One evaluation of early childhood developmental status in low- and middle-income countries (LMICs) estimates that 15.7% of children are significantly delayed in their cognitive development, 26.3% in socioemotional development and 36.8% in either or both (D. C. McCoy, personal communication). Stunting, low wealth and living in a rural area are significantly associated with neurodevelopmental delay; most of the children live in Africa and eastern Asia. Fortunately, neurodevelopmental science is benefitting from rapidly expanding technologies for the integration of the environmental (for example, infectious disease, nutritional and carer quality), brain-related (for example, developmental neuroscience and brain imaging) and genetic (for example, epigenetic modelling and genomic big data) domains that drive neurodevelopment. Figure 1 illustrates the mutually interactive nature of these three developmental domains, along with the current strategic areas of research at the environment–brain–gene interface (Box 1).

Figure 1: Mutually interactive domains — environment, gene and brain — interact in terms of environment–gene socioevolutionary processes (epigenetic), environment–brain moment-by-moment neurocognition (neuropsychology), and gene–brain universal brain and behavioural processes in child neurodevelopment.
figure 1

The foundation for this multi-level interaction is brain plasticity as shaped by risk and resilience in child neurodevelopment, which occurs at the evolutionary (physical environment), cultural (social environment), individual (brain) and neuronal genotype (genetic) levels. The most strategic points of research opportunity as presented in this Review occur at the intersections between brain, gene and environment.

Advances in developmental science have triggered a reconceptualization of neurodevelopment based on the recognition that developmental processes are a part of child health in the broader context of communicable and non-communicable disease1. The developmental origins of the health and disease hypothesis proposes that the physiological processes of developmental plasticity operate in early childhood, but have the potential for adverse consequences in later life2. Consequently, childhood — particularly early childhood — is a high-priority target for both preventive and remediating interventions to address the pervasive developmental needs in LMICs (D. C. McCoy, personal communication).

In this Review, we describe several high-impact findings that have emerged from research in low-resource settings that pertain to the developmental milieu of the child, its relationship to the brain and behavioural neurodevelopmental integrity of the child (neurodevelopmental disorders), and the genetic and epigenetic underpinnings that can drive this relationship. As we review key scientific advances in each of these three domains, we propose strategic areas of ongoing and future research that could provide innovative models to fuel significant advances and evidence-based interventions for meeting the developmental needs of children. We conclude by summarizing ways in which this model (environment, brain and gene) provides rich opportunities for a more global approach to child-development science, making it possible to achieve the UNICEF mandate of full child health and development for all3,4.

Approaches to malaria

In 2013 there were around 198 million cases of malaria of which 584,000 resulted in death5. A child dies from malaria every minute, and one in four survivors present with significant neurodevelopmental impairment6,7. However, cognitive rehabilitation, speech and physical therapy, and carer-training interventions can improve cognitive performance and behaviour of treated mother–child pairs8,9. As rehabilitation approaches are evaluated, there is mounting evidence of the neurocognitive benefits of computerized cognitive rehabilitation training (CCRT) in African children with a brain injury as a result of severe malaria and in those with HIV-related brain injury10. Dissemination and implementation science must now inform innovative approaches to bring such interventions to scale in low-resource communities. Mobile network health (mHealth) research opportunities are a high priority, given the ever-increasing access that children and adolescents in low-resource settings have to mobile-based internet and computing technologies11. Another key strategic research opportunity is to evaluate the impact of neurocognitive rehabilitation interventions such as CCRT, on the enhancement of brain-development neuroprotective factors12,13 (Box 2). We can then evaluate the extent to which such brain-based biomarkers mediate the neuropsychological benefits of CCRT, along with how neurocognitive rehabilitative interventions diminish biomarkers of brain inflammation (such as tumour necrosis factor-α (TNF-α) and creatinine).

Approaches to paediatric HIV

Globally, roughly 3.4 million children live with HIV infection and are at high risk of significant neurodevelopmental disabilities. Of these, almost 90% live in Africa where only 24% of infected children have access to anti-retroviral (ARV) treatment14. These children's environmental risk factors are compounded by poor nutrition owing to protein and specific micronutrient deficiencies15. They also often have parasitic, respiratory and enteric diarrheal infections16. Such compounded risk exists whether a child is infected with HIV (proximal risk) or lives in a household or community where HIV has a persisting and significant disruptive impact (distal risk)17. Multifaceted risk for all kinds of early developmental insults (for example, infection, malnutrition and poverty) demands that children in low-resource communities need a comprehensive package of assessments and interventions to holistically enhance their development16.

Recent epigenetic evidence suggests that chronic poverty may 'shrink' children's brains over successive generations as documented by longitudinal multigeneration brain-imaging research18. These considerations justify the junction between the environment and the brain as a highly strategic point of intervention. This is further illustrated by the strategic importance of implementation-science research in designing an effective comprehensive package of services for antenatal and postnatal care. This is evident when considering at-risk adolescent mothers in LMICs and the heightened risk of neurodisability in their infants. Pregnancy in adolescence is associated with premature delivery, stillbirth, fetal distress, birth asphyxia, low birth weight and miscarriage19. Furthermore, if the adolescent mother is also suffering from malnutrition these risks are compounded for the infant — long-term effects as a consequence of low birth weight include stunting, poor neurodevelopmental outcomes, and increased susceptibility to cardiovascular and metabolic diseases such as obesity and diabetes20. In fact, there is evidence that environmental factors such as nutrition can alter epigenetic modifications and thus play a part in the development of these disorders later in life21. The maternal microbiome is also important to infant health outcomes, including the risk of pre-term birth, the development of gastrointestinal diseases such as irritable bowel syndrome, and the development of the immune system22.

One of the greatest public health initiatives developed in the modern era of infectious disease is the prevention of mother-to-child transmission (PMTCT) of HIV. These interventions have reduced perinatal infection of children born to infected mothers from more than 30% to less than 1%23. However, there are still gestational neurodevelopmental risks associated with early exposure to ARVs24. One of the most exciting developments in the treatment of HIV has been the development and anti-retroviral characterization of VRC01. This is a potent and broadly neutralizing anti-HIV monoclonal antibody that prevents HIV-1 transmission from plasmacytoid dendritic cells to CD4 T lymphocytes25. Once proven safe for infants, such therapies should be administered as soon as possible after the diagnosis of HIV in infants, and the long-term neurodevelopmental and neurocognitive protective benefits of such innovative treatment strategies should be evaluated. These therapies could also be effective in the prevention of HIV transmission.

Trauma-associated psychiatric illness

Another strategic research opportunity is to further evaluate how maternal depression is associated with widespread changes in DNA methylation in their offspring26,27. Such epigenetic processes can result in heightened risk of depression and anxiety disorders in children as they become adults28. How best to package and bring to scale a strategic set of intervention services that address this remains a neglected area in high-impact implementation science. Likewise, populations traumatized through conflict and genocide can pass on psychiatric disorders transgenerationally. This may be partly mediated by the hormonal effects of maternal stress on neuropsychiatric risk for children in utero in regions where women have been traumatized through sexual violence in conflict zones (glucocorticoid-mediated inducement of cytokine inflammatory responses causing methylation of DNA in children in utero). Such intergenerational epigenetic mechanisms of psychiatric disorders necessitate evidence-based and sustainable community-wide treatment strategies to address these disorders within the foundational mother–child caring fabric of that society29. Task shifting will be a crucial strategy in addressing such community mental health support services30,31. There is evidence to support the effectiveness of a year-long maternal carer training programme for children who are affected by HIV in rural Uganda to facilitate child development in low-resource settings, while remediating maternal depression and enhancing carer functionality8,11.

Nodding syndrome

The beginning of the millennium was marked by the manifestation of the enigmatic condition nodding syndrome, which affects school-age children, and is reported in South Sudan, northern Uganda and southern Tanzania. This condition is characterized by episodes of repetitive nodding (dropping forward of the head) often coupled with seizure-like behaviours (for example, convulsions or staring spells) that occur during attempted feeding32,33. Nodding syndrome is also characterized by stunted brain growth, including significant brain atrophy near the hippocampal and glia matter of the brain and significant cerebellar involvement. This is accompanied by lifelong profound neurodisability, severe behavioural problems and high mortality34.

The nodding is caused by an atonic seizure, but the aetiology of this seizure is unknown, although associations with other developmental conditions have been established. Nodding syndrome is most prevalent in areas with high infection rates of the parasitic worm Onchocerca volvulus — a nematode carried by black fly of the genus Simulium — the bites of which can cause onchocerciasis, a highly prevalent type of blindness caused by infection. Other reports suggest an association between the syndrome and malnutrition35. Future research of this syndrome must focus on understanding the aetiology so that it can be prevented, diagnosed early and treated effectively. Emerging diseases that profoundly affect children, such as nodding syndrome, provide an important opportunity for developing diagnostic, management and intervention techniques adapted to LMICs that, in turn, can be used for the prevention of worldwide outbreaks of diseases that lead to severe disability.

Although nodding disease is highly localized, such enigmatic disorders that arise from time-to-time and result in profound neurodisability are important because they reveal the urgent need to develop scientific models that can be seamlessly integrated into emerging disciplines. These include geographical ecological mapping, maps of parasitology dispersion, genotypic mapping across populations and their geographical dispersions, and geographically mapped epidemiological risk of infectious disease. These multi-layered models must then be integrated with neuropathogenic mechanism models that include sensitive and specific brain inflammatory markers, as well as the corresponding neuropsychological sequelae of such central nervous system inflammatory markers.

Malnutrition and disease

Childhood malnutrition, both through prenatal and perinatal maternal micronutrient deficiencies36, infant micronutrient deficiencies37, and protein–calorie deficiency, imposes a heavy burden on neurodevelopment38,39,40. The primary effects of malnutrition have been associated with elevated mortality, morbidity, and risk of cognitive and socioemotional impairment. Although it has been extensively researched, and interventions have been attempted, malnutrition remains a serious challenge to children's development in LMICs. Efforts have not yet succeeded in eliminating malnutrition or in successfully bringing interventions to scale41. Secondary effects of malnutrition are associated with vulnerability to microbial pathogens that can also severely disrupt neurodevelopment42,43.

Enteric infections

The aetiology of malnutrition is complex. In particular, malnutrition might result from enteric infections of bacteria that are highly prevalent in LMICs, and include both well-known (Escherichia coli, Vibrio cholerae, and species of Salmonella, Shigella and anaerobic streptococci)44 and emerging pathogens (enteroaggregative E. coli, Cryptosporidium and Giardia)45. These infections can significantly affect childhood brain or behavioural development, presumably through damage to the gut microbiota. This can lead to intestinal inflammation that diminishes intestinal absorption, and protein and micronutrient deficiencies compounded by recurring dehydration and malaise46,47. This field of research has also significantly advanced our understanding of the inter-relationships between genetics (for example, neuroprotective APOE polymorphisms), enteric diseases, nutritional malabsorption and neurodevelopment in young children48,49.

An important research opportunity provided by this work involves the clinical evaluation of the neurodevelopmental benefits of micronutrient interventions to enteric disease, including whether glutamine works better than glucose as a key ingredient of oral rehydration and repair therapy (ORRT)50. Glutamate intervention may be more effective in the repair of intestinal barrier functions and hence improve child development as well as the absorption of ARV drugs in children with HIV.

Food-borne neurotoxins and nutritional malabsorption

Konzo disease is a permanent, irreversible, upper-motor neuron disorder, occurring primarily in rural areas of sub-Saharan Africa that are dependent on bitter varieties of cassava (Manihot esculenta; an annual crop cultivated for its edible starchy tuberous root, which is a major source of carbohydrates and, therefore, a food staple). Epidemiological studies have documented konzo outbreaks — mostly in women and children — in periods of food insecurity that have been brought about by drought, displacement by war or conflict, or other factors that have led to the insufficient processing of cassava tubers. The insufficient breakdown of linamarin compounds that contain cyanide result in neurological damage and seem to lead to outbreaks of konzo, which has been documented mostly in the Congo, Central African Republic, Mozambique and Tanzania51,52,53 with a prevalence of between 0.1% and 17% in affected villages54. Studies have recently documented neurocognitive impairments in children with konzo. Furthermore, even children who do not show signs of konzo, but who live in konzo-affected households may have neurocognitive impairment of working memory and learning ability55.

Konzo offers an important opportunity for integrative neurodevelopmental science. Neuroinflammatory markers of brain injury from cyanide toxicity and inflammatory markers of microbiota destruction in the gut from cyanide toxicity need to be mapped on sensitive neurocognitive impairment indicators in children. Konzo offers a rare opportunity to test integrative models of nutritional toxicity in the brain and gut against a backdrop of malnutrition and corresponding micronutrient deficiencies. Gauging their comparative weighting in the mediation of neurodevelopmental disability within the cognitive and neuromotor domains will allow us to determine the effectiveness of prevention and treatment strategies. Since konzo is entirely preventable, health education and promotion intervention methods should be evaluated at the community-wide level in terms of the benefit to disability-adjusted life years56.

Treating hydrocephalus in LMICs

Hydrocephalus, the abnormal accumulation of cerebrospinal fluid in the cerebral ventricles, has multiple causes, and is especially prevalent in LMICs. Failure to treat the condition almost always leads to death or severe neurodevelopmental disability. Higher birth rates and limited perinatal care contribute to a greater burden of care for hydrocephalus in LMICs57 (for example, there are 100,000–250,000 new infant cases of hydrocephalus annually in sub-Saharan Africa alone41). In addition to the expected burden of congenital hydrocephalus in LMICs, climate-driven neonatal ventriculitis of unknown pathogenesis has recently been identified as one of the chief causes of infant hydrocephalus (60% of cases in Uganda)58,59,60,61. In sub-Saharan Africa, rates of neonatal sepsis are estimated to be 170 per 1,000 births, with a corresponding mortality of 10 deaths per 1,000 births62. For survivors with post-infectious hydrocephalus (PIH), neurodevelopmental consequences of the primary brain injury can be devastating even before hydrocephalus develops. One-third of those with PIH remain profoundly disabled at five years, even after successful surgery63. However, innovative surgical techniques have been pioneered and developed in Uganda that have revolutionized the treatment of hydrocephalus worldwide63,64,65.

The standard treatment for hydrocephalus has long been the implantation of tubing that drains cerebrospinal fluid from the ventricles to the peritoneal cavity (ventriculoperitoneal shunt). However, this treatment creates lifelong dependence on an unreliable implanted device that often requires an emergency operation when it fails (40% failure within 2 years of the original implantation)66. An effective, minimally invasive treatment method (endoscopic third ventriculostomy (ETV) combined with endoscopic choroid plexus cauterization (CPC)) that avoids shunt-dependence in most infants was developed in Uganda as an alternative65. The procedure — the safety and efficacy of which were demonstrated initially in LMICs and then in high-income countries65 — combines two techniques. These involve creating a new opening through the floor of the third ventricle and reducing the choroid plexus tissue in the lateral ventricles by cauterization. Building the capacity to achieve universal access to optimal and affordable hydrocephalus treatment for infants in LMICs is an ongoing challenge64. Present efforts involve task shifting by training non-physician medical officers to undertake the shunt placement, allowing neurosurgeons to focus on the more complex third ventriculostomy procedures. Dissemination and implementation research is needed to test the effectiveness of this task-shifting approach.

Genetic studies of neurodevelopment

The genome has a substantial role in the aetiology of neurodevelopmental disorders. These disorders can be classified into six major categories (Box 2). There is abundant evidence that the disorders in all six categories may affect many facets of child development. The disorders in categories 1–3 are typically severe and impose multiple developmental challenges from birth. These conditions are referred to as congenital conditions (their broad definition also includes conditions that result from various challenges in pregnancy, such as severe micronutrient deficiency, for example folate deficiency). Given what is known about the prevalence of these conditions in high-income countries, estimates suggest that at least 7.6 million children are born annually with severe congenital conditions, and that the number is especially high in LMICs67.

The disorders in categories 4–6 include common multifactorial conditions with onset in early childhood. Of note, less than 50% of countries have policies for the control of these conditions67. These conditions currently constitute a substantial health challenge in high-income countries, but are substantially understudied, under diagnosed and underserved in LMICs. Although limited, the relevant research in LMICs unfolds in a number of dimensions, converging around the understanding that economic development and changes in lifestyle have led, or are leading to, a rapid increase in the observed prevalence of these multifactorial disorders. In other words, as people's environment improves, the role and prominence of genetic and genomic factors will increase. Common disorders include conditions that are attributable to epigenetic influences68 (for example, DNA methylation and histone modification).

In this context, two epigenetic mechanisms have been highlighted. The first mechanism connects nutritional challenges to the manifestation of metabolic syndromes. This happens through a causal link between nutrient restrictions in utero and in early childhood, lack of clean water and sanitation, and high levels of infectious organisms in the environment. These can lead to epigenetic changes in pathways related to metabolism, blood pressure and glucose regulation69. The second mechanism is the link between psychological stress and the glucocorticoid-mediated inducer of the cytokine inflammatory response70. Both exemplify the developmental origins of the health-and-disease hypothesis and its relevance to the aetiology of neurodevelopmental disability in LMICs71.

Key research and training priorities related to these six disorder categories are: determining their global prevalence; training scientists in appropriate molecular technologies and sustaining this increased human-resource capacity by providing ongoing support and training to keep up with the rapid technological advances in the field; developing methods for cheap and reliable diagnoses of the widest possible range of congenital conditions and identification of the broadest possible range of risk factors for complex multifactorial disorders; developing practical, accessible and inexpensive procedures for family-planning counselling (preconception and post-delivery); and continuing to build the capacity and infrastructure needed to initiate cutting-edge, relevant research that is comparable with that taking place in high-income countries. These research and training priorities should translate into public health services that can help couples in family planning and the resource mobilization needed to nurture children with such disorders.


We have outlined significant scientific findings and challenges that have emerged from research in LMICs. We have provided strategic research examples and areas of research opportunity (aetiology and intervention) at the junction between the environment, brain and gene. The dynamic interactions among these three domains are at the foundation of brain neurodevelopment in children (Box 1). New technologies are providing ever more sensitive biomarkers that can be related to the brain and behavioural neurodevelopmental integrity of the child. New technologies are also emerging that link the regional and global surveillance of neurodisability to environmental risk, and these can be integrated with the genetic and epigenetic underpinnings that drive this relationship.

Future approaches must accommodate the use of new data gathered by innovative technologies, offering fresh approaches to old problems in child development in LMICs. These new approaches will prove to be especially strategic at the points of interface and integration between the environment, gene and brain (Fig. 1). New models that can effectively integrate these three domains into a comprehensive and cohesive paradigm must have the following hallmarks.

Interdisciplinary approaches

Research in LMICs needs to take into account the complex multifactorial causes of neurodisability (Fig. 1). Environmental risk from natural disasters, social unrest, poverty and infection can offset a child's neurodevelopmental trajectory at the points of interface between gene and brain, during the antenatal and postnatal stages of development. In addition to toxins from the diet such as cassava-based cyanide in konzo, environmental toxins from mining and biomass fuels are leading to levels of exposure that can affect child neurodevelopment across entire communities. These complex systems of developmental risk factors call for comprehensive interdisciplinary approaches to understand the developmental trajectories so as to identify children at serious risk of neurodisability and those in need of intervention in LMICs. For example, maternal health programmes need to work closely with early childhood programmes to ensure an optimal prenatal environment for the developing foetus, improved pregnancy outcomes, and effective parental and community-wide interventions to enhance child development. Another example is the engagement of parents in child-awareness programmes to facilitate their cognitive and socioemotional development. We have cited examples of specific family-based72 and community-based73 interventions that have been successfully used in LMICs. Such interventions involve the integration of anthropology, public health education and promotion, social and media science, and developmental paediatric research.

Employment of new technologies

New technologies are enhancing research approaches in LMICs, presenting enormous potential to transform health-care delivery. The development of new mobile technologies for surveillance, assessment and treatment are particularly needed in LMICs, where mobile phone ownership is rapidly rising. Computerized interventions are already being used for the treatment of children with cerebral malaria and HIV. New and improved surgical techniques will be crucial for saving lives and altering atypical developmental trajectories. The miniaturization of diagnostic technologies that provide data for integrative risk maps, which can be integrated at the population level with genome distributions, will allow for effective population surveillance and public health intervention at a community-wide level.

Implementation research

Research in LMICs cannot be divorced from the health systems and the cultural context in which the populations are situated. Research on the implementation of evidence-based prevention and intervention is needed. Scientifically sound interventions scaled up to the community and national level will require working with governmental and non-governmental partners to ensure sustainability. As already noted, significant advances have been made by 'task shifting' in resource-constrained settings in order to, for example, delegate health-care tasks to health workers with lower qualifications. Such strategies have shown especially promising results in dealing with mental health gaps, but can be effectively applied for the rehabilitative care of neurodisability in children74. Task-shifting strategies, however, need to be evaluated for approaches within dissemination and implementation science. It is only in this manner that we will achieve the UN moral and legal mandate of full childhood neurodevelopment as a basic human right for all.