A high proportion of the children (40.1%) were considered to be at risk for delay based on international standards. This was driven mostly by the literacy and numeracy domain, with only 7% of children considered developmentally on track. This is similar to the most recent published subnational MICS for Madagascar (2012) in which 7.4% of children were on track in this domain, compared to 94.2% in physical, 80.3 in socio-emotional and 83.7% in the learning domain(11). Early literacy in western cultures is associated with better school performance, and early language skills are associated with cognitive development in both low and high resource settings(21–23). Relative poverty has been shown in both wealthy and poor settings to be associated with lower performance on language scales. A study by Fernald et al. in Madagascar demonstrated that poverty relative to others was also associated with lower development, particularly in the language domain(24). Although relative poverty was not associated with lower development in our study after adjusting for other factors including playing with the child and increase in stimulation by adults, this finding aligns with a 2000 study in the US(25) that found that effect of poverty on development was completely mediated by health and home stimulation factors. Poverty’s independent influence on development merits further study.
There may be differences in cultural norms around when the foundations of literacy are laid down. Conversations with parents in Ifanadiana suggest that although physical and socio-emotional development are the focus of attention by caregivers, children are not expected to learn to read until they start public school at age 5. Parents in Ifanadiana often have low literacy themselves, with limited capacity to prioritize and model literacy and numeracy for young children outside of school. And culturally, tools to stimulate development of this domain, such as children’s picture books, are both very rare in Malagasy languages, and very expensive to obtain (personal communication, LRR). Another possible explanation of this phenomenon is that the ECDI literacy/numeracy domain is simply inappropriate in this setting. The numeracy component in particular, asks if children know the name and recognize the symbols of numbers up to 10, which is more advanced for the same age on comparable tools, which assess a child’s ability to simply count to 10. (McCoy). The use of the ECDI tool in the international MICS surveys implies that despite obvious cultural and linguistic differences, all children globally should be measured in a similar fashion. Further research is needed on whether the ECDI literacy domain is relevant in this context.
A high proportion of children in our cohort were stunted—50.5% had moderate or severe and 19.0% were severely stunted. Stunting is so closely associated with developmental risk that it is used as a proxy in international estimates(1). In this setting of very high stunting rates, chronic malnutrition was not statistically significantly associated with developmental risk when controlling for other factors. In an analysis of MICS data from multiple countries, stunting was observed to be a good predictor for some, but not all domains of the ECDI; the relationship between stunting and literacy-numeracy varied by country and high vs. low breast feeding status(26). These results highlight the importance of the PNAN III strategy for reducing malnutrition in these areas, as many cognitive effects of chronic malnutrition may only become apparent as deficits in metacognitive skills (such as memorization, concentration and attention) at a later age when a child starts school and learning may be compromised.
Malnutrition reduction must include not just prevention of new cases but effective treatment of prevalent cases to assist in counteracting growth faltering. In other areas with high rates of stunting, robust malnutrition interventions targeted at smaller areas have been successful at reducing stunting at the population or community level(10, 27). However, important hidden factors contributing to growth-faltering, such as gender roles, age at onset of stunting, breast feeding duration, maternal mental health, dietary diversity and micronutrient availability can vary from site to site; thus formative research in communities should be undertaken to understand how national programs can be adapted on bolstered to have greatest effects.
Adult disengagement was also quite high in this cohort, with almost 25% of children getting no stimulating activities at all from adults. In this setting, engagement by an adult other than the mother or father was quite important. Unfortunately, the standardized questionnaire from the MICS (which we used in this study) does not indicate the specific relationship of the other adult to the child, so we were unable to identify which other adult engaged with the child. Many households in Madagascar as well as in this cohort include multiple generations. The median number of people per household in our cohort was 5, with a range from 1 to 18. Children in Ifanadiana, especially those born to young or unmarried mothers, are frequently raised in a grandparent-headed household. When a mother marries, she may take her young children with her, but often they are left in the household of the maternal grandparents (FR, personal communication), and in some areas of Madagascar, families practice a custom of giving the firstborn child to a grandparent to raise(LRR, personal communication). Grandmothers or aunts, thus, are often responsible for young children, especially those of younger mothers. It is reasonable to suppose that most often, a grandmother is the most deeply engaged “other” adult. Evidence-based interventions that incorporate both nutrition and development stimulation activities could be developed or adapted using community-based participatory research in this setting.
Although only 5% of our population were born to adolescent mothers, being under 20 at the birth of the child was associated with delay compared to both international standards and to peers. Around the world, mothers play a central role in the development of their children. In Madagascar, as in many countries, socialization and discipline of a child until approximately age 5 is generally the purview of the mother or other female relatives (28). Early motherhood is associated in global studies with infant and child mortality, preterm birth and stunting, both of which are associated with reduced child development ((26, 29)). Girl child marriage (defined as marriage before the age of 18) has been associated with off-track development in multiple countries including Madagascar; much of this association is explained by disparities in advanced education and wealth (30). Supporting young women to marry and initiate childbirth later could, thus, have intergenerational benefits. In Madagascar, the legal minimum age of marriage is 18 years, but in rural areas especially, marriage and childbirth often happen earlier for a variety of cultural and economic reasons. This puts both the adolescent girls who become mothers and their babies at increased risk for worse economic, health and development outcomes. Additionally, understanding of both men’s and women’s fertility preferences can be combined with family planning and ready contraception availability in health centers. Such projects are underway in Ifanadiana through the efforts of PIVOT and the MOH.
Factors such as higher number of paternal supportive activities and not being left in the care of another child were associated with a higher proportion of risk for delay. While this may reflect some sort of detrimental or harsh parenting practices on the part of fathers, a more likely possibility is that if a child is more in need, fathers spend more time focused on the child at risk. Fathers in Ifanadiana are expected to provide for the family and provide some care in times of emergency, but development is not usually their purview (FR, personal communication). However, if a child is faltering in some way, the father may do more activities with the child than a child who is not. The same thing could be true of being left alone with another child. When children are being weaned (age 2, frequently because of the advent of another child) even in parent-headed rather than grandparent-headed households, mothers distance themselves and children then become the responsibility of older children in the household (FR, personal communication). This pattern of young children becoming the responsibility of older children around age 2 has been documented in other areas of Madagascar (28) and is common in other settings outside of Madagascar. Children whose development or behavior may be raising red flags in the family may be less likely to be left with other children than those children with development perceived as normal. Surprisingly, having an illness or injury in the last 4 weeks was positively associated with development, which could suggest that active and social children may be more likely to get ill or injured.
This study has several limitations. We conducted a cross-sectional analysis of population-level data to understand the burden of and factors associated with risk of delay in Ifanadiana district. Causal inference cannot be assumed. To the best of our knowledge, no deep ethnographic studies of the population of Ifanadiana exist, so our understanding of the familial dynamics and relationships in Ifanadiana is based on the work of PIVOT’s community health and social work team, which have daily interactions with local families and vulnerable patients. Therefore it is possible that some nuances of relationships have not been well-characterized. The ECDI is a brief population screen meant to provide population level estimates of risk of delay, and not designed to clinically diagnose actual delay or disability in individual children. Also, we rely on self-reported data—the ECDI is administered in an interview to each child’s mother or primary female care-giver, so reports of adult-provided stimulation activities are dependent on the mother’s knowledge, which could lead to misclassification of adult involvement. Nevertheless, the ECDI has been validated on multiple continents and is a standard tool used for a decade in population level surveys to estimate burden of risk of delay, so our findings can be compared to those from Madagascar as a whole and globally. Despite these limitations, our findings contribute to the small body of literature on the burden and risks of developmental delay in low-resource rural settings. Future qualitative explanatory studies including the use of culturally-relevant development screening tools as well as a longitudinal analysis of future waves of the IHOPE cohort data will help to establish directionality of effects and to propose areas of intervention and risk mitigation.