Patterns of breast cancer incidence and mortality in Marin County, California, have deviated markedly from those observed in other parts of the state over the past decade. These findings demonstrate the public health burden of breast cancer in this community, but also have the potential to serve as a source of hypothesis generation as regards the association between higher socioeconomic status and breast cancer. Cancer incidence patterns in Marin County have been informative in the past; the soaring incidence of endometrial cancer there in the 1970s  was ultimately linked to the utilization of estrogen therapy by postmenopausal women.
The higher breast cancer incidence in Marin County has been attributed to its unique and uniform sociodemographic characteristics , which correspond to a higher prevalence of women with known breast cancer risk factors. With a relatively small population for an urban county, Marin County was one of the wealthiest counties in the United States in 2000, with a median per-capita income of almost US$58,000, about 200% higher than the state and national medians . Marin County consistently ranks among counties in the nation with the most highly educated populations, with at least 44% of adults having a bachelor's degree, which is 220% higher than the national rate in 1990 . It also has a higher percentage of white, non-Hispanic residents (80%) than surrounding counties .
In accordance with its high socioeconomic status, Marin County has higher proportions of women with other established breast cancer risk factors, including lower parity  and later age at childbearing. In 1994, 71% of the live births to white women were to mothers aged 30 years and over in Marin County, as compared with 46% in California . In addition, 69% of all women aged 55–64 years and 47% of all women aged 45 and over reported using hormone replacement therapy as part of a community health survey conducted in Marin County in 2001 (RRE, unpublished data, 2002) although comparable figures are not available for the state. Prehn and West, using 1990 census data, calculated breast cancer incidence rates for aggregations of census block groups matched to Marin County on characteristics associated with higher breast cancer risk (percentage white population, urban status, average parity, median household income, percentage of persons with a college degree, percentage of persons with a working-class occupation, and percentage of households living below the poverty line), and found rates in the matched areas to be comparable with those in Marin County . In another study using interview-based information, higher breast cancer incidence rates among white women in the SFBA were also fully explained by the distribution of parity, age at first birth, months of breast-feeding, age at menarche, and age at menopause . Other previous analyses of nationwide variation in breast cancer incidence and mortality found most of it attributable to the distribution of known breast cancer risk factors [15, 16]. It thus seems likely that a substantial part of the excess incidence observed in Marin County is explained by a higher concentration of women with a higher breast cancer risk profile.
The increasing incidence of breast cancer might be explained by changes in the composition of the Marin County population over the past decade. Data from the US Census Bureau show that Marin County had relatively high levels of outmigration and had low levels of immigration during 1990–1998 compared with other parts of California . Examination of the age-specific population denominators used in this analysis  shows that the proportion of women aged 45–64 years, which includes the leading edge of the 'baby boomer' cohort, changed very little over the decade in Marin County, whereas this proportion increased substantially in comparison areas. This deviant pattern could be explained by the outmigration from Marin County of women aged 45–64 years and by the immigration of younger women. As realty data show that the median single family home price surged to $530,000 by the year 2000, making Marin County one of the least affordable counties in the nation , we speculate that women in the age group 45–64 years who may have had more children at younger ages left the county over the decade, while younger, perhaps professional, women who did not have children or who delayed child-bearing immigrated to or stayed in the county. Some support for this phenomenon is provided by the widening difference between Marin County and California in the proportion of live births to white mothers aged 30 years and older as compared with those aged younger than 30 years (ratio: 1.54 in 1994, and 1.60 in 2000) . The numbers of women in Marin County with a high breast cancer risk profile may thus have increased further over the decade.
We consider it doubtful that the observed trends are strongly biased by inaccurate cancer registry or population data. The trends observed in the rates were also observed in the case counts alone. Although some of the observed rate increases could be attributable to inaccuracy of the intercensal population estimates, particularly in the 45–64 age group, systematic population undercounts would presumably have produced similar increases in the incidence of other cancers, which we did not observe. Furthermore, the California Department of Finance population estimate for Marin County for the year 1999 (244,900) was less than 1% lower than the actual count performed by the US Census in 2000 (247,289). Regardless of this, incidence trends will be reassessed when revised population estimates benchmarked to the 2000 census become available.
Breast cancer incidence has been positively associated with socioeconomic status across race/ethnicity, geography, and time [19–23]. Recent data from California show that rates for white women in the highest socioeconomic quintile were 27% higher than the lowest quintile and were 13% higher than the overall rate . However, understanding of the biologic factors mediating this association remains incomplete. In a large, national, interview-based study conducted in the 1970s, adjustment for reproductive patterns, menopausal characteristics, hormone use, alcohol consumption, body mass index, height, and family history did not entirely account for the relations of income and education to breast cancer risk . To achieve a more contemporary understanding of the contributions of established breast cancer risk factors, it would be helpful to reassess the roles of factors whose prevalence has increased substantially over time, including delayed childbearing  and use of hormone replacement therapy . In addition, there are ongoing efforts in Marin County to identify novel risk factors relevant to this socioeconomic group. These efforts include a case–control study of adolescent exposures and experiences for which data collection was recently completed, and projects to identify environmental exposures unique to this group or to the county.
It is uncertain to what extent the socioeconomic gradient in breast cancer, and the patterns we have observed in Marin County, might be explained by better awareness of breast cancer, by access to health care, or by utilization of screening programs. Screening mammography has been linked previously to breast cancer incidence increases in the United States, most notably the 4% annual increase after its widespread adoption in the 1980s . In the state of Hawaii, mammography utilization accounted for 23% of the geographic variation in overall breast cancer incidence in 1992–1993 and for 36% of the variation among women aged 50–64 years .
At first glance, our observations of incidence excesses limited to early-stage cancers would suggest some relation to regional differences in screening. However, these patterns do not demonstrate other hallmarks of differential 'lead time bias'. There was no evidence of a Marin County deficit in later-stage cancers, mortality rates remained stable despite distinct mortality declines observed in comparison regions, and disparate age-specific incidence trends were observed within the larger age group (aged 40+ years) targeted by screening programs . In addition, mammography utilization data do not suggest substantially elevated rates of screening among Marin County women. A community health survey conducted in Marin County in 2001 found that 68% of women aged 40+ years received a mammogram in the past year; this proportion is comparable with or only slightly higher than those for other areas with lower breast cancer rates (RRE, unpublished data, 2002). There are no data available to explore other aspects of mammographic screening in this population, such as increased frequency or better sensitivity due to higher volume . The possibility of a real increase in small tumors is, however, underscored by reports of significant increases in small breast cancers among unscreened women in Australia .
United States cancer registries do not routinely collect information about patient income and education, making assessment of cancer trends by socioeconomic status very difficult . Our findings of rapidly increasing rates of breast cancer in Marin County over the past decade, however, along with evidence that high rates there are explained by the sociodemography of the county , cause us to speculate whether breast cancer rates are increasing similarly in other groups of highly educated and affluent women nationwide. While we can monitor incidence rates in Marin County because it is a county, we cannot conduct ongoing annual cancer surveillance in sociodemographically similar populations living in sub-county regions because of a lack of appropriate population estimates. When detailed 2000 Census data are released for subcounty areas, we will re-examine the relations between sociodemographic characteristics and geographic variation in breast cancer incidence. If our speculations are supported by these data, then Marin County statistics indeed serve as a 'canary in the gold mine' as regards breast cancer incidence in educated or affluent women, and they may also be prescient as regards breast cancer mortality. These observations should motivate us to intensify our efforts to learn more about the relationship between socioeconomic status and breast cancer.