Medicare Advantage Growth and Nursing Homes: Estimating Quality Spillovers

Traditionally, nursing homes have had few clear financial incentives to improve quality when reimbursement for care has come primarily from Medicaid and fee-for-service Medicare. In cases where facilities must contract with private payers, however, nursing homes become responsible for costs and outcomes in ways that are non-existent under fee-for-service payment models. Currently, enrollees in private Medicare Advantage (MA) plans account for more than one-third of all Medicare beneficiaries and constitute a growing proportion of nursing home users. Recent evidence suggests that MA may generate better post-acute care outcomes at lower costs relative to traditional Medicare, yet little evidence exists on whether these efficiencies spillover and benefit long-stay residents who frequently reside in the same nursing facilities.

In this paper, I examine how market level changes in MA enrollment impact the quality of care for long-stay residents in non-hospital-based nursing homes across the country. Exploiting idiosyncratic policy changes in the MA payment schedule between 2005 and 2011 (prior to the significant redesign of payment policies starting in 2012), I use an instrumental-variables framework to estimate causal effects of changes in MA penetration on a battery of commonly used measures of nursing home quality for long-stay residents. Primary data sources include MA Ratebook files, CMS state/county/contract files, and CMS Nursing Home Compare archives.

All first stage regressions demonstrate that payment changes are highly correlated with changes in MA enrollment and meet commonly accepted thresholds of statistical significance (F=24.44), suggesting that weak instruments are not a concern. IV results show that increasing MA penetration is generally associated with worsening quality for long-stay residents across a number of measures. For example, a 1 percentage point increase in MA penetration is associated with a statistically significant 0.18 percentage point increase in the proportion of long-stay residents with pressure ulcers (p<0.001), or approximately a 1.70% increase. Back of the envelope calculations suggest that the national prevalence of pressure ulcers would have been 21% lower had there been no increases in MA enrollment over the study period. In a secondary analysis, I explore potential pathways to explain these results, and find that staffing hours per resident day decrease with increasing MA penetration. This is consistent with facilities cutting staff in response to lower payment rates from MA relative to traditional Medicare, and also the wide body of literature showing a strong relationship between staffing and quality.

These findings suggest that estimates of potential long-term savings from MA on government spending for post-acute care need to account for the quality impacts on long-stay residents. Moreover, seeing as most of these nursing home residents are covered by Medicaid, the costs associated with negative spillovers on quality will be principally borne by individual states, rather than the federal government, leading to broader normative questions about the redistributive fiscal implications of increasing MA enrollment.