A Randomized Trial to Promote Mammography in the Medicaid Population

ABSTRACT BODY:

Introduction: Cancer screening rates in the Medicaid population are substantially lower than rates among privately insured individuals. Mammography remains a central focus for disease prevention, and disparities in breast cancer screening rates exist in Medicaid-enrolled women. While there are internally valid interventions focused on mammography, very few evidence-based interventions are easily and regularly translated to population-level practice. The use of financial incentives is a primary example of this research-practice gap. Financial incentives can influence cancer screening, but limited research addresses how to scale the use of incentives to population-based programs. In June 2014, we implemented a population-based randomized trial that used direct mail (DM), financial incentives, and a centralized patient navigation system to promote mammography among women enrolled in Minnesota’s Medicaid program who were overdue for breast cancer screening (based on their medical claims).

Methods: All women enrolled in Minnesota’s Medicaid program were included if they (1) were between age 50 and 74, and (2) had not received a mammogram within a 3-year period prior to the study (N=27,273). This target population was randomized to either an intervention or delayed intervention group. The intervention group received innovative and persuasive DM materials coupled with a $20 incentive for using their Medicaid benefit to get screened, and the DM mailer instructed women to call a central number for assistance. If participants called, patient navigators set up mammography appointments via three-way calls. The presence or absence of a billing claim with appropriate CPT codes was used as evidence of screening (1=screened, 0=not screened). We hypothesized that the intervention would inspire women to get a mammogram shortly after receiving DM, and we compared treatment versus control for a cross-section of time within the first four months after implementation using logistic regression. However, we gathered twelve months of claims data and assess time to screening using survival analysis in order to examine the hypothesis that screening differences would be present early in the intervention.  

Results: Survival function demonstrated there were screening differences between treatment and control within the first three months after program implementation. Screening rates converged after approximately four to five months. The odds of getting screened for breast cancer were 1.15 (95% CI=1.06, 1.25) as high for treatment as for control adjusting for disability status in cross-sectional analysis. 

Conclusions: Findings compare favorably to prior experience with direct mail as well as the experience of other investigators despite targeting a particularly hard-to-reach population. This intervention may reach insured individuals who generally remain outside the health care system and has potential for nationwide use in other state’s Medicaid programs.