Despite an apparent scientific consensus that US racial/ethnic groups have disparate distributions of breast cancer estrogen receptor (ER) status, previous studies of this topic are limited by: (1) the sizable proportion of cases with unknown ER status (typically 15-40%). (2) the paucity of socioeconomic data. limiting ability to control for confounding by socioeconomic position, and (3) the near exclusive focus on black/white differences. Suggesting, however, that estimates of racial/ethnic disparities in ER status based on such data might be biased, evidence indicates: (a) ER status is more frequently missing among women of color and/or less affluent women, and (b) risk factors for ER status may themselves be socially patterned, e.g., exposure to hormone replacement therapy, for the last 40 years more common among white and more affluent women, may increase likelihood of developing an ER-positive tumor. Because ER status is a key tumor biomarker relevant to both breast cancer treatment and survival, it is important to gauge accurately the extent to which estimates of racial/ethnic disparities in ER status are real or biased. Accordingly, to assess whether estimates of racial/ethnic disparities in breast cancer ER status and incidence of breast cancer by ER status are biased by missing data on ER status and lack of control for socioeconomic position, we will employ multivariate and sensitivity analysis techniques to analyze a multi- racial/ethnic population-based data set consisting of all incident cases of invasive primary breast cancer among women (n = 42,240 including 26,491 white non-Hispanic; 4,102 black; 4,961 Hispanic, and 4,970 Asian and Pacific Islander cases) occurring between 1998 and 2002 and included in the records of two US population-based cancer registries: the Los Angeles Cancer Surveillance Program and the Northern California Cancer Center. Records include data on race/ethnicity, age at diagnosis, and tumor characteristics and are linked to 2000 census tract socioeconomic data. Analyses will: (1) assess the distribution of ER status (positive, negative, or missing) by race/ethnicity, socioeconomic position, age at diagnosis, and tumor size and tumor stage, to estimate probabilities for the distribution of ER status among those within unknown status; and (2) using multivariate and sensitivity analysis, assess whether missing ER data and lack of control for socioeconomic position bias estimates of racial/ethnic disparities in breast cancer ER status and breast cancer incidence by ER status. Results will provide the first evidence whether estimates of racial/ethnic disparities in ER status are biased due to missing data on ER status and lack of inclusion of socioeconomic data, and will do so in a large multi-racial/ethnic cohort. Findings will provide knowledge needed to address health disparities, a key objective of the National Cancer Institute, the National Institutes of Health, and Healthy People 2010, while also addressing the objectives of PAR- 04-159, by using """"""""existing data"""""""" to determine """"""""applicability of biomarkers of tumor initiation and progression for epidemiologic studies."""""""" ? ? ? ? ?
| Krieger, Nancy; Chen, Jarvis T; Ware, James H et al. (2008) Race/ethnicity and breast cancer estrogen receptor status: impact of class, missing data, and modeling assumptions. Cancer Causes Control 19:1305-18 |
