Articles

Global, regional, and national causes of child mortality in 2008: a systematic analysis
Robert E Black, Simon Cousens, Hope L Johnson, Joy E Lawn, Igor Rudan, Diego G Bassani, Prabhat Jha, Harry Campbell, Christa Fischer Walker, Richard Cibulskis, Thomas Eisele, Li Liu, Colin Mathers, for the Child Health Epidemiology Reference Group of WHO and UNICEF*

Summary
Background Up-to-date information on the causes of child deaths is crucial to guide global eorts to improve child survival. We report new estimates for 2008 of the major causes of death in children younger than 5 years. Methods We used multicause proportionate mortality models to estimate deaths in neonates aged 027 days and children aged 159 months, and selected single-cause disease models and analysis of vital registration data when available to estimate causes of child deaths. New data from China and India permitted national data to be used for these countries instead of predictions based on global statistical models, as was done previously. We estimated proportional causes of death for 193 countries, and by application of these proportions to the country-specic mortality rates in children younger than 5 years and birth rates, the numbers of deaths by cause were calculated for countries, regions, and the world. Findings Of the estimated 8795 million deaths in children younger than 5 years worldwide in 2008, infectious diseases caused 68% (5970 million), with the largest percentages due to pneumonia (18%, 1575 million, uncertainty range [UR] 1046 million1874 million), diarrhoea (15%, 1336 million, 0822 million2004 million), and malaria (8%, 0732 million, 0601 million0851 million). 41% (3575 million) of deaths occurred in neonates, and the most important single causes were preterm birth complications (12%, 1033 million, UR 0717 million1216 million), birth asphyxia (9%, 0814 million, 0563 million0997 million), sepsis (6%, 0521 million, 0356 million0735 million), and pneumonia (4%, 0386 million, 0264 million0545 million). 49% (4294 million) of child deaths occurred in ve countries: India, Nigeria, Democratic Republic of the Congo, Pakistan, and China. Interpretation These country-specic estimates of the major causes of child deaths should help to focus national programmes and donor assistance. Achievement of Millennium Development Goal 4, to reduce child mortality by two-thirds, is only possible if the high numbers of deaths are addressed by maternal, newborn, and child health interventions. Funding WHO, UNICEF, and Bill & Melinda Gates Foundation.
Lancet 2010; 375: 196987 Published Online May 12, 2010 DOI:10.1016/S01406736(10)60549-1 See Comment page 1941 *Members listed at end of paper Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (Prof R E Black MD, H L Johnson PhD, C Fischer Walker PhD, L Liu PhD); London School of Hygiene and Tropical Medicine, London, UK (Prof S Cousens MA); Saving Newborn Lives/Save the Children, Cape Town, South Africa (J E Lawn PhD); University of Edinburgh Medical School, Edinburgh, UK (Prof I Rudan MD, Prof H Campbell MD); Centre for Global Health Research, Li Ka Shing Knowledge Institute, St Michaels Hospital, University of Toronto, Toronto, ON, Canada (D G Bassani PhD, Prof P Jha MD); Global Malaria Programme (R Cibulskis PhD) and Department of Health Statistics and Informatics (C Mathers PhD), WHO, Geneva, Switzerland; and Department of International Health and Development, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA (T Eisele MD) Correspondence to: Prof Robert E Black, Department of International Health, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe Street, Baltimore, MD 21205, USA rblack@jhsph.edu

Introduction
Child mortality has been declining worldwide as a result of socioeconomic development and implementation of child survival interventions, yet 88 million children die every year before their fth birthday.1 The aim of UN Millennium Development Goal 4 (MDG 4) is to reduce mortality of children younger than 5 years by two-thirds between 1990 and 2015, but many countries, especially in south Asia and sub-Saharan Africa, are not on track to meet this target.1 An acceleration of the decline in mortality is possible with expansion of interventions targeting the important causes of death.24 In view of the short time left to meet MDG 4, demand is increasing for frequently updated national data on the causes of child mortality to guide national and global programmatic priorities and research. Updates of total mortality in children younger than 5 years are published every year, and the most recent estimates are for 2008. WHO and UNICEFs Child Health Epidemiology Reference Group (CHERG) undertook the last comprehensive review of the causes of child mortality worldwide for 200003.5 The Countdown to 2015 Initiative
www.thelancet.com Vol 375 June 5, 2010

used these estimates to produce country proles for 68 countries of low and middle income to assess their progress towards MDG 4.6 Availability of more recent data and improved methods will enable updated estimates of cause-specic child mortality. We present estimates of the distribution of causes of child deaths in 2008 for 193 countries, with aggregated regional and global totals.

Methods
Mortality rates in children younger than 5 years
Figure 1 summarises the process used to develop the estimates. Methods to estimate the country-specic mortality rates in children younger than 5 years (5q0) have been developed and agreed by the Inter-agency Group for Child Mortality Estimation (IGME), which consists of representatives of WHO, UNICEF, UN Population Division, World Bank, and academic institutions.7,8 Every year, IGME assesses and incorporates data from all available surveys, censuses, and vital registration systems to estimate change in the mortality rate of children younger than 5 years in each country.

1969

Articles

Countries with high death registration

Vital registration data Total number of deaths in children aged 059 months Survey data adjusted for changes in U5MR through 2008 for country or region Livebirths, population aged 0 and 14 years

U5MR for each country Livebirths, population aged 0 and 14 years NMR for each country

Surveys, censuses, vital registration data UN Population Division data

Countries with low death registration

UN Population Division data

Total number of deaths in neonates aged 027 days

Total number of deaths in children aged 159 months

Estimation of causes of death in neonates aged 027 days

Estimation of causes of death in children aged 159 months

Method used to produce cause-specic proportionate mortality Countries with >80% of deaths recorded as part of routine vital registration data Countries with 80% vital registration data combined with: NMR <15 per 1000 livebirths Multicause model with use of vital registration data Mean of vital registration multicause model (above) and high-mortality multicause model (below) Multicause model with use of community-based studies with 5 causes of death and countrylevel covariates for 2008 Registered deaths grouped according to standard ICD-10 codes

Method used to produce cause-specic proportionate mortality Registered deaths grouped according to standard ICD-10 codes Countries with >85% of deaths recorded as part of routine vital registration data Countries with 85% vital registration data combined with: Multicause model with use of vital registration data Mean of vital registration multicause model (above) and high-mortality multicause model (below) Multicause model with use of community-based studies with 2 causes of death and countrylevel covariates for 2008 U5MR <26 per 1000 livebirths and GNI >$7510

NMR 1520 per 1000 livebirths

U5MR 2635 per 1000 livebirths and GNI >$7510

NMR >20 per 1000 livebirths

U5MR >35 per 1000 livebirths and GNI <$7510

Neonatal tetanus model

WHO estimates for malaria, measles, and pertussis UNAIDS estimates for AIDS

Number of deaths by cause

Figure 1: Overview of procedures for estimation of deaths by cause in children younger than 5 years of age U5MR=mortality rate in children younger than 5 years. NMR=neonatal mortality rate. GNI=gross national income per person (international dollars). ICD-10=International Classication of Diseases, 10th revision.

Child mortality estimates that we have used are generally consistent with those previously published,9 apart from those for several high-income countries in which death registration data became available for a more recent year. These published estimates also incorporate a new adjustment in which data for 17 countries with high HIV prevalence were revised to correct for bias in survey data from deceased mothers; such bias is particularly a problem in settings with high AIDS mortality. Data were also adjusted for misreporting of the date of birth and the estimated change in child deaths due to AIDS.10
1970

Estimated livebirths, and de-facto numbers of children aged 0 and 14 years for 2008 are taken from the UN Population Division 2008 revision.11 Total deaths in children younger than 5 years for 2008 were estimated by application of the IGME-estimated mortality rates for children aged 0 and 14 years to the de-facto population for these age-groups.

Neonatal mortality rates

Figure 1 summarises the process used to develop the estimates. Methods used to estimate mortality rates in
www.thelancet.com Vol 375 June 5, 2010

Articles

Sites contributing data to the verbal autopsy model Vital registration data Vital registration data-based model Verbal autopsy data-based model Average of vital registration and verbal autopsy models National sample death registration with verbal autopsy data National verbal autopsy model

Figure 2: Methods used to estimate causes of death in neonates aged 027 days in 2008

neonates (aged 027 days) have been revised and updated for 2008. The previous WHO method for estimation12 has been revised to incorporate the eect of the projected change in the mortality rate of children younger than 5 years from the years for which data are available until 2008. The WHO database of mortality rates in pairs of neonates and children younger than 5 years that were recorded from death registration systems and nationally representative household surveys has been updated to include 3263 country-year datapoints across 167 countries and all WHO regions, of which 966 country-years are from survey data.13 For countries without high coverage of death registration but with suitable survey data, we applied a regression model to data from 1990 onwards, after adjustment to match the estimated trend in the mortality rate of children younger than 5 years, according to the equation:

Analysis and application of vital registration data

Data from vital registration systems on causes of deaths for neonates and children aged 159 months were extracted from the WHO mortality database, with adjustment for incomplete coverage if needed.8 We included countries from the database in the analysis if they had adequate registration of deaths (80% for neonates, 85% for children aged 159 months). Data closest to 2008 were used, but for very small countries (mostly islands in the Caribbean and Pacic), we used a mean of the number of deaths during the 3 years (in one case 5 years) closest to 2008 to obtain a more stable estimate of mortality by cause. In a few cases, causes of deaths in neonates and children aged 159 months were imputed from totals for children aged 04 years. For this imputation, we used the mean cause-specic ratio of neonatal deaths to deaths in children aged 159 months from other countries, accounting for 04% (0037 million/8795 million) of deaths in children younger than 5 years. Causes of deaths were categorised according to the International Classication of Diseases, 10th revision (ICD-10; webappendix pp 13). Some deaths that were coded to causes inappropriate for the neonatal period were reassigned (webappendix pp 46). Additionally, we reassigned neonatal deaths as congenital abnormalities if they were classied according to ICD-10 as due to endocrine disorders, nutritional and metabolic diseases, or diseases of the nervous, digestive, circulatory, musculoskeletal, and genitourinary systems, because these disorders are probably caused by congenital

log(Pr[NMR/1000])=+1log(Pr[U5MR/1000])+ 2(log[Pr(U5MR/1000)])2+3Xi
where NMR is neonatal mortality rate, U5MR is mortality rate in children younger than 5 years, Xi is one for country i and zero otherwise, and 3 is a countrylevel xed eect. For countries without data for mortality rates in both neonates and children younger than 5 years, this regression model was run with aggregated regional data with regional xed eects, rather than country-level xed eects, to predict the 2008 mortality-rate ratio of neonates to children younger than 5 years.
www.thelancet.com Vol 375 June 5, 2010

See Online for webappendix

1971

Articles

Sites contributing data to the verbal autopsy model Vital registration data Vital registration data-based model Verbal autopsy data-based model Average of vital registration and verbal autopsy models National sample death registration with verbal autopsy data National verbal autopsy model

Figure 3: Methods used to estimate causes of death in children aged 159 months in 2008

malformations. Deaths reported with ill-dened causes (ICD, 9th revision, chapter 16, or 10th revision, chapter 18, on symptoms, signs, and abnormal clinical and laboratory ndings not classied elsewhere) were reassigned to other causes in proportion to the number of reported deaths. The term birth asphyxia refers to intrapartum-related neonatal deaths, excluding preterm births and lethal congenital abnormalities. This term will be used to allow comparison with previous reports, but we recognise recommendations to change to terms such as death from intrapartum-related causes14 that might eventually become widely adopted. A model was developed for countries with low neonatal mortality and no useable death registration data by use of death registration data for low-mortality countries with adequate registration, as previously reported.15,16 This model was used for countries with less than 15 neonatal deaths per 1000 livebirths, and the highmortality model, described below, was used for countries with more than 20 neonatal deaths per 1000 livebirths. For countries with 1520 neonatal deaths per 1000 livebirths, both models were tted and a mean of the two results was used. For countries with inadequate death registration, and with fewer than 26 deaths in children younger than 5 years per 1000 livebirths or gross national income (GNI) per person at purchasing power parity of more than $7510 (international dollars), we estimated proportional causes of death in children aged 159 months with a multicause multinomial logistic regression model that used death registration data from
1972

97 countries to identify causes of death in children aged 159 months. This model included covariates for mortality rates in children younger than 5 years, GNI per person, and regional indicator variables for Europe and Latin America and the Caribbean, according to WHO regional classications. For countries with inadequate death registration, and with 2635 deaths in children younger than 5 years per 1000 livebirths and GNI per person of more than $7510, we took the mean of estimates from this model with those from the model used for high-mortality countries, described below, to estimate the causes of death in children aged 159 months.

Estimation of causes of deaths in high-mortality countries without complete vital registration data
To estimate causes of neonatal deaths, the multicause model for neonatal deaths15,16 was revised to include additional study data from sites contributing data (gure 2), and was rerun with updated covariate data for 2008. Causespecic results were adjusted country-by-country to t the estimated number of neonatal deaths for 2008. We also estimated causes of deaths in children aged 159 months. A full description of the methods in countries with incomplete death certication, 35 or more deaths in children younger than 5 years per 1000 livebirths, and GNI per person of $7510 or less is provided elsewhere.17 In brief, we analysed 81 datapoints from community-based mortality studies in which at least two causes of death had been reported in children aged 159 months (gure 3). Eligible studies were identied from systematic searches of
www.thelancet.com Vol 375 June 5, 2010

Articles

Estimated number (UR; millions) Neonates aged 027 days Preterm birth complications Birth asphyxia Sepsis Other Pneumonia* Congenital abnormalities Diarrhoea Tetanus Children aged 159 months Diarrhoea Pneumonia* Other infections Malaria Other non-communicable diseases Injury AIDS Pertussis Meningitis Measles Congenital abnormalities 1257 (07741886) 1189 (07891415) 0753 (04792830) 0732 (06010851) 0228 (01430606) 0279 (01740738) 0201 (01860215) 0195 () 0164 (01100728) 0118 (00750180) 0104 (00780160)
Malaria 8%

Pneumonia 14% Other non-communicable diseases 4%* 4% Preterm birth complications 12%

1033 (07171216) 0814 (05630997) 0521 (03560735) 0409 (03180883) 0386 (02640545) 0272 (02050384) 0079 (00570211) 0059 (00320083)

Birth asphyxia 9% Neonatal deaths 41% Sepsis 6% Other 5% Congenital abnormalities 3% 14% 1% Tetanus 1%

Other infections 9%

Meningitis 2% Pertussis 2% AIDS 2%

Injury 3% Measles 1%

Diarrhoea

Figure 4: Global causes of child deaths Data are separated into deaths of neonates aged 027 days and children aged 159 months. Causes that led to less than 1% of deaths are not presented. *Includes data for congenital abnormalities.

Uncertainty range (UR) is dened as the 25975 centile. =data unavailable. *Estimated number of deaths in children younger than 5 years overall is 1575 million (UR 1046 million1874 million). Estimated number of deaths in children younger than 5 years overall is 0376 million (UR 0283 million0580 million). Estimated number of deaths in children younger than 5 years overall is 1336 million (UR 0822 million2004 million). Uncertainty range is based on UNAIDS estimated lower and upper bounds for deaths in children younger than 15 years. Crowcroft and colleagues20 sensitivity analysis presents extreme upper and lower values for various inputs.

Table 1: Estimated numbers of deaths by cause in 2008

published reports and unpublished sources; studies were included if they were done after 1979, had a duration of 12 months or a multiple of 12 months, and recorded at least 25 deaths in children younger than 5 years, with each death represented once, and less than 25% of deaths due to unknown causes. Deaths were grouped into one of seven categories: pneumonia, diarrhoea, malaria, injury, meningitis or encephalitis, measles, or other known causes. Deaths attributed to neonatal causes, AIDS, or undetermined causes were excluded. Since we used only deaths due to known causes in the estimation, we assumed that the proportional distribution of deaths with undetermined causes was the same as that for deaths with known causes. Deaths attributed to more than one of the seven causes were reallocated on the basis of the relative importance of single causes in the same studies. Deaths attributed to malnutrition were reallocated to one of the ve infection categories on the basis of the relative importance of each of these causes. Similar to the high-mortality model for neonatal deaths,16 a study-based multinomial logistic regression model was applied to country-level data to estimate causes of child death.17 We used ordinary least squares regression
www.thelancet.com Vol 375 June 5, 2010

to identify possible explanatory variables for the log-ratio models of the proportion of each of the six causes of death relative to the proportion of the reference cause. After selection of pneumonia as the reference cause and nal covariates for the six log-ratio models, studies were given a weight proportional to the inverse of the square root of the total number of deaths. These studies were then included in the study-level multinomial logistic regression model with robust standard errors to obtain parameter estimates. We applied country-level covariate data from public databases (eg, WHO, UNICEF, World Bank) for 2008 to the study-based multinomial logistic regression model to estimate country-level cause-specic fractions. These fractions were then multiplied by the total number of non-AIDS deaths in children aged 159 months for each country. The country-level estimates of deaths by cause were adjusted for estimated eects of recently scaled up interventions: pneumonia and meningitis estimates were adjusted for use of the Haemophilus inuenzae type b vaccine, and malaria estimates were adjusted for use of insecticide-treated bednets. Countrylevel estimates of causes of death were then combined with cause-specic data from WHO technical programmes and deaths due to AIDS from UNAIDS, and were adjusted to the estimated total number of deaths in children aged 159 months, as described below.

Estimation of deaths due to malaria, pertussis, measles, tetanus, meningitis, and AIDS
For countries without complete vital registration data, cause fractions from the neonatal and child multicause models were replaced by cause-specic inputs from WHO technical programmes for malaria, pertussis, measles, and tetanus. The resulting cause fractions were adjusted
1973

Articles

Proportion of deaths by age 027 days (neonates) 159 months WHO regions Africa Americas Eastern Mediterranean Europe Southeast Asia Western Pacic

Figure 5: Distribution of 8795 million child deaths in 2008 by WHO region and by age

to sum to one. WHO has prepared revised and updated malaria mortality estimates for children younger than 5 years for 2008 by use of the data sources and methods described in the World Malaria Report 2009.18 Updated estimates of pertussis cases were prepared by the WHO Department of Immunization, Vaccines and Biologicals (IVB) by use of WHO and UNICEF estimates for vaccination coverage in 2008, interpolated for missing data.19 For countries without complete data for death registration, deaths from pertussis were based on a natural history model of vaccine coverage and age-specic casefatality rates from community-based studies.20 Measles incidence and mortality for 2008 were estimated from a revised natural history model of routine vaccination coverage,19 supplementary immunisation activities, reported measles cases, estimates of notication eciency, and estimates of age-specic case-fatality rates.21 Estimation of deaths from measles in India is described below. IVB and CHERG have developed a revised statistical model that predicts the odds of neonatal death due to tetanus in high-burden countries. This model is based on WHO estimates of literacy in women, and the proportions of births that are protected from tetanus and are delivered by skilled birth attendants.9,22 We used the neonatal tetanus model for high-burden countries for 2008 estimates. For low-burden countries without complete death registration data or estimates from the neonatal tetanus model, we used the CHERG neonatal multinomial logistic regression model15,16 to estimate neonatal deaths due to tetanus. For high-mortality countries, we used results from the multicause model for children aged
1974

159 months to estimate deaths due to meningitis or encephalitis, apart from in a few countries with low meningitis mortality, for which previously published estimates were used.23,24 Estimates for deaths due to AIDS were done by UNAIDS.10

Estimation of causes of deaths in India and China

The Registrar General of India has introduced substantially revised verbal autopsy methods into the sample registration system as part of Indias continuing Million Death Study.25,26 To estimate child deaths by cause for India, CHERG collaborated with the Million Death Study team to use data from a nationally representative sample of more than 123 000 deaths (23 000 child deaths) in 200103. Causes of deaths in neonates and children aged 159 months were assigned to categories (webappendix pp 13), and weighted by rural and urban subdivisions of each state to provide nationally representative cause fractions. Estimates of deaths due to measles in children younger than 5 years from India's Million Death Study were substantially smaller than were those derived from the natural history model.19 In the Million Death Study, verbal autopsy rules included deaths due to measles for which either an underlying or an intermediate cause of death had been assigned, but some deaths from measles might have been misclassied. The natural history model estimates might have been too high, mainly because they are heavily aected by an estimated case-fatality rate. We calculated the mean of estimates from the Million Death Study and the natural history model to provide a
www.thelancet.com Vol 375 June 5, 2010

Articles

Africa (4199 million deaths) Other non-communicable diseases 2%* Other infections 9%

Pneumonia 15% 3%

Preterm birth complications 8% Birth asphyxia 8%

Americas (0284 million deaths) Other non-communicable diseases 14%*

Pneumonia 2% 10%

Preterm birth complications 18%

Neonatal deaths 29%

Meningitis 2% Pertussis 2% AIDS 4%

Sepsis 5% Congenital abnormalities 2% Other 1% Tetanus 1% 1% Diarrhoea 18%

Neonatal deaths 48% Other infections 12%

Congenital abnormalities 9%

Birth asphyxia 7%

Malaria 16%

Measles 1% Injury 2% Pneumonia 14% 5%

Meningitis 1% Pertussis 1% AIDS 1% Injury 6% Diarrhoea 7% Europe (0148 million deaths) Other non-communicable diseases 14%* Pneumonia 3% 11%

Other 7% Sepsis 5%

Eastern Mediterranean (1239 million deaths) Other non-communicable diseases 4%*

Preterm birth complications 14%

Preterm birth complications 18%

Other infections 10%

Neonatal deaths 45%

Birth asphyxia 10%

Neonatal deaths 53% Other infections 9%

Meningitis 2% Pertussis 2% Malaria 3% Injury 3% Congenital abnormalities 5% 17% Diarrhoea Southeast Asia (2390 million deaths) Other non-communicable diseases 4%* Other infections 3% Meningitis 2% Pertussis 4% Malaria 1% Injury 4% Measles 3% Other 9% Neonatal deaths 54% Birth asphyxia 11% Pneumonia 13% 8% Preterm birth complications 14% Other 2% 1% Tetanus 1% Sepsis 7%

Congenital abnormalities 11%

Meningitis 2% Injury 6% Diarrhoea 5% Sepsis 3% Western Pacic (0534 million deaths) Other 10% Birth asphyxia 8%

Pneumonia 16% 2%

Preterm birth complications 15%

Other non-communicable diseases 10%* Birth asphyxia 14% Neonatal deaths 52% Other infections 8% Meningitis 2% Injury 8% 12% 2% Diarrhoea Sepsis 7% Congenital abnormalities 2% Tetanus 1% Diarrhoea 4% Sepsis 2% Other 14% Congenital abnormalities 5%

Figure 6: Regional causes of child deaths Data are separated into deaths of neonates aged 027 days and children aged 159 months. Causes that led to less than 1% of deaths are not presented. *Includes data for congenital abnormalities.

www.thelancet.com Vol 375 June 5, 2010

1975

Articles

All cause by age <5 years Africa Algeria Angola Benin Botswana Burkina Faso Burundi Cameroon Cape Verde Central African Republic Chad Comoros Congo (Brazzaville) Cte dIvoire Democratic Republic of the Congo Equatorial Guinea Eritrea Ethiopia Gabon Ghana Guinea Guinea-Bissau Kenya Lesotho Liberia Madagascar Malawi Mali Mauritania Mauritius Mozambique Namibia Niger Nigeria Rwanda So Tom and Principe Senegal Seychelles Sierra Leone South Africa Swaziland Tanzania The Gambia Togo Uganda Zambia Zimbabwe Americas Antigua and Barbuda Argentina Barbados 19 10 451 33 14 6104 21 6 4347 12 30 184 165 344 39 267 1448 117 069 44 786 89 461 342 25 936 98 624 2274 15 537 79 165 553 529 3478 10 237 321 432 3029 54 723 54 326 12 164 188 928 4567 20 328 71 086 56 348 100 403 12 344 289 110 420 2425 120 938 1 076 613 41 234 493 48 933 13 42 671 72 553 2873 174 959 6239 19 985 189 990 76 812 35 583 17 258 36 299 11 150 741 25 694 11 661 21 939 193 7255 22 084 898 4303 29 267 162 690 996 3032 121 594 1099 22 672 17 039 2910 50 033 2210 6425 23 964 17 193 28 463 4830 166 37 565 1034 26 874 298 154 14 197 167 16 035 9 9959 21 316 638 59 243 2067 7020 45 007 19 557 10 758 12 925 129 045 28 117 706 91 374 33 124 67 523 149 18 682 76 540 1377 11 234 49 899 390 839 2482 7205 199 839 1930 32 052 37 288 9254 138 896 2358 13 904 47 122 39 155 71 940 7514 123 72 855 1391 94 064 778 459 27 037 326 32 898 4 32 712 51 237 2235 115 717 4172 12 965 144 983 57 255 24 825 027 days 159 months

AIDS

Diarrhoea

Pertussis

Tetanus

Measles

39 2951 530 0 995 1095 4362 0 1758 2850 0 727 3247 6281 107 449 9080 294 1437 1014 231 8855 774 568 81 7693 1116 67 2 15 504 441 577 29 457 281 0 710 0 638 33 149 1407 14 889 84 1152 9542 9179 7541 0 16 1

3876 41 403 5114 101 22 071 10 568 14 484 28 4484 21 552 457 2222 10 424 102 679 326 2188 73 341 180 5193 7524 2321 38 802 451 3493 15 964 6169 19 261 1936 5 13 105 153 24 575 201 368 9305 72 7240 0 8904 6293 240 20 211 867 2324 30 391 11 222 3291 0 94 0

421 2073 1389 0 817 157 672 4 817 4273 26 108 927 10 866 150 68 3535 99 605 1611 283 1432 57 651 747 353 2114 148 0 1229 60 3429 32 353 142 1 350 0 1050 1 17 1507 26 164 6472 568 1808 0 50 0

187 657 190 5 1251 256 187 3 87 1901 6 20 320 1029 7 102 4898 10 244 328 31 440 9 77 268 191 568 84 0 779 7 1897 8619 160 2 338 0 169 139 5 565 36 122 510 117 60 0 0 0

220 1129 98 0 35 494 719 20 1 5 0 0 1 4095 327 168 187 27 1115 1827 290 1568 0 385 0 0 9 0 0 0 158 106 707 468 3 1458 0 2027 1 0 0 59 20 4060 400 2680 0 0 0

1976

www.thelancet.com Vol 375 June 5, 2010

Articles

Meningitis

Malaria

Pneumonia

Other infections*

Preterm birth complications

Birth asphyxia

Neonatal sepsis Congenital abnormalities

Other nonInjury communicable diseases

317 2685 1007 139 1407 1025 2405 13 482 1532 89 306 1922 10 612 61 190 6797 73 594 1016 355 3674 125 347 2004 1590 1062 274 7 2590 50 2835 17 637 1155 18 461 0 621 607 63 5414 67 375 4101 1528 796 0 94 0

0 13 911 9165 15 23 875 4142 17 038 1 3715 18 338 0 3705 16 704 94 196 969 29 21 786 888 14 365 12 832 2147 20 666 0 3163 2494 9376 20 918 1645 0 13 789 124 21 714 217 357 2436 4 9167 0 5520 51 1 28 745 1447 5130 42 549 11 709 1222 0 0 0

5862 33 078 7616 177 24 374 7459 16 115 49 5254 19 010 499 2489 13 222 112 655 410 1977 48 892 329 5712 9245 2191 30 406 585 3449 15 130 7376 18 802 2413 21 19 447 334 26 319 177 212 6272 127 8735 1 8444 6250 337 25 005 1002 2949 25 751 11 323 4728 1 743 1

3452 26 858 4125 133 17 827 6796 11 248 30 2856 10 238 319 1685 7073 61 928 171 1739 34 791 110 4159 4229 1352 27 630 420 2120 10 202 5936 10 647 1307 36 10 568 160 15 243 108 805 5753 57 6200 3 5337 4212 198 17 135 736 1605 17 513 9793 2514 1 1049 4

6620 10 427 3981 329 7586 3072 7523 83 2234 5699 368 1726 9532 54 669 356 1097 28 439 456 6625 4327 942 14 417 736 2063 8062 5649 8416 1565 66 11 067 439 8543 84 164 3746 61 5047 5 3269 8973 243 17 058 707 2257 13 734 5699 4094 9 3090 12

3824 9531 2470 153 5874 3354 5782 39 1861 6340 216 1000 6975 36 045 257 684 36 295 249 6207 4410 744 14 581 599 1656 6344 4556 6838 1169 42 9405 203 6823 85 417 4260 40 3768 1 2459 5113 130 17 043 486 1720 12 806 5467 2926 1 574 5

1708 7628 964 59 3109 2726 3511 16 1113 2879 129 581 4591 23 497 153 413 28 681 128 4923 3433 617 11 479 475 1238 4646 3321 3944 789 15 6176 94 3062 62 221 3471 26 2395 0 1562 1720 73 13 138 291 950 9713 4486 1546 0 593 2

2279 2945 1171 149 1958 836 2204 32 613 1281 64 499 2406 15 055 118 336 6645 130 2028 1126 202 4226 202 457 1655 1590 2074 376 63 2994 137 1467 22 210 1055 19 1451 2 537 2378 83 5407 171 672 3769 1664 1235 4 2572 4

678 5182 560 116 2869 1074 1509 15 352 1274 59 230 778 9158 17 321 9637 23 435 643 208 4225 39 367 2055 1166 2238 263 22 1746 26 2276 13 300 1032 48 768 1 1065 2288 48 2912 82 216 2338 1352 498 1 879 1

700 4886 888 72 3020 1732 1703 11 309 1452 43 238 1042 10 764 47 477 8429 32 1081 762 249 6529 95 293 1435 1381 2398 308 12 2020 39 2072 15 786 1697 15 845 0 1071 1377 28 5930 177 331 6742 2306 643 1 696 2 (Continues on next page)

www.thelancet.com Vol 375 June 5, 2010

1977

Articles

All cause by age <5 years (Continued from previous page) Belize Bolivia Brazil Canada Chile Colombia Costa Rica Cuba Dominica Dominican Republic Ecuador El Salvador Grenada Guatemala Guyana Haiti Honduras Jamaica Mexico Nicaragua Panama Paraguay Peru Saint Kitts and Nevis Saint Lucia Saint Vincent and the Grenadines Suriname The Bahamas Trinidad and Tobago Uruguay USA Venezuela Eastern Mediterranean Afghanistan Bahrain Djibouti Egypt Iran Iraq Jordan Kuwait Lebanon Libya Morocco Oman Pakistan Qatar Saudi Arabia Somalia Sudan Syria 310 739 171 2241 44 806 45 571 42 159 3070 557 860 2435 23 591 699 464 886 122 12473 75 933 138 357 9625 63 124 81 861 27 175 26 147 23 687 1993 295 499 1364 15 086 431 284 149 56 6946 23 978 52 532 4693 247 615 89 1380 17 631 19 424 18 472 1077 262 361 1071 8505 268 180 736 65 5527 51 955 85 825 4932 144 14 169 66 624 2210 2231 18 103 775 750 13 7346 7022 2259 30 15 401 827 19 168 6144 1651 36 367 3845 1626 4315 147 68 14 48 25 257 75 697 798 35 238 10 805 62 6506 33 112 1319 1168 11 046 494 383 10 4156 3190 940 25 4754 316 6736 3114 454 14 733 1762 707 2335 7647 11 36 19 115 31 484 401 19 260 5922 82 7663 33 513 892 1063 7057 281 366 3 3190 3832 1319 4 10 647 511 12 432 3030 1197 21 634 2084 919 1979 7121 4 12 7 143 44 213 396 15 978 4883 027 days 159 months

AIDS

Diarrhoea

Pertussis

Tetanus

Measles

6 28 151 0 3 51 0 1 0 96 64 88 0 427 48 873 84 61 83 25 11 43 191 0 0 0 6 1 43 1 6 32 17 0 136 33 242 0 0 0 6 0 22 0 425 0 0 165 3350 0

11 2158 3543 1 6 750 11 10 0 677 388 82 0 2937 126 3887 605 231 2002 343 98 378 658 0 0 1 15 1 4 16 33 777 89 709 1 418 2146 4480 4871 119 3 21 89 2917 13 74 209 2 597 16 576 14726 449

0 262 36 0 15 21 7 0 0 9 0 0 0 312 0 1541 4 1 6 22 27 3 0 0 0 0 0 0 0 0 40 10 1093 0 21 155 11 2325 0 0 4 0 201 0 7180 0 0 2991 1184 1

0 52 213 0 0 83 0 0 0 18 31 12 0 228 1 439 21 10 79 25 0 19 131 0 0 0 0 0 0 0 0 65 2766 0 9 518 173 348 11 0 0 0 279 7 5604 0 0 2068 546 60

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1822 0 5 155 0 98 0 0 4 0 3 0 275 0 0 3848 7 1

1978

www.thelancet.com Vol 375 June 5, 2010

Articles

Meningitis

Malaria

Pneumonia

Other infections*

Preterm birth complications

Birth asphyxia

Neonatal sepsis Congenital abnormalities

Other noncommunicable diseases

Injury

1 359 977 15 26 272 5 12 0 77 154 100 0 502 11 490 141 25 131 103 21 74 461 0 0 0 8 1 3 9 175 119 7557 0 35 1843 449 973 45 0 25 10 319 4 6218 0 47 1317 2481 128

0 3 38 0 0 12 0 0 0 4 0 0 0 0 7 166 2 0 0 0 0 0 17 0 0 0 2 0 0 0 0 1 46 0 4 0 1 0 0 0 0 0 0 0 539 0 0 4440 34 434 0

16 2582 5846 30 125 2215 39 81 1 1353 1198 309 4 3114 142 3862 1084 374 4551 757 217 725 2290 0 0 0 26 7 40 57 890 1118 80 694 2 417 4748 7313 8570 329 21 66 214 4063 47 84 210 7 1227 14 418 21 533 1053

19 1740 10 985 328 205 3042 77 115 1 851 842 224 3 1622 83 1342 678 249 3459 438 237 467 1413 2 4 5 27 9 79 112 5119 888 50 853 13 306 3855 4189 2341 263 28 62 211 2191 41 34 420 7 907 7681 11 744 886

26 2247 15 369 604 547 3490 178 72 4 1884 1529 475 11 2861 131 2108 1323 221 6346 841 257 1039 3650 5 25 9 57 9 151 160 10 590 2511 14 240 40 252 14 803 12 097 9533 1082 157 260 727 4853 225 77 392 30 3925 6169 24 994 2524

13 1784 6719 267 106 1283 72 51 1 831 618 200 4 714 78 1908 720 87 2481 323 105 546 1457 4 11 3 26 11 46 34 1687 935 16 165 11 202 2520 4004 4933 190 18 51 147 3440 44 67 613 6 735 6414 9576 524

5 1169 4379 41 63 1252 43 65 1 322 116 26 1 145 54 1126 325 15 1771 62 85 197 273 0 1 3 15 7 50 45 911 869 12 977 2 141 597 1760 1951 48 9 9 27 1912 8 56 337 1 161 4139 2510 90

21 650 9754 541 785 3136 236 183 3 683 1095 420 6 797 67 459 570 156 7989 495 390 442 2319 1 5 2 31 14 144 220 8001 2035 5792 73 162 8146 6422 3461 598 265 202 533 2356 178 31 690 32 2355 2159 4976 2075

9 700 5210 255 217 1196 80 102 0 338 545 171 1 973 43 503 348 140 4269 237 73 215 988 1 1 1 26 9 75 57 3770 425 14 882 12 81 3253 2576 616 244 31 84 297 411 83 10 795 23 1461 1651 2466 1137

15 436 3403 128 133 1300 26 57 0 204 443 149 1 769 37 462 241 81 3200 173 105 166 920 1 1 2 20 8 63 86 4013 1019 12 126 17 52 2037 1852 2141 140 25 65 179 624 49 7976 14 1058 1896 3830 699 (Continues on next page)

www.thelancet.com Vol 375 June 5, 2010

1979

Articles

All cause by age <5 years (Continued from previous page) Tunisia United Arab Emirates Yemen Europe Albania Andorra Armenia Austria Azerbaijan Belarus Belgium Bosnia and Herzegovina Bulgaria Croatia Cyprus Czech Republic Denmark Estonia Finland France Georgia Germany Greece Hungary Iceland Ireland Israel Italy Kazakhstan Kyrgyzstan Latvia Lithuania Luxembourg Macedonia Malta Moldova Monaco Montenegro Netherlands Norway Poland Portugal Romania Russia San Marino Serbia Slovakia Slovenia Spain Sweden Switzerland 650 3 1114 350 6175 1245 584 482 795 233 41 408 283 97 204 3090 1638 2943 374 675 14 345 655 2350 9661 4675 210 209 14 250 27 805 2 65 851 209 2644 378 2721 16 689 1 889 407 59 2150 340 342 165 1 649 203 3114 627 289 255 332 146 20 198 172 51 115 1680 1018 1686 191 371 6 189 318 1266 5066 2138 116 89 5 149 9 360 1 39 522 99 1560 187 1322 9119 0 565 176 39 1054 175 222 485 2 465 147 3060 618 295 227 463 88 21 210 111 46 89 1410 620 1257 183 304 8 156 337 1084 4595 2538 94 120 9 101 18 445 1 26 329 110 1084 190 1399 7570 1 324 231 20 1097 165 120 3490 461 56 834 1964 295 27 280 1527 166 29 554 027 days 159 months

AIDS

Diarrhoea

Pertussis

Tetanus

Measles

0 0 0 0 0 4 0 10 0 0 0 0 0 0 0 0 1 0 0 4 2 0 0 0 0 0 0 16 12 0 0 0 0 0 0 0 0 0 0 0 4 0 15 0 0 0 0 4 1 0

167 4 11 480 12 0 18 1 708 6 9 6 14 1 0 5 2 0 0 24 90 7 0 2 0 0 0 1 227 652 2 0 0 9 0 18 0 0 0 6 6 0 37 196 0 3 1 0 0 3 0

0 0 3407 0 0 0 0 36 0 2 0 0 0 0 0 0 0 0 2 0 0 0 0 0 1 5 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 1 0 0

17 0 1374 2 0 0 0 8 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 446 0 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

1980

www.thelancet.com Vol 375 June 5, 2010

Articles

Meningitis

Malaria

Pneumonia

Other infections*

Preterm birth complications

Birth asphyxia

Neonatal sepsis Congenital abnormalities

Other noncommunicable diseases

Injury

152 4 1124 7 0 65 7 172 25 20 3 15 1 0 11 3 2 2 42 26 50 3 16 1 10 8 20 203 162 1 4 0 0 1 13 0 0 14 4 20 5 23 279 0 9 5 0 50 5 1

0 0 263 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

363 23 10 251 117 0 187 3 1349 55 8 67 170 10 2 22 4 2 8 50 304 44 28 46 0 0 10 35 1672 1039 15 13 0 14 2 182 0 6 14 1 119 23 895 1184 0 53 45 1 37 8 3

263 26 3881 54 0 95 43 798 187 91 37 94 31 1 60 18 11 22 660 160 310 6 125 3 27 86 347 4434 643 22 17 1 4 1 40 0 2 116 26 158 16 366 2935 1 65 16 11 392 43 68

1062 145 9529 93 1 292 88 1329 120 83 117 137 47 10 56 95 9 40 408 438 965 130 165 1 53 122 552 101 859 14 24 4 111 6 35 0 17 150 30 800 84 329 2130 0 351 113 14 373 40 80

205 26 6631 43 0 114 32 609 76 46 48 73 17 2 52 26 10 12 361 210 199 16 31 1 16 38 165 67 544 52 15 2 25 2 54 0 7 99 35 186 39 98 1352 0 111 18 5 145 34 33

43 5 3001 4 0 22 1 281 25 13 6 16 11 0 10 0 4 6 84 89 47 0 4 1 5 10 62 0 186 3 9 0 6 0 46 0 1 53 6 108 5 11 448 0 11 3 7 90 11 10

714 134 2787 145 1 186 112 441 378 189 114 145 85 14 93 88 26 75 810 188 819 135 172 3 167 259 721 1412 311 63 74 3 63 8 248 1 18 242 59 914 122 574 3980 0 196 118 12 641 113 94

298 61 1218 104 1 82 38 279 161 68 52 87 15 8 48 34 10 26 442 71 320 23 76 3 47 88 340 865 130 20 18 2 11 3 57 0 9 111 24 183 51 160 1937 0 61 56 6 284 68 35

206 34 1442 68 0 49 25 152 211 57 32 44 15 4 50 12 20 13 208 55 181 32 37 0 18 28 108 663 132 17 36 2 8 4 113 0 4 53 18 148 31 229 2231 0 30 31 3 133 12 18 (Continues on next page)

www.thelancet.com Vol 375 June 5, 2010

1981

Articles

All cause by age <5 years (Continued from previous page) Tajikistan Turkey Turkmenistan UK Ukraine Uzbekistan Southeast Asia Bangladesh Bhutan Burma India Indonesia Maldives Nepal North Korea Sri Lanka Thailand Timor-Leste Western Pacic Australia Brunei Cambodia China Cook Islands Federated States of Micronesia Fiji Japan Kiribati Laos Malaysia Marshall Islands Mongolia Nauru New Zealand Niue Palau Papua New Guinea Philippines Samoa Singapore Solomon Islands South Korea Tonga Tuvalu Vanuatu Vietnam Total Worldwide 8 795 349 3 575 053 5 220 296 1407 51 32 349 369 289 7 109 313 3620 100 10 209 3519 47 2037 10 360 1 6 13 971 72 760 108 111 548 2409 56 8 228 20 836 801 26 11 260 205 710 3 42 149 1261 35 3441 1748 19 685 7 245 1 3 5306 32 788 47 48 221 987 25 3 94 12 982 606 25 21 089 163 578 3 67 164 2359 65 6768 1771 28 1352 3 116 0 3 8665 39 972 61 63 327 1422 31 5 134 7854 182 936 1199 123 562 1 829 826 173 036 163 36 822 18 246 6239 14035 3924 113 884 523 49 119 1 003 767 80 140 90 22 578 9373 3165 9971 1901 69 053 676 74 443 826 060 92 895 73 14 244 8873 3073 4064 2023 12 000 29 698 5352 4324 7259 21 200 4286 19 147 2339 2399 3794 10 892 7714 10 551 3014 1925 3465 10 308 027 days 159 months

AIDS

Diarrhoea

Pertussis

Tetanus

Measles

27 0 0 1 64 47 60 0 663 8585 635 0 279 14 6 273 0 0 0 107 1191 0 0 2 0 0 10 131 0 2 0 0 0 0 362 32 0 0 0 0 0 0 0 682 201 236

2304 389 691 4 59 2549 20 155 169 15 841 237 482 26 120 14 5416 2202 202 266 529 5 1 2306 11 420 0 5 17 46 17 758 35 4 92 0 1 0 0 728 4852 8 0 21 5 4 0 15 471 1 336 289

9 32 2 6 3 9 1731 7 948 86 182 284 0 764 232 6 15 7 0 0 12 401 0 0 0 0 0 45 9 0 1 0 1 0 0 63 77 0 0 1 0 0 0 0 44 195 465

23 234 11 0 0 27 2674 9 121 13 248 662 0 408 18 21 53 46 0 0 124 50 0 0 1 0 0 265 42 0 0 0 0 0 0 154 588 0 0 2 0 0 0 1 134 61 023

5 45 1 0 0 0 1622 0 409 81 275 545 0 74 0 33 15 214 0 0 172 467 0 0 0 0 0 316 9 0 0 0 0 0 0 286 216 0 0 3 0 0 0 3 376 117 623

The sum of data for individual causes, dierent age-groups, or individual countries does not match the total values in some cases because of rounding. The totals for individual causes dier from those in table 1 life (eg, congenital abnormalities, tetanus, sepsis). *Includes some deaths due to preterm birth complications, birth asphyxia, and other perinatal causes.

Table 2: Estimated numbers of deaths by cause in children younger than 5 years by WHO region and country

1982

www.thelancet.com Vol 375 June 5, 2010

Articles

Meningitis

Malaria

Pneumonia

Other infections*

Preterm birth complications

Birth asphyxia

Neonatal sepsis Congenital abnormalities

Other noncommunicable diseases

Injury

374 706 127 76 152 630 3912 44 1971 31 607 3031 4 785 634 33 37 71 13 1 1153 7443 0 5 2 15 2 436 13 1 74 0 7 0 0 658 2481 3 2 26 12 0 0 9 715 166 100

0 0 0 0 0 0 3244 0 2168 6292 1389 0 8 0 0 70 449 0 0 251 2 0 0 0 0 0 19 6 0 0 0 0 0 0 1018 40 0 0 34 0 0 0 5 28 732 049

2493 4152 1219 147 315 4554 25 978 291 16 293 371 605 38 331 26 5299 3649 647 1468 409 32 1 9097 62 229 1 32 41 225 24 2735 215 11 581 2 17 0 1 3135 17 351 21 14 141 86 9 1 53 2079 1 575 257

2034 2113 859 265 1147 2971 17 719 146 11 298 267 011 25 617 18 3011 2021 605 1215 316 240 7 5976 60 113 1 14 31 407 16 1923 275 7 422 1 42 0 1 1819 12 370 14 3 79 247 7 1 21 1893 1 062 071

1788 9014 950 1549 823 4660 30 547 167 17 288 248 993 33 258 42 6168 3854 1382 4175 485 316 7 3556 56 074 1 19 71 321 15 879 766 8 280 3 93 0 1 1588 13 883 22 28 98 630 11 1 45 5618 1 072 544

1101 3316 538 306 521 2142 32 375 129 13 175 191 607 17 516 18 6591 2129 511 1575 629 131 4 3236 62 117 0 8 27 178 8 1082 299 3 104 2 38 0 1 1593 5014 9 1 34 121 5 1 17 2142 828 800

612 552 311 39 173 956 29 464 87 10 803 121 395 8633 8 5304 1277 92 330 452 20 1 2443 7128 0 3 4 66 3 563 39 1 30 1 12 0 0 869 1269 2 3 9 85 1 0 3 366 522 576

429 5326 334 1109 2047 1600 6788 45 2757 54 823 9612 15 1440 1154 1155 2897 122 333 18 744 36 989 1 6 55 1400 7 371 911 4 134 1 89 0 1 471 6122 12 39 35 505 9 1 19 4025 375 897

440 2366 156 646 990 519 2374 56 2145 56 360 3857 10 467 612 541 1003 105 208 4 2058 24 198 1 13 40 577 4 515 471 5 217 0 26 0 1 802 5348 12 16 50 424 7 1 27 1296 254 529

361 1452 153 175 963 536 4293 49 27 682 53 362 3545 7 808 447 1005 643 90 109 7 1113 39 466 1 4 22 387 4 292 300 2 102 0 35 0 1 424 3118 6 6 16 293 4 1 9 966 293 890

because some of the other neonatal causes were split into individual causes in table 2. Additionally, some deaths from causes that are mainly associated with the neonatal period occurred after the rst month of

www.thelancet.com Vol 375 June 5, 2010

1983

Articles

provisional total of 81 000 deaths due to measles, pending further consideration of both approaches. The Million Death Study also assigned a substantially greater proportion of child deaths to malaria than did WHO estimates of total malaria cases and case fatality.18 Pending further analysis of data from the Million Death Study, we used WHO estimates of deaths from malaria in children younger than 5 years for 2008. We used the same method as for other countries to estimate deaths due to pertussis because of the diculties associated with use of verbal autopsy to dierentiate pertussis from other respiratory infections. Together with cause-specic inputs from WHO technical programmes and UNAIDS, the resulting cause-specic inputs for India were adjusted to t the estimated total deaths in neonates and children aged 159 months. Neonatal deaths in India were estimated to account for 54% of deaths in children younger than 5 years in 2008; this percentage was calculated by use of the method described above for estimation of the neonatal mortality rate with data from the Million Death Study for 200405, and three nationally representative surveys for 199899, 200203, and 200506. For China, causes of child deaths were based on estimates of cause fractions for deaths in neonates and children aged 159 months as previously described,27 adjusted to our estimates for total number of deaths in neonates and children aged 159 months in China in 2008. WHO technical programme estimates for deaths caused by malaria, tetanus, pertussis, and measles were used because these causes accounted for small proportions of child deaths and were not generally included as specic causes in published data from China.

receipt of feedback, or in some cases more recent data, the estimates were nalised.

Role of the funding source

The sponsor of the study had no role in the study design, data collection, data analysis, data interpretation, or the decision to submit for publication. All authors had complete access to data, and the corresponding author had nal responsibility for the decision to submit for publication.

Results
Of 8795 million child deaths that occurred in 2008,1 68% (5970 million) were caused by infectious diseases. The total numbers of deaths by cause are listed in tables 1 and 2 and in webappendix pp 1325, and the distribution of deaths by cause is shown in gure 4. The most important infectious diseases were pneumonia in neonates and older children, diarrhoea, and malaria. Deaths occurring in the neonatal period (aged 027 days) accounted for 41% (3575 million) of all deaths in children younger than 5 years. In this age-group, the greatest single causes of death were preterm birth complications and birth asphyxia, but collectively infectious causes were also important, especially sepsis and pneumonia. Distribution of deaths (gure 5) and their causes (gure 6) varied widely across the WHO regions. The largest numbers of deaths were in the African region (4199 million) and in the southeast Asian region (2390 million). These two regions had diering patterns of causes of death: a lower proportion of neonatal deaths occurred in the African region than in the southeast Asian region (29%, 1224 million vs 54%, 1295 million); and a higher proportion of deaths in Africa were due to malaria (16%, 0677 million) and AIDS (4%, 0181 million) than in southeast Asia, in which about 1% (0024 million) were due to these two causes combined. In the Americas, Europe, Asia, a high proportion of child deaths occurred during the neonatal period, ranging from 48% (0137 million/0284 million) in the Americas to 54% (1295 million/2390 million) in southeast Asia, with preterm birth complications and birth asphyxia as leading causes (gure 6). In countries with low neonatal mortality rates, congenital causes became proportionately more important. In all children younger than 5 years, the most important single causes of death were pneumonia, diarrhoea, and preterm birth complications (table 2, gure 4). Other important causes were birth asphyxia and malaria. 92% (0677 million) of deaths due to malaria, and 90% (0181 million) of deaths due to AIDS occurred in the African region. Successful vaccination programmes have reduced the worldwide total for deaths caused by measles and tetanus, but each of these diseases was still responsible for about 1% of deaths worldwide in 2008 (table 2). Numbers of deaths by cause in neonates, children aged 159 months, and all children younger than 5 years are
www.thelancet.com Vol 375 June 5, 2010

Estimation of uncertainty
Methods used to obtain uncertainty estimates for the multicause models are described elsewhere.16,17 Briey, we used a jackknife analysis to estimate the standard error of the models out-of-sample predictions. Monte Carlo simulations (1000 iterations) were used to perturb country-level estimates based on these standard errors. Uncertainty ranges (URs) for the simulations, dened as the 25975 centile, provide an indication of the uncertainty in the estimates. This approach to uncertainty estimation captures, to some extent, the misclassication of deaths by verbal autopsy and the variability across studies. However, it does not capture uncertainty related to the estimation of mortality in children younger than 5 years that is presented elsewhere,7,8 or that related to the analysis of vital registration data that comprises a very small proportion of the total deaths. The uncertainty for the estimation of deaths due to AIDS,10 malaria,18 measles,21 pertussis,20 and tetanus (Cousens S, personal communication) were derived from single-cause disease models. Figures 2 and 3, and webappendix pp 712 show the methods used for each country. The preliminary estimates were sent by WHO to all countries, and after
1984

Articles

provided for 193 countries in table 2 and webappendix pp 1325. 43% (274392 million/634176 million) of all children younger than 5 years worldwide reside in ve countriesIndia, Nigeria, Democratic Republic of the Congo, Pakistan, and Chinaand these countries accounted for 49% (4294 million) of all deaths in this age-group in 2008. These countries were responsible for high proportions of global totals for neonatal causes of death: birth asphyxia (53%, 0443 million), sepsis (52%, 0271 million), preterm birth complications (49%, 0521 million), and congenital abnormalities (43%, 0161 million). The highest proportions of deaths due to pneumonia occurred in India, Nigeria, Democratic Republic of the Congo, Pakistan, and Afghanistan, which collectively accounted for 52% (0826 million). 51% (0676 million) of deaths caused by diarrhoea occurred in India, Nigeria, Afghanistan, Pakistan, and Ethiopia. Deaths caused by malaria were concentrated in subSaharan Africa, with Nigeria, Democratic Republic of the Congo, Uganda, Sudan, and Tanzania accounting for 57% (0417 million) of deaths. Similarly, 51% (0103 million) of deaths due to AIDS occurred in South Africa, Nigeria, Mozambique, Tanzania, and Uganda. Injuries were shown to be important preventable causes of death in nearly all countries, with 32% (0093 million) of such deaths in India and China.

Discussion
Collectively, the most important causes of death in children younger than 5 years were infectious diseases, especially pneumonia, diarrhoea, and malaria. The most important single causes of death were pneumonia, diarrhoea, and preterm birth complications. Two-fths of deaths occurred in the neonatal period, during which the greatest single causes of death were preterm birth complications and birth asphyxia, but collectively, infectious diseases were also important. Numbers of deaths varied widely across WHO regions, with most deaths recorded in Africa and southeast Asia. Despite a continuing increase in the population of children younger than 5 years, the mortality rate is declining: 8795 million deaths occurred in 2008 versus 106 million per year during 200003.1,5 With greater declines in mortality in children aged 159 months than in neonates, the proportion of deaths in neonates has increased from 37% in 200003,5 to 41% (3575 million) of 8795 million deaths in children younger than 5 years in 2008. Thus, the main causes of death in the neonatal periodpreterm birth complications, birth asphyxia, and sepsis and pneumoniahave become even more important. In children aged 159 months, the two most important causes of death remain the same as in previous estimates, diarrhoea and pneumonia, but the percentage of deaths in children younger than 5 years attributable to each cause has reduced by 2025%.5 This reduction is partly due to the smaller proportion of deaths occurring in children aged 159 months, but also to new data
www.thelancet.com Vol 375 June 5, 2010

showing that the previous estimate of deaths due to diarrhoea in China was probably too high (12% vs new estimate of 31% [0011 million/0369 million]).27,28 Caution is essential with any comparison between previous and 2008 estimates of causes of death because additional data and changes in analytical methods could result in increased accuracy of estimates, but not a true indication of a time trend for certain diseases. The concentration of all-cause child deaths and deaths due to some specic causes, such as diarrhoea, pneumonia, malaria, and AIDS, in a small set of countries is striking. This result is partly related to the large populations of children younger than 5 years in these countries, but also some diseases are concentrated because of epidemiological and social conditions. Success in disease control eorts in these countries is essential if MDG 4 goals are to be achieved. However, nearly all countries still face the challenge to reduce child deaths from preventable conditions, irrespective of their number or cause. These national estimates of the causes of child death in 2008 should help to identify priority interventions for child survival, and how to allocate national and international resources. Undernutrition, including stunting, severe wasting, deciencies of vitamin A and zinc, and suboptimum breastfeeding, is not presented as a direct cause of death in these statistics, but has been found to be an underlying cause in a third of deaths in children younger than 5 years.29 In countries with adequate vital registration data, malnutrition was rarely listed as a cause of death, and verbal autopsy classication systems greatly underestimate the role of these nutritional deciencies. Thus, we chose to allocate the few deaths reported to be caused by malnutrition to major infectious diseases that often precipitate severe wasting. Successful implementation of interventions to prevent the development of undernutrition and micronutrient deciencies and to treat severe acute malnutrition would substantially reduce child mortality and improve the health and development of surviving children.4,30 These estimates are based on some advances in both data and methodology (panel). Vital registration data are available for additional countries and further verbal autopsy data could be included because new studies have been done, and a change to the methods could allow analysis of studies that were previously excluded. The use of national data for the worlds two largest countries (India and China) has been a major advance, but additional work is needed with data from these countries to fully represent the present state of major causes. The multinomial logistic regression model that is now used to estimate the causes of deaths in children aged 159 months, as was done previously for neonatal deaths,16 is a substantial improvement on the use of single-cause models. Singlecause models predicted collective death estimates that exceeded the yearly total number of deaths, so that adjustments needed to be made to t these estimates
1985

Articles

Panel: Major changes in estimation methods in 2008 compared with 2000035 New estimates of national mortality rate in children younger than 5 years and in neonates Multicause models led to increased datapoints from 102 to 148 National data used for India and China Multicause model used instead of single-cause models for age-group of 159 months (similar to previous multicause neonatal model) Adjustments in modelled estimates to account for recent scale-up of vaccination with Haemophilus inuenzae type b vaccine and distribution of insecticide-treated bednets Estimates of AIDS, neonatal pneumonia and sepsis, meningitis, pertussis, and non-communicable diseases added to previously presented causes Provision of uncertainty bounds for global numbers of child deaths from major causes

with study-specic and national covariates. These estimates include substantial uncertainty, but we believe that they are still useful for planning national health and nutrition eorts. CHERG will continue to work with WHO and UNICEF to identify data, improve estimation methods, and encourage use of resulting cause-specic child mortality numbers and rates in national and global health planning and priority setting. We anticipate updating these estimates every year to complement yearly updates in total deaths in children younger than 5 years. We challenge countries and programmes to advance the quality and consistency of data on causes of death, and, most importantly to use such data in the design of programmes to achieve maximum progress in the crucial few years before 2015.
Contributors REB coordinated the analyses and prepared all drafts of the report, incorporating comments by coauthors. All authors contributed to the conceptualisation, analysis, and completion of the estimates, and to editing of the report. CHERG provided assistance with new methods and critique of provisional estimates of the causes of child mortality. Child Health Epidemiology Reference Group of WHO and UNICEF Robert E Black (Chair), Jennifer Bryce, Laura Cauleld, Christa Fischer Walker, Hope Johnson, Henry Kalter, Joanne Katz, Li Liu, and Ne Walker (Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA); Diego Bassani and Prabhat Jha (University of Toronto, ON, Canada); Zulqar Bhutta (Aga Khan University, Karachi, Pakistan); Thomas Eisele (Tulane University School of Public Health, New Orleans, LA, USA); Harry Campbell, Igor Rudan, and Evropi Theodoratou (University of Edinburgh, Edinburgh, UK); Simon Cousens and Veronique Filippi (London School of Hygiene and Tropical Medicine, London, UK); Majid Ezzati (Harvard University, Boston, MA, USA); Claudio Lanata (Nutrition Research Institute, Lima, Peru); Joy Lawn (Saving Newborn Lives/Save the Children, Cape Town, South Africa); Herbert Peterson (University of North Carolina, Chapel Hill, NC, USA); and Richard Steketee (PATH Malaria Vaccine Initiative, Ferney Voltaire, France). Conicts of interest We declare that we have no conicts of interest. Acknowledgments The study was supported by WHO and UNICEF funding for meetings, and by a grant from the Bill & Melinda Gates Foundation to the US Fund for UNICEF for CHERG; the neonatal estimates were also supported by a grant from the Bill & Melinda Gates Foundation to Saving Newborn Lives/Save the Children; and the Global Fund to Fight AIDS, Tuberculosis and Malaria provided funding for validation of WHO estimates of country burdens. Throughout the development of the estimates, technical input was provided by WHO sta, including Rajiv Bahl, Ties Boerma, Thomas Cherian, Olivier Fontaine, Fiona Gore, Raymond Hutubessy, Jose Martines, Lulu Muhe, and Shamim Qazi, and by UNICEF sta, including David Brown, Mark Young, and Mickey Chopra. We thank Mikkel Oestergaard and Mie Inoue of WHO for collaboration on estimates of the total number of deaths in neonates and children younger than 5 years; Doris Ma Fat and Veronique Joseph of WHO for analysis of WHO death registration data; Cynthia Boschi Pinto of WHO and Nancy Binkin of UNICEF for coordinating the participation of their respective institutions; Carolyn Weidemann for coordinating the grant in support of CHERG from the Bill & Melinda Gates Foundation; Kit Chan and Yan Guo for collaboration on the estimates from China; Rajesh Kumar and Shally Awasthi for collaboration on the estimates from India; Alma Adler and Shefali Oza for their contributions to updating of the neonatal databases and early versions of the neonatal models; Martin Aryee for assistance with the diarrhoea mortality analyses; and Laura Lamberti for preparation of the gures.

within the total. Of the seven causes of death (including the group of other known causes) in the multinomial logistic regression model of deaths in children aged 159 months, ve were used directly for the estimates of 80% of the deaths. Malaria estimates were taken from other sources for 2008,18 but the high correlation (data not shown) between these estimates and the results of the multinomial logistic regression model provide encouragement that the multicause model can be used for malaria mortality in the future. The multinomial logistic regression model was used for estimation of 98% of neonatal deaths, the exception being tetanus. The major limitation for these estimates continues to be the scarcity of data on the causes of child death in the countries with the highest mortality. Ideally, recent data would be available for all countries through medically certied vital registration, but gures 2 and 3 show how few countries have adequate vital registration. Unfortunately data of sucient completeness from this source is available for only 76 countries, covering 4% (0344 million/8795 million) of deaths in children younger than 5 years in 2008. For other countries, various statistical models have to be used to estimate the cause distribution of deaths. The gap in evidence is most acute for sub-Saharan Africa, in which data based on verbal autopsy methods are available from few countries. The countries with high mortality and the least resources are those that most need the information to target disease control programmes and primary health-care services. Yet these countries are least likely to have adequate data of their owneg, only one country in sub-Saharan Africa has a vital registration system with reasonable completeness and a few others have nationally representative survey-based estimates of the causes of child mortality. Therefore, our estimates for many countries are derived from statistical modelling
1986

www.thelancet.com Vol 375 June 5, 2010

Articles

References 1 You D, Wardlaw T, Salama P, Jones G. Levels and trends in under-5 mortality, 19902008. Lancet 2010; 375: 10003. 2 Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS, The Bellagio Child Survival Study Group. How many child deaths can we prevent this year? Lancet 2003; 362: 6571. 3 Darmstadt GL, Bhutta ZA, Cousens S, Adam T, Walker N, de Bernis L, for the Lancet Neonatal Survival Steering Team. Evidence-based, cost-eective interventions: how many newborn babies can we save? Lancet 2005; 365: 97788. 4 Bhutta ZA, Ahmed T, Black RE, et al, for the Maternal and Child Undernutrition Study Group. What works? Interventions for maternal and child undernutrition and survival. Lancet 2008; 371: 41740. 5 Bryce J, Boschi-Pinto C, Shibuya K, Black RE, the WHO Child Health Epidemiology Reference Group. WHO estimates of the causes of death in children. Lancet 2005; 365: 114752. 6 Countdown Coverage Writing Group, on behalf of the Countdown to 2015 Core Group. Countdown to 2015 for maternal, newborn, and child survival: the 2008 report on tracking coverage of interventions. Lancet 2008; 371: 124758. 7 UNICEF, WHO, World Bank, UN Population Division. Levels and trends of child mortality in 2006: estimates developed by the Inter-agency Group for Child Mortality Estimation. http://www. childinfo.org/les/infant_child_mortality_2006.pdf (accessed Feb 10, 2010). 8 WHO. WHO mortality database: tables. http://www.who.int/ healthinfo/morttables/en/index.html (accessed Feb 10, 2010). 9 UNICEF. The state of the worlds children 2009. New York, NY: UNICEF, 2009. http://www.unicef.org/sowc (accessed Nov 4, 2009). 10 UN. 2008 report on the global AIDS epidemic. http://www.unaids. org/en/KnowledgeCentre/HIVData/GlobalReport/2008/2008_ Global_report.asp (accessed Dec 29, 2009). 11 UN Population Division. World population prospects: the 2008 revision. New York, NY: United Nations, 2009. 12 WHO. Neonatal and perinatal mortality: country, regional and global estimates. Geneva: World Health Organization, 2006. http:// www.who.int/entity/making_pregnancy_safer/publications/ neonatal.pdf (accessed March 3, 2010). 13 WHO. Input dataset for neonatal mortality estimates. http://www. who.int/healthinfo/statistics/MORT_2008_neonatal_empirical.xls (accessed May 5, 2010). 14 Committee on Obstetric Practice, American College of Obstetricians and Gynecologists. Inappropriate use of the terms fetal distress and birth asphyxia. Int J Gynaecol Obstet 1998; 61: 30910. 15 Lawn JE, Cousens S, Zupan J, for the Lancet Neonatal Survival Steering Team. 4 million neonatal deaths: When? Where? Why? Lancet 2005; 365: 891900. 16 Lawn JE, Wilczynska-Ketende K, Cousens SN. Estimating the causes of 4 million neonatal deaths in the year 2000. Int J Epidemiol 2006; 35: 70618.

17

18 19

20

21

22

23

24

25

26

27

28 29

30

Johnson HL, Liu L, Fischer-Walker C, Black RE. Estimating the distribution of causes of death among children age 159 months in high mortality countries with incomplete death certication. Int J Epidemiol (in press). WHO. World Malaria Report 2009. Geneva: World Health Organization, 2009. WHO. WHO/UNICEF estimates of national immunization coverage. http://www.who.int/immunization_monitoring/routine/ immunization_coverage/en/index4.html (accessed April 25, 2010). Crowcroft NS, Stein C, Duclos P, Birmingham M. How to best estimate the global burden of pertussis? Lancet Infect Dis 2003; 3: 41318. Wolfson LJ, Strebel PM, Gacic-Dobo M, Hoekstra EJ, McFarland JW, Hersh BS, for the Measles Initiative. Has the 2005 measles mortality reduction goal been achieved? A natural history modelling study. Lancet 2007; 369: 191200. WHO. WHO/UNICEF review of national immunization coverage, 19802008. Geneva: World Health Organization, 2009. http://www. who.int/immunization_monitoring/en/globalsummary/ wucoveragecountrylist.cfm (accessed April 25, 2010). OBrien KL, Wolfson LJ, Watt JP, et al, for the Hib and Pneumococcal Global Burden of Disease Study Team. Burden of disease caused by Streptococcus pneumoniae in children younger than 5 years: global estimates. Lancet 2009; 374: 893902. Watt JP, Wolfson LJ, OBrien KL, et al, for the Hib and Pneumocococcal Global Burden of Disease Study Team. Burden of disease caused by Haemophilus inuenzae type b in children younger than 5 years: global estimates. Lancet 2009; 374: 90311. Jha P, Gajalakshmi V, Gupta PC, et al, RGI-CGHR Prospective Study Collaborators. Prospective study of one million deaths in India: rationale, design, and validation results. PLoS Med 2006; 3: e18. Jha P, Jacob B, Gajalakshmi V, et al, RGI-CGHR Investigators. A nationally representative case-control study of smoking and death in India. N Engl J Med 2008; 358: 113747. Rudan I, Chan KY, Zhang JSF, et al, on behalf of WHO/UNICEFs Child Health Epidemiology Reference Group (CHERG). Causes of deaths in children younger than 5 years in China in 2008. Lancet 2010; 375: 108389. WHO. World health statistics 2008. Geneva: World Health Organization, 2008. Black RE, Allen LH, Bhutta ZA, et al, for the Maternal and Child Undernutrition Study Group. Maternal and child undernutrition: global and regional exposures and health consequences. Lancet 2008; 371: 24360. Victora CG, Adair L, Fall C, et al, for the Maternal and Child Undernutrition Study Group. Maternal and child undernutrition: consequences for adult health and human capital. Lancet 2008; 371: 34057.

www.thelancet.com Vol 375 June 5, 2010

1987