Receipt of Regular Primary Care and Early Cancer Detection in Appalachia.

Abstract

The objective was to examine the impact of regular primary care encounters (PCE) on early breast cancer detection in an Appalachian sample of Medicare beneficiaries diagnosed 2006-2008. Determinants of PCE were investigated and a mediation analysis was conducted where PCE was a mediator to cancer stage. A total of 3,589 cases were identified from Appalachian areas in Pennsylvania, Kentucky, Ohio, and North Carolina, and health care services were examined 2 months to 2 years prior to diagnosis. A regular care PCE variable was constructed with 4 ordinal levels: none, any, "annual," and "semi-annual." Association of PCE with stage, mortality and covariables was conducted using ordinal logistic regressions and Cox Proportional Hazards survival models. Sixty-eight percent of the cases had semi-annual PCE. Regular PCE was strongly associated with late-stage cancer rates (39%-13% by increasing PCE level, P < .0001) and 5-year all-cause mortality (42%-24%, P < .0001). Subgroup analysis revealed variations by hypertension and urban status, with nonhypertensives with no PCE being at particularly increased risk. Significant determinants of PCE included age, rural/urban status, comorbidity, dual Medicaid insurance, Appalachian region economic classification, state, select comorbidities, hypertension, and minimum distance to provider. Mediation analysis results were consistent with lower number of comorbidities leading to increased late cancer detection due to patients having a decreased PCE. PCE is an important determinant of cancer detection, with a dose-response relationship. Variations exist by geography and hypertension. Comorbidity may influence both PCE and late-stage rates with partial mediation through PCE.