• Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Breast cancer incidence is higher among black women than white women before age 40 years, but higher among white women than black women after age 40 years (black-white crossover). We used newly available population-based data to examine whether the age-specific incidences of breast cancer subtypes vary by race and ethnicity. We classified 91908 invasive breast cancers diagnosed in California between January 1, 2006, and December 31, 2009, by subtype based on tumor expression of estrogen receptor (ER) and progesterone receptor (PR)-together referred to as hormone receptor (HR)-and human epidermal growth factor receptor 2 (HER2). Breast cancer subtypes were classified as ER or PR positive and HER2 negative (HR(+)/HER2(-)), ER or PR positive and HER2 positive (HR(+)/HER2(+)), ER and PR negative and HER2 positive (HR(-)/HER2(+)), and ER, PR, and HER2 negative (triple-negative). We calculated and compared age-specific incidence rates, incidence rate ratios, and 95% confidence intervals by subtype and race (black, white, Hispanic, and Asian). All P values are two-sided. We did not observe an age-related black-white crossover in incidence for any molecular subtype of breast cancer. Compared with white women, black women had statistically significantly higher rates of triple-negative breast cancer at all ages but statistically significantly lower rates of HR(+)/HER2(-) breast cancers after age 35 years (all P < .05). The age-specific incidence of HR(+)/HER2(+) and HR(-)/HER2(+) subtypes did not vary markedly between white and black women. The black-white crossover in breast cancer incidence occurs only when all breast cancer subtypes are combined and relates largely to higher rates of triple-negative breast cancers and lower rates of HR(+)/HER2(-) breast cancers in black vs white women.
    CancerSpectrum Knowledge Environment 07/2012; 104(14):1094-101. DOI:10.1093/jnci/djs264 · 15.16 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In a famous article, Simpson described a hypothetical data example that led to apparently paradoxical results. We make the causal structure of Simpson's example explicit. We show how the paradox disappears when the statistical analysis is appropriately guided by subject-matter knowledge. We also review previous explanations of Simpson's paradox that attributed it to two distinct phenomena: confounding and non-collapsibility. Analytical errors may occur when the problem is stripped of its causal context and analyzed merely in statistical terms.
    International Journal of Epidemiology 03/2011; 40(3):780-5. DOI:10.1093/ije/dyr041 · 9.20 Impact Factor