Cancer Incidence and Mortality among Children and Adolescents in the United States, 2001-2003 Jun Li, MD MPH Epidemic Intelligence Service Officer Centers for Disease Control and Prevention Division of Cancer Prevention and Control Epidemiology and Applied Research Branch Good morning, today I will be presenting a study of cancer incidence and mortality among children and adolescents in the United State, 2001-2003 Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention

Background Childhood (0-19 years) cancer accounts for 1% of the total cancer cases in U.S. Most common disease-related cause of death among youth aged 1-19 years 2007 estimates 1 10,400 new cases 1,500 deaths childhood cancer accounts for about 1% of the total cancer cases in the United States, however, it is the most common disease-related cause of death among youth aged 1-19 years. According to American Cancer Society, about 10,400 new cases and 1,500 deaths were estimated for children in 2007. 1 Cancer Facts & Figure 2007

Background More meaningfully grouped by morphology using International Classification of Childhood Cancer (ICCC) ICCC based on International Classification of Diseases for Oncology (ICD-O) ICD-O-3, issued in 2000, became effective in 2001. Unlike adulthood cancer, childhood cancers are more meaningfully grouped by morphology using International Classification of Childhood Cancer (ICCC). ICCC is based on the International Classification of Diseases for Oncology (ICD-O). The third edition of ICD-O was issued in 2000 and became effective in 2001.

Background SEER: Surveillance, Epidemiology, and End Results This figure was generated from data of Surveillance, Epidemiology, and End Results (SEER) Program. It illustrates the trend of childhood cancer incidence and mortality from 1975 to 2000. While there was a diseasing in cancer mortality, the overall tendency for cancer incidence was upward. There was a slight increase from late 1970s to early 1990s indicated by the red arrow. SEER: Surveillance, Epidemiology, and End Results

Background Incidence rates not well described after using ICD-O-3 Recent cancer death rates not well characterized Geographic differences in cancer incidence and mortality not well elucidated Although there was a slight increase in childhood cancer incidence, recent incidence rate has not been well described, especially after utilizing ICD-O-3. Recent cancer death has not been well characterized, either. In addition, geographic differences in cancer incidence and mortality have not been well elucidated.

Objectives Describe recent cancer incidence and mortality rates Identify demographic and geographic variations in cancer incidence and mortality So the objectives of this study is to Describe recent cancer incidence and mortality rates And, identify demographic and geographic variation in cancer incidence and mortality

Methods---Incidence Cancer incidence data from National Program of Cancer Registry (NPCR) and SEER Cover 43 states and the District of Columbia, representing 90% of U.S. population Cases definition: Children (0-14 years) and adolescents (15-19 years) diagnosed with a primary malignant neoplasm Cancer incidence data were from two federally-funded population-based cancer registries: NPCR and SEER Data covered 43 states and the District of Columbia, representing over 90% of U.S. population. Cases were restricted to children (0-14 years) and adolescents (15-19 years) diagnosed with a primary malignant neoplasm in the U.S. from 2001 to 2003.

Methods---Incidence Cancers stratified by sex, age, race, ethnicity, and U.S. Census region Age-adjusted incidence rates (per million), rate ratios, and 95% CI estimated by SEER-Stat p-values < 0.05 for significance except for racial and regional analyses In the analysis, patients were stratified by sex, age, race, ethnicity, and U.S. Census region. Age-adjusted incidence rates (per million), rate ratios, and 95% CI were estimated by SEER*Stat software. p-values < 0.05 were used to determine significance, except for racial and regional analyses.

Methods---Incidence In racial and regional analyses White and the Northeast as reference groups Bonferroni correction was applied Cut-point of p-value was 0.0167 Multivariate analyses Negative binomial modeling Adjust for sex, age, race, ethnicity, region, and year In racial and regional analysis, we used White and the Northeast as reference groups. Because of multiple comparisons, the Bonferroni correction was applied and significance was determined by p-values < 0.0167. Negative binomial modeling was also used to evaluate these demographic and geographic variations with adjusting sex, age, race, ethnicity, U.S. Census region, and year simultaneously.

Methods---Mortality Mortality data from NCHS Cover 50 states and the District of Columbia, representing 100% of U.S. population Stratified analyses for cancer mortality same as incidence Mortality data were from National Center for Health Statistics. Data were reported from 50 states and the District of Columbia, representing 100% of U.S. population. Stratified analyses for cancer mortality were as same as that for cancer incidence

Results I Cancer Incidence Stratify analysis Multivariate modeling II Cancer Mortality Now I will be presenting the results

Distribution of Childhood Cancer Incidence by ICCC-3 Groups, U. S This pie chart illustrated the distribution of childhood cancer incidence by 12 ICCC groups. About 36,000 childhood cancer cases were diagnosed in the United States during the study years. Among them, leukemias were the most common cancer (accounting for 26.3%), followed by CNS neoplasms (17.6%) and lymphomas (14.6%). These 3 diseases accounted for nearly 60% of all childhood cancer cases. CNS: Central Nervous System

Incidence Rate of All Cancers Combined by Sex and Age, 2001-2003 ** ** This figure showed that for all cancer combined, male had significantly higher incidence rate than female; and adolescents had significantly higher incidence rate than children. **P<0.05

Incidence Rate of All Cancers Combined by Race, Ethnicity, and Region, 2001-2003 Category   Cases Rate (95% CI) White (ref) 29,934 173.2 (171.3 - 175.2) Black 4,004 117.9 (114.2 - 121.6) ** API 1,328 131.4 (124.5 - 138.7) ** AIAN 273 97.3 (86.1 - 109.8) ** Hispanic 6,729 163.8 (159.9 - 167.8) Non-Hispanic 29,717 166.3 (164.4 - 168.2) Northeast (ref) 7,676 179.1 (175.1 - 183.2) Midwest 9,011 165.5 (162.1 - 168.9) ** South 1,1082 158.7 (155.7 - 161.6) ** West 8,677 165.3 (161.8 - 168.8) ** This table showed incidence rate of all cancers combined stratified by Race, Ethnicity, and Census Region. For all cancers combined, whites had significantly higher rate of cancer than blacks, Asian Pacific Islanders, and American Indians/Alaska Natives, respectively. Incidence rate for the Northeast was significantly higher than the Midwest, South, and West, respectively. There was no significant difference for the incidence rate between Hispanics and non-Hispanics. **Rate is significantly different from the rate of reference groups: white or the Northeast. P<0.0167 ref: reference group

Multivariate Modeling* for All ICCC-3 Groups Combined, 2001-2003 Characteristic χ2 P RR 95% CI Sex 34.88 <.0001 Male vs. Female 1.11 1.07-1.14 Age 148.65 15-19 vs. 0-14 1.35 1.31-1.40 Race 288.74 Black vs. White 0.66 0.63-0.69 API vs. White 0.73 0.69-0.78 AI/AN vs. White 0.55 0.48-0.62 Negative binomial regression showed incidence rate of all childhood cancers combined varied significantly by sex, age, and race. These findings were consistent with the stratified analyses. *Adjusting for sex, age, race, ethnicity, region and year Underlining denotes reference group

Multivariate Modeling* for All ICCC-3 Groups Combined, 2001-2003 Characteristic χ2 P RR 95% CI Ethnicity 27.6 <.0001 Hispanic vs. Non-Hispanic 0.9 0.87-0.94 Region 8.36 0.0391 Midwest vs. Northeast 0.93 0.89-0.98 South vs. Northeast 0.95 0.91-1.00 West vs. Northeast 0.96 0.92-1.01 Multivariate modeling also showed that incidence rate varied significantly by ethnicity and U.S. Census region. Interestingly, we found after adjusting for all the independent variables simultaneously, variation by ethnicity became statistically significant, which was insignificant in the stratified analyses. (After adjusting all these independent variables the relative ranking of incidence rates of childhood cancers across regions changed, as shown by the magnitude of rate ratio.) *Adjusting for sex, age, race, ethnicity, region and year Underlining denotes reference group

Distribution of Childhood Cancer Deaths, U.S., 2001-2003 This pie chart illustrated the distribution of childhood cancer death by primary cancer site. About 6,700 childhood cancer deaths were identified in the United States during the study year. Among them, leukemias were the most common cause of death (accounted for 26.7%), followed by CNS neoplasms (24.8%). These two cancers accounted for more than 50% of all childhood cancer deaths. CNS: Central Nervous System 17

Death Rate of All Cancers Combined by Sex and Age, 2001-2003 ** ** This figured showed that for all cancer combined, male had significantly higher death rate than female; and adolescents had significantly higher death rate than children. **P<0.05

Death Rate of All Cancers Combined by Race, Ethnicity, and Region, 2001-2003 Category   Cases Rate (95% CI) White (ref) 5,353 28.3 (27.5 - 29.0) Black 1,047 26.6 (25.0 - 28.3) API 259 24.2 (21.3 - 27.3) ** AIAN 71 21.2 (16.6 - 26.9) ** Hispanic (ref) 1,291 30.2 (28.6 - 31.9) Non-Hispanic 5,421 27.1 (26.4 - 27.9) ** West (ref) 1,772 30.7 (29.3 - 32.2) Northeast 1,099 25.6 (24.1 - 27.1) ** Midwest 1,453 26.4 (25.1 - 27.8) ** South 2,406 27.6 (26.5 - 28.7) ** This table showed death rate of all cancers combined stratified by Race, Ethnicity, and Census Region. For all cancers combined, white had significantly higher rate than API and AIAN, respectively. The rate for the West was significantly higher than the Northeast, Midwest, and South, respectively. Hispanic had significantly higher death rate than non-Hispanics. **Rate is significantly different from the rate of reference groups. P<0.0167 ref: reference group

Conclusions Incidence and death rates varied by sex, age, race, and ethnicity. Incidence rates significantly higher for Boys Adolescents (15-19 years) Whites Non-Hispanics Northeast Census region In conclusion, consistent with past studies, the childhood cancer incidence and death rates varied by sex, age, race, and ethnicity. For all cancers combined, incidence rates were significantly higher for boys adolescents (15-19 years) Whites Non-Hispanics The youth living in the Northeast

Conclusions Death rates significantly higher for Boys Adolescents (15-19 years) Hispanics West Census region Compared with Non-Hispanics, Hispanics had significantly lower incidence rate, but higher death rate. For all cancers combined, death rates were significantly higher for boys adolescents (15-19 years) Hispanics the youth living in the West Compared with Non-Hispanics, Hispanics had significantly lower incidence rate, but significantly higher death rate

Conclusions Northeast: highest incidence rate, lowest death rate West: highest death rate, second lowest incidence rate Compared with rates between 1995 and 1999 1 Incidence rate slightly increased Death rate slightly decreased We also found, the youth living in the Northeast had the highest incidence rate, but the lowest death rate. The youth living in the West had the highest death rate, but the second lowest incidence rate. Compared with rates between 1995 and 1999 1 We found that Incidence rate slightly increased Death rate slightly decreased 1 Cancer. 2002 May 15;94(10):2766-92

Strengths First study to describe cancer incidence using data representing 90% of U.S. population Describe cancer incidence rate with new ICCC Addressed geographic variation in childhood cancer incidence and mortality To our knowledge, this is the first study to describe the childhood cancer incidence using data covering over 90% of U.S. population. Describe cancer incidence rate with new ICCC coding scheme. It also identified geographic variation in childhood cancer incidence and mortality by U.S. Census region

Limitations Small cancer case number limits ability to study the geographic variation at state and county level Underestimation of API, AI/AN, and Hispanics The study findings are subject to several limitations: First, small case number limits ability to study the geographic variation at state and local county levels Second, Asian Pacific Islanders, American Indian/Alaska Natives and Hispanics may be underestimated.

Recommendations Maintain surveillance systems to track and understand changes in incidence and mortality Need survival study to address the geographic and ethnic discordance between incidence and mortality Last, recommendations: Continued surveillance systems should be well maintained to better track and understand changes in childhood cancer incidence and mortality Survival studies are needed to better understand the geographic and ethnic discordance between cancer incidence and mortality.

Acknowledgements Centers for disease control and prevention: Sherri L. Stewart, PhD Lori A. Pollack, MD/MPH Trevor D. Thompson, BS Jacqueline W. Miller, MD Christie R. Eheman, PhD Mary C. White, ScD I would like to acknowledge the following people for their assistance in this study.