The Effects of Medicare Advantage Contract Concentration on Patient Nursing Home Outcomes

Medicare Advantage (MA) contracts receive capitated payment which incentivizes care management and efficiency in the care delivery process. The Medicare Modernization Act of 2004 allowed MA contracts to form preferred provider networks and to concentrate their patients among those preferred providers to take advantage of “economies of scale” in managing patient care. However, the effect of such concentration on patients’ welfare is not clear. Prior research has found that MA contracts pay providers at a much lower rates than traditional Medicare and selectively guide patients to lower quality providers. In this study, we focus on the skilled nursing facility (SNF) industry to assess the patient health implications if the treating SNF has a higher share of patients enrolled in the same MA contract. We use an instrumental variable approach to estimate the causal effect.

We use the Minimum Dataset (MDS) to assess all MA patients admitted to a SNF without a prior SNF stay in the past year, from 2012-2014. The MDS contains detailed patient assessment information for all patients regardless of MA enrollment status. We link MDS patients to Medicare enrollment file to identify their MA enrollment, to the Medicare Healthcare Effectiveness Data and Information Set (HEDIS) data to identify patients’ enrolled MA contract and to the On-line Survey & Certification Automated Record (OSCAR) to gather SNF characteristics.

Our primary outcomes of interest are 180-day survival post SNF admission, 30-day hospital readmissions, 30-day discharge to home, and if a patient became a long stay patient (100+ days in SNF). Our primary explanatory variable is the prevalent concentration of a given patient’s MA contract in the SNF they are admitted to. This is calculated by taking a count of patients in different contracts from the MDS on the first Thursday of April in a given year, when SNFs are expected to have the highest occupancy rate. This variable is endogenous however, as SNF concentration may be driven by MA contract selection effects, and may reflect unmeasured components of patient health and SNF quality. To address this, we instrument SNF concentration with the concentration of a patient’s MA contract in their residential zip code. As SNF choice is primarily driven by a patient’s distance to a SNF, the zip code level concentration is strongly correlated with the SNF concentration while being plausibly unrelated to the outcomes of interest. Using this instrument, we fit linear probability models adjusting for patient characteristics, SNF characteristics, and MA contract and zip code fixed effects.

We find that a one percentage point increase in a patient’s MA contract concentration in a SNF increases the probability of being discharged within 30 days by 0.45 percentage points, and decreases the probability of becoming a long stay patient by 0.35 percentage points, however, also decreases the probability of a patient surviving 180 days’ post admission by 0.17 percentage points. These results support the hypothesis that MA contract concentration does reduce length of post-acute SNF care, however, it may worsen the long term survival outcomes of a patient.