Leukemia in Atlanta, Georgia

LEUKEMIA IN ATLANTA, GEORGIA G. E. A lan D ever* This paper is an exploratory intra-urban m edical geographic study. Its purpose is to illustrate the spatial pattern of acute/sub-acute leu­ kemia incidence in the city of Atlanta. (1) Further, the paper will analyze various social, econom ic, and dem ographic variables that may be related to the spatial pattern of leukemia. (2) The variables selected for testing are those that reflect a hypothesis o f a viral etiology. (3) The viral etiology hypothesis reflects the theory that leukemia is caused by an infectious agent. If this is so, one w ould expect higher rates o f leukemia in areas that are conducive to the transmission o f an infectious disease. Thus, areas o f high and low leukemia incidence are designated as high and low risk areas. The correlations obtained for all areas (census tracts) are statistically significant. Subsequent analysis of the low risk areas showed significantly higher correlations, whereas the correlations were poor for the high risk areas. Therefore, meaningful associations for low risk areas can be determined but the factors relating to leukemia high risk areas are not so apparent. SPATIAL DISTRIBUTION OF LEUKEMIA IN ATLANTA. The spatial pattern o f acute/sub-ac^te leukemia in Atlanta reveals a clear pattern o f high rate and low rate areas (Figure 1). The high rates of leukemia, which range from 9.0 to 16.9 per 100,000, appear to cluster near the central city with one or tw o of these areas randomly distributed. The low rates range from 0.0 to 3.0 per 100,000 and show a contiguous rather than a clustered distribution. These areas o f high and low leukemia rates may b e considered areas o f high and low risk. (4) These areas are investi­ gated to determine the relationship o f leukemia risk to selected social, econom ic, and dem ographic variables. The map of leukemia incidence is based on all acute and sub-acute cases for all ages for the years 1960 through 1969. The 1965 estimated population o f Atlanta was used to calculate the rates. (5) The frequency polygon o f the incidence rates (Figure 1) shows a positive skewed dis­ tribution, indicating that there are many low rates with relatively few high rates. It is apparent from the spatial pattern and the frequency polygon that the distribution of leukemia incidence reflects both geo­ graphical and statistical peculiarities. This study accounts for these peculiarities through the use of step-wise regression and correlation analysis. *Dr. D ever is assistant professor o f geography at G eorgia State University. The paper was accepted for publication in June 1972. 92 So u t h e a s t e r n G e o g r a p h e r L E U K E M IA INCIDENCE 1960 - 1969 00 30 90 169 R A T EP E R1 0 0.00 0 Figure 1 METHOD OF ANALYSIS. Three different approaches are em ployed to identify the relationship of selected variables to the incidence rates of leukemia: (1 ) analysis of all census tracts in the city of Atlanta; (2 ) analysis of the high risk census tracts; and (3 ) analysis of the low risk Vol. X II, No. 2 93 census tracts. The m ethod of analysis used is step-wise regression and correlation. The variables that have been selected for testing for associ­ ations with leukemia incidence are shown in Table 1. The table also indicates the hypothesized relationships of the selected variables to leukemia which reflect the theory of a viral etiology, and in a few instances the established relationships are shown. TABLE 1 H YPO TH ESIZED RELATIO N SH IP O F SE LE C TE D VARIABLES TO LE U K E M IA : A TLA N TA , G EO RG IA Variables Exact Definition of the Variable Hypothesized relationship with Y X 101 Percent of Non-W hite Inverse X 102 Infant Mortality Inverse X 103 Percent of Housing W ithout Plumbing Direct X 104 M edian Value of the House Inverse X 105 Median Rent Inverse Xioe...