Breast cancer and Medicaid: Effects of enrollment, managed care, and misclassification on detection and treatment

Objectives. To assess the quality of breast cancer care under Medicaid and quantify biases in measurement.; Methods. Breast cancers were obtained from the statewide, population-based California Cancer Registry and linked to Medicaid records to ascertain enrollment, aid category, and health plan. Socioeconomic status and health care access were measured by aggregate data for Medical Services Study Area (California Health Care Policy Commission) and census block group (1990 Census). Women ages 30–64 enrolled in Medicaid in 1993 were categorized by duration of enrollment during the 12 months preceding diagnosis, and compared to other women. Logistic regression measured the effect of duration of enrollment on the odds of late-stage disease, controlling for demographic, socioeconomic, access, and tumor characteristics. Women ages 40–64 enrolled in Medicaid in 1997–1998 were categorized by enrollment during the 24 months preceding diagnosis. Outcomes of interest were the likelihood of being diagnosed with Stage 0–I disease, of receiving breast conserving surgery (BCS) among Stage 0–II cases, and of receiving radiation therapy among BCS recipients. Women in Medicaid managed care (MMC) were compared to women in Medicaid fee-for-service (MFFS); logistic regression controlled for duration of enrollment, aid category, education, demographic and tumor characteristics, and for stage at diagnosis, county screening intensity. Registry information on health insurer was compared to enrollment records for the 25-month period surrounding diagnosis. Logistic regression evaluated the association of demographic and socioeconomic factors with misclassification, and measured the risk of late-stage disease using Medicaid status from enrollment records and from the registry.; Results. Women enrolled in Medicaid were significantly less likely to be diagnosed at an early stage or to receive recommended treatment than other women with breast cancer. Duration of enrollment was a significant predictor of breast cancer stage. No evidence was found that MMC care was more successful at screening than MFFS, or that it provided a higher quality of care. Registry information on Medicaid status was substantially misclassified, but introduced minimal bias in evaluating cancer stage.; Conclusions. This is the first assessment of the effect of MMC on cancer detection and treatment. It provides insights that could improve future research in this area.