Menu
Menu
Menu
Menu
Background: Rapid demographic, epidemiological, and nutritional transitons have brought a pressing need to track progress in adolescent health. Here, we present country-level estimates of 12 headline indicators from the Lancet Commission on adolescent health and wellbeing, from 1990 to 2016. Methods: Indicators included those of health outcomes (disability-adjusted life-years [DALYs] due to communicable, maternal, and nutritional diseases; injuries; and non-communicable diseases); health risks (tobacco smoking, binge drinking, overweight, and anaemia); and social determinants of health (adolescent fertility; completion of secondary education; not in education, employment, or training [NEET]; child marriage; and demand for contraception satisfied with modern methods). We drew data from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2016, International Labour Organisation, household surveys, and the Barro-Lee education dataset. Findings: From 1990 to 2016, remarkable shifts in adolescent health occurred. A decrease in disease burden in many countries has been offset by population growth in countries with the poorest adolescent health profiles. Compared with 1990, an additional 250 million adolescents were living in multi-burden countries in 2016, where they face a heavy and complex burden of disease. The rapidity of nutritional transition is evident from the 324·1 million (18%) of 1·8 billion adolescents globally who were overweight or obese in 2016, an increase of 176·9 million compared with 1990, and the 430·7 million (24%) who had anaemia in 2016, an increase of 74·2 million compared with 1990. Child marriage remains common, with an estimated 66 million women aged 20–24 years married before age 18 years. Although gender-parity in secondary school completion exists globally, prevalence of NEET remains high for young women in multi-burden countries, suggesting few opportunities to enter the workforce in these settings. Interpretation: Although disease burden has fallen in many settings, demographic shifts have heightened global inequalities. Global disease burden has changed little since 1990 and the prevalence of many adolescent health risks have increased. Health, education, and legal systems have not kept pace with shifting adolescent needs and demographic changes. Gender inequity remains a powerful driver of poor adolescent health in many countries. Funding: Australian National Health and Medical Research Council, and the Bill & Melinda Gates Foundation.
We populated the 12 headline indicators for adolescent health as defined by the Lancet Commission on adolescent health and wellbeing (table 1),1 hereafter referred to as the Commission, with global, country-level, and disease-group data. In selecting data sources, we considered coverage, data quality, international comparability, and recency of data. We first reviewed relevant data available at the Institute of Health Metrics and Evaluation (IHME). Although some data might be available elsewhere, IHME has extensively catalogued primary health data for each country across the world. For their Global Burden of Diseases, Injuries, and Risk Factors Study (GBD), they have harmonised estimates and used models to fill data gaps to produce estimates that are updated annually for 333 health outcomes and 84 risks and determinants by country or territory, sex, and 5-year age group.18, 19 We used data from GBD 2016, which included 195 countries and territories. Specific methods and definitions of GBD 2016 are described elsewhere.18, 19, 20, 21 Data for some indicators of social determinants were not readily available at IHME and we reviewed other relevant global data collections to obtain these.7, 13 Definitions and data availability for 12 headline indicators of adolescent health from the Lancet Commission on adolescent health and wellbeing DALYs=disability-adjusted life-years. IHME=Institute for Health Metrics and Evaluation. IOTF=World Obesity Federation. BMI=body-mass index. NEET=not in education, employment, or training. In the Commission, the indicators were organised into three main categories: health outcomes, health risks, and social determinants of health. Brief definitions of each indicator and the data sources used are shown in table 1 (full definitions of each indicator and data sources are in the appendix). Indicators were generally defined for adolescents aged 10–24 years. This age definition reliably captures developmental stage during biological, social, and neurocognitive transitions.3 We reported indicators for adolescents in 195 locations, as defined by IHME, comprising 188 countries (as classified by the UN) and seven territories (Puerto Rico, American Samoa, Bermuda, Greenland, Guam, Northern Mariana Islands, and the Virgin Islands). Hereafter, countries refers to both UN countries and territories. Because the population of these 195 countries is more than 99·9% of the global population, in this analyses we report our data as global estimates. The Commission defined three country disease groups (country groups) on the basis of adolescent disease burden to represent different stages of the epidemiological transition (and therefore different patterns of health need) for adolescents.1 Countries were defined as multi-burden if adolescents (aged 10–24 years, both sexes combined) of that country had a burden of communicable, maternal, and nutritional conditions (ie, group 1 conditions; table 1) of 2500 disability-adjusted life-years (DALYs) or more per 100 000 adolescents. This threshold was set using GBD 2013 data to capture countries where DALYs caused by group 1 conditions in 2013 were at least double the average rate for countries where non-communicable diseases were predominant, and countries where group 1 conditions accounted for at least 20% of the total burden among those aged 10–24 years. Countries were defined as injury excess if adolescents were estimated to have 2500 DALYs or more per 100 000 adolescents due to injury and less than 2500 DALYs per 100 000 population due to group 1 conditions. This threshold was set to identify those countries that had a low burden of group 1 conditions but that had a burden of injuries at least twice that of countries where non-communicable diseases were predominant, and countries where injury accounted for least 20% of the total burden for adolescents. Countries were defined as non-communicable disease predominant if both injuries and group 1 conditions each contributed less than 2500 DALYs per 100 000 adolescents to the disease burden. We used these definitions to define country groups in 1990 and 2016; country groupings reported are for 2016 unless otherwise specified. Hence, with progression through the epidemiological transition, a country would be expected to move up from the multi-burden group to the injury excess group, and from the injury-excess group to the non-communicable disease-predominant group. These country groupings broadly corresponded to socioeconomic development (defined in the appendix). The median Socio-demographic Index (SDI) value for multi-burden countries was 0·45 (range 0·19–0·74), which was generally lower than injury-excess countries (SDI median 0·71, range 0·47–0·88) and the non-communicable disease-predominant countries (SDI median 0·8, range 0·45–0·94). The country groupings also broadly corresponded with World Bank income levels, with almost all (30 [97%] of 31) low-income countries in 2016 classified as multi-burden and almost all (52 [88%] of 59) high-income countries classified as non-communicable disease predominant (appendix). We report the most recent estimate for each indicator at a country level. For each indicator we identified the countries with the lowest and highest observed values and ranked each country in between, presenting this ranking as a heat map. We also report estimates for each indicator for the three country groups and globally. We report group estimates using country-level estimates for 2016 for indicators drawn from IHME, 2010 estimates for secondary education, and the most recent country-specific data available for child marriage (2003–16), and not in education, employment, or training (NEET; 2005–16; appendix). For indicators drawn from IHME data, we generated group counts as the sum of counts in each country. We then used these group counts of numerator and denominator to generate estimates of group prevalence. For the indicators of secondary education, child marriage, and NEET, the group prevalence estimates reflect only countries for which data were available, and we calculated the count estimates (estimated number of adolescents) using the prevalence derived from the available data and applied that to the total denominator population of that group. For the indicator of child marriage, data were not available for many countries in the non-communicable disease-predominant group and data were mostly missing for high-income countries. To estimate the count and prevalence of child marriage in the non-communicable disease-predominant country group, we assumed that non-communicable disease-predominant countries without an estimate had a prevalence equivalent to the lowest prevalence observed in the rest of the group. Data for child marriage were of sufficient coverage for injury-excess and multi-burden countries. In addition to estimating the group counts on the basis of observed data, we estimated global counts from two scenarios to distinguish changes between 1990 and 2016 due to shifts in health and due to population change—ie, stable demography and changing epidemiology, and changing demography with fixed epidemiology. We populated indicators associated with NEET and child marriage from primary data; however, standard errors (and therefore confidence intervals) were not readily available for these estimates. We drew data for estimates for the other indicators from modelled data based on multiple individual primary data sources. Uncertainty estimates for these estimates (distinct from confidence intervals in that they represent uncertainty derived from primary data, model estimation, and model specification) were available at a country level for adolescents but were not readily available for manipulated data (including aggregate country groupings). Uncertainty estimates at a country level for indicators drawn from IHME (other than adolescent fertility) are provided in the appendix. Adolescent fertility in GBD 2016 was modelled using a hybrid approach of modelling the total fertility rate directly and then fitting UN World Population Prospects age patterns of fertility to those estimates; uncertainty estimates for adolescent fertility were not available. Uncertainty estimates were not available for indicators of injury and non-communicable disease burden because of the redistribution of self-harm from the IHME level 1 injury group to the level 1 non-communicable disease group because of its association with mental disorders. To enable assessment of data quality, we provide uncertainty estimates for the IHME level 1 groups and self-harm separately. Furthermore, we provide uncertainty estimates for all-cause years of life lost (YLL) and years lost due to disability (YLD) for adolescents to allow an assessment of the quality of data of the two broad components of DALYs. We report annual rates of change for each indicator at a country and group level. Linear regression models were fitted to available data points for each indicator and location, and we used the β coefficient to estimate the annual rate of change (expressed as percentage change). We used linear regression (rather than the estimates at each extreme of the time series) to account for fluctuations in some indicators over time (eg, an increase in DALYs due to war or conflict). We used data points for 1990–2015 in intervals of 5 years plus 2016 for data drawn from IHME. For data drawn from Barro-Lee, an educational attainment dataset that covers the period 1950–2010, we used intervals of 5 years for data from the period 1990–2010. Rates of change for NEET and child marriage could not be caulated because of restricted data over time. The funders provided salaries for research staff and had no role in study design, data collection, data analysis, data interpretation, or writing of the report. PSA, SJCH, KLF, ECK, and GCP had access to all the data; AHM, NJK, SL, CMSI, and TV had access to all the data as provided by IHME; and all other authors able to access the data as requested. The corresponding author had final responsibility for the decision to submit for publication.
