Over the last 4 decades, childhood cancer mortality declined in most developed areas of the world. However, scant information is available from middle-income and developing countries. The authors analyzed and compared patterns in childhood cancer mortality in 24 developed and middle-income countries in America, Asia, and Oceania between 1970 and 2007.
Childhood age-standardized annual mortality rates were derived from the World Health Organization (WHO) database for all neoplasms, bone and kidney cancer, non-Hodgkin lymphoma (NHL), and leukemias.
Since 1970, rates for all childhood cancers dropped from approximately 8 per 100,000 boys to 3 per 100,000 boys and from 6 per 100,000 girls to 2 per 100,000 girls in North America and Japan. Latin American countries registered rates of approximately 5 per 100,000 boys and 4 per 100,000 girls for 2005 through 2007, similar to the rates registered in more developed areas in the early 1980s. Similar patterns were observed for leukemias, for which the mortality rates were 0.81 per 100,000 boys and 0.55 per 100,000 girls in North America, 0.86 per 100,000 boys and 0.68 per 100,000 girls in Japan, and 1.98 per 100,000 boys and 1.65 per 100,000 girls in Latin America for 2005 through 2007. Bone cancer rates for 2005 through 2007 were approximately 2-fold higher in Argentina than in the United States. During the same period, Mexico registered the highest rate for kidney cancer and Colombia registered the highest rate for NHL, whereas the lowest rates were registered by Japan for kidney and by Japan and the United States for NHL.
Patterns and trends in childhood cancer have been updated up to 2007 for 30 European countries.1 Mortality from all neoplasms steadily declined in the European Union (EU), reaching rates of approximately 3 per 100,000 children for 2005 through 2007. However, within Europe, appreciable differences in trends still persist, and the mortality rates in eastern and southern European countries for the middle 2000s were similar to those in western and northern European countries for the early 1990s. Thus, differences in the adoption and impact of therapeutic achievements for childhood cancers across Europe have persisted.2, 3
Scant information is available from other areas of the world and particularly from middle-income and developing countries. From the 1960s to the end of the 1980s, a 50% decline in childhood cancer mortality was observed in the United States and Canada, and substantial declines also were observed in other developed countries, such as Australia, Israel, and Japan (reaching rates between 3.5 and 4.5 per 100,000 in the late 1980s).4 Over the same period, the pattern was less favorable for other areas of the world, including countries of Latin America (Uruguay, Cuba, Argentina, Costa Rica), Kuwait, New Zealand, and Singapore, which registered the highest childhood cancer mortality rates in the late 1980s (6 to 7.5 per 100,000 boys and 5 to 6 per 100,000 girls).4 In the same countries, an excess in mortality from childhood leukemias also was observed. The smaller and later decline in childhood cancer mortality in several middle- and low-income and developing countries reflects a delay in the adoption of and a lack of accessibility to newer therapies in those countries.1, 5 The objectives of the current study were to monitor recent patterns in childhood cancer mortality in various areas of the world, to update trends up to 2007 in 24 countries, and to provide an overview of trends for all childhood cancers and leukemias since 1970 using joinpoint regression analysis.6
MATERIALS AND METHODS
We derived death certification data for childhood cancers and estimates of the resident population at ages 0 to 14 years (further subdivided in 4 age groups: 0 [newborn], ages 1-4 years, ages 5-9 years, and ages 10-14 years) for the period from 1970 through 2007 using the World Health Organization (WHO) database.7 This included reliable information for leukemias and all childhood neoplasms in Canada and the United States, in 11 Latin American countries (Argentina, Brazil, Chile, Colombia, Costa Rica, Cuba, Ecuador, Mexico, Puerto Rico, Uruguay, and Venezuela), in 9 Asian countries (Hong Kong Special Administrative Region [SAR] of the People's Republic of China, Israel, Japan, Kuwait, Republic of Korea, Singapore, Azerbaijan, Kazakhstan, and Uzbekistan), and in Australia, and New Zealand. For the other childhood cancer sites that we considered (ie, bone and articular cartilage, kidney and other urinary sites [predominantly Wilms tumor] and non-Hodgkin lymphoma [NHL]), reliable mortality data were available for 8 countries only, ie, Canada, the United States, Argentina, Brazil, Colombia, Mexico, Japan, and the Republic of Korea. Cuba and Uzbekistan reported reliable data only for NHL, and Kazakhstan reported reliable data only for bone and articular cartilage cancer.
Data were available from 1970 up to 2007, except for Brazil (since 1979); Azerbaijan, Kazakhstan, and Uzbekistan (since 1981); and the Republic of Korea (since 1985). In Table 1, the last year with available data is indicated for each country. In a few countries, data were missing for 1 or more calendar years. No estimations were made for missing data.
Table 1. Age-Standardized 0–14 Years (World Standard Population) Mortality Rates From All Childhood Cancers and Leukemias Per 100,000 Boys and Girls in 24 Selected American, Asian, and Oceanian Countries in 1990 Through 1994, 2000 Through 2004, 2005 Through 2007, and Average Annual Number of Certified Deaths for the Most Recent Period
Standardized Mortality Rate
No. of Deaths
Standardized Mortality Rate
No. of Deaths
SAR indicates Special Administrative Region.
When data were not available for the whole period, the years for which they were available are indicated in parentheses.
Estimated coverage of deaths/causes of deaths <90%.
During the calendar period that we considered, 3 different revisions of the International Classification of Diseases (ICD) were used.8-10 For most countries, there were no major changes in the classification or coding of the cancers between various ICD revisions. We recoded classification of cancer deaths—for all calendar periods and countries—according to the 10th revision of the ICD. It was not possible to obtain reliable death certification data for neoplasms of the nervous system because of difficulties in histopathologic classification and changes in the classification of neuroblastoma, which is coded in part to the organ affected (chiefly, the adrenal gland; ie, with cancers of endocrine organs), in part to connective and soft-tissue sarcomas, and in part to the nervous system.3 We did not consider mortality from other childhood cancers (ie, eye [retinoblastoma] and Hodgkin lymphomas) because of the limited number of deaths in most countries.
Estimates of the resident populations for the corresponding calendar periods, based on official censuses, were extracted from the same WHO database.7 For a few American countries, data were unavailable for selected years in the WHO database; for 1995, 2000, and 2005, data were extracted from Pan American Health Organization publications11, 12; and, for the other missing years, data were estimated by interpolation.
From the matrices of certified deaths and resident population, we computed age-standardized mortality rates at ages 0 to 14 years per 100,000 boys and per 100,000 girls using the direct method on the basis of the world standard population.13 We also computed age-standardized rates for 2 areas on the American continent (North America, including Canada, the United States, and Puerto Rico; and Latin America, including all other American countries that were considered in our analysis). Puerto Rico was pooled with Canada and the United States because of similarities with those countries.
To identify significant changes in the trends for all childhood neoplasms and leukemias, we performed joinpoint regression analyses using the software provided by the Surveillance Research Program of the US National Cancer Institute.14 The objective of this analysis was to identify possible points when a significant change in the linear slope of the trend (on a log-scale) was detected over the study period.6 In joinpoint analysis, the best fitting points, called “joinpoints,” are chosen at which the rate changes significantly. The analysis starts with the minimum number of joinpoints (eg, 0 joinpoints; namely, a straight line) and tests whether 1 or more joinpoints (up to 4) are significant and must be added to the model. Each significant joinpoint that indicates a change in the slope (if any) is retained in the final model. To describe linear trends by period, the estimated annual percent change (APC) is then computed for each of those trends by fitting a regression line to the natural logarithm of the rates using calendar year as a regressor variable. The average APC (AAPC), based on an underlying joinpoint model, also was calculated: It was estimated as the geometric weighted average of the APC with the weights equal to the lengths of each time interval segment.15 This summary measure simplifies the trends comparison between sexes and countries with the same range of years.
Table 1 shows the age-standardized (ages 0-14 years) mortality rates from all childhood cancers and leukemias per 100,000 boys and per 100,000 girls in 24 countries of the Americas, Asia, and Oceania during the periods from 1990 to 1994, from 2000 to 2004, and from 2005 to 2007, and the average annual number of certified deaths for the most recent period. The histograms of the age-standardized (ages 0-14 years) mortality rates for each individual country considered in the period 2005–2007 are presented in Figure 1. In the interpretation and discussion of these results, due attention must be paid to random variation, mainly for smaller countries and rare cancers, and to problems of reliability of death certification, particularly for some countries in Latin America and the former Soviet Union. With these cautions in mind, during 1990 to 1994, mortality rates from all childhood neoplasms varied by a factor of approximately 2 between the highest rates (in Kazakhstan; approximately 7 per 100,000 boys and 6 per 100,000 girls) and the lowest rates (approximately 3-4 per 100,000 boys and 2.5-3.5 per 100,000 girls) in the most developed countries, ie, North America, Hong Kong SAR, Japan, Australia, Israel, and Puerto Rico. In most countries that we considered, mortality from all childhood neoplasms declined over the last 15 years, even with some important exceptions (Brazil, Ecuador, and Mexico), for which mortality rates increased until the most recent years for both sexes. For 2005 to 2007, the variation between the highest rate and the lowest rate remained similar to the variation for 1990 to 1994 (by a factor of approximately 3), but the highest rates were lower than those for 1990 to 1994 (approximately 6 per 100,000 boys and 5 per 100,000 girls), and the lowest rates were approximately 2 per 100,000 for both sexes. Whereas the pattern of countries with the lowest rates remained unchanged over the last 15 years, the highest rates in 2005 to 2007 were registered for Mexico, Venezuela, and Ecuador for both sexes.
Similar geographic patterns and temporal trends were observed for leukemias, which accounted for approximately 33% of all childhood cancer mortality. For 1990 to 1994, among boys, the highest mortality rates were reported in Azerbaijan, Singapore, Kazakhstan, and Venezuela (approximately 3 per 100,000 boys). Among girls, the highest rates were reported for Asian and Latin American countries and ranged from 1.69 per 100,000 girls in Argentina to 2.38 per 100,000 girls in Colombia; whereas the lowest rates were registered in most developed countries, ie, North America, Hong Kong SAR, Japan, Australia, Israel, and Puerto Rico (approximately 1 per 100,000 girls). For 2005 to 2007, the geographic distribution of mortality from leukemia remained similar to that for 1990 to 1994, but the most developed countries registered rates lower than 1 per 100,000 boys and 0.8 per 100,000 girls.
Table 2 gives the age-standardized (ages 0-14 years) mortality rates from bone cancers, kidney cancers, and NHL per 100,000 boys and per 100,000 girls in 8 larger countries (>30 millions inhabitants) in the periods 1990 to 1994, 2000 to 2004, and 2005 to 2007, and the average annual number of certified deaths for the most recent period. For the other countries, the number of deaths from bone and kidney cancers and from NHL were very few; thus, it was not possible to calculate reliable mortality rates. Again, the highest rates were registered in less developed countries. For bone cancers during 2005 to 2007, the mortality rates were approximately 2-fold higher in Argentina (0.21 per 100,000 boys and 0.24 per 100,000 girls) than in the United States (0.13 per 100,000 for both sexes). Over the same period, mortality rates from kidney cancer and from NHL in boys varied by a factor of approximately 3 between the highest rates in Mexico (0.16 per 100,000 boys) for kidney cancers and in Colombia for NHL (0.37 per 100,000 boys) and the lowest rates in Japan and the United States (0.05-0.09 per 100,000 boys for kidney cancers and 0.13 per 100,000 boys for NHL). Similar patterns were observed in girls.
Table 2. Age-Standardized 0–14 Years (World Standard Population) Mortality Rates From Selected Childhood Cancers Per 100,000 Boys and Girls in 8 Selected Countries in 1990 Through 1994, 2000 Through 2004, 2005 Through 2007, and Average Annual Number of Certified Deaths for the Most Recent Period
Standardized Mortality Rate
No. of Deaths
Standardized Mortality Rate
No. of Deaths
When data were not available for the whole period 2005–2007, the years for which they were available are indicated in parentheses.
Estimated coverage of deaths/causes of deaths <90%.
Figure 2 shows the trends in age-standardized (ages 0-14 years) mortality rates from all childhood cancers and from leukemias for 1985 to 2007 in Latin America, the larger middle-income geographic area that was considered in the current analysis; and in North America and Japan, the 2 major developed areas that were considered. Although mortality rates from all childhood neoplasms steadily declined in North America and Japan over the calendar period considered, no changes were evident in Latin America. Mortality rates for all childhood cancers dropped from about 8 to 3 per 100,000 boys and from 6 to 2 per 100,000 girls in North America and Japan; whereas, in more recent years, Latin American countries registered rates of about 5 per 100,000 boys and 4 per 100,000 girls, similar to the rates registered in more developed areas during the early 1980s. Similar patterns were observed for leukemias (Fig. 2), with mortality rates of 0.81 per 100,000 boys and 0.55 per 100,000 girls in North America, 0.86 per 100,000 boys and 0.68 per 100,000 girls in Japan, and 1.98 per 100,000 boys and 1.65 per 100,000 girls in Latin America for 2005 to 2007.
Figure 3 and Table 3 show the findings from the joinpoint analysis for the age-standardized (ages 0-14 years) mortality rates for all childhood cancers and leukemias in 12 selected countries over the period from 1970 to 2007. Mortality from all childhood cancers declined steadily in Canada, the United States, Australia, and Japan over the entire period, with an AAPC of approximately −3% for both sexes. The declines were smaller in Argentina, Chile, Cuba, and Korea (AAPC, approximately −2%). The picture was less favorable in the largest Latin America countries, ie, Brazil, Colombia, and Venezuela, for which AAPCs were between −0.3% and −0.5%, and was unfavorable for Mexico, where the AAPC was 0.8 in girls and 1% in boys. For leukemias, the declines were >4% per year in Canada, the United States, Japan, and Korea; and approximately 2% to 3% in Argentina, Chile, and Cuba. Like for all cancers, the declines were less marked in Brazil, Colombia, and Venezuela (≤−1%), and an upward trend was registered in Mexico (AAPC = 1.5% in boys and 1% in girls).
Table 3. Joinpoint Analysis for All Childhood Cancer Mortality and Leukemias in Boys and Girls From Selected American, Asian, and Oceanian Countries: 1970 Through 2007
The current update of geographic patterns and temporal trends in childhood cancer mortality in various countries worldwide demonstrates that, over the last 35 years, the decline in mortality for all cancers has been >60% in the United States, Canada, and Puerto Rico, corresponding to the prevention of approximately 2800 deaths per year. These declines in mortality were caused by the improved management of childhood cancers, including the adoption of effective, multidrug chemotherapy protocols; supportive measures to overcome toxicity; advancements in radiotherapy and bone marrow transplantation; and improved diagnostic techniques, even in the absence of any single breakthrough. Over the period from 1990 to 2004, in the United States, the 10-year relative survival rate for childhood acute lymphoblastic leukemia increased from 73% to 84%, the 10-year relative survival rate for other leukemias increased from 39% to 59%, and the 10-year relative survival rate for NHL increased from 73% to 87%. The survival rate for patients with HL was >95%.16 In contrast, for childhood bone and kidney cancers, there was little evidence of progress in mortality over the last decade, even in most developed areas of the world. This was caused at least in part by the impact of conservative treatment17 and the lack of substantial changes in chemotherapy over the period.18
In the United States during the late 1990s, the 5-year relative survival rate for all races and both sexes combined from osteosarcoma and Ewing sarcoma still was approximately 65%, whereas the rate for kidney cancer was >90% and did not change appreciably.19 However, the number of deaths from these neoplasms represent a limited proportion of all childhood cancers, ie, approximately 2% (for kidney cancer) to 5% (for bone cancer). Comparable declines in childhood cancer mortality were observed in other developed areas, such as Australia, Israel, Japan, and New Zealand. However, the pattern was less favorable in Latin America.
The inconsistent cancer trends observed in various Latin American countries, including the largest ones (Mexico, Colombia, Brazil, and Venezuela), probably reflect improved diagnosis and registration of childhood cancers over the last few decades,20-22 which would lead to increasing rates, along with the impact of treatment, which would lead to declining rates. Childhood cancer mortality in those countries in the early 1980s, when therapies had still a limited impact on mortality, was appreciably lower than in North America, indicating the existence of substantial under registration. In such a situation, the real trends in mortality are not interpretable for those countries.
Brazil and Colombia also have problems of population coverage, and this introduces additional uncertainties regarding childhood cancer mortality and trends.23 Other Latin American countries (ie, Argentina and Chile) that have a longer tradition of reliable population estimates and death certification have favorable childhood cancer trends, although to a lesser extent than in more developed countries of the world.1
Except for a few high-resource countries, Asia is represented very little. In addition, in the current analysis, data were considered for a few selected republics of the former Soviet Union (Azerbaijan, Kazakhstan, and Uzbekistan). Azerbaijan and Kazakhstan had some decline in mortality mainly from childhood leukemias over recent years, but their rates remained higher than those in more developed countries. Therefore, despite potential problems in death certification quality and validity,7, 24, 25 like in Latin America, there is ample scope for further improvement in the management of children with cancer in those areas.
Recent trends and rates in childhood cancer mortality in North America (and Puerto Rico), Japan, and Oceania were similar and were even more favorable than in the EU,1, 26 including several of the largest countries in western Europe (Italy, Spain, and the United Kingdom). Substantial declines in childhood cancer mortality in the United States were observed despite some suggestions of a moderate increase in incidence (0.6% per year during 1975 to 2002), which may have been caused at least in part by improved diagnostic accuracy.27 In any case, because there is no consistent indication of an appreciable decline in incidence,19, 28 the key reason for the decline in mortality is improved treatment. Therefore, modern, effective treatment for childhood cancers and leukemias continued to be adopted earlier and were more widespread in North America and Japan than in several other developed areas of the world, including a few major western European countries.5
In developed areas, >70% of childhood cancers are cured.29 However, despite some progress,30 the cure rate remains considerably lower in Latin America, where childhood cancers accounted for >6000 deaths in the 10 major countries in 2005, and the current rates are over twice those of North America. Thus, at least 3000 cancer deaths in children aged <15 years would be avoidable through the widespread adoption of adequate, modern treatment in those countries. Mexico, the country that had the highest rates in 2005,31 did not have a single pediatric oncologist up to the year 2000 in 13 of its 32 states.32
Despite the favorable pattern in developed areas reported over the last 3 decades, childhood cancers continue to represent an important cause of death worldwide. In Australia, for 2004 to 2006, cancer was the second leading cause of death among children ages 1 to 14 years28 and the fourth leading cause in Japan among children aged 0 to 14 years,26 confirming the key role of childhood cancer care and the importance of advancements in its management worldwide. Improvements in the adoption of current integrated treatment protocols in Latin American and other lower- and middle-income countries worldwide would prevent a substantial proportion of childhood cancer deaths.
We thank Mrs. I. Garimoldi for editorial assistance.
CONFLICT OF INTEREST DISCLOSURES
This work was conducted with contributions from the Italian and Swiss Leagues Against Cancer, the Swiss Foundation for Research Against Cancer, and the Italian Association for Cancer Research. Dr. Bertuccio was supported by a fellowship from the Italian Foundation for Cancer Research.