NBER Center for Aging and Health Research

2022. Amitabh Chandra. How Do Informed Consumers Consume Healthcare? Evidence From Medicare.

Patients are generally far less knowledgeable about medical conditions, services, and treatment options than their healthcare providers. This project considers how that information asymmetry influences care by comparing the healthcare used by Medicare patients who formerly worked as medical professionals (doctors and nurses) to the healthcare used by socioeconomically similar Medicare patients who did not work in medical professions (lawyers). We will consider differences in the use of many specific healthcare services, such as emergency visits, hospital care, chemotherapy, drug adherence, palliative care, diagnostic testing, and opioid use. We will also compare other aspects of care, such as provider choice (e.g., hospital quality, specialist use, travel time), and Medicare Advantage and Part D plan choice. The study will be the most comprehensive assessment to date of the role of patient expertise, or lack thereof, in healthcare decision making in the United States.

2022. David C. Chan. Beliefs, Self-Interest, and Altruism in COVID-19 Vaccination.

This project explores the determinants of vaccine take-up among veterans in the United States. Using a combination of observational data, survey data, and informational treatments, we aim to disentangle the economic roots of vaccine hesitancy into self-interested concerns about the benefits and risks of vaccination, altruistic concerns about protecting others, and beliefs that social networks and politics may influence. A new survey will elicit information on political beliefs, social networks and responses to hypothetical benefits and risks of a vaccine. In addition to analyzing the survey data independently, we will explore potential linkages between these data and the VA Corporate Data Warehouse, which contains a wealth of information on veterans’ health care utilization, diagnoses, and health outcomes. We will also explore spillovers of beliefs about COVID-19 vaccination onto beliefs and behavior related to flu vaccination.

2022. Engy Ziedan. Drivers of Racial Segregation in Physician Services: Implications for Health Outcomes Equity.

This project examines whether segregation by patient race and other demographics occurs in the context of primary care; and then considers the implications of segregation for disparities in health care use and outcomes. It documents the allocation of Medicare beneficiaries and other patients by race across primary care providers and, if we find racial segregation exists, we will explore contributory factors. We will determine the extent to which the delivery of physician services is racially segregated, and whether the degree of segregation has changed over time. We will identify the extent to which racial segregation varies across geographic areas and explore the drivers of those geographic differences. We will then evaluate whether racial segregation in the delivery of primary care is driven by the geographic location of physician practices and patient residences; or whether patient selection across providers, conditional on geographic location, contributes to racial segregation.

2022. Gopi Shah Goda. The Impacts of COVID-19 Illnesses on U.S. Workers.

This project investigates whether the post-acute sequelae of COVID-19, colloquially referred to as “long COVID” and involving lingering symptoms from COVID-19 illness, has meaningfully depressed labor force participation. To overcome the limited data on long COVID, we construct an indicator of COVID-19 illness, based on the 60 percent increase in work absences due to illness since the start of the pandemic, and the high correlation between illness-related absences and reported cases of COVID-19. Because respondents in the monthly CPS are surveyed on a rotating basis over the course of 16 months, it is possible to follow individuals with illness-related work absences over a period of approximately one year, allowing for a longitudinal analysis of short- and medium-term consequences of suspected COVID-19 illness on labor market outcomes. We aim to quantify the causal link between illness-related work absences and labor market/disability outcomes using an event-study approach that estimates within-person changes in labor market behavior in the periods prior to and following absence.

Priority Research Areas: The Economics of COVID-19

2021. Jonathan Gruber. Assessing the Impact of The False Claims Act on Healthcare Delivery.

The US federal government prosecute most health care fraud and abuse cases under a privatized whistleblowing law called the False Claims Act (FCA). Under the FCA, individuals who have private information about misreporting or overbilling to the Medicare program can sue providers directly in federal civil court. The Department of Justice participates in many successful cases, but whistleblowers can also pursue cases on their own. Since the 1986 enactment of the FCA, the FCA has produced over 7,000 healthcare whistleblower lawsuits, from which the government has recovered more than $32 Billion.
A number of FCA cases have targeted hospices for admitting ineligible patients. Medicare covers hospice for patients with a life expectancy less than 6 months. In 2018, the hospice program served more than 1.5 million Medicare beneficiaries, including more than half of all decedents, spending $19.2 billion. Whistleblowers and the Department of Justice have pursued cases against many for-profit hospice chains for admitting patients who are not terminally ill, including many Alzheimer’s and dementia patients. They charge that administrators pressured clinicians to admit patients and punished or fired those who refused. Long-stay patients may be particularly profitable because they require less ongoing care.
The objectives of our analysis are to 1) estimate differences in admission patterns between defendant hospices and controls and 2) estimate the impact of hospice admission for the marginal patient on life expectancy and Medicare spending. We will identify estimates based on Medicare beneficiaries’ geographical proximity to a defendant hospice and hospices’ proximity to an annual cap on reimbursements, which is determined in-part by random variation in the length of stays of previously admitted patients.

2021. Kate Bundorf. Health Care System Transformation and COVID-19: The Case of Medically Complex Older Adults.

This proposed pilot project is part of a broader research agenda examining how the movement toward value-based care in North Carolina impacts low-income, disabled, middle-aged adults (45-64) covered by Medicaid. North Carolina currently has a small number of innovative care models that were implemented through Medicaid waivers with the goal of delivering more patient-centered, coordinated care for this population. In addition, the State is on the cusp of transitioning its Medicaid population to mandatory managed care in July of 2021. Our project will have two aims based on this policy environment, positioning our team for a larger grant.

Aim 1: To determine the extent to which non-COVID-related care changed during the COVID-19 pandemic for community-based, low-income adults with Medicaid in North Carolina and the implications for health outcomes.

Aim 2: To evaluate the feasibility of a future study of the effects of the mandatory transition of non-dual Medicaid enrollees to MMC in North Carolina on health care costs, use and outcomes.

2021. Nicholas W. Papageorge. Genes, Alzheimer’s Disease and Related Dementias (ADRD) and Economic Outcomes.

Recent advances in behavioral genetics have led to the discovery of specific genetic variants that predict complex health outcomes, such as Alzheimer’s disease (AD) (Lambert et al, 2013; Kunkle et al, 2019). We aim to study associations between these variants, summarized as a linear index known as a polygenic score, and economic outcomes. Loosely, a higher polygenic score for AD (henceforth: the AD score) means an individual possesses more of the genetic variants associated with AD. We will also study one particular marker – APOE ε4 – separately given its substantial established links to AD. To examine relationships between genetic measures and individual outcomes, we will use data from the Health and Retirement Study (HRS), which follows a nationally representative sample of U.S. adults above age 50. The data include detailed health and economic variables for older Americans. Moreover, a large portion of the sample has been genotyped, which means the AD score and other genetic measures can be computed for thousands of individuals and paired with detailed lifecycle health and economic information. The overarching goal of the proposed research is to understand how a genetic predisposition for AD affects lifecycle behavior and outcomes; through which particular channels this occurs; and, ultimately, what policies or measures could mitigate the potential detrimental effects of a genetic predisposition for AD. Our study has three specific aims.
• We will estimate descriptive relationships between our genetic measures for AD, the development of Alzheimer’s disease and related dementias (ADRD), and economic behavior and outcomes, including employment, labor income, retirement age and wealth accumulation.
• We will explore mechanisms that explain any empirical associations between genetic variants related to AD and economic outcomes. For example, we find a negative relationship between genetic variants for AD and wealth at retirement. Additional analyses can help determine whether this relationship is due to medical expenditures, poor financial and investment decisions, or other factors.
• We will provide initial evidence on whether policies or institutional features, such as defined benefit pension (DB) schemes or default long-term care insurance (LTCI) provision, could potentially counteract negative consequences of genetic variants related to AD.

2021. Yu-Chu Shen, Renee Y. Hsia. Does the Expansion of Hospitals with Percutaneous Coronary Intervention (PCI) Capacity have Differential Effects on Procedure Volume and Outcomes for Heart Attack Patients in Advantaged and Disadvantaged Communities?.

Using all-payer patient discharge data from California between 2006 and 2019, we propose to look at the following:
• Has the probability of being treated at low PCI-volume hospitals changed over time differentially between hospitals serving disadvantaged and advantaged communities?
• Have disparities in treatment and health outcomes between those treated in low-volume and high-volume PCI hospitals grown over time?
• Does the extent of disparities in treatment and health outcomes between those treated in low-volume and high-volume PCI depend on whether the patients are in disadvantaged or advantaged communities?

2021. Yu-Chu Shen, Renee Hsia. Access Disparities in the Adoption of Stroke Treatment Technology.

This pilot project explores geographic variations in the adoption of stroke treatment technology across hospitals,
and the extent to which more rapid adoption in some hospitals may exacerbate health disparities between
advantaged and disadvantaged communities. The study focuses on the dissemination of advanced stroke
treatment capabilities across hospitals between 2008 and 2019; and how race, ethnicity, income, population
density, and geography relate to a patient’s access to different levels of care.

Priority Research Areas: Health Trends and Disparities

2020. Kosali Simon. Public Reporting of NH Antipsychotic Use: Changes in the Reporting of Exclusionary Diagnoses?.

In this proposed pilot study, we aim to examine the use of exclusionary diagnoses around the 20-resident threshold for the long-stay APM quality measure to understand whether NHs have been gaming CMS’ public reporting system for APMs. The ultimate goal is to collect preliminary data in order to write a larger NIH funded grant proposal that will include Medicare/Medicaid claims data that document diagnoses outside of the NH (e.g.
physicians and hospitals) and to study the issue of gaming for other quality measures reported on NH Compare. We hope to next link both the intended and unintended effects of this policy on APM access to the health outcomes of NH residents, to better understand how these policies affect health at later life.

Priority Research Areas: The Dynamics of the Health Care Ecosystem

2020. Mireille Jacobson, Julie Zissimopoulos. The Role of Medicare's Annual Wellness Visit in the Assessment of Cognitive Health.

Dementia and mild cognitive impairment (MCI) are widely considered under-diagnosed, with some estimates suggesting that between 15 and 40% of dementia cases and an even higher share of MCI cases are undiagnosed (Taylor et al., 2009; Amjad et al., 2018). Clinical diagnosis often occurs late in the disease trajectory, hindering timely treatment of reversible causes of memory loss. Late diagnosis also complicates a patient’s need to
develop clear and consistent medical, legal, and financial plans.
Medicare payment policy for the routine assessment of cognitive impairment in the primary care setting is a potentially potent tool to improve early detection. As part of a new benefit created under the Affordable Care Act (ACA), Medicare now covers an Annual Wellness Visit (AWV) that requires, among other things, an assessment to detect cognitive impairment. Providers are given little guidance on how to do this assessment beyond “direct observation” and, if appropriate, a “brief validated structured cognitive assessment.”1 This assessment is just one of many required components of the AWV, which should include a health risk assessment, an enumeration of current providers and medications, a depression screening, more routine height and weight measurement,
and so on. Importantly, providers bill Medicare for the visit as a whole and are not required to submit documentation of specific details of the visit.
Perhaps not surprisingly, anecdotal evidence suggests that many physicians do not follow the AWV structure. Providers are more inclined to discuss issues such as the management of high blood pressure or vaccinations at the expense of screening for memory loss or depression.2 In addition, most providers, who are reimbursed on a fee-for-service basis and thus get paid irrespective of outcomes, do not face strong incentives to ensure followup
testing is performed if a cognitive test reveals memory deficits.
A few recent studies have analyzed take-up of the AWV benefit (Ganguli et al. 2017), its relationship to use of preventive services and depression screening (Jensen et al. 2015; Pfoh et al. 2015), and its correlation with measures of cognitive care (Fowler et al. 2018; 2019 Alzheimer’s Disease Facts and Figures Special Report on Detection in the Primary Care Setting). Yet, our understanding of whether the AWV has increased assessment for cognitive impairment in the Medicare population; how assessments get done; and what happens if someone shows signs of impaired cognition is extraordinarily limited. The proposed pilot will seed a broader project to fill in the gaps in our understanding of the AWV and the required cognitive assessment component of this visit. The specific aims of the broad project are to:
• Aim 1: Characterize who gets a cognitive assessment in Medicare during an AWV
• Aim 2: Determine why some beneficiaries get screened while others do not
• Aim 3: Describe health care service use after a beneficiary screens positive for cognitive impairment
• Aim 4: Develop and test a nudge to improve take-up and/or targeting of cognitive assessments to appropriate demographic groups

Priority Research Areas: Cognitive Aging and the Demography of Dementia

2020. William Evans. Evaluating the Ability of a Senior Companion Program to Improve Senior Health.

This proposal is for a planning grant for a randomized controlled trial that evaluates the effect on health of senior companion services. This study will be a collaboration between researchers at the Wilson Sheehan Lab for Economic Opportunities (LEO) at the University of Notre Dame and a partnering social service agency. The project development team will vet providers interested in conducting a rigorous evaluation and form a research
partnership with the organization best fit for such an evaluation. The search for potential organizations will be conducted by LEO with the assistance of our partner organizations. LEO coordinates with local social service agencies to conduct random assignment experiments of their innovative programs designed to assist those in poverty. LEO has partnered with three national networks (Catholic Charities USA, Lutheran Services of America and Goodwill Industries) and we plan to work with these organization to help vet potential partners. Organizations will be considered fit to participate in research if they meet the following three criteria. First, the partner must currently operate a large and active SCP geared toward low-SES individuals. Second, the organization’s SCP must exhibit an excess demand for services. This can be gauged by the length of the program’s current waitlist.
Finally, the organization must be willing to use a lottery to allocate SCP services in place of a waitlist. In our experience, partnering with an organization that meets these three criteria will ensure a successful rigorous study. During this planning stage, our research team will also design and program two surveys that will be used during the full evaluation. The first is the intake survey that will be used during the enrollment process and the
second is the follow-up survey that will accompany the evaluation.
The ultimate objective of this project is to provide precise estimates of the impact of SCP programs on seniors’ health status and independent living. For the purposes of this study, independent living and health status will be measured using both self-reported surveys and administrative data on emergency department and in-patient admissions. The randomized controlled trial (RCT) that will provide senior companions to a sample of around 200 predominantly low-SES homebound seniors. An additional 200 seniors will comprise the control group. This study pool will be recruited from the partnering organization’s current waitlist. Following informed consent protocols appropriate to the study, clients choosing to participate will enter a lottery to determine their placement in the treatment versus control group. which will determine if they will receive a study senior companion. Ideally,
the partnering organization would currently operate a lengthy wait list (e.g., a two-year wait to enter the program); this way, clients selected to the control group can expect a waiting period as long as clients who choose not to participate in the study. LEO has extensive experience working with social service providers. They have 30 active/completed projects in 25 different locations across the country. Their service partners have included social service organizations, county and city governments, and juvenile justice centers.

Priority Research Areas: Determinants of Health, Well-Being and Longevity

2020. Janet Currie. New Modes of Health Care Delivery—Effects on Patients and Providers.

We live in an era of rapid changes in the structures and mechanisms of health care delivery. One striking change has been the development of new types of outpatient facilities including retail clinics, urgent care centers, and ambulatory surgical centers. We aim to examine the impact of these non-traditional care settings on the on the utilization of care and on patient outcomes among the elderly. These new care settings may serve as substitutes or complements for care provided in more traditional settings, such as emergency, inpatient, and primary care office settings. They may therefore provide an opportunity to reduce cost and improve access to care. However, they may also make it more likely that a patient seeks care in a setting that is not optimal for their condition,
which could lead to higher costs and worse outcomes. We will use individual medical histories from longitudinal Medicare claims data to account for underlying differences in patient health status, and we will exploit variation in the closing and opening of facilities in order to study the way that patients are selected into different modes of care, substitution patterns across modes, and impacts on patient outcomes.
A second related objective will be to study the impact of new modes of medical care delivery on traditional providers. At the facility level, does increased competition from new providers change merger and acquisition patterns or the types of services offered? At the individual physician level, can we see impacts in terms of where they practice, or the quality of medical decision making? Because it has data on patients, physicians, and facility
type, the Medicare claims data is well suited for this type of analysis. The project should shed light not only on the specific issue of how new care modes help or harm Medicare patients, but also on the more general question of whether increasing competition in a setting that is still far from perfectly competitive improves welfare.

Priority Research Areas: The Dynamics of the Health Care Ecosystem