Integrating Medicare And Medicaid Data To Improve Care Quality And Advance Health Equity Among Dual-Eligible Beneficiaries
KEY POINTS
|
People who qualify for Medicare and Medicaid on the basis of a combination of low-income status and age, disability, or end-stage renal disease, also known as dual eligibles, remain a focus of research and policy making because of their complex insurance coverage and high health care costs. Nationally, there are approximately twelve million dual eligibles: approximately nine million with full Medicaid (which covers Medicare’s out-of-pocket costs and services such as long-term care) and three million with partial Medicaid (which covers the Medicare Part B premium and, in some cases, cost sharing). Although dual eligibles only represent 19 percent of Medicare enrollees and 14 percent of Medicaid enrollees, they account for 34 percent and 30 percent of Medicare and Medicaid spending, respectively. This is because dual eligibles are socially and medically at risk and, thus, have costly and complex health care needs. More than two-thirds of dual eligibles have three or more chronic conditions, 41 percent have a behavioral health disorder, and 46 percent are Black, Hispanic, or other people of color.
A long-standing concern has been that coverage for dual eligibles is complex and is poorly configured to equitably address this population’s needs. Specifically, Medicare is the primary payer for outpatient services, hospital care, and prescription drugs for dual eligibles, whereas Medicaid pays for long-term care, including home and community-based services and care in nursing facilities, with few financial or administrative mechanisms to coordinate care between the two programs. As such, dual eligibles may receive care that is fragmented, inefficient, and inequitable.
Policy makers continue to test new models to integrate Medicare and Medicaid coverage for dual eligibles, with the goal of improving the quality and efficiency of care. These models include Medicare-Medicaid Plans, Medicare Advantage Dual-Eligible Special Needs Plans (D-SNPs), and new several subclasses of Medicare Advantage D-SNPs that attain different levels of integration with Medicaid (coordination-only, highly integrated, and fully integrated Medicare Advantage D-SNPs). The Medicaid and CHIP Payment and Access Commission has called for refinement of these models and for an explicit focus on equity in integration policy.
However, researchers’ ability to evaluate integrated models and inform equity-focused policy remains a challenge because few data resources comprehensively measure needs for, use of, or spending on care among dual eligibles. Existing data sources also provide an incomplete picture of patient needs for or experiences with care in areas that are most salient to this population (for example, coordinating hospital care and long-term care). Further, some data sources are limited in the extent to which they enable analysts to identify underserved subpopulations of dual eligibles. A growing body of research documents substantial health care disparities within the dual-eligible population, making it critical to measure equity of care among different subpopulations of dual eligibles.
In this brief we review existing data resources that can be used to study the dual-eligible population, along with linkages among these data (see exhibit 1). We review the strengths and limitations of these sources. Finally, we identify areas in which dedicated data infrastructure could play a role in informing equity-focused care for dual eligibles.
EXHIBIT 1 Administrative, survey, and assessment data available to study care among people dually eligible for Medicare and Medicaid
|
|
Data set includes measures of… |
Linkage availablea |
Regularly updated |
|||||
|
Race and ethnicity |
Socioeconomic status or social needs |
Unmet care need |
Use of care |
Health care spending |
Health status |
|||
|
Administrative data |
||||||||
|
Medicare fee-for-service files |
○ |
− |
− |
● |
● |
● |
○ |
● |
|
Medicare and Medicaid Linked Enrollee Analytic Data Source |
○ |
− |
− |
● |
● |
● |
● |
− |
|
Medicaid Transformed Analytic Files |
○ |
− |
− |
● |
● |
● |
○ |
● |
|
Medicare Advantage encounter data |
○ |
− |
− |
● |
● |
● |
○ |
● |
|
Survey data |
||||||||
|
Medicare Current Beneficiary Survey |
● |
● |
● |
● |
● |
● |
○ |
● |
|
Medicare Consumer Assessment of Healthcare Providers and Systems Survey |
● |
○ |
● |
○ |
− |
● |
● |
● |
|
Health Outcomes Survey |
● |
● |
− |
− |
− |
● |
○ |
● |
|
Health and Retirement Study |
● |
● |
● |
● |
● |
● |
○ |
● |
|
National Health and Aging Trends Study |
● |
● |
○ |
○ |
− |
● |
● |
● |
|
Assessment data |
||||||||
|
Minimum Data Set |
● |
○ |
● |
● |
− |
● |
● |
● |
|
Home Health Outcome and Assessment Information Set |
● |
○ |
● |
○ |
− |
● |
● |
● |
Source: Authors’ analysis. Notes: A filled circle denotes “yes,” an open circle denotes “yes, but limited,” and a minus sign denotes “no.” aData support linkage to other sources of Medicare and Medicaid data.
Administrative data, which include enrollment data, fee-for-service health care claims, and managed care encounter records, are key data sources for examining care in Medicare and Medicaid. These data capture information about use of care, diagnoses, and some demographic information. They also identify enrollment in managed care plans, including integrated models for dual eligibles. Although Medicare and Medicaid administrative data both include dually eligible enrollees, data for each program are constructed and distributed separately, which can make it difficult to analyze the full range of care received by dual eligibles. Moreover, because Medicaid is administered by states, Medicaid data quality can vary from state to state.
Administrative Data Linkage And Integration
The Centers for Medicare and Medicaid Services (CMS) and researchers have long explored strategies and opportunities to integrate Medicare and Medicaid data to assess outcomes and improve care for dual eligibles. The Chronic Conditions Data Warehouse serves as a repository for national Medicare and Medicaid administrative data. However, linkages of Medicare and Medicaid administrative data are difficult to implement, often requiring complex, multistep matching algorithms to analyze linked files.
The only Chronic Conditions Data Warehouse data set explicitly designed to link Medicare and Medicaid administrative data for dual eligibles is the Medicare-Medicaid Linked Enrollee Analytic Data Source. The data source provides longitudinal (2006–12) demographic, enrollment, chronic condition, service use, and spending data for dual eligibles. However, Medicare-Medicaid Linked Enrollee Analytic Data Source files have not been updated continuously since 2012, and a more recent version of the data source, from 2016, lacks many variables from earlier years. At present, the future of the Medicare-Medicaid Linked Enrollee Analytic Data Source is unknown.
Recent Enhancements To Administrative Data
Reflecting CMS’s strategy to advance health equity among dual eligibles through improved coordination between Medicare and Medicaid, recent enhancements to administrative data provide new opportunities to analyze care and health care disparities among dual eligibles. In 2020 CMS introduced new Medicaid Transformed Medicaid Statistical Information System Analytic Files, which facilitate analyses of the Medicaid population, including dual eligibles. CMS has also begun to make Medicare Advantage encounter data available to researchers. Medicare Advantage data are particularly critical for studying care among dual eligibles, given that approximately 40 percent of dual eligibles are enrolled in Medicare Advantage plans, including Medicare Advantage D-SNPs. In addition, these Medicare and Medicaid administrative files are now linkable at the person level on a common encrypted identifier from the Chronic Conditions Data Warehouse—a relatively new capability resulting from the availability of Transformed Medicaid Statistical Information System Analytic Files in the Chronic Conditions Data Warehouse Virtual Research Data Center. Importantly, this data enhancement enables more comprehensive analyses of care among dual eligibles.
Administrative Data Quality And Limitations
Despite CMS’s recent enhancements to administrative data sources, challenges remain. For example, the quality of Transformed Medicaid Statistical Information System Analytic Files data varies across states. These data quality concerns can limit analysts’ ability to conduct state-by-state comparisons of care where Medicaid is the primary payer for dual eligibles (for example, home and community-based services). In addition, one analysis found that Transformed Medicaid Statistical Information System Analytic Files do not reliably identify dually eligible individuals in all states. It also found that Transformed Medicaid Statistical Information System Analytic Files alone could not be used to identify enrollees of integrated plans (identifying enrollees of these plans required linkage to Medicare data).
Administrative data also present challenges for analyzing health care disparities among dual eligibles. Specifically, race and ethnicity data are considered to be of concern or “unusable” in several states’ Transformed Medicaid Statistical Information System Analytic Files administrative files. The quality of Medicare administrative data also remains a concern, as race and ethnicity data are imputed for a large proportion of Medicare enrollees. These imputation methods have known limitations for identifying racial and ethnic groups that have proportionately smaller representation in the population (for example, American Indian or Alaska Native enrollees), thus limiting reliable analyses of health care disparities in some subpopulations of dual eligibles.
Last, administrative data are limited in their ability to allow analysts to identify underserved subpopulations. Specifically, these data only capture health information for people who use care, making it difficult to identify individuals with unmet care needs. Moreover, administrative data provide little information about health-related social needs (for example, housing insecurity, food insecurity, social support) that influence access to care and contribute to health care disparities within the dually eligible population.
Several surveys of individuals with Medicare, including dual eligibles, capture demographic and health characteristics, as well as patient-reported outcomes such as access to care and unmet care needs—information not reported in administrative data. We describe five surveys available at different levels of granularity and with different administrative data linkage capabilities.
Medicare Current Beneficiary Survey
The Medicare Current Beneficiary Survey (MCBS) is a national survey of approximately 14,000 individuals with Medicare per year. Participants are followed up to four years and new cohorts are added annually. Strengths of the MCBS include its collection of demographic data (including respondent-reported race and ethnicity) and respondent-reported measures of physical and mental health, access to care, and satisfaction with care. A redesign of the survey in 2015 resulted in the collection of additional socioeconomic variables, including food insecurity and detailed income data.
The MCBS is available in two formats: a public use file, which contains a condensed set of variables and aggregates response categories for some variables (to protect respondent confidentiality), and a limited data set containing detailed variables. The limited data set includes linkages to insurance variables from administrative data. These variables distinguish full versus partial Medicaid enrollment and identify enrollment in Medicare Advantage plans, including Medicare Advantage D-SNPs. These data have enabled researchers to compare care for dual eligibles across different forms of Medicare coverage and by race and ethnicity. The limited data set can also be linked to the MCBS Cost Supplement file, which reports detailed health care expenditures for a subset of respondents.
The MCBS has some limitations for studying dual eligibles. First, small samples preclude analyses of some subpopulations, including subclasses of Medicare Advantage D-SNPs attaining different levels of Medicaid integration. Second, the survey does not assess aspects of care that may be important for dual eligibles (for example, transitions of care between hospitals and nursing facilities). Third, although MCBS includes some variables from administrative data, researchers cannot link MCBS to complete Medicare or Medicaid administrative data.
Consumer Assessment Of Healthcare Providers And Systems Surveys
Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys focus on Medicare enrollees’ experiences with care. Separate but closely related surveys are administered to enrollees in traditional Medicare and Medicare Advantage plans (including Medicare Advantage D-SNPs). CAHPS surveys assess patient experiences with care in several domains, including provider communication, care coordination, and access to care. Some items capture salient aspects of care for dual eligibles (for example, help managing care among different providers and services). The survey also collects respondents’ demographic characteristics (including race and ethnicity), health and functional status, and immunizations.
CAHPS survey data are distributed as research identifiable files linkable to Medicare administrative data. These linkages enable researchers to identify enrollees in Medicaid and Medicare Advantage plans (including Medicare Advantage D-SNPs), and thus are useful for examining variations in care among dual eligibles by type of Medicare coverage. CAHPS surveys can also support comparisons of performance between Medicare Advantage D-SNPs and—for sufficiently large plans—analyses of plan-level disparities by race and ethnicity. Such analyses could enable researchers to identify features of Medicare Advantage D-SNPs associated with more equitable care among dual eligibles.
Although CAHPS surveys are widely used to examine patient-reported quality of care, these data support more limited assessments of salient aspects of care for dual eligibles (for example, coordination of hospital care and long-term care). Further, CAHPS surveys collect a comparatively smaller set of socioeconomic variables than other Medicare surveys (for example, they do not assess income).
Medicare Health Outcomes Survey
The Health Outcomes Survey is a survey of individuals enrolled in Medicare Advantage plans and Medicare-Medicaid Plans that have least 500 members. The survey collects an extensive set of patient-reported mental and physical health measures, including functional status (for example, difficulty performing activities of daily living) and frailty (for example, balance and falls). The Health Outcomes Survey interviews participants in a baseline year and again two years later, enabling monitoring of health changes. CMS uses frailty data collected in the Health Outcomes Survey to risk-adjust payment rates to Medicare-Medicaid Plans, certain integrated Medicare Advantage D-SNPs serving enrollees with a high level of frailty, and Programs of All-Inclusive Care for the Elderly, which serve frail community-dwelling individuals, most of whom are dual eligibles.
The Health Outcomes Survey collects information about demographics (race and ethnicity) and some socioeconomic factors and health-related social needs (for example, education, income, and living circumstances). However, it asks fewer questions about patients’ care from providers, and it does not directly assess patient-reported access to care or care coordination.
The Health Outcomes Survey is available as a limited data set, which incorporates some administrative variables that identify Medicaid, Medicare Advantage, and Medicare Advantage D-SNP enrollment.
Health And Retirement Study
The Health and Retirement Study is a national longitudinal survey of US adults ages fifty and older. The survey includes cohorts of older adults living in the community who are followed over time through biennial surveys (follow-up continues if people move into a nursing facility). The 2020 survey includes seven cohorts of people born between 1920 and 1965. Across cohorts, each survey wave includes approximately 20,000 people. Black and Hispanic adults are oversampled.
The Health and Retirement Study does not exclusively sample individuals who are enrolled in Medicare or those who are dually eligible. However, because of the survey’s focus on older adults, it disproportionately includes Medicare enrollees. More than 80 percent of respondents with Medicare provide their beneficiary identification number to facilitate linkage of the survey to administrative data.
A distinguishing feature of the Health and Retirement Study is its collection of an extensive set of demographic, economic, and health variables. Demographic measures include respondent-reported race and ethnicity and education. The Health and Retirement Study asks respondents to report sources of income and wealth (usually within ranges). The RAND Corporation developed imputation methods to estimate household-level income and wealth from these data, which have been used by researchers to investigate issues such as Medicaid enrollment among low-income individuals with Medicare. Imputed income and wealth variables are provided in a public use file, known as the RAND Health and Retirement Study Longitudinal File, which harmonizes variables across survey waves.
The Health and Retirement Study also includes an array of health, cognitive status, and functional status measures, assessed through survey questionnaires, screening items (for example, telephone interview for cognitive status), and physiologic assessments (for example, blood samples). Further, the Health and Retirement Study collects information about insurance coverage, patient-reported access to care, and out-of-pocket medical spending.
Restricted use files can be linked to some Medicare and Medicaid administrative data. Linked Medicare claims for inpatient, outpatient, and professional services are available for people in traditional Medicare, but are not currently available for Medicare Advantage. Linkages to Medicaid data are available through 2012, but linkage to more recent Transformed Medicaid Statistical Information System Analytic Files data are not currently available. However, because the Health and Retirement Study includes adults ages fifty and older, it only captures a subset of the younger dual-eligible population (ages 21–64), which is an important group because these individuals qualify for Medicare on the basis of the presence of a disability or end-stage renal disease.
National Health And Aging Trends Study
The National Health and Aging Trends Study is a national longitudinal survey of Medicare enrollees ages sixty-five and older designed to measure physical and cognitive functioning, economic and social well-being, and end-of-life care. An initial cohort of approximately 12,000 participants has been surveyed annually since 2011; a second cohort of approximately 7,000 participants was added in 2015. The survey oversamples individuals enrolled in Medicare who identify as Black.
The study collects an expansive set of health, function, and psychosocial measures, including disease history, self-care, mobility, and participation in social activities. One unique feature of the National Health and Aging Trends Study is its inclusion of an end-of-life survey module, administered to family members of decedents, which examines topics such as symptom management and emotional support. The survey also assesses demographic characteristics (including race and ethnicity) and economic indicators (including food insecurity, income, and savings).
Restricted-use National Health and Aging Trends Study files can be linked to Medicare and Medicaid administrative data. Linkages are available for Medicaid Transformed Medicaid Statistical Information System Analytic Files, traditional Medicare, and Medicare Advantage data, enabling researchers to construct a more complete picture of care than is currently feasible with the Health and Retirement Study. However, a limitation of the National Health and Aging Trends Study is that it cannot be used to analyze the dual-eligible population younger than sixty-five.
Two national standardized assessments provide data to study people receiving care in nursing facilities or home health care. These data are important for studying care among dual eligibles because long-term services and supports and home and community-based services account for approximately 75 percent of this population’s Medicaid spending. Moreover, assessment data collect information on all individuals receiving care in these settings, irrespective of payer, thus capturing dually eligible individuals enrolled in traditional Medicare and Medicare Advantage plans.
Minimum Data Set
The Minimum Data Set is a standardized assessment of all residents of CMS-certified nursing facilities. Assessments are completed by facility staff once every three months, or more frequently as a result of charges in patient circumstances. The Minimum Data Set collects a rich set of clinical measures that track the health, functional and cognitive status, and psychosocial well-being of residents—for example, diagnoses, mobility, geriatric syndromes, and participation in social activities. The Minimum Data Set includes information about residents’ receipt of therapy services, such as physical or speech therapy. These assessments of functional status and needs for therapeutic services underlie CMS’s new Patient Driven Payment Model for nursing facility care.
The Minimum Data Set also captures information about care planning—for example, if residents discussed leaving the facility to be cared for in the community—as well as end-of-life care. Patient demographics, including race and ethnicity, are included. The Minimum Data Set can be linked to Medicare and Medicaid administrative data.
Home Health Outcome And Assessment Information Set
The Outcome and Assessment Information Set is designed to measure patient health, needs for care, and health outcomes among individuals receiving home health care from CMS-certified home health providers. Assessments are completed at sixty-day intervals by providers and are based on evaluations of patients and interviews with patients or caregivers. The Outcome and Assessment Information Set captures demographic variables including race and ethnicity, diagnoses, assessments of physical and cognitive function, and information about support from caregivers. The Outcome and Assessment Information Set files can be linked to Medicare and Medicaid administrative data.
Outcome and Assessment Information Set data primarily capture postacute home health care services (for example, care after a hospital stay). However, other home and community-based services used extensively by dual eligibles, such as personal assistance services, are not captured.
Administrative, survey, and assessment data provide opportunities to examine—and ultimately advance—health equity among dual eligibles, yet these data sources have key limitations. However, there are opportunities to address these limitations.
First, administrative data sources could be improved through efforts to collect complete and validated race and ethnicity data. For example, CMS and states could improve collection of race and ethnicity data directly from individuals to reduce missing data and the reliance on algorithms to impute race and ethnicity. Second, integrating measures of health-related social needs into administrative data could aid in better understanding and addressing these needs among dual eligibles, including subpopulations of dual eligibles with different health-related social needs. Under a recently finalized rule, CMS will begin requiring all SNPs, including Medicare Advantage D-SNPs, to include within their annual health risk assessment screenings for housing, food, and transportation insecurity. Incorporating these assessments into administrative data could enable analyses to evaluate how well Medicare Advantage D-SNPs address enrollees’ health-related social needs. Third, further work is needed to validate Medicare Advantage encounter and Medicaid Transformed Medicaid Statistical Information System Analytic Files data and assess the reliability of patient-level linkages across files over time. Fourth, work is needed to standardize how Transformed Medicaid Statistical Information System Analytic Files capture Medicaid-covered services used extensively by dual eligibles (such as home and community-based services) to support reliable state-by-state comparisons of service use. Finally, development of a comprehensive linked Medicare and Medicaid database for dual eligibles, similar to the Medicare-Medicaid Linked Enrollee Analytic Data Source but incorporating Medicare Advantage, Transformed Medicaid Statistical Information System Analytic Files, and assessment data, could help make these data sources more accessible to researchers.
There are also opportunities to improve survey data. First, oversampling of dual eligibles in Medicare surveys, including dual eligibles from traditionally underserved subpopulations (for example, Black, Hispanic, or other populations) and those in integrated coverage models, could support more robust analyses of health equity in these models. Second, surveys of dual eligibles could be tailored to focus on care needs that are often unique to this population (for example, transitions of care between acute and long-term care settings). Finally, survey items could be added to explore whether dual eligibles have unmet care coordination needs.
Although the Improving Medicare Post-Acute Care Transformation Act of 2014 requires CMS to use existing assessment data instruments to collect standardized data from long-term care facilities and home health agencies, the process of developing standardized data elements to reflect the heterogeneity of the patient population is still underway. This standardization process should include the development of data elements specific to dual eligibles—for example, availability of housing and social supports that influence whether dual eligibles can be discharged successfully from a long-term care facility into the community. Moreover, assessment data should include more in-depth measures of home and community-based services, such as personal assistance services, used by dual eligibles.
Policy makers are pursuing reforms to provide more equitable care to dual eligibles, which is a priority, given this population’s medical and social risk factors. However, the limitations of existing data resources have made it difficult to comprehensively study care among dual eligibles and to evaluate whether models that integrate coverage for dual eligibles are successful in advancing health equity. Improvements to administrative, survey, and assessment data could expand researchers’ opportunities to study and, ultimately, inform equity-focused improvements in care among dual eligibles.
All briefs go through peer review before publication.