An evolving geography of breast cancer incidence: Are patients or practices the cause?

Between 1985 and 2009, breast cancer screening emerged from an uncommon practice to a staple of women’s health care.  Using Connecticut Tumor Registry data from this 25-year period, we examined where 53,808 women lived when diagnosed with invasive breast cancers.  Geographic variation during 5 distinct periods (1985-89, 1990-94, 1995-99, 2000-04 and 2005-09) was evaluated using a spatial scan statistic (SaTScan).  Over this period, statewide incidence of breast cancer increased 23% (from 145.9 to 179.3 cases annually per 100,000 at-risk women).  Piecewise regression analysis discerned little change in the statewide rate between 1985 and 1994, but estimated an increase of 3.63 cases per 100,000 at-risk women per year from 1995 through 2009.   Accompanying this, we found substantial geographic rate variation around the State in 1985-89 that affected nearly 40% of it's at-risk population.  Such variation diminished significantly through 2000-04 (i.e., fewer and less severe ‘clusters’ affecting fewer than 17% of the P-A-R ) and we speculate this trend occurred with uptake of disease screening practices that added new cases by depleting longstanding reservoirs of sub-clinical disease in communities where health service utilization, initially, may have been sub-optimal.  A stabilization of geographic rate variation observed through 2009 ultimately may reflect true differences in concentration of populations-at-risk across Connecticut.  Our findings prompt consideration of how evolution of health service utilization may artifactually affect geographic presentation of disease over time and why it is necessary to differentiate spatial variation attributable to population and health system characteristics.