Modifying effect of serious mental illness on breast cancer care among female Medicaid recipients

Objective: To analyze the variation in breast cancer (BCa) care among female Medicaid recipients with and without serious mental illness (SMI). Methods: We created a cohort of BCa patients aged 18-64 years using Pennsylvania Medicaid claims between 1995 and 2004. Date of first claim for BCa was the index-date. We identified those with a claim for SMI. Clinical data was obtained from Pennsylvania Cancer Registry. BCa with SMI and BCa only groups were retrospectively followed for one year pre and up to two years post the index-date. Screening and health resource utilization (HRU) patterns were compared. We fitted GLM log-link models to explore the marginal burden of SMI in BCa patients. Cox proportional hazard model was used to study the effect of SMI on survival. Results: Our cohort consisted of 10866 BCa patients. Of these, 1340 had a diagnosis of SMI. Differences in the demographic and clinical attributes were observed between groups. Unadjusted comparisons of HRU showed that BCa with SMI group had higher mean inpatient cost ($2132 vs. 3802, p<.0001), higher outpatient cost ($1021 vs. $1541, p<.0001), higher proportion with inpatient visits (36% vs. 45%, p<.0001) and higher proportion with outpatient visits (51% vs. 71%, p<.0001). Results of GLM log-link model showed that BCa with SMI group had higher total cost. Being a BCa patient with SMI was not a predictor of mortality. Conclusions: SMI has a modifying effect on BCa care among female Medicaid recipients. Early detection and treatment for SMI can help reduce the burden on Medicaid.