Spatial variation in breast cancer incidence in Connecticut: Does geographic access to mammography matter

Introduction. Breast cancer is a major health burden worldwide. Among women, breast cancer is the most commonly diagnosed cancer and the second leading cause of cancer death. Incidence rates for breast cancer in US women have been on the rise for decades. Research has suggested that the increased incidence over time may be due in part to more aggressive screening campaigns in the population. Mammography access in the U.S. varies based on geographic and structural (sociodemographic, racial, and economic) barriers. The current study aims to answer if spatial variation in breast cancer incidence in Connecticut is related to the geographic availability of mammography.;Methods. Data for the numbers of incident cases of invasive breast cancer in women for all of Connecticut were obtained for the years 2003 to 2007 from the Connecticut Tumor Registry by town of residence. Age-adjusted standardized incidence ratios (SIRs) were calculated and cluster analysis was performed. Since the population size varies greatly among these small mapping units, Bayesian hierarchical modeling was applied to produce a map of smoothed SIRs with an overlay of all licensed mammography clinics in the state. This can be used to identify geographic differences in breast cancer incidence in Connecticut as it relates to the availability of mammography. Poisson regression was run with categorized covariates for distance to nearest mammography facility to answer the question of whether breast cancer incidence rates in Connecticut are related to the geographic availability of mammography.;Results. The smoothed map (based on a convolution model with distance covariates) shows that although close proximity to a mammography facility does not necessarily coincide with higher incidence, towns with less geographic accessibility do tend to have lower SIRs. Poisson regression, adjusted for socioeconomic variables, revealed that living in a town a distance of ≥ 8 miles from a mammography facility was a significant predictor for lower incidence of breast cancer. There is a noticeable negative trend as distance to the nearest mammography facility increases, with SIRs falling from 1.00 (ref.) to 0.95 (5 - < 8 miles) to 0.87 (≥ 8 miles). Unadjusted analysis suggests confounding of distance to the nearest mammography facility with income and/or education.;Conclusions. Visual display of the geographic difference in breast cancer incidence in Connecticut in relation to the location of screening facilities reveals geographic access seems to impact incidence. This is confirmed through regression analysis, where longer distance to mammography facility is a significant predictor for lower incidence of breast cancer. A negative trend emerges: As distance to the nearest mammography facility increases, SIRs for breast cancer are reduced. Further unadjusted regression analysis reveals that Connecticut towns with less geographic access to mammography screening tend to have lower socioeconomic status.