Child and Adolescent Health From 1990 to 2015

Importance Comprehensive and timely monitoring of disease burden in all age groups, including children and adolescents, is essential for improving population health. Objective To quantify and describe levels and trends of mortality and nonfatal health outcomes among children and adolescents from 1990 to 2015 to provide a framework for policy discussion. Evidence Review Cause-specific mortality and nonfatal health outcomes were analyzed for 195 countries and territories by age group, sex, and year from 1990 to 2015 using standardized approaches for data processing and statistical modeling, with subsequent analysis of the findings to describe levels and trends across geography and time among children and adolescents 19 years or younger. A composite indicator of income, education, and fertility was developed (Socio-demographic Index [SDI]) for each geographic unit and year, which evaluates the historical association between SDI and health loss. Findings Global child and adolescent mortality decreased from 14.18 million (95% uncertainty interval [UI], 14.09 million to 14.28 million) deaths in 1990 to 7.26 million (95% UI, 7.14 million to 7.39 million) deaths in 2015, but progress has been unevenly distributed. Countries with a lower SDI had a larger proportion of mortality burden (75%) in 2015 than was the case in 1990 (61%). Most deaths in 2015 occurred in South Asia and sub-Saharan Africa. Global trends were driven by reductions in mortality owing to infectious, nutritional, and neonatal disorders, which in the aggregate led to a relative increase in the importance of noncommunicable diseases and injuries in explaining global disease burden. The absolute burden of disability in children and adolescents increased 4.3% (95% UI, 3.1%-5.6%) from 1990 to 2015, with much of the increase owing to population growth and improved survival for children and adolescents to older ages. Other than infectious conditions, many top causes of disability are associated with long-term sequelae of conditions present at birth (eg, neonatal disorders, congenital birth defects, and hemoglobinopathies) and complications of a variety of infections and nutritional deficiencies. Anemia, developmental intellectual disability, hearing loss, epilepsy, and vision loss are important contributors to childhood disability that can arise from multiple causes. Maternal and reproductive health remains a key cause of disease burden in adolescent females, especially in lower-SDI countries. In low-SDI countries, mortality is the primary driver of health loss for children and adolescents, whereas disability predominates in higher-SDI locations; the specific pattern of epidemiological transition varies across diseases and injuries. Conclusions and Relevance Consistent international attention and investment have led to sustained improvements in causes of health loss among children and adolescents in many countries, although progress has been uneven. The persistence of infectious diseases in some countries, coupled with ongoing epidemiologic transition to injuries and noncommunicable diseases, require all countries to carefully evaluate and implement appropriate strategies to maximize the health of their children and adolescents and for the international community to carefully consider which elements of child and adolescent health should be monitored.

R educing mortality among children younger than 5 years has been a focus of significant international attention for several decades, beginning with the Convention on the Rights of the Child, accelerating during the Millennium Development Goal era, and continuing with the Sustainable Development Goals (SDGs). 1-3 Global progress in reducing death in children younger than 5 years has been substantial, 4 but much less attention has been focused on quantifying and minimizing mortality burden among older children and adolescents. 5 Likewise, nonfatal health outcomes have received comparatively little attention despite the fact that injuries, noncommunicable diseases (NCDs), and acquired chronic conditions with childhood onset profoundly affect long-term health trajectories, future health care needs, intellectual development, and economic and productivity prospects. [6][7][8] High return on investment is expected when evidencebased interventions are implemented to address the health and well-being of children and adolescents. 9 During the past decades, the world experienced rapid economic changes along with declines in fertility and greater longevity in many countries, collectively leading to marked changes in global demographics. 10,11 The identification of successes, unmet needs, and emerging challenges must therefore consider sociodemographic information to contextualize levels and trends of disease burden. 5,12 This information can guide prevention and intervention efforts, tracking and allocation of resources for health and other youth-centric services (eg, education), and monitoring progress for countries at all points on the spectrum of economic development.
Two comprehensive reports on the burden of diseases and injuries in young persons were published following the Global Burden of Diseases, Injuries, and Risk Factors (GBD) 2013 Study. 13,14 The first report covered children and adolescents 19 years or younger; the second described disease burden in young persons aged 10 to 24 years. 15 In the present study-an extension of GBD 2015-we again focus on children and adolescents 19 years or younger, extending the data to 2015 and to 195 countries and territories. We present results separately by sex, describe the epidemiologic factors of several highly disabling conditions that arise from multiple GBD causes, report levels and trends in pregnancy complications among adolescents, and evaluate the association between metrics of disease burden and the Socio-demographic Index (SDI), a composite indicator of development status generated for GBD 2015.
Briefly, we quantified an extensive set of health loss metrics-with corresponding uncertainty intervals (UIs)from 1990 to 2015 for 20 age groups and both sexes in 195 countries and territories. For the present study, we further analyzed levels and trends for children and adolescents 19 years or younger, which includes the first 7 age groups of the GBD 2015 analyses. Health loss metrics in this analysis include all-cause mortality, cause-specific mortality (deaths and years of life lost [YLLs]), nonfatal health outcomes (prevalence and years lived with disability [YLDs]), and total disease burden (disability-adjusted life years [DALYs]). Countries and territories were hierarchically organized into 21 regions and 7 super-regions, which are aggregates of the 21 regions in the GBD location hierarchy. The GBD cause list organizes all diseases and injuries into a 4-level hierarchy. The first level has 3 categories: (1) communicable, maternal, neonatal, and nutritional disorders (group I conditions); (2) NCDs; and (3) injuries. Level 2 of the hierarchy has 21 cause groups, while levels 3 (166 causes) and 4 (261 causes) contain more disaggregated causes and cause groups. The full GBD cause list with corresponding International Classification of Diseases (ICD)-9 and ICD-10 codes is available in previous publications on cause-specific mortality and nonfatal health outcomes. 16,17 Our all-cause and cause-specific mortality analyses used systematic approaches to address data challenges such as variation in both death certification practices and coding schemes, inconsistent age group reporting, and misclassification of human immunodeficiency virus (HIV) or AIDS. Each death was assigned to a single underlying cause. Cause-of-death ensemble modeling was the most widely used statistical tool for estimating cause-specific mortality across GBD 2015. Cause-of-death ensemble modeling uses a train-test-test approach to evaluate a wide range of families of statistical models, maximizing out-of-sample predictive validity of final models. Years of life lost were calculated by multiplying counts of age-specific death and normative life expectancy at the age of death. 16 Analyses of nonfatal health outcomes used detailed epidemiologic data from systematic reviews of the literature, hospital and claims databases, health surveys, case notification systems, cohort studies, and disease-specific registries. DisMod-MR 2.1, a statistical modeling method developed in-house, was the most widely used statistical method in GBD 2015; it is a Bayesian meta-regression tool that synthesizes all available data, adjusting for different case definitions or sampling strategies, to generate internally consis-tent results for prevalence, incidence, remission, and excess mortality in each population. 22 Each most-specific cause was paired with a variable number of mutually exclusive and collectively exhaustive sequelae, which quantify the main outcomes (including asymptomatic states) of diseases and injuries and are the units of analysis for nonfatal health outcomes. Years lived with disability were calculated as the product of sequela-specific prevalence and corresponding GBD disability weights derived from population surveys with more than 60 000 respondents. 23,24 Disability weights were assumed to be invariant by geography, but the distribution of sequelae-and therefore cumulative disability per case-varies by geography, year, sex, and age. Finally, we adjusted for comorbid illness using a microsimulation framework within each population and proportionally adjusting YLDs for each comorbid condition. Disabilityadjusted life years are the sum of YLLs and YLDs. 17 We developed the SDI for GBD 2015, as described previously, to characterize epidemiologic transitions more robustly than is possible with analyses based only on income. 4,[16][17][18][19] The SDI is a composite measure of developmental status as it is associated with health, calculated as the geometric mean of the following 3 indicators: total rate of fertility, log income per capita, and mean years of education among those 15 years or older. Socio-demographic Index scores were scaled from 0 (highest fertility, lowest income, and lowest education) to 1 (highest income, highest education, and lowest fertility), and each geographical unit was assigned an SDI score for each year. We analyzed the average association between SDI score and all-ages rates of YLLs, YLDs, and DALYs for all level 2 and level 3 causes. For comparisons across SDI quintiles, each geographical unit was assigned to a single quintile according to its SDI in 2015 (eFigure 1 in the Supplement).
For all results, 95% UIs were derived from 1000 draws of the posterior distribution at each analytic step and represent the ordinal 25th and 975th draws. Unlike confidence intervals, which capture only sampling error, UIs provide a means of also capturing other sources of uncertainty owing to model specification (eg, parameter selection) and estimation (eg, data adjustments from nonreference categories and β values for covariates). Cumulative and annualized rates of change were calculated on point estimates, and corresponding UIs were derived from the same calculations performed at the draw level.
We present results as both total numbers to illustrate the absolute magnitude of burden, and all-age rates, to compare across geographical areas with differently sized populations. We completed age standardization for ages 19 years or younger for the 10 highest-ranked global causes of death and disability to help compare across populations with different age structures; all other results are presented as total number and all-ages rates only. Results for the global level, along with SDI quintile and region in order of decreasing SDI, are presented in the main article. Results for each country and territory are contained in the Supplement and are available online at http://vizhub.healthdata.org/gbdcompare by age group and sex.

All-Cause Mortality and Cause-Specific Mortality in Children and Adolescents
Total deaths and the age-standardized mortality rate (per 100 000 population) for all causes combined, as well as the 10 largest level 3 causes of death globally, are shown for children and adolescents 19 years or younger in 1990 and 2015 in Table 1. Corresponding country-level results, with uncertainty and cumulative percent change, are in eTable 1 in the Supplement for children and adolescents 19 years or younger and eTable 2 in the Supplement for children and adolescents 5 years or younger. In 2015, there were 7.26 million (95% UI, 7.14 million to 7.39 million) deaths among children and adolescents globally, of which 5.82 million (95% UI, 5.69 million to 5.95 million) occurred among children younger than 5 years, 463 000 (95% UI, 453 000-473 000) among those aged 5 to 9 years, 391 000 (95% UI, 383 000-402 000) among children aged 10 to 14 years, and 588 000 (95% UI, 567 000-610 000) among those aged 15 to 19 years.
As can be seen in Table 1, mortality in children and adolescents 19 years or younger decreased in all SDI quintiles, but inequality increased. Nearly 75% of all deaths among children and adolescents in 2015 occurred in the 2 lowest SDI quintiles (compared with 61% in 1990), while only 1.6% occurred in the highest SDI quintile (compared with 2.1% in 2015). Agestandardized rates of death declined from 1990 to 2015 at similar rates of 55% and 56% in the 2 lowest and highest SDI quintiles, respectively, while they declined by 63% in middle and high-middle SDI quintiles. South Asia accounted for 2.21 million (95% UI, 2.15 million to 2.27 million) child and adolescent deaths, 30.4% of the global total and the most of any region. Next were Western sub-Saharan Africa (1.68 million; 95% UI, 1.61 million to 1.76 million [23.1%]), Eastern sub-Saharan Africa (1.11 million; 95% UI, 1.07 million to 1.14 million [15.3%]), North Africa and the Middle East (529 000; 95% UI, 499 000-562 000 [7.3%]), and central sub-Saharan Africa (463 000; 95% UI, 408 000-524 000 [6.4%]). Geographical patterns of mortality in children younger than 5 years were similar to those in children and adolescents 19 years or younger but with a slightly greater concentration of mortality burden in the 2 lowest SDI quintiles (77% of total). Mortality rates (per 100 000 population) varied from a low of 26.0 (95% UI, in the high-income Asia Pacific region to a high of 666 (95% UI, 638-696) in Western sub-Saharan Africa among all children and adolescents 19 years or younger and from 58.8 (95% UI, in the high-income Asia Pacific region to 2133 (95% UI, 2029-2245) in Western sub-Saharan Africa for children 5 years or younger in 2015.

Cause-Specific Mortality
As seen in Table 1 across the entire age range, rankings were dominated by those affecting the youngest children. Globally, the most common causes of death were neonatal preterm birth complications (mortality rate, 31.4 per 100 000 population; 95% UI, 29.1-34.2 deaths per 100 000 population), lower respiratory tract infections (LRIs) (31.1; 95% UI,  (8.7;95% UI,, and HIV and AIDS (8.1;95% UI,. With the exception of the infectious causes (malaria, diarrheal diseases, and meningitis) each cause was highly ranked in all regions. Rankings of the 25 leading level 3 causes of death among children and adolescents 19 years or younger, disaggregated by sex, are shown in Figure 1. Besides the causes listed above, others ranking in the top 10 in specific regions included hemoglobinopathies and hemolytic anemias (in Western sub-Saharan Africa, where sickle cell disease is the largest level 4 cause of hemoglobinopathies), as well as selected infections (measles, HIV and AIDS, whooping cough, intestinal infectious disease, sexually transmitted infections excluding HIV [ie, congenital syphilis], and encephalitis) and injuries (drowning, road injuries, direct effects of war [ie, collective violence] and natural disasters, exposure to mechanical forces, aspiration of a foreign body, and fire).

Differences in Causes of Death by Geography, Age, and Sex
We found important differences in mortality patterns for each of the 7 component age groups 19 years or younger in 2015 (eFigure 2A-G in the Supplement). During the neonatal period (ie, 6 days or less and 7-27 days), rankings across SDI quintiles and regions were broadly similar; mortality was dominated by neonatal complications, congenital anomalies, and LRIs. Divergence began to appear during the postneonatal period (ie, 28-364 days), when acquired infections such as LRIs, diarrhea, malaria, and meningitis predominated in lower-SDI geographical areas and congenital anomalies and sudden infant death syndrome predominated in higher-SDI geographical areas. Protein-energy malnutrition also emerged as an important cause of death in the postneonatal period in several regions, especially in males, a trend that continued into children aged 1 to 4 years, where it ranked fourth globally in both sexes. Malaria, LRIs, and diarrhea were the 3 highestranked causes of death in children aged 1 to 4 years; because protein-energy malnutrition and other forms of malnutrition raise the mortality risk for each, the effect of malnutrition is even higher than that reflected in results for protein-energy malnutrition alone. Geographic heterogeneity was also observed in other causes of death in children aged 1 to 4 years for both females and males at the global level, including measles (concentrated in the lowest 3 SDI quintiles, particularly Oceania and Southeast Asia), leukemia, road injuries, and drowning (all concentrated in the 3 highest SDI quintiles).
Geographical differences in causes of death in 2015 were more pronounced with increasing age (ie, 5-9 years, 10-14 years, and 15-19 years). Congenital anomalies and cancers (leukemia, brain cancer, and other neoplasms [eg, sarcomas]) were highly ranked in high-SDI regions in all age groups, simultaneously reflecting continued risk of mortality beyond the time of initial diagnosis and lower overall risk of mortality in the population. Intestinal infectious disease was highly ranked globally (second in children aged 5-9 years for both males and females), driven primarily by very large mortality numbers in South Asia and Southeast Asia. Human immunodeficiency virus and AIDS rose to be ranked first globally among children aged 10 to 14 years, driven almost entirely by epidemics in the Caribbean and sub-Saharan Africa. Diarrhea, LRIs, malaria, and protein-energy malnutrition remained important causes of death throughout all age groups but were largely limited except in geographical areas with lower SDIs. Five level 3 causes of maternal mortality-hemorrhage, hypertensive disorders, indirect causes, other direct causes, and the combined category of abortion, ectopic pregnancy, and miscarriage-were in the top 25 causes of maternal mortality globally in females aged 15 to 19 years, reflecting the high burden of maternal mortality among adolescents in the 2 lowest SDI quintiles.
The ranking of injuries as causes of death increased consistently with age and with increasing SDI; all injuries except self-harm ranked higher in males than females. Road injuries were the leading injury-associated cause of death in all age groups, rising to first globally among all causes for both sexes in adolescents aged 15 to 19 years. Drowning was the next highest-ranked cause of injury-associated death in children aged 5 to 9 years (ninth overall among females and sixth among males) and 10 to 14 years (eighth overall in females and third in males), while self-harm (second overall in females and third in males) and interpersonal violence (14th overall in females and second in males) were the next most common injuryassociated causes of death among adolescents aged 15 to 19 years. The direct mortality burden of war was extremely large in North Africa and the Middle East, where it ranked second for each sex among children aged 1 to 4 years and first in all subsequent age groups in 2015.

Leading Causes of Nonfatal Health Outcomes in Children and Adolescents
Total prevalent cases and the age-standardized prevalence rate (per 100 000 population) for all causes combined, as well as the 10 leading level 3 causes with the most YLDs globally, are shown for children and adolescents 19 years or younger in Table 2. Corresponding country-level results for 1990 and 2015, with uncertainty and mean annualized rates of change, are in eTable 3 in the Supplement for children and adolescents 19 years or younger and eTable 4 in the Supplement for children 5 years or younger. In 2015, nonfatal health outcomes caused 154 million (95% UI, 117 million to 196 million) YLDs among children and adolescents, of which 33.3 million (95% UI, 23.5 million to 45.3 million) were in children 5 years or younger, 35.0 million (95% UI, 24.9 million to 47.4 million) in those aged 5 to 9 years, 40.9 million (95% UI, 29.8 million to 54.9 million) in those aged 10 to 14 years, and 44.4 million (95% UI, 32.9 million to 58.0 million) in those aged 15 to 19 years.

Disability Burden From Conditions With Multiple Causes
Many clinical conditions cause significant disease burden in children and adolescents, but because they can arise from multiple causes, their effect is not obvious when examining causes of GBD. Examples that would be in the top 10 global causes of YLDs if considered alone are anemia, developmental intellectual disability, epilepsy, hearing loss, and vision loss. For example, while iron-deficiency anemia was the leading level 3 cause of disability, it accounted for only about two-thirds of total anemia in children and adolescents 19 years or younger in 2015 (eTable 5 in the Supplement), and each case tended to be less severe than other etiologic causes of anemia. Infectious diseases, hemoglobinopathies, malaria, hookworm, gynecologic conditions, and gastritis and duodenitis were other important causes of anemia in children and adolescents. Neonatal disorders were the most common nonidiopathic cause of both developmental intellectual disability (eTable 6 in the Supplement) and epilepsy (eTable 7 in the Supplement). Autism, iodine deficiency, and congenital disorders were important causes of intellectual disability, while much of the rest of intellectual disability and much of nonidiopathic epilepsy were secondary to infectious causes, especially malaria and meningitis. Hearing and vision loss also contributed to the disease burden within children and adolescents 19 years or younger, with age-associated and other hearing loss accounting for most hearing loss burden (eTable 8 in the Supplement). For vision loss, a range of causes contributed to the burden among children and adolescents 19 years or younger, including neonatal disorders and nutritional deficiencies (eTable 9 in the Supplement).

Pregnancy Complications in Adolescents
Mortality was the primary driver of health loss owing to maternal disorders in adolescents. The global maternal mortality ratio per 100 000 live births was 278 (95% UI, 229-339) and 142 (95% UI, 123-166) in 2015 for children and adolescents aged 10 to 14 and 15 to 19 years, respectively, causing 1343 (95% UI, 1105-1640) and 26 855 (95% UI, 23 254-31 521) maternal deaths. Both age groups had a maternal mortality ratio higher than the global aggregate of 132 (95% UI, 117-153) seen in women aged 25 to 29 years (eTable 10 in the Supplement and GBD 2015 maternal mortality publication 18 ). The mean annualized decline in the maternal mortality ratio among adolescents aged 10 to 19 years was only 1.4% (95% UI, 0.8%-2.0%), which was slower than the global improvement rate of 2.6% for overall maternal mortality. Maternal hemorrhage was the highest-ranked level 3 cause of maternal mortality globally, driven largely by its prominence in low-SDI geographical areas where teenage pregnancy and the burden of maternal mortality are the highest (eFigures 3 and 4 in the Supplement). Other top-ranked causes of maternal mortality included maternal hypertensive disorders, other direct maternal disorders (eg, pulmonary embolism, cardiomyopathy, and surgical and anesthetic complications), and the combined category of abortion, ectopic pregnancy, and/or miscarriage. The risk of nonfatal complications during pregnancy is also higher in adolescents than in women in their 20s (eFigure 5 in the Supplement). Abortion, ectopic pregnancy, and/or miscarriage is the most common disabling outcome of pregnancy among adolescents, followed by maternal hemorrhage, maternal hypertensive disorders, maternal sepsis and other maternal infections, and obstructed labor.

Ranking and Trends of DALYs in Children and Adolescents
Ranking of the 25 leading level 3 causes of DALYs in 1990DALYs in , 2005DALYs in , and 2015, along with the changes in total number, all-ages rate, and age-standardized rate, are shown in Figure 2 for children and adolescents 19 years or younger. Corresponding DALY rankings disaggregated by SDI quintile are in eFigure 6A-E in the Supplement. Between 1990 and 2005, more than 40% declines in DALYs in children and adolescents 19 years or younger were seen for LRIs, diarrheal diseases, measles, tetanus, drowning, and neonatal hemolytic disease and other neonatal jaundice; similar declines between 2005 and 2015 were seen for malaria, measles, tetanus, and neonatal hemolytic disease and other neonatal jaundice. The most significant increase was for HIV and AIDS, which increased by close to 600% to rank 11th globally in 2005, a ranking that stayed largely static through 2015 despite a nearly 30% drop in DALYs from 2005 to 2015. Malaria and iron-deficiency anemia were the other group I conditions with significantly increased DALYs between 1990 and 2005; DALYs for both conditions also subsequently decreased significantly by 2015. Several NCDs increased in ranking from 1990 to 2015 for children and adolescents 19 years or younger. Some diseases-including congenital anomalies, asthma, and hemoglobinopathies and hemolytic anemiasincreased in ranking despite registering decreased agestandardized DALY rates for each time period. In contrast, other causes-including sense organ diseases, skin diseases, and mental and substance abuse disorders such as depression, anxiety, and conduct disorder-increased in ranking with largely unchanged or slightly increased global age-standardized DALY rates from 1990 to 2015. Road injuries and drowning were the 2 highest-ranking injuries in terms of DALYs despite significant decreases from 1990 to 2015. Figure 3A shows the mean association between SDI and causespecific YLLs and YLDs from 1990 to 2015 for all level 2 causes. Nonlinearity of associations at times followed the nonlinearity of the SDI itself. There is a clear and substantial downward gradient in child and adolescent health loss with increasing SDI. Years of life lost are the dominant component of DALYs in the geographical areas with the lowest SDI, a trend that continues until an SDI of roughly 0.80, after which YLDs become responsible for a larger proportion of DALYs. There is also a clear increase in the all-ages rate of YLDs in the geographical areas with the highest SDI to the point where, at the highest SDI, 67% of all DALYs are owing to nonfatal health outcomes. Figure   shows the corresponding information displayed as a proportion of total rates of YLL and YLD owing to each level 2 cause at each SDI level. For most level 2 causes, the proportion of all YLLs owing to group I causes decreases with increasing SDI. The exceptions are neonatal disorders and HIV and AIDS and tuberculosis, which increased in relative importance with increasing SDI. In the geographical areas with the highest SDI, self-harm and interpersonal violence, other NCDs, and neoplasms were responsible for an increasing proportion of YLLs. The proportion of YLDs owing to group I causes similarly decreased with increasing SDI, while the proportion owing to NCDs generally increased. Most level 3 causes followed this same pattern, with 2 notable exceptions among the top causes of child and adolescent DALYs: congenital anomalies  1990, 2005, and 2015. Lines connecting the boxes illustrate changes in ranking. Any cause that appears in the top 25 in any year is listed, along with its ranking during each year. Group I causes (infectious, neonatal, nutritional, and maternal) are shown in gray, noncommunicable diseases in red, and injuries in green. Changes in total DALYs are in the first column next to 2005, followed by changes in all-ages rates of DALYs, and age-standardized rates of DALYs. Statistically significant differences appear in bold. HIV indicates human immunodeficiency virus, and STI, sexually transmitted infection.

HIV and AIDS and tuberculosis
A, Expected association between rates of YLL and YLD with SDI for the 21 GBD level 2 causes in children and adolescents 19 years or younger, both sexes, 1990-2015. Each geography is assigned an SDI value for each year, and nonlinear spline regressions are used to find the average relationship between SDI and cause-specific burden rates. B, Expected association between rates of YLL and YLD and SDI for the 21 GBD level 2 causes as a proportion of total rates of YLL and YLD in children and adolescents 19 years or younger, both sexes, 1990-2015. Each geographical area is assigned an SDI value for each year, and nonlinear spline regressions are used to find the mean association between SDI and cause-specific rates of disease burden. HIV indicates human immunodeficiency virus. (eFigure 7A in the Supplement) and neonatal disorders (eFigure 7B in the Supplement). For congenital anomalies, there was a consistent decrease in the rate of YLLs, with increasing SDI for all causes, but increases in the rate of YLDs for most level 3 congenital anomalies, especially congenital heart anomalies and other congenital anomalies. For neonatal disorders, increasing SDI was associated with consistent improvements in neonatal sepsis, hemolytic disease of the newborn, and other neonatal disorders but not for preterm birth complications. Rates of DALYs for preterm birth complications increased initially, and there was little suggestion of further improvement beyond an SDI of 0.85 for any neonatal disorder, especially preterm birth complications. The mean association between SDI and sex-specific rates of DALYs of level 3 causes also showed broad consistency in that the burden owing to most causes decreased with increasing SDI (eFigure 8A-V in the Supplement). Exceptions were the level 3 causes that consistently increased with increasing SDImusculoskeletal and mental and substance abuse disorders-as well as those that either peaked in high-middle and middle SDI or improved only marginally until the highest SDI levels-asthma, epilepsy, migraine, road injuries, self-harm, interpersonal violence, and neoplasms such as leukemia, lymphoma, brain cancer, and other neoplasms. Sex differences were notable in that some conditions, such as neonatal disorders, neoplasms, nutritional deficiencies, hemoglobinopathies and hemolytic anemias, epilepsy, and all categories of injuries, had higher all-ages rates of DALYs in males, while others, such as migraine, gastrointestinal disorders, musculoskeletal disorders, and congenital anomalies, were higher in females.

Discussion
We found widespread reductions in total disease burden among children and adolescents, but progress has been unequal. In 2015, an even greater share of global mortality burden was concentrated in the lowest-SDI countries than it was in 1990, and there has been a significant increase in the global proportion of nonfatal disease burden. Global reductions in disease burden from infectious, neonatal, maternal, and nutritionassociated causes have been accompanied by a growing importance of NCDs and injuries. Neonatal disorders and congenital anomalies remain large and, in some cases, growing problems in many countries. Nutritional deficiencies, along with infections such as HIV and AIDS, diarrhea, LRIs, malaria, intestinal infectious diseases, and vaccine-preventable diseases, also still cause enormous health loss in some countries. Some populations of children and adolescents have been struck by the massive effect of war on health, while others are struggling with the detrimental effects of road injuries, drowning, self-harm, and interpersonal violence, and the growing importance of NCDs such as mental and substance use disorders, cancer, congenital anomalies, and hemoglobinopathies. Leading causes of death and disability vary as a function of age, sex, and SDI status, so the precise challenges for each country may be very different. The SDGs set a series of absolute timebound targets for improving health, the environment, and so-cietal development. By design, the SDG agenda is broader than the Millennium Development Goal framework, in which there was only one child target (Millennium Development Goal 4: reduce mortality by two-thirds in children younger than 5 years). By broadening the agenda and shifting to absolute targets, the SDGs provide an excellent starting point to judge progress at the country level and should rightfully focus attention on the countries with the most progress yet to achieve. 25 Tracking the entire spectrum of disease in children and adolescents facilitates monitoring of the SDGs but can also highlight non-SDG health challenges, and it should be used to inform final decision making with respect to SDG indicators.
Besides neonatal mortality and mortality in children younger than 5 years, 3 other health-associated SDGs are important for understanding disease burden in children and adolescents given the age pattern of these conditions. Examples include infection-associated SDG targets addressing malaria, HIV and tuberculosis, neglected tropical diseases, severe malnutrition, and access to safe water, sanitation, and hygiene. Our analysis shows that these SDG indicators are improving for children and adolescents in most geographical areas. Sustainable Development Goals focusing on reproductive health targets such as maternal mortality, adolescent fertility, universal access to modern contraception, skilled birth attendance, and neonatal support services are also an important part of the discussion about child and adolescent health loss given the continued importance of teenage pregnancy, pregnancyassociated complications, and maternal death in many settings as well as the intricate links between maternal and child health. Many of the injury-associated SDG targets also are relevant to tracking child health progress, including those on disaster preparedness, road injuries, poisoning, self-harm, interpersonal violence, and war.
Most of the SDG targets addressing NCDs, such as those associated with mental health conditions and cardiovascular diseases, not only focus on NCDs that are primarily problems in adults but also specifically exclude children and adolescents. This exclusion is problematic, especially because of the growing importance found in our study of mental health, substance abuse, and self-harm among children and adolescents. Half of all mental illnesses begin by age 14 years and three-quarters begin by the mid-20s. If untreated, these conditions can predispose to self-harm and severely influence children's development, educational attainment, and long-term fulfillment and economic potential. 26 Other SDGs, such as those addressing education and sex equality, are not specifically associated with health but can have a significant effect on the health of children and adolescents.
One possible explanation for growing inequality in disease burden among children and adolescents is that many of the geographical areas with the lowest SDIs have not historically been significant recipients of development assistance for health (DAH). 27,28 Although development assistance for child and newborn health has been among the fastestgrowing focus areas of DAH since 1990 and is one of the few areas in which funding has continued to increase since 2010, it has been uneven. For example, most countries in Central and Western sub-Saharan Africa received less than half the DAH per all-cause DALY in 2010 to 2012 received by countries in Southern sub-Saharan Africa, Eastern sub-Saharan Africa, and Central Latin America. 29 Many DAH programs have concentrated on funding widespread delivery of proven preventive measures, including vaccines; nutritional support; maternal education; improved water, sanitation, and hygiene; and treatment of diarrhea, LRIs, and HIV and AIDS. Synergistic maternal health programs have also enjoyed strong support during this period. These programs must continue and be expanded, especially in low-SDI settings in which progress in child and adolescent health has been comparatively slow, but focus must also be turned to strengthening health systems, especially in geographical areas in which such systems are underdeveloped or have been weakened by conflict. This finding leads to another possible explanation for growing inequality; namely, that there has been inadequate focus on increasing local health system capacity and capabilities. High-SDI settings have seen improved prevention and better outcomes for many childhood illnesses, such as neonatal disorders, as well as many congenital birth defects, injuries, and cancers, such as leukemia. Improved treatment has led to increasing numbers of children now reaching adulthood with ongoing medical needs. [30][31][32][33][34] These medical advances have not been as readily realized in geographical areas with lower SDIs or in older children and adolescents. 35,36 Sustained investment is needed to improve prevention, diagnosis, and treatment for causes not traditionally targeted by DAH, including strengthening of workforces and facilities, [37][38][39] cooperation between health centers in the same region, 40 clinician specialization, [41][42][43][44] improved surgical and anesthetic care, 34 evidence-based case management for life-threatening complications, 45,46 injury prevention, and active screening to identify and treat high-risk children (eg, those with congenital heart disease [47][48][49] and hemoglobinopathies 50,51 ). Integration of health care services across facilities and specialties to meet the varied needs of children and adolescents is paramount to successful intervention programs, especially in geographical areas with middle SDIs where there is a high likelihood of comorbid NCDs, injuries, and group I conditions. Indeed, the World Health Organization Global Strategy for Women's, Children's, and Adolescents' Health (2016-2030) 9 addresses both explanations for growing inequality. In addition to advocating for programs supported by DAH, it recommends specific actions to address the interplay between environment, economics, and health through multisector action including investments and policy implementation, leadership engagement at regional and country levels, and building resilience of health systems through workforce development, human capital investment, sex equality, and youth empowerment.
One aspect of child and adolescent health that is not comprehensively addressed by the World Health Organization Global Strategy for Women's, Children's, andAdolescents' Health (2016-2030) 9 is injuries. Minimizing the burden of injuries in children and adolescents requires the implementation of specific prevention policies where they do not exist and support for treatment and rehabilitation of injured youth. First and foremost, this implementation necessitates strict limita-tions on child labor and elimination of child slavery. 52 To reduce suicide, Sri Lanka and South Korea have both restricted the availability of pesticides with good effect, 53,54 while Australia has taken the step of implementing comprehensive mental health screening and treatment. 55 Along with suicide, interpersonal violence may be partially addressed via implementation of more stringent gun regulations, 56 although many other social, demographic, and societal factors, such as poverty, education, and drug use, also must be addressed in policies aimed at reducing homicide. 57,58 Road injuries, falls, and interpersonal violence are common etiologic causes of traumatic brain injury, a condition that can have significant mortality and morbidity. 59,60 Policies requiring seat belts, appropriate use of car seats, and helmets for cyclists can greatly reduce the risk of traumatic brain injury following road injuries. 61 Timely hospital care and ongoing rehabilitation services are necessary to minimize the long-term health burden of traumatic brain injury, 62,63 although optimal approaches are still in development. 64 In addition to effective medical programs, comprehensive community-based approaches are needed to maximize the development and empowerment of children and adolescents at the population level. One approach that could serve as a model-and may be possible in many high-SDI countries-is the Healthy Child, Healthy Future campaign in Northern Ireland. 65 The central tenet of the program is to promote a shift from an approach to child health that relies on medical screening to identify children with treatable conditions to an approach that also emphasizes promotion of general health, primary prevention of diseases and injuries, and active intervention for children at risk. Involving parents, schools, and communities in this effort is seen as a crucial part because they are the entities who have the most influence on the physical and social environment of children, and their own personal behaviors are likely to have a direct effect on children in their care.

Limitations
This analysis has several limitations. First, it is based on the GBD 2015 study results and is therefore subject to the same potential problems as the overall study, including variations in availability and quality of data, variations in coding practices, delays in release of data after their collection, paucity of data following conflict and natural disasters, and incomplete information or potential underreporting on nonfatal health outcomes in many geographical areas. Despite potential limitations, drawing conclusions on levels of and trends in child and adolescent health loss from this central study at least ensures that comparisons are all internally consistent with one another. Second, in this study we concentrated on only the leading global causes of fatal and nonfatal health loss in the aggregate age groups of children and adolescents 19 years or younger. This approach allows reporting on general patterns of health loss in children and adolescents, but given that childhood is a period of rapid development and change, it does not highlight less common conditions and may obscure some agespecific and geography-specific subtleties of epidemiologic factors of disease and injury. Third, data availability for children aged 5 to 14 years is not as robust as that for those 5 years or younger and individuals aged 15 to 19 years. Estimates of allcause mortality were thus based primarily on extensive data sources in children 5 years or younger and individuals aged 15 to 19 years, with estimates for intervening age groups derived from a large collection of empirical life tables and use of survival history data from surveys and censuses. 16 We found broad agreement between our all-cause mortality estimates and primary data sources from the Sample Registration System and Demographic and Health Surveys in India, which supports the GBD findings. Hill and colleagues 66 have argued against this approach on the basis that it could underestimate mortality in populations in which interventions targeted at children 5 years or younger have not also led to improved survival in children aged 5 to 14 years. The implication is that such underestimation would dilute the perceived importance of causes that disproportionately affect children older than 5 years (eg, injuries, NCDs, and maternal disorders), but even if that were the case, it should have minimal effect on the interpretation of levels and trends within each age group or country. Fourth, given the geography-centric approach to GBD analyses, we have limited ability to evaluate disease burden in subpopulations that are not geographically based, such as refugees and many indigenous peoples. Fifth, while GBD 2015 has analyzed the total burden and underlying etiologic causes of several broad conditions, such as anemia and developmental intellectual disability, there are others that may be particularly important in children and adolescents that have not received the same level of scrutiny, including, for example, traumatic brain injury, hydrocephalus, and sepsis. Sixth, our quantification of disease burden focuses only on affected individuals and therefore does not capture indirect effects, including effects on education and development of child labor, 52 the long-term effects of war, 67 or the burden on parents, siblings, and communities of caring for ill or injured children and adolescents. 68

Conclusions
Timely, robust, and comprehensive assessment of disease burden among children and adolescents provides information that is essential to health policy decision making in countries at all points along the spectrum of economic development. Understanding the burden of disease and how it is changing helps identify context-specific successes, unmet needs, and future challenges. Child and adolescent health has dramatically improved from 1990 to 2015 throughout the world. International attention and investment have led to improvements in many infectious and nutritional diseases, but progress has been uneven. The burden of disease during childhood and adolescence is now even more concentrated in lower-SDI countries and territories than it was a generation ago, and there is an ongoing epidemiologic transition with a growing relative burden of NCDs and injuries. If we are going to continue the current pace of improvement in child and adolescent health, we must invest in better data collection, continue to monitor trends in population disease burden, and adapt health systems to meet the ongoing and changing needs of children and adolescents so that all can have a chance to grow up to be healthy.