When being insured is not enough: The effect of insurance type on survival in cervical cancer (060)

Abstract

<b>Objectives:</b> To investigate disparities in early-stage diagnosis and survival by insurance type in adults aged <65 years with cervical cancer. <b>Methods:</b> Adults aged <65 years diagnosed with cervical cancer between 2004-2014 with private, Medicaid, or no insurance in the National Cancer Databases, were eligible. Patients with other insurance or missing stages were excluded. Adjusted odds ratio (AOR) and 95% CI for stage I diagnosis were estimated using logistic modeling, with relative importance weighted by dominance analysis. Propensity score balancing was applied to sequentially adjust demographic and socioeconomic (SES) factors, histology, stage, and treatment by insurance. Survival and adjusted hazard ratios (aHRs) were compared before and after sequential balancing using the weighted Kaplan-Meier method and Cox modeling, respectively. Excess relative risk (ERR) was calculated using aHR-1. The individual contribution of factors to ERR was then estimated after each sequential adjustment. <b>Results:</b> There were 67,608 evaluable patients, among which 60.6% had private insurance, 27.2% had Medicaid, and 12.3% were uninsured. The early-stage diagnosis was largely explained by age and insurance with a normalized weight of 61% and 28%, respectively (Figure 1A). Overall, 62.3%, 45.3%, and 41.6% of patients with private insurance, Medicaid coverage, or no insurance were diagnosed at stage I, respectively. Patients with Medicaid coverage or the uninsured were 51% (p<0.0001) or 56% (p<0.0001) less likely to present with early-stage than those with private insurance, respectively. Moreover, uninsured patients had approximately 10% lower odds of a stage I diagnosis compared to patients with Medicaid coverage (p<0.0001). Overall survival and 5-year survival rates before (Figure 1B) and after sequential balancing (Figure 1C) were superior for patients with private insurance and similar for those with Medicaid versus no insurance. Figure 1D illustrates the sequential reduction in aHR and ERR associated with Medicaid/uninsured versus private insurance. Demographic and SES factors accounted for 1.2%, histology for 7.7%, stage for 46.9%, and treatment for 8.7% of the survival disparity in the Medicaid/uninsured group versus private insurance, with 35.5% remaining unexplained. <b>Conclusions:</b> Stage is the most significant predictor of survival among patients with cervical cancer in our study, and early-stage diagnosis is significantly higher among women with private insurance. Although Medicaid patients had a 10% higher likelihood of stage I diagnosis than uninsured, survival among Medicaid and uninsured patients was similar and significantly inferior to private insurance. Unexplained factors accounted for 35.5% of ERR in survival and merit further investigation. These factors could help explain why Medicaid coverage does not lead to the same outcomes as seen with private insurance. Additionally, future studies are needed to evaluate the effect of the Affordable Care Act on outcomes.Fig. 1Insurance effects on early-stage diagnosis <b>(</b>A, D<b>)</b> and survival <b>(</b>BD<b>)</b> in adults <65 years with cervical cancer. Relative importance (Normalized Weight %) of demographic and socioeconomic factors for early-stage diagnosis was weighted by dominance analysis (A). Insurance differences in unadjusted survival (B) and in adjusted survival (C) were evaluated using weighted Kaplan-Meier methods with pairwise differences in 5-year survival, hazard ratio (HR) or adjusted HR (AHR) and 95% confidence interval (CI) included in the embedded tables, respectively with superior survival for private insurance (<i>P<0.</i>0001) and statistical similar survival for Medicaid vs no insurance (<i>P>0.05</i>). The proportion diagnosed with stage I disease, 5-year survival rate, adjusted hazard ratio (AHR) with 95% CI, excessive relative risk (ERR=AHR-1) and the individual contribution to the total EEC associated with Medicaid/uninsured relative to private insurance were estimated after sequentially adjustment for demographics and socioeconomic status (SES) factors followed by histology then stage and finally by treatment (D).