Where did expanded eligibility for Medicare Savings Programs in Massachusetts lead to growth in program enrollment? What can we say about the demographics of places where expanded eligibility led to higher enrollment?
With these initial questions, my project set out to study what happened after a group of low-income seniors in Massachusetts organized for and won a dramatic expansion of eligibility for Medicare Savings Programs (MSP) in 2020 and again in 2023.
These seniors and millions like them nationwide are falling off of a "health care cliff" in old age. Assistance paying for health care premiums, deductibles, and co-pays was greatly expanded to the "working poor" aged 19-64 through the Affordable Care Act and Covid-era reforms. For people earning over 100% of the federal poverty level, though, turning 65 meant being cut off health care assistance since the ACA and other measures did not update eligibility rules for the elderly. Many Massachusetts Senior Action Council members I worked with went from spending 3% of their monthly income on health care to 30%, overnight. These individuals were typically low-wage workers like LPNs, rail car cleaners, rehab aides, and factory line workers who have too much retirement income to qualify for Medicaid but not enough to afford the needs of daily life. Expanding eligibility for Medicare Savings Programs, which return approximately $7,300 of benefits to seniors' pockets annually, is the lever this group of seniors identified and lobbied their elected officials to use to address this problem.
States can adjust eligibility for MSP (by raising the income limits and raising/removing asset limits) while the federal government picks up most of the tab, making this a fiscally doable way for states to alleviate hardship for seniors. This is what Massachusetts did in 2020 and 2023, setting up a clear way for me to examine data before and after these changes in order to see if eligibility expansion really led to increased enrollment and therefore material relief for seniors.
Turning my research interest into a functional dataset posed several challenges. The first was finding and adapting administrative data from the Centers for Medicare and Medicaid Services (CMS) at a meaningful scale and across my time period of interest. I am very familiar with the different forms of Medicare Savings Programs from my time lobbying the Massachusetts legislature about them, but even so I wasn't sure the best way to "count" enrollment--raw numbers, annualized growth rate, beneficiaries per capita? In the end, I modeled my statistic of "enrollment rate per 100,000 Medicare beneficiaries" off of AARP's research on this topic.
To do so, I wrote a function to process quarterly CMS MSP enrollment data from 2015-2023 into a dataframe with consistent variable names, which required writing a function to deal with inconsistencies over the nine years of data. Then, I used Census data to find out how many Medicare beneficiaries lived in Massachusetts counties in each of those years (from Census table C27006). Because I was originally interested in 2023 data to see if effects of the 2023 MSP expansion were noticeable, it was important for me to retrieve Census data for 2023 to create my enrollment rate statistic. However, the American Community Survey 5-year survey tables only exist through 2022. I had to rely on the 1-year ACS for 2023 Medicare beneficiary numbers, which are only produced for geographies with 60,000 people or more. That leaves out two very low population Massachusetts counties (Nantucket and Dukes, home to Martha's Vineyard). Ultimately, I was happy with the enrollment rate statistic I was able to produce for the vast majority of time and geography of interest.
I decided to narrow to looking at the effect of the 2020 MSP expansion for the rest of my project. Both 2020 and 2023 were years of dramatic upheaval in the realm of health care, and especially so for Medicaid. 2020 of course saw Covid-19 and the freezing of Medicaid rolls for the health emergency. 2023, then, was the year of "unwinding" those protections, which meant dramatic declines in enrollment due to new ineligibility or administrative issues.
These factors unquestionably confound the story that can be told about MSP expansion policy by looking at enrollment numbers alone. My project proceeds to develop tools and outputs that could be useful to a non-researcher working on senior health care, like someone in my previous community organizing role or in a legislator's office.
The first static plot, plot1_enrollment_time1, shows statewide MSP quarterly enrollment rates, 2015-2023. Covid and Medicaid unwinding are almost certainly driving most of the story here, but it is helpfully sobering to recognize that MSP policy is not the main story of elder enrollment in Medicaid. We know MSP is severly under-enrolled (that is, most seniors eligible do not apply to the program and therefore don't receive the benefits), so this could serve to visualize that for decision makers.
The second plot is a map, plot2_enrollment_map which seems of particular interest to health care enrollment assisters or senior-focused organizations or organizers. The map tells us that different parts of Massachusetts experienced change in enrollment post-MSP expansion very differently. With that context depicted on a map, I can imagine that groups and decision makers in Springfield, for example, could begin to zero in on what is working or not to explain the relative uptake of MSP.
The relative value of the map for showing county-level change inspired my second Shiny app, which allows users to map demographic characteristics of counties and see them alongside the MSP enrollment change post-2020 MSP expansion. This is a gesture toward my research question: "What can we say about the demographics of places where expanded eligibility led to higher enrollment? I used Census data to look at number of Medicare beneficiaries 65+ per capita, how white/nonwhite a county is. and the county's overall population. A more developed tool could select even better demographic indicators, or perhaps draw from survey data about senior well-being and economics.
The first Shiny app recreates a chart similar to the first static plot, charting enrollment rate over time by county, using plotly.
My statistical analysis is not rigorous but sought to fit a linear model to explain variation in MSP enrollment rate through demographic variables and the overall policy expansion in 2020. Results suggested that post-2020 enrollment rates went down, and that less white counties could be expected to see higher enrollment. I don't put much stock in these, but I hope that the data wrangling I performed and questions I pursued could be taken up in a more developed research design to get at the question of Where expanded eligibility for Medicare Savings Programs in Massachusetts led to growth in program enrollment and what drove that enrollment.
Finally, my text analysis was a preliminary attempt to look at how legislators describe MSP policy. Do they use terms of health care and health outcomes, or economic relief and cost reduction? I was hoping to analyze source material like public comments or committee hearing materials from the Massachusetts Legislature, but it releases very little data publicly. Instead, I analyzed a number of bill summaries from Congress.gov (pulled from this query). With a lay user in mind once again, I used tidytext and ggwordcloud to create a wordcloud to show the most common terms used in Congress which revealed a mix of health and economic terminology.
Community organizers and community members don't typically look into causal relationships between policy demands, but I see great value in finding out what is actually "won" by a campaign win like the MSP expansion victories in Massachusetts over the last four years. Further research, especially if done in concert with seniors and policy makers leading on this issue, could draw out lessons about what does and does not deliver much-needed relief in the form of lower health care costs. Perhaps academic researchers could also tap into even better data from MassHealth or other agencies to perform analysis.