Article OPEN https://doi.org/10.1038/s41586-019-1545-0 Mapping 123 million neonatal, infant and child deaths between 2000 and 2017 A list of authors and their affiliations appears in the online version of the paper. Since 2000, many countries have achieved considerable success in improving child survival, but localized progress remains unclear. To inform efforts towards United Nations Sustainable Development Goal 3.2—to end preventable child deaths by 2030—we need consistently estimated data at the subnational level regarding child mortality rates and trends. Here we quantified, for the period 2000–2017, the subnational variation in mortality rates and number of deaths of neonates, infants and children under 5 years of age within 99 low- and middle-income countries using a geostatistical survival model. We estimated that 32% of children under 5 in these countries lived in districts that had attained rates of 25 or fewer child deaths per 1,000 live births by 2017, and that 58% of child deaths between 2000 and 2017 in these countries could have been averted in the absence of geographical inequality. This study enables the identification of high-mortality clusters, patterns of progress and geographical inequalities to inform appropriate investments and implementations that will help to improve the health of all populations. Gains in child survival have long served as an important proxy meas- in 99 countries at policy-relevant subnational scales (first and second ure for improvements in overall population health and development1,2. administrative levels) for each year from 2000 to 2017. We fit a geo- Global progress in reducing child deaths has been heralded as one of statistical discrete hazards model to a large dataset that is composed the greatest success stories of global health3. The annual global num- of 467 geo-referenced household surveys and censuses, representing ber of deaths of children under 5 years of age (under 5)4 has declined approximately 15.9 million births and 1.1 million deaths of children from 19.6 million in 1950 to 5.4 million in 2017. Nevertheless, these from 2000 to 2017. Our model includes socioeconomic, environmental advances in child survival have been far from universally achieved, and health-related spatial covariates with known associations to child particularly in low- and middle-income countries (LMICs)4. Previous mortality and uses a Gaussian process random effect to exploit the subnational child mortality assessments at the first (that is, states or correlation between data points near each other across dimensions of provinces) or second (that is, districts or counties) administrative level space, time and age group, which helps to mitigate the limitations asso- indicate that extensive geographical inequalities persist5–7. ciated with data sparsity in our estimations. For this study, we report Progress in child survival also diverges across age groups4. Global U5MR as the expected number of deaths per 1,000 live births, reflecting reductions in mortality rates of children under 5—that is, the under-5 the probability of dying before the age of 5 for a given location and year. mortality rate (U5MR)—among post-neonatal age groups are greater than those for mortality of neonates (0–28 days)4,8. It is relatively Unequal rates of child mortality unclear how these age patterns are shifting at a more local scale, pos- The risk of a newborn dying before their fifth birthday varies tremen- ing challenges to ensuring child survival. To pursue the ambitious dously based on where in the world, and within their country, they are Sustainable Development Goal (SDG) of the United Nations9 to “end born. Across the 99 countries in this study, we estimate that U5MR preventable deaths of newborns and children under 5” by 2030, it is varied as much as 24-fold at the national level in 2017, with the highest vital for decision-makers at all levels to better understand where, and rate in the Central African Republic of 123.9 deaths (95% uncertainty at what ages, child survival remains most tenuous. interval, 104.9–148.2) per 1,000 live births, and the lowest rate in Cuba of 5.1 deaths (4.4–6.0)4. We observed large subnational variation within Precision public health and child mortality countries in which overall U5MR was either high or comparatively Country-level estimates facilitate international comparisons but mask low. For example, in Vietnam, rates across second administrative units important geographical heterogeneity. Previous assessments of mortal- (henceforth referred to as ‘units’) varied 5.7-fold, from 6.9 (4.6–9.8) in ity of children under 5 have noted significant within-country heteroge- the Tenth District in Hồ Chí Minh City to 39.7 (28.1–55.6) in Mường neity, particularly in sub-Saharan Africa5,7,10–14, as well as in Brazil15, Tè District in the Northwest region (Figs. 1b, 2). Iran16 and China17. Understanding public health risks at more granular Decreases in U5MR between 2000 and 2017 were evident to some subpopulation levels is central to the emerging concept of precision extent throughout all units (Figs. 1a, b, 2). No unit showed a significant public health18, which uses “the best available data to target more effec- increase in U5MR in this period, and in most units U5MR decreased tively and efficiently interventions…to those most in need”18. Efforts greatly, even in units in which the mortality risk was the highest. Out to produce high-resolution estimates of mortality of children under 5, of 17,554 units, 60.3% (10,585 units) showed a significant (defined determinants at scales that cover the multiple countries are emerging, as 95% uncertainty intervals that did not overlap) decrease in U5MR including for vaccine coverage19,20, malaria21, diarrhoea22 and child between 2000 and 2017. Across units in 2000, U5MR ranged from 7.5 growth failure23,24. In a previous study, we produced comprehensive (5.0–10.6) in Santa Clara district, Villa Clara province, Cuba, to 308.4 estimates of African child mortality rates at a 5 × 5-km scale for 5-year (274.9–348.4) in the Sabon Birni Local Government Area of Sokoto intervals5. For areas outside of Africa, in which 72% of the world’s chil- State, Nigeria. By 2017, the unit with the highest estimated U5MR dren live and 46% of global child deaths occurred in 20174, subnational across all 99 countries was Garki Local Government Area, Jigawa state, heterogeneity remains mostly undescribed25. Nigeria, at 195.1 (158.6–230.9). Overall, the total percentage of units Here we produce estimates of death counts and mortality rates of chil- with a U5MR higher than 80 deaths per 1,000 live births decreased from dren under 5, infants (under 1 years of age) and neonates (0–28 days) 28.9% (5,070) of units in 2000 to 7.0% (1,236) in 2017. Furthermore, 1 7 O CT O B ER 2 0 1 9 | V O L 5 7 4 | N A T U RE | 3 5 3 RESEARCH Article Mortality rate Mortality rate a per 1,000 b per 1,000 live births live births >200 >200 100 100 25 25 5 5 Posterior probability of meeting SDG for c under-5 mortality in 2017 1.00 d Neonatal/ under-5 0.75 mortality ratio 0.50 >0.75 0.25 0.50 0 <0.25 Fig. 1 | U5MR estimates in 99 LMICs. a, U5MR at the second 1,000 live births for children under 5 in 2017. d, Proportion of mortality administrative level in 2000. b, U5MR at the second administrative level of children under 5 occurring in the neonatal (0–28 days) group at the in 2017. c, Modelled posterior exceedance probability that a given second second administrative level in 2017. administrative unit had achieved the SDG 3.2 target of 25 deaths per 32% of units, representing 11.9% of the under-5 population in the 99 deaths in 13% (2,226) of units, mostly within high-mortality areas in countries, had already met SDG 3.2 for U5MR with a 90% certainty sub-Saharan Africa. threshold (Fig. 1c). For neonatal mortality, 34% of units met the target of ≤12 deaths per 1,000 live births (Extended Data Fig. 1). Within Distribution of under-5 deaths may not follow rates countries, successes were mixed in some cases. For example, Colombia, The goal of mortality-reduction efforts is ultimately to prevent prema- Guatemala, Libya, Panama, Peru and Vietnam had all achieved SDG ture deaths, and not just to reduce mortality rates. Across the countries 3.2 for U5MR at the national level by 2017, but each country had units studied here, there were 3.5 million (41%) fewer deaths of children that did not achieve the goal with 90% certainty (Fig. 1c). under 5 in 2017 than in 2000 (5.0 million compared to 8.5 million). At Successful reductions in child mortality were also observed through- the national level, the largest number of child deaths in 2017 occurred out entire countries. For example, in 43 LMICs across several world in India (1.04 (0.98–1.10) million), Nigeria (0.79 (0.65–0.96) million), regions, the worst-performing unit in 2017 had a U5MR that was lower Pakistan (0.34 (0.27–0.41) million) and the Democratic Republic of than the best-performing unit in 2000 (Fig. 2). Nearly half of these the Congo (0.25 (0.21–0.31) million) (Fig. 3a). Within these countries, countries were in sub-Saharan Africa. Rwanda showed notable progress the geographical concentration of the deaths of the children varied. In during the study period, reducing mortality from 144.0 (130.0–161.6) Pakistan, over 50% of child deaths in 2017 occurred in Punjab province, in its best-achieving district in 2000 (Rubavu) to 57.2 (47.4–72.1) in which had a U5MR of 63.3 (54.1–76.0) deaths per 1,000 live births its worst-achieving district in 2017 (Kayonza). These broad reduc- (Fig. 3b). By contrast, 50% of child deaths in the Democratic Republic tions in U5MR have also led to a convergence of absolute subnational of the Congo in 2017 occurred across 9 out of 26 provinces. Such geographical inequalities, although relative subnational inequalities findings are in a large part artefacts of how borders are drawn around appear to be mostly unchanged between 2000 and 2017 (Fig. 2 and various at-risk populations (the provinces above account for 53% and Supplementary Fig. 6.12). Despite this success, the highest U5MRs in 63%, respectively, of the under-5 population that is at risk in these two 2017 were still largely concentrated in areas in which rates were highest countries), but can have a real impact at the level at which planning in 2000 (Fig. 1a, b). We observed estimated U5MR ≥ 80 across large occurs. Some concentrated areas with apparent high absolute numbers geographical areas in Western and Central sub-Saharan Africa, and of deaths highlighted by local-level estimates become less noticeable within Afghanistan, Cambodia, Haiti, Laos and Myanmar (Fig. 1b). when reporting at aggregated administrative levels; for example, areas Deaths of neonates (0–28 days of age) and post-neonates (28–364 across Guatemala, Honduras and El Salvador are visually striking hot- days of age) have come to encompass a larger fraction of overall mor- spots in Fig. 3d, but less so in Fig. 3b, c. tality of children under 5 in recent years. By 2017 (Fig. 1d), neonatal Our estimates indicate that targeting areas with a ‘high’ U5MR of mortality increased as a proportion of total deaths of children under 80 will have a lower overall effect than in previous years owing to 5 in 91% (90) of countries and for 83% (14,656) of units compared to the reductions in mortality rates. In 2000, 23.7% of child deaths— 2000. In almost all places where U5MR decreased, the share of the representing 2.0 (1.7–2.4) million deaths—occurred in regions in which mortality burden increased in the groups of children with younger ages. U5MR was less than 80 that year (Fig. 4). By comparison, in 2017, Similarly, the mortality of infants (<1 year) has increased relative to 69.5% of child deaths occurred in areas in which U5MR was below 80. the mortality for children who are 1–4 years of age in many areas. For A growing proportion of deaths of children under 5 are occurring in example, in the Diourbel Region, Senegal, infant mortality constituted ‘low’-mortality areas; 7.3% (5.1–10.2) of all deaths of children under 5 54.4% (52.4–56.6) of total mortality of children under 5 in 2000; by in 2017 occurred in locations in which the U5MR was below the SDG 2017, the relative contribution of infant mortality was 73.2% (70.3– 3.2 target rate of 25, compared to 1.2% (0.9–1.6) in 2000. For instance, 75.8). This shift towards mortality predominantly affecting neonates Lima, Peru, has a U5MR in the 8th percentile of units in this study, yet and infants was not as evident in all locations; mortality for children it ranks in the 96th percentile of highest number of deaths of children aged 1–4 years was responsible for more than 30% of overall under-5 under 5. 3 5 4 | N A T U RE | V O L 5 7 4 | 1 7 O CT O B ER 2 0 1 9 Article RESEARCH a 300 Under-5 mortality per 1,000 live births 200 100 0 CUB CRI TUN SLV PSE VNM NIC COL IRN PRY JOR JAM LBY HND VEN PER TTO PAN EGY BLZ ECU SYR CPV KGZ MAR UZB GTM IRQ GUY IDN PHL SUR TKM NPL DOM STP BTN BOL BGD KHM MNG ZAF TLS IND NAM GAB KEN MMR SDN ERI SWZ YEM COM GMB PNG DJI SEN MRT AFG ZWE HTI GNQ ETH ZMB LAO COG MWI LSO AGO GNB TGO MOZ MDG BDI CMR LBR TJK PAK CIV COD BEN GIN SOM SSD NER SLE TCD MLI CAF THA LKA DZA BWA RWA GHA UGA TZA NGA BFA Central Europe, eastern Europe and central Asia North Africa and Middle East Southeast Asia, East Asia and Oceania b GBD super region Latin America and the Caribbean South Asia Sub-Saharan Africa 4 U5MR relative to national mean 3 2 1 0 CUB CRI TUN SLV PSE VNM NIC COL IRN PRY JOR JAM LBY HND VEN PER TTO PAN EGY BLZ ECU SYR CPV KGZ MAR UZB GTM IRQ GUY IDN PHL SUR TKM NPL DOM STP BTN BOL BGD KHM MNG ZAF TLS IND NAM GAB KEN MMR SDN ERI SWZ YEM COM GMB PNG DJI SEN MRT AFG ZWE HTI GNQ ETH ZMB LAO COG MWI LSO AGO GNB TGO MOZ MDG BDI CMR LBR TJK PAK CIV COD BEN GIN SOM SSD NER SLE TCD MLI CAF THA LKA DZA BWA RWA GHA UGA TZA NGA BFA Country Fig. 2 | Geographical inequality in U5MR across 99 countries for 2000 in each country. Thus, each bar represents the extent of geographical and 2017. a, Absolute inequalities. Range of U5MR estimates in second inequality in U5MRs estimated for each country. Bars indicating the administrative-level units across 99 LMICs. b, Relative inequalities. Range range in 2017 are coloured according to their Global Burden of Disease of ratios of U5MR estimates in second administrative-level units relative super-region. Grey bars indicate the range in U5MR in 2000. The diamond to country means. Each dot represents a second administrative-level unit. in each bar represents the median U5MR estimated across second The lower bound of each bar represents the second administrative-level administrative-level units in each country and year. A coloured bar that is unit with the lowest U5MR in each country. The upper end of each bar shorter than its grey counterpart indicates that geographical inequality has represents the second administrative-level unit with the highest U5MR narrowed. Despite population growth, child deaths have declined due to the deaths are estimated to have occurred, we provide an important evi- outpaced decline in U5MR. For example, there were a total of 8.5 dence base for looking both to the past, for examples of success, and (7.2–10.0) million deaths of children under 5 in the countries in this towards the future, in order to identify where precision public-health study in 2000; had the 2017 under-5 population been exposed to the initiatives could save the most lives. same U5MRs that were observed in 2000, there would have been 10.6 The epidemiological toll of child mortality should be considered both (9.0–12.5) million deaths in 2017. Instead, we observed 5.0 (3.8–6.6) in terms of total deaths and as rates of mortality. Focusing only on million deaths in 2017 (Extended Data Fig. 5). mortality rates can effectively mask areas in which rates are compara- Finally, we combine estimates of subnational variation in mortality tively low but child deaths are high owing to large population sizes. The rates and populations to gain a better understanding of the impact number of deaths that occur in high-risk areas has declined, and most of geographical inequality. Overall, 2.7 (2.5–2.9) million deaths, or under-5 deaths in recent years have occurred in lower-risk areas. This 54% of the total number of deaths of children under 5, would have ‘prevention paradox’28 could indicate that whole-population interven- been averted in 2017 had all units had a U5MR that matched the best- tions could have a larger overall impact than targeting high-risk areas29. performing unit in each respective country (Extended Data Fig. 2). At the same time, strategies that target resources to those locations that Over the 2000–2017 period, this number is 71.8 (68.5–74.9) million have the highest number of child deaths risk leaving behind some of deaths, or 58% (55–61) of the total number of deaths of children under the world’s most marginalized communities: remote, more-sparsely 5. Total deaths attributable to inequality in this scenario ranged from populated places in which, relative to the number of children born 13 (6–24) deaths in Belize to 0.84 (0.72–0.99) million deaths in India. each year, a large number of children die before their fifth birthday. Furthermore, had all units met the SDG 3.2 target of 25 deaths per Instead, by considering subnational measures of both counts and rates 1,000, an estimated 2.6 (2.3–2.8) million deaths of children under 5 of deaths of children under 5, decision-makers can better tailor child would have been averted in 2017. health programs to align with local contexts, norms and needs. Rural communities with high rates but low counts may benefit from ‘last- Discussion mile’ initiatives to provide effective health services to populations who This study offers a comprehensive, geospatially resolved resource for lack adequate access to care. By contrast, locations with low rates but national and subnational estimates of child deaths and mortality rates high counts may require programs that focus on alleviating the cost of for 99 LMICs, where 93% of the world’s child deaths4 occurred in 2017. care, unsafe environmental exposures or health risks that are uniquely Gains in child survival varied substantially within the vast majority of associated with urban slums30. The SDGs have pointed the global countries from 2000 to 2017. Countries such as Vietnam, for example, development agenda towards progress in child survival. Our analysis showed more than fivefold variation in mortality rates across second indicates that reaching the SDG 3.2 targets of 25 child deaths per 1,000 administrative-level units. The inconsistency of successes, even at live births and 12 neonatal deaths per 1,000 live births will require subnational levels, indicates how differences in health policy, finan- only modest improvements or have already been achieved by some cial resources, access to and use of health services, infrastructure, and units; however, these targets are ambitious for other units in which economic development ultimately contribute to millions of lives cut child mortality remains high. It is worth noting that many countries short25–27. By providing detailed maps that show precisely where these contain areas that fit both of these profiles. For example, 11 countries 1 7 O CT O B ER 2 0 1 9 | V O L 5 7 4 | N A T U RE | 3 5 5 RESEARCH Article a Number of deaths (×103) b Number of deaths (×103) >500 >50 200 20 50 5 5 0.5 c Number d Number of deaths of deaths (×103) per grid cell >5 >10 2 4 0.5 1 0.05 Fig. 3 | Estimated number of children under 5 who died within 99 c, Number of deaths in each second administrative-level unit. d, Number countries in 2017. a, Number of deaths of children under 5 in each of deaths of children under 5 in each 5 × 5-km grid cell. Note that scales country. b, Number of deaths in each first administrative-level unit. vary for each aggregation unit. had at least 1 unit that had already met SDG 3.2 with high certainty, Notably, irrespective of income level or location, some causes of neo- and at least 1 unit that had not. Subnational estimates can empower natal death (for example, chromosomal anomalies and severe preterm countries to benchmark gains in child survival against their own sub- birth complications) remain difficult to prevent completely with cur- national exemplars as well as advances that have been achieved by their rent medical technologies. Ultimately, large gains in neonatal mortal- peers. Through our counterfactual analysis we showed that even if all ity will require serious investment in health system strengthening39. units had met the SDG 3.2 goal in 2017, there would still have been Affordable approaches to preventing the majority of neonatal deaths 2.4 million deaths of children under 5, indicating that ‘ending pre- in LMICs exist and there are success stories with lessons learned to ventable child deaths’ is more complex than simply meeting a target apply40–44, but decisions about which approaches to take must be based threshold. Future research efforts must address the causes of child mor- on the local epidemiological and health system context. In the absence tality in local areas and more precisely identify causes of child deaths of spatially detailed cause of death data, subnational neonatal mortality that are amenable to intervention. To that end, new and innovative estimates can indicate dominant causes and thus serve as a useful proxy data-collection efforts, such as the ongoing Child Health and Mortality to guide prioritization of interventions45. Prevention Surveillance network, offer promising prospects by applying The accuracy and precision of our estimates were primarily deter- high-validity, pathology-based methods alongside verbal autopsies to mined by the timeliness, quantity and quality of available data. In Sri determine the cause of death31. Lanka, for example, there were no available surveys, and the wide This study offers a unique platform to support the identification of uncertainty intervals surrounding estimates reflect the dearth of availa- local success stories that could be replicated elsewhere. In Rwanda, ble evidence in that country (Extended Data Figs. 3, 4). In certain areas, for example, the highest U5MR at the district level in 2017 was 60.2% this decreased the confidence that we had in claiming that a specific (52.0–67.8%) lower than the lowest U5MR at the district level in 2000. subnational area met the SDG 3.2 target (Fig. 1c). This issue is most Such gains have been partially credited to focused investments in concerning in cases in which estimated mortality rates are high, thus the country’s poorest populations, expanding the Mutuelles de santé helping to identify locations in which it would be most useful to focus insurance program, and developing a strong workforce of community future data-collection efforts. High mortality rates with large uncer- health workers who provide evidence-based treatment and health pro- tainty intervals were estimated across much of Eastern and Central motion32,33. Nepal and Cambodia are among the exemplars for consid- sub-Saharan Africa, and in Cambodia, Laos, Myanmar and Papua New erably decreasing subnational inequalities in child survival since 2000. Guinea (Extended Data Figs. 3, 4). Furthermore, ongoing conflict in In an era when narrowing disparities within countries is as important countries such as Syria, Yemen and Iraq pose substantial challenges as reducing national-level gaps, these results provide the evidence base to collecting more contemporaneous data, and our estimates may not to inform best practices and stimulate national conversations about fully capture the effects of prolonged civil unrest or war46,47. Further related social determinants. methodological and data limitations are discussed in the Methods. Neonatal mortality rates have also declined but failed to keep pace The accurate estimation of mortality is also a matter of equity; with reductions in mortality rates of older children, leading to a higher highly refined health surveillance is common in high-income coun- proportion of deaths of children under 5 occurring within the first four tries, whereas in LMICs, in which rates of child mortality are the high- weeks of life: from 37.4% (37.1–37.7) in 2000 to 43.7% (43.1–44.3%) in est, surveillance that helps to guide investments in health towards the 2017. This trend is probably related to the increase in scale of routine areas with the greatest need is less routine48. Ideally, all countries would programs and improved infrastructure (for example, vaccination34, and have high-quality, continuous, and complete civil and vital registration water and sanitation35) and the introduction of effective interventions systems that capture all of the births, deaths and causes of death at to target communicable diseases (for example, malaria control36 and the appropriate geographical resolution49. In the meantime, analyses prevention of mother-to-child transmission of HIV37). These inter- such as this serve to bridge the information gap that exists between ventions have tended to target amenable causes of mortality that are low-mortality countries with strong information systems and countries more common in older children under 5 rather than dominant causes that face a dual challenge of weaker information systems and higher of neonatal mortality, such as prematurity and congenital anomalies38. disease burden. 3 5 6 | N A T U RE | V O L 5 7 4 | 1 7 O CT O B ER 2 0 1 9 Article RESEARCH a 2000 Number of under-5 deaths (×105) SDG 3.2 target 5 4 2017 3 2 1 0 0 25 100 200 300 U5MR (per 1,000) b 2017 Number of under-5 deaths (×105) SDG 3.2 target 5 4 2000 3 2 1 0 0 25 100 200 300 U5MR (per 1,000) Eastern Europe and North Africa and Southeast Asia, East Asia GBD super region central Asia Middle East and Oceania Latin America and South Asia Sub-Saharan Africa the Caribbean Fig. 4 | Number of deaths of children under 5, distributed across of all bars represents the total number of deaths across the 99 countries. level of U5MR, in 2000 and in 2017, across 99 countries. Bar heights a, Deaths of children under 5 in 2000. b, Deaths of children under 5 in represent the total number of deaths of children under 5 within all second 2017. The dotted line in the 2000 plot is the shape of the distribution in administrative-level units with corresponding U5MR. Bins are a width of 2017, and the dotted line in the 2017 plot represents the distribution in 5 deaths per 1,000 live births. The colour of each bar represents the global 2000. region as defined by the subset legend map. As such, the sum of heights By harnessing the unprecedented availability of geo-referenced data 11. Dwyer-Lindgren, L. et al. Small area estimation of under-5 mortality in Bangladesh, Cameroon, Chad, Mozambique, Uganda, and Zambia using and developing robust statistical methods, we provide a high-resolution spatially misaligned data. Popul. Health Metr. 16, 13 (2018). atlas of child death counts and rates since 2000, covering countries 12. Wakefield, J. et al. Estimating under-five mortality in space and time in a that account for 93% of child deaths. We bring attention to subna- developing world context. Stat. Methods Med. Res. https://doi. org/10.1177/0962280218767988 (2018). tional geographical inequalities in the distribution, rates and absolute 13. Burke, M., Heft-Neal, S. & Bendavid, E. Sources of variation in under-5 mortality counts of child deaths by age. These high-resolution estimates can help across sub-Saharan Africa: a spatial analysis. Lancet Glob. Health 4, e936–e945 decision-makers to structure policy and program implementation and (2016). 14. Macharia, P. M. et al. Sub national variation and inequalities in under-five facilitate pathways to end preventable child deaths50 by 2030. mortality in Kenya since 1965. BMC Public Health 19, 146 (2019). 15. Sousa, A., Hill, K. & Dal Poz, M. R. Sub-national assessment of inequality trends Online content in neonatal and child mortality in Brazil. Int. J. Equity Health 9, 21 (2010). Any methods, additional references, Nature Research reporting summaries, source 16. Mohammadi, Y. et al. Measuring Iran’s success in achieving Millennium data, extended data, supplementary information, acknowledgements, peer review Development Goal 4: a systematic analysis of under-5 mortality at national and subnational levels from 1990 to 2015. Lancet Glob. Health 5, e537–e544 (2017). information; details of author contributions and competing interests; and statements of 17. Wang, Y. et al. Under-5 mortality in 2851 Chinese counties, 1996–2012: a data and code availability are available at https://doi.org/10.1038/s41586-019-1545-0. subnational assessment of achieving MDG 4 goals in China. Lancet 387, 273–283 (2016). Received: 28 March 2019; Accepted: 6 August 2019; 18. Horton, R. Offline: in defence of precision public health. Lancet 392, 1504 (2018). Published online 16 October 2019. 19. Takahashi, S., Metcalf, C. J. E., Ferrari, M. J., Tatem, A. J. & Lessler, J. The geography of measles vaccination in the African Great Lakes region. Nat. 1. Reidpath, D. D. & Allotey, P. Infant mortality rate as an indicator of population Commun. 8, 15585 (2017). health. J. Epidemiol. Community Health 57, 344–346 (2003). 20. Utazi, C. E. et al. High resolution age-structured mapping of childhood 2. United Nations General Assembly. United Nations Millennium Declaration: vaccination coverage in low and middle income countries. Vaccine 36, Resolution adopted by the General Assembly. A/RES/55/2 (UN General 1583–1591 (2018). Assembly, 2000). 21. Gething, P. W. et al. Mapping Plasmodium falciparum mortality in Africa between 3. Centers for Disease Control and Prevention. Ten Great Public Health 1990 and 2015. N. Engl. J. Med. 375, 2435–2445 (2016). Achievements—Worldwide, 2001–2010. https://www.cdc.gov/mmwr/preview/ 22. Reiner, R. C. Jr et al. Variation in childhood diarrheal morbidity and mortality in mmwrhtml/mm6024a4.htm (2011). Africa, 2000–2015. N. Engl. J. Med. 379, 1128–1138 (2018). 4. GBD 2017 Mortality Collaborators. Global, regional, and national age-sex- 23. Osgood-Zimmerman, A. et al. Mapping child growth failure in Africa between specific mortality and life expectancy, 1950–2017: a systematic analysis for the 2000 and 2015. Nature 555, 41–47 (2018). Global Burden of Disease Study 2017. Lancet 392, 1684–1735 (2018). 24. Amoah, B., Giorgi, E., Heyes, D. J., van Burren, S. & Diggle, P. J. Geostatistical 5. Golding, N. et al. Mapping under-5 and neonatal mortality in Africa, 2000–15: a modelling of the association between malaria and child growth in Africa. Int. J. baseline analysis for the Sustainable Development Goals. Lancet 390, Health Geogr. 17, 7 (2018). 2171–2182 (2017). 25. Bishai, D. M. et al. Factors contributing to maternal and child mortality 6. Pezzulo, C. et al. Geospatial Modeling of Child Mortality across 27 Countries in reductions in 146 low- and middle-income countries between 1990 and 2010. sub-Saharan Africa. DHS Spatial Analysis Reports No. 13 (USAID, 2016). PLoS ONE 11, e0144908 (2016). 7. Li, Z. et al. Changes in the spatial distribution of the under-five mortality rate: 26. Marmot, M. Social determinants of health inequalities. Lancet 365, 1099–1104 small-area analysis of 122 DHS surveys in 262 subregions of 35 countries in (2005). Africa. PLoS ONE 14, e0210645 (2019). 27. Victora, C. G. et al. Countdown to 2015: a decade of tracking progress for 8. Lawn, J. E., Cousens, S. & Zupan, J. 4 million neonatal deaths: When? Where? maternal, newborn, and child survival. Lancet 387, 2049–2059 (2016). Why? Lancet 365, 891–900 (2005). 28. Rose, G. Strategy of prevention: lessons from cardiovascular disease. Br. Med. J. 9. World Health Organization. SDG 3: Ensure healthy lives and promote wellbeing (Clin. Res. Ed.) 282, 1847–1851 (1981). for all at all ages. https://www.who.int/sdg/targets/en/ (2019). 29. Mackenbach, J. P., Lingsma, H. F., van Ravesteyn, N. T. & Kamphuis, C. B. M. The 10. Dwyer-Lindgren, L. et al. Estimation of district-level under-5 mortality in Zambia population and high-risk approaches to prevention: quantitative estimates of using birth history data, 1980–2010. Spat. SpatioTemporal Epidemiol. 11, their contribution to population health in the Netherlands, 1970–2010. Eur. J. 89–107 (2014). Public Health 23, 909–915 (2013). 1 7 O CT O B ER 2 0 1 9 | V O L 5 7 4 | N A T U RE | 3 5 7 RESEARCH Article 30. Agarwal, S. & Taneja, S. All slums are not equal: child health conditions among 44. Ntigurirwa, P. et al. A health partnership to reduce neonatal mortality in four the urban poor. Indian Pediatr. 42, 233–244 (2005). hospitals in Rwanda. Glob. Health 13, 28 (2017). 31. Farag, T. H. et al. Precisely tracking childhood death. Am. J. Trop. Med. Hyg. 97, 45. Knippenberg, R. et al. Systematic scaling up of neonatal care in countries. 3–5 (2017). Lancet 365, 1087–1098 (2005). 32. Farmer, P. E. et al. Reduced premature mortality in Rwanda: lessons from 46. GBD 2015 Eastern Mediterranean Region Neonatal, Infant, and under-5 success. Br. Med. J. 346, f65 (2013). Mortality Collaborators. Neonatal, infant, and under-5 mortality and morbidity 33. Gurusamy, P. S. R. & Janagaraj, P. D. A success story: the burden of maternal, burden in the Eastern Mediterranean region: findings from the Global Burden neonatal and childhood mortality in Rwanda - critical appraisal of interventions of Disease 2015 study. Int. J. Public Health 63, 63–77 (2018). and recommendations for the future. Afr. J. Reprod. Health 22, 9–16 (2018). 47. Wagner, Z. et al. Armed conflict and child mortality in Africa: a geospatial 34. McGovern, M. E. & Canning, D. Vaccination and all-cause child mortality from analysis. Lancet 392, 857–865 (2018). 1985 to 2011: global evidence from the Demographic and Health Surveys. Am. 48. Mikkelsen, L. et al. A global assessment of civil registration and vital statistics J. Epidemiol. 182, 791–798 (2015). systems: monitoring data quality and progress. Lancet 386, 1395–1406 35. Cheng, J. J., Schuster-Wallace, C. J., Watt, S., Newbold, B. K. & Mente, A. An (2015). ecological quantification of the relationships between water, sanitation and 49. AbouZahr, C. et al. Civil registration and vital statistics: progress in the data infant, child, and maternal mortality. Environ. Health 11, 4 (2012). revolution for counting and accountability. Lancet 386, 1373–1385 (2015). 36. Steketee, R. W. & Campbell, C. C. Impact of national malaria control scale-up 50. Annan, K. Data can help to end malnutrition across Africa. Nature 555, 7 programmes in Africa: magnitude and attribution of effects. Malar. J. 9, 299 (2010). (2018). 37. Kiragu, K., Collins, L., Von Zinkernagel, D. & Mushavi, A. Integrating PMTCT into maternal, newborn, and child health and related services: experiences from the Publisher’s note Springer Nature remains neutral with regard to jurisdictional global plan priority countries. J. Acquir. Immune Defic. Syndr. 75, S36–S42 (2017). claims in published maps and institutional affiliations. 38. GBD 2017 Causes of Death Collaborators. Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980–2017: a systematic analysis for the Global Burden of Disease Open Access This article is licensed under a Creative Commons Study 2017. Lancet 392, 1736–1788 (2018). Attribution 4.0 International License, which permits use, sharing, 39. Pasha, O. et al. A combined community- and facility-based approach to adaptation, distribution and reproduction in any medium or improve pregnancy outcomes in low-resource settings: a Global Network format, as long as you give appropriate credit to the original author(s) and the cluster randomized trial. BMC Med. 11, 215 (2013). source, provide a link to the Creative Commons license, and indicate if 40. Horton, S. et al. Ranking 93 health interventions for low- and middle-income changes were made. The images or other third party material in this article are countries by cost-effectiveness. PLoS ONE 12, e0182951 (2017). included in the article’s Creative Commons license, unless indicated otherwise 41. Simmons, L. E., Rubens, C. E., Darmstadt, G. L. & Gravett, M. G. Preventing in a credit line to the material. If material is not included in the article’s preterm birth and neonatal mortality: exploring the epidemiology, causes, and Creative Commons license and your intended use is not permitted by statutory interventions. Semin. Perinatol. 34, 408–415 (2010). regulation or exceeds the permitted use, you will need to obtain permission 42. Darmstadt, G. L. et al. Evidence-based, cost-effective interventions: how many directly from the copyright holder. To view a copy of this license, visit http:// newborn babies can we save? Lancet 365, 977–988 (2005). creativecommons.org/licenses/by/4.0/. 43. Saugstad, O. D. Reducing global neonatal mortality is possible. Neonatology 99, 250–257 (2011). © The Author(s) 2019 3 5 8 | N A T U RE | V O L 5 7 4 | 1 7 O CT O B ER 2 0 1 9 Article RESEARCH Roy Burstein1,612, Nathaniel J. Henry1,612, Michael L. Collison1, Laurie B. Marczak1, Amber Muluneh378, Ghulam Mustafa379,380, Behnam Nabavizadeh381, Mehdi Naderi382, Ahamarshan Sligar1, Stefanie Watson1, Neal Marquez1, Mahdieh Abbasalizad-Farhangi2, Masoumeh Jayaraman Nagarajan383,384, Azin Nahvijou385, Farid Najafi386, Vinay Nangia387, Duduzile Edith Abbasi3, Foad Abd-Allah4, Amir Abdoli5, Mohammad Abdollahi6, Ibrahim Abdollahpour7,8, Ndwandwe388, Nahid Neamati353, Ionut Negoi389,390, Ruxandra Irina Negoi391,392, Josephine W. Rizwan Suliankatchi Abdulkader9, Michael R. M. Abrigo10, Dilaram Acharya11,12, Oladimeji M. Ngunjiri393, Huong Lan Thi Nguyen394, Long Hoang Nguyen395, Son Hoang Nguyen395, Katie R. Adebayo13, Victor Adekanmbi14, Davoud Adham15, Mahdi Afshari16, Mohammad Aghaali17, Nielsen396,397, Dina Nur Anggraini Ningrum610,611, Yirga Legesse Nirayo398, Molly R. Nixon1, Keivan Ahmadi18, Mehdi Ahmadi19, Ehsan Ahmadpour20, Rushdia Ahmed21,22, Chalachew Chukwudi A. Nnaji388,399, Marzieh Nojomi60,400, Mehdi Noroozi401, Shirin Nosratnejad402, Jean Genet Akal23, Joshua O. Akinyemi24, Fares Alahdab25, Noore Alam26, Genet Melak Alamene27, Jacques Noubiap403, Soraya Nouraei Motlagh234, Richard Ofori-Asenso404,405, Felix Akpojene Kefyalew Addis Alene28,29, Mehran Alijanzadeh30, Cyrus Alinia31, Vahid Alipour32,33, Syed Ogbo406, Kelechi E. Oladimeji303,407, Andrew T. Olagunju408,409, Meysam Olfatifar410, Solomon Mohamed Aljunid34,35, Mohammed J. Almalki36,37, Hesham M. Al-Mekhlafi38,39, Khalid Olum411,412, Bolajoko Olubukunola Olusanya413, Mojisola Morenike Oluwasanu414, Obinna E. Altirkawi40, Nelson Alvis-Guzman41,42, Adeladza Kofi Amegah43, Saeed Amini44, Arianna Onwujekwe415, Eyal Oren416,417, Doris D. V. Ortega-Altamirano418, Alberto Ortiz419,420, Maever Loreche Amit45,46, Zohreh Anbari44, Sofia Androudi47, Mina Anjomshoa48, Fereshteh Osayomwanbo Osarenotor421, Frank B. Osei422,423, Aaron E. Osgood-Zimmerman1, Stanislav Ansari49, Carl Abelardo T. Antonio50,51, Jalal Arabloo33, Zohreh Arefi52, Olatunde Aremu53, S. Otstavnov424,425, Mayowa Ojo Owolabi426, Mahesh P. A.427, Abdol Sattar Pagheh122, Smita Bahram Armoon54,55, Amit Arora56,57, Al Artaman58, Anvar Asadi59, Mehran Asadi-Aliabadi60, Pakhale428, Songhomitra Panda-Jonas429, Animika Pandey121, Eun-Kee Park430, Hadi Amir Ashraf-Ganjouei7,61, Reza Assadi62, Bahar Ataeinia63, Sachin R. Atre64,65, Beatriz Paulina Parsian353, Tahereh Pashaei142, Sangram Kishor Patel431,432, Veincent Christian Filipino Ayala Quintanilla66,67, Martin Amogre Ayanore68, Samad Azari33, Ebrahim Babaee60, Arefeh Pepito433, Alexandre Pereira434,435, Samantha Perkins1, Brandon V. Pickering1, Thomas Babazadeh69, Alaa Badawi70,71, Soghra Bagheri3, Mojtaba Bagherzadeh72, Nafiseh Pilgrim436, Majid Pirestani437, Bakhtiar Piroozi370, Meghdad Pirsaheb3, Oleguer Plana- Baheiraei73,74, Abbas Balouchi75, Aleksandra Barac76,77, Quique Bassat78,79, Bernhard T. Ripoll438, Hadi Pourjafar439,440, Parul Puri302, Mostafa Qorbani441, Hedley Quintana100, Baune80, Mohsen Bayati81, Neeraj Bedi37,82, Ettore Beghi83, Masoud Behzadifar84, Meysam Mohammad Rabiee442, Navid Rabiee72, Amir Radfar443,444, Alireza Rafiei445,446, Fakher Behzadifar85, Yared Belete Belay86, Brent Bell1, Michelle L. Bell87, Dessalegn Ajema Berbada88, Rahim447,448, Zohreh Rahimi449, Vafa Rahimi-Movaghar450, Shadi Rahimzadeh63, Fatemeh Robert S. Bernstein89,90, Natalia V. Bhattacharjee1, Suraj Bhattarai91,92, Zulfiqar A. Bhutta93,94, Rajati451, Sree Bhushan Raju452, Azra Ramezankhani453,454, Chhabi Lal Ranabhat455,456, Davide Ali Bijani95, Somayeh Bohlouli96, Nicholas J. K. Breitborde97,98, Gabrielle Britton99,100, Annie J. Rasella457,458, Vahid Rashedi459, Lal Rawal56,460, Robert C. Reiner Jr1,141, Andre M. N. Browne101, Sharath Burugina Nagaraja102, Reinhard Busse103, Zahid A. Butt104,105, Josip Renzaho461, Satar Rezaei462, Aziz Rezapour33, Seyed Mohammad Riahi463,464, Ana Isabel Car106,107, Rosario Cárdenas108, Carlos A. Castañeda-Orjuela109,110, Ester Cerin111,112, Wagaye Ribeiro465, Leonardo Roever466, Elias Merdassa Roro467,468, Max Roser469, Gholamreza Fentahun Chanie113, Pranab Chatterjee114, Dinh-Toi Chu115, Cyrus Cooper116,117, Vera M. Roshandel324,470, Daem Roshani471, Ali Rostami472, Enrico Rubagotti473,474, Salvatore Costa118, Koustuv Dalal119,120, Lalit Dandona1,121, Rakhi Dandona1,121, Farah Daoud1, Ahmad Rubino475, Siamak Sabour463, Nafis Sadat1, Ehsan Sadeghi59, Reza Saeedi476, Yahya Safari3, Daryani122, Rajat Das Gupta21, Ian Davis1, Nicole Davis Weaver1, Dragos Virgil Davitoiu123,124, Roya Safari-Faramani477, Mahdi Safdarian450,478, Amirhossein Sahebkar479,480, Mohammad Jan-Walter De Neve125, Feleke Mekonnen Demeke126, Gebre Teklemariam Demoz127,128, Kebede Reza Salahshoor481, Nasir Salam482, Payman Salamati450,483, Farkhonde Salehi484, Saleh Salehi Deribe129,130, Rupak Desai131, Aniruddha Deshpande1, Hanna Demelash Desyibelew132, Sagnik Zahabi485,486, Yahya Salimi386, Hamideh Salimzadeh324, Joshua A. Salomon487, Evanson Dey133, Samath Dhamminda Dharmaratne1,134, Meghnath Dhimal135, Daniel Diaz136,137, Leila Zondani Sambala388, Abdallah M. Samy488, Milena M. Santric Milicevic489, Bruno Piassi Sao Doshmangir138, Andre R. Duraes139,140, Laura Dwyer-Lindgren1,141, Lucas Earl1, Roya Jose490, Sivan Yegnanarayana Iyer Saraswathy491,492, Rodrigo Sarmiento-Suárez493, Benn Ebrahimi142, Soheil Ebrahimpour69, Andem Effiong143, Aziz Eftekhari144,145, Elham Ehsani- Sartorius141,494, Brijesh Sathian495,496, Sonia Saxena497, Alyssa N. Sbarra1, Lauren E. Schaeffer1, Chimeh146, Iman El Sayed147, Maysaa El Sayed Zaki148, Maha El Tantawi149,150, Ziad El- David C. Schwebel498, Sadaf G. Sepanlou324,325, Seyedmojtaba Seyedmousavi499,500, Faramarz Khatib151, Mohammad Hassan Emamian152, Shymaa Enany153, Sharareh Eskandarieh7, Shaahmadi501, Masood Ali Shaikh502, Mehran Shams-Beyranvand63,503, Amir Shamshirian504, Oghenowede Eyawo154,155, Maha Ezalarab1, Mahbobeh Faramarzi156, Mohammad Fareed157, Morteza Shamsizadeh505, Kiomars Sharafi3, Mehdi Sharif506,507, Mahdi Sharif-Alhoseini450, Roghiyeh Faridnia158, Andre Faro159, Ali Akbar Fazaeli160, Mehdi Fazlzadeh161,162, Netsanet Hamid Sharifi508, Jayendra Sharma509, Rajesh Sharma510, Aziz Sheikh511,512, Chloe Shields1, Fentahun163, Seyed-Mohammad Fereshtehnejad164,165, João C. Fernandes166, Irina Filip167,168, Mika Shigematsu513, Rahman Shiri514, Ivy Shiue515, Kerem Shuval198, Tariq J. Siddiqi275, João Florian Fischer169, Nataliya A. Foigt170, Masoud Foroutan171, Joel Msafiri Francis172, Takeshi Pedro Silva118, Jasvinder A. Singh516,517, Dhirendra Narain Sinha518,519, Malede Mequanent Fukumoto173,174, Nancy Fullman1, Silvano Gallus175, Destallem Gebremedhin Gebre176,177, Sisay378,520, Solomon Sisay521, Karen Sliwa403, David L. Smith1,141, Ranjani Somayaji522,523, Tsegaye Tewelde Gebrehiwot178, Gebreamlak Gebremedhn Gebremeskel179,180, Bradford D. Moslem Soofi524, Joan B. Soriano525,526, Chandrashekhar T. Sreeramareddy527, Agus Gessner181,182, Birhanu Geta183, Peter W. Gething101, Reza Ghadimi184, Keyghobad Ghadiri3, Sudaryanto528, Mu’awiyyah Babale Sufiyan529, Bryan L. Sykes530, P. N. Sylaja531, Rafael Mahsa Ghajarzadeh185, Ahmad Ghashghaee186, Paramjit Singh Gill187, Tiffany K. Gill188, Nick Tabarés-Seisdedos532,533, Karen M. Tabb534, Takahiro Tabuchi535, Nuno Taveira536,537, Golding609, Nelson G. M. Gomes189,190, Philimon N. Gona191, Sameer Vali Gopalani192,193, Mohamad-Hani Temsah538,539, Abdullah Sulieman Terkawi540,541, Zemenu Tadesse Tessema378, Giuseppe Gorini194, Bárbara Niegia Garcia Goulart195, Nicholas Graetz1, Felix Greaves196,197, Kavumpurathu Raman Thankappan542, Sathish Thirunavukkarasu543, Quyen G. To544, Manfred S. Green198, Yuming Guo199,200, Arvin Haj-Mirzaian201,202, Arya Haj-Mirzaian201,203, Marcos Roberto Tovani-Palone545, Bach Xuan Tran546, Khanh Bao Tran547,548, Irfan Ullah549,550, Brian James Hall204, Samer Hamidi205, Hamidreza Haririan206, Josep Maria Haro207,208, Milad Muhammad Shariq Usman275, Olalekan A. Uthman551, Amir Vahedian-Azimi552,553, Pascual R. Hasankhani209, Edris Hasanpoor210, Amir Hasanzadeh211,212, Hadi Hassankhani213,214, Hamid Valdez554,555, Job F. M. van Boven556,557, Tommi Juhani Vasankari558, Yasser Vasseghian462, Yimam Hassen215,216, Mohamed I. Hegazy4, Delia Hendrie217, Fatemeh Heydarpour218, Thomas Yousef Veisani559, Narayanaswamy Venketasubramanian560,561, Francesco S. Violante562,563, R. Hird199,219, Chi Linh Hoang220, Gillian Hollerich1, Enayatollah Homaie Rad221,222, Mojtaba Sergey Konstantinovitch Vladimirov564,565, Vasily Vlassov566, Theo Vos1,141, Giang Thu Vu220, Hoseini-Ghahfarokhi223, Naznin Hossain22,224, Mostafa Hosseini225, Mehdi Hosseinzadeh33,226, Isidora S. Vujcic77, Yasir Waheed567, Jon Wakefield568, Haidong Wang1,141, Yafeng Wang569, Mihaela Hostiuc123,227, Sorin Hostiuc228,229, Mowafa Househ230,231, Mohamed Hsairi232, Yuan-Pang Wang570, Joseph L. Ward571, Robert G. Weintraub572,573, Kidu Gidey Olayinka Stephen Ilesanmi233, Mohammad Hasan Imani-Nasab234, Usman Iqbal235, Seyed Sina Weldegwergs398, Girmay Teklay Weldesamuel574, Ronny Westerman575, Charles Shey Naghibi Irvani236, Nazrul Islam237,238, Sheikh Mohammed Shariful Islam239,240, Mikk Wiysonge576,577, Dawit Zewdu Wondafrash578,579, Lauren Woyczynski1, Ai-Min Wu580, Jürisson241, Nader Jafari Balalami242, Amir Jalali243, Javad Javidnia244, Achala Upendra Gelin Xu581, Abbas Yadegar355, Tomohide Yamada582, Vahid Yazdi-Feyzabadi583,584, Jayatilleke245,246, Ensiyeh Jenabi247, John S. Ji248, Yash B. Jobanputra249, Kimberly Johnson1, Christopher Sabo Yilgwan585,586, Paul Yip587,588, Naohiro Yonemoto589, Javad Yoosefi Lebni331, Jost B. Jonas250,251, Zahra Jorjoran Shushtari252, Jacek Jerzy Jozwiak253,254, Ali Kabir255, Amaha Mustafa Z. Younis590,591, Mahmoud Yousefifard592, Hebat-Allah Salah A. Yousof593, Kahsay256, Hamed Kalani257,258, Rohollah Kalhor259, Manoochehr Karami260, Surendra Chuanhua Yu569,594, Hasan Yusefzadeh595, Erfan Zabeh596,597, Telma Zahirian Moghadam33,598, Karki261,262, Amir Kasaeian263,264, Nicholas J. Kassebaum1,265, Peter Njenga Keiyoro266, Grant Sojib Bin Zaman599,600, Mohammad Zamani601, Hamed Zandian598,602, Alireza Zangeneh603, Rodgers Kemp1,267, Roghayeh Khabiri268,269, Yousef Saleh Khader270, Morteza Abdullatif Taddese Alemu Zerfu604,605, Yunquan Zhang606,607, Arash Ziapour331, Sanjay Zodpey608, Khafaie271, Ejaz Ahmad Khan272, Junaid Khan273, Muhammad Shahzeb Khan274,275, Young-Ho Christopher J. L. Murray1,141,613 & Simon I. Hay1,141,613* Khang276,277, Khaled Khatab278,279, Amir Khater280, Mona M. Khater281, Alireza Khatony3, Mohammad Khazaei282, Salman Khazaei260, Maryam Khazaei-Pool283, Jagdish Khubchandani284, Neda Kianipour3,285, Yun Jin Kim286, Ruth W. Kimokoti287, Damaris K. Kinyoki1,141, Adnan Kisa288,289, Sezer Kisa290, Tufa Kolola291, Soewarta Kosen292, Parvaiz A. 1 Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, USA. Koul293, Ai Koyanagi79,294, Moritz U. G. Kraemer295,296, Kewal Krishan297, Kris J. Krohn1, 2 Department of Nutrition, Tabriz University of Medical Sciences, Tabriz, Iran. 3Kermanshah Nuworza Kugbey298,299, G. Anil Kumar121, Manasi Kumar300,301, Pushpendra Kumar302, University of Medical Sciences, Kermanshah, Iran. 4Department of Neurology, Cairo University, Desmond Kuupiel303,304, Ben Lacey305,306, Sheetal D. Lad307, Faris Hasan Lami308, Anders O. Cairo, Egypt. 5Department of Parasitology and Mycology, Jahrom University of Medical Larsson309,310, Paul H. Lee311, Mostafa Leili282, Aubrey J. Levine1, Shanshan Li199, Lee-Ling Sciences, Jahrom, Iran. 6The Institute of Pharmaceutical Sciences, Tehran University of Medical Lim312,313, Stefan Listl314,315, Joshua Longbottom316, Jaifred Christian F. Lopez45,317, Stefan Sciences, Tehran, Iran. 7Multiple Sclerosis Research Center, Tehran University of Medical Lorkowski318,319, Sameh Magdeldin320,321, Hassan Magdy Abd El Razek322, Muhammed Magdy Sciences, Tehran, Iran. 8Department of Epidemiology, Arak University of Medical Sciences, Arak, Abd El Razek323, Azeem Majeed196, Afshin Maleki142, Reza Malekzadeh324,325, Deborah Iran. 9Department of Public Health, Ministry of Health, Riyadh, Saudi Arabia. 10Research Carvalho Malta326, Abdullah A. Mamun327, Navid Manafi328,329, Ana-Laura Manda330, Morteza Department, Philippine Institute for Development Studies, Quezon City, The Philippines. Mansourian331, Francisco Rogerlândio Martins-Melo332, Anthony Masaka333, Benjamin Ballard 11 Department of Preventive Medicine, Dongguk University, Gyeongju, South Korea. Massenburg334, Pallab K. Maulik335,336, Benjamin K. Mayala1, Mohsen Mazidi337, Martin 12 Department of Community Medicine, Kathmandu University, Devdaha, Nepal. 13College of McKee338, Ravi Mehrotra339, Kala M. Mehta340, Gebrekiros Gebremichael Meles88, Walter Medicine, University College Hospital, Ibadan, Nigeria. 14School of Medicine, Cardiff University, Mendoza341, Ritesh G. Menezes342, Atte Meretoja343,344, Tuomo J. Meretoja345,346, Tomislav Cardiff, UK. 15School of Health, Ardabil University of Medical Science, Ardabil, Iran. Mestrovic347,348, Ted R. Miller217,349, Molly K. Miller-Petrie1, Edward J. Mills350, George J. 16 Department of Community Medicine, Zabol University of Medical Sciences, Zabol, Iran. Milne351, G. K. Mini352, Seyed Mostafa Mir353,354, Hamed Mirjalali355, Erkin M. 17 Department of Epidemiology and Biostatistics, Qom University of Medical Sciences, Qom, Mirrakhimov356,357, Efat Mohamadi358, Dara K. Mohammad359,360, Aso Mohammad Iran. 18School of Pharmacy, University of Lincoln, Lincoln, UK. 19Environmental Technologies Darwesh361, Naser Mohammad Gholi Mezerji362, Ammas Siraj Mohammed363, Shafiu Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. 20Department Mohammed364,365, Ali H. Mokdad1,141, Mariam Molokhia366, Lorenzo Monasta367, Yoshan of Parasitology and Mycology, Tabriz University of Medical Sciences, Tabriz, Iran. 21James P. Moodley303, Mahmood Moosazadeh368, Ghobad Moradi369,370, Masoud Moradi3,59, Yousef Grant School of Public Health, Brac University, Dhaka, Bangladesh. 22Health Systems and Moradi371, Maziar Moradi-Lakeh60, Mehdi Moradinazar59, Paula Moraga372, Lidia Morawska373, Population Studies Division, International Centre for Diarrhoeal Disease Research Bangladesh, Abbas Mosapour353,374, Seyyed Meysam Mousavi375, Ulrich Otto Mueller376,377, Atalay Goshu Dhaka, Bangladesh. 23Department of Medical Laboratory Science, Bahir Dar University, Bahir RESEARCH Article Dar, Ethiopia. 24Epidemiology and Medical Statistics, University of Ibadan, Ibadan, Nigeria. National Institute of Health, Bogota, Colombia. 110Epidemiology and Public Health Evaluation 25 Evidence Based Practice Center, Mayo Clinic Foundation for Medical Education and Research, Group, National University of Colombia, Bogota, Colombia. 111Mary Mackillop Institute for Rochester, MN, USA. 26Prevention Division, Queensland Health, Herston, Queensland, Australia. Health Research, Australian Catholic University, Melbourne, Victoria, Australia. 112School of 27 School of Health Sciences, Madda Walabu University, Bale Goba, Ethiopia. 28Institute of Public Public Health, University of Hong Kong, Hong Kong, China. 113Department of Obstetrics and Health, University of Gondar, Gondar, Ethiopia. 29Research School of Population Health, Gynaecology, University of Gondar, Gondar, Ethiopia. 114Division of Epidemiology, National Australian National University, Canberra, Australian Capital Territory, Australia. 30Qazvin Institute of Cholera and Enteric Diseases, Kolkata, India. 115Faculty of Biology, Hanoi National University of Medical Sciences, Qazvin, Iran. 31Department of Health Care Management and University of Education, Hanoi, Vietnam. 116Department of Rheumatology, University of Oxford, Economics, Urmia University of Medical Science, Urmia, Iran. 32Health Economics Department, Oxford, UK. 117Medical Research Council Lifecourse Epidemiology Unit, University of Iran University of Medical Sciences, Tehran, Iran. 33Health Management and Economics Southampton, Southampton, UK. 118Applied Molecular Biosciences Unit (UCIBIO), University of Research Center, Iran University of Medical Sciences, Tehran, Iran. 34Department of Health Porto, Porto, Portugal. 119Institute of Public Health Kalyani, Kalyani, India. 120School of Health Policy and Management, Kuwait University, Safat, Kuwait. 35International Centre for Casemix Science, Orebro University, Orebro, Sweden. 121Public Health Foundation of India, Gurugram, and Clinical Coding, National University of Malaysia, Bandar Tun Razak, Malaysia. 36Faculty of India. 122Toxoplasmosis Research Center, Mazandaran University of Medical Sciences, Sari, Iran. Public Health and Tropical Medicine, Jazan University, Jazan, Saudi Arabia. 37Jazan University, 123 Department of General Surgery, Carol Davila University of Medicine and Pharmacy, Jazan, Saudi Arabia. 38Medical Research Center, Jazan University, Jazan, Saudi Arabia. Bucharest, Romania. 124Department of Surgery, Clinical Emergency Hospital St Pantelimon, 39 Department of Medical Parasitology, Sana’a University, Sana’a, Yemen. 40King Saud Bucharest, Romania. 125Heidelberg Institute of Global Health (HIGH), Heidelberg University, University, Riyadh, Saudi Arabia. 41Research Group in Health Economics, Universidad de Heidelberg, Germany. 126Bahir Dar University, Bahir Dar, Ethiopia. 127School of Pharmacy, Cartagena, Cartagena, Colombia. 42Research Group in Hospital Management and Health Aksum University, Aksum, Ethiopia. 128Addis Ababa University, Addis Ababa, Ethiopia. 129School Policies, Universidad de la Costa, Barranquilla, Colombia. 43Biomedical Science, University of of Public Health, Addis Ababa University, Addis Ababa, Ethiopia. 130Department of Global Cape Coast, Cape Coast, Ghana. 44Health Services Management Department, Arak University of Health and Infection, Brighton and Sussex Medical School, Brighton, UK. 131Division of Medical Sciences, Arak, Iran. 45Department of Epidemiology and Biostatistics, University of the Cardiology, Atlanta Veterans Affairs Medical Center, Decatur, GA, USA. 132Public Health Philippines Manila, Manila, The Philippines. 46Online Programs for Applied Learning, Johns Nutrition, Bahir Dar University, Bahir Dar, Ethiopia. 133Centre for Atmospheric Sciences, Indian Hopkins University, Baltimore, MD, USA. 47Department of Medicine, University of Thessaly, Institute of Technology Delhi, New Delhi, India. 134Department of Community Medicine, Volos, Greece. 48Social Determinants of Health Research Center, Rafsanjan University of University of Peradeniya, Peradeniya, Sri Lanka. 135Health Research Section, Nepal Health Medical Sciences, Rafsanjan, Iran. 49Research Center for Evidence Based Medicine-Health Research Council, Kathmandu, Nepal. 136Center of Complexity Sciences, National Autonomous Management and Safety Promotion Research Institute, Tabriz University of Medical Sciences, University of Mexico, Mexico City, Mexico. 137Facultad de Medicina Veterinaria y Zootecnia, Tabriz, Iran. 50Department of Health Policy and Administration, University of the Philippines Autonomous University of Sinaloa, Culiacan Rosales, Mexico. 138Department of Health Policy Manila, Manila, The Philippines. 51Department of Applied Social Sciences, Hong Kong and Economy, Tabriz University of Medical Sciences, Tabriz, Iran. 139School of Medicine, Federal Polytechnic University, Hong Kong, China. 52Department of Health Promotion and Education, University of Bahia, Salvador, Brazil. 140Diretoria Médica, Roberto Santos General Hospital, Tehran University of Medical Sciences, Tehran, Iran. 53School of Health Sciences, Birmingham Salvador, Brazil. 141Department of Health Metrics Sciences, School of Medicine, University of City University, Birmingham, UK. 54School of Nursing and Midwifery, Saveh University of Washington, Seattle, WA, USA. 142Environmental Health Research Center, Kurdistan University Medical Sciences, Saveh, Iran. 55Social Determinants of Health Research Center, Saveh of Medical Sciences, Sanandaj, Iran. 143Clinical Epidemiology and Biostatistics, University of University of Medical Sciences, Saveh, Iran. 56School of Science and Health, Western Sydney Newcastle, Newcastle, New South Wales, Australia. 144Department of Toxicology and University, Penrith, New South Wales, Australia. 57Oral Health Services, Sydney Local Health Pharmacology, Tabriz University of Medical Sciences, Tabriz, Iran. 145Department of Basic District, Sydney, New South Wales, Australia. 58Department of Community Health Sciences, Sciences, Maragheh University of Medical Sciences, Maragheh, Iran. 146National Institute for University of Manitoba, Winnipeg, Manitoba, Canada. 59Research Center for Environmental Health Researchers, Tehran University of Medical Sciences, Tehran, Iran. 147Medical Research Determinants of Health, Kermanshah University of Medical Sciences, Kermanshah, Iran. Institute, Alexandria University, Alexandria, Egypt. 148Department of Clinical Pathology, 60 Preventive Medicine and Public Health Research Center, Iran University of Medical Sciences, Mansoura University, Mansoura, Egypt. 149Pediatric Dentistry and Dental Public Health, Tehran, Iran. 61Faculty of Medicine, Tehran University of Medical Sciences, Tehran, Iran. Alexandria University, Alexandria, Egypt. 150Preventive Dental Sciences, Imam Abdulrahman 62 Education Development Center, Mashhad University of Medical Sciences, Mashhad, Iran. Bin Faisal University, Dammam, Saudi Arabia. 151Department of Public Health Sciences, 63 Non-communicable Diseases Research Center, Tehran University of Medical Sciences, Tehran, Karolinska Institutet, Stockholm, Sweden. 152Ophthalmic Epidemiology Research Center, Iran. 64Center for Clinical Global Health Education, Johns Hopkins University, Baltimore, MD, Shahroud University of Medical Sciences, Shahroud, Iran. 153Department of Microbiology and USA. 65Dr D. Y. Patil Medical College, Pune, India. 66The Judith Lumley Centre, La Trobe Immunology, Suez Canal University, Ismailia, Egypt. 154Epidemiology and Population Health, University, Melbourne, Victoria, Australia. 67General Office for Research and Technological British Columbia Centre for Excellence in HIV/AIDS, Vancouver, British Columbia, Canada. Transfer, Peruvian National Institute of Health, Lima, Peru. 68Department of Family and 155 Faculty of Health Sciences, Simon Fraser University, Burnaby, British Columbia, Canada. Community Health, University of Health and Allied Sciences, Ho, Ghana. 69Center for Infectious 156 Babol University of Medical Sciences, Babol, Iran. 157College of Medicine, Imam Muhammad Diseases Research, Babol, Iran. 70Public Health Risk Sciences Division, Public Health Agency of Ibn Saud Islamic University, Riyadh, Saudi Arabia. 158Department of Parasitology, Mazandaran Canada, Toronto, Ontario, Canada. 71Department of Nutritional Sciences, University of Toronto, University of Medical Sciences, Sari, Iran. 159Department of Psychology, Federal University of Toronto, Ontario, Canada. 72Department of Chemistry, Sharif University of Technology, Tehran, Sergipe, Sao Cristovao, Brazil. 160Social Determinants of Health Research Center, Hamadan Iran. 73Tissue Engineering and Applied Cell Sciences Division, Tarbiat Modares University, University of Medical Sciences, Hamadan, Iran. 161Environmental Health Engineering, Tehran Tehran, Iran. 74Division of Diseases, Advanced Technologies Research Group, Tehran, Iran. University of Medical Sciences, Tehran, Iran. 162Department of Environmental Health 75 School of Nursing and Midwifery, Iran University of Medical Sciences, Tehran, Iran. 76Clinic for Engineering, Ardabil University of Medical Science, Ardabil, Iran. 163Department of Public Infectious and Tropical Diseases, Clinical Center of Serbia, Belgrade, Serbia. 77Faculty of Health Nutrition, Bahir Dar University, Bahir Dar, Ethiopia. 164Department of Neurobiology, Medicine, University of Belgrade, Belgrade, Serbia. 78Barcelona Institute for Global Health, Karolinska Institutet, Stockholm, Sweden. 165Division of Neurology, University of Ottawa, Ottawa, University of Barcelona, Barcelona, Spain. 79Catalan Institution for Research and Advanced Ontario, Canada. 166Center for Biotechnology and Fine Chemistry, Catholic University of Studies (ICREA), Barcelona, Spain. 80Department of Psychiatry, Melbourne Medical School, Portugal, Porto, Portugal. 167Psychiatry Department, Kaiser Permanente, Fontana, CA, USA. Melbourne, Victoria, Australia. 81Health Human Resources Research Center, Shiraz University of 168 Department of Health Sciences, A.T. Still University, Mesa, AZ, USA. 169Department of Public Medical Sciences, Shiraz, Iran. 82Department of Community Medicine, Gandhi Medical College Health Medicine, Bielefeld University, Bielefeld, Germany. 170Institute of Gerontology, National Bhopal, Bhopal, India. 83Department of Neuroscience, Mario Negri Institute for Pharmacological Academy of Medical Sciences of Ukraine, Kyiv, Ukraine. 171Abadan School of Medical Sciences, Research, Milan, Italy. 84Social Determinants of Health Research Center, Lorestan University of Abadan, Iran. 172Clinical Medicine and Wits Reproductive Health and HIV Institute, University of Medical Sciences, Khorramabad, Iran. 85Hepatitis Research Center, Lorestan University of the Witwatersrand, Johannesburg, South Africa. 173Gene Expression & Regulation Program, Medical Sciences, Khorramabad, Iran. 86Pharmacoepidemiology and Social Pharmacy, Mekelle Cancer Institute (W.I.A.), Philadelphia, PA, USA. 174Department of Dermatology, Kobe University, University, Mekelle, Ethiopia. 87School of Forestry and Environmental Studies, Yale University, Kobe, Japan. 175Department of Environmental Health Science, Mario Negri Institute for New Haven, CT, USA. 88Department of Public Health, Arba Minch University, Arba Minch, Pharmacological Research, Milan, Italy. 176Mekelle University, Mekelle, Ethiopia. 177Dr Tewelde Ethiopia. 89Hubert Department of Global Health, Emory University, Atlanta, GA, USA. Legesse Health Sciences College, Mekelle, Ethiopia. 178Department of Epidemiology, Jimma 90 Department of Global Health, University of South Florida, Tampa, FL, USA. 91London School of University, Jimma, Ethiopia. 179School of Nursing, Mekelle University, Mekelle, Ethiopia. Hygiene & Tropical Medicine, London, UK. 92Nepal Academy of Science & Technology, Patan, 180 Nursing Department, Aksum University, Aksum, Ethiopia. 181Vaccines Department, Pfizer, Nepal. 93The Centre for Global Child Health, Hospital for Sick Children, University of Toronto, Collegeville, PA, USA. 182Agency of Preventive Medicine, Paris, France. 183Department of Toronto, Ontario, Canada. 94Center of Excellence in Women and Child Health, Aga Khan Pharmacy, Wollo University, Dessie, Ethiopia. 184Health Research Institute, Babol University of University, Karachi, Pakistan. 95Social Determinants of Health Research Center, Babol University Medical Sciences, Babol, Iran. 185Department of Neurology, Tehran University of Medical of Medical Sciences, Babol, Iran. 96Department of Veterinary Medicine, Karaj Islamic Azad Sciences, Tehran, Iran. 186Department of Health Services Management, Iran University of University, Kermanshah, Iran. 97Department of Psychology, Ohio State University, Columbus, Medical Sciences, Tehran, Iran. 187Unit of Academic Primary Care, University of Warwick, OH, USA. 98Psychiatry and Behavioral Health Department, Ohio State University, Columbus, OH, Coventry, UK. 188Adelaide Medical School, University of Adelaide, Adelaide, South Australia, USA. 99Neuroscience Department, Institute for Scientific Research and High Technology Australia. 189Department of Chemistry, University of Porto, Porto, Portugal. 190REQUIMTE/LAQV, Services, City of Knowledge, Panama. 100Gorgas Memorial Institute for Health Studies, Panama, Porto, Portugal. 191Nursing and Health Sciences Department, University of Massachusetts Panama. 101Big Data Institute, Li Ka Shing Centre for Health Information and Discovery, Boston, Boston, MA, USA. 192Department of Biostatistics and Epidemiology, University of University of Oxford, Oxford, UK. 102Department of Community Medicine, Employees’ State Oklahoma, Oklahoma City, OK, USA. 193Department of Health and Social Affairs, Government of Insurance Model Hospital, Bangalore, India. 103Department for Health Care Management, the Federated States of Micronesia, Palikir, Federated States of Micronesia. 194Occupational and Technical University of Berlin, Berlin, Germany. 104School of Population and Public Health, Environmental Epidemiology Section, Cancer Prevention and Research Institute, Florence, Italy. University of British Columbia, Vancouver, British Columbia, Canada. 105Al Shifa School of 195 Postgraduate Program in Epidemiology, Federal University of Rio Grande do Sul, Porto Public Health, Al Shifa Trust Eye Hospital, Rawalpindi, Pakistan. 106Centre for Population Health Alegre, Brazil. 196Department of Primary Care and Public Health, Imperial College London, Sciences, Nanyang Technological University, Singapore, Singapore. 107Global Ehealth Unit, London, UK. 197Health Improvement Directorate, Public Health England, London, UK. 198School Imperial College London, London, UK. 108Department of Population and Health, Metropolitan of Public Health, University of Haifa, Haifa, Israel. 199School of Public Health and Preventive Autonomous University, Mexico City, Mexico. 109Colombian National Health Observatory, Medicine, Monash University, Melbourne, Victoria, Australia. 200Department of Epidemiology Article RESEARCH and Biostatistics, Zhengzhou University, Zhengzhou, China. 201Department of Pharmacology, 285 School of Public Health, Kermanshah University of Medical Sciences, Kermanshah, Iran. Tehran University of Medical Sciences, Tehran, Iran. 202Obesity Research Center, Research 286 School of Medicine, Xiamen University Malaysia, Sepang, Malaysia. 287Department of Institute for Endocrine Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran. Nutrition, Simmons College, Boston, MA, USA. 288Department of Health Management and 203 Department of Radiology, Johns Hopkins University, Baltimore, MD, USA. 204Global and Health Economics, Kristiania University College, Oslo, Norway. 289Department of Health Community Mental Health Research Group, University of Macau, Macao, China. 205School of Services Policy and Management, University of South Carolina, Columbia, SC, USA. 290Nursing Health and Environmental Studies, Hamdan Bin Mohammed Smart University, Dubai, United and Health Promotion, Oslo Metropolitan University, Oslo, Norway. 291Department of Public Arab Emirates. 206Tabriz University of Medical Sciences, Tabriz, Iran. 207Biomedical Research Health, Debre Berhan University, Debre Berhan, Ethiopia. 292Independent Consultant, Jakarta, Networking Center for Mental Health Network (CIBERSAM), Madrid, Spain. 208Research and Indonesia. 293Department of Internal and Pulmonary Medicine, Sheri Kashmir Institute of Development Unit, San Juan de Dios Sanitary Park, Sant Boi De Llobregat, Spain. 209School of Medical Sciences, Srinagar, India. 294CIBERSAM, San Juan de Dios Sanitary Park, Sant Boi De Nutrition and Food Sciences, Tabriz University of Medical Sciences, Tabriz, Iran. 210Healthcare Llobregat, Spain. 295Department of Zoology, University of Oxford, Oxford, UK. 296Medical School, Management, Maragheh University of Medical Sciences, Maragheh, Iran. 211Department of Harvard University, Boston, MA, USA. 297Department of Anthropology, Panjab University, Microbiology, Tehran University of Medical Sciences, Tehran, Iran. 212Department of Chandigarh, India. 298Family and Community Health, University of Health and Allied Sciences, Microbiology, Maragheh University of Medical Sciences, Maragheh, Iran. 213School of Nursing Ho, Ghana. 299Psychology and Health Promotion, University of Kwazulu-Natal, Durban, South and Midwifery Tabriz University of Medical Sciences, Tabriz, Iran. 214Independent Consultant, Africa. 300Department of Psychiatry, University of Nairobi, Nairobi, Kenya. 301Department of Tabriz, Iran. 215Public Health Department, Mizan-Tepi University, Teppi, Ethiopia. 216Unit of Psychology, University College London, London, UK. 302International Institute for Population Epidemiology and Social Medicine, University Hospital Antwerp, Antwerp, Belgium. 217School of Sciences, Mumbai, India. 303Department of Public Health Medicine, University of Kwazulu- Public Health, Curtin University, Bentley, Western Australia, Australia. 218Medical Biology Natal, Durban, South Africa. 304Nursing, St John of God Hospital, Duayaw Nkwanta, Ghana. Research Center, Kermanshah University of Medical Sciences, Kermanshah, Iran. 219Population 305 Nuffield Department of Population Health, University of Oxford, Oxford, UK. 306Oxford Health, Baker Heart and Diabetes Institute, Melbourne, Victoria, Australia. 220Center of Biomedical Research Centre, National Institute for Health Research (NIHR), Oxford, UK. 307 Excellence in Behavioral Medicine, Nguyen Tat Thanh University, Ho Chi Minh, Vietnam. Department of Pediatrics, Post Graduate Institute of Medical Education and Research, 221 Social Determinants of Health Research Center, Guilan Road Trauma Research Center, Chandigarh, India. 308Department of Community and Family Medicine, Academy of Medical Guilan University of Medical Sciences, Rasht, Iran. 222Guilan Road Trauma Research Center, Science, Baghdad, Iraq. 309Department of Medical Sciences, Uppsala University, Uppsala, Guilan University of Medical Sciences, Rasht, Iran. 223Radiology and Nuclear Medicine Sweden. 310Department of Clinical Chemistry and Pharmacology, Uppsala University Hospital, Department, Kermanshah University of Medical Sciences, Kermanshah, Iran. 224Department of Uppsala, Sweden. 311School of Nursing, Hong Kong Polytechnic University, Hong Kong, China. Pharmacology and Therapeutics, University of Dhaka, Dhaka, Bangladesh. 225Department of 312 Department of Medicine, University of Malaya, Kuala Lumpur, Malaysia. 313Department of Epidemiology and Biostatistics, Tehran University of Medical Sciences, Tehran, Iran. Medicine and Therapeutics, The Chinese University of Hong Kong, Shatin, China. 226 314 Computer Science Department, University of Human Development, Sulaimaniyah, Iraq. Department of Dentistry, Radboud University, Nijmegen, The Netherlands. 315Section for 227 Department of Internal Medicine, Bucharest Emergency Hospital, Bucharest, Romania. Translational Health Economics, Heidelberg University Hospital, Heidelberg, Germany. 228 316 Faculty of Dentistry, Carol Davila University of Medicine and Pharmacy, Bucharest, Romania. Department of Vector Biology, Liverpool School of Tropical Medicine, Liverpool, UK. 229 317 Clinical Legal Medicine, National Institute of Legal Medicine Mina Minovici, Bucharest, Alliance for Improving Health Outcomes, Quezon City, The Philippines. 318Institute of Romania. 230Division of Information and Computing Technology, Hamad Bin Khalifa University, Nutrition, Friedrich Schiller University Jena, Jena, Germany. 319Competence Cluster for Doha, Qatar. 231Qatar Foundation for Education, Science and Community Development, Doha, Nutrition and Cardiovascular Health (NUTRICARD), Jena, Germany. 320Physiology Department, Qatar. 232Faculty of Medicine Tunis, Medicine School of Tunis, Baab Saadoun, Tunisia. Suez Canal University, Ismailia, Egypt. 321Proteomics and Metabolomics Unit, Suez Canal 233 Department of Community Medicine, University of Ibadan, Ibadan, Nigeria. 234Department of University, Ismailia, Egypt. 322Department of Cardiology, Damietta University, Damietta, Egypt. Public Health, Lorestan University of Medical Sciences, Khorramabad, Iran. 235Global Health 323 Ophthalmology Department, Aswan Faculty of Medicine, Aswan, Egypt. 324Digestive Diseases and Development Department, Taipei Medical University, Taipei City, Taiwan. 236Research Research Institute, Tehran University of Medical Sciences, Tehran, Iran. 325Non-communicable Institute for Endocrine Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran. Diseases Research Center, Shiraz University of Medical Sciences, Shiraz, Iran. 326Department of 237 MRC Epidemiology Unit, University of Cambridge, Cambridge, UK. 238Harvard University, Maternal and Child Nursing and Public Health, Federal University of Minas Gerais, Belo Boston, MA, USA. 239Institute for Physical Activity and Nutrition, Deakin University, Burwood, Horizonte, Brazil. 327Institute for Social Science Research, The University of Queensland, Victoria, Australia. 240Sydney Medical School, University of Sydney, Sydney, New South Wales, Brisbane, Queensland, Australia. 328Ophthalmology Department, Iran University of Medical Australia. 241Institute of Family Medicine and Public Health, University of Tartu, Tartu, Estonia. Sciences, Tehran, Iran. 329Department Ophthalmology, University of Manitoba, Winnipeg, 242 Psychosis Department, Babol Nushirvani University of Technology, Babol, Iran. 243Psychiatric Manitoba, Canada. 330Surgery Department, Emergency University Hospital Bucharest, Department, Kermanshah University of Medical Sciences, Kermanshah, Iran. 244Department of Bucharest, Romania. 331Department of Health Education and Health Promotion, Iran University Medical Mycology, Mazandaran University of Medical Sciences, Sari, Iran. 245Faculty of of Medical Sciences, Tehran, Iran. 332Campus Caucaia, Federal Institute of Education, Science Graduate Studies, University of Colombo, Colombo, Sri Lanka. 246Institute of Medicine, and Technology of Ceará, Caucaia, Brazil. 333Faculty of Health and Education, Botho University- University of Colombo, Colombo, Sri Lanka. 247School of Midwifery, A.T. Still University, Mesa, Botswana, Gaborone, Botswana. 334Division of Plastic Surgery, University of Washington, AZ, USA. 248Environmental Research Center, Duke Kunshan University, Kunshan, China. Seattle, WA, USA. 335School of Medicine, University of New South Wales, Sydney, New South 249 Department of Medicine, University of Miami, Atlantis, FL, USA. 250Department of Wales, Australia. 336Research Department, The George Institute for Global Health, New Delhi, Ophthalmology, Heidelberg University, Heidelberg, Germany. 251Beijing Institute of India. 337Department of Biology and Biological Engineering, Chalmers University of Technology, Ophthalmology, Beijing Tongren Hospital, Beijing, China. 252Social Determinants of Health Gothenburg, Sweden. 338Department of Health Services Research and Policy, London School of Research Center, University of Social Welfare and Rehabilitation Sciences, Tehran, Iran. Hygiene & Tropical Medicine, London, UK. 339Preventive Oncology Department, National 253 Faculty of Medicine and Health Sciences, University of Opole, Opole, Poland. 254Department Institute of Cancer Prevention and Research, Noida, India. 340Department of Epidemiology and of Family Medicine and Public Health, University of Opole, Opole, Poland. 255Minimally Invasive Biostatistics, University of California San Francisco, San Francisco, CA, USA. 341Peru Country Surgery Research Center, Iran University of Medical Sciences, Tehran, Iran. 256Department of Office, United Nations Population Fund (UNFPA), Lima, Peru. 342Forensic Medicine Division, Nutrition and Dietetics, Mekelle University, Mekelle, Ethiopia. 257Mazandaran University of Imam Abdulrahman Bin Faisal University, Dammam, Saudi Arabia. 343Neurocenter, Helsinki Medical Sciences, Sari, Iran. 258Isfahan University of Medical Sciences, Isfahan, Iran. 259Social University Hospital, Helsinki, Finland. 344School of Health Sciences, University of Melbourne, Determinants of Health Research Center, Qazvin University of Medical Sciences, Qazvin, Iran. Parkville, Victoria, Australia. 345Breast Surgery Unit, Helsinki University Hospital, Helsinki, 260 Department of Epidemiology, Hamadan University of Medical Sciences, Hamadan, Iran. Finland. 346University of Helsinki, Helsinki, Finland. 347Clinical Microbiology and Parasitology 261 Research and Development, Australian Red Cross Blood Service, Sydney, New South Wales, Unit, Dr Zora Profozic Polyclinic, Zagreb, Croatia. 348University Centre Varazdin, University Australia. 262School of Public Health and Community Medicine, University of New South Wales, North, Varazdin, Croatia. 349Pacific Institute for Research & Evaluation, Calverton, MD, USA. Sydney, New South Wales, Australia. 263Hematologic Malignancies Research Center, Tehran 350 Health, Evidence and Impact, McMaster University, Hamilton, Ontario, Canada. University of Medical Sciences, Tehran, Iran. 264Hematology-Oncology and Stem Cell 351 Department of Computer Science and Software Engineering, University of Western Australia, Transplantation Research Center, Tehran University of Medical Sciences, Tehran, Iran. Perth, Western Australia, Australia. 352Department of Public Health, Amrita Institute of Medical 265 Department of Anesthesiology & Pain Medicine, University of Washington, Seattle, WA, USA. Sciences, Kochi, India. 353Department of Clinical Biochemistry, Babol University of Medical 266 Odel Campus, University of Nairobi, Nairobi, Kenya. 267Michigan State University, East Sciences, Babol, Iran. 354Golestan University of Medical Sciences, Gorgan, Iran. 355Foodborne Lansing, MI, USA. 268Tabriz Health Management Research Center, Tabriz University of Medical and Waterborne Diseases Research Center, Shahid Beheshti University of Medical Sciences, Sciences, Tabriz, Iran. 269National Institute for Health Research (NIHR), Tehran University of Tehran, Iran. 356Faculty of General Medicine, Kyrgyz State Medical Academy, Bishkek, Medical Sciences, Tehran, Iran. 270Department of Public Health and Community Medicine, Kyrgyzstan. 357Department of Atherosclerosis and Coronary Heart Disease, National Center of Jordan University of Science and Technology, Ramtha, Jordan. 271Social Determinants of Health Cardiology and Internal Disease, Bishkek, Kyrgyzstan. 358Health Equity Research Center, Tehran Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. University of Medical Sciences, Tehran, Iran. 359Department of Medicine Huddinge, Karolinska 272 Epidemiology and Biostatistics Department, Health Services Academy, Islamabad, Pakistan. Institutet, Stockholm, Sweden. 360Department of Food Technology, College of Agriculture, 273 Population Studies, International Institute for Population Sciences, Mumbai, India. Salahaddin University-Erbil, Erbil, Iraq. 361Information Technology Department, University of 274 Department of Internal Medicine, John H. Stroger Jr Hospital of Cook County, Chicago, IL, Human Development, Sulaimaniyah, Iraq. 362Department of Biostatistics, Hamadan University USA. 275Department of Internal Medicine, Dow University of Health Sciences, Karachi, Pakistan. of Medical Sciences, Hamadan, Iran. 363School of Pharmacy, Haramaya University, Harar, 276 Institute of Health Policy and Management, Seoul National University, Seoul, South Korea. Ethiopia. 364Institute of Public Health, Heidelberg University, Heidelberg, Germany. 365Health 277 Department of Health Policy and Management, Seoul National University, Seoul, South Systems and Policy Research Unit, Ahmadu Bello University, Zaria, Nigeria. 366Faculty of Life Korea. 278Faculty of Health and Wellbeing, Sheffield Hallam University, Sheffield, UK. Sciences and Medicine, King’s College London, London, UK. 367Clinical Epidemiology and 279 Department of Arts and Sciences, Ohio University, Zanesville, OH, USA. 280Internal Medicine Public Health Research Unit, Burlo Garofolo Institute for Maternal and Child Health, Trieste, and Gastroenterology Department, National Hepatology and Tropical Research Institute, Cairo, Italy. 368Health Sciences Research Center, Mazandaran University of Medical Sciences, Sari, Egypt. 281Department of Medical Parasitology, Cairo University, Cairo, Egypt. 282Department of Iran. 369Department of Epidemiology and Biostatistics, Kurdistan University of Medical Environmental Health Engineering, Hamadan University of Medical Sciences, Hamadan, Iran. Sciences, Sanandaj, Iran. 370Social Determinants of Health Research Center, Kurdistan 283 Department of Public Health, Mazandaran University of Medical Sciences, Sari, Iran. University of Medical Sciences, Sanandaj, Iran. 371Department of Epidemiology, Iran University 284 Department of Nutrition and Health Science, Ball State University, Muncie, IN, USA. of Medical Sciences, Tehran, Iran. 372Department of Mathematical Sciences, University of Bath, RESEARCH Article Bath, UK. 373International Laboratory for Air Quality and Health, Queensland University of Yonsei University, Wonju, South Korea. 457Institute of Public Health, Federal University of Bahia, Technology, Brisbane, Queensland, Australia. 374Department of Clinical Biochemistry, Tarbiat Salvador, Brazil. 458Gonçalo Moniz Institute, Oswaldo Cruz Foundation, Salvador, Brazil. Modares University, Tehran, Iran. 375Department of Health Management and Economics, Tehran 459 School of Behavioral Sciences and Mental Health, Iran University of Medical Sciences, University of Medical Sciences, Tehran, Iran. 376Federal Institute for Population Research, Tehran, Iran. 460Social Science and Psychology, Western Sydney University, Penrith, New South Wiesbaden, Germany. 377Center for Population and Health, Wiesbaden, Germany. Wales, Australia. 461School of Social Sciences and Psychology, Western Sydney University, 378 Department of Epidemiology and Biostatistics, University of Gondar, Gondar, Ethiopia. Penrith, New South Wales, Australia. 462Research Center for Environmental Determinants of 379 Department of Pediatric Medicine, Nishtar Medical University, Multan, Pakistan. Health (RCEDH), Kermanshah University of Medical Sciences, Kermanshah, Iran. 463Department 380 Department of Pediatrics & Pediatric Pulmonology, Institute of Mother & Child Care, Multan, of Epidemiology, Shahid Beheshti University of Medical Sciences, Tehran, Iran. 464Department Pakistan. 381Department of Urology, Tehran University of Medical Sciences, Tehran, Iran. of Epidemiology, Birjand University of Medical Sciences, Birjand, Iran. 465EPIUnit, University of 382 Operating Room Department, Kermanshah University of Medical Sciences, Kermanshah, Porto, Porto, Portugal. 466Department of Clinical Research, Federal University of Uberlândia, Iran. 383Research and Analytics, Initiative for Financing Health and Human Development, Uberlândia, Brazil. 467Public Health, Addis Ababa University, Addis Ababa, Ethiopia. Chennai, India. 384Research and Analytics, Bioinsilico Technologies, Chennai, India. 385Cancer 468 Department of Public Health, Wollega University, Nekemte, Ethiopia. 469Martin School, Research Center of Cancer Institute, Tehran University of Medical Sciences, Tehran, Iran. University of Oxford, Oxford, UK. 470Golestan Research Center of Gastroenterology and 386 Department of Epidemiology & Biostatistics, Kermanshah University of Medical Sciences, Hepatology, Golestan University of Medical Sciences, Gorgan, Iran. 471Epidemiology and Kermanshah, Iran. 387Suraj Eye Institute, Nagpur, India. 388Cochrane South Africa, South African Biostatistics, Kurdistan University of Medical Sciences, Sanandaj, Iran. 472Infectious Diseases Medical Research Council, Cape Town, South Africa. 389Emergency Hospital of Bucharest, Carol and Tropical Medicine Research Center, Babol University of Medical Sciences, Babol, Iran. Davila University of Medicine and Pharmacy, Bucharest, Romania. 390General Surgery 473 School of Biotechnology, Ikiam Amazon Regional University, Tena, Ecuador. 474Department Department, Carol Davila University of Medicine and Pharmacy, Bucharest, Romania. of Ocean Science and Engineering, Southern University of Science and Technology, Shenzhen, 391 Anatomy and Embryology Department, Carol Davila University of Medicine and Pharmacy, China. 475Department of Biomedical Sciences, University of Sassari, Sassari, Italy. Bucharest, Romania. 392Department of Cardiology, Cardio-aid, Bucharest, Romania. 476 Department of Health, Safety and Environment (HSE), Shahid Beheshti University of Medical 393 Department of Biological Sciences, University of Embu, Embu, Kenya. 394Institute for Global Sciences, Tehran, Iran. 477Faculty of Public Health, Kermanshah University of Medical Sciences, Health Innovations, Duy Tan University, Hanoi, Vietnam. 395Center for Excellence in Behavioral Kermanshah, Iran. 478Department of Neuroscience, Iran University of Medical Sciences, Tehran, Health, Nguyen Tat Thanh University, Ho Chi Minh, Vietnam. 396Global Health Department, Iran. 479Neurogenic Inflammation Research Center, Mashhad University of Medical Sciences, University of Washington, Seattle, WA, USA. 397Department of Pediatrics, University of Mashhad, Iran. 480Biotechnology Research Center, Mashhad University of Medical Sciences, Washington, Seattle, WA, USA. 398Clinical Pharmacy Unit, Mekelle University, Mekelle, Ethiopia. Mashhad, Iran. 481Department of Anatomical Sciences, Kermanshah University of Medical 399 Public Health Science Department, School of Public Health and Family Medicine, University Sciences, Kermanshah, Iran. 482Department of Pathology, Al-Imam Mohammad Ibn Saud of Cape Town, Cape Town, South Africa. 400Department of Community and Family Medicine, Islamic University, Riyadh, Saudi Arabia. 483School of Health and Policy Management, Faculty of Iran University of Medical Sciences, Tehran, Iran. 401University of Social Welfare and Health, York University, Toronto, Ontario, Canada. 484Taleghani Hospital, Kermanshah University Rehabilitation Sciences, Tehran, Iran. 402Department of Health Economics, Tabriz University of of Medical Sciences, Kermanshah, Iran. 485Department of Radiology and Nuclear Medicine, Medical Sciences, Tabriz, Iran. 403Department of Medicine, University of Cape Town, Cape Town, Kermanshah University of Medical Sciences, Kermanshah, Iran. 486Taleghani Hospital, South Africa. 404Centre of Cardiovascular Research and Education in Therapeutics, Monash Kermanshah, Iran. 487Center for Health Policy & Center for Primary Care and Outcomes University, Melbourne, Victoria, Australia. 405Independent Consultant, Accra, Ghana. Research, Stanford University, Stanford, CA, USA. 488Department of Entomology, Ain Shams 406 Translational Health Research Institute, Western Sydney University, Penrith, New South University, Cairo, Egypt. 489Centre School of Public Health and Health Management, University Wales, Australia. 407Center for the Aid Program of Research in South Africa (CAPRISA) TB and of Belgrade, Belgrade, Serbia. 490Post-graduate Program in Infectious Diseases and Tropical HIV Pathogenesis Unit, United Nations Programme on HIV/AIDS (UNAIDS), Durban, South Medicine, Federal University of Minas Gerais, Belo Horizonte, Brazil. 491Department of Africa. 408Department of Psychiatry and Behavioural Neurosciences, McMaster University, Community Medicine, PSG Institute of Medical Sciences and Research, Coimbatore, India. Hamilton, Ontario, Canada. 409Department of Psychiatry, University of Lagos, Lagos, Nigeria. 492 PSG-FAIMER South Asia Regional Institute, Coimbatore, India. 493Department of Health and 410 Gastroenterology and Liver Disease Research Center, A.C.S. Medical College and Hospital, Society, Faculty of Medicine, University of Applied and Environmental Sciences, Bogotá, Tehran, Iran. 411Department of Food Science and Postharvest Technology, Gulu University, Gulu, Colombia. 494Faculty of Infectious and Tropical Diseases, London School of Hygiene & Tropical Uganda. 412Ghent University, Ghent, Belgium. 413Centre for Healthy Start Initiative, Lagos, Medicine, London, UK. 495Surgery Department, Hamad Medical Corporation, Doha, Qatar. Nigeria. 414Department of Health Promotion and Education, University of Ibadan, Ibadan, 496 Faculty of Health & Social Sciences, Bournemouth University, Bournemouth, UK. 497School Nigeria. 415Department of Pharmacology and Therapeutics, University of Nigeria Nsukka, of Public Health, Imperial College London, London, UK. 498Department of Psychology, Enugu, Nigeria. 416University of Washington, Seattle, WA, USA. 417Graduate School of Public University of Alabama at Birmingham, Birmingham, AL, USA. 499Center of Expertise in Health, San Diego State University, San Diego, CA, USA. 418Center for Health Systems Research, Microbiology, Tehran University of Medical Sciences, Tehran, Iran. 500Invasive Fungi Research National Institute of Public Health, Cuernavaca, Mexico. 419School of Medicine, Autonomous Center, Mazandaran University of Medical Sciences, Sari, Iran. 501Department of Health University of Madrid, Madrid, Spain. 420Department of Nephrology and Hypertension, The Promotion and Education, Alborz University of Medical Sciences, Karaj, Iran. 502Independent Institute for Health Research Foundation Jiménez Díaz University Hospital, Madrid, Spain. Consultant, Karachi, Pakistan. 503School of Medicine, Dezful University of Medical Sciences, 421 Environmental Management and Toxicology, University of Benin, Benin City, Nigeria. Dezful, Iran. 504Medical Laboratory Sciences, Mazandaran University of Medical Sciences, Sari, 422 Faculty of Geoinformation Science and Earth Observation, University of Twente, Enschede, Iran. 505Chronic Diseases (Home Care) Research Center, Hamadan University of Medical The Netherlands. 423Department of Mathematics and Statistics, University of Energy and Sciences, Hamadan, Iran. 506Department of Laboratory Sciences, Karaj Islamic Azad University, Natural Resources, Sunyani, Ghana. 424Analytical Center, Moscow Institute of Physics and Kermanshah, Iran. 507Department of Basic Sciences, Karaj Islamic Azad University, Technology, Dolgoprudny, Russia. 425Committee for the Comprehensive Assessment of Medical Kermanshah, Iran. 508HIV/STI Surveillance Research Center, Kerman University of Medical Devices and Information Technology, Health Technology Assessment Association, Moscow, Sciences, Kerman, Iran. 509Policy and Planning Division, Ministry of Health, Riyadh, Saudi Russia. 426Institute for Advanced Medical Research and Training, University of Ibadan, Ibadan, Arabia. 510University School of Management and Entrepreneurship, Delhi Technological Nigeria. 427Department of Tb & Respiratory Medicine, Jagadguru Sri Shivarathreeswara University, New Delhi, India. 511Division of General Internal Medicine and Primary Care, Harvard University, Mysore, India. 428Department of Medicine, Ottawa Hospital Research Institute, University, Boston, MA, USA. 512Usher Institute of Population Health Sciences and Informatics, Ottawa, Ontario, Canada. 429Heidelberg University, Heidelberg, Germany. 430Department of University of Edinburgh, Edinburgh, UK. 513National Institute of Infectious Diseases, Tokyo, Medical Humanities and Social Medicine, Kosin University, Busan, South Korea. 431Research Japan. 514Finnish Institute of Occupational Health, Helsinki, Finland. 515Institute of Medical and Evaluation, Population Council, New Delhi, India. 432Indian Institute of Health Management Epidemiology, Martin Luther University Halle-Wittenberg, Halle, Germany. 516Department of Research University, Jaipur, India. 433Center for Research and Innovation, Ateneo De Manila Epidemiology, University of Alabama at Birmingham, Birmingham, AL, USA. 517Department of University, Pasig City, The Philippines. 434Department of Genetics, Harvard University, Boston, Medicine, University of Alabama at Birmingham, Birmingham, AL, USA. 518Department of MA, USA. 435Laboratory of Genetics and Molecular Cardiology, University of São Paulo, Sao Epidemiology, School of Preventive Oncology, Patna, India. 519Department of Epidemiology, Paulo, Brazil. 436Department of Cardiology, University of Bern, Bern, Switzerland. Healis Sekhsaria Institute for Public Health, Mumbai, India. 520Department of Physiotherapy 437 Parasitology and Entomology Department, Tarbiat Modares University, Tehran, Iran. and Occupational Therapy, Næstved-Slagelse-Ringsted Hospitals, Slagelse, Denmark. 438 521 National Centre for Register-Based Research, Aarhus University, Aarhus, Denmark. Medical Division, German Leprosy and TB Relief Association Ethiopia, Addis Ababa, Ethiopia. 439 522 Department of Nutrition and Food Sciences, Maragheh University of Medical Sciences, Department of Medicine, University of Washington, Seattle, WA, USA. 523Department of Maragheh, Iran. 440Department of Public Health, Maragheh University of Medical Sciences, Medicine, University of Calgary, Calgary, Alberta, Canada. 524Social Development and Health Maragheh, Iran. 441Non-communicable Diseases Research Center, Alborz University of Medical Promotion Research Center, Kermanshah University of Medical Sciences, Kermanshah, Iran. Sciences, Karaj, Iran. 442Biomedical Engineering, Amirkabir University of Technology, Tehran, 525 Hospital Universitario de la Princesa, Autonomous University of Madrid, Madrid, Spain. Iran. 443College of Graduate Health Sciences, A.T. Still University, Mesa, AZ, USA. 444Medichem, 526 Centro de Investigación en Red de Enfermedades Respiratorias (CIBERES), Institute of Barcelona, Spain. 445Molecular and Cell Biology Research Center, Mazandaran University of Health Carlos III, Madrid, Spain. 527Division of Community Medicine, International Medical Medical Sciences, Sari, Iran. 446Department of Immunology, Mazandaran University of Medical University, Kuala Lumpur, Malaysia. 528Department of Nursing, Muhammadiyah University of Sciences, Sari, Iran. 447Endocrinology and Metabolism Molecular-Cellular Sciences Institute, Surakarta, Kartasura, Indonesia. 529Department of Community Medicine, Ahmadu Bello Tehran University of Medical Sciences, Tehran, Iran. 448Thalassemia and Hemoglobinopathy University, Zaria, Nigeria. 530Department of Criminology, Law and Society, University of Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. 449Department California Irvine, Irvine, CA, USA. 531Neurology Department, Sree Chitra Tirunal Institute for of Clinical Biochemistry, Kermanshah University of Medical Sciences, Kermanshah, Iran. 450Sina Medical Sciences and Technology, Trivandrum, India. 532Carlos III Health Institute, Biomedical Trauma and Surgery Research Center, Tehran University of Medical Sciences, Tehran, Iran. Research Networking Center for Mental Health Network (CIBERSAM), Madrid, Spain. 451 533 Department of Health Education & Promotion, Kermanshah University of Medical Sciences, Department of Medicine, University of Valencia, Valencia, Spain. 534School of Social Work, Kermanshah, Iran. 452Department of Nephrology, Nizam’s Institute of Medical Sciences, University of Illinois, Urbana, IL, USA. 535Cancer Control Center, Osaka International Cancer Hyderabad, India. 453Prevention of Metabolic Disorders Research Center, Shahid Beheshti Institute, Osaka, Japan. 536University Institute ‘Egas Moniz’, Monte Da Caparica, Portugal. University of Medical Sciences, Tehran, Iran. 454Critical Care Quality Improvement Research 537 Research Institute for Medicines, Faculty of Pharmacy of Lisbon, University of Lisbon, Lisbon, Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran. 455Policy Research Portugal. 538Department of Pediatrics, King Saud University, Riyadh, Saudi Arabia. 539College of Institute, Kathmandu, Nepal. 456Institute for Poverty Alleviation and International Development, Medicine, Alfaisal University, Riyadh, Saudi Arabia. 540Anesthesiology Department, University of Article RESEARCH Virginia, Charlottesville, VA, USA. 541Syrian Expatriate Medical Association (SEMA), 578 Department of Pharmacology and Toxicology, Mekelle University, Mekelle, Ethiopia. Charlottesville, VA, USA. 542Department of Public Health and Community Medicine, Central 579 Department of Pharmacology, Addis Ababa University, Addis Ababa, Ethiopia. 580Zhejiang University Kerala, Kasaragod, India. 543Nanyang Technological University, Singapore, Spine Research Center, Wenzhou Medical University, Wenzhou, China. 581School of Medicine, Singapore. 544School of Exercise and Nutrition Sciences, Queensland University of Nanjing University, Nanjing, China. 582Department of Diabetes and Metabolic Diseases, Technology, Brisbane, Queensland, Australia. 545Department of Pathology and Legal Medicine, University of Tokyo, Tokyo, Japan. 583Department of Health Management, Policy and University of São Paulo, Sao Paulo, Brazil. 546Department of Health Economics, Hanoi Medical Economics, Kerman University of Medical Sciences, Kerman, Iran. 584Health Services University, Hanoi, Vietnam. 547Department of Molecular Medicine and Pathology, University of Management Research Center, Kerman University of Medical Sciences, Kerman, Iran. Auckland, Auckland, New Zealand. 548Clinical Hematology and Toxicology, Military Medical 585 Department of Pediatrics, University of Jos, Jos, Nigeria. 586Department of Pediatrics, Jos University, Hanoi, Vietnam. 549Gomal Center of Biochemistry and Biotechnology, Gomal University Teaching Hospital, Jos, Nigeria. 587Centre for Suicide Research and Prevention, University, Dera Ismail Khan, Pakistan. 550TB Culture Laboratory, Mufti Mehmood Memorial University of Hong Kong, Hong Kong, China. 588Department of Social Work and Social Teaching Hospital, Dera Ismail Khan, Pakistan. 551Division of Health Sciences, University of Administration, University of Hong Kong, Hong Kong, China. 589Department of Warwick, Coventry, UK. 552Department of Education and Health, Trauma Research Center, Psychopharmacology, National Center of Neurology and Psychiatry, Tokyo, Japan. 590Health Tehran, Iran. 553Critical and Intensive Care Department, Trauma Research Center, Tehran, Iran. Economics & Finance, Global Health, Jackson State University, Jackson, MS, USA. 591Research 554 Argentine Society of Medicine, Buenos Aires, Argentina. 555Velez Sarsfield Hospital, Buenos Center for Public Health, Tsinghua University, Peking, China. 592Prevention of Cardiovascular Aires, Argentina. 556University Medical Center Groningen, University of Groningen, Groningen, Disease Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran. The Netherlands. 557Department of General Practice, University Medical Center Groningen, 593 Medical Parasitology Department, Cairo University, Cairo, Egypt. 594Global Health Institute, Groningen, The Netherlands. 558Ukk Institute, Tampere, Finland. 559Psychosocial Injuries Wuhan University, Wuhan, China. 595Department of Health Management and Economics, Research Center, Ilam University of Medical Sciences, Ilam, Iran. 560Raffles Neuroscience A.C.S. Medical College and Hospital, Tehran, Iran. 596Department of Electrical Engineering, Centre, Raffles Hospital, Singapore, Singapore. 561Yong Loo Lin School of Medicine, National Sharif University of Technology, Tehran, Iran. 597Electrical Engineering, Institute for Research University of Singapore, Singapore, Singapore. 562Department of Medical and Surgical in Fundamental Sciences, Tehran, Iran. 598Social Determinants of Health Research Center, Sciences, University of Bologna, Bologna, Italy. 563Occupational Health Unit, Sant’orsola Ardabil University of Medical Science, Ardabil, Iran. 599Maternal and Child Health Division, Malpighi Hospital, Bologna, Italy. 564Department of Information Technologies and International Centre for Diarrhoeal Disease Research Bangladesh, Dhaka, Bangladesh. 600 Management, Moscow Institute of Physics and Technology, Dolgoprudny, Russia. Department of Medicine, Monash University, Melbourne, Victoria, Australia. 601Student 565 Department of Information and Internet Technologies, I. M. Sechenov First Moscow State Research Committee, Babol University of Medical Sciences, Babol, Iran. 602Department of Medical University, Moscow, Russia. 566Department of Health Care Administration and Community Medicine, Ardabil University of Medical Science, Ardabil, Iran. 603Social Economy, National Research University Higher School of Economics, Moscow, Russia. Development and Health Promotion Research Center, Kermanshah University of Medical 567 Foundation University Medical College, Foundation University, Rawalpindi, Pakistan. Sciences, Kermanshah, Iran. 604Maternal and Child Wellbeing Unit, African Population Health 568 Department of Statistics, University of Washington, Seattle, WA, USA. 569Department of Research Centre, Nairobi, Kenya. 605Public Health Department, Dilla University, Dilla, Ethiopia. Epidemiology and Biostatistics, Wuhan University, Wuhan, China. 570Department of 606 Department of Preventative Medicine, Wuhan University, Wuhan, China. 607School of Psychiatry, University of São Paulo, Sao Paulo, Brazil. 571Institute of Child Health, University Public Health, Wuhan University of Science and Technology, Wuhan, China. 608Indian Institute College London, London, UK. 572Cardiology Department, Royal Children’s Hospital, of Public Health, Public Health Foundation of India, Gurugram, India. 609School of Melbourne, Victoria, Australia. 573Murdoch Childrens Research Institute, Melbourne, Victoria, BioSciences, University of Melbourne, Parkville, Victoria, Australia. 610State University of Australia. 574School of Nursing, Aksum University, Aksum, Ethiopia. 575Competence Center of Semarang, Public Health Science Department, Kota Semarang, Indonesia. 611Graduate Mortality-Follow-Up, Federal Institute for Population Research, Wiesbaden, Germany. Institute of Biomedical Informatics, Taipei Medical University, Taipei City, Taiwan. 612These 576 Cochrane South Africa, Medical Research Council South Africa, Cape Town, South Africa. authors contributed equally: Roy Burstein, Nathaniel J. Henry. 613These authors jointly 577 Department of Global Health, Stellenbosch University, Cape Town, South Africa. supervised this work: Christopher J. L. Murray, Simon I. Hay. *e-mail: sihay@uw.edu RESEARCH Article Methods that ages were represented in seven mutually exclusive bins (0, 1–5, 6–11, 12–23, Overview. We fitted a discrete hazards geostatistical model51,52 with correlated 24–35, 36–47 and 48–59 months), each with its own assumed constant mortality space–time–age errors and made predictions to generate joint estimates—with probability. The model explicitly accounted for variation across age bin, year and uncertainty—of the probability of death (the number of deaths per live births) space through inclusion of both fixed and random effects. Indicator variables for and the number of deaths for children aged 0–28 days (neonates), children under each age bin were included to form a discrete baseline mortality hazard function, 1 year old (infants) and children under 5 years old at the subnational level for 99 representing the risk of mortality in discrete bins from birth to 59 months of age LMICs for each year from 2000 to 2017. The analytical process is summarized in with covariates set at their means. Baseline hazard functions were allowed to vary the flowchart in Extended Data Fig. 6. We made estimates at a grid-cell resolution in space and time in response to changing covariate values, as well as in response of approximately 5 × 5-km and then produced spatially aggregated estimates at to linear effect on year. To model this relationship, we estimated the effect of each the first (that is, states or provinces) and second (that is, districts or counties) covariate value on the risk of mortality. These estimated effects were then applied administrative levels, as well as the country level. to the gridded surface of covariate values to make predictions across the entire Countries were selected for inclusion in this study based on their socio- study geography. We also included a Gaussian random effect across countries to demographic index (SDI) published in the Global Burden of Disease study (GBD)53. account for larger-scale variations due to political or institutional effects, as well as The SDI is a measure of development based on income per capita, educational a Gaussian random effect for each data source to account for source-specific biases. attainment and fertility rates among women under 25 years old. We primarily aimed Finally, we included a Gaussian process random effect with a covariance matrix to include all countries in the middle, low–middle or low SDI quintiles, with several structured to account for remaining correlation across age, time and physical space. exceptions. Brazil and Mexico were excluded despite middle SDI status owing to As such, estimates at a specific age, time or place benefitted from drawing predic- the availability of high-quality vital registration data in these countries, which have tive strength from data points nearby in all of these dimensions. served as the basis for existing subnational estimates of child mortality. Because this For each modelling region, we fitted one such discrete hazards model with a study did not incorporate vital registration data sources (see ‘Limitations’), Brazil binomial data likelihood. All data were prepared such that we counted or estimated and Mexico were not estimated directly; instead, state-level estimates from the GBD the number of children entering into (n) and dying within (Y) each period–age 2017 study were directly substituted in figures where appropriate4. Albania and bin from each GPS-point location (s) in each survey (k) within each country (c). Moldova were excluded despite middle SDI status owing to geographical disconti- The number of deaths for children in age band (a) in year (t) at location (s) was nuity with other included countries and lack of available survey data. North Korea assumed to follow a binomial distribution: was excluded despite low–middle SDI status owing to geographical discontinuity and insufficient data. As countries with high–middle SDI status in 2017, China and Ya , s, t ~ binomial(na , s, t , Pa , s, t ) Malaysia were excluded from this analysis. Libya was included despite high–middle where Pa,s,t is the probability of death in age bin a, conditional on survival to that SDI status to create better geographical continuity. Island nations with populations age bin for a particular space–time location. Using a generalized linear regression under 1 million were excluded because they typically lacked sufficient survey data modelling framework, a logit link function is used to relate P to a linear combi- or geographical continuity for a geospatial analytic approach to be advantageous nation of effects: over a national approach. Supplementary Figure 3.1 shows a map of the countries included in this study and Supplementary Table 3.1 lists the countries. 7 Data. We extracted individual records from 555 household sample survey and logit(Pa , s, t ) = β 0 + ∑ Iaβa + β 2Xs, t + β 3t + νc [s] + ν k [s] + Za , s, t 1 census sources. Records came in the form of either summary birth histories (SBHs) a=2 or complete birth histories (CBHs). All input data were subject to quality checks, which resulted in the exclusion of 82 surveys and censuses owing to quality con- The first term, β0, is an intercept, representing the mean for the first age band when cerns (see Supplementary Information section 3.2 for more details). Data on life all covariates are equal to zero, whereas βa1 are fixed effects for each age band, and mortality experiences from CBH sources can be tabulated directly into discrete representing the mean overall hazard deviation for each age band from the inter- period and age bins, thus allowing for period-specific mortality estimations, known cept, when all other covariates are equal to zero. β2 are the effects of geospatial as the synthetic cohort method54–56. For SBH data, we used indirect estimation57 to covariates (Xs,t), which we describe in detail in Supplementary Information section estimate age-specific mortality probabilities and sample sizes and assign them to 4. β3 is an overall linear temporal effect to account for overall temporal trends specific time periods. Complete details are available in Supplementary Information within the region. All geospatial covariates were centred and scaled by subtracting section 3.3.2. their mean and dividing by their standard deviations. Each v term represents uncor- In all cases, after pre-processing, each data point provided a number of deaths related Gaussian random effects: νc [s] ~ normal(0, σc2) is a country-level random and a sample size for an age bin in a specific year and location. We referenced effect applied to all locations (s) within a country (c); ν k [s] ~ normal(0, σk2) is a data all data points to GPS coordinates (latitude and longitude) wherever possible. In source-level random effect for the survey (k) from which the data at location s were cases in which GPS data were unavailable, we matched data points to the smallest observed. Data source-level random effects were used to account for systematic possible areal unit (also referred to as ‘polygons’). All polygon data were spatially variation or biases across data sources and were included in model fitting but not resampled into multiple GPS coordinates and weighted based on the population in prediction from fitted models. The term Za,s,t ~ Gaussian process(0, K) is a distribution following a previously described procedure5,22,23,58 and described correlated random effect across age, space and time, and is modelled as a four-di- in Supplementary Information section 3.4. Our combined global dataset contained mensional mean zero Gaussian process with covariance matrix K. This term approximately 15.9 million births and 1.1 million child deaths. A complete list of accounts for structured residual correlation across these spatial–age–temporal data sources is provided in Supplementary Table 8.1. dimensions that are not accounted for by any of the model’s other fixed or random In addition to data on child mortality, we used a number of spatial data sources effects. This structure was chosen, because the hazard probability for each age for this analysis. These included a suite of geospatial covariates, population esti- group is expected to vary in space and time, and such spatiotemporal correlations mates and administrative boundaries68. These sources and processing procedures are likely to be similar across ages. K is constructed as a separable process across are described in Supplementary Information section 4. age, space and time (K = Σa ⊗ Σt ⊗ Σs) . The continuous spatial component is Spatial covariates. We extracted values from each of 10 geospatial covariates at modelled with a stationary isotropic Matérn covariance function, and the age and each data point location. Geospatial covariates are spatial data represented at the temporal effects were each assumed to be discrete auto-regressive order 1. We 5 × 5-km grid-cell resolution. The covariates were travel time to the nearest city, provide further details on model fitting and specification in Supplementary educational attainment of maternal-aged women, the ratio of population of chil- Information section 5.1. dren under 5 to women of reproductive age (ages 15–49 years old), the mass per We assigned priors to all model parameters and performed maximum a pos- cubic meter of air of particles with a diameter less than 2.5 μm, total population, teriori inference using Template Model Builder60 software in R version 3.4. We a binary indicator of urbanicity, intensity of lights at night, the proportion of chil- fitted the model separately for each of 11 world regions (see Supplementary Fig. dren aged 12–23 months who had received the third dose of diphtheria–pertussis– 3.1), owing to memory constraints and to allow model parameters to vary across tetanus vaccine, incidence rate of Plasmodium falciparum-associated malaria in epidemiologically distinct regions. children under 5 and prevalence of stunting in children under 5 (see Supplementary Post-estimation. Using the joint precision matrix and point estimates, we generated Information). All covariate values were centred on their means and scaled by their 1,000 draws from all model parameters using a multivariate-normal approxima- standard deviations. Covariates typically had global spatial coverage and values that tion. These model parameter draws were used to predict corresponding draws vary by year. More details of the spatial covariates can be found in Supplementary of mortality probabilities across all age groups for each grid cell in each year. In Information section 4. other words, for each age bin in each year we estimated 1,000 gridded surfaces of Analysis. Geostatistical model. To synthesize information across various sources, mortality probability estimates, each surface corresponding to one draw from the and to make consistent estimates across space and time, we fitted discrete haz- posterior parameter estimates61. All subsequent post-estimation procedures were ards51,52 geostatistical models59 to our data. The models were discrete in the sense carried out across draws to propagate model uncertainty. We used these estimated Article RESEARCH spatiotemporal gridded surfaces of age-specific mortality probabilities to produce administrative unit level. Furthermore, discrete localized mortality shock events various final resulting data products. could be missing in our analysis due to the lack of data and selection biases in From the fitted model parameters, we produced posterior mortality probability surveys and censuses, and spatiotemporal smoothing. Fatal discontinuities are estimates for each age group for each 5 × 5-km grid cell for each year from 2000 explicitly accounted for at the national or province level by calibration to GBD to 2017. We combined gridded age group estimates to obtain infant (under 1) estimates. In all, 0.35% (0.4 million) of the 123 million deaths over this period were and child (under 5) mortality estimates at each gridded location. Using a con- attributed to fatal discontinuities. version from mortality probability to mortality rates, and using a gridded surface On the modelling side, we integrated point and areal data into a continuous of population, we also estimated the number of deaths that occurred in each age model by constructing pseudo-points from areal data. Modelling approaches that group at each location in each year. For both mortality probabilities and counts, integrate point and areal data as part of a joint model likelihood function are in we multiplied out corresponding gridded estimates by a constant to ensure that at development66 but are currently computationally infeasible at the large geograph- the national—and in two countries, the first administrative-level unit—aggregated ical scales at which we currently model. Furthermore, we divided our models into estimates for each age group and year were calibrated such that they equalled esti- 11 regional fits (see Supplementary Fig. 3.1), as a full model that encompasses mates in the GBD study4. This calibration allowed us to take advantage of national all 99 countries would be computationally infeasible due to memory constraints. data sources, such as vital registration, that could not be used in this study. We also Splitting up modelling in this way had the benefit of enabling parameters to vary aggregated grid-cell-level estimates to first and second administrative-level units across epidemiologically distinct world regions. A preferred model, however, would using gridded population surface to weight estimates. These steps are described be fitted to all data simultaneously with parameters that are spatially variable. in Supplementary Information section 5.2. The separable model used for age–space–time correlations is a common par- Model validation. We used fivefold cross-validation to assess and compare model simonious assumption afforded in applying spatiotemporal geostatistical models performance with respect to estimating local trends of age-specific mortality. Each due to efficient computation and inference; however, it yields the assumption of fold was created by combining complete surveys into subsets of approximately 20% fully symmetric covariance. The symmetry implicit in the separable model dictates, of data sources from the input data. Holding out entire surveys at a time served for example, that (holding age constant for simplicity) the covariance between as a comparable approximation to the type of missingness in our data, essentially the observations at (location 1, time 1) and (location 2, time 2) is the same as the helping us check how well our model estimates of mortality probabilities compared covariance between (location 1, time 2) and (location 2, time 1). Given our avail- to empirical estimates of mortality probability from an unobserved data source able data density in space–age–time, we believe that attempting to parameterize a that did not inform the model. more complex non-separable model would be challenging both computationally For each posterior draw, we aggregated to administrative units. Using data and inferentially, and it is not clear whether there would be much to benefit from aggregated to the administrative unit and aggregated estimate pairs, we calcu- the extra complications. lated the difference between out-of-sample empirical data estimates and modelled There are several limitations to address with respect to the use of covariates in estimates, and we report the following summary metrics: mean error, which serves the model. Most of the geospatial covariates that we used in the geostatistical model as a measure of bias, the square root of mean errors, which serves as a measure of were themselves estimates produced from various geospatial models. Some of those the total variation in the errors, the correlation and 95% coverage. At the second estimated surfaces used covariates that were also included in our model in their administrative-level unit for under-5 mortality, our out-of-sample 95% coverage estimation process. As such, we emphasize that our model is meant to be predictive, was 93%, correlation was 0.78, mean absolute error was 0.015 and mean error was and that drawing inference from fitted coefficients across these highly correlated −0.0011. These results indicate a good overall fit, with minimal bias. This proce- covariates is problematic and not recommended. Furthermore, we assumed no dure and the full validation results are discussed in Supplementary Information measurement error in the covariate values and assumed that the functional form section 5.3. between mortality and all covariates was linear in logit space. In certain locations, Limitations. This work should be assessed in full acknowledgement of several we used covariate values for prediction that were outside the observed range of the data and methodological limitations. We exclusively used CBH and SBH data from training data. As we explore in Supplementary Information section 5.4.2, however, household survey and census data sources. Ideally, estimates of child mortality these areas represent a relatively small proportion of the population. should incorporate all available data, including data from administrative vital Finally, we used a method for indirect estimation of SBHs that was recently registration systems. Vital registration systems are commonly present in many developed and validated57. As such, indirect estimation was carried out as a middle-income and all high-income countries. There are known data-quality issues pre-processing step before fitting the geostatistical model. We attempted to prop- with vital registration sources in many middle-income countries48,62 that add com- agate various forms of uncertainty that could be introduced in this step, which plications to their inclusion in our modelling procedure. For example, systems may resulted in halving the total effective sample size across all SBH data. In future, we not capture all deaths, and this level of ‘underreporting’ probably varies in space, aim to fully integrate such processing into the statistical model; such methods are time and age. In addition, underreporting is probably negatively correlated with in development67, but are not yet computationally feasible at scale. mortality, and could contribute substantial bias to estimates. Statistical methods Reporting summary. Further information on research design is available in must be developed to jointly estimate—and adjust for—underreporting in vital the Nature Research Reporting Summary linked to this paper. registration data before such data can be used in geospatial models of child mor- tality. Promising work has begun in this domain in specific countries63, but further Data availability advancement will be necessary to improve estimates across a time series and across The findings of this study are supported by data that are available from pub- many countries at once. lic online repositories, data that are publicly available upon request of the data We assume that SBH and CBH data were retrospectively representative in the provider and data that are not publicly available due to restrictions by the data locations in which they were collected. As such, we assume that survey respond- provider. Non-publicly available data were used under a license for the current ents did not migrate. High-spatial-resolution migration estimates with which to study, but may be available from the authors upon reasonable request and with adjust estimates do not yet exist, and many of the data sources that we use do permission of the data provider. A detailed table of data sources and availability not collect information on migration. We conducted a focused sensitivity anal- can be found in Supplementary Table 8.1. The full output of the analyses is publicly ysis (Supplementary Information section 5.4.4) for migration in six countries, available in the Global Health Data Exchange (GHDx; http://ghdx.healthdata.org/ and found that although our results were generally robust, there was variation record/ihme-data/lmic-under5-mortality-rate-geospatial-estimates-2000-2017) by country. Furthermore, despite providing high-quality retrospective data from and can be explored using custom data visualization tools (https://vizhub. representative samples of households, birth history data can suffer from certain healthdata.org/lbd/under5). non-sampling issues, such as survival/selection biases64 and misplacement of births65. We did not attempt to make corrections to data, and they were used as-is. 51. Lindgren, F., Rue, H. & Lindström, J. An explicit link between Gaussian fields and Furthermore, retrospective birth history data will—by design—have a changing Gaussian Markov random fields: the stochastic partial differential equation composition of maternal ages depending on the time since the survey. This was approach. R. Stat. Soc. 73, 423–498 (2011). 52. Allison, P. D. Discrete-time methods for the analysis of event histories. Sociol. minimized by limiting retrospective trends to up to 17 years. Methodol. 13, 61 (1982). Although we collated a large geo-referenced database of survey data on child 53. Institute for Health Metrics and Evaluation. Global Burden of Disease Study mortality, these data represented about 1% (1.1 million) of total deaths of children 2017 (GBD 2017) Socio-Demographic Index (SDI) 1950–2017. http://ghdx. under 5 in study areas over the period. Where data do not exist or are not available healthdata.org/record/ihme-data/gbd-2017-socio-demographic-index-sdi- in certain locations, mean estimates are informed from smoothing to nearby 1950%E2%80%932017 (GHDx, 2019). estimates and covariates. As such, there could be additional small-scale heterogeneity 54. Ahmad, O. B., Lopez, A. D. & Inoue, M. The decline in child mortality: a reappraisal. Bull. World Health Organ. 78, 1175–1191 (2000). that is not picked up by our model. Wider uncertainty intervals in areas with 55. Somoza, J. L. Illustrative Analysis: Infant and Child Mortality in Colombia. no data account for these potential unknowns, and our 95% coverage estimates World Fertility Survey Scientific Report No. 10 (International Statistical Institute, in out-of-sample predictive tests appear to be well-calibrated at the second 1980). RESEARCH Article 56. Rutstein, S. O. Infant and child mortality: levels, trends and demographic 68. WorldPop Dataset (WorldPop, accessed 25 July 2017); http://www.worldpop. differentials. World Fertility Survey Scientific Report No. 43 (International org.uk/data/get_data/ Statistical Institute, 1984). 57. Burstein, R., Wang, H., Reiner, R. C. Jr & Hay, S. I. Development and validation Acknowledgements This work was primarily supported by grant OPP1132415 of a new method for indirect estimation of neonatal, infant, and child from the Bill & Melinda Gates Foundation. mortality trends using summary birth histories. PLoS Med. 15, e1002687 (2018). Author contributions S.I.H. and R.B. conceived and planned the study. R.B., 58. Graetz, N. et al. Mapping local variation in educational attainment across Africa. S.I.H., M.C., N.H., J.L, A.B., N.G. and S.W. identified and obtained data for this Nature 555, 48–53 (2018). analysis. M.C., N.H., J.L, A.B. and S.W. extracted, processed and geo-positioned 59. Diggle, P. & Ribeiro, P. J. Model-based Geostatistics (Springer, 2007). the data. R.B. and N.H. carried out the statistical analyses with assistance 60. Kristensen, K., Nielsen, A., Berg, C. W., Skaug, H. & Bell, B. M. TMB: automatic and input from S.I.H., A.O.-Z. and N.M. R.B., N.H., M.C., S.W., L.W., K.J. and differentiation and Laplace approximation. J. Stat. Softw. 70, 1–21 (2016). L.E. prepared figures and tables. R.B. and L.B.M. wrote the first draft of the 61. Patil, A. P., Gething, P. W., Piel, F. B. & Hay, S. I. Bayesian geostatistics in health manuscript with assistance from S.I.H., S.P., L.E.S. and N.D.W. and all authors cartography: the perspective of malaria. Trends Parasitol. 27, 246–253 contributed to subsequent revisions. All authors provided intellectual input into (2011). aspects of this study. 62. Phillips, D. E. et al. A composite metric for assessing data on mortality and causes of death: the vital statistics performance index. Popul. Health Metr. 12, Competing interests This study was funded by the Bill & Melinda Gates 14 (2014). Foundation. Co-authors employed by the Bill & Melinda Gates Foundation 63. Schmertmann, C. P. & Gonzaga, M. R. Bayesian estimation of age-specific provided feedback on initial maps and drafts of this manuscript. Otherwise, the mortality and life expectancy for small areas with defective vital records. funders of the study had no role in study design, data collection, data analysis, Demography 55, 1363–1388 (2018). data interpretation, writing of the final report, or decision to publish. The 64. Walker, N., Hill, K. & Zhao, F. Child mortality estimation: methods used to adjust corresponding author had full access to all of the data in the study and had final for bias due to AIDS in estimating trends in under-five mortality. PLoS Med. 9, responsibility for the decision to submit for publication. e1001298 (2012). 65. Silva, R. Child mortality estimation: consistency of under-five mortality rate Additional information estimates using full birth histories and summary birth histories. PLoS Med. 9, Supplementary information is available for this paper at https://doi.org/ e1001296 (2012). 10.1038/s41586-019-1545-0. 66. Wilson, K. & Wakefield, J. Pointless spatial modeling. Biostatistics https://doi. Correspondence and requests for materials should be addressed to S.I.H. org/10.1093/biostatistics/kxy041 (2018). Peer review information Nature thanks Paloma Botella and the other, 67. Wilson, K. & Wakefield, J. Child mortality estimation incorporating anonymous, reviewer(s) for their contribution to the peer review of this work summary birth history data. Preprint at https://arxiv.org/abs/1810.04140 Reprints and permissions information is available at http://www.nature.com/ (2018). reprints. Article RESEARCH Extended Data Fig. 1 | Neonatal and infant mortality rates in 2017. units in 2017. Note that the ranges in the keys are different for the two a, b, Maps showing the mortality rates of neonates (a; birth to 28 days of maps. age) and infants (b; under 1 year of age) across second administrative-level RESEARCH Article Extended Data Fig. 2 | Impact of inequality on U5MR. a, Potential of under-5 deaths in each country. The red portion of each bar indicates reduction in the number of deaths that would have occurred if all second the number of deaths ‘attributed’ to geographical inequality in mortality administrative-level units in the 20 countries with the greatest number of rates, whereas the blue portion represents the number of deaths that would deaths of children under 5 in 2017 realized a homogenous U5MR that was remain in the scenario in which all second administrative-level units equal to that of the lowest observed mortality rate in that country. In total, within countries had the same mortality rate as the best-performing unit. 66% of under-5 deaths could have been averted if all countries maintained b, Locations of under-5 deaths ‘attributable’ to geographical inequality, mortality rates equal to the second administrative-level unit with lowest across all second administrative-level units in each country. Each country mortality. If this reference rate is set to the lowest observed rate across all has one unit highlighted with a green diamond, which is the reference of the 99 countries that were included in this study, 95% of under-5 deaths unit, or the location with the lowest mortality rate in the country in 2017. could have been averted. The size of each bar represents the total number Article RESEARCH Extended Data Fig. 3 | Relative uncertainty in U5MR estimates for 2017. 22.9 and 44.8, and 190.6 deaths per 1,000 live births. Areas in which our Relative uncertainty in second administrative-level estimates compared estimates are more uncertain are coloured with a scale of increasing blue with mean estimated U5MRs in each second administrative-level unit hue, whereas areas in which the mean estimates of U5MR are high are for 2017. Mean rates and relative uncertainty are split into population- coloured with a scale of increasing red hue. Purple areas have high, but weighted quartiles. These cut-off points indicate the relative uncertainty uncertain, estimates of U5MRs. White areas have low relative mortality, minimum, 25th, 50th and 75th percentiles, and maximum, which are with fairly certain estimates. Relative uncertainty is defined as the ratio of 0.29, 0.51, 0.63 and 0.86, and 3.12, respectively. The under-5 mortality the width of the 95% uncertainty interval to the mean estimate. minimum, 25th, 50th and 75th percentiles, and maximum are 1.4, 13.0, RESEARCH Article Extended Data Fig. 4 | Lower and upper uncertainty interval boundaries for U5MR mortality estimates in 2017. a, b, Lower (a) and upper (b) 95% uncertainty intervals for U5MR estimates across the second administrative-level units in 99 countries. Article RESEARCH Extended Data Fig. 5 | The counteracting forces of population change and mortality rate decline on total number of under-5 deaths. Arrow plots show the mortality rate strata (bins of 10 per 1,000 livebirths) in 2000. RESEARCH Article Extended Data Fig. 6 | Flowchart summarizing analytical process. Standard demographic notation were used. n, length of age bin; x, starting age of age bin; d, number of deaths; q, probability of death; a, average time lived in age bin by those who died in the age bin. nature research | reporting summary Corresponding author(s): Simon Hay Last updated by author(s): Jul 9, 2019 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist. Statistics For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable. For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above. Software and code Policy information about availability of computer code Data collection No primary data collection was carried out for this analysis. Data analysis This analysis was carried out using R version 3.5.0. The main statistical model used the Template Model Builder (TMB) software version 1.7.14 in R. All code used for these analyses is publicly available online at https://github.com/ihmeuw/lbd/tree/u5mr-2019. For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. Data Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability The findings of this study are supported by data available in public online repositories, data publicly available upon request of the data provider, and data not October 2018 publicly available due to restrictions by the data provider. Non-publicly available data were used under license for the current study but may be available from the authors upon reasonable request and with permission of the data provider. A detailed table of data sources and availability can be found in Supplementary Table 8.1. The full output of the analyses are publicly available in the Global Health Data Exchange (http://ghdx.healthdata.org/) and can further be explored via data visualization tools (https://healthdata.org/lbd/under5). 1 nature research | reporting summary Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf Life sciences study design All studies must disclose on these points even when the disclosure is negative. Sample size This observational study incorporated all available survey data sources that met the inclusion criteria as described in the Methods section. The combined dataset used in this analysis contained 25.1 million births and 1.3 million child deaths. Data exclusions Surveys were excluded due to missingness greater than 10% in date of birth or death and children ever born or died, unrealistic geographic trends compared to other surveys in nearby country-years, inability to match the microdata to geographic locations, or non-standard methodology. A full list of excluded surveys is included in Supplementary Table 8.2. Replication This is an observational study using many years of survey and surveillance data and could be replicated. Randomization This analysis is an observational mapping study, there were no experimental groups. Blinding Blinding was not relevant to this study, as it was an observational study using survey and surveillance data. Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Antibodies ChIP-seq Eukaryotic cell lines Flow cytometry Palaeontology MRI-based neuroimaging Animals and other organisms Human research participants Clinical data October 2018 2