Articles National, regional, and global causes of mortality in 5–19-year-olds from 2000 to 2019: a systematic analysis Li Liu*, Francisco Villavicencio*, Diana Yeung, Jamie Perin, Gerard Lopez, Kathleen L Strong, Robert E Black Summary Background Investments in the survival of older children and adolescents (aged 5–19 years) bring triple dividends for Lancet Glob Health 2022; now, their future, and the next generation. However, 1·5 million deaths occurred in this age group globally in 2019, 10: e337–47 nearly all from preventable causes. To better focus the attention of the global community on improving survival of See Comment page e299 children and adolescents and to guide effective policy and programmes, sound and timely cause of death data are *Authors contributed equally crucial, but often scarce. Department of Population, Family, and Reproductive Health (L Liu PhD), Department Methods In this systematic analysis, we provide updated time-series for 2000–19 of national, regional, and global cause of International Health (L Liu, of death estimates for 5–19-year-olds with age-sex disaggregation. We estimated separately for countries with high F Villavicencio PhD, versus low mortality, by data availability, and for four age-sex groups (5–9-year-olds [both sexes], 10–14-year-olds [both D Yeung MSPH, J Perin PhD, sexes], 15–19-year-old females, and 15–19-year-old males). Only studies reporting at least two causes of death were Prof R E Black MD), Johns Hopkins Bloomberg included in our analysis. We obtained empirical cause of death data through systematic review, known investigator School of Public Health, tracing, and acquisition of known national and subnational cause of death studies. We adapted the Bayesian Least Baltimore, MD, USA; Centre for Absolute Shrinkage and Selection Operator approach to address data scarcity, enhance covariate selection, produce Demographic Studies, more robust estimates, offer increased flexibility, allow country random effects, propagate coherent uncertainty, and Universitat Autònoma de Barcelona, Barcelona, Spain improve model stability. We harmonised all-cause mortality estimates with the UN Inter-agency Group for Child (F Villavicencio); Department of Mortality Estimation and systematically integrated single cause estimates as needed from WHO and UNAIDS. Maternal, Newborn, Child and Adolescent Health and Ageing, WHO, Geneva, Switzerland Findings In 2019, the global leading specific causes of death were road traffic injuries (115 843 [95% uncertainty (G Lopez MPH, K L Strong PhD) interval 110 672–125 054] deaths; 7·8% [7·5–8·1]); neoplasms (95 401 [90 744–104 812]; 6·4% [6·1–6·8]); malaria Correspondence to: (81 516 [72 150–94 477]; 5·5% [4·9–6·2]); drowning (77 460 [72 474–85 952]; 5·2% [4·9–5·5]); and diarrhoea (72 679 Dr Li Liu, Department of [66 599–82 002], 4·9% [4·5–5·3]). The leading causes varied substantially across regions. The contribution of Population, Family and communicable, maternal, perinatal, and nutritional conditions declined with age, whereas the number of deaths Reproductive Health, and associated with injuries increased. The leading causes of death were diarrhoea (51 630 [47 206–56 235] deaths; Department of International Health, Johns Hopkins 10·0% [9·5–10·5]) in 5–9-year-olds; malaria (31 587 [23 940–43 116]; 8·6% [6·6–10·4]) in 10–14-year-olds; self-harm Bloomberg School of Public (32 646 [29 530–36 416]; 13·4% [12·6–14·3]) in 15–19-year-old females; and road traffic injuries (48 757 [45 692–52 625]; Health, Baltimore, MD 21205, 13·9% [13·3–14·3]) in 15–19-year-old males. Widespread declines in cause-specific mortality were estimated across USA lliu26@jhu.edu age-sex groups and geographies in 2000–19, with few exceptions like collective violence. Interpretation Child and adolescent survival needs focused attention. To translate the vision into actions, more investments in the health information infrastructure for cause of death and in the related life-saving interventions are needed. Funding Bill & Melinda Gates Foundation and WHO. Copyright © 2022 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Introduction from maternal, communicable, and non-communicable Older children and adolescents (aged 5–19 years) make causes, and injuries for 2016–30.6,7 The Global Strategy up a quarter of the world population.1 Investments in the for Women’s, Children’s, and Adolescent’s Health survival and health of this age group bring triple (2016–30) has survival as a main objective and adolescent dividends: their health now, their future health, and the mortality as one of its 16 core indicators.8,9 These and health of the next generation.2,3 However, 1·48 million other global initiatives10–12 have recognised that inter­ deaths occurred in this age group globally in 2019,4 nearly ventions aimed at these crucial years can lead to far- all from preventable causes.5 Following the achievements reaching, multisectoral benefits in promoting lifetime made in reducing mortality in children younger than survival and health, alleviating poverty, and stimulating 5 years in the Millennium Development Goals era economies.13 (2000–15), the survival of older children and adolescents To better focus the attention of the global community has gained attention. Sustainable Development Goal 3 on improving survival of older children and adolescents emphasises healthy lives and wellbeing for all ages, and to guide effective policy and programmes, sound and and has targets for reducing premature mortality timely cause of death data are needed. However, these www.thelancet.com/lancetgh Vol 10 March 2022 e337 Articles Research in context Evidence before this study ages, including older children and adolescents. The limitations To fill the crucial knowledge gap regarding causes of deaths in associated with the Global Burden of Disease estimates are also older children and adolescents (aged 5–19 years), several the case for Global Health Estimates. estimation exercises have been done. We searched PubMed, Added value of this study Scopus, EMBASE, Web of Science, Global Health Index Medicus, We provide independent cause of death estimates for Global Health OVID, IndMed, PAHO, Popline, and Cochrane for 5–19-year-olds as a whole, 5–9-year olds and 10–14-year olds of studies published between Jan 1, 1980 and Dec 31, 2017, using both sexes, 15–19-year-old females, and 15–19-year-old males the search terms (appendix pp 11–12) related to causes of at national, regional, and global levels. We estimate separately deaths, ages 5–19, and high mortality countries. Nationally for countries with low and high mortality burden. For the low representative data have been used to examine causes of mortality countries, we use high-quality vital registration data deaths in 5–14-year-olds in China, India, Brazil, and Mexico as is or use them as model inputs. For the high mortality annually for 2005–16. However, this only covered about 40% of countries, we use verbal autopsy studies reporting at least two the global population aged 5–14 years. The Global Burden of causes of death from the general population as model inputs. Disease study produced estimates for all countries with time- series. However, it used multiple sources of cause of death data Implications of all the available evidence non-differentially, including those from vital registration and Older children and adolescent causes of deaths are under- verbal autopsy studies, and those from general and special studied. Severe and persistent empirical data gaps exist. More populations. A third study is the Global Health Estimates research is needed to fill these gaps. Based on current research, routinely updated by WHO. Using independent life tables, focused attention on survival in 5–19-year-olds is particularly Global Health Estimates combined WHO single cause needed during the Sustainable Development Goal era, in which programme estimates where available and Global Burden of many global development priorities are competing. Disease estimates to generate cause of death estimates for all data are scarce in low-income and middle-income grouped countries in nine regions according to the UN countries (LMICs) due to incomplete or non-existing Inter-agency Group for Child Mortality Estimation civil registration and vital statistics systems.14 Compared (IGME; appendix pp 3–6).4,17 with children younger than 5 years, national data on cause of death in older children and adolescents are Country modelling particularly scarce due to poor attention, commitment, For the 67 high-quality vital registration countries, we and resources. We fill this important gap by providing used vital registration data as reported or with minor cause of death estimates for 5–19-year-olds at national, adjustments. Following the same criteria as the WHO’s regional, and global levels through a systematic analysis. Global Health Estimates (GHE),16 high-quality vital registration countries are those with at least 5 years of Methods high quality and complete cause of death data for Overview people aged 15 years and older. For these countries, We provide annual cause of death estimates for vital registration data were taken from the WHO 5–19-years-olds from 2000 to 2019 for 195 countries mortality database and adjusted for missing years, See Online for appendix (appendix pp 3–6). We classify the countries into three garbage codes, ill-defined codes, miscoded causes, and categories, depending on their data availability and programme data. mortality: countries with high-quality vital registration For the 51 low mortality modelled (LMM) countries, data, countries without high-quality vital registration empirical high-quality vital registration data were used data and with low mortality, and countries without as model inputs. To parameterise the model, covariates high-quality vital registration data and with high were selected on the basis of a literature review of mortality (appendix pp 3–6). For the last two categories, sociodemographic variables and risk factors related to we estimated cause-specific mortality fractions (CSMFs) cause of deaths for 5–19-year-olds (appendix pp 8–10). using Bayesian multinomial-logistic models. Con­ We estimated the CSMFs by implementing a Bayesian sidering cause of death profile differentials and data multinomial-logistic model with country random effects.18 availability, we estimated separately for 5–9-year-olds Multinomial-logistic models have been previously used to of both sexes, 10–14-year-olds of both sexes, 15–19-year- estimate the CSMFs in a frequentist framework.19,20 The old females, and 15–19-year-old males. Our cause of Bayesian paradigm incorporates three major improve­ death categories (appendix p 7) include any cause of ments: a data driven covariate selection process, the death reported for 3% or more of global deaths in each inclusion of country random effects when nationally of the four age-sex groups according to previous representative data are available, and more coherent estimates.15,16 To derive regional mortality estimates, we uncertainty propagation. Additionally, a Least Absolute e338 www.thelancet.com/lancetgh Vol 10 March 2022 Articles Shrinkage and Selection Operator (LASSO)21 was used to produced by the WHO Immunization, Vaccines and impose a penalty on the model coefficients, enhance Biologicals Department30 into endemic and epidemic model stability, and prevent overfitting when covariates components (appendix p 20). We adopted pulmonary increase.18 The best LASSO parameter was identified and extrapulmonary tuberculosis estimates produced by through cross-validation using out-of-sample prediction the Global Tuberculosis Program31 and split them out of with the least root mean squared error (appendix the modelled lower-respiratory infections and other pp 24–25). To estimate CSMFs for each year in each CMPN fractions. We also used the median of the lower- country using the estimated parameters from the Bayesian respiratory tract infections and other CMPN fractions LASSO model, we developed a prediction database with a from high-quality vital registration data as the residual complete time series of covariates from 2000 to 2019 for lower-respiratory tract infections and other CMPN, all 195 countries (appendix pp 8–10). To convert the assuming high-quality vital registration countries have CSMFs into cause-specific mortality rates and death the minimum fractions of these causes. We then counts, we used country estimates of all-cause mortality squeezed pro rata the single cause estimates of HIV/ envelopes from UN-IGME (appendix pp 16–17).4,17 AIDS, endemic measles, extrapulmonary tuberculosis, For the 76 high mortality modelled (HMM) countries, and the residual other CMPN into the modelled other we did a systematic review for verbal autopsy studies CMPN; pulmonary tuberculosis and residual lower- (appendix pp 11–13). In addition, we procured additional respiratory tract infections were added into the modelled verbal autopsy studies through known investigator lower-respiratory tract infections (appendix p 20). If for a tracing (eg, the Million Death Study,5 the 2019 Pakistan given country single-cause estimates were not available Verbal Autopsy Social Autopsy Study [Bhutta ZA, for 2000–19, we assumed zero. University of Toronto, Totonto, ON, Canada, personal Crisis estimates were separately produced and com­munication], Country wide Mortality Surveillance for redistributed into cause of death categories depending Action [COMSA] in Mozambique,22 and a study from on the nature of the crisis (appendix p 21) and whether or Madagascar23). We created separate databases for each of not they were included in the UN-IGME envelope.4 To the four age-sex groups in which any studies providing avoid excessive crisis deaths, for each country-year we information for the relevant group were included. Similar capped the crisis fraction to the maximum fraction Bayesian LASSO models used for the LMM countries reported in the GHE16 for that country-year. were applied to the HMM countries, with different covariates as inputs (appendix pp 8–10). In addition, on China the basis of empirical evidence we only modelled malaria For China, we relied on empirical data from the Disease in individuals aged 5–14 years for HMM countries, and Surveillance Points, a sample registration system for all assumed zero malaria deaths in individuals aged 15 years ages covering 323·8 million people in China (Liu Y, and older. For high mortality country-years in which Chu Y, Yeung D, et al, unpublished data). Specifically, we malaria incidence was zero according to the Global aggregated Disease Surveillance Points cause of death Malaria Program estimates,24 malaria deaths were also data to be consistent with the cause grouping used here assumed to be zero. Considering the epidemiological and adjusted the empirical Disease Surveillance Points evidence that deaths due to malaria in 5–14-year-olds do data for the complex sample design. We also applied the not exceed the proportions of deaths due to malaria in UN-IGME envelopes at the national level for 2004–19 to 1–59-month-olds,25,26 we capped malaria CSMFs in this address under-counting32 and facilitate international group at the 1–59-month level for the corresponding comparison. We then extrapolated the 2004 CSMFs at the country-year and redistributed excess malaria deaths pro provincial levels for 2000–03 when empirical data were rata between the remaining causes post-hoc.27 The missing. For HIV/AIDS, we incorporated UNAIDS modelling process is summarised in the appendix (p 19). estimates following the same procedure described above. Details of the China national and subnational cause of Single-cause estimates death estimates are available elsewhere (Liu Y, Chu Y, In addition to modelled causes, we incorporated Yeung D, et al, unpublished data). estimates of single causes with low burden or irregular patterns. These include HIV/AIDS, tuberculosis, Uncertainty analysis measles, and crisis (collective violence, natural disasters, We assessed the uncertainty of the cause-specific and infectious disease epidemics). We calculated mortality estimates based on the uncertainty of the model UNAIDS estimated HIV/AIDS fractions on the basis of outputs, UN-IGME envelopes, and single-cause esti­ the envelopes used in Spectrum28 and applied them to mates. Uncertainty for modelled proportions were based the UN-IGME envelopes (the total number of deaths in on the posterior distribution of the Bayesian LASSO each age-sex group in each country-year).29 We split HIV/ model parameters and random effect estimates.18 We AIDS out of the modelled fraction of other communicable, accounted for uncertainty in the total number of deaths maternal, perinatal, and nutritional conditions (other and mortality rates due to all causes by taking samples CMPN). We split estimates of deaths due to measles from the posterior distribution of the UN-IGME mortality www.thelancet.com/lancetgh Vol 10 March 2022 e339 Articles For the GATHER guidelines see envelopes.4 Uncertainty for the estimates of measles, We followed the GATHER guidelines for global health http://gather-statement.org tuberculosis, HIV/AIDS, and crisis were obtained by estimates and included the GATHER checklist for sampling from the corresponding uncertainty intervals transparency and replicability (appendix p 22). We did on Monte Carlo simulations (appendix pp 17–18). our analyses using the open-source statistical software R (version 4.1.1)33 and implemented the Bayesian Maternal (0·8%) modelling in JAGS (version 4.3.0),34 with wrapper Collective violence (1·4%) Measles (1·5%) functions from the R2jags package (version 0.7.1).35 Natural disasters (0·1%) HIV/AIDS (2·3%) Lower-respiratory tract infections (4·0%) These estimates have been reviewed and endorsed by Other injuries (10·5%) representatives of WHO Member States through Injuries (32·9%) Tuberculosis (4·8%) country consultation. Road traffic injuries (7·8%) Diarrhoea (4·9%) Role of the funding source The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of Malaria (5·5%) the report. Drowning (5·2%) Communicable, maternal, perinatal, Results and nutritional In total, we used 12 642 cause of death input datapoints. conditions (38·5%) Self-harm (4·2%) They include 7710 datapoints from 67 countries for high- quality vital registration countries, 4241 empirical high- quality vital registration data points from 58 countries Interpersonal Other communicable, violence (3·6%) maternal, perinatal, (excluding those with <15 deaths) for the LMM and nutritional conditions (14·8%) countries, and 691 verbal autopsy study datapoints from 29 countries for the HMM countries (appendix p 23). Not all LMM and HMM countries were represented Congenital (1·2%) Other non-communicable Cardiovascular (2·6%) in the input data due to data availability. In 2019, 1 480 227 diseases (14·6%) Digestive system (3·8%) deaths (95% uncertainty interval [UI] 1 426 105–1 598 461) Neoplasms (6·4%) occurred in 5–19-year-olds globally. 119 802 (8·1%) global Non-communicable diseases (28·6%) deaths in 5–19-year-olds were reported in high-quality Figure 1: Global causes of death in 5–19-year-olds in 2019 vital registration countries, 84 414 (5·7%) were reported in LMM countries, 1 221 418 (82·5%) were reported in Deaths in 2000 Deaths in 2019 HMM countries, and 54 593 (3·7%) were reported in China. The breakdown of the estimation inputs by All 5–19-year-olds age-sex groups and its comparison to outputs are Road traffic injuries 162 355 (156 666–169 358) 115 843 (110 672–125 054) presented in the appendix (p 23). The geographical Neoplasms 107 826 (103 578–112 917) 95 401 (90 744–104 812) distribution of input data is shown in the appendix Malaria 116 250 (108 080–125 305) 81 516 (72 150–94 477) (p 14). On the basis of the out-of-sample prediction with Drowning 131 420 (123 226–140 348) 77 460 (72 474–85 952) the SD of the random effect set at 0·07, the best LASSO Diarrhoea 144 702 (136 371–154 976) 72 679 (66 599–82 002) penalty for each model is reported in the appendix Tuberculosis 124 546 (104 041–128 195) 70 829 (55 460–74 740) (pp 24–25). Self-harm 68 702 (65 555–72 518) 61 903 (57 290–67 651) Of the 1·48 million deaths in 5–19-year-olds, 569 930 Lower-respiratory tract infections 82 132 (77 353–93 930) 58 883 (53 980–71 901) (38·5%) deaths due to communicable, maternal, perinatal, Digestive system 101 716 (96 832–109 140) 56 323 (52 820–62 076) and nutritional (CMPN) conditions were reported, 423 878 Interpersonal violence 55 568 (53 408–58 265) 53 540 (50 355–57 928) (28·6%) deaths due to non-com­municable diseases were Cardiovascular 51 261 (47 872–55 491) 38 634 (35 273–43 091) reported, and 486 403 (32·9%) deaths due to injuries were HIV/AIDS 34 351 (31 018–38 400) 33 527 (30 744–38 872) reported (figure 1). The leading specific causes of deaths Measles 57 732 (47 964–68 795) 21 759 (16 589–31 468) in 2019 were road traffic injuries (115 843 [95% UI Collective violence 18 226 (17 308–18 964) 21 034 (18 058–24 940) 110 672–125 054] deaths; 7·8% [7·5–8·1]), neoplasms Congenital 19 117 (17 390–21 270) 17 555 (15 862–19 462) (95 401 [90 744–104 812]; 6·4% [6·1–6·8]), malaria (81 516 Maternal 18 056 (16 187–20 367) 12 167 (10 744–14 082) [72 150–94 477]; 5·5% [4·9–6·2]), drowning (77 460 Natural disasters 2270 (2149–2 398) 992 (926–1 063) [72 474–85 952]; 5·2% [4·9–5·5]), and diarrhoea (72 679 Other CMPN 317 199 (305 712–349 677) 218 570 (206 447–253 063) [66 599–82 002]; 4·9% [4·5–5·3]; figure 1; table). Other non-communicable diseases 305 954 (295 723–324 203) 215 965 (206 872–234 160) The mortality rates, number of deaths, and cause of Other injuries 202 328 (193 615–212 492) 155 631 (143 573–176 297) death distribution in 5–19-year-olds also varied between All cause 2 121 733 (2 080 030–2 194 627) 1 480 227 (1 426 105–1 598 461) regions in 2019 (figure 2). Of the nine regions presented, (Table continues on next page) west and central Africa, south Asia, and eastern and southern Africa have the highest number of deaths e340 www.thelancet.com/lancetgh Vol 10 March 2022 Articles in 5–19-year-olds. CMPN conditions are collectively Deaths in 2000 Deaths in 2019 dominant in west and central Africa and eastern and southern Africa, whereas injuries cause a higher (Continued from previous page) proportion of deaths in other regions. The Middle East 5–9-year-olds and north Africa was uniquely characterised by having Diarrhoea 105 893 (99 981–112 931) 51 630 (47 206–56 235) collective violence as the leading cause of death. The Malaria 81 802 (76 321–88 314) 49 934 (44 427–56 095) leading cause of death in west and central Africa was Lower-respiratory tract infections 51 637 (47 003–60 188) 33 851 (29 448–41 005) malaria, whereas the leading cause of death in south Asia Drowning 53 273 (48 645–58 670) 33 086 (29 589–36 514) was road traffic injuries and HIV/AIDS in eastern and Neoplasms 30 113 (28 469–31 995) 28 449 (25 901–30 925) southern Africa (appendix pp 26–27). Road traffic injuries 41 132 (38 539–44 167) 27 989 (25 521–30 332) Unique age patterns exist globally in the causes of Tuberculosis 36 710 (27 277–42 457) 22 369 (15 760–26 180) deaths in 5–19-year-olds. The top three leading specific Measles 57 737 (47 964–68 795) 21 764 (16 589–31 468) causes in each age-sex group in 2019 were diarrhoea, Congenital 19 121 (17 390–21 270) 17 556 (15 862–19 462) malaria, and lower-respiratory tract infections in Digestive system 40 244 (37 176–43 844) 16 124 (14 255–18 161) 5–9-year-olds; malaria, neoplasms, and lower-respiratory HIV/AIDS 19 173 (15 950–22 492) 8349 (7339–9270) tract infections in 10–14-year-olds; self-harm, neoplasms, Collective violence 3890 (3772–4009) 5082 (4012–6097) and cardiovascular diseases in females aged 15–19 years; Natural disasters 618 (567–665) 331 (296–371) and road traffic injuries, interpersonal violence, and self- Other CMPN 134 544 (124 684–151 495) 86 965 (77 417–96 835) harm in males aged 15–19 (table). Other non-communicable diseases 115 353 (109 958–121 534) 67 718 (62 492–73 228) The regional age-specific cause of death patterns are Other injuries 79 095 (73 610–85 149) 45 778 (41 548–50 039) even more diverse (figure 3). As expected, overall All cause 870 327 (850 443–895 917) 516 971 (487 332–549 117) the proportion of CMPN conditions reduced with age 10–14-year-olds across all regions. However, in west and central Africa Malaria 34 455 (28 378–41 164) 31 587 (23 940–43 116) and eastern and southern Africa, CMPN conditions Neoplasms 29 796 (26 662–33 202) 26 820 (23 288–34 760) accounted for more than half of the age-specific deaths Lower-respiratory tract infections 30 507 (26 560–36 943) 25 034 (21 127–34 916) in 5–14-year-olds compared with less than a quarter in Drowning 48 477 (41 623–55 624) 23 610 (20 129–30 816) other regions. West and central Africa has a particularly Road traffic injuries 34 548 (30 618–39 376) 22 542 (19 431–28 636) large proportion of deaths in 5–14-year-olds due to Diarrhoea 38 815 (32 487–46 474) 21 047 (16 751–28 938) malaria. HIV/AIDS causes a high proportion of deaths HIV/AIDS 7092 (5637–8618) 12 068 (9792–16 967) in 10–19-year-olds in eastern and southern Africa. Digestive system 22184 (19 438–27 815) 10 595 (8673–14 877) Drowning is the leading cause of death in 5–9-year-olds Tuberculosis 11 575 (7206–15 214) 7823 (4737–10 010) in south Asia. In 15–19-year-old females, self-harm is the Collective violence 4890 (4436–5048) 5835 (3350–9273) leading cause of death in south Asia, west and central Natural disasters 746 (649–833) 298 (258–344) Africa, eastern and southern Africa, and east Asia and Other CMPN 77 870 (69 730–92 122) 61 919 (52 009–84 811) Pacific, whereas collective violence is the leading cause Other non-communicable diseases 92 261 (84 162–108 025) 58 396 (51 936–72 679) of death in the Middle East and north Africa, and road traffic injuries and neoplasms are the leading specific Other injuries 57 247 (50 972–64 685) 60 572 (49 709–79 719) causes in the remaining regions (North America, Latin All cause 490 461 (455 318–543 872) 368 151 (326 403–462 580) America and the Caribbean, western Europe, and 15–19-year-old females eastern Europe and central Asia). Across the nine Self-harm 36 989 (34 805–39 688) 32 646 (29 530–36 416) regions, maternal causes contributed to the highest Neoplasms 21 425 (19 655–23 587) 18 298 (16 233–21 042) fraction of deaths in west and central Africa and Latin Cardiovascular 25 848 (23 167–29 399) 16 953 (14 823–19 879) America and the Caribbean. In 15–19-year-old males, Road traffic injuries 19 697 (18 036–21 630) 16 559 (14 483–19 573) road traffic injuries were the leading specific cause in Tuberculosis 35 496 (28 260–40 586) 15 369 (10 655–18 586) five regions (east Asia and Pacific, eastern and southern Digestive system 20 920 (18 509–23 706) 12 937 (11 098–15 128) Africa, south Asia, western Europe, and eastern Europe Maternal 18 047 (16 187–20 367) 12 161 (10 744–14 082) and central Asia). In the same age-sex group, Interpersonal violence 9097 (8073–10 309) 10 437 (8 964–12 467) interpersonal violence is the most prevalent cause of HIV/AIDS 5045 (4264–5649) 5949 (5132–6931) death in Latin America and the Caribbean, self-harm— Drowning 6725 (5971–7656) 4531 (3915–5344) closely followed by road traffic injuries—is the most Collective violence 3347 (3009–3694) 3315 (2936–3805) common in North America, and collective violence is the Natural disasters 321 (286–360) 129 (107–153) most common in the Middle East and north Africa Other CMPN 53 887 (47 765–64 529) 28 460 (23 872–37 904) (appendix pp 28–31). Other non-communicable diseases 51 472 (48 164–55 557) 47 105 (42 787–52 882) At a national level, India, Nigeria, and the Demographic Other injuries 25 761 (23 196–29 006) 18 373 (15 677–21 614) Republic of the Congo had the highest number of deaths All cause 334 080 (325 183–346 764) 243 225 (227 530–263 597) in 5–19-year-olds in 2019 (appendix pp 41–42). In these (Table continues on next page) settings, most deaths occurred in children aged 5–9 years. www.thelancet.com/lancetgh Vol 10 March 2022 e341 Articles The leading causes in this age group were drowning, Deaths in 2000 Deaths in 2019 diarrhoea, and tuberculosis in India; malaria, diarrhoea, (Continued from previous page) and lower-respiratory tract infections in Nigeria; and 15–19-year-old males malaria, measles (2019 was an outbreak year), and Road traffic injuries 66 979 (64 779–69 881) 48 757 (45 692–52 625) diarrhoea in the Demographic Republic of the Congo Interpersonal violence 46 464 (44 609–48 813) 43 102 (40 579–46 263) (appendix pp 43–45). Self-harm 31 720 (29 935–33 922) 29 255 (27 258–32 024) From 2000 to 2019, the total number of deaths in Tuberculosis 40 769 (31 917–45 086) 25 265 (18 457–28 629) 5–19-year-olds declined from 2·12 million (95% UI Neoplasms 26 487 (25 130–28 109) 21 834 (20 379–23 910) 2·08–2·19) to 1·48 million (1·43–1·60) globally. The Cardiovascular 25 407 (23 618–27 596) 21 686 (19 662–24 290) decline is in general widespread across age, time, and Digestive system 18 364 (17 182–19 838) 16 665 (15 179–18 688) cause of death (figure 4). In the four age-sex groups, the Drowning 22 948 (21 615–24 510) 16 239 (14 785–18 005) fastest decline was observed in measles for 5–9-year-olds HIV/AIDS 3044 (2657–3585) 7154 (6246–8210) at an annual average rate of reduction (AARR) of 5·57 Collective violence 6096 (5701–6519) 6795 (6109–7591) [95% UI 3·30–7·21], followed by diseases of the digestive Natural disasters 593 (541–649) 232 (200–266) system in 5–14-year-olds (5·25 [4·74–5·81] for 5–9-year- Other CMPN 50 912 (45 771–63 072) 41 237 (36 167–53 863) olds and 3·97 [2·19–5·41] for 10–14-year-olds). In both Other non-communicable diseases 46 861 (44 638–49 359) 42 747 (39 705–46 396) 15–19-year-old females and males, tuberculosis had the Other injuries 40 219 (38 045–42 820) 30 913 (28 396–34 100) largest AARR: 4·72 [3·14–6·63] for females and 2·94 All cause 426 860 (416 713–441 895) 351 881 (331 858–377 736) [1·51–4·61] for males. Deaths due to natural disasters were high in 2004 Data are number of deaths (95% uncertainty intervals). CMPN=communicable, maternal, perinatal, and nutritional in Indonesia, Sri Lanka, Thailand, and India due to conditions. the Indian Ocean tsunami; in 2008 in China due to Table: Global overall and sex-age group specific causes of death in 2000 and 2019 the Wenchuan earthquake; and 2010 in Haiti due to the Léogâne earthquake (appendix pp 36–37, 41, 43). Collective violence shows higher mortality in 2019 than 2000 in all age-sex groups except 15–19-year-old females. A higher Western Europe 10 341 deaths 4·6% Eastern Europe 40·9% 54·5% and central Asia 25 486 deaths 11·1% North America 41·8% 3·2% 47·0% Middle East and 59·3% 37·5% north Africa 18 022 deaths 71 490 deaths 15·8% 49·5% East Asia 34·7% 17·8% and Pacific 40·6% 160 602 Regional all-cause mortality 41·6% deaths rate per 1000 population aged 5–19 years No data 11·2% <3 21·7% 3–4·9 5–9·9 54·3% 34·5% 16·7% 44·6% 10–24·9 17·2% 33·7% ≥25 Latin America 66·1% Regional group-specific and Caribbean South Asia mortality fractions in 83 289 deaths 368 267 deaths 5–19-year-olds in 2019 West and central Africa Communicable, maternal, 439 440 deaths 25·4% perinatal, and nutritional 48·3% conditions Non-communicable diseases 26·3% Injuries Eastern and southern Africa 303 281 deaths Figure 2: Regional all-cause mortality rates and group-specific mortality fractions in 5–19-year-olds in 2019 e342 www.thelancet.com/lancetgh Vol 10 March 2022 Articles Maternal Measles HIV/AIDS Lower-respiratory tract infections Tuberculosis Diarrhoea Malaria Other CMPN Congenital Cardiovascular Digestive system Neoplasms Other non-communicable diseases Interpersonal violence Self-harm Drowning Road traffic injuries Other injuries Natural disasters Collective violence World West and central Africa 100 80 Mortality fractions (%) 60 40 20 0 Eastern and southern Africa Middle East and north Africa 100 80 Mortality fractions (%) 60 40 20 0 South Asia East Asia and Pacific 100 80 Mortality fractions (%) 60 40 20 0 Latin America and Caribbean North America 100 80 Mortality fractions (%) 60 40 20 0 Eastern Europe and central Asia Western Europe 100 80 Mortality fractions (%) 60 Figure 3: Global and regional 40 causes of mortality fractions by age-sex group in 2019 20 Data used to create figure reported in the appendix 0 5−9-year-olds 10−14-year-olds 15−19-year-old 15−19-year-old 5−9-year-olds 10−14-year-olds 15−19-year-old 15−19-year-old (pp 28–31). females males females males CMPN=communicable, Age-sex group Age-sex group maternal, perinatal, and nutritional conditions. www.thelancet.com/lancetgh Vol 10 March 2022 e343 Articles Maternal Tuberculosis Congenital Other non-communicable diseases Road traffic injuries Measles Diarrhoea Cardiovascular Interpersonal violence Other injuries HIV/AIDS Malaria Digestive system Self-harm Natural disasters Lower-respiratory tract infections Other CMPN Neoplasms Drowning Collective violence 5–9-year-olds 15–19-year-old females 1·2 1·0 Mortality rate per 1000 population 0·8 0·6 0·4 0·2 0 10–14-year-olds 15–19-year-old males 1·2 1·0 Mortality rate per 1000 population 0·8 0·6 0·4 0·2 0 2000 2005 2010 2015 2019 −5·0 0 7·5 2000 2005 2010 2015 2019 −5·0 0 7·5 Year AARR Year AARR Figure 4: Global cause-specific mortality rates and AARR by age-sex group, 2000–19 Natural disasters not included in the calculation of AARR due to its idiosyncratic tendency. Horizontal lines in AARR panels denote 95% uncertainty intervals. AARR=annual average rate of reduction. CMPN=communicable, maternal, perinatal, and nutritional conditions. mortality rate due to collective violence was estimated in Discussion 2011–19 compared with the previous decade with a peak in In this study, we estimated the distribution of causes 2014 across all age-sex groups, primarily due to conflicts of death for 5–19-year-olds across age-sex groups, in Syria. HIV/AIDS mortality increased between 2000 and geographies, and time. Globally in 2019, the leading 2019; however, it was higher in 2010 than 2019. specific causes are road traffic injuries, neoplasms, Widespread decline in cause-specific mortality rates was malaria, drowning, and diarrhoea. However, the leading also estimated across regions in 2000–19. For example, causes vary substantially across regions. The contribution malaria mortality decreased quickly in 5–9-year-olds and of CMPN conditions declines with age, whereas the effect gradually in 10–14-year-olds in west and central Africa of injuries increases. The leading causes in each age-sex (appendix p 32). Road traffic injuries mortality declined group globally are diarrhoea in 5–9-year-olds, malaria in steadily in eastern Europe and central Asia, North America, 10–14-year-olds, self-harm in females aged 15–19 years, and western Europe in 15–19-year-old males; however, and road traffic injuries in males aged 15–19 years. interpersonal violence mortality had a flatter trend in the Widespread declines in cause-specific mortality were Middle East and north Africa in the same age-sex group estimated across age-sex groups and geographies. (appendix pp 33–40). At a country level, India had the However, mortality due to collective violence increased in fastest decline in deaths due to diarrhoea in 5–9-year-olds the past decade (2010–19) compared with the 2000s. (appendix p 43); Nigeria and the Demographic Republic of Road traffic injuries are the most common cause of the Congo had the fastest decline in malaria mortality in death in all 5–19-year-olds, and in 15–19-year-old males the same age group (appendix pp 44–45). specifically. The AARR of road traffic injuries was e344 www.thelancet.com/lancetgh Vol 10 March 2022 Articles around 2% across all four age-sex groups. Compared cause of death in 5–19-year-olds. Of the 1·48 million with causes with high AARR (eg, measles in 5–9-years- deaths in 5–19-year-olds in 2019, 1·36 (91·1%) occurred olds) mortality rates due to road traffic injuries have in countries without high quality vital registration, declined at a moderate pace from 2000 to 2019. Due to and causes of 1·22 (82·5%) million deaths were increasing urbanisation and motorisation in LMICs, estimated using a small number of verbal autopsy deaths due to road traffic injuries might experience a studies. Of the verbal autopsy studies, few are nationally slower decline or even increase in the future.36 To representative; most are done at the subnational or accelerate the decline in road traffic injuries mortality, community levels. All of which show deep disparities in legislation (eg, drunk driving legislation coupled with data availability. Therefore, empirical data collection enforcement, through policing or technology) could be through national and subnational verbal autopsy studies effective.36 and the set-up and maintenance of sample registration Neoplasms are the second leading cause of death in systems, similar to COMSA-Mozambique,22 should be a 10–14-year-olds of both sexes and 15–19-years-old females. priority in LMICs. This will be particularly synergistic The pace of neoplasm mortality decline has been slow now that we have an estimation model allowing across the age-sex groups. Some of the decline could be more influence from nationally representative verbal explained by advancement in access to early diagnosis autopsy studies. and therapeutics (eg, for chronic myeloid leukaemia and Compared with children younger than 5 years,27 cause non-Hodgkin lymphoma).37–39 But continued advance­ of death in 5–19-year-olds is characterised by many more ments in prevention, diagnosis, and therapeutics are causes. As a result, we needed to model a higher number needed for other neoplasms.37 Neoplasm mortality is of causes (eg, 11 modelled causes in 15–19-year-old probably underestimated due to undiagnosis and females compared with eight modelled causes in children underdiagnosis in LMICs.40,41 aged 1–59 months). We chose to only estimate causes Drowning is a significant cause of death in individuals accounting for at least 3% of the age-sex-specific burden younger than 15 years and should be a particular focus globally14,15 to balance between the number of causes for child and adolescent survival programmes in south modelled, data availability, and model stability. As a Asia and east Asian and Pacific. Evidence of drowning result, we estimated large fractions of other CMPN, other intervention effectiveness targeting older children non-communicable diseases, and other injuries. We also (>5 years old) is scarce,42 especially in LMICs, but the adopted single cause estimates and the limitations existing studies suggest the effectiveness of swimming associated with those estimates are carried over. instruction in school-aged children (4–12 years old).43,44 Additionally, fewer studies and a smaller number of Communicable diseases, such as malaria and diarrhoea total deaths per study were available as model inputs in are some of the most important specific causes of death 5–19-year-olds compared with individuals younger than in LMICs, particularly in west and central Africa, eastern 5 years. For example, 520 cause of death input data points and southern Africa, and south Asia. The reasons why were available for 1–59-month-olds compared with 145 in diarrhoea remains high in 5–14-year-olds could be that in 15–19-year-old males.27 The average number of total these settings, cholera and some other causes of diarrhoea deaths per study were 1091 in neonates and 185 in (eg, enterotoxigenic Escherichia coli) occur throughout 15–19-year-old males. This led to modelling instability childhood and adulthood. Individuals aged 5–19 years, and estimates are sensitive to modelling decisions. Partly like those younger than 5 years, still die from insufficient due to data scarcity, our estimates are sensitive to the use of therapeutic oral rehydration solutions, zinc, modelling group assignment. For selected countries with antibiotics, and immunisations (such as rotavirus vaccine a probability of death from 5 to 19 around a cutoff of 10 for infants and cholera vaccine for older children). per 1000 population, the estimates could be quite Integrated community and school case management and different if they were classified as HMM countries versus vaccinations could be effective in reducing malaria and LMM countries. We attempted model averaging to diarrhoea mortality.27,45 address this uncommon discontinuity, but the results Collective violence is the only cause of death with a were too volatile to be plausible.27 largely consistent increase across all four age-sex We chose to produce estimates for 2000–19; we groups and the numbers are likely underestimated. intentionally left out 2020 due to restricted knowledge on This is primarily driven by conflicts in fragile and the direct and indirect effects of COVID-19. This conflict-affected countries.46 These countries not only highlights the acute and persistent data gap49 in child and have direct and indirect conflict-related mortality, but adolescent mortality and cause of death. also have prolonged disparities in health intervention Older children and adolescent survival need focused coverage compared with countries in the same attention because cause of death in these age groups geographical region that have not been as severely reflect changes in health risks with increasing age. affected by war.47,48 Improving survival in these age groups requires We faced a number of challenges during the multisectoral action from outside of the health estimation process. First, a major data gap exists for domain—including education, transportation and road www.thelancet.com/lancetgh Vol 10 March 2022 e345 Articles infra­structure, water and sanitation, and law enforce­ 4 United Nations Inter-agency Group for Child Mortality Estimation ment, all of which have important, complementary roles (UN IGME). Levels and trends in child mortality. Report 2020, estimates developed by the United Nations inter-agency group for in reducing mortality.12 From a life course perspective, child mortality estimation. New York, NY: United Nations we need to build on the survival gains achieved through Children’s Fund, 2020. focused investment on reducing mortality in children 5 Fadel SA, Boschi-Pinto C, Yu S, et al. Trends in cause-specific mortality among children aged 5–14 years from 2005 to 2016 in younger than 5 years mortality by extending those India, China, Brazil, and Mexico: an analysis of nationally resources to prevent and treat injuries and non- representative mortality studies. Lancet 2019; 393: 1119–27. communicable diseases in 5–19-year-olds. The global 6 UN General Assembly. Transforming our world: the 2030 Agenda for Sustainable Development. 2015. https://sdgs.un. health community has become increasingly engaged org/2030agenda (accessed Nov 4, 2021). in advocacy and actions to improve survival in this 7 WHO. Sustainable Development Goals (SDGs). 2021. https://www. age group.8,50 To translate these visions into actions, who.int/health-topics/sustainable-development-goals#tab=tab_2 (accessed Nov 4, 2021). investments are needed to build country capacity in 8 WHO. The global strategy for women’s, children’s and adolescents’ birth, death, and cause of death registration, sample health. https://www.who.int/data/maternal-newborn-child- registration, and surveys, so that life-saving inter­ adolescent-ageing/global-strategy-data (accessed April 24, 2021). ventions can be provided where they are most needed to 9 Partnership for maternal, newborn & child health report. Partnership for maternal, newborn & child health report: improve survival across the life course for all children commitments to the Every Women Every Child global strategy for and adolescents. women’s children’s and adolescents’ health. Geneva: World Health Organization, 2018. Contributors 10 WHO. Leading the realization of human rights to health and LL and REB conceptualised the study. DY led the data procurement and through health: report of the High-Level Working Group on the preparation processes with covariate preparation supported by GL and Health and Human Rights of Women, Children and Adolescents. KLS. FV led the data analyses with support from JP. All authors had Geneva: World Health Organization, 2017. access to the data. FV, DY, and JP verified the underlying data. LL, FV, 11 Costello A, Peterson S, Rasanathan K, Daelmans B, Bahl R. Where’s and DY wrote the first draft. All authors contributed to the subsequent the leadership? Future commitments of UNICEF and WHO for revisions of the draft. LL had final responsibility for the decision to global child health. BMJ 2018; 362: k3219. submit for publication. 12 Strong KL, Pedersen J, White Johansson E, et al. Patterns and trends in causes of child and adolescent mortality 2000–2016: Declaration of interests setting the scene for child health redesign. BMJ Glob Health 2021; LL, FV, DY, JP, and REB report grants from the Bill & Melinda Gates 6: e004760. Foundation. All other authors declare no competing interests. 13 Black RE, Liu L, Hartwig F, et al. Health and development from Data sharing preconception to 20 years of age to build human capital. Lancet The source code, primary inputs and cause of death data collected and (in press). For the GitHub repository see estimated are publicly available for research purposes on GitHub. 14 Mikkelsen L, Phillips DE, AbouZahr C, et al. A global assessment of https://github.com/panchoVG/ civil registration and vital statistics systems: monitoring data quality Acknowledgments and progress. Lancet 2015; 386: 1395–406. Mort5to19 The study is generously supported by the Bill & Melinda Gates 15 GBD 2016 Causes of Death Collaborators. Global, regional, and Foundation (OPP1172551). We thank the families and researchers who national age-sex specific mortality for 264 causes of death, have contributed to the 5–19-year-old cause of death literature through 1980–2016: a systematic analysis for the Global Burden of Disease survey interviews, vital registration, and scientific investigations. Study 2016. Lancet 2017; 390: 1151–210. We thank our collaborators at WHO for their technical inputs and 16 WHO. Global health estimates 2016: deaths by cause, age, sex, by disease-specific mortality estimates, including Doris Ma Fat and country and by region, 2000–2016. Geneva: World Health Gretchen Stevens for causes of death in countries with adequate vital Organization, 2018. registration, Noor Abdisalan and John Aponte for malaria estimates in 17 Masquelier B, Hug L, Sharrow D, et al. Global, regional, and non-endemic countries, Minal Patel for measles estimates among national mortality trends in youth aged 15–24 years between 1990 5–9-year-olds, and Pete Dodd, Philippe Glaziou, and Ben Marais for and 2019: a systematic analysis. Lancet Glob Health 2021; estimates of pulmonary and non-pulmonary tuberculosis. We also thank 9: e409–17. Mary Mahy and Juliana Daher from UNAIDS for HIV estimates, and 18 Mulick AR, Oza S, Prieto-Merino D, Villavicencio F, Cousens S, Colin Mathers for crisis estimates. We are grateful to Simon Cousens, Perin J. A Bayesian hierarchical model with integrated Amy Mulick, and David Prieto-Merino for their support with the covariate selection and misclassification matrices to estimate Bayesian model. We also acknowledge the contribution of our Graduate neonatal and child causes of death. medRxiv 2021; published online Feb 12. https://doi.org/10.1101/2021.02.10.21251488 Research Assistants Yijing Feng, Xuejuan Ning, Jingyao Hong, and (preprint). Dana Sarnak, who worked tirelessly during the process of systematic 19 Liu L, Johnson HL, Cousens S, et al. Global, regional, and national review of the literature detailing the cause of death in 5–19-year-olds and causes of child mortality: an updated systematic analysis for 2010 extracting study covariates. We thank Christoph Keller and with time trends since 2000. Lancet 2012; 379: 2151–61. Virginie Buchard at NASA for providing PM2·5 covariate data. We 20 Liu L, Oza S, Hogan D, et al. Global, regional, and national causes appreciate the feedback provided by in-country stakeholders during the of under-5 mortality in 2000–15: an updated systematic analysis WHO country consultation process. Thanks to Brittany Furgal for the with implications for the Sustainable Development Goals. Lancet administrative and editorial support. 2016; 388: 3027–35. References 21 Park T, Casella G. The Bayesian Lasso. J Am Stat Assoc 2008; 1 United Nations Department of Economic and Social Affairs. World 103: 681–86. Population Prospects 2019. 2019. https://population.un.org/wpp/ 22 Amouzou A, Kante A, Macicame I, et al. National sample vital (accessed April 24, 2021). registration system: a sustainable platform for COVID-19 and other 2 WHO. Global Accelerated Action for the Health of Adolescents infectious diseases surveillance in low and middle-income (AA-HA!): guidance to support country implementation. Geneva: countries. J Glob Health 2020; 10: 020368. World Health Organization, 2017. 23 Masquelier B, Pison G, Rakotonirina J, Rasoanomenjanahary A. 3 WHO. Why invest in adolescent health? 2021. https://www.who.int/ Estimating cause-specific mortality in Madagascar: an evaluation of maternal_child_adolescent/topics/adolescence/why-invest/en/ death notification data from the capital city. Popul Health Metr 2019; (accessed April 14, 2021). 17: 8. e346 www.thelancet.com/lancetgh Vol 10 March 2022 Articles 24 WHO. World malaria report 2020: 20 years of global progress and 38 Gramatges MM, Rabin KR. The adolescent and young adult with challenges. Geneva: World Health Organization, 2020. cancer: state of the art—acute leukemias. Curr Oncol Rep 2013; 25 Abdullah S, Adazu K, Masanja H, et al. Patterns of age-specific 15: 317–24. mortality in children in endemic areas of sub-Saharan Africa. 39 Vishnu P, Aboulafia DM. AIDS-related non-Hodgkin’s lymphoma Am J Trop Med Hyg 2007; 77: 99–105. in the era of highly active antiretroviral therapy. Adv Hematol 2012; 26 Brooker SJ, Clarke S, Fernando D, et al. Malaria in middle 2012: 485943. childhood and adolescence. In Bundy DAP, Silva N, Horton S, 40 Gupta S, Howard SC, Hunger SP, et al. Treating childhood cancer Jamison DT, Patton GC, eds. Child and adolescent health and in low- and middle-income countries. In: Gelband H, Jha P, development. Washington DC: World Bank Group, 2017. Sankaranarayanan R, Horton S, eds. Cancer: disease control 27 Perin J, Mulick AR, Yeung D, et al. Global, regional, and national priorities, 3rd edn. Washington DC: The World Bank, 2015. causes of under-5 mortality in 2000–19: an updated systematic 41 Atun R, Bhakta N, Denburg A, et al. Sustainable care for children analysis with implications for the Sustainable Development Goals. with cancer: a Lancet Oncology Commission. Lancet Oncol 2020; Lancet Child Adolesc Health 2021; published online Nov 7. 21: e185–224. https://doi.org/10.1016/S2352-4642(21)00311-4. 42 Wallis BA, Watt K, Franklin RC, Taylor M, Nixon JW, Kimble RM. 28 UNAIDS. Quick start guide for spectrum. 2020. https://www. Interventions associated with drowning prevention in children and unaids.org/sites/default/files/media_asset/QuickStartGuide_ adolescents: systematic literature review. Inj Prev 2015; 21: 195–204. Spectrum_en.pdf (accessed Nov 4, 2021). 43 Mecrow TS, Linnan M, Rahman A, et al. Does teaching children to 29 UNAIDS. UNAIDS Data, 2020. 2020. https://www.unaids.org/ swim increase exposure to water or risk-taking when in the water? sites/default/files/media_asset/2020_aids-data-book_en.pdf Emerging evidence from Bangladesh. Inj Prev 2015; 21: 185–88. (accessed Nov 3, 2021). 44 Rahman F, Bose S, Linnan M, et al. Cost-effectiveness of an injury 30 Simons E, Ferrari M, Fricks J, et al. Assessment of the 2010 global and drowning prevention program in Bangladesh. Pediatrics 2012; measles mortality reduction goal: results from a model of 130: e1621–28. surveillance data. Lancet 2012; 379: 2173–78. 45 Fischer Walker CL, Sack D, Black RE. Etiology of diarrhea in older 31 WHO. Global tubuerculosis report 2020. Geneva: World Health children, adolescents and adults: a systematic review. Organization, 2020. PLoS Negl Trop Dis 2010; 4: e768. 32 Guo K, Yin P, Wang L, et al. Propensity score weighting for 46 World Bank. FY21 list of fragile and conflict-affected situations. addressing under-reporting in mortality surveillance: a proof-of- 2021. https://pubdocs.worldbank.org/en/888211594267968803/ concept study using the nationally representative mortality data in FCSList-FY21.pdf (accessed Nov 4, 2021). China. Popul Health Metr 2015; 13: 16. 47 Akseer N, Wright J, Tasic H, et al. Women, children and adolescents 33 R Core Team. R: a language and environment for statistical in conflict countries: an assessment of inequalities in intervention computing. 2021. https://www.r-project.org/ (accessed Nov 4, 2021). coverage and survival. BMJ Glob Health 2020; 5: e002214. 34 Plummer M. JAGS: a program for analysis of Bayesian graphical 48 Wagner Z, Heft-Neal S, Bhutta ZA, Black RE, Burke M, Bendavid E. models using Gibbs sampling. 3rd international workshop on Armed conflict and child mortality in Africa: a geospatial analysis. distributed statistical computing. 2003. https://www.r-project.org/ Lancet 2018; 392: 857–65. conferences/DSC-2003/Proceedings/ (accessed Feb 27, 2019). 49 WHO, UN Department of Economic and Social Affairs. Technical 35 Su YS, Yajima M. R2jags: Using R to run ‘JAGS’. R package Advisory Group on COVID-19 Mortality Assessment. 2021. https:// version 0.7-1. 2020. https://cran.r-project.org/web/packages/R2jags/ www.who.int/data/technical-advisory-group/covid-19--mortality- (accessed Feb 23, 2021). assessment/introduction (accessed Jan 28, 2022). 36 Staton C, Vissoci J, Gong E, et al. Road traffic injury prevention 50 Qiao J, Wang Y, Li X, et al. A Lancet Commission on 70 years of initiatives: a systematic review and metasummary of effectiveness women’s reproductive, maternal, newborn, child, and adolescent in low and middle income countries. PLoS One 2016; 11: e0144971. health in China. Lancet 2021; 397: 2497–536. 37 World Economic Forum. This film is causing the cost of cancer drugs in China to fall. 2018. https://www.weforum.org/ agenda/2018/07/price-of-cancer-drugs-in-china-set-to-drop- following-hit-film/ (accessed Nov 4, 2021). www.thelancet.com/lancetgh Vol 10 March 2022 e347