K. Kafadar et Jw. Tukey, UNITED-STATES CANCER DEATH RATES - A SIMPLE ADJUSTMENT FOR URBANIZATION, International statistical review, 61(2), 1993, pp. 257-281
A statistical study of death rates due to various forms of cancer in t
he United States begins with an adjustment of the rates in terms of a
measure of urbanization. The rates used in this study were collected a
nd age-adjusted by the National Cancer Institute (NCI); they cover the
years 1950-1969. Only the rates for trachea, bronchus and lung combin
ed are analyzed here. The county rates are analyzed separately for eac
h of five disk-shaped regions of the country, selected to cover most a
reas with relatively high rates. A measure of urbanization based on th
e size of the largest place in the county is defined and then used to
stratify the counties in each disk. Considering first only Poisson var
iation, Poisson intervals for the stratum rates in each disk are deriv
ed and compared across regions. The analysis, largely exploratory in n
ature, is a first step. When continued, it should provide insights abo
ut geographic gradations that might later prove to be related to envir
onmental influences. This study treats only the rates from cancer of t
he trachea, bronchus and lung, for white males in selected parts of th
e country, but it is clear that the methodology may be applied to othe
r sites, populations, or areas.