Abstract
Recent controversy about the high breast cancer incidence rates in Marin County, California (1) illustrates some of the difficulties in interpreting cancer trends in small geographic areas. Breast cancer incidence reportedly increased by 8.0% per year in the 1991 to 1997 period among non-Hispanic white women aged 45 to 64 years in Marin County, California, with no significant increase among other age groups in Marin county or other counties within the San Francisco Bay area (SFBA) (2). Clark et al. reported similar findings for the 1990 to 1999 time period (3). As was first noted by Edwards (1), a limitation of the two published analyses is that the denominators (number of women at risk) used to calculate the incidence rates after 1990 were projected from the 1990 census. This approach potentially under- or over-estimates the population in small geographic areas during periods of rapid migration (4). We therefore recalculated the rates using population estimates based on interpolation between the 1990 and 2000 census (5). The latter indicate that the number of non-Hispanic white women aged 45 to 64 years in Marin County increased by 29% from 1992 to 1999 (5), not by 10% as was projected from the 1990 census (6). We also calculated the estimated annual percent change in incidence rates from 1992 to 1999 among non-Hispanic white women residing in Marin County and in the other four counties in the San Francisco bay area (SFBA) in the 1992 to 1999 period, using rates based on projected populations from the 1990 census (6), and in the 1992 to 2000 period, using interpolated population estimates from the 1990 and 2000 census (5). The joinpoint regression analysis assumed one linear trend during this time period (7).
