The age peak in childhood acute lymphoblastic leukemia

BACKGROUND: White children have a much higher incidence rate of acute lymphoblastic leukemia (ALL) than do African American children. This discrepancy, coupled with the geographic and temporal variations in the incidence of childhood ALL, have led to speculation that factors associated with socioeconomic status (SES) may play an important role in its etiology. Because most of the variation is accounted for by the occurrence of a peak in incidence between the ages of 2 and 5 years, the purpose of this study was to compare the SES of children diagnosed with ALL between the peak ages of 2-5 years with those children diagnosed at other ages (birth-1 year and 6-14 years). METHODS: Patients included 4210 children who were diagnosed with ALL between January 1, 1989 and December 31, 1991 by a member institution of the Children's Cancer Group or the Pediatric Oncology Group. Of these children, 3614 were white and 596 were African American. The SES of a case was defined as the SES of the child's zip code of residence at the time of diagnosis. Five sociodemographic variable categories for each zip code were obtained from the 1990 U.S. Census including per capita income, number of housing units by household income, number of housing units by level of urbanization, number of persons older than 25 years by educational attainment, and number of persons by occupation. Mean values were compared for white children versus African American children, and peak ages (2-5 years) versus nonpeak ages (birth-1 year and 6-14 years) for both whites and African Americans. In addition, Wilcoxon's rank sum tests were performed. RESULTS: There were statistically significant differences between the means in each of the socioeconomic categories when African Americans were compared with whites. However, within race, the means of the SES variables for white children diagnosed during the peak ages (2-5 years) were not significantly different from children diagnosed at other ages (birth-1 year and 6-14 years). Similarly, all but one of the comparisons for African American children yielded nonstatistically significant results. Similar results were obtained from the Wilcoxon's rank sum tests. CONCLUSIONS: The results of this analysis suggest that age differences in childhood ALL incidence may not be solely accounted for by SES differences.