.

We will perform regression analyses to determine how exposure variables are related to GMs and MOS. General linear model analyses will be performed to understand the effect and interactions of predefined variables and covariates on GMs. In case of multiple subgroup comparisons, we will divide the significant level of 0.05 by the total number of subgroups of the analysis (Bonferroni correction).

GMs and MOS will be rated by at least one community health worker, the first (MT) and the third author (CE). Fleiss’ kappa will be calculated to assess the agreement among the three scorers when assigning categorical ratings to GMs. ICC statistics will be applied to examine the agreement of the MOS among the scorers.

Longer-term outcomes

The listed longer-term outcomes are those we expect to achieve in a follow-up evaluation period of 3–5 years, depending on the availability of funds. Details will be determined later, as they can change according to requirements. Long-term outcomes should reflect an improved child health and nutritional status, improved motor, cognitive and language abilities, a better supply of vision and hearing aids and orthoses, proper management in case of epilepsy, access to MRI, EEG, X-ray and other diagnostic procedures, as well as adequate psychological treatment and counselling—in other words: a proper management of neurological and developmental disorders and disabilities and a strengthened health system.

Patient and public involvement

It has not been appropriate to involve infants and their families or the public in the design, or conduct, or reporting or dissemination plans of G.A.N.E.S.H.

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).

After parents/caregivers have received detailed information on G.A.N.E.S.H., they are assured of confidentiality and anonymity in reports and publications generated as part of G.A.N.E.S.H. Participation is voluntary, and families are assured that refusing to participate will not affect their access to care. If parents agree to participate, their (written) consent is obtained. Illiterate parents are informed in the presence of a literate witness and consent with their fingerprint. All study data are stored in secure databases. Paper files are kept in locked filing cabinets, and computer and video files are stored on password-protected servers. Only staff members associated with the G.A.N.E.S.H. programme have access to the data.

Results are submitted to peer-reviewed journals and presented at national and international conferences. Beyond that we aim for G.A.N.E.S.H. to serve as a model for early identification of and intervention in children with (or a high-risk for) neurological and developmental disorders and disabilities to have a sustainably impact on healthcare plans for low-income settings. Therefore, dissemination to healthcare professionals and policy makers is crucial.

All results presented are group data; individual participants are not identifiable. We intend to analyse and publish GMA and MOS results as well as 12-month to 24-month outcomes as standalone research articles in peer-reviewed journals. In accordance with our main aim to improve the quality and quantity of interactions with beneficiary families, the main research questions (in chronological order) to answer and discuss are as follows:

(1) Is it feasible for trained community health workers to launch G.A.N.E.S.H. and apply GMA in a rural, low-income setting? Key aspects: success of training, acceptance by ASHAs and ANMs, acceptance in the community, challenges, interscorer agreement, frequent staff meetings, expert supervision. Limitations/risks: parental refusal, cultural setting in case of male community health worker, migration of trained staff, priority of nutritional aspects over GMA.

(2) Does G.A.N.E.S.H. improve the frequency of home visits for infants and their mothers and the extent and level of counselling for families in need, and does it help to identify abnormal motor behaviour and malnutrition? Key aspects: identification of abnormal GMs and reduced MOS, parental compliance with further paediatric and neurological examination, launching of targeted family-oriented intervention, targeted health education, provision of food supplementation and micronutrients. Limitations/risks: no compliance with further assessment and intervention because early markers are subtle; families see the need to consult a doctor only in case of (frequent) seizures, acute infection or malformations like cleft palate and congenital talipes equinovarus. The need for prevention and early intervention is difficult to grasp for parents, and is often impeded by cultural hierarchical family structure.

(3) Are we able to identify early on (especially at 3–5 months) children who will have an adverse outcome around 12–24 months of age? Key aspects: predictive value of GMA, sustainable family acceptance and compliance with targeted strategies of early intervention. Limitations/risks: the nutritional status, especially after weaning from breastfeeding (second half of the first year of life) might have a greater impact on the development than the integrity of the nervous system determined by GMA.