Contextual Factors in Aging Working Group

LTSS delivery and financing are complex and pose unique challenges to how we understand and design reform efforts. LTSS may be delivered in individuals’ homes, in community settings, may be interwoven within supported housing or residential care facilities, or may drive transitions to long-stay nursing homes. As persons with disabilities often have co-occurring chronic medical conditions, the availability, quality, and coordination with medical care is also a critical consideration. Health outcomes are not only affected by “formal” services that are financed and/or regulated by federal and state programs and private insurers, but by the environment in which individuals “live, learn, work, play, worship, and age.” These environments encompass characteristics of family and social networks, housing, transportation, and land use, among others. Although individuals’ social and economic, service delivery, and built and natural physical environments may contribute to important indicators of LTSS access and quality, including – and perhaps especially among older adults with disability who are socially disadvantaged – little attention has been directed at conceptualizing or comprehensively understanding how the LTSS environmental context affects health and well-being of older adults with disabilities. This directory catalogs contextual measures of the LTSS landscape. 

LTSS Framework 

Measures are organized and constructed around key domains in a conceptual framework of LTSS delivery developed by Fabius et al. (2023).1 This conceptual framework identifies environmental domains that contribute to LTSS use, care quality, and care experiences. Each domain operates at different environmental levels (societal, community context, neighborhood). Domains: 

1. Social and Economic: This domain captures measures of economic and sociocultural factors.  

2. Health Care and Social Services Delivery Domain: This domain captures measures of financing, direct care workforce, family caregiving, and social services. 

3. Built and Natural Physical and Environment Domain: This domain captures measures of transportation and land use, communication systems infrastructure, and housing. 

We use the U.S. Census Geographic Areas Reference Manual (GARM) to define geographic entities of interest for these contextual measures. Geographic entities can represent legal and administrative boundaries related to governance (e.g., states, counties, cities, and congressional districts) or statistical boundaries (e.g., metropolitan statistical areas, ZIP Codes, Census Tracts, Census Blocks). Table 1 defines each geographic area. 

Table 1 – Geographic Entities, Type, and Definition 

Detailed information about each contextual measure is provided on the following pages including: geographic entity, year(s) available, variable type, data source, and approach used to define and create the measure. Measures may be linked to other data sources using the appropriate geographic entity.  

Operationalizing Continuous Measures 

Our recent work2 categorized continuous measures using tertiles. Percentages across geographic entities were divided into tertiles based on our sample from the 2015 National Health and Trends Study. If information for 2015 was not available, data from the closest year was used. For example, states’ No Wrong Door Score information was only available for 2016. Research teams may decide to categorize continuous measures differently, as appropriate for their data. We provide examples of measures where this approach was used.  

Summary of Measures by Domain and Year(s) Available 

Measure: State minimum wage 

Geographic entity: State 

Year(s) available: 2013-2022   

Variable type: Continuous  

Source: Department of Labor, supplemented with the Department of Labor minimum wage history page. 2022 information supplemented by Labor Law Center.  

Why this measure is important: This measure provides a comparison of the state’s minimum wage level to the federal minimum wage level, which could be used as a general measure of each state’s labor position.  

Background and Definitions: Minimum wage is defined differently at the federal and state levels. Federal law sets a minimum wage with the Fair Labor Standards Act which serves as the preempted floor that applies to all states without minimum wage laws. However, the federal minimum wage only applies to employees of businesses with annual gross revenue of $500,000 or more.  The federal minimum wage does not apply to tipped employees, workers with disabilities, students, those under 20 years old, and apprentices. This makes it theoretically possible to pay someone less than the federal minimum wage.  

Each state may set a minimum wage that exceeds the federal minimum wage level, subject to exceptions such as employers’ number of employees, annual revenues, and coverage of health insurance.  

Approach: Most information for each state was taken directly from the Department of Labor website. However, not every state uses a single value for their minimum wage. In general, we took the minimum wage values that likely affect the greatest number of people in each state.  For example, in states where the minimum wage law applies to employers of 4 or more, we assumed that to be the minimum wage statewide. In states where the minimum wage law applies to employers of 6 or more, we assumed that to be the minimum wage statewide.  

For states with minimum wages that are lower than the federal wage law, we took the federal wage to be the minimum wage.  

State-specific notes: 

1. Minnesota: There are two sets of minimum wage laws, one that applies to employers with annual revenues of more than $500,000 and one that applies to employers with annual revenues of less than $500,000. If the table from Department of Labor indicated two wages, which was the case for years prior to 2020, the two wages were averaged. For 2020 onwards, the higher wage was used.  

2. Montana: There are two sets of minimum wages, one that applies to businesses with annual sales greater than $110,000 and another that applies to businesses with annual sales less than $110,000. The latter minimum wage is lower than the federal minimum wage, which applies to those that the federal law does not cover. For this, we use the higher minimum wage for all years.  

3. Nevada: There are two minimum wages, one that applies to employers that provide health insurance and another that applies to employers that do not provide health insurance. The two rates are averaged.  

4. Ohio: There are two sets of minimum wage laws, one that applies to those with annual revenues of over $342,000 and one that applies to enterprises with annual revenues of less than $342,000. The two wages are averaged for years 2016 and prior, and the higher wage was taken for years after 2016.  

5. Oklahoma: There are two sets of minimum wage, one that applies to businesses with annual sales greater than $100,00 or business with ten or more full time employees at one location and another that applies to all other employers. The latter minimum wage laws are lower than the federal minimum wage law. Therefore, the higher minimum wage is taken, which is the same level as the federal minimum wage.  

References:  

• US Department of Labor, Wage and Hour Division. State minimum wage laws. Published July 1, 2023. Accessed September 25, 2023. http://www.dol.gov/agencies/whd/minimum-wage/state 

• https://www.laborlawcenter.com/state-minimum-wage-rates/ 

Measure: Percentage of people living below poverty line in the previous 12 months 

Geographic entity: Census tract 

Year(s) available: 2015 

Variable type: Continuous 

Range: 0-100.00% 

Mean: 15.25% 

Median: 12.09% 

Source: All data downloaded directly from data.census.gov, specifically the B17005 table.  

Why this measure is important: This measure reflects the degree to which individuals living in a geographic area are affected by extreme poverty during a particular year.  

Background and Definitions: The definition of poverty used here is the one measured and defined by the US Census Bureau, detailed here.  Specifically, this metric defines those living below the poverty level as people with income in the past 12 months below a specific poverty threshold based on the size of the family and age of the members.  

The data from the ACS provides further categories of many other factors, including sex, in labor force, and employment status, which can be helpful for any researchers interested in those variables.  

Approach: For each census tract, the total number of “Income in the past 12 months below poverty level” is divided by the total number of people indicated as residing within that geography, generating a continuous variable throughout all the census tracts.  

Missing data: 835 census tracts (1.1%) have missing values for this measure. 

References:  

• US Census Bureau. American Community Survey, Table B17005: Poverty status in the past 12 months of individuals by sex by employment status. 2015: ACS 5-Year Estimates Detailed Tables. Accessed September 25, 2023. https://data.census.gov/table?q=B17005&g=010XX00US$1400000&y=2015&d=ACS+5-Year+Estimates+Detailed+Tables&tid=ACSDT5Y2015.B17005 

• https://www.census.gov/topics/income-poverty/poverty/guidance/poverty-measures.html 

Measure: Percentage of children who receive public assistance in the form of SSI, cash assistance, or SNAP in the previous 12 months.  

Geographic entity: Census tract 

Year(s) available: 2015 

Variable type: Continuous 

Range: 0.00 - 100.00% 

Mean: 28.91% 

Median: 25.00% 

Source: All data downloaded from data.census.gov, specifically table B09010 “Receipts of Supplemental Security Income (SSI), Cash Public Assistance Income, or Food Stamps/SNAP in the Past 12 Months by Household Type for Children Under 18 Years in Households”, 5-year estimate for the year 2015.  

Why this measure is important: The measure is a proxy that captures the neighborhood relative economic depression via the percentage of households with children that receive any type of public assistance.  

Background and Definitions: Children can receive public assistance through both federal and state programs. Supplemental Security Income (SSI) is a federally administered program for people 65 or older and children with disabilities. State-administered programs can vary in eligibility and coverage and include: Temporary Assistance to Needy Families (TANF), general assistance, Women, Infants, and Children (WIC), and the Supplemental Nutrition Program (SNAP). In the ACS, cash public assistance refers to: TANF, general assistance, WIC, and others. For TANF, it is typically only awarded to families with dependent children, and not for single adults. In general, children may be eligible for public assistance even if their parents are not. Children can be eligible for SNAP even if their parents are not eligible.  

According to the population reference bureau, the US Census Bureau defines a household as “all the people who occupy a single housing unit, regardless of their relationship to one another.” 

Approach: The number of children (under the age of 18) who received public assistance in household is divided by the total number of children in the census tract to calculate a percentage for the entire census tract. 

Missing data: 1,131 census tracts (1.5%) have missing values for this measure. 

References:  

• https://data.census.gov/table?q=B09010&g=010XX00US$1400000&y=2015 

• https://www.census.gov/topics/income-poverty/public-assistance/about.html 

• https://www.prb.org/resources/what-is-a-household/#:~:text=Children%2C%20Youth%2C%20and%20Families%20U.S.,their%20relationship%20to%20one%20another. 

Measure: Social Deprivation Index (SDI) 

Geographic entity: Census tract 

Year(s) available: 2015 

Variable type: Discrete continuous  

Range: 1 to 100 

Mean: 51.69 

Median: 53 

Source: We used 2011-2015 American Community Survey (ACS) 5-year estimates for the SDI at the Census-tract level. This measure may be recalculated for any 5-year ACS estimates and different geographies, including counties and ZIP Code Tabulation Areas (ZCTA).  

Why this measure is important: The SDI is a composite measure that captures levels of disadvantage across small geographies and has been used to indicate areas where health care resources are needed. This can be important for measure for understanding differences in health outcomes among older adults and other vulnerable populations.  

Background and Definitions: The SDI has been used to examine how socio-economic factors are associated with health outcomes. 

Approach: The SDI is a composite measure of seven area-level characteristics gathered by the ACS: (1) percent living in poverty, (2) percent with less than 12 years of education, (3) percent single-parent household, (4) percent living in the rented housing unit, (5) percent living in the overcrowded housing unit, (6) percent of households without a car, and (7) percent unemployed adults under 65 years of age. The methods to construct the SDI are described in more detail elsewhere.1  

Missing data: 1,577 census tracts (2.1%) have missing values for this measure. 

References: 

1. Butler DC, Petterson S, Phillips RL, Bazemore AW. Measures of Social Deprivation That Predict Health Care Access and Need within a Rational Area of Primary Care Service Delivery. Health Serv Res. 2013;48(2 Pt 1):539-559. doi:10.1111/j.1475-6773.2012.01449.x 

Measure: Percentage of people in the civilian workforce who are unemployed 

Geographic entity: Census tract   

Year(s) available: 2015 

Variable type: Continuous 

Range: 0.00% to 100.00% 

Mean: 9.02%  

Median: 7.67%  

Source: All data downloaded directly from data.census.gov, specifically the B23025 table.  

Why this measure is important: This measure captures unemployment for the civilian workforce within a particular geography.  

Background and Definitions: There are many ways to calculate unemployment levels in a particular area, as well as many ways to define the labor force. For this measure, the civilian labor force is defined here. The definition of employment is defined here. The choice to use the civilian labor force is three-fold. First, the data from the ACS did not provide unemployment information specific to the total labor force. Second, for most purposes, the civilian labor force which excludes the armed forces is more appropriate for analyses that focus on adequacy of the long-term services and support delivery environment. Third, using the total population as the denominator would be inappropriate as not everyone is in the labor force.  

Approach: The percentage is calculated by dividing the number of unemployed individuals in the civilian labor force by the total number of those in the civilian labor force in every available census tract, generating a continuous number reflecting percentage of unemployment across all the census tracts.  

Missing data: 802 census tracts (1.1%) have missing values for this measure. 

References:  

• US Census Bureau. American Community Survey, Table B23025: Employment status for the population 16 years and over. 2015: ACS 5-Year Estimates Detailed Tables. Accessed September 25, 2023. https://data.census.gov/table?q=b23025&tid=ACSDT5Y2020.B23025 

• https://www.census.gov/quickfacts/fact/note/US/LFE041220 

Measure: The number of home health aides and personal care aides per 1000 jobs  

Geographic entity: Metropolitan Statistical Area and Nonmetropolitan Area 

Year(s) available: 2015 

Variable type: Continuous 

Range: 0.968 to 125.657 

Mean: 16.46 

Median: 13.97 

Source: The main data is sourced from the Bureau of Labor Statistics (BLS) page, namely the Occupational Employment and Wage Statistics data, for a particular year. However, to transform the data set to be linked with a data set with only census tract information, additional supplemental data are needed. The reference file for how metropolitan statistical areas is defined can be found here, also on the U.S. Bureau of Labor Statistics website. This is useful for connecting the FIPS codes to the Metropolitan Statistical Area code (MSA code). Converting from census tract to FIPS state and county code is simple, as the census tract code contains the FIPS state and county information. The New England states (Maine, New Hampshire, Vermont, Massachusetts, Connecticut, and Rhode Island) require an additional resource that converts census tract information into the corresponding township names. The resource we used can be found here, on the Missouri Census Data Center website, specifically we used the Geocorr 2014 version as it was closest to our 2015 data set. Select the six New England states and the corresponding source and target geography to generate the reference file.  

Why this measure is important: This measure captures the supply of the direct care workforce present in an MSA and non-MSA. This measure describes the availability of direct care workers in a particular geographic area to support older adults with LTSS needs. 

Background and Definitions: We defined direct care workers in OEWS data tables using two occupations: Home Health Aide and Personal Care Aide. This measure reports the number of direct care workers per 1000 jobs in a given MSA or non-MSA. The data source defines “jobs” as any full-time or part-time job in the area.  

Approach: To calculate the number of direct care workers per 1000 jobs, we summed up the number of home health aides and personal care aides for each MSA and non-MSA.  The count of the number of people in those two occupations may be summed or averaged for an estimate of the number of direct care workforce in that MSA, but typically summation is the approach to use, as there are many areas in which one occupation is missing data or has no count. 

Occupational Employment and Wage Statistics data (OEWS) provided by the bureau of labor only provides information at the metropolitan and nonmetropolitan statistical area level, meaning any data with other geographical definitions will first need to be converted to the MSA or non-MSA level. Mapping the census tracts to MSA and non-MSA areas is straight forward except for the six New England States. Those six states have archaic township designations that do not align perfectly with MSA definitions and require adjustment. In short, New England townships may span counties and MSA designation, meaning an intermediary township match must be conducted first to accurately map any census tract data into the correct MSA code.  

The following description applies to linking census tract level data to OEWS data. Processes for MSA and non-MSA areas are the same. 

The first step involves separating the data set into two files, Non-New England States and New England States. The New England States must be treated separately as the census tract information for those states does not match completely with the MSA areas the OEWS uses. The linking process for the Non-New England States and the New England States are distinct and will be described separately.  

For the Non-New England states, the FIPS state and county code (5 digits) can be isolated from the census tract number. After the FIPS codes are isolated, they can be mapped onto their corresponding MSA code via the reference file from the Bureau of Labor Statistics. With the matched MSA codes, the data set can then be merged with the OEWS data files from the BLS.  

For New-England states, the census tract is used instead of the five-digit FIPS code. Using the reference file from the Missouri Census Data Center, each census tract can be matched with the corresponding township names. From the township names, they are then matched with the information in the MSA reference file for the corresponding MSA codes. Note that there are some township names that do not match due to phrasing e.g., Bridgeport City versus Bridgeport Town versus Bridgeport which all refer to the same location. After adjusting for those irregularities, the MSA codes are then used to find the corresponding data from the BLS OEWS data tables. There are some missing cells in the OEWS data tables, which we replaced by the state average of that value for whichever state that missing cell was in. Note that 0s are not replaced, only missing values are.  

References:  

• US Bureau of Labor Statistics. Occupational employment and wage statistics: Tables created by BLS. Published April 27, 2023. Accessed September 25, 2023. https://www.bls.gov/oes/tables.htm 

• https://www.bls.gov/sae/additional-resources/metropolitan-statistical-area-definitions.htm 

• https://mcdc.missouri.edu/applications/geocorr2014.html 

Measure: The mean hourly wage of home health aides and personal care aides 

Geographic entity: Metropolitan Statistical Area and Nonmetropolitan Area 

Year(s) available: 2015 

Variable type: Continuous 

Range: 8.09 to 20.37 

Mean: 10.66 

Median: 10.49 

Source: The main data is sourced from the Bureau of Labor Statistics (BLS) page, namely the Occupational Employment and Wage Statistics data, for a particular year. However, to transform the data set to be linked with a data set with only census tract information, additional supplemental data are needed. The reference file for how metropolitan statistical areas is defined can be found here, also on the U.S. Bureau of Labor Statistics website. This is useful for connecting the FIPS codes to the Metropolitan Statistical Area code (MSA code). Converting from census tract to FIPS state and county code is simple, as the census tract code contains the FIPS state and county information. The New England states (Maine, New Hampshire, Vermont, Massachusetts, Connecticut, and Rhode Island) require an additional resource that converts census tract information into the corresponding township names. The resource we used can be found here, on the Missouri Census Data Center website, specifically we used the Geocorr 2014 version as it was closest to our 2015 data set. Select the six New England states and the corresponding source and target geography to generate the reference file.  

Why this measure is important: This measure captures the average hourly wage of the direct care workforce in an MSA and non-MSA. This measure describes compensation of direct care workers in a particular geographic area to support older adults with LTSS needs.  

Background and Definitions: We defined direct care workers in OEWS data tables using two occupations: Home Health Aide and Personal Care Aide. This measure reports the average hourly wage of workers in these occupations in each MSA or non-MSA.  

Approach: To calculate average hourly wage for direct care workers in each MSA or non-MSA, we averaged the hourly wages for all home health aides and personal care aides in each geographic unit. 

Mapping the census tracts to metropolitan and non-metropolitan areas is straight forward except for the six New England States. Those six states have archaic township designations that do not align perfectly with MSA definitions and require adjustment. In short, New England townships may span counties and MSA designation, meaning an intermediary township match must be conducted first to accurately map any census tract data into the correct MSA code. Occupational Employment and Wage Statistics data (OEWS) provided by the bureau of labor only provides information at the metropolitan and nonmetropolitan statistical area level, meaning any data with other geographical definitions will first need to be converted to the MSA level.  

The following description applies to linking census tract level data to OEWS data. Processes for MSA and non-MSA areas are the same. 

The first step involves separating the data set in question into two files, Non-New England States and New England States. The New England States must be treated separately as the census tract information for those states does not match neatly with the MSA areas the OEWS uses. The linking process for the Non-New England States and the New England States are distinct and will be described separately.  

For the Non-New England states, the FIPS state and county code (5 digits) can be isolated from the census tract number. After the FIPS codes are isolated, they can be mapped onto their corresponding MSA code via the reference file from the Bureau of Labor Statistics. With the matched MSA codes, the data set can then be merged with the OEWS data files from the BLS.  

For New-England states, the census tract is used instead of the five-digit FIPS code. Using the reference file from the Missouri Census Data Center, each census tract can be matched with the corresponding township names. From the township names, they are then matched with the information in the MSA reference file for the corresponding MSA codes. Note that there are some township names that do not match due to phrasing e.g., Bridgeport City versus Bridgeport Town versus Bridgeport which all refer to the same location. After adjusting for those irregularities, the MSA codes are then used to find the corresponding data from the BLS OEWS data tables.  

There are some missing cells in the OEWS data tables, which we replaced by the state average of that value for whichever state that missing cell was in. Note that 0s are not replaced, only missing values are.  

References:  

• US Bureau of Labor Statistics. Occupational employment and wage statistics: Tables created by BLS. Published April 27, 2023. Accessed September 25, 2023. https://www.bls.gov/oes/tables.htm 

• https://www.bls.gov/sae/additional-resources/metropolitan-statistical-area-definitions.htm 

• https://mcdc.missouri.edu/applications/geocorr2014.html 

Measure: Presence of MLTSS program for adults aged 65 or older dually-enrolled in Medicaid and Medicare 

Geographic entity: State 

Year(s) available: 2011-2019 

Variable type: Binary 

Source: Mathematica Policy Research’s “Medicaid Managed Long-Term Services and Supports: Summative Evaluation Report”1 was used to identify states with presence of any MLTSS program(s) for adults aged 65 or older. We specifically used Table II.1 (pp. 12-16)1 for our measure of MLTSS presence from 2011-2019. The table details the MLTSS programs included in the evaluation by program features including, state, program start date, and target populations of the program. Because the Mathematica Evaluation report excluded MLTSS programs implemented through FAI demonstrations, we relied on three other reports2-4 to confirm the presence of MLTSS in states that only implemented MLTSS through FAI demonstrations. 

Why this measure is important: This state-level measure captures which state Medicaid programs used MLTSS to deliver care to older adults (aged 65 or older) enrolled in Medicaid from 2011 through 2019. It identifies states that added and discontinued MLTSS over this time period and indicates state Medicaid programs using different payment and delivery approaches for LTSS.  

Background and Definitions: Medicaid MLTSS is the delivery of LTSS through a managed care delivery model where state Medicaid agencies contract with managed care organizations to provide LTSS through a fixed capitated payment. States with MLTSS aim to improve care quality and control costs for Medicaid beneficiaries with disabilities. MLTSS programs can target specific populations (older adults, persons with physical disabilities, and individuals with intellectual or developmental disabilities). This measure focuses on MLTSS programs that include adults aged 65 or older as a target population of the program. 

Approach: We used Table II.1 (pp. 12-16) from the evaluation report to identify states with at least one MLTSS program where adults aged 65 or older were one of the target populations of the program. The table details the MLTSS programs included in the evaluation by program features including, state, program start date, and target populations of the program. The list of MLTSS programs in this table reflected the national landscape of programs as of September 2019. The evaluation excluded MLTSS programs implemented under the Financial Alignment Initiative (FAI) for Medicare-Medicaid dual enrollees and states with these programs are not included in our measure.2 The evaluation also excluded Rhode Island’s Rhody Health Options program because the program ended before the evaluation study period (see footnote 1, p. xi). We rely on the Table II.1 notes to understand MLTSS activity over time, including earlier MLTSS programs that were consolidated.  

We used the program start date to identify the presence of MLTSS programs for the years 2011-2019. For example, Arizona’s MLTSS program started January 1, 1989, so we assume that MLTSS for adults ages 65 and older was available for Arizona enrollees for all years 2011-2019. We count the presence of MLTSS in a certain calendar year if the program was in place for at least 6 months of that year. For example, Kansas’s KanCare started January 1st, 2013, so we determine MLTSS presence in Kansas for the entire 2013 year. However, because Florida’s Statewide Medicaid Managed Care program started August 1, 2013, we determine that MLTSS was not available in Florida for the entire year of 2013 and indicate its first full year presence as 2014. States not listed in the table are assumed to not have MLTSS. 

Seven states had two programs that target adults aged 65 and older. While these programs have different start dates, we consider the presence of MLTSS in the state based on the start date of the oldest program. For example, Tennessee has two MLTSS programs that target adults aged 65 and older (TennCare CHOICES and Employment and Community First CHOICES). TennCare CHOICES started first in 2010 so we determined MLTSS presence in Tennessee for all years 2011 through 2019. 

State-specific notes 

1. Hawaii: The table notes indicated that Hawaii had a MLTSS program with 1115 waiver authority beginning in 2009 which was combined with the QUEST program in 2015. For this reason, we assume that the start date of MLTSS presence in Hawaii as January 1, 2009. 

2. Illinois: The table notes indicated that the state’s first MLTSS program (Integrated Care Program started May 1, 2011) for adults aged 65 and older and two other programs started in subsequent years which were integrated as HealthChoice on 1/1/2018. Furthermore, we consider the start date of MLTSS presence in Illinois to be May 1, 2011. 

3. New Mexico: The table notes indicated that New Mexico had a MLTSS program in 2008 and that it was consolidated into Centennial Care in 2014. For this reason, we assume that the start date of MLTSS presence to be January 1, 2008. 

4. Pennsylvania: We consider the first MLTSS program for adults 65 and older to be Community Health Choices (start date: 1/1/2018). While Pennsylvania has had a MLTSS since 2009 that includes adults aged 65 and older, the table notes state that the program is specific to adults 21 and older with a diagnosis of autism spectrum disorder. 

5. Rhode Island: While the evaluation report1 excluded Rhode Island’s MLTSS program because it ended before the study period, we include and consider this program for the years that it was available: November 2013-September 2018 

Final Counts by Year: 50 states & District of Columbia (n=51). Number of states with MLTSS presence for adults aged 65 or older by year: 

1. 2011 - 13 states 

2. 2012 - 14 states 

3. 2013 - 15 states 

4. 2014 - 20 states 

5. 2015 - 20 states 

6. 2016 - 21 states 

7. 2017 - 21 states 

8. 2018 - 23 states 

9. 2019 – 23 states 

References: 

1. Wysocki, A., Libersky, J., Gellar, J., Miller, D., Liu, S., Luo, M., Tourtellotte, A., & Lipson, D. (2020). Medicaid Managed Long-Term Services and Supports: Summative Evaluation Report (Medicaid Section 1115 Demonstrations Summative Evaluation Report). Mathematica. https://www.mathematica.org/publications/medicaid-managed-long-term-services-and-supports-summative-evaluation-report 

2. Kasten, J., Lipson, D., Saucier, P., & Libersky, J. (2017). Who Enrolls in Medicaid Managed Care Programs that Cover Long-Term Services and Supports?: Implications of Enrollee Diversity for a National Cross-State Evaluation (Medicaid 1115 Demonstrations, pp. 1-8). Mathematica Policy Research. https://www.medicaid.gov/medicaid/downloads/1115-ib1-508-mltss-enrollment.pdf  

3. Lewis, E., Eiken, S., Amos, A., & Saucier, P. (2018). The Growth of Managed Long-Term Services and Supports Programs: 2017 Update. Truven Health Analytics.  

4. Chapter 3: Managed Long-Term Services and Supports: Status of State Adoption and Areas of Program Evolution (pp. 47–78). (2018). [Report to Congress on Medicaid and CHIP]. MACPAC. https://www.macpac.gov/wp-content/uploads/2018/06/June-2018-Report-to-Congress-on-Medicaid-and-CHIP.pdf 

Measure: Medicaid Home and Community-based Services (HCBS) Generosity 

Geographic entity: State 

Year(s) available: 2011-2019 

Variable type: Continuous 

Source: The Truven Health Analytics report that the measure was drawn from provider information about Medicaid LTSS expenditures at the national and state levels. Information includes a description of trends in total LTSS expenditures and LTSS spending for HCBS. 

Why this measure is important: Medicaid HCBS generosity reflects the extent that states prioritize community-based services for their Medicaid beneficiaries with LTSS needs. There is substantial variation across states who have shifted Medicaid funding to HCBS for its Medicaid beneficiaries with LTSS needs. This variation may have implications for the resources, services, and care experiences of older adults eligible for Medicaid. 

Background and Definitions: The Supreme Court decision on Olmstead vs. L.C. in 1999—a civil rights case regarding discrimination against individuals with intellectual disabilities—determined that individuals have the right to live in the community rather than institutions if so desired. This decision had implications for state Medicaid programs which are the primary payer for LTSS for the elderly and disabled. Since this decision, states have worked to rebalance Medicaid financing from institutions to HCBS through Medicaid Waivers and State Plan Amendments.  

Approach: Medicaid HCBS generosity is a percentage that is calculated by dividing a state’s Medicaid LTSS expenditures on HCBS by a state’s total Medicaid LTSS expenditures. 

References:  

• Eiken S, Sredl K, Burwell B, Amos A. Medicaid Expenditures for Long-Term Services and Supports (LTSS) in FY 2015. Truven Health Analytics; 2017. 

• Fabius CD, Ogarek J, Shireman TI. Racial Disparities in Medicaid Home and Community-Based Service utilization among White, Black, and Hispanic adults with multiple sclerosis: Implications of state policy. Journal of Racial and Ethnic Health Disparities. 2019;6(6):1200-1207. 

Measure: Medicaid Per Capita Expenditures for Aged Population by U.S. state 

Geographic entity: State 

Year(s) available: 2018 and 2019 

Variable type: Continuous ($) 

2018 Range: $8172 to $37695 

2018 Mean: $18356 

2018 Median: $17885 

2019 Range: $8877 to $36881 

2019 Mean: $18634.80  

2019 Median: $18748  

Source: We used publicly available data from the Medicaid & CHIP Scorecard.1 The Scorecard publishes state data on the administration and outcomes of Medicaid and the Children’s Health Insurance Program (CHIP) including: enrollment, expenditures, and program improvement efforts. Per capita expenditures rely on spending reported to the Medicaid Budget and Expenditure System (MBES) and the number of enrollees and expenditures reported in the Transformed Medicaid Statistical Information System (T-MSIS). For MBES, states report these data to these systems using Form CMS-64. For T-MSIS, to construct denominators for each eligibility group. Estimates are available for five eligibility groups: children, adults (non-expansion, non-disabled, under age 65), aged, people with disabilities, and adults (ACA Medicaid expansion).  

Why this measure is important: This measure captures variation in Medicaid spending for older adults (aged 65 or older) across states. This measure could be used to describe and understand geographic variation in outcomes among older adults enrolled in Medicaid. 

Background and Definitions: Medicaid provides health coverage to over 80 million persons in the U.S. and has multiple eligibility groups, including the “aged” population which includes adults aged 65 or older. Health covered includes doctors’ visits, hospital care, and medical equipment for those who are eligible for their state’s Medicaid program. For older adults enrolled in both Medicaid and Medicare, the two programs can work together to cover health care costs. Medicaid often pays for services that Medicare does not include, such as transportation or LTSS.  

Approach: We used Table 1 on the Medicaid.gov website1 for 2018 and 2019. We gathered the per capita expenditure estimates of the aged population for each state and the District of Columbia. The Table 1 estimates of state per capita expenditures reflect total Medicaid spending for the aged population divided by the number of enrollees in the aged eligibility category.  

References: 

1. Medicaid Per Capita Expenditures | Medicaid. Accessed September 7, 2022. https://www.medicaid.gov/state-overviews/scorecard/how-much-states-spend-per-medicaid-enrollee/index.html. 

Measure: Medicare Advantage penetration rate  

Geographic entity: State 

Year(s) available: 2015 

Variable type: Continuous  

Source: The Kaiser Family Foundation “Medicare Advantage 2015 Spotlight: Enrollment Market Update” issue brief.  

Why this measure is important: Medicare Advantage penetration, defined as the number of people enrolled in Medicare Advantage over all those enrolled in Medicare, can provide context on the Medicare landscape of that state. Medicare Advantage enrollees also offer supplemental benefits people could spend on LTSS related services. In addition, some Medicare Advantage plans are integrated with Medicaid LTSS for certain populations.   

Background and Definitions: By its nature Medicare Advantage plans vary dramatically across states and plan providers. Medicare Advantage includes multiple types of plans, the vast majority are HMO plans but also includes PPO and PFFS.  This measure counts all individuals enrolled in Medicare Advantage versus everyone enrolled in Medicare in general.  

Approach: Medicare Advantage penetration rate is already calculated in the table KFF provided and is defined as the number of people enrolled in Medicare Advantage over all those enrolled in Medicare within each state.  

References:  

• https://www.kff.org/medicare/issue-brief/medicare-advantage-2015-spotlight-enrollment-market-update/ 

Measure: Medicare reimbursement per beneficiary 

Geographic entity: Hospital Referral Region 

Year(s) available: 2015 

Variable type: Continuous 

Source: Dartmouth Atlas (https://data.dartmouthatlas.org/medicare-reimbursements/ ) 

Why this measure is important: Medicare reimbursement rates by hospital referral region (HRR) representative of the health care market for medical care. 

Background and Definitions: Dartmouth Atlas Medicare reimbursement rates are calculated from Medicare claims files from CMS. Fee-for-service patients enrolled in Medicare Parts A and B are included. Patients enrolled in health maintenance organizations (HMOs) are excluded from the analyses. Rates are adjusted for age, sex, and race based on the Medicare population. 

Approach: We include Medicare reimbursement rates by HRR, which represents regional health markets for tertiary medical care. Each HRR includes at least one hospital that includes major cardiovascular procedures and neurosurgery. There are 306 HRRs in the country (2023). For our analysis, we categorize this continuous measure into tertiles. 

References:  

• The Dartmouth Atlas of Health Care: 2018 Data Update, Kristen Bronner, Scottie Eliassen, Ashleigh King, Christopher Leggett, Sukdith Punjasthitkul, and Jonathan Skinner (2021) 

Measure: State No Wrong Door Score 

Geographic entity: State 

Year(s) available: 2016 

Variable type: Continuous 

Range: 0.00% - 92.0% 

Mean: 59.58% 

Median: 60.00% 

Source: AARP Report “Picking up the Pace of Change: A State Scorecard on Long-Term Services and Supports for Older Adults, People with Physical Disabilities, and Family Caregivers.” 

Why this measure is important: No Wrong Door functions refer to the characteristics of states’ “one-stop-shopping" models that help consumers and their families access public and private services regardless of which organization they contact. Operations and functions of each organization vary greatly which may impact older adults’ and families’ ability to access LTSS. 

Background and Definitions: The measure reflects No Wrong Door scores from 2016. It consists of a single statewide No Wrong Door System with a composite indicator (0%-100%) from 41 criteria across five areas that include the following: (1) state governance and administration [10 criteria], (2) target population [5 criteria], (3) public outreach and coordination and key referral sources [8 criteria], (4) person-centered counseling [9 criteria], and (5) streamlined eligibility for public programs [9 criteria]. The report provides a percentage for each state, as well as its rank relative to other states. A higher percentage indicates a higher-ranking score. 

Approach: State composite indicator (0%-100%) was categorized into three tertiles, with the third tertile reflecting the highest No Wrong Door Scores.  

References:  

• Reinhard S, Accius J, Houser A, Ujvari K, Alexis J, Fox-Grage W. Picking up the Pace of Change: A State Scorecard on Long-Term Services and Supports for Older Adults, People with Physical Disabilities, and Family Caregivers. American Association of Retired Persons Foundation; The Commonwealth Fund; The SCAN Foundation; 2017. 

Measure: Percentage of state’s total population that are enrolled in Medicaid 

Geographic entity: State 

Year(s) available: 2015 

Variable type: Continuous  

Range: 10.00% - 31.07% 

Mean: 19.50% 

Median: 19.04% 

Source: Data taken from US Census Bureau from the American Community Survey Tables for Health Insurance Coverage Table HI-05. “Health Insurance Coverage Status and Type of Coverage by State and Age for All People: 2015”  

Why this measure is important: As LTSS delivery is mostly under the realm of Medicaid, understanding what percentage of the state’s population is enrolled in Medicaid can give some context to said state’s volume of LTSS deliveries, amongst other correlations like the state’s proportion of Medicaid eligible population.  

Background and Definitions: Medicaid is the primary payer of LTSS in the US. Around 20% of total federal and state Medicaid spending went towards LTSS (in 2015). Understanding the landscape of Medicaid enrollment is important to understanding LTSS delivery. Medicaid is a joint federal and state program, meaning there is wide variability between states. This measure does not fully capture all the nuances and would be best supplemented by other contextual factors.  

The source table of this measure provides a host of other valuable contextual information, including age breakdown of Medicaid enrollment and other insurance coverage.  

Approach: The state percentage was calculated by dividing the total Medicaid enrollment in the state by the total population of the state.  

References:  

• https://www.census.gov/data/tables/time-series/demo/health-insurance/acs-hi.2015.html#list-tab-776654388 

Measure: Whether the state has an active paid family leave policy in place 

Geographic entity: State 

Year(s) available: 2011-2022 

Variable type: Binary   

Source: The most updated information about each state’s paid family leave can be found on the National Conference of State Legislatures website. Additional information and supplemental resources can be found on this page and this page respectively. For previous years, the information can be found on the Bipartisan Policy Center.   

Why this measure is important: This state level measure indicates whether potential family caregivers have protection to take time off to care for family members and is reflective of the state’s general labor stance.  

Background and Definitions: Paid family leave is synonymous with the previous terminology of paid parental leave, referring to laws that protect workers for taking time-off to care for a newborn, new child, care for a family member or attend to one’s own serious health conditions while guaranteeing some of that time to be compensated for eligible employees.  

There is no consistent definition of eligibility criteria or generosity (e.g., duration, degree of wage recovery). For example, California only guarantees 8 weeks for parental or family caregiving reasons but grants 52 weeks of paid personal medical leave. In contrast, the District of Columbia offers 12 weeks of coverage starting since 2022. This measure is operationalized as simplified binary indicator for the presence of any paid family leave and does not capture state-by-state differences in policy design or change over time.  

Approach: The presence of a paid family leave policy in a state is denoted in two stages: when the law was enacted, and when the law becomes effective. A score of “1” indicates that the law was enacted and a score of “2” indicates that the law was in effect. However, note that not all information regarding when the law was passed versus enacted is available, and therefore using the year that the law was passed is a more consistent approach.  

References:  

• National Conference of State Legislatures. State family and medical leave laws. Published September 9, 2022. Accessed September 25, 2023. https://www.ncsl.org/labor-and-employment/state-family-and-medical-leave-laws 

• https://www.ncsl.org/research/labor-and-employment/paid-family-leave-resources.aspx 

• https://www.ncsl.org/research/labor-and-employment/time-off-to-care.aspx 

• https://bipartisanpolicy.org/explainer/state-paid-family-leave-laws-across-the-u-s/#:~:text=These%20programs%20are%20active%20in,or%20considered%20paid%20family%20leave. 

Measure: Whether the state has a paid sick leave policies in place   

Geographic entity: State 

Year(s) available: 2011-2021 

Variable type: Binary   

Source: The most updated information about each state’s paid sick leave can be found on the National Conference of State Legislature Website, with additional information found on the page describing paid family and medical leave laws. Additional information about paid time-off in general can be found on the same website, here.  

Why this measure is important: This state level measure indicates the state’s stance on guaranteed paid sick leave for certain workers, which is reflective of the state’s general stance on worker protection.  

Background and Definitions: Paid sick leave laws specifically refer to guaranteed paid time off for an individual to recover from a moderately severe, short-term medical condition. This contrasts with paid medical leave which refers to attending to more serious health conditions.  

There is no strict definition of what paid sick leave covers. For instance, these policies often are not applicable to every employer. Whether the policy applies to part-time employees is inconsistent. How these policies are formulated also vary wildly across states, with different accrual rates. For this measure, we defined a state as having a paid sick leave policy if the state has any paid sick leave law regardless of how generous the policy is.  

Approach: The presence of a paid sick leave policy in a state is denoted in two stages: when the law was enacted, and when the law becomes effective. A score of 1 indicates that the law was enacted and a score of 2 indicates that the law was in effect. However, note that not all information regarding when the law was passed versus enacted is available, and therefore using the year that the law was passed is a more consistent approach.  

References:  

• National Conference of State Legislatures. Paid sick leave. Published July 21, 2020. Accessed September 25, 2023. https://www.ncsl.org/labor-and-employment/paid-sick-leave 

• https://www.ncsl.org/research/labor-and-employment/time-off-to-care.aspx 

Measure: State Unemployment Modernization Laws 

Geographic entity: State 

Year(s) available: 2016 

Variable type: Binary 

Source: Raising Expectations: A State-by-State Analysis of Laws That Help Working Family Caregivers (2018) 

Why this measure is important: Amendments to state unemployment insurance (UI) laws that cover attending to certain compelling family caregiving needs within the definition of “good cause” for leaving employment. This makes caregivers who are ready to return to work eligible for UI benefits during their job search. These supports are one part of broader strategies to support family caregivers of older adults and people living with disabilities that may impact use of LTSS as well care experiences. 

Background and Definitions: The measure reflects whether a state offers unemployment insurance benefits to individuals who lost or left their jobs due to a family member’s illness or disability and are ready and able to begin working again (“UI modernization”). Five points were awarded to states that amend UI laws to include attending to certain compelling family caregiving needs within the definition of “good cause” for leaving employment, making caregivers who are ready to return to work eligible for UI benefits during their job search.  

Approach: A value of “1” indicates that a state had modernization laws and “0” for those that did not have a modernization law. 

References:  

• Raising Expectations: A State-by-State Analysis of Laws That Help Working Family Caregivers. National Partnership for Women & Families; 2018. 

Measure: Percentage of people in household with computer and broadband internet subscription overall all people in households 

Geographic entity: Census tract 

Year(s) available: 2017 

Variable type: Continuous  

Range: 0.00% - 100.00% 

Mean: 80.65% 

Median: 83.07% 

Source: All data downloaded directly from data.census.gov, specifically table S2802 “Types of Internet Subscriptions by Selected Characteristics”, 5-year estimate for the year 2017.  

Why this measure is important: Having a computer and internet connection are essential to accessing health care, health information, personal finances, media, and maintaining social relationships.  Households without a computer and internet connection face barriers to accessing an increasingly virtual and digital landscape.   

Background and Definitions: This measure captures whether households had a computer and internet subscription. The ACS defines an internet subscription as any paid service that allows access to the internet including cellular data plan, broadband, fiber optics, DSL, or other services. “Computer” is broadly defined and includes smartphones and tablets.  

Approach: The data table from the American Community Survey provides census tract level account for the total people living in a household, the number of people living a household with a computer and broadband internet subscription, as well as other categorizations based on age, race, and other factors. This measure is constructed simply by dividing the number of people in households with a broadband internet subscription with the total number of people in households.  

Missing data: 1,716 census tracts (2.4%) have missing values for this measure. 

References:  

• https://data.census.gov/table?q=s2802&g=010XX00US$1400000&y=2017 

Measure: Percentage of people in household with computer and broadband internet subscription overall all people in households 

Geographic entity: Census tract 

Year(s) available: 2017 

Variable type: Continuous  

Range: 0.00% - 100.00% 

Mean: 80.65% 

Median: 83.07% 

Source: All data downloaded directly from data.census.gov, specifically table S2802 “Types of Internet Subscriptions by Selected Characteristics”, 5-year estimate for the year 2017.  

Why this measure is important: Having a computer and internet connection are essential to accessing health care, health information, personal finances, media, and maintaining social relationships.  Households without a computer and internet connection face barriers to accessing an increasingly virtual and digital landscape.   

Background and Definitions: This measure captures whether households had a computer and internet subscription. The ACS defines an internet subscription as any paid service that allows access to the internet including cellular data plan, broadband, fiber optics, DSL, or other services. “Computer” is broadly defined and includes smartphones and tablets.  

Approach: The data table from the American Community Survey provides census tract level account for the total people living in a household, the number of people living a household with a computer and broadband internet subscription, as well as other categorizations based on age, race, and other factors. This measure is constructed simply by dividing the number of people in households with a broadband internet subscription with the total number of people in households.  

Missing data: 1,716 census tracts (2.4%) have missing values for this measure. 

References:  

• https://data.census.gov/table?q=s2802&g=010XX00US$1400000&y=2017 

Measure: The median housing age of the structures built in that area 

Geographic entity: Census tract 

Year(s) available: 2015 

Variable type: Continuous (discrete)  

Source:  All data downloaded from data.census.gov, specifically table B25035 “Median Year Structure Built”, 1-year estimate 

Why this measure is important: This measure describes the status of the housing stock in that geographic area, which may reflect the level of investment, need for maintenance, economic status, accessibility, environmental hazards present in the housing environment. This measure could influence the ability of older adults to “age in place” at home. 

Background and Definitions: Housing age is a crude measure but does offer an insight into various potentially connected barriers. An area with predominately older structures may indicate the need for investments and maintenance in the area. Older structures may be less accessible and require more adaptations for older adults and people with disabilities to age in place at home. 

Approach: Housing age is calculated by subtracting 2015 by the median year of structures built in that census tract, which is used as the median housing age in that geographic area.  

The percentages across all census tracts are then divided into tertiles, which will be the tertile of the entire population. If there is a sample or subpopulation of interest, the tertiles can be selectively generated for the percentages specific to those in the sample.   

Missing data: Approximately 2 percent of all census tracts have this information missing.  

References:  

• https://data.census.gov/table?q=B25035&y=2015&tid=ACSDT1Y2015.B25035 

Measure: Presence of State Coordinating Council (SCC) by U.S. state in 2014 

Geographic entity: State 

Year(s) available: 2014 

Variable type: Binary 

Source: We used the working definition developed by the National Conference of State Legislatures (NCSL) for their 2015 report1 of state human service transportation coordinating councils in all 50 states. Data for this report were collected in 2014 and reflect the status of SCCs in that year.  

Why this measure is important: This state-level measure captures the policy and planning environment for transportation services used by persons with mobility challenges. The presence of SCC may indicate better transportation options and coordination across state agencies, private companies, and public transit.  

Background and Definitions: The report1 defines state coordinating council as: “…[a] forum where government agencies—and, in some cases, other stakeholder groups—work together to make specialized transportation services more effective, efficient, and accessible to the people who need them. Coordinating councils are, in the simplest sense, groups of diverse organizations that actively work together on an ongoing basis to better coordinate and provide transportation services to people who have mobility challenges. Coordinating councils operate at all levels of government—federal, state, regional and local” (p. 6). Three defining features of SCCs. States have a SCC if it is characterized by all three features.  

1. Multidisciplinary: Coordinates among diverse transportation and human services providers  

2. Statewide: Coordinates across the entire state with a focus on state agencies, policies, and programs.  

3. Ongoing: Engages in active, ongoing coordination throughout the year.  

Approach: Using the report, we created a binary variable for each U.S. state and the District of Columbia where “1” indicates that the state has a SCC and “0” indicates that the state does not have a SCC. According to the report, Idaho was the only state to have two state coordinating councils in 2014. We do not distinguish this characteristic in our measure and code the state as a “1.” Seven states (Illinois, Missouri, Nebraska, New York, North Carolina, Tennessee, Wyoming) established a SCC but it was inactive at the time of the report. These states were coded as “0.” The District of Columbia (DC) is not a U.S. state and was not included in the report. We assigned DC as a “0” in our datafile. While this data represents the year 2014, we merged with 2015 round 5 NHATS data on the assumption that there is minimal difference between the two years.  

References: 

1. Rall J, Myers A. State Human Service Transportation Coordinating Councils: An Overview and State Profiles. National Conference of State Legislatures; 2015:1-73. Accessed July 18, 2022.  https://www.ncsl.org/Portals/1/Documents/transportation/SCC_transportation_final02.pdf 

Measure: Presence of Web Accessibility Laws for Older Adults and People with Disabilities 

Geographic entity: State 

Year(s) available: 2014 

Variable type: Binary 

Source: We use Yang et al. 20151 to develop a state-level measure of web accessibility policies. Yang et al. 2015 examined internet accessibility legislation in the US and noted which states have these policies in place (Table 1). While unspecified in the article, we assume that the state legislations were current as of 2014.  

Why this measure is important: This state-level measure captures the web accessibility policy environment for older adults and persons with disabilities and fits under the communication infrastructure construct in the LTSS framework. The presence of web accessibility laws mean that public-facing government websites must conform to a set of accessibility standards to ensure that older adults and persons with disabilities can navigate information and services critical to their health and well-being.  

Background and Definitions: No federal laws require all publicly accessible websites to implement accessibility standards for older adults and persons with disabilities.1 However, some states implement legislation which add standards and requirements regarding web accessibility. States draw from the following legal foundations to develop their state laws and policies on web accessibility.   

1. The Americans with Disabilities Act (ADA): Title III of the ADA defines places of public accommodation. While the federal statute does not include the Internet as a public place, some state laws do and determine that Internet websites are places of public accommodation.  

2. Section 508 of the Rehabilitation Act and Access Board Recommendations: Section 508 refers to procurement and puts accessibility requirements on electronic and information technology purchased by the federal government. In this way, it forces businesses to procure technology that complies with accessibility standards.  

3. The World Wide Web Consortium’s Web Content Accessibility Guidelines (WCAG): The guidelines are more inclusive and expansive and include: text alternatives for non-text content, captions for multimedia presentations, content that reduces the risk of seizures, and keyboard accessibility.1   

Of note, these web accessibility policies, standards, and requirements only apply to state or state affiliated entities. Private entities and businesses are exempt from these requirements.  

Approach: Using the Table 1 in Yang et al. 2015, we created a binary variable for each U.S. state and the District of Columbia based on Column 2. “1” indicates that the state has a web accessibility policy available in 2014 and “0” indicates that the state did not have a policy in 2014. A state web accessibility policy could include any of the legal foundations described in the previous section. We do not distinguish states by the legal foundation used to inform their state policy.  While this data represents the year 2014, we merged with 2015 round 5 NHATS data on the assumption that there is minimal difference between the two years.  

References: 

1. Yang YT, Chen B. Web Accessibility for Older Adults: A Comparative Analysis of Disability Laws. The Gerontologist. 2015;55(5):854-864. doi:10.1093/geront/gnv057