Estimating the Effects of Medicare Advantage on the Health of Aging Population in Nursing Homes

Brown University, Providence RI

Investigators

Abstract

Older adults aged 65 or older, eligible for Medicare, must choose between traditional Fee-For-Service Medicare or Medicare Advantage (MA). MA has grown rapidly, covering 54% of Medicare beneficiaries in 2024. This growth has extended to nursing home (NH) residents, where about 23% of those turning 65 require NH care. In 2021, over 70% of NH residents were long-stay (LS), with the share of MA-enrolled LS residents rising from 13.3% in 2011 to 31.5% in 2021. To address the needs of LS residents, new care models such as Institutional Special Needs Plans (I-SNPs) have expanded. Despite the increase of MA and these new models in LTC, their impact on LS residents in NHs remains understudied. My F99 projects address this gap by describing MA trends and assessing MA’s effects on NH care quality and health outcomes among LS residents using rigorous causal methods. My K00 projects will expand the work using the unique NH electronic medical records (EMR) and newly developed AI algorithms to capture comprehensive health outcomes. While Medicare doesn’t pay for long-term custodial care in NHs, it covers all other healthcare, like hospital, post-acute, and hospice care. This incentivizes MA plans to manage care for LS residents. This may help reduce wasteful care, but it may also limit LS residents’ access to necessary care, making the impacts of MA on their welfare and health unclear. LS residents, with high levels of chronic illness and limited ability to advocate for care, are particularly vulnerable to delays or denials of care under MA plans. Aim 1 (F99) will study the growth of MA and I-SNPs among LS residents and identify determinants of this growth (Aim 1a), and examine MA’s impact on NH care quality and health outcomes using novel causal methods, including shift- share instrumental variable (SSIV) and generalized synthetic controls (GSC), which have not yet been used in aging research (Aim 1b). Traditional analyses use Claims or the Minimum Data Set (MDS) to assess health outcomes of NH residents, but the MDS may underreport the outcomes. Aim 2 (K00) will develop natural language processing (NLP) models to analyze NH EMR, which is the largest nationwide databases, to identify health outcomes like cognitive and physical functioning, and to uncover details of managed care process among MA-enrolled LS residents (Aim 2a). Using the most comprehensive outcomes identified by the full administrative data and the largest NH EMR, I will reevaluate the effects of MA on the health of LS residents, applying GSC (Aim 2b). My research aligns with the NIA’s goal to understand the factors affecting the health and wellbeing of older populations, while promoting inclusion of underrepresented groups in aging research. With extensive training under the mentorship from Dr. Mor (aging research), Dr. Meyers (MA; Health Economist), Dr. Rahman (Economist), and Dr. Mehrotra (NLP), I will gain the expertise needed to become an impactful faculty researcher focused on aging, advancing research that addresses critical gaps in the care of aging populations.