Developmental disabilities among children younger than 5 years in 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016

1Early childhood, commonly defined as the first 5 years of life, is the fastest period of growth and the period in which the developing brain is most sensitive to stimulation and nurturing. This period of development is regarded as the foundation for subsequent educational and vocational attainment at the individual level, and for overall human capital and economic development at the population level. The UN's Millennium Development Goals (MDGs) focused largely on reducing under-5 mortality, especially in low-income and middle-income countries (LMICs). By contrast, the UN's Sustainable Development Goals (SDGs) from 2015 to 2030 envision improvements in the broader health status of children beyond survival., Alongside the general recognition of people with disabilities in several of the SDGs, SDG 4 specifically requires actions to monitor the proportion of children younger than 5 years who are achieving their developmental potential in health, education, and psychosocial wellbeing, disaggregated by disability, age, sex, geographic location, and other characteristics.

2Developmental disabilities are a group of conditions resulting from impairments that affect a child's physical, learning, or behavioural functioning. Affected children typically have sensory impairments (hearing and vision loss), epilepsy or seizures, cerebral palsy, attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD), intellectual disability, or other learning disorders. Children with developmental delays and disabilities are at greater risk of suboptimal health, educational attainment, and wellbeing than are children without such disabilities. However, epidemiological data on developmental disabilities to guide comprehensive health policy engagements at global, regional, and national levels are scarce. The most widely reported data to date suggest that roughly 250 million children are at risk of suboptimal development in LMICs., However, this number is based solely on children thought to be at risk of poor development because of stunting or extreme poverty and does not fully capture children with developmental disabilities. Another study based on UNICEF's Early Childhood Development Index estimated that 80·8 million children aged 3 or 4 years in LMICs had low cognitive or socioemotional development in 2010. That study was limited by the age group considered and the scope of developmental disabilities addressed. The Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) offers an independent source of robust age-specific data on non-fatal health outcomes., We aimed to estimate the prevalence of disability and years lived with disability (YLDs) among children younger than 5 years with developmental disabilities based on findings from GBD 2016. This analysis provides baseline data for monitoring the trends in these metrics over time among children younger than 5 years during the SDG era. Leading causes of specific developmental disabilities are also examined.

3Evidence before this studySince 2007, The Lancet Series on early childhood development has provided estimates of children at risk of suboptimal development in low-income and middle-income countries (LMICs). However, similar to previous estimates, the most recent estimate of 250 million at-risk children, which was derived from 141 countries, is restricted to those who are stunted or exposed to extreme poverty in LMICs and excludes the vast majority of children with developmental delays and disabilities. More comprehensive baseline data that directly describe dimensions of health related to neurocognitive development are required for monitoring the proportion of children younger than 5 years who are at risk of suboptimal development in health, learning, and wellbeing globally as mandated by the UN's Sustainable Development Goals (SDGs). The Global Burden of Diseases, Injuries, and Risk Factors Study 2016 (GBD 2016) produced comprehensive and comparable estimates of age-specific health disorders for 195 countries and territories from 1990 to 2016.Added value of this studyThis study reports GBD 2016 estimates of the prevalence and years lived with disability for developmental disabilities among children younger than 5 years, including epilepsy, intellectual disability, vision loss, hearing loss, autism spectrum disorder (ASD), and attention deficit hyperactivity disorder (ADHD). To our knowledge, this study is the first to investigate trends in the burden of these developmental disabilities between 1990 and 2016 and to provide baseline data for monitoring progress at the global, regional, and country levels under the SDGs.Implications of all the available evidenceAlthough under-5 mortality halved between 1990 and 2016, there has not been a corresponding improvement in non-fatal health outcomes among childhood survivors globally. The number of children younger than 5 years at risk of suboptimal development in LMICs is likely to exceed 350 million (roughly three in every five children), even without inclusion of all known disabilities in GBD 2016. The absence of any systematic attention to developmental disabilities has had greatest effect in sub-Saharan Africa, where the number of affected children increased by more than 70% between 1990 and 2016, despite an overall decrease in prevalence worldwide during this period. The SDGs now present a comprehensive framework for addressing the burden of developmental delays and disabilities among survivors of the leading causes of child mortality in the early years of life, especially in LMICs. More crucially, local health and educational systems should be appropriately equipped to support affected children and their families optimally. Although the prevalence of conditions such as ASD and ADHD typically peak at school age or later, some children will require timely intervention from early childhood. Global investment is needed to improve primary data sources for developmental disabilities to minimise uncertainty around the estimates of non-fatal health outcomes in most countries.

4Evidence before this study

5Since 2007, The Lancet Series on early childhood development has provided estimates of children at risk of suboptimal development in low-income and middle-income countries (LMICs). However, similar to previous estimates, the most recent estimate of 250 million at-risk children, which was derived from 141 countries, is restricted to those who are stunted or exposed to extreme poverty in LMICs and excludes the vast majority of children with developmental delays and disabilities. More comprehensive baseline data that directly describe dimensions of health related to neurocognitive development are required for monitoring the proportion of children younger than 5 years who are at risk of suboptimal development in health, learning, and wellbeing globally as mandated by the UN's Sustainable Development Goals (SDGs). The Global Burden of Diseases, Injuries, and Risk Factors Study 2016 (GBD 2016) produced comprehensive and comparable estimates of age-specific health disorders for 195 countries and territories from 1990 to 2016.

6Added value of this study

7This study reports GBD 2016 estimates of the prevalence and years lived with disability for developmental disabilities among children younger than 5 years, including epilepsy, intellectual disability, vision loss, hearing loss, autism spectrum disorder (ASD), and attention deficit hyperactivity disorder (ADHD). To our knowledge, this study is the first to investigate trends in the burden of these developmental disabilities between 1990 and 2016 and to provide baseline data for monitoring progress at the global, regional, and country levels under the SDGs.

8Implications of all the available evidence

9Although under-5 mortality halved between 1990 and 2016, there has not been a corresponding improvement in non-fatal health outcomes among childhood survivors globally. The number of children younger than 5 years at risk of suboptimal development in LMICs is likely to exceed 350 million (roughly three in every five children), even without inclusion of all known disabilities in GBD 2016. The absence of any systematic attention to developmental disabilities has had greatest effect in sub-Saharan Africa, where the number of affected children increased by more than 70% between 1990 and 2016, despite an overall decrease in prevalence worldwide during this period. The SDGs now present a comprehensive framework for addressing the burden of developmental delays and disabilities among survivors of the leading causes of child mortality in the early years of life, especially in LMICs. More crucially, local health and educational systems should be appropriately equipped to support affected children and their families optimally. Although the prevalence of conditions such as ASD and ADHD typically peak at school age or later, some children will require timely intervention from early childhood. Global investment is needed to improve primary data sources for developmental disabilities to minimise uncertainty around the estimates of non-fatal health outcomes in most countries.

10This study complies with Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER) recommendations. An overview of the general modelling strategies for the non-fatal health outcomes of 328 diseases and conditions, as defined in International Classification of Diseases (ICD) codes 9 and 10, is provided in the appendix. Detailed descriptions of the modelling approaches for each of the four impairments (epilepsy, intellectual disability, hearing loss, and vision loss) and two GBD causes (ASD and ADHD) included in this report, along with specific data, diagnostic, and modelling considerations for each, are also provided in the appendix.

11The causes included in estimation of each of the impairments are shown in the appendix. Although cerebral palsy was not estimated as a separate impairment, estimates of its prevalence and YLDs are included in those of intellectual disability because of overlapping, cause-specific perinatal complications, including neonatal preterm birth complications, neonatal encephalopathy due to asphyxia and birth trauma, neonatal sepsis and other neonatal infections, and kernicterus after neonatal jaundice. We considered all developmental disorders or impairments as proxies for developmental disabilities in this study.

12The two metrics, prevalence and YLDs, associated with developmental disorders in the GBD study are collectively referred to as the burden of disability in this study, without prejudice towards any ethical, sociocultural, or public health constructs of disability in the literature. Prevalence estimates are stated in both absolute numbers and per 100 000 children, in keeping with the practice for all conditions in the GBD study. The term disability is used to describe the perceived short-term or long-term loss of health (and not welfare loss) associated with a condition, which is reflected through an estimated disability weight. Thus, YLDs do not strictly measure disability from a public health perspective based on WHO's International Classification of Functioning, Disability and Health (ICF). Rather, the metric seeks to provide a comparable measure of disease burden across diverse health conditions and impairments.

13Case definitions and diagnostic criteria were based on ICD-9 and ICD-10 codes complemented with relevant guidelines, such as the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV-TR and the Guidelines for Epidemiologic Studies on Epilepsy., Developmental intellectual disability was separated into five bands based on intelligence quotient scores: borderline, mild, moderate, severe, and profound. Hearing and visual impairments were similarly separated into bands of severity corresponding to frequency response and visual acuity cutoffs, respectively. The age at which conditions such as ASD and ADHD could be diagnosed in early childhood was also considered in input data evaluation and model development.

14The GBD estimation strategy is designed to provide a standardised analytical approach for estimating prevalence and YLDs by age, sex, cause, year, and location. The first step in the estimation of each condition was compilation of all available data inputs from systematic reviews of the literature, hospital and claims databases, health surveys, case notification systems, cohort studies, and multinational survey data. All input data for GBD 2016 are available at the Global Health Data Exchange. Effort was made to optimise the comparability of data derived from different sources using different methods, to find a consistent set of estimates across prevalence data, and to generate estimates for locations with sparse or no data by use of available information from other locations combined with covariates. We obtained prevalence estimates by age group from 1990 to 2016 using DisMod-MR 2.1, a statistical modelling, Bayesian meta-regression tool developed for the GBD project. DisMod-MR 2.1 synthesises epidemiological data for non-fatal health outcomes from disparate settings and sources, adjusting for different case definitions or diagnostic criteria and sampling methods to generate internally consistent estimates of prevalence, incidence, remission, and mortality by location, year, age group, and sex. Estimation in DisMod-MR 2.1 occurs sequentially at five levels: global, super-region, region, country, and subnational. Results from higher levels guide analyses at lower geographical levels. Model parameters, input data values, and fit statistics for each component model are viewable in the publicly available Epi Visualisation tool.

15After generating internally consistent estimates of incidence, prevalence, remission, and mortality for each condition, prevalent cases were distributed among a set of mutually exclusive and collectively exhaustive sequelae (ie, severity of the disease state). The complete sequela list for each cause and impairment are provided in their corresponding methods description in the appendix. After initial calculation of sequela-level prevalence, GBD ensured that the sum of all causes leading to each impairment was scaled to equal the total number of cases of the impairment. Because of the commonality of co-occurrence of epilepsy, blindness, and intellectual disability arising from many causes (eg, neonatal complications leading to cerebral palsy, long-term complications of cerebral malaria and meningitis), scaling the prevalence sums of each impairment was done sequentially. After scaling, each sequela was paired with a unique health state describing the associated disability. Each health state had a corresponding disability weight that was estimated with pairwise comparison methods that presented pairs of lay health-state descriptions to more than 60 000 respondents in open web-based surveys done among the general population in nine countries (Bangladesh, Hungary, Indonesia, Italy, Peru, Sweden, Tanzania, the USA, and the Netherlands). Disability weights ranged between 0 (perfect health) and 1 (death), and were assumed to be similar in different locations. By contrast, the distribution of sequelae varied by location, year, sex, and age. Sequela-level prevalence estimates were multiplied by the disability weights to generate YLDs. YLDs for each sequela were then adjusted for comorbidity with a microsimulation framework that assumed that comorbidity within each age group, sex, location, and year was independent.

16All computations in GBD were done 1000 times, each time drawing from the distribution of the sampling error of data inputs, the uncertainty of data corrections for measurement errors, the uncertainty in coefficients from model fit, and the uncertainty of severity distributions and disability weights. Uncertainty intervals (UIs) were defined by the 25th and 975th values of the ordered 1000 estimate values. Changes in estimates between locations or over time that were in the same direction in more than 950 of the 1000 samples were considered as significant. Final estimates for children younger than 5 years were compiled by summing the prevalence and YLD estimates in the first four GBD age groups: early neonatal (0–6 days), late neonatal (7–27 days), post-neonatal (28–364 days), and 1–4 years. The aggregate figures for each of the impairments are independent from one another so the total burden from all developmental impairments is somewhat less than the sum of each impairment because of comorbidities.

17The funder of the study had no role in study design, data collection, data analysis, data interpretation, or the writing of the report. All authors had full access to the data in the study and had final responsibility for the decision to submit for publication.

18The estimated global number of children younger than 5 years in GBD 2016 was 598·5 million in 1990, 600·2 million in 2005, 626·9 million in 2010, and 632·0 million in 2016. The total number of children with any of the six developmental disabilities after adjusting for comorbidity between intellectual disability and ASD was 53·0 million (95% UI 49·0–57·1 [8·9%, 8·2–9·5]) in 1990 compared with 52·9 million (48·7–57·3 [8·4%, 7·7–9·1]) in 2016. Of the children with developmental disabilities in 2016, 2·7 million (5·1%) lived in high-income countries and 50·2 million (94·9%) in LMICs (table 1). About 54% of children with any developmental disability were male, although the proportions of male and female children varied by type of impairment.

19Vision loss was the most prevalent developmental disability and decreased from 1990 to 2016 (figure 1). 26·4 million (95% UI 24·0–29·2) children had vision loss in 1990, corresponding to a prevalence of 4407 (95% UI 4006–4887) per 100 000 children. This figure declined to 25·2 million (22·7–28·4) in 2016, corresponding to a prevalence of 3991 (3595–4487) per 100 000 children. Hearing loss was the second most prevalent disability, affecting 15·0 million (13·0–17·1) children (prevalence 2511 [95% UI 2178–2859] per 100 000) in 1990. Although the prevalence of hearing loss decreased to 2445 (95% UI 2113–2793) per 100 000 children in 2016, the number of affected children increased slightly to 15·5 million (95% UI 13·4–17·7). ADHD was the least prevalent disability during the period, but increased from 835 171 (746 061–946 713) affected children in 1990 to 890 229 (794 104–1 022 157) in 2016. Intellectual disability was associated with the highest YLDs throughout the period, followed by epilepsy, hearing loss, vision loss, ASD, and ADHD (figure 1). Compared with the estimates in 1990, the YLDs for intellectual disability and epilepsy increased in 2016, whereas the YLDs for hearing loss showed a steady but modest decline up to 2016. Generally, the UIs for the YLDs were wider than those for prevalence (figure 1).

20Although the prevalence of developmental disabilities in children younger than 5 years declined from 1990 to 2006 in all regions except North America, the number of affected children increased by 71·3% in sub-Saharan Africa, 7·6% in north Africa and the Middle East, and 5·9% in North America (figure 2, table 1). These increases were offset globally by a decline in the number of children with disabilities in all other regions during this period, with southeast Asia, east Asia, and Oceania recording the largest decline of 34·5%. The six disabilities accounted for 3·8 million (95% UI 2·8–4·9) YLDs in 1990 compared with 3·9 million (2·9–5·2) in 2016 (table 1), representing 13·3% of the total 29·3 million (20·7–39·8) YLDs from all causes in children younger than 5 years in 2016. Sub-Saharan Africa recorded the highest increase in YLDs (91·1%), whereas southeast Asia, east Asia, and Oceania recorded the greatest decrease (−32·8%; figure 2).

21The geographical distribution of developmental disabilities in 2016 based on prevalence estimates is shown in figure 3. Unlike most developmental disabilities, ASD was least prevalent in western Europe and ADHD in south Asia. The highest number of children with intellectual disability (3·9 million [95% UI 3·1–4·7]), hearing loss (5·1 million [4·4–5·8]), and ASD (1·2 million [1·0–1·4]) were recorded in south Asia (table 2), accounting for 31·1%, 32·9%, and 25·4% of the global prevalence, respectively. The highest number of children with epilepsy (1·2 million [0·8–1·7]), vision loss (6·9 million [6·2–7·8]), and ADHD (284 197 [251 986–325 635]) were recorded in sub-Saharan Africa, representing 30·4%, 27·5%, and 31·9% of the global prevalence, respectively. The prevalence per 100 000 children of intellectual disability (2567 [95% UI 2099–3189]), vision loss (4947 [4324–5803]), and ASD (785 [645–948]) were highest in north Africa and the Middle East. Epilepsy, hearing loss, and vision loss were least prevalent in North America, whereas intellectual disability was least prevalent in Latin America. The highest prevalence of children with all disabilities was recorded in south Asia (9792 [8940–10 641] per 100 000) and the lowest in North America (4090 [3735–4467] per 100 000; table 1).

22Country-specific prevalence estimates for each disability are presented in the appendix. For all disabilities, India had the highest number of affected children, except for ADHD, for which the highest number of affected children was in China (table 3). On the African continent, Nigeria, Democratic Republic of the Congo, and Ethiopia were among the top ten countries with the highest number of children with any disability. The USA was one of the top ten countries for intellectual disability, hearing loss, ASD, and ADHD. The highest prevalence of each disability was found in LMICs, apart from ASD and ADHD, which were highest in Sweden and Australia. The top ten countries accounted for more than half of the global prevalence of all developmental disabilities except for ADHD (48·2%; table 3).

23The highest YLDs for all disabilities except for ASD were reported in sub-Saharan Africa (the highest YLDs for ASD were recorded in south Asia; table 2). The lowest YLDs for all developmental disabilities except ASD and ADHD were reported in North America. The lowest YLDs for ASD were reported in western Europe, and for ADHD in central Europe, eastern Europe, and central Asia. However, the highest YLDs per 100 000 population for intellectual disability and vision loss were reported in north Africa and the Middle East, while the highest YLDs per 100 000 population for hearing loss were recorded in south Asia. The top ten countries accounted for more than half of the YLDs for all disabilities except ADHD which amounted to 48·2%. On a regional basis, the highest YLDs per 100 000 population for the six disabilities combined was recorded in north Africa and the Middle East (758 [95% UI 540–1090]) and the lowest in North America (346 [254–451]; table 1).

24Country-specific YLDs for each disability are presented in the appendix. The highest YLDs for all disabilities except ADHD were found in India, while the highest YLDs for ADHD were recorded in China. The highest YLDs for epilepsy, intellectual disability, and vision loss were in Dominican Republic, for hearing loss in Yemen, and for ADHD and ASD in Australia (table 3).

25The underlying causes of the four disorders conceptually modelled as impairments in GBD 2016 are presented in the appendix. Refraction and accommodation disorders (85·2%) were the leading cause of vision loss globally. The major causes of hearing loss were otitis media (57·1%) and congenital anomalies (21·1%). Congenital anomalies (39·7%) and neonatal disorders, including preterm birth complications, infections, and birth asphyxia (21·0%), were the prominent known causes of intellectual disability, whereas idiopathic causes accounted for 29·0%. Neonatal disorders (51·8%) were the leading cause of epilepsy. The extent to which these causes contributed to each impairment varied across the regions and by country.

26About 52·9 million (8·4%) children worldwide had one of the six developmental disabilities investigated by the GBD group in 2016 compared with 53·0 million (8·9%) in 1990. This marginal change is in sharp contrast with the trend in under-5 mortality, which substantially declined from 11 million (95% UI 10·8–11·3) in 1990 to 5 million (4·8–5·2) in 2016. The disproportionately high burden of development disabilities in LMICs has been corroborated by a 2018 systematic review. The increasing burden reflects the absence of any systematic global initiative to curtail this burden among the increasing number of children who are surviving the first 5 years of life. It is likely that improving neonatal survival rates, particularly for those born prematurely, is contributing substantially to this burden, especially during the MDG era.

27As with most health conditions, the dearth of population-based data for developmental disabilities, especially in LMICs, has led to a reliance on statistical estimation of trends in health outcomes to guide health policy and interventions. This practice is underpinned by the view that the absence of evidence is not evidence of absence. In this study, we aimed to distil the best available epidemiological data on the global burden of developmental disabilities among children younger than 5 years from the extensive GBD 2016 database to facilitate an objective assessment of progress in promoting early childhood development under the SDG framework.

28Although a direct comparison with our study is impracticable because of differences in methods, Lu and colleagues estimated a declining trend between 2007 and 2010 in the number of developmentally at-risk children exposed to stunting or extreme poverty in LMICs. When combining data from our study with data from other studies, of early childhood development based on stunting, extreme poverty, and low cognition, we obtained a crude estimate of 350 million for the number of children at risk of not realising their developmental potential in LMICs, or three in every five children younger than 5 years (appendix). This estimate is likely to be an underestimate since we have excluded prominent disabilities such as idiopathic cerebral palsy and other motor disorders from the study. However, given that this estimate includes neither separate estimates of cerebral palsy nor quantification of the role of home life and access to education, there is at best sparse information about the effects of modest growth and nutritional deficits on the long-term neurocognitive development of young children. Our study also suggests that the burden is higher in males than in females for most developmental disabilities.

29Globally, YLDs associated with developmental disabilities have not improved since 1990. The only exception perhaps was the modest decline in vision loss, which might be attributable to effects of the Vision 2020 global initiative, which was launched in 1999 by WHO. Although hearing and vision loss were the most prevalent among the six disabilities from 1990 to 2016, intellectual disability and epilepsy were associated with the highest YLDs in children younger than 5 years. It is possible that rehabilitation with glasses and hearing aids alleviated the effects of impairment in some children with sensory losses, whereas children with intellectual disability and epilepsy experience greater societal stigma and support services are often inadequate in LMIC settings. However, over the entire life course, hearing loss was found to be the third leading cause of YLDs globally in 2016, and the second leading cause if combined with vision loss. Recognising the need for comprehensive early detection and intervention for all children with or at risk of any developmental disability from birth is perhaps more crucial than in later years, given the substantial YLDs associated with these conditions over a life course.

30Summary global estimates often mask significant variations at the regional and national levels. Although minimal decline was observed between 1990 and 2016 globally, the burden of all developmental disabilities substantially increased in sub-Saharan Africa and in north Africa and the Middle East. Moreover, south Asia and sub-Saharan Africa, which frequently account for the highest child mortality globally, had the highest number of children with developmental disabilities, although south Asia showed some modest progress. This finding is consistent with other reports on the risk of suboptimal development among children younger than 5 years.,, For example, regardless of the differences in case definition, India, China, and Nigeria were the three leading countries with the highest number of children with or at risk of developmental disabilities. In 2016, Democratic Republic of the Congo displaced Nigeria from the third position for vision loss. Except for the USA, China, and Brazil, the ten leading countries for developmental disabilities were predominantly from south Asia and sub-Saharan Africa. This pattern of disease burden exemplifies the ethical rationale of the disability-inclusive framework of the SDGs that seeks to promote and ensure safety nets for the survivors of acute childhood illnesses in LMICs to set them on the trajectory of optimal early childhood development.

31Given the variability in population sizes, the countries with the highest prevalence were usually not among the leading countries by actual number of children. Population-based data in some high-income countries, such as the USA, in fact show significant increasing trends in the prevalence of most developmental disabilities due in part to increases in the diagnosis and recognition of developmental disorders such as ASD and ADHD.,, This trend is consistent with the observed increases in the overall burden of developmental disabilities between 1990 and 2016 in North America. Moreover, the prevalence of all disabilities in the USA in GBD 2016 (4·1%, 95% UI 3·8–4·5) is comparable to the 5·2% reported among children younger than 6 years in the study by Houtrow and colleagues. The low prevalence of ADHD that we found globally compared with other disabilities could be partly attributable to the fact that no incidence was assumed before age 2 years in our modelling strategy (appendix). Also, except for severely affected children, the onset of ASD symptoms typically occurs by age 3 years, but may not fully manifest until school age or later. Additional information about the GBD method for identification and classification of ASD, which is applicable to ADHD, can be found in an earlier report by Baxter and colleagues. Although the GBD estimates are quite conservative, our findings underscore the need for intervention in some infants with ASD and ADHD from early childhood.

32Identification of the major causes, as well as the effects, of developmental disabilities should be an urgent priority in regions with the largest prevalence and absolute burden for several reasons.,,, First, most of the countries in those regions lack functional health-care and social-care systems to support children with disabilities. Second, the lifetime costs of supporting children with disabilities are substantial and might further impoverish many poor families and communities. Third, societal stigma and unfavourable cultural beliefs place children at great risk of neglect, violence, or even infanticide. Fourth, educational and vocational opportunities are limited and rarely allow full economic participation and independence of children with developmental disabilities when they transition into adulthood. A range of interventions for children with or at risk of developmental disabilities are well described in the literature,,,, and include primary prevention aimed at reducing the incidence of developmental disabilities, secondary prevention through early detection of disabilities at a time when the brain is still very sensitive to interventions and change, and tertiary prevention through comprehensive community-based rehabilitation programmes (now termed community-based inclusive development), as exemplified by some early childhood development initiatives in high-burden LMICs such as Bangladesh and India., Some disabilities might be less amenable than others to prevention. For example, the GBD study did not investigate the causes of ASD and ADHD, which are often attributed to complex interactions between genetic and environmental factors. However, several risk factors amenable to intervention have been reported in the literature and merit attention and efforts to equip local health and educational systems to provide requisite services for the affected children and their families. The newly launched Nurturing Care Framework for Early Childhood Development by WHO and its partners to facilitate optimal early childhood development in LMICs should necessarily also prioritise actions to address the specific needs of children with or at risk of developmental disabilities under the SDG era, especially in nations with a high burden. A cursory or symbolic reference to developmental disabilities in this framework will not be adequate or effective in galvanising the required attention by policy makers for the affected children and their families.

33The disability-inclusive aspirations of the SDGs agenda are reinforced by the UN's Conventions on the Rights of the Child and the Rights of Persons with Disabilities. Since 2016, the GBD study has introduced a component for monitoring health-related targets and indicators periodically under the SDGs. Presently, 50 health-related indicators that directly involve health services, health outcomes, and risk factors with well established causal links have been identified, of which 37 are monitored in GBD 2016. Children with developmental disabilities are not yet captured within this framework because SDG 4 (to ensure inclusive and equitable quality education and promote lifelong learning opportunities for all) is completely omitted from the considered health-related indicators. Moreover, developmental disabilities have not yet been explicitly linked with SDG 3, which seeks to ensure healthy lives and promote wellbeing at all ages. It is hoped that the growing evidence on health-related causes of developmental disabilities reported in this study, and elsewhere, will facilitate the development of specific and explicit health-related indicators to address the special needs of the affected children.,

34Similar to any modelling endeavour in epidemiology, GBD 2016 has limitations that have been extensively described elsewhere in line with the GATHER reporting guidelines., For example, most of the uncertainty in the YLD estimates are likely to result from limitations in the calculations of disability weights. These uncertainties might be minimised in the future by removing some of the ambiguities in lay descriptions, providing better definitions for disorders in surveys, and increasing the general volume and quality of survey data in this area. Additionally, the non-fatal models, especially for conditions such as ASD and ADHD, continue to rely on sparse data in many regions and exemplify the unique challenges associated with measuring disability in childhood, particularly for more subtle and difficult to diagnose conditions such as ASD and ADHD., Ultimately, a measure of functional limitations consistent with ICF is needed to complement the YLDs for developmental disabilities in childhood.

35Additional limitations of this study deserve mention. First, direct and separate prevalence and YLDs estimates for other disabilities, such as cerebral palsy, communication, and motor disorders, were not provided, which will have resulted in underestimation of the overall number of children with developmental disabilities. There is no evidence to suggest, for example, that there has been any improvement in the global burden of cerebral palsy between 1990 and 2016., Second, age-standardised prevalence estimates and severity patterns of impairments were not reported in this analysis but are freely available at the Global Health Data Exchange for download and interactive visualisation for all the 195 countries and territories in GBD 2016. Third, we did not estimate the incidence of developmental delays and disabilities. Finally, it was difficult to completely and precisely account for children with multiple disabilities and across multiple developmental domains. These overlaps were partially captured by estimation of prevalence of sequelae (most children with ASD also have intellectual disability), ensuring the total number of cases of each impairment equals the sum across all GBD causes, and accounting for the overlap between epilepsy, blindness, and intellectual disability. Therefore, the sum of the impairments and causes is not equal to the total number of children with developmental disability. Some residual overlap is certainly likely, however, especially for hearing and vision loss. We believe that the findings in the GBD study provide valuable baseline data for action and further refinements as more and better primary input sources become available for all developmental disabilities. This viewpoint is without prejudice to other ICF-oriented approaches to quantifying the actual burden of developmental disabilities at the family and societal levels. The annual updating of GBD allows for expedient incorporation of such insights and data as they become available. Additionally, because future GBD iterations will continue to refine the methods, incorporate new data sources, and reanalyse the entire time series, our estimates for 1990–2016 might not be identical to those in subsequent GBD reports.

36In conclusion, although the burden of mortality among children younger than 5 years has been halved between 1990 and 2016, there has been no corresponding improvement in non-fatal health outcomes among children with developmental disabilities globally. This lack of improvement might be attributed to absent or inadequate systematic policies and interventions to address the needs of survivors of childhood illnesses who develop life-long disabilities from early childhood, especially in sub-Saharan Africa. This report underscores the importance of developing explicit health-related indicators for monitoring progress to address the needs and rights of children with or at risk of developmental disabilities within the framework of the disability-inclusive mandates of the SDGs and beyond. More crucially, local health and educational systems should be appropriately equipped to support affected children and their families optimally.