Articles ∂@克 Neonatal,postneonatal,childhood,and under-5 mortality for 187 countries,1970-2010:a systematic analysis of progress towards Millennium Development Goal 4 Julie Knoll Rajaratnam,Jake R Marcus,AbrahamD Flaxman,Haidong Wang,Alison Levin-Rector,Laura Dwyer,Megan Costa,Alan DLopez, Christopher JL Murray Summary Lancet2010:375:1988-2008 Background Previous assessments have highlighted that less than a quarter of countries are on track to achieve This online publication has Millennium Development Goal 4(MDG 4),which calls for a two-thirds reduction in mortality in children younger been corrected. than 5 years between 1990 and 2015.In view of policy initiatives and investments made since 2000,it is important to The corrected version first appeared at TheLancet.com on see if there is acceleration towards the MDG 4 target.We assessed levels and trends in child mortality for 187 countries August 17,2010 from1970to2010. Published Online May24.2010 Methods We compiled a database of 16174 measurements of mortality in children younger than 5 years for D0t10.101650140 187 countries from 1970 to 2009,by use of data from all available sources,including vital registration systems, 6736(10)60703-9 summary birth histories in censuses and surveys,and complete birth histories.We used Gaussian process regression See Comment page 1941 to generate estimates of the probability of death between birth and age 5 years.This is the first study that uses Institute for Health Metrics Gaussian process regression to estimate child mortality,and this technique has better out-of-sample predictive validity and Evaluation,University of Washington,Seattle,WA,USA than do previous methods and captures uncertainty caused by sampling and non-sampling error across data types. (JK Rajaratnam PhD, Neonatal,postneonatal,and childhood mortality was estimated from mortality in children younger than 5 years by JR Marcus BAAD Flaxman PhD, use of the 1760 measurements from vital registration systems and complete birth histories that contained specific HWang PhD, information about neonatal and postneonatal mortality. A Levin-Rector BSPH, LDwyer BA,M Costa BA ProfC]L Murray MD);and Findings Worldwide mortality in children younger than 5 years has dropped from 11.9 million deaths in 1990 to School of Population Health, 7.7 million deaths in 2010,consisting of 3.1 million neonatal deaths,2.3 million postneonatal deaths,and 2.3 million University of Queensland, childhood deaths(deaths in children aged 1-4 years).33.0%of deaths in children younger than 5 years occur in south Brisbane,QLD,Australia (ProfA D Lopez PhD) Asia and 49.6%occur in sub-Saharan Africa,with less than 1%of deaths occurring in high-income countries.Across Correspondence to 21 regions of the world,rates of neonatal,postneonatal,and childhood mortality are declining.The global decline Prof Christopher ]L Murray. from 1990 to 2010 is 2.1%per year for neonatal mortality,2.3%for postneonatal mortality,and 2.2%for childhood nstitute for Health Metrics and mortality.In 13 regions of the world,including all regions in sub-Saharan Africa,there is evidence of accelerating Evaluation,University of declines from 2000 to 2010 compared with 1990 to 2000.Within sub-Saharan Africa,rates of decline have increased Washington,23015th Avenue, Suite 600 Seattle by more than 1%in Angola,Botswana,Cameroon,Congo,Democratic Republic of the Congo,Kenya,Lesotho,Liberia, WA 98121,USA Rwanda,Senegal,Sierra Leone,Swaziland,and The Gambia. cjlm@uw.edu Interpretation Robust measurement of mortality in children younger than 5 years shows that accelerating declines are occurring in several low-income countries.These positive developments deserve attention and might need enhanced policy attention and resources Funding Bill Melinda Gates Foundation. Introduction The MDG 4 target has shifted the focus from tracking There are only 5 years left to achieve Millennium levels of child mortality to assessing whether countries Development Goal 4(MDG 4),which calls for a two-thirds are reducing child mortality at the 4.4%rate per year reduction in mortality in children younger than 5 years needed to achieve the two-thirds reduction in 25 years. between 1990 and 2015.Regular assessment of levels and Accurate assessments of rates of change need more trends in child mortality is essential for countries to ascertain robust measurement with narrower uncertainty intervals their progress towards this goal and to take action to meet it. than do assessments of levels.Although there have been Previous appraisals of mortality in children younger than substantial investments in the collection of data such as 5 years suggest that few countries are on track to meet summary and complete birth histories to measure child MDG 4.In each of these studies,no more than 26%of mortality,assessments of trends have varied substantially low-income and middle-income countries examined were from year to year and from source to source.For example. deemed to be on track to reach this target.Groups such as the list of the ten countries with the fastest rates of decline the Countdown to 2015 Initiative'have therefore tried to in child mortality between 1990 and 2007,as reported by rally support to accelerate progress in child mortality UNICEF in 2008,*UNICEF in 2009,and the UN 1988 www.thelancet.com Vol 375 June 5,2010
1988 www.thelancet.com Vol 375 June 5, 2010 Articles Lancet 2010; 375: 1988–2008 This online publication has been corrected. The corrected version fi rst appeared at TheLancet.com on August 17, 2010 Published Online May 24, 2010 DOI:10.1016/S0140- 6736(10)60703-9 See Comment page 1941 Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, USA (J K Rajaratnam PhD, J R Marcus BA, A D Flaxman PhD, H Wang PhD, A Levin-Rector BSPH, L Dwyer BA, M Costa BA, Prof C J L Murray MD); and School of Population Health, University of Queensland, Brisbane, QLD, Australia (Prof A D Lopez PhD) Correspondence to: Prof Christopher J L Murray, Institute for Health Metrics and Evaluation, University of Washington, 2301 5th Avenue, Suite 600, Seattle, WA 98121, USA cjlm@uw.edu Neonatal, postneonatal, childhood, and under-5 mortality for 187 countries, 1970–2010: a systematic analysis of progress towards Millennium Development Goal 4 Julie Knoll Rajaratnam, Jake R Marcus, Abraham D Flaxman, Haidong Wang, Alison Levin-Rector, Laura Dwyer, Megan Costa, Alan D Lopez, Christopher J L Murray Summary Background Previous assessments have highlighted that less than a quarter of countries are on track to achieve Millennium Development Goal 4 (MDG 4), which calls for a two-thirds reduction in mortality in children younger than 5 years between 1990 and 2015. In view of policy initiatives and investments made since 2000, it is important to see if there is acceleration towards the MDG 4 target. We assessed levels and trends in child mortality for 187 countries from 1970 to 2010. Methods We compiled a database of 16 174 measurements of mortality in children younger than 5 years for 187 countries from 1970 to 2009, by use of data from all available sources, including vital registration systems, summary birth histories in censuses and surveys, and complete birth histories. We used Gaussian process regression to generate estimates of the probability of death between birth and age 5 years. This is the fi rst study that uses Gaussian process regression to estimate child mortality, and this technique has better out-of-sample predictive validity than do previous methods and captures uncertainty caused by sampling and non-sampling error across data types. Neonatal, postneonatal, and childhood mortality was estimated from mortality in children younger than 5 years by use of the 1760 measurements from vital registration systems and complete birth histories that contained specifi c information about neonatal and postneonatal mortality. Findings Worldwide mortality in children younger than 5 years has dropped from 11·9 million deaths in 1990 to 7·7 million deaths in 2010, consisting of 3·1 million neonatal deaths, 2·3 million postneonatal deaths, and 2·3 million childhood deaths (deaths in children aged 1–4 years). 33·0% of deaths in children younger than 5 years occur in south Asia and 49·6% occur in sub-Saharan Africa, with less than 1% of deaths occurring in high-income countries. Across 21 regions of the world, rates of neonatal, postneonatal, and childhood mortality are declining. The global decline from 1990 to 2010 is 2·1% per year for neonatal mortality, 2·3% for postneonatal mortality, and 2·2% for childhood mortality. In 13 regions of the world, including all regions in sub-Saharan Africa, there is evidence of accelerating declines from 2000 to 2010 compared with 1990 to 2000. Within sub-Saharan Africa, rates of decline have increased by more than 1% in Angola, Botswana, Cameroon, Congo, Democratic Republic of the Congo, Kenya, Lesotho, Liberia, Rwanda, Senegal, Sierra Leone, Swaziland, and The Gambia. Interpretation Robust measurement of mortality in children younger than 5 years shows that accelerating declines are occurring in several low-income countries. These positive developments deserve attention and might need enhanced policy attention and resources. Funding Bill & Melinda Gates Foundation. Introduction There are only 5 years left to achieve Millennium Development Goal 4 (MDG 4), which calls for a two-thirds reduction in mortality in children younger than 5 years between 1990 and 2015. Regular assessment of levels and trends in child mortality is essential for countries to ascertain their progress towards this goal and to take action to meet it. Previous appraisals of mortality in children younger than 5 years suggest that few countries are on track to meet MDG 4.1–3 In each of these studies,1–3 no more than 26% of low-income and middle-income countries examined were deemed to be on track to reach this target. Groups such as the Countdown to 2015 Initiative3 have therefore tried to rally support to accelerate progress in child mortality.4–7 The MDG 4 target has shifted the focus from tracking levels of child mortality to assessing whether countries are reducing child mortality at the 4·4% rate per year needed to achieve the two-thirds reduction in 25 years. Accurate assessments of rates of change need more robust measurement with narrower uncertainty intervals than do assessments of levels. Although there have been substantial investments in the collection of data such as summary and complete birth histories to measure child mortality, assessments of trends have varied substantially from year to year and from source to source. For example, the list of the ten countries with the fastest rates of decline in child mortality between 1990 and 2007, as reported by UNICEF in 2008,8 UNICEF in 2009,9 and the UN
Articles A Norway B日Salvador ☐Vital registration 200 Demographic and Health Surveys Centers for Disease Control 14 and Prevention Reproductive 150 Health Surveys Census 100 50 的型 ☐Vital registration ▣Other 。。 ◆。◆。。 0+ D Liberia □Census 3507 ☐Other 250 300 250 200 150 150 1。 200 100 100 Demographicand Health Surveys Census Household deaths Other Malaria indicator survey 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 Year Year Figure 1:Empirical data sources and estimated under-5 mortality from 1970 to 2010 for selected countries Dashed lines indicate uncertainty intervals.Hollow cirdes represent outliers.Under-5 mortality is defined as the probability of death between birth and age 5years Upward-pointing triangles are direct estimates from complete birth histories.Downward-pointing triangles are indirect estimates from summary birth histories. Population Division (UNPD)in 2009,have only three development assistance for health,'the expansion of countries in common:Portugal,Vietnam,and the insecticide-treated net coverage,activity of the GAVI Maldives.In 2008,UNICEF reported that Thailand had Alliance,and rollout of antiretroviral drugs,"there are the fastest rate of decline in the world,leading researchers many reasons to hope that accelerations might be to undertake a case study of this success."But in 2009. occurring in some countries. UNICEF reported that Thailand had only the 47th fastest In this study,we examined levels,rates of decline,and rate of decline;in a UNPD report,the country had the accelerations and decelerations in rates of decline in fourth fastest rate of decline.Such confusion about the neonatal,postneonatal,childhood,and under-5 mortality true extent of progress can foster policy inaction in from 1970 to 2010 in 187 countries.This study was aided countries,precisely at a time when targeted,effective by four important developments since the previous programmes are needed most.Variation in the studies were done.First,we made use of data that have assessments of rates of decline indicates the availability been newly released or acquired during an intensive and use of different datasets,different analytical methods, 3-year effort to obtain access to microdata (individual and different decisions about data quality by the analysts. level data)and tabulated data sources.Second,we used Evidence from several low-income countries suggests new methods to analyse data from summary birth that in some countries,declines in mortality in children histories with reduced bias and measurement error. younger than 5 years might have accelerated since Third,we applied new data synthesis methods with 2000,-#whereas in others,the rate of decline might be enhanced predictive validity to combine data from slowing.During the 25 years of the MDG 4 target, several sources and capture both sampling and non- countries are likely to experience accelerations and sampling error patterns.This new method requires decelerations in rates of decline.Acceleration matters many fewer subjective inputs to estimation,ensuring because it could be an early indication of policy or that the output is strongly grounded in empirical data programme success.The need to use the best datasets and is as reproducible as possible.Finally,we took and the most valid methods for assessing child mortality advantage of more data and better models with improved over time is only intensified when trying to detect such predictive validity to analyse country patterns of accelerations and decelerations.In view of the scale-up in neonatal,postneonatal,and childhood mortality. www.thelancet.com Vol 375 June 5,2010 1989
Articles www.thelancet.com Vol 375 June 5, 2010 1989 Population Division (UNPD) in 2009,10 have only three countries in common: Portugal, Vietnam, and the Maldives. In 2008, UNICEF reported that Thailand had the fastest rate of decline in the world, leading researchers to undertake a case study of this success.11 But in 2009, UNICEF reported that Thailand had only the 47th fastest rate of decline;9 in a UNPD report, the country had the fourth fastest rate of decline.10 Such confusion about the true extent of progress can foster policy inaction in countries, precisely at a time when targeted, eff ective programmes are needed most. Variation in the assessments of rates of decline indicates the availability and use of diff erent datasets, diff erent analytical methods, and diff erent decisions about data quality by the analysts. Evidence from several low-income countries suggests that in some countries, declines in mortality in children younger than 5 years might have accelerated since 2000,12–14 whereas in others, the rate of decline might be slowing. During the 25 years of the MDG 4 target, countries are likely to experience accelerations and decelerations in rates of decline. Acceleration matters because it could be an early indication of policy or programme success. The need to use the best datasets and the most valid methods for assessing child mortality over time is only intensifi ed when trying to detect such accelerations and decelerations. In view of the scale-up in development assistance for health,15 the expansion of insecticide-treated net coverage,16 activity of the GAVI Alliance,17 and rollout of antiretroviral drugs,18 there are many reasons to hope that accelerations might be occurring in some countries. In this study, we examined levels, rates of decline, and accelerations and decelerations in rates of decline in neonatal, post neonatal, childhood, and under-5 mortality from 1970 to 2010 in 187 countries. This study was aided by four important developments since the previous studies were done. First, we made use of data that have been newly released or acquired during an intensive 3-year eff ort to obtain access to microdata (individuallevel data) and tabulated data sources. Second, we used new methods to analyse data from summary birth histories with reduced bias and measurement error.19 Third, we applied new data synthesis methods with enhanced predictive validity to combine data from several sources and capture both sampling and nonsampling error patterns. This new method requires many fewer subjective inputs to estimation, ensuring that the output is strongly grounded in empirical data and is as reproducible as possible. Finally, we took advantage of more data and better models with improved predictive validity to analyse country patterns of neonatal, postneonatal, and childhood mortality. Figure 1: Empirical data sources and estimated under-5 mortality from 1970 to 2010 for selected countries Dashed lines indicate uncertainty intervals. Hollow circles represent outliers. Under-5 mortality is defi ned as the probability of death between birth and age 5 years. Upward-pointing triangles are direct estimates from complete birth histories. Downward-pointing triangles are indirect estimates from summary birth histories. Under-5 mortality (per 1000) Under-5 mortality (per 1000) Year Vital registration Census Other Norway 16 0 1970 1980 1990 2000 2010 4 6 8 12 14 10 300 0 50 100 150 250 200 A Year Demographic and Health Surveys Centers for Disease Control and Prevention Reproductive Health Surveys Census Vital registration Other El Salvador 0 1970 1980 1990 2000 2010 50 150 100 200 B C Laos D Liberia ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Demographic and Health Surveys Census Household deaths Other Malaria indicator survey 350 0 50 100 150 250 200 300
Articles Methods were available from summary birth histories,we Data sources estimated under-5 mortality by use of the combined By use of improved methods,we substantially updated method developed by Rajaratnam and colleagues.If the database of measurements for under-5 mortality microdata were not available,but we were able to obtain (defined as the probability of death between birth and age tabulated data for children who died and children ever- 5 years)used by Murray and colleagues'in 2007 to include born by mother's age,we applied the maternal age newly released or obtained data,as well as reanalysed cohort-derived method.We analysed 545 surveys with microdata from many of the sources included in the 2007 summary birth history microdata or tabulated data and database.We retained measurements from the original 256 surveys with complete birth history microdata.If database if we were not able to reanalyse the source data. microdata or tabulated data were not available,we The database now contains 7933 more measurements included estimated values of under-5 mortality from than did the 2007 analysis.Data for mortality in children reports,such as preliminary DHS reports. younger than 5 years were derived from a range of We also analysed survey and census data for deaths in sources,including vital registration systems,sample the household.We adjusted estimates on the basis of registration systems,surveys,and censuses.A full list of household deaths from single surveys by use of the See Online for webappendix data types and sources is provided in webappendix growth balance method."When completeness of death pp211-15. reporting was estimated to be more than 100%,we Survey measurements of under-5 mortality in the adjusted the death rates downwards.with the logic that database consist of data from complete and summary respondents might be telescoping deaths-ie,including birth histories.Under-5 mortality from complete birth deaths that occurred outside the recall period in the histories in continuing survey programmes such as the period of recall.Child death registration is usually Demographic and Health Surveys(DHS)were computed lower than is adult death registration,so estimates from pooled data across all such surveys in a given corrected upward(24 in total)must be viewed as lower country (World Fertility Surveys,as the precursor to bound estimates of child mortality.Sensitivity of our DHS,were included with DHS).The pooling approach results to inclusion of these sources is presented in mitigates some of the concerns of bias in the complete webappendix pp 10-11. birth histories,such as selection bias for surveys in Our final database contained 17208 measurements, countries with high prevalence of HIV.%If microdata consisting of 10555 reanalysed measurements from summary birth histories,1455 from complete birth 18 UN Population Division (2009) histories,79 from household deaths,3626 from vital ●UNICEF(2009)9 平Murray et al(2007j registration systems,and 1493 from various other sources.1034 of the total measurements were classified 16 ●● as outliers on the basis of examination of country plots and in some cases because of known biases in the data. We used two criteria for identifying outliers:rates of child mortality that were far beyond the plausible range 14 12 10 N in view of a country's level of development,and rates of child mortality that were substantially inconsistent with other sources of information for the same country that cannot be explained by a known mortality shock. 香分量量量神◆◆◆ Generally,we favoured the inclusion of data points rather than their exclusion.We also excluded countries with populations of less than 50000 from the analysis Overall,we produced yearlyestimates ofunder-5mortality from 16 174 empirical measurements for 187 countries. Estimation of under-5 mortality For each country,we generated a time series of estimates of under-5 mortality by synthesising the empirical data estimates with an analytical technique called Gaussian processregression(GPR).2"Detailsoftheimplementation of this technique are shown in webappendix pp 1-8. Briefly,we applied Loess regression of the log of under-5 mortality in a country as a function of time and 1970 1980 1990 2000 2010 an indicator variable for measurements from vital Year registration data to allow for under-registration of child Figure 2:Worldwide number of deaths in children younger than 5 years from 1970 to 2010 deaths.This predicted series was then updated by the data 1990 www.thelancet.com Vol 375 June 5,2010
Articles 1990 www.thelancet.com Vol 375 June 5, 2010 Methods Data sources By use of improved methods, we substantially updated the database of measurements for under-5 mortality (defi ned as the probability of death between birth and age 5 years) used by Murray and colleagues1 in 2007 to include newly released or obtained data, as well as reanalysed microdata from many of the sources included in the 2007 database. We retained measurements from the original database if we were not able to reanalyse the source data. The database now contains 7933 more measurements than did the 2007 analysis. Data for mortality in children younger than 5 years were derived from a range of sources, including vital registration systems, sample registration systems, surveys, and censuses. A full list of data types and sources is provided in webappendix pp 211–15. Survey measurements of under-5 mortality in the database consist of data from complete and summary birth histories. Under-5 mortality from complete birth histories in continuing survey programmes such as the Demographic and Health Surveys (DHS) were computed from pooled data across all such surveys in a given country (World Fertility Surveys, as the precursor to DHS, were included with DHS). The pooling approach mitigates some of the concerns of bias in the complete birth histories, such as selection bias for surveys in countries with high prevalence of HIV.20 If microdata were available from summary birth histories, we estimated under-5 mortality by use of the combined method developed by Rajaratnam and colleagues.19 If microdata were not available, but we were able to obtain tabulated data for children who died and children everborn by mother’s age, we applied the maternal age cohort-derived method.19 We analysed 545 surveys with summary birth history microdata or tabulated data and 256 surveys with complete birth history microdata. If microdata or tabulated data were not available, we included estimated values of under-5 mortality from reports, such as preliminary DHS reports. We also analysed survey and census data for deaths in the household. We adjusted estimates on the basis of household deaths from single surveys by use of the growth balance method.21 When completeness of death reporting was estimated to be more than 100%, we adjusted the death rates downwards, with the logic that respondents might be telescoping deaths—ie, including deaths that occurred outside the recall period in the period of recall.22,23 Child death registration is usually lower than is adult death registration, so estimates corrected upward (24 in total) must be viewed as lower bound estimates of child mortality. Sensitivity of our results to inclusion of these sources is presented in webappendix pp 10–11. Our fi nal database contained 17 208 measurements, consisting of 10 555 reanalysed measurements from summary birth histories, 1455 from complete birth histories, 79 from household deaths, 3626 from vital registration systems, and 1493 from various other sources. 1034 of the total measurements were classifi ed as outliers on the basis of examination of country plots and in some cases because of known biases in the data. We used two criteria for identifying outliers: rates of child mortality that were far beyond the plausible range in view of a country’s level of development, and rates of child mortality that were substantially inconsistent with other sources of information for the same country that cannot be explained by a known mortality shock. Generally, we favoured the inclusion of data points rather than their exclusion. We also excluded countries with populations of less than 50 000 from the analysis. Overall, we produced yearly estimates of under-5 mortality from 16 174 empirical measurements for 187 countries. Estimation of under-5 mortality For each country, we generated a time series of estimates of under-5 mortality by synthesising the empirical data estimates with an analytical technique called Gaussian process regression (GPR).24–27 Details of the implementation of this technique are shown in webappendix pp 1–8. Briefl y, we applied Loess regression of the log of under-5 mortality in a country as a function of time and an indicator variable for measurements from vital registration data to allow for under-registration of child Figure 2: Worldwide number of deaths in children younger than 5 years from 1970 to 2010 deaths. This predicted series was then updated by the data UN Population Division (2009)10 UNICEF (2009)9 Murray et al (2007)1 Current study 0 1970 1980 1990 Year Estimated number of deaths (millions) 2000 2010 8 6 10 12 14 16 18 See Online for webappendix
Articles within each country by use of GPR.Our GPR model 60 Neonatal mortality improves on previous approaches to synthesising Postneonatal mortality measurements for under-5 mortality by providing the 一Childhood mortality flexibility to track observed trends in the data and a coherent,empirical framework for distinguishing these 50 real trends from fluctuations caused by sampling and non-sampling errors.Uncertainty in the measurements because of non-sampling error was captured in the model by a set of variance parameters,one for each type of data 40 source.These parameters were estimated on the basis of the degree to which a source tends to disagree with the other sources.The more that a particular source conflicted with other sources,the higher the variance parameter will be and thus the more uncertain the measurements.The model takes this uncertainty into account as well as the sampling uncertainty and accordingly smooths data that are noisy(ie,with large variation because of sampling or 20- small numbers)and uncertain.For countries with reliable data,the GPR estimates closely track the observed data. As described in webappendix pp 8-10,we assessed the validity of this strategy to synthesise data sources by 01 1970 1980 1990 2000 2010 undertaking two types of out-of sample predictive validity Year tests.When 20%of surveys and 20%of vital registration country-years were excluded from the analysis,the GPR Figure 3:Worldwide neonatal,postneonatal,and childhood mortality from 1970to2010 predictions for these withheld data had a median relative Seetext for definitions of neonatal,postneonatal,and chilhood mortality. error of 7.3%.When the last 10 years of data for every country with at least 20 years of data were excluded,GPR shocks were not included in the GPR estimation.We predictions had a median relative error of 10.9%. added the shocks back in by replacing the GPR estimate Webappendix p 212 shows the predictive validity of GPR with the mean of the empirical measurements in the compared with four other methods on performance in year of the shock.For countries in which shocks were the two types of out-of-sample predictive validity tests. expected but not present in the data (surveys are often The four methods consist of a Loess-based approach and unable to include regions heavily affected by confict). spline-based approaches with varying numbers of knots. such as the Democratic Republic of the Congo,Ethiopia, GPR outperforms all four of these approaches. and Sudan,we used province-level data from the One benefit of GPR is that it improves on previous Complex Emergency Database (CE-DAT)to generate attempts to estimate uncertainty in under-5 mortality national estimates that were more indicative of mortality that have ignored the many sources of uncertainty beyond shocks."Unfortunately,because of the wide variation in model specification for an ad-hoc choice of model.'The the CE-DAT data.this effort did not substantially alter uncertainty in this application depends on the sample the GPR estimates for these countries;therefore,we did sizes of the measurement instruments.non-sampling not include these data points.Because of the difficulties error for a given data source,and aspects of the Gaussian inherent in obtaining reliable data that cover periods of process.Details are provided in webappendix pp 1-17 mortality shocks,the systematic process we applied in these scenarios is imperfect and remains a limitation to Wars,earthquakes,and other mortality shocks our analysis.The effect of the Haiti 2010 earthquake was Periods that were affected by mortality shocks were estimated by applying the same percentage increase in identified by data from the Uppsala Conflict Data mortality,compared with the previous year,recorded in Program and the Centre for Research on the the Armenia 1988 earthquake. Epidemiology of Disasters.Deaths caused by conflict or natural disaster from each database,in addition to Analysis of trends population estimates from the UN,were used to We computed yearly rates of change in under-5 mortality generate a variable of war/disaster deaths per head.Any and examined rates over time for each country.The year with a value higher than a threshold of one death average rate of decline needed to meet MDG 4 is 4.4% per 10000 population was coded as a mortality shock per year,and across different periods between 1990 and year.We also examined the data from each country to 2010,we looked at how many countries were experiencing identify aberrations in observed mortality that were declines at that rate or greater. consistent with a historical record of conflict or disaster. Concerns have also been raised that mortality from Empirical data points from years classified as mortality HIV might be differentially affecting mothers whose www.thelancet.com Vol 375 June 5,2010 1991
Articles www.thelancet.com Vol 375 June 5, 2010 1991 within each country by use of GPR. Our GPR model improves on previous approaches to synthesising measurements for under-5 mortality by providing the fl exibility to track observed trends in the data and a coherent, empirical framework for distinguishing these real trends from fl uctuations caused by sampling and non-sampling errors. Uncertainty in the measurements because of non-sampling error was captured in the model by a set of variance parameters, one for each type of data source. These parameters were estimated on the basis of the degree to which a source tends to disagree with the other sources. The more that a particular source confl icted with other sources, the higher the variance parameter will be and thus the more uncertain the measurements. The model takes this uncertainty into account as well as the sampling uncertainty and accordingly smooths data that are noisy (ie, with large variation because of sampling or small numbers) and uncertain. For countries with reliable data, the GPR estimates closely track the observed data. As described in webappendix pp 8–10, we assessed the validity of this strategy to synthesise data sources by undertaking two types of out-of-sample predictive validity tests. When 20% of surveys and 20% of vital registration country-years were excluded from the analysis, the GPR predictions for these withheld data had a median relative error of 7·3%. When the last 10 years of data for every country with at least 20 years of data were excluded, GPR predictions had a median relative error of 10·9%. Webappendix p 212 shows the predictive validity of GPR compared with four other methods on performance in the two types of out-of-sample predictive validity tests. The four methods consist of a Loess-based approach and spline-based approaches with varying numbers of knots. GPR outperforms all four of these approaches. One benefi t of GPR is that it improves on previous attempts to estimate uncertainty in under-5 mortality that have ignored the many sources of uncertainty beyond model specifi cation for an ad-hoc choice of model.1 The uncertainty in this application depends on the sample sizes of the measurement instruments, non-sampling error for a given data source, and aspects of the Gaussian process. Details are provided in webappendix pp 1–17. Wars, earthquakes, and other mortality shocks Periods that were aff ected by mortality shocks were identifi ed by data from the Uppsala Confl ict Data Program and the Centre for Research on the Epidemiology of Disasters.28–31 Deaths caused by confl ict or natural disaster from each database, in addition to population estimates from the UN,10 were used to generate a variable of war/disaster deaths per head. Any year with a value higher than a threshold of one death per 10 000 population was coded as a mortality shock year. We also examined the data from each country to identify aberrations in observed mortality that were consistent with a historical record of confl ict or disaster. Empirical data points from years classifi ed as mortality shocks were not included in the GPR estimation. We added the shocks back in by replacing the GPR estimate with the mean of the empirical measurements in the year of the shock. For countries in which shocks were expected but not present in the data (surveys are often unable to include regions heavily aff ected by confl ict), such as the Democratic Republic of the Congo, Ethiopia, and Sudan, we used province-level data from the Complex Emergency Database (CE-DAT) to generate national estimates that were more indicative of mortality shocks.32 Unfortunately, because of the wide variation in the CE-DAT data, this eff ort did not substantially alter the GPR estimates for these countries; therefore, we did not include these data points. Because of the diffi culties inherent in obtaining reliable data that cover periods of mortality shocks, the systematic process we applied in these scenarios is imperfect and remains a limitation to our analysis. The eff ect of the Haiti 2010 earthquake was estimated by applying the same percentage increase in mortality, compared with the previous year, recorded in the Armenia 1988 earthquake. Analysis of trends We computed yearly rates of change in under-5 mortality and examined rates over time for each country. The average rate of decline needed to meet MDG 4 is 4·4% per year, and across diff erent periods between 1990 and 2010, we looked at how many countries were experiencing declines at that rate or greater. Concerns have also been raised that mortality from HIV might be diff erentially aff ecting mothers whose Figure 3: Worldwide neonatal, postneonatal, and childhood mortality from 1970 to 2010 See text for defi nitions of neonatal, postneonatal, and childhood mortality. 1970 Probability of death (per 1000) 0 20 30 40 50 60 Neonatal mortality Postneonatal mortality Childhood mortality 1980 1990 Year 2000 2010
Articles 1970 1980 1990 2000 2010 Asia Pacific,high income Brunei 53-4(48-2-600) 214(18.5-247) 11-4(10-0-131) 92(8-0-10-9) 7-5(56-10-3引 Japan 17-8(17-7-18-0) 11-6(114-117) 6-6(6-5-6-8) 44(43-45) 33(2-8-39) Singapore 29-2(27-9-305) 153(14-4-16.3) 7-7(7-2-83) 4-1(3-7-44) 2-5(1-933别 South Korea 56-5(521-610) 18-7(14-2-24-2) 113(8-4-14-7 93(6-8-12.2) 5-13-4-7-6) Asia,central Armenia 103-1(88-6-122-8) 68.9(65-7-72-1) 509(48-7-532) 32-260-7-33-8) 193(157-23-8) Azerbaijan 149-8(109-1-206-8) 109-8(99-7-121-9) 78-6(733-83-2) 56-252-5-596) 33-8(26-6-427) Georgia 94-0(77-9-1130) 58.9(55-7-623) 44-8(42-6-47-0) 35-2(33-2-37-2) 239(19-0-293) Kazakhstan 69-8(609-785) 653(61-8-68-9) 53-3(50-3-56-0) 44-1(40-8-475) 313(25-4-40-0) Kyrgyzstan 139-3126-5-152-6) 982(941-1028) 69-5(66-8-72-6 50-9(46-7-54-5) 425(36-0-52-4) Mongolia 157-4(147-0-169-4) 126-3(1212-131-5) 94-5(909-98-5) 58-5(55-7-614) 33-8(29-4-387) Tajikistan 177-5(142-2-210-6) 1435(135-2-152-6) 108.4(102-2-114-3) 81-4(76-8-86-1) 49-7(39-7-604) Turkmenistan 184-7(137-2-239-7) 124-8(111-7-139-9) 98-4(89-4-108.5) 74-7(644-86-3) 26.0(189-34-8 Uzbekistan 81-6(70592-2) 774(74-1-81-4) 659(62-6-68.5) 573(54-8-60-4) 435(36.7-51-6) Asia,east China 85-7(733-99-8) 48.5(430-54-3) 39-6(35-8-43-3) 320(28-5-36-5) 154(118-20-1) North Korea 1144(62.1-210-7) 70.5(38-3-129-7) 509(27-7-93-8) 45-9(24-9845) 33-4(18-1-615) Taiwan 27-4(26-9-27-9) 15-3(15-0-15-6) 8-9(87-9-2) 8-6(8-3-8-9) 62(52-740 Asia,south Afghanistan 286-9(260-0-314-9) 212-0(196.9-230-1) 163-5(1542-1735) 141-0(131-6-150-4) 1213(1090-1347刀 Bangladesh 234-4(227-4-242-7刀 191-0(186-5-195-2) 139-5(136-7-142-5) 87-9(85-5-90-1) 55-9(51-6-60-1) Bhutan 278-1(246-8-310-7) 207-7(190-3-2294) 144-8(134-9-1567) 881(81-4-957) 541(47-4-61-8) India 197-8(191-9-203-4) 150-9(147-8-154-0) 114-3(112-0-116-7) 84-6(82-5-86-8) 62-6(58-2-67-3) Nepal 261-7(2523-271-7) 210-2(2047-215-8) 137-3(133-6-141-0) 79753-809) 47-3(42-6-523) Pakistan 169-2(162-6-176-5) 134-9(129-0-141-0) 113-3(109-0-117-7 946(89-9-98-8) 80.3705-923) Asia,southeast Cambodia 169-2(140-7-203-6) 1734(1648-183-6) 121-2(115-5-127-3) 105-4(99-8-1117) 59-7(42-4-84.8) Burma 162-6(139-0-19040 125-6(106-5-144-1) 1202(103-5-142-9) 79-6(68-3-924) 55-039-7-76-0) Indonesia 156-8(152-5-161-0) 104-1(101-7-106-8) 71.5(69-8-733) 48-8(47-4-504) 366(324-411) Laos 218-8(158.4-296-1) 184-7(1654-207-9) 144-5(126-9-164-7) 98-376-3-127-0) 68.3(47-1-97-9) Malaysia 496(47-0-52-3) 28-0(25-6-30-6) 16-4(151-17-9) 9-6(8-7-10-6) 51(4-1-63) Maldives 247-1(226-3-272-5) 1727(160-7-187-0) 88-5(80-8-967) 34-0(29-7-39-0) 14-0(111-17-9 Mauritius 84-0(81-7-864) 438(42-2-453) 24-4(23-3-255) 19-0(18-1-20-1) 13-0(10-7-157刃 Philippines 87-5(83-8-909) 78.876-3-813) 541(52-4-55-9) 38-3(36-5-40-7) 28-6(23-6-33-9) Seychelles 79-9(70-1-91-7) 29-8(26-1-34-6) 18-9(163-217) 17-3(14-3-21-2) 12-5(8-7-18-6) Sri Lanka 66-8(63-0-70-7) 437(412-465) 35-4(32-1-392) 17-6(15-8-194) 10-17-2-14-0) Thailand 87-9(845-91-3) 492(47-3-51-5) 233(22-1-247刀 13-8(131-14-7) 8.9(7-6-10-5) Timor-Leste 202-1(177-4-2297刀 153-2(142-0-167-5) 101-4(92-7-110-3) 826(74-0-925) 632(46-7-83-7) Vietnam 87-377-0-992) 67.9(644-71-6) 46-3(43-8-48.8) 21-8(19-9-24-0) 12.9(9-1-18-6) Australasia Australia 23-1(22-6-23-6) 14-6(14-1-15-0) 9-9(96-10-3) 6-4(6-2-6-7) 4-73-956) New Zealand 21-5(20-7-22-3) 16-5(15-8-17-3) 11-1(10-6-117) 7-975-83) 58(5-0-6.9) (Continues on next page) children have high mortality,resulting in selection HIV-negative women in the 10 years before the survey bias."This bias is expected to have a larger effect farther in 21 DHS with HIV testing that can be linked to the back in time before the survey,-which would tend to complete birth histories.The difference in the under-5 reduce the estimated trend in child mortality.Our use, mortality rate ranges from 135 (uncertainty interval in nearly all countries,of overlapping surveys to 72-200)per 1000 higher in HIV-positive women than in estimate the levels and trends in child mortality should HIV-negative women to 73 (4-141)per 1000 lower in substantially attenuate this effect.As described in more HIV-positive women than in HIV-negative women.This detail in the webappendix,we have further examined variation is related to the correlated socioeconomic the under-5 mortality rate in HIV-positive women and status and location of HIV-positive women relative to 1992 www.thelancet.com Vol 375 June 5,2010
Articles 1992 www.thelancet.com Vol 375 June 5, 2010 children have high mortality, resulting in selection bias.14 This bias is expected to have a larger eff ect farther back in time before the survey,33–35 which would tend to reduce the estimated trend in child mortality. Our use, in nearly all countries, of overlapping surveys to estimate the levels and trends in child mortality should substantially attenuate this eff ect. As described in more detail in the webappendix, we have further examined the under-5 mortality rate in HIV-positive women and HIV-negative women in the 10 years before the survey in 21 DHS with HIV testing that can be linked to the complete birth histories. The diff erence in the under-5 mortality rate ranges from 135 (uncertainty interval 72–200) per 1000 higher in HIV-positive women than in HIV-negative women to 73 (4–141) per 1000 lower in HIV-positive women than in HIV-negative women. This variation is related to the correlated socioeconomic status and location of HIV-positive women relative to 1970 1980 1990 2000 2010 Asia Pacifi c, high income Brunei 53·4 (48·2–60·0) 21·4 (18·5–24·7) 11·4 (10·0–13·1) 9·2 (8·0–10·9) 7·5 (5·6–10·3) Japan 17·8 (17·7–18·0) 11·6 (11·4–11·7) 6·6 (6·5–6·8) 4·4 (4·3–4·5) 3·3 (2·8–3·9) Singapore 29·2 (27·9–30·5) 15·3 (14·4–16·3) 7·7 (7·2–8·3) 4·1 (3·7–4·4) 2·5 (1·9–3·3) South Korea 56·5 (52·1–61·0) 18·7 (14·2–24·2) 11·3 (8·4–14·7) 9·3 (6·8–12·2) 5·1 (3·4–7·6) Asia, central Armenia 103·1 (88·6–122·8) 68·9 (65·7–72·1) 50·9 (48·7–53·2) 32·2 (30·7–33·8) 19·3 (15·7–23·8) Azerbaijan 149·8 (109·1–206·8) 109·8 (99·7–121·9) 78·6 (73·3–83·2) 56·2 (52·5–59·6) 33·8 (26·6–42·7) Georgia 94·0 (77·9–113·0) 58·9 (55·7–62·3) 44·8 (42·6–47·0) 35·2 (33·2–37·2) 23·9 (19·0–29·3) Kazakhstan 69·8 (60·9–78·5) 65·3 (61·8–68·9) 53·3 (50·3–56·0) 44·1 (40·8–47·5) 31·3 (25·4–40·0) Kyrgyzstan 139·3 (126·5–152·6) 98·2 (94·1–102·8) 69·5 (66·8–72·6) 50·9 (46·7–54·5) 42·5 (36·0–52·4) Mongolia 157·4 (147·0–169·4) 126·3 (121·2–131·5) 94·5 (90·9–98·5) 58·5 (55·7–61·4) 33·8 (29·4–38·7) Tajikistan 177·5 (142·2–210·6) 143·5 (135·2–152·6) 108·4 (102·2–114·3) 81·4 (76·8–86·1) 49·7 (39·7–60·4) Turkmenistan 184·7 (137·2–239·7) 124·8 (111·7–139·9) 98·4 (89·4–108·5) 74·7 (64·4–86·3) 26·0 (18·9–34·8) Uzbekistan 81·6 (70·5–92·2) 77·4 (74·1–81·4) 65·9 (62·6–68·5) 57·3 (54·8–60·4) 43·5 (36·7–51·6) Asia, east China 85·7 (73·3–99·8) 48·5 (43·0–54·3) 39·6 (35·8–43·3) 32·0 (28·5–36·5) 15·4 (11·8–20·1) North Korea 114·4 (62·1–210·7) 70·5 (38·3–129·7) 50·9 (27·7–93·8) 45·9 (24·9–84·5) 33·4 (18·1–61·5) Taiwan 27·4 (26·9–27·9) 15·3 (15·0–15·6) 8·9 (8·7–9·2) 8·6 (8·3–8·9) 6·2 (5·2–7·4) Asia, south Afghanistan 286·9 (260·0–314·9) 212·0 (196·9–230·1) 163·5 (154·2–173·5) 141·0 (131·6–150·4) 121·3 (109·0–134·7) Bangladesh 234·4 (227·4–242·7) 191·0 (186·5–195·2) 139·5 (136·7–142·5) 87·9 (85·5–90·1) 55·9 (51·6–60·1) Bhutan 278·1 (246·8–310·7) 207·7 (190·3–229·4) 144·8 (134·9–156·7) 88·1 (81·4–95·7) 54·1 (47·4–61·8) India 197·8 (191·9–203·4) 150·9 (147·8–154·0) 114·3 (112·0–116·7) 84·6 (82·5–86·8) 62·6 (58·2–67·3) Nepal 261·7 (252·3–271·7) 210·2 (204·7–215·8) 137·3 (133·6–141·0) 77·9 (75·3–80·9) 47·3 (42·6–52·3) Pakistan 169·2 (162·6–176·5) 134·9 (129·0–141·0) 113·3 (109·0–117·7) 94·6 (89·9–98·8) 80·3 (70·5–92·3) Asia, southeast Cambodia 169·2 (140·7–203·6) 173·4 (164·8–183·6) 121·2 (115·5–127·3) 105·4 (99·8–111·7) 59·7 (42·4–84·8) Burma 162·6 (139·0–190·4) 125·6 (106·5–144·1) 120·2 (103·5–142·9) 79·6 (68·3–92·4) 55·0 (39·7–76·0) Indonesia 156·8 (152·5–161·0) 104·1 (101·7–106·8) 71·5 (69·8–73·3) 48·8 (47·4–50·4) 36·6 (32·4–41·1) Laos 218·8 (158·4–296·1) 184·7 (165·4–207·9) 144·5 (126·9–164·7) 98·3 (76·3–127·0) 68·3 (47·1–97·9) Malaysia 49·6 (47·0–52·3) 28·0 (25·6–30·6) 16·4 (15·1–17·9) 9·6 (8·7–10·6) 5·1 (4·1–6·3) Maldives 247·1 (226·3–272·5) 172·7 (160·7–187·0) 88·5 (80·8–96·7) 34·0 (29·7–39·0) 14·0 (11·1–17·9) Mauritius 84·0 (81·7–86·4) 43·8 (42·2–45·3) 24·4 (23·3–25·5) 19·0 (18·1–20·1) 13·0 (10·7–15·7) Philippines 87·5 (83·8–90·9) 78·8 (76·3–81·3) 54·1 (52·4–55·9) 38·3 (36·5–40·7) 28·6 (23·6–33·9) Seychelles 79·9 (70·1–91·7) 29·8 (26·1–34·6) 18·9 (16·3–21·7) 17·3 (14·3–21·2) 12·5 (8·7–18·6) Sri Lanka 66·8 (63·0–70·7) 43·7 (41·2–46·5) 35·4 (32·1–39·2) 17·6 (15·8–19·4) 10·1 (7·2–14·0) Thailand 87·9 (84·5–91·3) 49·2 (47·3–51·5) 23·3 (22·1–24·7) 13·8 (13·1–14·7) 8·9 (7·6–10·5) Timor-Leste 202·1 (177·4–229·7) 153·2 (142·0–167·5) 101·4 (92·7–110·3) 82·6 (74·0–92·5) 63·2 (46·7–83·7) Vietnam 87·3 (77·0–99·2) 67·9 (64·4–71·6) 46·3 (43·8–48·8) 21·8 (19·9–24·0) 12·9 (9·1–18·6) Australasia Australia 23·1 (22·6–23·6) 14·6 (14·1–15·0) 9·9 (9·6–10·3) 6·4 (6·2–6·7) 4·7 (3·9–5·6) New Zealand 21·5 (20·7–22·3) 16·5 (15·8–17·3) 11·1 (10·6–11·7) 7·9 (7·5–8·3) 5·8 (5·0–6·9) (Continues on next page)
Articles 1970 1980 1990 2000 2010 (Continued from previous page) Caribbean Antigua and Barbuda 441(38-950-0) 302(255-35-9) 181(15-4-21-1) 23-2(202-27-0) 13-8(108-17-9) Barbados 715(65-9-77-7) 404(36.9-44-7) 264(23-6-294) 21-7(18-9-25-1) 106(8-1-13-7刀 Belize 127-6(109-6-1450) 78-0(71-0-85-9) 43-6(40-0-46-9) 31-0(287-34-1) 22-7(192-27-8) Cuba 38935-8-43-0) 23-3(21-6-25-3) 139(12-8-151) 8-6(7-9-94) 5-2(4-5-61) Dominica 631(58-6-68-0) 22-9(203-257) 24-1(213-26-6) 21-0(17.9-243) 16-5(12-5-215) Dominican Republic 1187(115-5-121-9) 84-4(824-86-5) 57-0(55-7-58.5) 38-3(37-1-394) 27-5(24-6-30-7) Grenada 80-0715-89-6) 45-0(391-52-0) 267(23-8-30-0) 21-6(183-24-9) 11-6(91-14-7 Guyana 72-0(67-4-76-6) 69-2(63-8-75-2) 61-0(57-2-64-6 46-8(43-9-49-9) 38-0(32.9-44-7刀 Haiti 229-6(2200-2409) 198-6(192-8-204-8) 149-2(1451-154-0) 998(95-9-103-3) 102-6(900-119-6) Jamaica 57-0(53-2-60-9) 45-0(42-4-48-1) 339(32-0-359) 254(23-5-27-2) 18-4(154-22-0) Saint Lucia 77-0713-84-0) 39-1(364-41-7刀 232(212-25-1) 164(149-184) 11-4(8-8-14-6) Saint Vincent and the 80-174-6-85-9) 601(55-8-65-1) 24-8(22.2-27-9) 24-0(21.6-26-8) 23-2(18-9-28-8) Grenadines Suriname 718(62-3-831) 60-5(56-565-0) 464(43-1-49-6) 43739-6-48-4) 35-5(28-5-43-7刀 The Bahamas 68-8(615-76-3) 49-9(457-544) 36-032-8-396) 183(16-3-21-0) 15-8(12-8-19-5) Trinidad and Tobago 51.6(48.8-54-4) 37-2(35-2-393) 30-2(281-32-4) 319(295-346) 25-4(204-31-4) Europe,central Albania 653(50-1-84-7) 42-5(38-3-47-3) 40-1(37-6-42-7刀 22-3(20-5-24-1) 15-1(12-1-19-0) Bosnia and Herzegovina 58.6(402-84-7刀 343(26-2-44-3) 17-716-9-18-7) 104(97-111) 7-9(62-98) Bulgaria 32.2(294-355) 239(21.9-260) 18.3(16-6-20-0) 17-4(15-9-19-0) 10-7(8-7-133) Croatia 40-8(29-7-56-2) 24-5(20-6-30-1) 13-4(12-8-14-0) 85(8-1-9-0) 5-4(47-6-2) Czech Republic 293(26-3-32-5) 21-1(195-22-9) 142(13-0-15-5) 74(67-83) 41(3-5-5-0) Hungary 38-0(34-3-41-5) 25-1(221-27-70 16-7(15-0-18-1) 10-2(93-113) 5-5(44-6-7八 Macedonia 59-6(413-86-4) 407(315-52.1) 26-2(23-6-28-9) 18-5(16-8-204) 11-8(93-15-0) Montenegro 82.6(56-2-119-0) 45-6(318-64-8) 25-1(18-5-34-1) 13-9(12-5-156) 947-7-11-8) Poland 365(36-0-369) 24-9(245-25-3 19-0(18-6-19-3) 96(93-99) 645-1-82) Romania 50-4(45-7-55-6) 36-3(32-5-408) 304(27-2-33-9) 22-2(20-0-24-8) 15-2(12-4-18-8) Serbia 37-7(26-0-546) 21-7(15-7-30-8) 127(9-6-16-6) 7-8(758-1) 4-0(3-5-4-6) Slovakia 31-9(30-8-33-0) 23-4(22-6-24-3) 14-2(135-14-9) 102(95-109) 6-6(5-1-8.6) Slovenia 42-6(31-5-579) 213(18-0-25-4) 10-6(99-113) 54(4959) 3-2(2-6-3-8) Europe,eastern Belarus 32-0(29-4-349) 27-5(25-5-29-7) 21-5(19-9-23-1) 169(15-7-18-1) 10-4(8-8-12-4) Estonia 23-6(21-8-25-8 21-8(20-0-23-8) 18-0(16-4-19-8) 11-0(99-12.1) 63(53-7-4) Latvia 232(213-25-3) 217(198-237) 173(157-19-0) 14-8(13-5-16-3) 9-57-7-113) Lithuania 24-9(23-8-26-0) 20-319-4-21.2) 14-4(13-7-15-2) 10-9(10-2-116) 6-8(57-7-9) Moldova 65-0(547-71) 56-2(51-6-615) 35133-0-37-6) 23-0(21-6-24-6) 13-7(11-7-16-1) Russia 354(32-0-391) 32-8(297-36-4) 26-7(237-297) 21.6(19-3-242) 145(12-0-17-3引 Ukraine 30-5(28-0-333) 29-2(27-2-31-3) 21-6(20-3-230) 19-7(18.6-20-9) 15-5(13-0-18.5) (Continues on next page) HIV-negative women.We have simulated the effect of Neonatal,postneonatal,and childhood death rates this range on our estimates in view of the types of We divided the estimates of under-5 mortality generated analytical methods we apply,and find that biases even by GPR into estimates of neonatal (the probability of in the presence of HIV seroprevalence of 20%range death before age 1 month),postneonatal(the probability from underestimation of ten per 1000 to overestimation of death before age 1 year conditional on surviving to age of six per 1000.Trends in the past 10-15 years are largely 1 month),and childhood (the probability of death from unaffected.Because of the enormous variation in the age 1 year to age 5 years)risks of death by use of a potential bias and,the further confounding of this two-step modelling process in which we first predicted association by scale-up of prevention of mother-to-child sex-specific under-5 mortality and then predicted the sex- transmission and antiretroviral drugs,we have opted to specific neonatal,postneonatal,and childhood risks of not apply any standard correction to the estimated rates death.We modelled the age breakdown by use of separate of child mortality. models for boys and girls because different levels of www.thelancet.com Vol 375 June 5,2010 1993
Articles www.thelancet.com Vol 375 June 5, 2010 1993 HIV-negative women. We have simulated the eff ect of this range on our estimates in view of the types of analytical methods we apply, and fi nd that biases even in the presence of HIV seroprevalence of 20% range from underestimation of ten per 1000 to overestimation of six per 1000. Trends in the past 10–15 years are largely unaff ected. Because of the enormous variation in the potential bias and, the further confounding of this association by scale-up of prevention of mother-to-child transmission and antiretroviral drugs, we have opted to not apply any standard correction to the estimated rates of child mortality. Neonatal, postneonatal, and childhood death rates We divided the estimates of under-5 mortality generated by GPR into estimates of neonatal (the probability of death before age 1 month), postneonatal (the probability of death before age 1 year conditional on surviving to age 1 month), and childhood (the probability of death from age 1 year to age 5 years) risks of death by use of a two-step modelling process in which we fi rst predicted sex-specifi c under-5 mortality and then predicted the sexspecifi c neonatal, postneonatal, and childhood risks of death. We modelled the age breakdown by use of separate models for boys and girls because diff erent levels of 1970 1980 1990 2000 2010 (Continued from previous page) Caribbean Antigua and Barbuda 44·1 (38·9–50·0) 30·2 (25·5–35·9) 18·1 (15·4–21·1) 23·2 (20·2–27·0) 13·8 (10·8–17·9) Barbados 71·5 (65·9–77·7) 40·4 (36·9–44·7) 26·4 (23·6–29·4) 21·7 (18·9–25·1) 10·6 (8·1–13·7) Belize 127·6 (109·6–145·0) 78·0 (71·0–85·9) 43·6 (40·0–46·9) 31·0 (28·7–34·1) 22·7 (19·2–27·8) Cuba 38·9 (35·8–43·0) 23·3 (21·6–25·3) 13·9 (12·8–15·1) 8·6 (7·9–9·4) 5·2 (4·5–6·1) Dominica 63·1 (58·6–68·0) 22·9 (20·3–25·7) 24·1 (21·3–26·6) 21·0 (17·9–24·3) 16·5 (12·5–21·5) Dominican Republic 118·7 (115·5–121·9) 84·4 (82·4–86·5) 57·0 (55·7–58·5) 38·3 (37·1–39·4) 27·5 (24·6–30·7) Grenada 80·0 (71·5–89·6) 45·0 (39·1–52·0) 26·7 (23·8–30·0) 21·6 (18·3–24·9) 11·6 (9·1–14·7) Guyana 72·0 (67·4–76·6) 69·2 (63·8–75·2) 61·0 (57·2–64·6) 46·8 (43·9–49·9) 38·0 (32·9–44·7) Haiti 229·6 (220·0–240·9) 198·6 (192·8–204·8) 149·2 (145·1–154·0) 99·8 (95·9–103·3) 102·6 (90·0–119·6) Jamaica 57·0 (53·2–60·9) 45·0 (42·4–48·1) 33·9 (32·0–35·9) 25·4 (23·5–27·2) 18·4 (15·4–22·0) Saint Lucia 77·0 (71·3–84·0) 39·1 (36·4–41·7) 23·2 (21·2–25·1) 16·4 (14·9–18·4) 11·4 (8·8–14·6) Saint Vincent and the Grenadines 80·1 (74·6–85·9) 60·1 (55·8–65·1) 24·8 (22·2–27·9) 24·0 (21·6–26·8) 23·2 (18·9–28·8) Suriname 71·8 (62·3–83·1) 60·5 (56·5–65·0) 46·4 (43·1–49·6) 43·7 (39·6–48·4) 35·5 (28·5–43·7) The Bahamas 68·8 (61·5–76·3) 49·9 (45·7–54·4) 36·0 (32·8–39·6) 18·3 (16·3–21·0) 15·8 (12·8–19·5) Trinidad and Tobago 51·6 (48·8–54·4) 37·2 (35·2–39·3) 30·2 (28·1–32·4) 31·9 (29·5–34·6) 25·4 (20·4–31·4) Europe, central Albania 65·3 (50·1–84·7) 42·5 (38·3–47·3) 40·1 (37·6–42·7) 22·3 (20·5–24·1) 15·1 (12·1–19·0) Bosnia and Herzegovina 58·6 (40·2–84·7) 34·3 (26·2–44·3) 17·7 (16·9–18·7) 10·4 (9·7–11·1) 7·9 (6·2–9·8) Bulgaria 32·2 (29·4–35·5) 23·9 (21·9–26·0) 18·3 (16·6–20·0) 17·4 (15·9–19·0) 10·7 (8·7–13·3) Croatia 40·8 (29·7–56·2) 24·5 (20·6–30·1) 13·4 (12·8–14·0) 8·5 (8·1–9·0) 5·4 (4·7–6·2) Czech Republic 29·3 (26·3–32·5) 21·1 (19·5–22·9) 14·2 (13·0–15·5) 7·4 (6·7–8·3) 4·1 (3·5–5·0) Hungary 38·0 (34·3–41·5) 25·1 (22·1–27·7) 16·7 (15·0–18·1) 10·2 (9·3–11·3) 5·5 (4·4–6·7) Macedonia 59·6 (41·3–86·4) 40·7 (31·5–52·1) 26·2 (23·6–28·9) 18·5 (16·8–20·4) 11·8 (9·3–15·0) Montenegro 82·6 (56·2–119·0) 45·6 (31·8–64·8) 25·1 (18·5–34·1) 13·9 (12·5–15·6) 9·4 (7·7–11·8) Poland 36·5 (36·0–36·9) 24·9 (24·5–25·3) 19·0 (18·6–19·3) 9·6 (9·3–9·9) 6·4 (5·1–8·2) Romania 50·4 (45·7–55·6) 36·3 (32·5–40·8) 30·4 (27·2–33·9) 22·2 (20·0–24·8) 15·2 (12·4–18·8) Serbia 37·7 (26·0–54·6) 21·7 (15·7–30·8) 12·7 (9·6–16·6) 7·8 (7·5–8·1) 4·0 (3·5–4·6) Slovakia 31·9 (30·8–33·0) 23·4 (22·6–24·3) 14·2 (13·5–14·9) 10·2 (9·5–10·9) 6·6 (5·1–8·6) Slovenia 42·6 (31·5–57·9) 21·3 (18·0–25·4) 10·6 (9·9–11·3) 5·4 (4·9–5·9) 3·2 (2·6–3·8) Europe, eastern Belarus 32·0 (29·4–34·9) 27·5 (25·5–29·7) 21·5 (19·9–23·1) 16·9 (15·7–18·1) 10·4 (8·8–12·4) Estonia 23·6 (21·8–25·8) 21·8 (20·0–23·8) 18·0 (16·4–19·8) 11·0 (9·9–12·1) 6·3 (5·3–7·4) Latvia 23·2 (21·3–25·3) 21·7 (19·8–23·7) 17·3 (15·7–19·0) 14·8 (13·5–16·3) 9·5 (7·7–11·3) Lithuania 24·9 (23·8–26·0) 20·3 (19·4–21·2) 14·4 (13·7–15·2) 10·9 (10·2–11·6) 6·8 (5·7–7·9) Moldova 65·0 (54·7–77·1) 56·2 (51·6–61·5) 35·1 (33·0–37·6) 23·0 (21·6–24·6) 13·7 (11·7–16·1) Russia 35·4 (32·0–39·1) 32·8 (29·7–36·4) 26·7 (23·7–29·7) 21·6 (19·3–24·2) 14·5 (12·0–17·3) Ukraine 30·5 (28·0–33·3) 29·2 (27·2–31·3) 21·6 (20·3–23·0) 19·7 (18·6–20·9) 15·5 (13·0–18·5) (Continues on next page)
Articles 1970 1980 1990 2000 2010 (Continued from previous page) Europe,western Andorra 16-311-5-231) 11-7(8.3-166) 9-4(6-6-133) 71(5-0-101) 493-4-6.9) Austria 29-4(28.5-30-2) 17-2(16-5-17-8) 10-8(103-11-3) 6-1(5-7-6-5) 3-93-3-4-6) Belgium 25-5(24-8-26-2) 153(147-15-8 102(98-10-7) 6-2(59-6.5) 43(36-53) Cyprus 26-6(22-5-31-1) 20-0(17.9-22-1) 12-8(11-7-14-0) 6-6(59-7-3) 28(2-3-34) Denmark 16-4158-171) 10-1(9-6-10-7) 8-7(8-2-92) 5-8(556-2) 41(3-4-49) Finland 15-9(151-16-8) 8-07-5-85) 7-1(6-6-7.5) 45(4-1-48) 3-02.5-37) France 192(19-0-195) 14-7(144-149) 95(93-97) 54(53-5-6) 393-2-48) Germany 25-8(25-5-26-1) 14-0(13-8-14-3) 8-9(87-9-1) 525-0-53) 4-13-5-4-8) Greece 28-7(27-8-29-40 19-0(183-19-7) 11-2(10-6-11-6) 6-9(6-6-7-3) 373-0-45 Iceland 16-4(14-9-18-0) 9-7(8-6-10-9) 6-7(5-9-77) 41(3-5-4-8) 2-6(2-1-33引 Ireland 23-422-5-24-3) 14-9(143-15-6) 99(92-10-5) 74(7-0-7-8) 42(35-51) Israel 27-2(26-3-28.2) 19-0(183-19-7) 12-1(11-6-12-6) 6-9(6-6-7-3) 4739-57) Italy 34-334-0-34-7 17-0(167-17-3) 10-0(98-10.2) 6-4(6-2-6-6) 33(2-8-4-0) Luxembourg 233(215-25-7) 13-9(12-5-154) 9-1(82-10-2) 51(46-5-8) 2-9(23-37刃 Malta 31-5(28-8-34-5) 16.9(15-2-187) 11-3(10-0-12-8) 7-2(6-1-84) 5-2(40-6.9) Netherlands 16-4(16-0-16.9) 114(11-0-11-9) 8-9(8-6-9-2) 6-5(6-3-6-8) 43(37-49) Norway 15-8(151-16-4) 10-5(9-9-11-0) 8-8(8-4-93) 5-0(47-5-3) 34(2-8-40) Portugal 744732-75-6) 294(28-6-30-2) 15-0(14-4-15-6 8-07-6-85) 33(2-6-4-3引 Spain 32-8(32-4-332) 143(141-14-6) 9-2(90-9-5) 57(55-5-9) 383-1-46) Sweden 12-7(12-2-133) 89(85-93) 7-2(6.9-7-6 49(46-5-2) 27(2-2-33) Switzerland 18-7(18-1-194) 10-6(10-1-11-1) 8-8(84-9-2) 6-0(5-7-63引 42(3-5-5-0) UK 21-8(21-5-221) 152(15-0-15-5) 9-7(95-99) 6-8(6-6-6-9) 53(4-5-6-2) Latin America,Andean Bolivia 220-7(212-7-2299) 153-0(1481-158-0) 103-9(100-3-107-5) 62-258-6-65-6) 46-7399-57-0) Ecuador 129-1(125-8-132-7 82-5(80-4-84-6 501(487-51-8) 34-0321-35-7 21.0(18.0-24-3) Peru 159-0(154-8-1632) 114-1(111-6-117-0) 72-7(70-9-74-6) 40-7(392-423) 24-6(21-6-28-9) Latin America,central Colombia 86-8(84-0-898) 52.8(51-2-544) 333(32-0-345) 24-7(23-6-258) 153131-17-7) Costa Rica 73-6(70-5-7-2) 33-131-5-34-8) 214(20-2-22-6) 14-8(13-7-16-0) 8.77-3-10-2) ElSalvador 1604(152-6-168-1) 110-7(106.5-116-3) 59-4(566-62-1) 34-232-3-36-3) 192(16-7-22-3) Guatemala 175-1(1698-1811) 120-7(117-8-124-1) 75-8(73-6-77-70 49-9(47-7-52-5) 31-9(27-6-36-9) Honduras 149-5(142-3-158.5) 92-9(90-3-95-8) 56-3(54-2-58-1) 36-6(353-382) 22-6(197-254) Mexico 107-5(104-3-110-6) 70-0(67-9-72-2) 41-9(40-2-438) 25-8(244-27-4) 16-6(14-7-18-4) Nicaragua 166-2(160-2-172-7) 1023(99-2-105-2) 63-9(62-0-66-0) 39337-8-40-8) 26.6(23-1-31-0) Panama 54-7(52-5-57-2) 36-8(35-2-38-7) 27-8(26-3-293) 233(21-7-25-10 18-0(153-20-9) Venezuela 58-8(55-6-62-2) 38-636-4-40-9) 294(27-5-31-1) 232215-25-0) 161(132-194) Latin America,southern Argentina 72-6(72-0-733) 37-7(37-3-38-2) 28-0(27-6-28.3) 19-7(194-200) 12.9105-15-8) Chile 92-1(854-98-8) 37-3(34-4-40-3) 18-2(16-7-19-8) 11-1(10-3-12-1) 6.5(54-7-9) Uruguay 56-0(54-5-577) 40-7(39-2-421) 22-2(212-233) 16-5(157-17-3) 11.5(9-3-14-4) Latin America,tropical Brazil 120-8(117-2-124-7) 83-6(807-861) 52-0(503-54-1) 30-8(28-5-32-9) 19-9(17-3-23-0) Paraguay 73-9(71-4-769) 56-6(54-9-58-7) 37-7(36-1-39-1) 27-7(26.2-295) 20-9(17-9-24-8) (Continues on next page) mortality at each age are seen between the sexes. finer age groups cannot be based on combined under-5 especially in the neonatal period in which male mortality mortality alone. generally exceeds female mortality."The level of We estimated sex-specific under-5 mortality by modelling mortality at each age for both sexes combined is a the relation between the ratio of male under-5 mortality to function of the level for each sex and the relative size of female under-5 mortality and level of mortality for both the population of each sex;modelling mortality for the sexes combined by use of vital registration data 1994 www.thelancet.com Vol 375 June 5,2010
Articles 1994 www.thelancet.com Vol 375 June 5, 2010 mortality at each age are seen between the sexes, especially in the neonatal period in which male mortality generally exceeds female mortality.36,37 The level of mortality at each age for both sexes combined is a function of the level for each sex and the relative size of the population of each sex; modelling mortality for the fi ner age groups cannot be based on combined under-5 mortality alone. We estimated sex-specifi c under-5 mortality by modelling the relation between the ratio of male under-5 mortality to female under-5 mortality and level of mortality for both sexes combined by use of vital registration data 1970 1980 1990 2000 2010 (Continued from previous page) Europe, western Andorra 16·3 (11·5–23·1) 11·7 (8·3–16·6) 9·4 (6·6–13·3) 7·1 (5·0–10·1) 4·9 (3·4–6·9) Austria 29·4 (28·5–30·2) 17·2 (16·5–17·8) 10·8 (10·3–11·3) 6·1 (5·7–6·5) 3·9 (3·3–4·6) Belgium 25·5 (24·8–26·2) 15·3 (14·7–15·8) 10·2 (9·8–10·7) 6·2 (5·9–6·5) 4·3 (3·6–5·3) Cyprus 26·6 (22·5–31·1) 20·0 (17·9–22·1) 12·8 (11·7–14·0) 6·6 (5·9–7·3) 2·8 (2·3–3·4) Denmark 16·4 (15·8–17·1) 10·1 (9·6–10·7) 8·7 (8·2–9·2) 5·8 (5·5–6·2) 4·1 (3·4–4·9) Finland 15·9 (15·1–16·8) 8·0 (7·5–8·5) 7·1 (6·6–7·5) 4·5 (4·1–4·8) 3·0 (2·5–3·7) France 19·2 (19·0–19·5) 14·7 (14·4–14·9) 9·5 (9·3–9·7) 5·4 (5·3–5·6) 3·9 (3·2–4·8) Germany 25·8 (25·5–26·1) 14·0 (13·8–14·3) 8·9 (8·7–9·1) 5·2 (5·0–5·3) 4·1 (3·5–4·8) Greece 28·7 (27·8–29·4) 19·0 (18·3–19·7) 11·2 (10·6–11·6) 6·9 (6·6–7·3) 3·7 (3·0–4·5) Iceland 16·4 (14·9–18·0) 9·7 (8·6–10·9) 6·7 (5·9–7·7) 4·1 (3·5–4·8) 2·6 (2·1–3·3) Ireland 23·4 (22·5–24·3) 14·9 (14·3–15·6) 9·9 (9·2–10·5) 7·4 (7·0–7·8) 4·2 (3·5–5·1) Israel 27·2 (26·3–28·2) 19·0 (18·3–19·7) 12·1 (11·6–12·6) 6·9 (6·6–7·3) 4·7 (3·9–5·7) Italy 34·3 (34·0–34·7) 17·0 (16·7–17·3) 10·0 (9·8–10·2) 6·4 (6·2–6·6) 3·3 (2·8–4·0) Luxembourg 23·3 (21·5–25·7) 13·9 (12·5–15·4) 9·1 (8·2–10·2) 5·1 (4·6–5·8) 2·9 (2·3–3·7) Malta 31·5 (28·8–34·5) 16·9 (15·2–18·7) 11·3 (10·0–12·8) 7·2 (6·1–8·4) 5·2 (4·0–6·9) Netherlands 16·4 (16·0–16·9) 11·4 (11·0–11·9) 8·9 (8·6–9·2) 6·5 (6·3–6·8) 4·3 (3·7–4·9) Norway 15·8 (15·1–16·4) 10·5 (9·9–11·0) 8·8 (8·4–9·3) 5·0 (4·7–5·3) 3·4 (2·8–4·0) Portugal 74·4 (73·2–75·6) 29·4 (28·6–30·2) 15·0 (14·4–15·6) 8·0 (7·6–8·5) 3·3 (2·6–4·3) Spain 32·8 (32·4–33·2) 14·3 (14·1–14·6) 9·2 (9·0–9·5) 5·7 (5·5–5·9) 3·8 (3·1–4·6) Sweden 12·7 (12·2–13·3) 8·9 (8·5–9·3) 7·2 (6·9–7·6) 4·9 (4·6–5·2) 2·7 (2·2–3·3) Switzerland 18·7 (18·1–19·4) 10·6 (10·1–11·1) 8·8 (8·4–9·2) 6·0 (5·7–6·3) 4·2 (3·5–5·0) UK 21·8 (21·5–22·1) 15·2 (15·0–15·5) 9·7 (9·5–9·9) 6·8 (6·6–6·9) 5·3 (4·5–6·2) Latin America, Andean Bolivia 220·7 (212·7–229·9) 153·0 (148·1–158·0) 103·9 (100·3–107·5) 62·2 (58·6–65·6) 46·7 (39·9–57·0) Ecuador 129·1 (125·8–132·7) 82·5 (80·4–84·6) 50·1 (48·7–51·8) 34·0 (32·1–35·7) 21·0 (18·0–24·3) Peru 159·0 (154·8–163·2) 114·1 (111·6–117·0) 72·7 (70·9–74·6) 40·7 (39·2–42·3) 24·6 (21·6–28·9) Latin America, central Colombia 86·8 (84·0–89·8) 52·8 (51·2–54·4) 33·3 (32·0–34·5) 24·7 (23·6–25·8) 15·3 (13·1–17·7) Costa Rica 73·6 (70·5–77·2) 33·1 (31·5–34·8) 21·4 (20·2–22·6) 14·8 (13·7–16·0) 8·7 (7·3–10·2) El Salvador 160·4 (152·6–168·1) 110·7 (106·5–116·3) 59·4 (56·6–62·1) 34·2 (32·3–36·3) 19·2 (16·7–22·3) Guatemala 175·1 (169·8–181·1) 120·7 (117·8–124·1) 75·8 (73·6–77·7) 49·9 (47·7–52·5) 31·9 (27·6–36·9) Honduras 149·5 (142·3–158·5) 92·9 (90·3–95·8) 56·3 (54·2–58·1) 36·6 (35·3–38·2) 22·6 (19·7–25·4) Mexico 107·5 (104·3–110·6) 70·0 (67·9–72·2) 41·9 (40·2–43·8) 25·8 (24·4–27·4) 16·6 (14·7–18·4) Nicaragua 166·2 (160·2–172·7) 102·3 (99·2–105·2) 63·9 (62·0–66·0) 39·3 (37·8–40·8) 26·6 (23·1–31·0) Panama 54·7 (52·5–57·2) 36·8 (35·2–38·7) 27·8 (26·3–29·3) 23·3 (21·7–25·1) 18·0 (15·3–20·9) Venezuela 58·8 (55·6–62·2) 38·6 (36·4–40·9) 29·4 (27·5–31·1) 23·2 (21·5–25·0) 16·1 (13·2–19·4) Latin America, southern Argentina 72·6 (72·0–73·3) 37·7 (37·3–38·2) 28·0 (27·6–28·3) 19·7 (19·4–20·0) 12·9 (10·5–15·8) Chile 92·1 (85·4–98·8) 37·3 (34·4–40·3) 18·2 (16·7–19·8) 11·1 (10·3–12·1) 6·5 (5·4–7·9) Uruguay 56·0 (54·5–57·7) 40·7 (39·2–42·1) 22·2 (21·2–23·3) 16·5 (15·7–17·3) 11·5 (9·3–14·4) Latin America, tropical Brazil 120·8 (117·2–124·7) 83·6 (80·7–86·1) 52·0 (50·3–54·1) 30·8 (28·5–32·9) 19·9 (17·3–23·0) Paraguay 73·9 (71·4–76·9) 56·6 (54·9–58·7) 37·7 (36·1–39·1) 27·7 (26·2–29·5) 20·9 (17·9–24·8) (Continues on next page)
Articles 1970 1980 1990 2000 2010 (Continued from previous page) North Africa and Middle East Algeria 1724(162-2-1835) 1009(960-106.6) 524(49-8-55-1) 34-1(31-3-37-2) 19-3(14-3-25-1) Bahrain 70-2(633-78-7) 29-6(26-8-32-5) 19-0(17-2-21-0) 12-2(10-9-135) 7-4(6-1-8-8) Egypt 2367(2312-242-8) 157-9(154-8-160-9) 854(83-7-87-0) 456(444-46-7) 247(22-4-27-4) Iran 1831(162-3-211-5) 107-2(100-7-114-2) 65-5(621-69-7刀 46-7(43-2-50-8) 31-1(24-2-38-9) Iraq 111-2(98-4-126-9) 78-2(73-3-83-8) 58.4(55-7-61-7刀 42-7(40-2-450) 31-6(26.9-36-6) Jordan 86.6(82-5-91-2) 54-4(523-56-6) 334(32-1-34-8) 23-7(22-6-24-9) 14-1(11-6-167) Kuwait 52-8(47-8-583) 35-0(31-2-395) 128(113-14-8) 124(11-0-14-1) 7-8(6-0-10-3) Lebanon 68.0(60-3-771) 43-4(40-4-47-0) 314(28-5-34-5) 13-8(12-5-153) 102(7-3-144) Libya 1184(111-8-1252) 61-8(58-6-65-1) 37-6356-398) 22-1(194-25-0) 129(9916-6) Morocco 178.2(173-3-1831) 124-0(121-0-126-9) 76-974-8-787) 50-5(48-6-52-3) 32-4(27-9-37-7) Occupied Palestinian 116-1(90-1-147-2) 70-6(62-5-79-6) 41-9(38-945-0) 293(26-1-32-8) 221(17-8-27-5) territories Oman 190-0(165-8-2163 97-6(85-3-115-2) 37-130-5-455) 159(117-219) 93(6.6-134) Qatar 523(43-6-644) 27-5(24-6-30-9) 159(14-5-17-2) 12-8(116-141) 10-5(88-12-6) Saudi Arabia 204-9(1673-260-1) 71-9(633-819) 29-5(24-8-35-2) 21-8(18.0-26-8) 15-0(11-5-19-8) Syria 853(813-894) 48-2(45-8-50-4) 31-8(30-6-332) 188(17-7-19-8) 11-4(9-6-13-7刀 Tunisia 147-4(142-2-153-4) 80-7(78-3-83-6) 47-4(45-4-49-8) 27-.2(245-29-8) 15-2(118-194) Turkey 194-1(185-5-203-8) 123-9(119-3-128-7) 713(68-4-74-2) 40-0(377-42-5) 29-2(22-6-391) United Arab Emirates 81171-991-1) 38-1(32-3-454) 16-1(12-9-202) 7-0(5-2-99) 3-0(21-43) Yemen 285-1(275-0-296.3) 188-1(181-9-195-5) 128.3(123-1-1333) 93-2(88-7-98-1) 60-0(508-69-6) North America,high income Canada 26-1(257-26-6) 12-9(12-6-13-2) 8-8(8.6-9-1) 6-6(6.4-68) 4-9(4-0-6.0) USA 257(255-259) 16-0(158-16-1) 11-6(11-5-11-7) 8-3(8-2-8-4) 6758-76) Oceania F球 55-0(52-0-58-3) 42-1(38-0-46-4) 35-2(30-5-407) 30-8(26-3-364) 26-6(21-6-32-8) Federated States of 115-6(87-4-147-0 86-5(73-3-102-3) 563(48-5-66-9) 37-7(294-48-0) 25-7(194-34-1) Micronesia Kiribati 1343(117-1-152-8) 102-9(84-8-1251) 758(686-83-6) 592(534-653) 46-4(36-1-60-3) Marshall Islands 49538.3-64-3) 49-8(41-1-60-1) 50-8(44-0-58-6) 47-4(423-5370 37-5(31-2-449) Papua New Guinea 133-1(121-1-147-3引 106-0(995-113-8) 100-2(931-108-0) 91-7(77-6-1083) 82-7(65-7-106.9) Samoa 57-0(51-9-62.9) 36-5(312-42-0) 29-6(24-9353) 24-5(21.3-28-2) 18-6(153-22-8) Solomon Islands 892(78-6-101-4) 56-3(47-0-67-6) 384(33-8-43-5) 34-6(31-3-38.6) 294(23-9-35-6) Tonga 365(32-4-41.6) 29-5(25-5-33-9) 25-8(20-7-32-6) 22-2(17-6-285) 19-2(15-0-246) Vanuatu 116.2(100-0-135-1) 81-0(71-5-914) 52-8(45-6-60-2) 349(29-2-41-5) 23-0(18-3-29-6) Sub-Saharan Africa,central Angola 305-9(267-8-356-3) 266-3(2508-283-4 236-3(222-6-2482) 193-6(1811-206-7) 134-8(1138-155-1) Central African Republic 205-6(193-8-216.1) 178-7(172-3-1853) 163-8(158-1-1701) 153-5(141-2-165-8) 138-1(117-2-160-9) Congo 144-9127-6-1639) 122-4(1145-131-7刃 1094(1035-1153) 1145(107-6-122.2) 107-5(92-4-1241) Democratic Republic of the 241-6(2161-270-2) 207-8(198-0-217-9) 182-9(176-1-190.0) 165-1(1543-176-0) 131-1(117-2-145-6 Congo Equatorial Guinea 209-6(183-8-2388) 191-1(180-5-201-7) 178-7(1694-1877) 180-3(166-7-195-7) 1801(155-5-210.8) Gabon 1678(152-0-185-5 118-1(112-5123-6) 93-8(893-97-6) 83-1(773-89-3引 68-3(58-5-79-8) (Continues on next page) (3253 points)and DHS complete birth history data covariate,and regional and national random effects on (525 points)from 147 countries.This model produced intercept and slope. predictions of the ratio of male to female under-5 mortality We then modelled the probability of an under-5 death that were then combined with the sex ratio at birth (from occurring during the neonatal,postneonatal,and UNPD"if available,and assuming 1.05 if not available)to childhood periods to the sex-specific under-5 mortality form a system of two equations that could be solved to rate,again by use of data from vital registration systems generate estimates of under-5 mortality for boys and (1234points)and DHS complete birth histories(526 points) under-5 mortality for girls.This model included year as a from 122 countries.These models also included regional www.thelancet.com Vol 375 June 5,2010 1995
Articles www.thelancet.com Vol 375 June 5, 2010 1995 (3253 points) and DHS complete birth history data (525 points) from 147 countries. This model produced predictions of the ratio of male to female under-5 mortality that were then combined with the sex ratio at birth (from UNPD10 if available, and assuming 1·05 if not available) to form a system of two equations that could be solved to generate estimates of under-5 mortality for boys and under-5 mortality for girls. This model included year as a covariate, and regional and national random eff ects on intercept and slope. We then modelled the probability of an under-5 death occurring during the neonatal, postneonatal, and childhood periods to the sex-specifi c under-5 mortality rate, again by use of data from vital registration systems (1234 points) and DHS complete birth histories (526 points) from 122 countries. These models also included regional 1970 1980 1990 2000 2010 (Continued from previous page) North Africa and Middle East Algeria 172·4 (162·2–183·5) 100·9 (96·0–106·6) 52·4 (49·8–55·1) 34·1 (31·3–37·2) 19·3 (14·3–25·1) Bahrain 70·2 (63·3–78·7) 29·6 (26·8–32·5) 19·0 (17·2–21·0) 12·2 (10·9–13·5) 7·4 (6·1–8·8) Egypt 236·7 (231·2–242·8) 157·9 (154·8–160·9) 85·4 (83·7–87·0) 45·6 (44·4–46·7) 24·7 (22·4–27·4) Iran 183·1 (162·3–211·5) 107·2 (100·7–114·2) 65·5 (62·1–69·7) 46·7 (43·2–50·8) 31·1 (24·2–38·9) Iraq 111·2 (98·4–126·9) 78·2 (73·3–83·8) 58·4 (55·7–61·7) 42·7 (40·2–45·0) 31·6 (26·9–36·6) Jordan 86·6 (82·5–91·2) 54·4 (52·3–56·6) 33·4 (32·1–34·8) 23·7 (22·6–24·9) 14·1 (11·6–16·7) Kuwait 52·8 (47·8–58·3) 35·0 (31·2–39·5) 12·8 (11·3–14·8) 12·4 (11·0–14·1) 7·8 (6·0–10·3) Lebanon 68·0 (60·3–77·1) 43·4 (40·4–47·0) 31·4 (28·5–34·5) 13·8 (12·5–15·3) 10·2 (7·3–14·4) Libya 118·4 (111·8–125·2) 61·8 (58·6–65·1) 37·6 (35·6–39·8) 22·1 (19·4–25·0) 12·9 (9·9–16·6) Morocco 178·2 (173·3–183·1) 124·0 (121·0–126·9) 76·9 (74·8–78·7) 50·5 (48·6–52·3) 32·4 (27·9–37·7) Occupied Palestinian territories 116·1 (90·1–147·2) 70·6 (62·5–79·6) 41·9 (38·9–45·0) 29·3 (26·1–32·8) 22·1 (17·8–27·5) Oman 190·0 (165·8–216·3) 97·6 (85·3–115·2) 37·1 (30·5–45·5) 15·9 (11·7–21·9) 9·3 (6·6–13·4) Qatar 52·3 (43·6–64·4) 27·5 (24·6–30·9) 15·9 (14·5–17·2) 12·8 (11·6–14·1) 10·5 (8·8–12·6) Saudi Arabia 204·9 (167·3–260·1) 71·9 (63·3–81·9) 29·5 (24·8–35·2) 21·8 (18·0–26·8) 15·0 (11·5–19·8) Syria 85·3 (81·3–89·4) 48·2 (45·8–50·4) 31·8 (30·6–33·2) 18·8 (17·7–19·8) 11·4 (9·6–13·7) Tunisia 147·4 (142·2–153·4) 80·7 (78·3–83·6) 47·4 (45·4–49·8) 27·2 (24·5–29·8) 15·2 (11·8–19·4) Turkey 194·1 (185·5–203·8) 123·9 (119·3–128·7) 71·3 (68·4–74·2) 40·0 (37·7–42·5) 29·2 (22·6–39·1) United Arab Emirates 81·1 (71·9–91·1) 38·1 (32·3–45·4) 16·1 (12·9–20·2) 7·0 (5·2–9·9) 3·0 (2·1–4·3) Yemen 285·1 (275·0–296·3) 188·1 (181·9–195·5) 128·3 (123·1–133·3) 93·2 (88·7–98·1) 60·0 (50·8–69·6) North America, high income Canada 26·1 (25·7–26·6) 12·9 (12·6–13·2) 8·8 (8·6–9·1) 6·6 (6·4–6·8) 4·9 (4·0–6·0) USA 25·7 (25·5–25·9) 16·0 (15·8–16·1) 11·6 (11·5–11·7) 8·3 (8·2–8·4) 6·7 (5·8–7·6) Oceania Fiji 55·0 (52·0–58·3) 42·1 (38·0–46·4) 35·2 (30·5–40·7) 30·8 (26·3–36·4) 26·6 (21·6–32·8) Federated States of Micronesia 115·6 (87·4–147·0) 86·5 (73·3–102·3) 56·3 (48·5–66·9) 37·7 (29·4–48·0) 25·7 (19·4–34·1) Kiribati 134·3 (117·1–152·8) 102·9 (84·8–125·1) 75·8 (68·6–83·6) 59·2 (53·4–65·3) 46·4 (36·1–60·3) Marshall Islands 49·5 (38·3–64·3) 49·8 (41·1–60·1) 50·8 (44·0–58·6) 47·4 (42·3–53·7) 37·5 (31·2–44·9) Papua New Guinea 133·1 (121·1–147·3) 106·0 (99·5–113·8) 100·2 (93·1–108·0) 91·7 (77·6–108·3) 82·7 (65·7–106·9) Samoa 57·0 (51·9–62·9) 36·5 (31·2–42·0) 29·6 (24·9–35·3) 24·5 (21·3–28·2) 18·6 (15·3–22·8) Solomon Islands 89·2 (78·6–101·4) 56·3 (47·0–67·6) 38·4 (33·8–43·5) 34·6 (31·3–38·6) 29·4 (23·9–35·6) Tonga 36·5 (32·4–41·6) 29·5 (25·5–33·9) 25·8 (20·7–32·6) 22·2 (17·6–28·5) 19·2 (15·0–24·6) Vanuatu 116·2 (100·0–135·1) 81·0 (71·5–91·4) 52·8 (45·6–60·2) 34·9 (29·2–41·5) 23·0 (18·3–29·6) Sub-Saharan Africa, central Angola 305·9 (267·8–356·3) 266·3 (250·8–283·4) 236·3 (222·6–248·2) 193·6 (181·1–206·7) 134·8 (113·8–155·1) Central African Republic 205·6 (193·8–216·1) 178·7 (172·3–185·3) 163·8 (158·1–170·1) 153·5 (141·2–165·8) 138·1 (117·2–160·9) Congo 144·9 (127·6–163·9) 122·4 (114·5–131·7) 109·4 (103·5–115·3) 114·5 (107·6–122·2) 107·5 (92·4–124·1) Democratic Republic of the Congo 241·6 (216·1–270·2) 207·8 (198·0–217·9) 182·9 (176·1–190·0) 165·1 (154·3–176·0) 131·1 (117·2–145·6) Equatorial Guinea 209·6 (183·8–238·8) 191·1 (180·5–201·7) 178·7 (169·4–187·7) 180·3 (166·7–195·7) 180·1 (155·5–210·8) Gabon 167·8 (152·0–185·5) 118·1 (112·5–123·6) 93·8 (89·3–97·6) 83·1 (77·3–89·3) 68·3 (58·5–79·8) (Continues on next page)
Articles 1970 1980 1990 2000 2010 (Continued from previous page) Sub-Saharan Africa,east Burundi 229-8(219-7-241-0) 1921(184-1-199-0) 175-8(166-5-187-2) 150-1(141-5-158.6) 130-3(115-1-148-0) Comoros 197-8(182-1-2144) 159-5(1533-166-1) 115-6(1109-120-6) 834(751-92.8) 615(52-972-1) Djibouti 116-2(100-2-133-8) 1113(103-5-119-2) 110-0(104-6-116-1) 896(84-8-944) 66-7(589-74-6) Eritrea 232-0(217-7-248.5) 185-8(1783-191-8) 142-7(138.2-148-4) 100-6(93-0-108-7) 78.2(66-9-945) Ethiopia 255-4(240-1-270-5) 237-1(229-1-245-6) 201-9(194-9-2093) 136-9(130-7-142-3) 101-0(89-3-117-6) Kenya 142-0(137-2-1467) 112-2(1093-115-0) 103-8(101-2-106-8) 1004(959-104-4) 82272-0-91-9) Madagascar 188-4(1790-197-40 180-8(174-5-186-7) 158-0(153-0-163-2) 1045(993-109-6) 70.5(62.9-797刀 Malawi 343-9(330.3-358-1) 260-6(252-3-2701) 211-3(204-8-217-6) 148-4(1434-153-4) 96.8(843-111-6) Mozambique 271-3(255-9-287-9) 2315(223-2-241-1) 226-6(218.7-2361) 1635(156-1-170-8) 1337(114-3-157-9) Rwanda 2323(222.9-241-5) 2037(1945-2139) 169-4(161-7-177-6) 1660(158-0-174-6) 102-9(88.2-118-6) Somalia 206-3(178-7-240.6) 189-0(171-0-2054) 174-2(164-8-1841) 145-1(137-3-1533) 111-4(98.4-123-4) Sudan 152-1(146-8-157-3) 1412(136.4-1462) 118-1(111-4-124-3) 103-7(897-118-8) 921(77-5-110-3引 Tanzania 207-7(1995-2159) 1732(167-9178-0) 153-1(147-9-158-0) 127-1(1217-133-0) 98.4(85-6-116-2) Uganda 190-4(1838-196.3) 185-9(1815-190-7刀 167-8(1635-172-1) 1411(136.8-145.6 1167(105-0-127-5) Zambia 181-8(175-0-1894) 168-3(1632-173-4) 172-8(167.9-178-1) 143-4(138-4-149-1) 118.8(106-3-134-0) Sub-Saharan Africa,southern Botswana 98-4(92-4-103-8) 621(59-1-656) 542(48.5-611) 59-0(50-3-69-8) 49137-7-62-2) Lesotho 172-0(160-9-182-3) 1139(1091-118.8) 94-5(90-8-982) 1049(1002-109-7刀 96-878-8-115-6) Namibia 111-5(1051-118-2) 954(91-7-996) 72-0(693-75-2) 621(592-65-7) 549(45-6-644) SouthAfrica 132-8(1192-146-8) 86-8(81-2-92-2) 57-6(54-2-611) 37-337-0-37-6) 50-9(43-2-603) Swaziland 188-0(167-2-210-3) 122.3(115-0-130-5) 73-7(693-78.4) 99-3(93-7-1064) 1012(79-5-1260) Zimbabwe 111-3(106-5-116-4) 932(89996-9) 73-3(70-4-76-1) 73-7(70-1-77-4) 70-4(57-3-86-1 Sub-Saharan Africa,west Benin 280-6(269-5-293-1) 2221(215-0-228-1) 176-5(172-0-180-9) 137-0(132-7-141-8) 100-7(89-3-111-3) Burkina Faso 319-5(305-5-334-2) 250-2(242-5-258-0) 204-7(1982-211-0) 172-4(1654-1797刀 1337(1143-155-8) Cameroon 2143(205-8-223-6) 174-4(169-1-181.0) 143-5(138-4-148.3) 140-3(134-1-146-5) 1144(943-140-1) Cape Verde 116-2(107-0-130-5) 793(74-5-83-3) 58-2(54-8-61-3) 454(408-515) 323(269-391) Chad 260-0(2433-276-9) 2454(238-6-253-2) 210-8(204-9-216-5) 189-0(182-2-196-6) 168-7(149-1-190-0) Coted'lvoire 225-3217-1-234-8) 170-4(165-0-175-5) 149-2(145-3-153-6) 128-7(123-7-133-4) 107-3(953-121-5) Ghana 174-7(170-0-179-6) 1494(146-3-152-9) 122-2(119-5-125-0) 986(96-1-1011) 77-5722-83-4) Guinea 318-9(3007-336-9) 274-0(266-3-281.2) 226-8(220.9-232-5) 1742(168.4-180-2) 132-7(1193-146-1) Guinea-Bissau 295-0(263-4-3302) 262.4(2505-275-9) 233-2(224-9-242-0) 1911(182-4-199-6) 158.6(143-3-177-9) Liberia 271-6(257-9-285-6) 240-6(227-8-252-70 235-4(223-6-247-7 172-2(163-4-181-5) 98.9(86-6-113-0) Mali 370-33594-382-3) 3017(2943-309-7刀 254-0(247-2-260-0) 212-4(2057-2193) 161-2(142-6-179-5) Mauritania 1899(182-6-1967刀 143-1(138-9147-2) 112-8(109-4-116-2) 100-3(964-1047) 85-577-8-95-0) Niger 354-7(3414-368-2) 326-5(317-1-336-2) 297-0(2892-305-3) 2237216-0-231-5) 161-1(142-0-1853) Nigeria 224-4(214-1-236-5) 200-5(1922-208-7) 194-1(187-1-201-8) 177-0(169-2-185-8) 157-0(140-9-174-0) Sao Tome and Principe 132-1(1191-146-3) 114-8(106-6-1247刀 1094(1015-1183) 749(674-82-8) 62-2(50-2-76-2) Senegal 278-5(270-6-287-1) 203-1(197-4-2082) 146-5(142-0-150-4) 1235(119-1-128-2) 86.57-7-97-0) Sierra Leone 3514(3257-378-8) 2924(279-7-3061) 241-6(232-7-250-7) 196-9(188-8-207-2) 1391(1247-152-8) The Gambia 291-0(263-8-324-0j 209-8(195-4-224-6) 146-0(138-0-154-1) 120-8(113.9-128-1) 80-7(69-5-93-7刀 Togo 225-8(218-4-233-8) 175-0(170-5-180-2) 142-4(137-9-147-5) 1140(108.3-1196) 91-4(82-1-101.8) Countries are groupec n the basis of epidemiological profiles and geography 5 mortality is defined as the probability of dea ath between birth and age 5years. Table 1:Under-5 mortality(uncertainty interval)per 1000,by decade and national random effects on both intercept and slope explored and out-of-sample validity tests undertaken.The and the resulting probabilities were scaled to sum to best performing models from these analyses were 1.0 and converted to age-specific probabilities of death implemented.In out-of-sample predictive validity tests, conditional on survival to the beginning of the period.For with the models fitted repeatedly with data from random both steps,a range of different model specifications were sets of 80%of countries,the median relative error in the 1996 www.thelancet.com Vol 375 June 5,2010
Articles 1996 www.thelancet.com Vol 375 June 5, 2010 and national random eff ects on both intercept and slope and the resulting probabilities were scaled to sum to 1·0 and converted to age-specifi c probabilities of death conditional on survival to the beginning of the period. For both steps, a range of diff erent model specifi cations were explored and out-of-sample validity tests undertaken. The best performing models from these analyses were implemented. In out-of-sample predictive validity tests, with the models fi tted repeatedly with data from random sets of 80% of countries, the median relative error in the 1970 1980 1990 2000 2010 (Continued from previous page) Sub-Saharan Africa, east Burundi 229·8 (219·7–241·0) 192·1 (184·1–199·0) 175·8 (166·5–187·2) 150·1 (141·5–158·6) 130·3 (115·1–148·0) Comoros 197·8 (182·1–214·4) 159·5 (153·3–166·1) 115·6 (110·9–120·6) 83·4 (75·1–92·8) 61·5 (52·9–72·1) Djibouti 116·2 (100·2–133·8) 111·3 (103·5–119·2) 110·0 (104·6–116·1) 89·6 (84·8–94·4) 66·7 (58·9–74·6) Eritrea 232·0 (217·7–248·5) 185·8 (178·3–191·8) 142·7 (138·2–148·4) 100·6 (93·0–108·7) 78·2 (66·9–94·5) Ethiopia 255·4 (240·1–270·5) 237·1 (229·1–245·6) 201·9 (194·9–209·3) 136·9 (130·7–142·3) 101·0 (89·3–117·6) Kenya 142·0 (137·2–146·7) 112·2 (109·3–115·0) 103·8 (101·2–106·8) 100·4 (95·9–104·4) 82·2 (72·0–91·9) Madagascar 188·4 (179·0–197·4) 180·8 (174·5–186·7) 158·0 (153·0–163·2) 104·5 (99·3–109·6) 70·5 (62·9–79·7) Malawi 343·9 (330·3–358·1) 260·6 (252·3–270·1) 211·3 (204·8–217·6) 148·4 (143·4–153·4) 96·8 (84·3–111·6) Mozambique 271·3 (255·9–287·9) 231·5 (223·2–241·1) 226·6 (218·7–236·1) 163·5 (156·1–170·8) 133·7 (114·3–157·9) Rwanda 232·3 (222·9–241·5) 203·7 (194·5–213·9) 169·4 (161·7–177·6) 166·0 (158·0–174·6) 102·9 (88·2–118·6) Somalia 206·3 (178·7–240·6) 189·0 (171·0–205·4) 174·2 (164·8–184·1) 145·1 (137·3–153·3) 111·4 (98·4–123·4) Sudan 152·1 (146·8–157·3) 141·2 (136·4–146·2) 118·1 (111·4–124·3) 103·7 (89·7–118·8) 92·1 (77·5–110·3) Tanzania 207·7 (199·5–215·9) 173·2 (167·9–178·0) 153·1 (147·9–158·0) 127·1 (121·7–133·0) 98·4 (85·6–116·2) Uganda 190·4 (183·8–196·3) 185·9 (181·5–190·7) 167·8 (163·5–172·1) 141·1 (136·8–145·6) 116·7 (105·0–127·5) Zambia 181·8 (175·0–189·4) 168·3 (163·2–173·4) 172·8 (167·9–178·1) 143·4 (138·4–149·1) 118·8 (106·3–134·0) Sub-Saharan Africa, southern Botswana 98·4 (92·4–103·8) 62·1 (59·1–65·6) 54·2 (48·5–61·1) 59·0 (50·3–69·8) 49·1 (37·7–62·2) Lesotho 172·0 (160·9–182·3) 113·9 (109·1–118·8) 94·5 (90·8–98·2) 104·9 (100·2–109·7) 96·8 (78·8–115·6) Namibia 111·5 (105·1–118·2) 95·4 (91·7–99·6) 72·0 (69·3–75·2) 62·1 (59·2–65·7) 54·9 (45·6–64·4) South Africa 132·8 (119·2–146·8) 86·8 (81·2–92·2) 57·6 (54·2–61·1) 37·3 (37·0–37·6) 50·9 (43·2–60·3) Swaziland 188·0 (167·2–210·3) 122·3 (115·0–130·5) 73·7 (69·3–78·4) 99·3 (93·7–106·4) 101·2 (79·5–126·0) Zimbabwe 111·3 (106·5–116·4) 93·2 (89·9–96·9) 73·3 (70·4–76·1) 73·7 (70·1–77·4) 70·4 (57·3–86·1) Sub-Saharan Africa, west Benin 280·6 (269·5–293·1) 222·1 (215·0–228·1) 176·5 (172·0–180·9) 137·0 (132·7–141·8) 100·7 (89·3–111·3) Burkina Faso 319·5 (305·5–334·2) 250·2 (242·5–258·0) 204·7 (198·2–211·0) 172·4 (165·4–179·7) 133·7 (114·3–155·8) Cameroon 214·3 (205·8–223·6) 174·4 (169·1–181·0) 143·5 (138·4–148·3) 140·3 (134·1–146·5) 114·4 (94·3–140·1) Cape Verde 116·2 (107·0–130·5) 79·3 (74·5–83·3) 58·2 (54·8–61·3) 45·4 (40·8–51·5) 32·3 (26·9–39·1) Chad 260·0 (243·3–276·9) 245·4 (238·6–253·2) 210·8 (204·9–216·5) 189·0 (182·2–196·6) 168·7 (149·1–190·0) Côte d’Ivoire 225·3 (217·1–234·8) 170·4 (165·0–175·5) 149·2 (145·3–153·6) 128·7 (123·7–133·4) 107·3 (95·3–121·5) Ghana 174·7 (170·0–179·6) 149·4 (146·3–152·9) 122·2 (119·5–125·0) 98·6 (96·1–101·1) 77·5 (72·2–83·4) Guinea 318·9 (300·7–336·9) 274·0 (266·3–281·2) 226·8 (220·9–232·5) 174·2 (168·4–180·2) 132·7 (119·3–146·1) Guinea-Bissau 295·0 (263·4–330·2) 262·4 (250·5–275·9) 233·2 (224·9–242·0) 191·1 (182·4–199·6) 158·6 (143·3–177·9) Liberia 271·6 (257·9–285·6) 240·6 (227·8–252·7) 235·4 (223·6–247·7) 172·2 (163·4–181·5) 98·9 (86·6–113·0) Mali 370·3 (359·4–382·3) 301·7 (294·3–309·7) 254·0 (247·2–260·0) 212·4 (205·7–219·3) 161·2 (142·6–179·5) Mauritania 189·9 (182·6–196·7) 143·1 (138·9–147·2) 112·8 (109·4–116·2) 100·3 (96·4–104·7) 85·5 (77·8–95·0) Niger 354·7 (341·4–368·2) 326·5 (317·1–336·2) 297·0 (289·2–305·3) 223·7 (216·0–231·5) 161·1 (142·0–185·3) Nigeria 224·4 (214·1–236·5) 200·5 (192·2–208·7) 194·1 (187·1–201·8) 177·0 (169·2–185·8) 157·0 (140·9–174·0) Sao Tome and Principe 132·1 (119·1–146·3) 114·8 (106·6–124·7) 109·4 (101·5–118·3) 74·9 (67·4–82·8) 62·2 (50·2–76·2) Senegal 278·5 (270·6–287·1) 203·1 (197·4–208·2) 146·5 (142·0–150·4) 123·5 (119·1–128·2) 86·5 (77·7–97·0) Sierra Leone 351·4 (325·7–378·8) 292·4 (279·7–306·1) 241·6 (232·7–250·7) 196·9 (188·8–207·2) 139·1 (124·7–152·8) The Gambia 291·0 (263·8–324·0) 209·8 (195·4–224·6) 146·0 (138·0–154·1) 120·8 (113·9–128·1) 80·7 (69·5–93·7) Togo 225·8 (218·4–233·8) 175·0 (170·5–180·2) 142·4 (137·9–147·5) 114·0 (108·3–119·6) 91·4 (82·1–101·8) Countries are grouped into 21 regions on the basis of epidemiological profi les and geography. Under-5 mortality is defi ned as the probability of death between birth and age 5 years. Table 1: Under-5 mortality (uncertainty interval) per 1000, by decade
Articles A B -Asia Pacific,high income Europe,central 200 Latin America.Andean -Australasia -Europe,eastern -Latin Americ.central Europe,wester Latin America,southern (1000m North America. 150 Latin America,tropical high income -North Africa/Middle East 100 10 50 0 0 140 —Asia,central Caribbean 250 —Asia east -Oceania 120 一Asia,southeast -Sub-Saharan Africa, (000 southern 200 150 60 100 Asia.south 50 Sub-Saharan Africa,central 20 Sub-Saharan Africa,east Sub-Saharan Africa,west 0 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 Year Year Figure 4:Under-5 mortality from 1970 to 2010,by region Under-5 mortality is defined as the probability of death between birth and age 5 years first model for estimates from the knocked out countries child who dies under age 1 year,and IMR is the infant was 2.4%for boys and 3.1%for girls;in the same mortality rate.We used a similar method to estimate predictive validity tests for the second model,the median deaths in children aged between 1 year and 5 years. relative error was 7.1%for the neonatal model,9.1%for Deaths in children younger than 5 years were the sum the postneonatal model,and 8.8%for the childhood of deaths in infants younger than 1 year and deaths in model (the results of other predictive validity tests are children aged between 1 year and 5 years.This method of provided in webappendix pp 9-10).The uncertainty from computing deaths in children younger than 5 years is the GPR process was propagated through these two more accurate than is use of under-5 mortality and births models to estimate the uncertainty in our final neonatal, in the current year,because it better accounts for changing postneonatal,and childhood mortality estimates (further cohort sizes and mortality rates from year to year.We details are provided in webappendix p 8). validated this method by comparing it with other approaches (ie,use of under-5 mortality and births,and Aggregate numbers of deaths by converting infant and childhood probabilities of death To compute aggregate numbers of deaths for each country, to mortality rates and multiplying by population estimates) we combined estimates of neonatal and postneonatal in countries with complete vital registration data. mortality to obtain an estimate of the infant mortality In addition to computing under-5 mortality and rate.We obtained deaths in infants younger than 1 year by number of deaths by country,we generated results for applying the infant mortality rate(the probability of death 21 regions of the world.These regions were grouped on from birth to age 1 year)to the number of births in the the basis of epidemiological profiles and geography (see current and previous years with the following formula webappendix p 205 for regions).Analyses were (which gives more weight to the births in the current year undertaken in Stata (version 11.0),R citation (version because most deaths in infants younger than 1 year occur 2.9.0),and Python (version 2.5).We used the PyMC in the first few months of life): package (version 2.0)in Python to implement the Markov chain Monte Carlo sampling. Don-Bo*(1-aon)*IMRn+B*do*IMRom Role of the funding source where D.represents deaths under age 1 year,t is the The sponsors of the study had no role in study design, current year,B is births,d is the mean time lived by a data collection,data analysis,data interpretation,or www.thelancet.com Vol 375 June 5,2010 1997
Articles www.thelancet.com Vol 375 June 5, 2010 1997 fi rst model for estimates from the knocked out countries was 2·4% for boys and 3·1% for girls; in the same predictive validity tests for the second model, the median relative error was 7·1% for the neonatal model, 9·1% for the postneonatal model, and 8·8% for the childhood model (the results of other predictive validity tests are provided in webappendix pp 9–10). The uncertainty from the GPR process was propagated through these two models to estimate the uncertainty in our fi nal neonatal, postneonatal, and childhood mortality estimates (further details are provided in webappendix p 8). Aggregate numbers of deaths To compute aggregate numbers of deaths for each country, we combined estimates of neonatal and postneonatal mortality to obtain an estimate of the infant mortality rate. We obtained deaths in infants younger than 1 year by applying the infant mortality rate (the probability of death from birth to age 1 year) to the number of births in the current and previous years with the following formula (which gives more weight to the births in the current year because most deaths in infants younger than 1 year occur in the fi rst few months of life): where 1D0 represents deaths under age 1 year, t is the current year, B is births, 1a0 is the mean time lived by a child who dies under age 1 year, and IMR is the infant mortality rate. We used a similar method to estimate deaths in children aged between 1 year and 5 years. Deaths in children younger than 5 years were the sum of deaths in infants younger than 1 year and deaths in children aged between 1 year and 5 years. This method of computing deaths in children younger than 5 years is more accurate than is use of under-5 mortality and births in the current year, because it better accounts for changing cohort sizes and mortality rates from year to year. We validated this method by comparing it with other approaches (ie, use of under-5 mortality and births, and by converting infant and childhood probabilities of death to mortality rates and multiplying by population estimates) in countries with complete vital registration data. In addition to computing under-5 mortality and number of deaths by country, we generated results for 21 regions of the world. These regions were grouped on the basis of epidemiological profi les and geography (see webappendix p 205 for regions).38 Analyses were undertaken in Stata (version 11.0), R citation (version 2.9.0), and Python (version 2.5). We used the PyMC package (version 2.0) in Python to implement the Markov chain Monte Carlo sampling. Role of the funding source The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or Figure 4: Under-5 mortality from 1970 to 2010, by region Under-5 mortality is defi ned as the probability of death between birth and age 5 years. Under-5 mortality (per 1000) 0 10 20 30 0 50 100 150 40 200 Caribbean Oceania Sub–Saharan Africa, southern Asia, south Sub–Saharan Africa, central Sub–Saharan Africa, east Sub–Saharan Africa, west A B Europe, central Europe, eastern Europe, western North America, high income Asia Pacific, high income Australasia Latin America, Andean Latin Americ, central Latin America, southern Latin America, tropical North Africa/Middle East Under-5 mortality (per 1000) Year 0 1970 1980 1990 2000 2010 20 40 80 0 50 100 150 140 250 120 60 100 C Year 200 1970 1980 1990 2000 2010 D Asia, central Asia, east Asia, southeast 1D0(t) =B(t) * 1–1a0(t) *IMR(t) +B(t–1) *1a0(t–1) *IMR(t–1) ( )
