Abstract
Introduction: Breast cancer survival is affected by endogenous factors, i.e. genetic factors, and exogenous factors, i.e. diet and socioeconomic status (SES). In an attempt to begin to define, the mechanisms by which SES could affect survival we explored in a population served by a public hospital the potential effects of insurance status and income on the overall survival of breast cancer (BCa) patients.
Methods: 1627 female BCa patients diagnosed/treated between 1987 – 2006 at LSUHSC-S's Feist-Weiller Cancer Center (FWCC) were investigated to explore the potential effect of disparity of overall survival with insurance status, income, race, age, stage, body mass index (BMI) and year of diagnosis. Chi-Square test, Student t test,ANOVA and Student-Newman-Keuls test were used with actuarial methods and log rank tests used to estimate and compare overall survivals. Univariate and multivariate Cox regression were used to estimate the hazard ratios (HR) of risk factors with overall survival. All p-values < 0.05 were considered statistically significant.
Results: Age was statistically different between stages, insurance status, year of diagnosis and BMI status. Mean income was statistically different between race (p<0.01) and insurance status (p=0.033) but not statistically different among stage, year of diagnosis, or BMI status. The insured patients (commercial + Medicare) had higher proportions of stage 1 and stage 2 cancers. Those with unknown insurance had a higher proportion of unknown stage.
In univariate analysis, overall median survival was 10.5 years (white 11 yrs, black 9 yrs, P=0.029). Median survival times were 15.5, 12.5, 5.5, 1.5 and 5.5 years for stage 1, 2, 3, 4 and unknown stage. In the four 5 year cohorts from 1987 to 2006, the median survival time increased from 7 yrs to 10 yrs, to 11 yrs, to > 7 yrs (p<0.001). Comparing the decade 1987–1996 to1997–2006 the median survival times were 9 years and >10 years (p=0.011). The HR was 1.018 for age (p<0.001). Insurance status had a marked affect on survival with the median survival times (HR) of >20 (1), 16.5 (2.2), 7.5 (3.5) and 3.5 (6.1) years for insured, Medicaid, uninsured and unknown insurance respectively (p<0.001). Adjusting for race and stage, overall survival was comparable between FWCC and SEER.
Multivariate Cox regression found that race, stage, insurance status, age and years of diagnosis were significant predictors of overall BCa survival. The HRs were 1.77, 3.09, 8.79 and 2.22 for stage 2, stage 3, stage 4 and unknown stage, respectively, compared to stage 1. White patients had 22% lower risk of dying than blacks. Compared to diagnosis in 1987–1996, the risk of dying for diagnoses in 1997–2006 decreased 39% (HR=0.61). Adjusting for other factors, insurance was a major factor affecting survival with HRs of 1.94, 2.893 and 2.892 for Medicaid, uninsured and unknown insurance patients respectively compared to insured patients.
Conclusion: Insurance status proved to be an important BCa survival factor. After adjusting for confounding factors that might influence acquisition of insurance, i.e. age, race, stage, BMI and income, insurance status remained a highly significant survival factor with HR ranging for 1.9–2.9 for Medicaid or uninsured patients. The quality and availability of care was similar for all patients but the effects of other illnesses, compliance, and health literacy are being examined.
Second AACR International Conference on the Science of Cancer Health Disparities— Feb 3–6, 2009; Carefree, AZ