Measuring the health-related Sustainable Development Goals in 188 countries: a baseline analysis from the Global Burden of Disease Study 2015

Background In September, 2015, the UN General Assembly established the Sustainable Development Goals (SDGs). The SDGs specify 17 universal goals, 169 targets, and 230 indicators leading up to 2030. We provide an analysis of 33 health-related SDG indicators based on the Global Burden of Diseases, Injuries, and Risk Factors Study 2015 (GBD 2015). Methods We applied statistical methods to systematically compiled data to estimate the performance of 33 health-related SDG indicators for 188 countries from 1990 to 2015. We rescaled each indicator on a scale from 0 (worst observed value between 1990 and 2015) to 100 (best observed). Indices representing all 33 health-related SDG indicators (health-related SDG index), health-related SDG indicators included in the Millennium Development Goals (MDG index), and health-related indicators not included in the MDGs (non-MDG index) were computed as the geometric mean of the rescaled indicators by SDG target. We used spline regressions to examine the relations between the Socio-demographic Index (SDI, a summary measure based on average income per person, educational attainment, and total fertility rate) and each of the health-related SDG indicators and indices. Findings In 2015, the median health-related SDG index was 59·3 (95% uncertainty interval 56·8–61·8) and varied widely by country, ranging from 85·5 (84·2–86·5) in Iceland to 20·4 (15·4–24·9) in Central African Republic. SDI was a good predictor of the health-related SDG index (r2=0·88) and the MDG index (r2=0·92), whereas the non-MDG index had a weaker relation with SDI (r2=0·79). Between 2000 and 2015, the health-related SDG index improved by a median of 7·9 (IQR 5·0–10·4), and gains on the MDG index (a median change of 10·0 [6·7–13·1]) exceeded that of the non-MDG index (a median change of 5·5 [2·1–8·9]). Since 2000, pronounced progress occurred for indicators such as met need with modern contraception, under-5 mortality, and neonatal mortality, as well as the indicator for universal health coverage tracer interventions. Moderate improvements were found for indicators such as HIV and tuberculosis incidence, minimal changes for hepatitis B incidence took place, and childhood overweight considerably worsened. Interpretation GBD provides an independent, comparable avenue for monitoring progress towards the health-related SDGs. Our analysis not only highlights the importance of income, education, and fertility as drivers of health improvement but also emphasises that investments in these areas alone will not be sufficient. Although considerable progress on the health-related MDG indicators has been made, these gains will need to be sustained and, in many cases, accelerated to achieve the ambitious SDG targets. The minimal improvement in or worsening of health-related indicators beyond the MDGs highlight the need for additional resources to effectively address the expanded scope of the health-related SDGs. Funding Bill & Melinda Gates Foundation. GBD is an annual effort to measure the health of populations at regional, country, and selected subnational levels.33 GBD produces estimates of mortality and morbidity by cause, age, sex, and country for the period 1990 to the most recent year, reflecting all available data sources adjusted for bias. GBD also measures many health system characteristics, risk factor exposure, and mortality and morbidity attributable to these risks. In addition to providing highly detailed standardised information for many outcomes and risks, various summary measures are also computed, including disability-adjusted life-years (DALYs) and healthy life expectancy. For the present analysis, we used estimates from GBD 2015 to provide a baseline assessment for 188 countries. Further details on GBD 2015, which covers 1990–2015, are available elsewhere.34, 35, 36, 37, 38, 39 We defined health-related SDG indicators as indicators for health services, health outcomes, and environmental, occupational, behavioural, and metabolic risks with well established causal connections to health. Many of the 47 health-related SDG indicators selected by the IAEG-SDGs are produced as part of GBD. Table 1 outlines the ten goals, corresponding to 21 health-related targets and 33 health-related indicators included in this present iteration of GBD. This table also outlines the definition of the indicator used in this analysis; detailed descriptions of the estimation methods and data sources are given in the methods appendix pp 10–311. For the 14 health-related indicators that were not included in this analysis, their prospects for measurement in future iterations of GBD are described in table 2. Health-related SDG goals and targets proposed by the Inter-Agency and Expert Group on SDG Indicators, and health-related SDG indicators used in this analysis Detailed descriptions of the data sources and methods used to estimate each health-related SDG indicator are in the methods appendix pp 10–311. SDG=Sustainable Development Goal. MDG=Millennium Development Goal. IOTF=International Obesity Task Force. GBD=Global Burden of Disease Study. NCDs=non-communicable diseases. SEV=summary exposure value. WaSH=water, sanitation, and hygiene. JMP=Joint Monitoring Program. DALY=disability-adjusted life-year. PM2·5=fine particulate matter smaller than 2·5 μm. Health-related SDG indicators (proposed by the Inter-Agency and Expert Group on SDG Indicators) excluded in the present analysis, and measurement needs and strategy for future reporting, by SDG target SDG=Sustainable Development Goal. GBD=Global Burden of Disease. TRIPS=Agreement on Trade-Related Aspects of Intellectual Property Rights. DAH=development assistance for health. IHR=International Health Regulations. DHS=Demographic and Health Survey. ISIC=International Standard Industrial Classification. Direct outputs of GBD that are health-related SDG indicators include mortality disaggregated by age (under-5 and neonatal) and cause (maternal, cardiovascular disease, cancer, diabetes, chronic respiratory diseases, road injuries, self-harm, unintentional poisonings, exposure to forces of nature, interpersonal violence, and collective violence and legal intervention [ie, deaths due to law enforcement actions, irrespective of their legality]), as well as disease incidence (HIV, malaria, tuberculosis, and hepatitis B) and prevalence (neglected tropical diseases). The GBD comparative risk assessment includes measurement of exposure prevalence included as health-related SDG indicators (under-5 stunting, wasting, and overweight; tobacco smoking; harmful alcohol use; intimate partner violence; unsafe water, sanitation, and hygiene; household air pollution; and ambient particulate matter pollution), as well as deaths or disease burden attributable to risk factors selected as health-related SDG indicators (unsafe water, sanitation, and hygiene; household air pollution and ambient particulate matter pollution; and occupational risks). Underlying GBD outputs are a range of additional health determinants that contribute to the estimation of morbidity and mortality, for which data are systematically compiled and estimates are produced. For example, GBD comprehensively analyses data from household surveys on vaccine coverage and combines survey estimates with reported administrative data to produce time series of vaccine coverage for all countries from 1990 to 2015. Estimates of vaccine coverage are then included as predictors of vaccine-preventable morbidity and mortality in GBD. Additional health indicators produced as part of GBD and included as health-related SDG indicators in this analysis are: met need with modern contraception among women of reproductive age, adolescent birth rate, skilled birth attendance coverage, and universal health coverage (UHC) tracer interventions. For UHC tracer interventions, we developed an index based on the geometric mean of the coverage of a set of UHC tracer interventions: met need with modern contraception; antenatal care (one or more visits and four or more visits); skilled birth attendance coverage; in-facility delivery rates; vaccination coverage (three doses of diphtheria–pertussis–tetanus, measles vaccine, and three doses of oral polio vaccine or inactivated polio vaccine); tuberculosis case detection rate; coverage of antiretroviral therapy for populations living with HIV, and coverage of insecticide-treated nets for malaria-endemic countries. For selected indicators proposed by the IAEG-SDGs, we made modifications to the definition for clarity or on the basis of the definition used in GBD (table 1). For example, Indicator 2.2.2 proposes a measure of malnutrition that combined prevalence of wasting and overweight among children under age 5 years. As childhood wasting and overweight have very different determinants, we opted to report them separately. For childhood overweight, we report prevalence in children aged 2–4 years, the definition used in GBD based on thresholds set by the International Obesity Task Force.40 Further details on the estimation and data sources used for all indicators, compliant with Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER),41, 42 are included in the methods appendix pp 10–311. To identify broad patterns and more easily track general progress, we developed an overall health-related SDG index that is a function of the 33 health-related SDG indicators (referred to as the health-related SDG index). We also constructed two related indices: one reflecting the SDG health-related indicators previously included in the MDG monitoring framework (referred to as the MDG index) and one reflecting SDG health-related indicators not included in the MDGs (referred to as the non-MDG index). Three broad approaches can be used to create composite measures: normative, preference weighted, and statistical. Normative approaches combine each indicator based on first principles or an over-riding construct such as the contribution of each indicator to overall health. Preference-weighted approaches weight each indicator by expressed or elicited social preferences for the relative importance of different indicators. Statistical approaches seek to reduce a long set of variables or indicators into common components of variance using methods such as principal component analysis or factor analysis. In this case, because the SDGs reflect the collective vision of UN member states, we used a preference-weighted approach, assuming that each SDG target should be treated equally. To combine indicators, we adopted methods used to construct the Human Development Index,43 which include rescaling each indicator on a scale from 0 to 100 and then combining indicators using the geometric mean. The geometric mean allows indicators with very high values to partly compensate for low values on other indicators (referred to as partial substitutability). In the methods appendix pp 312–13, we describe results from alternative index construction methods (ie, principal component analysis; the arithmetic mean across targets referred to as complete substitutability; and the minimum value across targets referred to as zero substitutability). Quantitative targets for each of the health-related SDG indicators are not universally specified. As a result, we rescaled each health-related SDG indicator on a scale from 0 to 100, with 0 being the lowest (worst) value observed and 100 being the highest (best) value observed over the time period 1990–2015. We log-transformed mortality and morbidity before rescaling. We then estimated the health-related SDG index by first computing the geometric mean of each rescaled health-related SDG indicator for a given target, followed by the geometric mean of resulting values across all SDG targets. To avoid problems with indicator values close to 0, when computing indices we applied a floor of one to all indicators. This analytic approach weights each of the health-related SDG targets equally. In addition to the health-related SDG index, we also used the same methods to construct an index that represents 14 health-related SDG indicators that were previously MDG indicators and an index representing 19 non-MDG indicators (table 1). Uncertainty in the indicator and indices values was computed using a simulation analysis. As part of GBD 2015, we assessed cause-specific disease burden and risk exposure along the development spectrum, providing context on expected changes as countries progress to higher levels of income per person, higher educational attainment, and lower fertility.34, 37, 38, 39 We conducted a similar analysis by examining the relations of the overall health-related SDG index and each of the individual health-related SDG indicators with the Socio-demographic Index (SDI), a summary measure of development that uses lag-distributed income per person, average educational attainment in the population over age 15 years, and the total fertility rate. The SDI was constructed using the same method for the Human Development Index and the health-related SDG index. Each of the three components was first rescaled on a 0–1 scale, with 0 being the lowest (worst) value observed in the time period 1980–2015 and 1 being the highest (best) value observed. SDI was then computed as the geometric mean of these three rescaled components. To capture average relations, we used a spline regression (ie, piecewise linear regression with so-called knots specifying the intersection between pieces) of the health-related SDG indicators and health-related SDG index on SDI using the full set of data by country from 1990 to 2015. We also compared the health-related SDG indicators with the GBD 2015 estimates of healthy life expectancy38 to explore the relation between the SDGs and overall health achievement for each country. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.