Causes of child death: comparison of MCEE and GBD 2013 estimates
2015; Elsevier BV; Volume: 385; Issue: 9986 Linguagem: Inglês
10.1016/s0140-6736(15)61132-1
ISSN1474-547X
AutoresLi Liu, Robert E. Black, Simon Cousens, Colin Mathers, Joy E. Lawn, Daniel R Hogan,
Tópico(s)Neonatal Respiratory Health Research
ResumoChild cause of death (COD) estimates for 2013 produced by the Maternal and Child Epidemiology Estimation group (MCEE, formerly the Child Health Epidemiology Reference Group or CHERG) and the WHO were published (online on Sept 30, 2014) in The Lancet.1Liu L Oza S Hogan D et al.Global, regional, and national causes of child mortality in 2000–13, with projections to inform post-2015 priorities: an updated systematic analysis.Lancet. 2015; 385: 430-440Summary Full Text Full Text PDF PubMed Scopus (1920) Google Scholar Estimates of child COD from the Global Burden of Disease (GBD) Study were later published in The Lancet (online on Dec 17, 2014).2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar The GBD authors compared our previously published estimates for 20103Liu L Johnson HL Cousens S et al.Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000.Lancet. 2012; 379: 2151-2161Summary Full Text Full Text PDF PubMed Scopus (2641) Google Scholar with their new estimates for 2010 providing both differing causes and numbers of deaths than they published previously.2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar We believe it is more useful to have a comparison of both sets of estimates for 2013, as shown in the table.TableMCEE and GBD estimates of causes of deaths among children aged 1–59 months and 0–27 days, worldwide, in 20131–59 months0–27 daysMCEEGBDRelative differenceMCEEGBDRelative differencePneumonia800 (681–923)709 (629–791)11136 (84–219)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.197 (169–225)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.−45*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.Diarrhoea558 (429–731)475 (398–545)1520 (12–33)†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.45 (37–53)†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.−125†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.Malaria456 (351–546)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.570 (438–733)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.−25*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.0†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.17 (11–26)†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.NA†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.Injury324 (258–391)321 (278–372)1......Meningitis151 (125–185)121 (90–157)20......AIDS103 (76–142)†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.64 (59–72)†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.38†Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates.......Measles102 (74–166)82 (42–145)20......Pertussis60 (43–94)56 (21–127)7......Preterm birth complications......965 (615–1537)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.693 (554–854)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.28*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.Intrapartum complications......662 (421–1054)612 (492–724)8Sepsis......421 (269–688)342 (215–479)19Congenital......276 (175–438)247 (219–280)11Tetanus......49 (32–79)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.26 (12–39)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.47*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.Other967 (781–1134)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.1268 (NA)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.−31*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.232 (145–373)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.452 (NA)*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.−95*Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.Total35213666−4276126145Data are N in thousands (uncertainty range). GBD=Global Burden of Disease. MCEE=Maternal and Child Epidemiology Estimation. NA=not available.* Causes where the absolute value of the relative difference, defined as difference between the MCEE and GBD estimates divided by the MCEE estimates and multiplied by 100, is at least 25.† Causes where the 95% uncertainty ranges do not overlap between the MCEE and GBD estimates. Open table in a new tab The estimates have good agreement (≤5% difference) on the global number of all-cause deaths in the neonatal and 1–59 month periods. The agreement for numbers of deaths for major causes is mixed. In the 1–59 month period, six out of eight causes agree (relative difference is less than 25%). For two causes, malaria and AIDS, the relative difference is 25% or greater. The uncertainty ranges did not overlap for AIDS, but this was estimated to cause few deaths by both teams. The agreement for neonatal causes is poorer. For three causes—intrapartum complications, sepsis, and congenital—the estimates are fairly similar. The estimates differ for preterm birth complications and other neonatal conditions, which might be due to differences in International Classification of Diseases coding of neonatal causes by GBD as other causes instead of preterm birth complications. Although neonatal diarrhoea estimates are substantially different, diarrhoea is a minor cause of neonatal deaths in both sets of estimates. GBD estimated that globally 16 800 neonatal deaths were attributable to malaria in 2013. This is implausible. Malaria deaths are rare among children younger than one month,4Enweronu-Laryea CC Adjei GO Mensah B Duah N Quashie NB Prevalence of congenital malaria in high-risk Ghanaian newborns: a cross-sectional study.Malar J. 2013; 12: 17Crossref PubMed Scopus (23) Google Scholar, 5Falade C Mokuolu O Okafor H et al.Epidemiology of congenital malaria in Nigeria: a multi-centre study.Trop Med Int Health. 2007; 12: 1279-1287Crossref PubMed Scopus (71) Google Scholar and the previous GBD COD exercise did not report any neonatal malaria death in 2010.6Lozano R Naghavi M Foreman K et al.Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010.Lancet. 2013; 380: 2095-3028Summary Full Text Full Text PDF Scopus (9685) Google Scholar It is unclear what changes in input data or modelling strategy led to these non-trivial neonatal malaria estimates.Overall, given differences in model inputs and modelling strategies between the two estimation processes, the major cause-specific estimates are not greatly different. To further improve precision and resolution in the estimates, additional high quality data need to be collected and made accessible.We also compared estimates for the five countries with the largest burden of under-five deaths. Overall the number of all-cause deaths agree reasonably well across India, Nigeria, Pakistan, DR Congo, and China. The agreement for cause-specific deaths largely reflects that seen at the global level. For example, estimates of 1–59 month deaths due to diarrhoea and pneumonia tend to be higher globally in the MCEE estimates and are higher in three of the five countries, while GBD malaria estimates tend to be higher globally and in two of the three countries with a substantial malaria burden. MCEE estimated more postneonatal pneumonia and neonatal preterm birth complication deaths than GBD in four of the five countries.Moreover, we compared input data and methods. GBD dropped all studies using InterVA for analysis of verbal autopsy causes of death, but MCEE did not. GBD also used verbal autopsy studies from special populations (eg, refugees), which might skew the COD distribution; MCEE excluded all studies from special populations. In addition, GBD included studies of single COD that might be subject to over-reporting of the cause of interest. For example, studies reporting only malaria deaths might select highly endemic populations and could overestimate malaria burden. To ensure all cause fractions add to one, GBD relied on their CODCorrect programme.2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar It is unclear whether the systematic adjustment that CODCorrect imposed on cause fractions, which is primarily based on uncertainty associated with cause fractions, addressed the possible biases introduced by using single cause studies. Again, using malaria estimates as an example, a plethora of malaria single cause studies have been published, which might explain the relatively narrow uncertainty, thus the malaria fraction would need little reduction in CODCorrect (only −1·25%, webtable 24).2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar This might lead to an overestimation of malaria deaths. MCEE sought to address this issue by using verbal autopsy studies reporting two or more causes for postneonates and four or more causes for neonates.Furthermore, the GBD report2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar stated that MCEE estimated no deaths due to neonatal causes after the neonatal period. This is incorrect. We estimated1Liu L Oza S Hogan D et al.Global, regional, and national causes of child mortality in 2000–13, with projections to inform post-2015 priorities: an updated systematic analysis.Lancet. 2015; 385: 430-440Summary Full Text Full Text PDF PubMed Scopus (1920) Google Scholar, 3Liu L Johnson HL Cousens S et al.Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000.Lancet. 2012; 379: 2151-2161Summary Full Text Full Text PDF PubMed Scopus (2641) Google Scholar deaths in the 1–59 month period due to congenital malformation, preterm birth complications, and other neonatal causes, but these causes were collapsed into “other” in the 1–59 month age group for reporting. The GBD team questioned MCEE's partitioning of the world into several models (MCEE derives estimates separately for three country groupings based on all-cause mortality rates and availability of vital registration data). However, the GBD Study assumes modelling all countries simultaneously permits an accurate characterisation of widely divergent COD distributions across countries—eg, that patterns and determinants of COD in Sweden are informative for the COD distribution in South Sudan.Global and national estimations of causes of child death are complex endeavours and need to have full transparency and replicability to ensure confidence in the results and opportunities for improvement. We applaud the recent effort of GBD to publish a small portion of their source code for DisMod. However, the CODEm code used for the majority of causes is not available. Additionally, due to the unavailability of some important input data, replication of the estimates is impossible. As a result, understanding of the reasons for differences between MCEE and GBD estimates is more speculation than calculation. Models should be made accessible by provision at the time of publication of all inputs and analysis code. To this end, we made our input data and analysis for our 2013 estimates publicly available when our paper was published. We hope this will become standard for all.We declare no competing interests. Child cause of death (COD) estimates for 2013 produced by the Maternal and Child Epidemiology Estimation group (MCEE, formerly the Child Health Epidemiology Reference Group or CHERG) and the WHO were published (online on Sept 30, 2014) in The Lancet.1Liu L Oza S Hogan D et al.Global, regional, and national causes of child mortality in 2000–13, with projections to inform post-2015 priorities: an updated systematic analysis.Lancet. 2015; 385: 430-440Summary Full Text Full Text PDF PubMed Scopus (1920) Google Scholar Estimates of child COD from the Global Burden of Disease (GBD) Study were later published in The Lancet (online on Dec 17, 2014).2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar The GBD authors compared our previously published estimates for 20103Liu L Johnson HL Cousens S et al.Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000.Lancet. 2012; 379: 2151-2161Summary Full Text Full Text PDF PubMed Scopus (2641) Google Scholar with their new estimates for 2010 providing both differing causes and numbers of deaths than they published previously.2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar We believe it is more useful to have a comparison of both sets of estimates for 2013, as shown in the table. Data are N in thousands (uncertainty range). GBD=Global Burden of Disease. MCEE=Maternal and Child Epidemiology Estimation. NA=not available. The estimates have good agreement (≤5% difference) on the global number of all-cause deaths in the neonatal and 1–59 month periods. The agreement for numbers of deaths for major causes is mixed. In the 1–59 month period, six out of eight causes agree (relative difference is less than 25%). For two causes, malaria and AIDS, the relative difference is 25% or greater. The uncertainty ranges did not overlap for AIDS, but this was estimated to cause few deaths by both teams. The agreement for neonatal causes is poorer. For three causes—intrapartum complications, sepsis, and congenital—the estimates are fairly similar. The estimates differ for preterm birth complications and other neonatal conditions, which might be due to differences in International Classification of Diseases coding of neonatal causes by GBD as other causes instead of preterm birth complications. Although neonatal diarrhoea estimates are substantially different, diarrhoea is a minor cause of neonatal deaths in both sets of estimates. GBD estimated that globally 16 800 neonatal deaths were attributable to malaria in 2013. This is implausible. Malaria deaths are rare among children younger than one month,4Enweronu-Laryea CC Adjei GO Mensah B Duah N Quashie NB Prevalence of congenital malaria in high-risk Ghanaian newborns: a cross-sectional study.Malar J. 2013; 12: 17Crossref PubMed Scopus (23) Google Scholar, 5Falade C Mokuolu O Okafor H et al.Epidemiology of congenital malaria in Nigeria: a multi-centre study.Trop Med Int Health. 2007; 12: 1279-1287Crossref PubMed Scopus (71) Google Scholar and the previous GBD COD exercise did not report any neonatal malaria death in 2010.6Lozano R Naghavi M Foreman K et al.Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010.Lancet. 2013; 380: 2095-3028Summary Full Text Full Text PDF Scopus (9685) Google Scholar It is unclear what changes in input data or modelling strategy led to these non-trivial neonatal malaria estimates. Overall, given differences in model inputs and modelling strategies between the two estimation processes, the major cause-specific estimates are not greatly different. To further improve precision and resolution in the estimates, additional high quality data need to be collected and made accessible. We also compared estimates for the five countries with the largest burden of under-five deaths. Overall the number of all-cause deaths agree reasonably well across India, Nigeria, Pakistan, DR Congo, and China. The agreement for cause-specific deaths largely reflects that seen at the global level. For example, estimates of 1–59 month deaths due to diarrhoea and pneumonia tend to be higher globally in the MCEE estimates and are higher in three of the five countries, while GBD malaria estimates tend to be higher globally and in two of the three countries with a substantial malaria burden. MCEE estimated more postneonatal pneumonia and neonatal preterm birth complication deaths than GBD in four of the five countries. Moreover, we compared input data and methods. GBD dropped all studies using InterVA for analysis of verbal autopsy causes of death, but MCEE did not. GBD also used verbal autopsy studies from special populations (eg, refugees), which might skew the COD distribution; MCEE excluded all studies from special populations. In addition, GBD included studies of single COD that might be subject to over-reporting of the cause of interest. For example, studies reporting only malaria deaths might select highly endemic populations and could overestimate malaria burden. To ensure all cause fractions add to one, GBD relied on their CODCorrect programme.2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar It is unclear whether the systematic adjustment that CODCorrect imposed on cause fractions, which is primarily based on uncertainty associated with cause fractions, addressed the possible biases introduced by using single cause studies. Again, using malaria estimates as an example, a plethora of malaria single cause studies have been published, which might explain the relatively narrow uncertainty, thus the malaria fraction would need little reduction in CODCorrect (only −1·25%, webtable 24).2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar This might lead to an overestimation of malaria deaths. MCEE sought to address this issue by using verbal autopsy studies reporting two or more causes for postneonates and four or more causes for neonates. Furthermore, the GBD report2GBD 2013 Mortality and Causes of Death CollaboratorsGlobal, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013.Lancet. 2015; 385: 117-171Summary Full Text Full Text PDF PubMed Scopus (5239) Google Scholar stated that MCEE estimated no deaths due to neonatal causes after the neonatal period. This is incorrect. We estimated1Liu L Oza S Hogan D et al.Global, regional, and national causes of child mortality in 2000–13, with projections to inform post-2015 priorities: an updated systematic analysis.Lancet. 2015; 385: 430-440Summary Full Text Full Text PDF PubMed Scopus (1920) Google Scholar, 3Liu L Johnson HL Cousens S et al.Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000.Lancet. 2012; 379: 2151-2161Summary Full Text Full Text PDF PubMed Scopus (2641) Google Scholar deaths in the 1–59 month period due to congenital malformation, preterm birth complications, and other neonatal causes, but these causes were collapsed into “other” in the 1–59 month age group for reporting. The GBD team questioned MCEE's partitioning of the world into several models (MCEE derives estimates separately for three country groupings based on all-cause mortality rates and availability of vital registration data). However, the GBD Study assumes modelling all countries simultaneously permits an accurate characterisation of widely divergent COD distributions across countries—eg, that patterns and determinants of COD in Sweden are informative for the COD distribution in South Sudan. Global and national estimations of causes of child death are complex endeavours and need to have full transparency and replicability to ensure confidence in the results and opportunities for improvement. We applaud the recent effort of GBD to publish a small portion of their source code for DisMod. However, the CODEm code used for the majority of causes is not available. Additionally, due to the unavailability of some important input data, replication of the estimates is impossible. As a result, understanding of the reasons for differences between MCEE and GBD estimates is more speculation than calculation. Models should be made accessible by provision at the time of publication of all inputs and analysis code. To this end, we made our input data and analysis for our 2013 estimates publicly available when our paper was published. We hope this will become standard for all. We declare no competing interests. Global, regional, and national causes of child mortality in 2000–13, with projections to inform post-2015 priorities: an updated systematic analysisOur projection results provide concrete examples of how the distribution of child causes of deaths could look in 15–20 years to inform priority setting in the post-2015 era. More evidence is needed about shifts in timing, causes, and places of under-5 deaths to inform child survival agendas by and beyond 2015, to end preventable child deaths in a generation, and to count and account for every newborn and every child. Full-Text PDF Global, regional, and national age–sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013For most countries, the general pattern of reductions in age-sex specific mortality has been associated with a progressive shift towards a larger share of the remaining deaths caused by non-communicable disease and injuries. Assessing epidemiological convergence across countries depends on whether an absolute or relative measure of inequality is used. Nevertheless, age-standardised death rates for seven substantial causes are increasing, suggesting the potential for reversals in some countries. Important gaps exist in the empirical data for cause of death estimates for some countries; for example, no national data for India are available for the past decade. Full-Text PDF Causes of child death: comparison of MCEE and GBD 2013 estimates – Authors' replyInformation about causes of childhood mortality is an important input to global health policy making. Two sets of estimates are available from the Global Burden of Disease Study (GBD) 20131 and the Maternal and Child Epidemiology Estimation group (MCEE, formerly known as the Child Health Epidemiology Reference Group [CHERG]). Liu and colleagues2 make a comparison between estimates for 2013 from the two studies and speculate reasons for the differences. Full-Text PDF
Referência(s)