advanced search

<< Back To Home

CSU 67/2009: GLOBAL PATTERNS OF MORTALITY IN YOUNG PEOPLE

Wednesday, 28th of October 2009

CSU 67/2009: GLOBAL PATTERNS OF MORTALITY IN YOUNG PEOPLE
 
 Note to Readers
 
 Those readers interested in CSU 65/2009, on nonspecific effects of
 vaccines, may also be interested in a new methodological article at
 http://www3.interscience.wiley.com/cgi-bin/fulltext/122456888/HTMLSTART
 
 
 GLOBAL PATTERNS OF MORTALITY IN YOUNG PEOPLE
 
 Writing in The Lancet, Patton and colleagues review the evidence from
 developed and developing countries on mortality in young people. In the
 developing world, deaths among under-fives continue to outnumber those
 among adolescents and young adults. In many industrialized countries, the
 reverse is true. In all countries, deaths in adolescents occur after
 society has made large investments in education in the (disappointed)
 expectation that these will reap dividends when youth reach their
 economically productive years. This is especially unfortunate in developing
 countries, where public resources are even scarcer than in the
 industrialized world.
 
 Full text is at
 ttp://www.thelancet.com/journals/lancet/article/PIIS0140-6736
 (09)60741-8/fulltext
 
 Good reading.
 
 BD
 
 The Lancet, Volume 374, Issue 9693, Pages 881 - 892, 12 September 2009
 
 Global patterns of mortality in young people: a systematic analysis of
 population health data
 
 Original Text
 Prof George C Patton MD a b , Carolyn Coffey MSc a, Prof Susan M Sawyer MD
 a b, Russell M Viner PhD c, Dagmar M Haller PhD d, Krishna Bose PhD e, Prof
 Theo Vos PhD f, Jane Ferguson MSc e, Colin D Mathers PhD g
 
 Summary
 
 Background
 Pronounced changes in patterns of health take place in adolescence and
 young adulthood, but the effects on mortality patterns worldwide have not
 been reported. We analysed worldwide rates and patterns of mortality
 between early adolescence and young adulthood.
 
 Methods
 We obtained data from the 2004 Global Burden of Disease Study, and used
 all-cause mortality estimates developed for the 2006 World Health Report,
 with adjustments for revisions in death from HIV/AIDS and from war and
 natural disasters. Data for cause of death were derived from national vital
 registration when available; for other countries we used sample
 registration data, verbal autopsy, and disease surveillance data to model
 causes of death. Worldwide rates and patterns of mortality were
 investigated by WHO region, income status, and cause in age-groups of 10—14
 years, 15—19 years, and 20—24 years.
 
 Findings
 2·6 million deaths occurred in people aged 10—24 years in 2004. 2·56
 million (97%) of these deaths were in low-income and middle-income
 countries, and almost two thirds (1·67 million) were in sub-Saharan Africa
 and southeast Asia. Pronounced rises in mortality rates were recorded from
 early adolescence (10—14 years) to young adulthood (20—24 years), but
 reasons varied by region and sex. Maternal conditions were a leading cause
 of female deaths at 15%. HIV/AIDS and tuberculosis contributed to 11% of
 deaths. Traffic accidents were the largest cause and accounted for 14% of
 male and 5% of female deaths. Other prominent causes included violence (12%
 of male deaths) and suicide (6% of all deaths).
 
 Interpretation
 Present global priorities for adolescent health policy, which focus on
 HIV/AIDS and maternal mortality, are an important but insufficient response
 to prevent mortality in an age-group in which more than two in five deaths
 are due to intentional and unintentional injuries.
 
 Funding
 WHO and National Health and Medical Research Council.
 
 Introduction
 Adolescence is commonly regarded as a healthy time of life, with peaks in
 strength, speed, fitness, and many cognitive abilities. However, major
 shifts in health take place around puberty as new health risks with
 potentially life-threatening consequences become prominent.1—3 Reproductive
 maturity brings about risks for sexually transmitted diseases including
 HIV, and for women, particularly in low-income and middle-income countries,
 risks linked to pregnancy and child birth.4 Patterns of injury change with
 physical maturity, with young men in particular incurring trauma from war,
 violence, and traffic accidents. Puberty also indicates the onset of many
 mental health disorders of adulthood, which can be associated with a
 heightened risk of suicide.5 Consequently, health profiles change rapidly
 from early adolescence to young adulthood.
 
 Economic development typically brings about reductions in morbidity and
 mortality from communicable diseases, undernutrition, and maternal causes.6
 However, its effect on the health of young people might differ from that of
 older age-groups. For example, a long duration of education, with delays in
 marriage and parenthood, will probably reduce rates of early maternal
 death.7 Increased availability of motor vehicles is likely to raise the
 risk of traffic injuries.8 Furthermore, ease of access to psychoactive
 substances might heighten risks for mental health disorders and suicide.5
 The inexperience and ongoing neurodevelopment of adolescents might leave
 them vulnerable to some health risks associated with economic change.9
 No comprehensive studies of death in people aged 10—24 years have been
 done, even though this group consists of around 30% of the world's
 population.10 Reports have generally used country data to address overall
 mortality or specific causes of death.11—13 Some studies have compared
 mortality between countries14 or used longitudinal data,12 but none have
 taken a worldwide approach. Blum and Nelson-Nmari15 reported five leading
 causes of death in people aged 15—29 years (unintentional injuries,
 HIV/AIDS, other communicable diseases, violence, and suicide) and reported
 substantial regional variation in the number of deaths due to each cause.
 However, they were neither able to separate data into narrow age bands, nor
 extrapolate mortality rates to a regional or worldwide population.
 Similarly, studies of mortality in late childhood have grouped data for
 children aged 5—14 years, without scope for analysis of changes in patterns
 of death before and after puberty.16
 
 We examined whether changes in profiles of health during adolescent
 development is shown in shifts in mortality between early adolescence and
 young adulthood. We aimed to describe international rates and patterns of
 mortality between early adolescence and young adulthood, differences in
 youth mortality patterns between middle-income or low-income countries and
 that in high-income countries, and regional variations.
 
 Methods
 Data collection
 Data used in these analyses were derived from the Global Burden of Disease
 study17 for 2004, and were separated into 5-year groups up to age 24 years.
 We investigated changes in mortality in young adolescence (10—14 years),
 late adolescence (15—19 years), and young adulthood (20—24 years). WHO
 member states were classified into seven groups on the basis of income and
 region (panel). All countries with a gross national income per head of US
 $10 066 or higher, estimated by the World Bank, were defined as high-income
 countries, irrespective of region.18 This definition allowed some
 comparison of mortality patterns with economic development. Low-income or
 middle-income countries were grouped into regions according to WHO regional
 classifications, which are an indicator of an important administrative
 framework for international health policies (panel).
 
 Panel
 Countries grouped by WHO region and income per head, 2004
 High-income countries*
 Region of the Americas
 Bahamas, Canada, USA
 Mediterranean region
 Bahrain, Kuwait, Qatar, Saudi Arabia, United Arab Emirates
 European region
 Andorra, Austria, Belgium, Cyprus, Denmark, Finland, France, Germany,
 Greece, Iceland, Ireland, Israel, Italy, Luxembourg, Malta, Monaco,
 Netherlands, Norway, Portugal, San Marino, Slovenia, Spain, Sweden,
 Switzerland, UK
 Western Pacific region
 Australia, Brunei Darussalam, Japan, New Zealand, South Korea, Singapore
 
 Low-income and middle-income countries†
 African region
 Algeria, Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape
 Verde, Central African Republic, Chad, Comoros, Congo, Côte d'Ivoire,
 Democratic Republic of the Congo, Equatorial Guinea, Eritrea, Ethiopia,
 Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia,
 Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia,
 Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra
 Leone, South Africa, Swaziland, Togo, Uganda, Tanzania, Zambia, Zimbabwe
 
 Region of the Americas
 Antigua and Barbuda, Argentina, Barbados, Belize, Bolivia, Brazil, Chile,
 Colombia, Costa Rica, 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, Trinidad and Tobago, Uruguay, Venezuela
 
 Eastern Mediterranean region
 Afghanistan, Djibouti, Egypt, Iran, Iraq, Jordan, Lebanon, Libya, Morocco,
 Oman, Pakistan, Somalia, Sudan, Syria, Tunisia, Yemen
 
 European region
 Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria,
 Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan,
 Latvia, Lithuania, Poland, Moldova, Romania, Russian Federation, Serbia and
 Montenegro, Slovakia, Tajikistan, The former Yugoslav Republic of
 Macedonia, Turkey, Turkmenistan, Uzbekistan, Ukraine
 Southeast Asia region
 Bangladesh, Bhutan, North Korea, India, Indonesia, Maldives, Burma, Nepal,
 Sri Lanka, Thailand, Timor-Leste
 Western Pacific region
 Cambodia, China, Cook Islands, Fiji, Kiribati, Laos, Malaysia, Marshall
 Islands, Micronesia, Mongolia, Nauru, Niue, Palau, Papua New Guinea,
 Philippines, Samoa, Solomon Islands, Tonga, Tuvalu, Vanuatu, Vietnam
 Non-member states or territories
 American Samoa, Anguilla, Aruba, Bermuda, British Virgin Islands, Cayman
 Islands, Channel Islands, Faeroe Islands, Falkland Islands (Malvinas),
 French Guiana, French Polynesia, Gibraltar, Greenland, Guadeloupe, Guam,
 Holy See, Isle of Man, Liechtenstein, Martinique, Montserrat, Netherlands
 Antilles, New Caledonia, Northern Mariana Islands, West Bank and Gaza
 Strip, Pitcairn, Puerto Rico, Réunion, Saint Helena, Saint Pierre et
 Miquelon, Tokelau, Turks and Caicos Islands, Virgin Islands, Wallis and
 Futuna Islands, Western Sahara
 * WHO member states were classified as high income when their 2004 gross
 national income per head was US$10 066 or higher as estimated by the World
 Development Report 2004.18
 † Low-income and middle-income coutries include those with a 2004 gross
 national income per head lower than US$10 066.
 
 Estimates of all-cause mortality
 Life tables for the 192 WHO member states in 2004 were developed from
 death-registration data and sample-registration systems (India and China),
 and data for child and adult mortality from censuses and surveys, such as
 the Demographic and Health Surveys and UNICEF's Multiple Indicator Cluster
 Surveys. Age-specific and sex-specific death rates for countries were
 established with one of three standard approaches:19 routine life table
 methods for countries with complete vital registration, standard
 demographic methods to correct for under-registration of deaths, or model
 life tables when vital registration data were unavailable.20
 55 countries, 42 of which were in sub-Saharan Africa, had no information
 available. On the basis of predicted rates of early childhood mortality in
 2004, the most likely corresponding rates of adult mortality (excluding any
 HIV/AIDS deaths) were selected, along with uncertainty ranges, with use of
 regression models of child versus adult mortality from a set of almost 2000
 life tables judged to be of good quality.20
 For China, death rates recorded by age for children older than 5 years from
 the 2000 census were adjusted for an estimated under-reporting of 11·3% for
 boys and 18·1% for girls and predicted for 2004, with an assumed yearly
 rate of decrease in mortality of 1·5% on the basis of two intercensal
 periods: 1982—90 and 1990—2000.21 This projection takes into account other
 sources of data such as the Child Mortality Surveillance System.22 For
 India, all-cause mortality was derived from a time-series analysis of
 age-specific death rates from the sample registration system for 1990—2002,
 after correction for under-registration (88% completeness).23
 An early version of these life tables was published in the 2006 World
 Health Report.24 All-cause mortality rates for age and sex were further
 adjusted for the revised estimates of mortality from HIV/AIDS, conflict,
 and natural disasters.17 Total deaths by age and sex were estimated for
 each country by application of these rates to the estimated 2004 resident
 populations prepared by the United Nations Population Division in its 2006
 revision.
 
 Estimates of causes of death
 78 countries in 2004 had complete data (coverage of 85% or more, and most
 were coded to the third or fourth character of the International
 Statistical Classification of Disease and Related Health Problems version
 10 (ICD-10; table 1). We used vital registration data to estimate deaths by
 cause for 78 countries. However, usable death-registration data to
 establish cause of death were available for 112 countries, most of which
 were in the high-income group, and low-income and middle-income countries
 in Latin America and the Caribbean, Europe, and central Asia.
 
 Table 1Table image
 Data sources for cause-of-death estimation for 2004 Global Burden of
 Disease study17
 
 Statistics for cause of death are reported to WHO every year by country,
 year, cause, age, and sex. Deaths assigned to ICD codes for symptoms,
 signs, and ill-defined conditions, and some ill-defined codes in the
 cancer, cardiovascular disease, and injury chapters of ICD, were
 redistributed across defined causes. Percentages of deaths coded to these
 ill-defined causes varied from 4% in New Zealand to more than 30% in Sri
 Lanka and Thailand.19 When the latest available year for data was earlier
 than 2004, death-registration data from 1980 to the latest available year
 were used to project cause-specific trends for 2004. When distributions for
 very small countries were estimated, an average of data from the most
 recent 3 years was used to keep random variation to a minimum. For 34
 countries with less than 85% coverage, cause-of-death modelling was used to
 adjust proportions of deaths in the three broad groups of causes to account
 for likely biases in cause-of-death distribution.19 Cause-of-death
 modelling for calculation of cause-of-death composition used 1613
 country-years of historical death-registration data.
 
 For the 43 countries in the African region without useable
 death-registration data, regional patterns for broad-cause distribution
 were established on the basis of information that included the South
 African 2004 vital registration data, the Zimbabwe National Burden of
 Disease study 1997,25 INDEPTH26 (international network of field sites with
 continuous demographic evaluation of populations and their health) verbal
 autopsy data from seven sites in Africa for 1999—2002, data from
 Antananarivo in Madagascar for 1976—95, and Mozambique Maputo Central
 Hospital Mortuary data for 1993—2004. For broad disease groups I, II, and
 III, the proportional distributions of total deaths (excluding those from
 HIV/AIDS and war) for urban populations were calculated on the basis of
 averages from the Madagascar urban death-registration data, South African
 death-registration data for 2004, the Zimbabwean 1997 estimates, and
 cause-of-death modelling predictions for 2004. All African INDEPTH sites
 were rural, and thus these data were regarded as representative of rural
 populations. Distributions of rural deaths were based on averages from data
 from the INDEPTH sites, the distribution in South African rural provinces,
 and distributions predicted with cause-of-death modelling.
 To estimate deaths by cause for the 34 countries outside Africa without
 usable death-registration data, we used cause-of-death modelling to
 estimate proportions of deaths in broad disease groups (I—III) by age and
 sex. These models were based on all-cause mortality (excluding deaths from
 HIV/AIDS, war, and natural disasters), gross national income per head, and
 region.16 For countries without complete death-registration data, 21
 specific causes were further adjusted on the basis of epidemiological
 evidence from population-based studies, disease registers and notifications
 systems, and analyses by WHO programmes (more than 2700 datasets). These
 specific causes included HIV/AIDS, malaria, tuberculosis, cancers, drug
 dependence, war, and natural disasters. Almost a third of these datasets
 related to sub-Saharan Africa.
 
 Deaths were categorised with the framework of three broad-cause groups from
 previous burden-of-disease studies, but with maternal mortality as a
 distinct group to allow comparison between sex in group I causes of death.
 16 The four groups of death were: group IA, maternal mortality; group IB,
 communicable disease mortality including nutritional disorders; group II,
 non-communicable disease; and group III, injury including traffic
 accidents, fires, drowning, self-inflicted injury, and violence and war.
 Deaths assigned to ICD codes for symptoms, signs, and ill-defined
 conditions were proportionately redistributed to causes within groups I and
 II.
 
 Role of the funding source
 WHO funding allowed age-disaggregation of data. The corresponding author
 had full access to all the data in the study and had final responsibility
 for the decision to submit for publication.
 
 Results
 Of a worldwide population of 1·8 billion people aged 10—24 years, all-cause
 mortality was estimated at about 2·6 million deaths in 2004 (table 2). More
 than a third of deaths were in southeast Asia, with Africa having the next
 highest mortality (table 2). Relative risks (RR) for death in Africa were
 higher than in any other region, and nearly seven times higher than in
 high-income countries (table 2). Table 3 shows rates of all-cause mortality
 by the three age-groups for both sexes.
 
 Table 2Table image
 Estimated total number of deaths and all-cause mortality rates (per
 100 000) in people aged 10—24 years in 2004, classified by economic and
 geographic groupings
 
 Table 3Table image
 2004 all-cause mortality rates (per 100 000) stratified by country
 classification, age-group, and sex
 
 Relative risk of death in countries of low and middle income, compared with
 those of high income, was greatest in young adolescents, but the absolute
 difference was greatest in young adulthood (table 3). Males died at higher
 rates than did females in all age-groups and regions apart from Africa and
 southeast Asia (table 2). Risk in poor countries compared with high-income
 countries was higher for females than for males (RR 5·6 vs 2·8), largely
 because of the low rates for young women in high-income countries.
 
 Worldwide mortality rates were 2·4-fold higher in young adults than in
 young adolescents. High-income countries had the steepest relative rise in
 mortality from young adolescence to young adulthood (table 3). In
 low-income and middle-income countries, all-cause mortality in young adults
 was 2·4-times higher than in young adolescents (table 3). In the African
 region, all-cause mortality was high for all age-groups, with a female
 predominance most evident in the 15—19 year age-group. In low-income and
 middle-income countries in the American region, male mortality rose more
 than five-fold between early adolescence and young adulthood, and brought
 about a pronounced difference between sexes (RR 3·3; table 3).
 
 Low-income and middle-income countries in the eastern Mediterranean region
 had fairly high mortality in young adolescents, but did not have the
 pronounced rise that was noted for women aged 15—24 years in Africa and
 southeast Asia. In low-income and middle-income countries in the European
 region, mortality in young adolescents was low, but in males increased more
 than five times by young adulthood, with a resultant mortality gap compared
 with females. In the southeast Asian region, females died at greater rates
 than did males in all age-groups. Mortality in young adolescents varied
 from a high of around 22% in southeast Asia, Africa, and the eastern
 Mediterranean to a low of 13% in eastern Europe. Low-income and
 middle-income countries in the western Pacific region had the lowest
 regional mortality and an overall pattern similar to that of high-income
 countries (table 3).
 
 Causes of death were grouped into four broad categories. Figure 1 shows
 cause-specific mortality groups for high-income countries, low-income and
 middle-income countries by region, and the world. For worldwide deaths,
 group IA (maternal causes) and IB (communicable, perinatal, and nutritional
 causes) accounted for the highest proportion (48%) of mortality in young
 women aged 10—24 years. Overall group I deaths were almost 50% greater (RR
 1·47) in females than in males, but this excess was attributable to
 maternal deaths. Worldwide mortality for group IB in the 10—24 year
 age-group was very similar for females and males (1·01). Early adolescent
 rates of deaths for group II (non-communicable diseases; 21 per 100 000
 deaths per year) and group III (injuries; 26 per 100 000 per year) doubled
 by young adulthood (41 per 100 000 per year and 53 per 100 000,
 respectively).
 
 Figure 1 Full-size image (94K)
 Mortality rates (per 100 000) due to maternal, communicable,
 non-communicable, and injury causes
 Mortality rates shown by sex, country classification, and age-group: 10—14
 years (A), 15—19 years (B), and 20—24 years (C). Maternal=group IA.
 Communicable=group IB. Non-communicable=group II. Injury=group III. M=male
 deaths. F=female deaths. High income=high-income countries.
 LMICs=low-income and middle-income countries. AFR=African region.
 AMR=region of the Americas. EMR=eastern Mediterranean region. EUR=European
 region. SEAR=southeast Asia region. WPR=western Pacific region.
 
 For males, 28% of deaths were group I and 21% were group II. Worldwide,
 group III deaths were the most prominent and accounted for 51% of all male
 mortality. Group II deaths in males increased 2·6-fold (from 20 to 51 per
 100 000 per year) and group I 1·7-fold (from 39 to 63 per 100 000 per year)
 between early adolescence and young adulthood. An almost four-fold rise in
 male deaths in group III was reported between early adolescence and young
 adulthood (from 36 to 141 per 100 000 per year). Mortality rates in all
 categories were lower in high-income countries than in other countries, and
 the largest differences were in group I deaths (figure 1). Both group IA
 and IB deaths were uncommon in high-income countries, and accounted for
 only 4% of mortality.
 
 The African region had the highest rates for both sexes and had high rates
 in each category (figure 1). Of note were the large numbers of group I
 female deaths, which grew 2·6-fold (from 155 to 402 per 100 000) between
 early adolescence and young adulthood. In low-income and middle-income
 countries in the American region, the increase in male mortality between
 early adolescence and young adulthood was largely from an eight-fold rise
 in group III deaths, which accounted for 72% of all male deaths and 52% of
 all deaths in adolescents and young adults in the region. In the eastern
 Mediterranean region, mortality was moderately high for both sexes in each
 age-group, but a pronounced rise in group III male deaths was noted between
 early adolescence and young adulthood (figure 1).
 
 In low-income and middle-income countries in the European region, the male
 predominance of deaths was largely due to more than a seven-fold rise in
 group III deaths between early adolescence and young adulthood (from 24 to
 179 per 100 000), which accounted for 65% of male deaths and 48% of all
 deaths. In the southeast Asia region, group III causes were predominant
 (43%), with many deaths in groups I (36%) and II (21%). In western Pacific
 countries of low and middle incomes, 62% of male deaths and 42% of all
 deaths were due to group III causes.
 Figure 2 shows specific causes of group I deaths. Worldwide, maternal
 mortality accounted for 15% of all female deaths and 7% of all deaths in
 people aged 10—24 years. 11% of deaths were due to HIV/AIDS and
 tuberculosis, and rates of death from these diseases rose around five-fold
 between early adolescence and young adulthood (from 6·5 to 34 per 100 000
 per year). However, mortality from lower respiratory tract infections was
 lower in young adults than in young adolescents (figure 2). Figure 3 shows
 specific causes of group II deaths. Cardiovascular-related deaths were the
 largest subgroup within group II, accounting for 6% of all deaths in
 10—24-year-olds. 5% of all deaths were from cancer, and 5% were attributed
 to neuropsychiatric disorders and epilepsy.
 
 Figure 2 Full-size image (113K)
 Specific group I causes of death stratified by sex, age-group, and region
 Specific group I causes of death are tuberculosis, HIV/AIDS, acute lower
 respiratory infection (ALRI), and maternal deaths. M=male deaths. F=female
 deaths. (A) Data for world, high-income countries, and low-income and
 middle-income countries (LMICs). (B) LMICs grouped by WHO region.
 AFR=African region. AMR=region of the Americas. EMR=eastern Mediterranean
 region. EUR=European region. SEAR=southeast Asia region. WPR=western
 Pacific region.
 
 Figure 3 Full-size image (119K)
 Specific group II causes of death stratified by sex, age-group, and region
 Specific group II causes of death are cancer, cardiovascular disease (CVD),
 diabetes, epilepsy, and other neuropsychiatric causes. M=male deaths.
 F=female deaths. (A) Data for world, high-income countries, and low-income
 and middle-income countries (LMICs). (B) LMICs grouped by WHO region.
 AFR=African region. AMR=region of the Americas. EMR=eastern Mediterranean
 region. EUR=European region. SEAR=southeast Asia region. WPR=western
 Pacific region.
 Figure 4 shows specific causes of group III deaths. Traffic accidents were
 the largest contributor to group III in all age-groups and caused 14% of
 male and 5% of female deaths. 9% of young males died from violence and 6%
 of both sexes from suicide. Fire-related death was the third most common
 group III cause, accounting for 4% of female deaths. Drowning was common in
 both sexes and accounted for 5% of male and 2% of female deaths.
 
 Figure 4 Full-size image (117K)
 Specific group III causes of death stratified by sex, age-group, and region
 Specific group III causes of death are traffic accidents, fire-related
 deaths, drowning, self-inflicted injury, and violence. Violence refers to
 deaths from both violence in and outside of war. M=male deaths. F=female
 deaths. (A) Data for world, high-income countries, and low-income and
 middle-income countries (LMICs). (B) LMICs grouped by WHO region.
 AFR=African region. AMR=region of the Americas. EMR=eastern Mediterranean
 region. EUR=European region. SEAR=southeast Asia region. WPR=western
 Pacific region.
 
 In high-income countries, traffic accidents caused 32% of deaths in males
 aged 10—24 years. Violence and suicide accounted for 10% and 15% of male
 mortality, respectively. Young adults had higher rates of death due to
 traffic accidents (RR 7·7), suicide (16), and violence (18) than did young
 adolescents. In females, traffic accidents (27%) and suicide (12%) were the
 main causes of death. Traffic accidents, violence, and suicide accounted
 for almost 80% of group III deaths in both sexes and more than half of
 all-cause mortality.
 
 In low-income and middle-income countries, group I mortality was largely
 due to HIV/AIDS and tuberculosis, and in females, due to maternal causes.
 In group III, violence and traffic accidents were most prominent in males,
 and suicide and fire-related deaths in females. Cardiovascular-related
 mortality was the most common group II cause, with rates in low-income and
 middle-income countries 3·4-fold higher in males and 4·7-fold higher in
 females than in high-income countries. Mortality rates for epilepsy (RR
 4·3) and diabetes (3·4) were also higher in countries of low and middle
 income. In the African region, maternal mortality caused 26% of female
 deaths in those aged 10—24 years, with a clear rise in mortality with age (
 figure 2). Deaths due to HIV/AIDS and tuberculosis also increased with age,
 accounting for more than one in five of all deaths between early
 adolescence and young adulthood. The largest subgroup of group III causes
 in males was violence and war (16% of all deaths), followed by traffic
 accidents (8%), drowning (3%), and suicide (2%).
 In countries of low and middle income in the American region, high
 mortality rates for boys in late adolescence and young men were mainly due
 to violence. Such deaths were 26-fold higher in young men than in young
 adolescent males, and accounted for more than 42% of all deaths in those
 aged 15—24 years and for 9% of female deaths. Additionally, a rise of
 almost six-fold in traffic deaths and nine-fold in suicide was recorded for
 males between early adolescence and young adulthood.
 
 In low-income and middle-income countries in the eastern Mediterranean
 region, deaths due to violence and traffic accidents (17% and 16% of all
 male deaths, respectively) largely accounted for the rise in group III
 mortality with age. The main group III causes of female deaths were suicide
 (6%), traffic accidents (6%), fire-related (5%), and violence (4%).
 Maternal mortality contributed to the rise in group I deaths with age,
 accounting for 16% of all female deaths. For males, infectious diseases,
 especially tuberculosis, accounted for the rise in group I deaths with age.
 In European countries of low and middle income, the major group III causes
 of deaths in males aged 10—24 years were traffic accidents (17%), suicide
 (14%), and violence (9%), which largely contributed to the reported rise
 with age. In southeast Asian countries, male deaths in group I increased
 partly due to HIV/AIDS and tuberculosis, which constituted 10% of deaths in
 males aged 10—24 years. For females, the rise in group I deaths was largely
 the result of maternal mortality, with some contribution from HIV/AIDS and
 tuberculosis. For males, traffic accidents (9%), suicide (8%), and violence
 (4%) were the most common cause of group III deaths in this region. Fires
 (8%) and suicide (8%) contributed to the high group III mortality in
 females.
 
 In low-income and middle-income countries in the western Pacific, the main
 causes of group III deaths in males were traffic accidents (24%), drowning
 (14%), suicide (6%), and violence (6%). Increased mortality by age was
 mostly from traffic accidents (RR 9·2), which accounted for almost half of
 group III deaths. However, rates of drowning halved between early
 adolescence and young adulthood. For females, suicide (12%), road traffic
 accidents (11%), and drowning (11%) were the main group III causes of
 death. Cancer accounted for 9% and cardiovascular disease around 5% of all
 deaths.
 Table 4 shows the ten most common ICD-10 causes of mortality by age-group.
 Traffic accidents were the most frequent cause in all male age-groups and
 were the most common cause in young people overall. Suicide increased in
 both sexes in people aged 15—24 years, and was overall the second most
 common cause. Violence was common in males aged 15—24 years, and was the
 third most frequent cause overall.
 
 Table 4Table image
 Ten most common causes of death by sex and age-group
 Infectious diseases were frequent in both sexes in all age-groups. Deaths
 from lower respiratory tract infections were the most common cause in young
 adolescents, and although rates fell somewhat with age, it was the fourth
 most frequent cause of mortality overall. Rates of HIV/AIDS and
 tuberculosis increased substantially between early adolescence and young
 adulthood, and accounted for 17% of female and almost 14% of male deaths in
 young adulthood. Overall, tuberculosis and HIV/AIDS were ranked fifth and
 sixth, and accounted for more deaths than did traffic accidents. Maternal
 causes of death (maternal haemorrhage, sepsis, and abortion) were common in
 females aged 15—24 years.
 
 DiscussionFindings from this study have shown that only 3% of deaths were in
 high-income countries, even though 11% of young people live in these
 settings. Mortality rates in low-income and middle-incomes countries were
 almost four-fold higher than were those in high-income countries, a
 difference that was particularly pronounced for young women. Rates were
 highest in Africa and southeast Asia; these regions accounted for around
 two-thirds of worldwide deaths in young people, but constituted only 42% of
 the population. Female death rates were generally lower than were those for
 males, apart from in African and southeast Asian regions, where the
 increased number of female deaths was largely from maternal mortality, with
 deaths from suicide, fire, HIV/AIDS, and tuberculosis also contributing.
 High rates of death from injury in young women in southeast Asia and Africa
 contrasted greatly with low rates in low-income and middle-income countries
 of Central and South America, western Pacific, and eastern Europe.
 
 Major barriers exist to presentation of an adequate worldwide profile of
 mortality in any age-group.27 Less than a third of the world's populations
 have complete national registration data for cause-specific deaths.
 Countries with the lowest life expectancies have the greatest scarcity of
 data.16, 28 Sample registration systems provide a source of cause-specific
 mortality data for some countries with low and middle incomes.29 However,
 for many poor countries, including most of sub-Saharan Africa, verbal
 autopsy within recent surveys was the main source of data.29
 We included estimates of cause-specific deaths for regions with scarce
 death-registration data and wide uncertainty ranges, because exclusion of
 these regions would result in a potentially biased picture of worldwide
 mortality. In an early analysis of the 2001 Global Burden of Disease study,
 30 researchers estimated that uncertainty in all-cause mortality ranged
 from 1% for high-income countries to 15—20% for sub-Saharan Africa.
 Uncertainty ranges were generally large for deaths from specific diseases.
 Uncertainty for deaths from traffic accidents ranged from 3% for
 high-income countries to 25% for sub-Saharan Africa, and for stroke from
 10% for high-income countries to 30% for sub-Saharan Africa.24 In our
 analyses, uncertainty ranges are almost certainly larger than are those
 from the Global Burden of Disease study31 for all ages combined. Improved
 registration of deaths and development of alternative research methods for
 study of mortality in this age-group remains important.27 Since progress
 towards civil registration in many low-income countries will probably be
 slow, attainment of valid data for young people from these sources will be
 essential in the foreseeable future.32, 33
 
 Some caution is needed when regional mortality estimates are extrapolated
 to a country level. For example, in the western Pacific all-cause rates of
 death were low and cause-specific mortality was similar to that in
 high-income countries. However, these estimates are dominated by China—the
 most populous country in the region. Small countries in this region, such
 as Papua New Guinea, have patterns of death similar to those in regions of
 high mortality.34
 
 Regional differences in age of onset of sexual activity and availability
 and accessibility of condoms, contraception, safe abortion, antenatal and
 obstetric care, and HIV testing are some of the reasons for the rise in
 female mortality with age.4, 35 High maternal mortality rates in some
 low-income and middle-income countries have been previously described,36
 and are the target of the fifth Millennium Development Goal. Prominence of
 maternal death in this young age-group should be a further incentive for
 promotion of developmentally appropriate sexual and reproductive health
 services for young people.35 In view of the high number of deaths due to
 abortion, to ensure not only that contraception is available but also that
 all abortions are safe would do much towards reduction of mortality in
 young women.37 In regions without substantial maternal mortality, rates of
 female deaths generally remained low throughout adolescence and young
 adulthood.
 In males, injury deaths contributed most to the rise in mortality with age
 worldwide, but rates and patterns of causes varied substantially across
 regions. Injuries account for around 10% of worldwide mortality for all
 ages, but in people aged 10—24 years they accounted for more than 40% of
 all deaths and about half of male deaths.16, 38 The importance of injury
 deaths in young people in high-income countries has been recognised.12, 13
 However, injury mortality was most prominent in low-income and
 middle-income countries. This finding emphasises the need for heightened
 investment in injury-prevention programmes for this age-group.39 Diverse
 responses are needed because patterns of injury deaths varied substantially
 between regions. An increase in traffic-related deaths in many high-income
 countries up until the 1980s resulted in coordinated intersectoral policy
 responses, with a subsequent fall in mortality.9, 40 Typical preventive
 measures include investments in road infrastructure, compulsory seatbelts
 and helmets, and enforcement of legislation for use of alcohol and other
 drugs when driving.
 
 Further regional differences in priorities exist for prevention. In the
 western Pacific region, drowning caused most injury deaths in children aged
 10—14 years.41 Simple policy responses such as improved availability of
 swimming lessons in childhood and teaching about the hazards of water could
 prevent many deaths.42, 43 For deaths caused by violence, suicide, and
 accidental injury, policy responses will probably need to be complex and
 target a wide range of ages, but with particular benefits for this young
 age-group.44 In Central and South America, violence was the major cause of
 injury deaths in young men. Firearms are responsible for up to 97% of
 homicide deaths in countries in this region, making a strong case for
 strengthened gun control.45 However, without policies to address the
 illicit drug industry, few employment opportunities, urban segregation, and
 a culture of machismo, gun control alone is unlikely to be effective.46
 
 In eastern Europe, patterns of injury deaths in young men were complex,
 with homicide, traffic injuries, and suicide all common. The emergence of a
 large black or underground economy with a so-called criminological
 transition and exposure to violence might be one reason why rapid
 socioeconomic transition affects mortality in young men.47, 48 Furthermore,
 heightened alcohol misuse has been linked to changes in rates of suicide,
 accidental injury, and homicide, and could be a further target for
 prevention.49—51 In southeast Asia, the number of injury deaths in young
 women, particularly from fire-related death and suicide, was pronounced.
 Previous reports16, 52, 53 from India have attributed these deaths to
 suicide and accidents, but the role of violence from family members is
 important in many cases. Restricted access to pesticides in rural India
 will probably be an important element of suicide prevention.43, 54, 55
 
 Increases in mortality between early adolescence and young adulthood
 suggest major underlying shifts in health status that have so far attracted
 little attention from policy-makers. In some regions, including Africa and
 southeast Asia, group I causes of death rise in adolescence and young
 adulthood and thus are of major importance. Present global health policies
 for the prevention and management of HIV, and provision of access to
 information and services for sexual and reproductive health, will probably
 have major benefits.56 However, even in these regions, tuberculosis and
 lower respiratory tract infections cause more youth deaths than does
 HIV/AIDS, but have not yet attracted a similar response in policy.
 Importantly, group I causes of death were not prominent in most low-income
 and-middle-income countries. In the western Pacific and eastern Europe,
 although overall death rates were increased, patterns of death resembled
 those reported in high-income regions, suggesting a need for very different
 strategies.
 
 Contributors
 GCP conceived the idea for this report in consultation with WHO's authors
 (JF, KB, CDM) and developed the analytical plan with CC, CDM, DMH, SMS, and
 RMV. CDM was responsible for development of the dataset and advised about
 data analyses. CC and GCP undertook the data analysis in consultation with
 CDM and RMV. GCP prepared the initial draft in consultation with all
 authors. TV participated in drafting the report. All authors commented on
 subsequent drafts.
 
 Conflicts of interest
 RMV has undertaken paid consultation about adolescent health for WHO. All
 other authors declare that they have no conflicts of interest.
 
 Acknowledgments
 The age disaggregation of the Global Burden of Disease data was supported
 with funds from WHO. GCP is supported by a Senior Principal Research
 Fellowship from the National Health and Medical Research Council. We thank
 Professor John Carlin for his comments on an early version of the analysis.
 
 References
 1 Kleinert S. Adolescent health: an opportunity not to be missed. Lancet
 2007; 369: 1057-1058. Full Text | PDF(44KB) | CrossRef | PubMed
 2 Resnick MD, Bearman PS, Blum RW, et al. Protecting adolescents from harm:
 findings from the National Longitudinal Study on Adolescent Health. JAMA
 1997; 278: 823-832. PubMed
 3 Patton GC, Viner R. Pubertal transitions in health. Lancet 2007; 369:
 1130-1209. Summary | Full Text | PDF(153KB) | CrossRef | PubMed
 4 Ronsmans C, Graham WJ. Maternal mortality: who, when, where and why.
 Lancet 2006; 368: 1189-1200. Summary | Full Text | PDF(263KB) | CrossRef |
 PubMed
 5 Patel V, Flisher AJ, Hetrick SE, McGorry PD. Mental health of young
 people: a global public-health challenge. Lancet 2007; 369: 1302-1313.
 Summary | Full Text | PDF(132KB) | CrossRef | PubMed
 6 Beaglehole R, Bonita R. Public health at the crossroads. Cambridge:
 Cambridge University Press, 2004.
 7 Gluckman PD, Hanson MA. Evolution, development and timing of puberty.
 Trends Endocrinol Metab 2006; 17: 7-12. CrossRef | PubMed
 8 Kopits E, Cropper M. Traffic fatalities and economic growth. Accid Anal
 Prev 2005; 37: 169-178. CrossRef | PubMed
 9 Krug EG, Sharma GK, Lozano R. The global burden of injuries. Am J Public
 Health 2000; 90: 523-526. CrossRef | PubMed
 10 Goodburn EA, Ross DA. Young people's health in developing countries: a
 neglected problem and opportunity. Health Policy Plan 2008; 15: 137-144.
 CrossRef | PubMed
 11 Pan SY, Desmeules M, Morrison H, et al. Adolescent injury deaths and
 hospitalization in Canada: magnitude and temporal trends (1979—2003). J
 Adolesc Health 2007; 41: 84-92. CrossRef | PubMed
 12 Singh GK, Yu SM. Trends and differential in adolescent and young adult
 mortality in the United States, 1950 through 1993. Am J Public Health
 Nations Health 1996; 86: 560-564. PubMed
 13 Schlueter V, Narring F, Munch U, Michaud PA. Trends in violent deaths
 among young people 10—24 years olds, in Switzerland, 1969—1997. Eur J
 Epidemiol 2008; 19: 291-297. CrossRef | PubMed
 14 Borrell C, Pasarin MI, Cirera E, Klutke P, Pipitone E, Plasencia A.
 Trends in young adult mortality in three European cities: Barcelona,
 Bologna and Munich. 1986—1995. J Epidemiol Community Health 2001; 55:
 577-582. CrossRef | PubMed
 15 Blum RW, Nelson-Nmari K. The health of young people in a global context.
 J Adolesc Health 2004; 35: 402-418. CrossRef | PubMed