Long-Term Care: Who Gets It, Who Provides It, Who Pays, And How Much?

By H. Stephen Kaye, Charlene Harrington, and Mitchell P. LaPlante Long-Term Care: Who Gets It, Who Provides It, Who Pays, And How Much? doi: 10.1377/hlthaff.2009.0535 HEALTH AFFAIRS 29, NO. 1 (2010): 11 21 2010 Project HOPE The People-to-People Health Foundation, Inc. ABSTRACT Long-term care in the United States is needed by 10.9 million community residents, half of them nonelderly, and 1.8 million nursing home residents, predominantly elderly. Ninety-two percent of community residents receive unpaid help, while 13 percent receive paid help. Paid community-based long-term care services are primarily funded by Medicaid or Medicare, while nursing home stays are primarily paid for by Medicaid plus out-of-pocket copayments. Per person expenditures are five times as high, and national expenditures three times as high, for nursing home residents compared to community residents. This suggests that a redistribution of spending across care settings might produce substantial savings or permit service expansions. Long-term care services, whether provided in institutions or the community, are essential to the well-being of many elderly and nonelderly people with limitations in performing daily activities. Long-term care provided outside of institutions, known as personal assistance services, personal care services, or home and community-based services, also enables many people with disabilities to maintain their independence; avoid institutionalization; and participate in family, community, and economic activities. Noninstitutional long-term care can be purchased or obtained from family, friends, and other volunteer helpers. With projections indicating a doubling in the need for long-term care over the next forty years, 1 spending on publicly paid services already an ever-increasing share of ever-rising national health care expenditures is of grave concern to policymakers at the federal and state levels. Efforts to both improve the long-term care system and reduce spending are limited by a lack of information on how much is spent, for what services and in what settings, and the extent and nature of unpaid help that people receive. This paper is an attempt to partly fill that gap. Its findings, for example, indicate that most longterm care spending goes to the relatively small minority of long-term care recipients living in nursing homes, that the vast majority of community residents needing long-term care get only unpaid help, and that although about half of all long-term care recipients are under age sixtyfive, four-fifths of long-term care spending is for elderly recipients. Data Sources And Methods This paper explores the size and characteristics of the U.S. population needing help with daily activities, the nature of unpaid and paid providers of long-term care, sources of payment, and spending for those services, both individually and on a national level. We conducted analyses of public-use data sets from five nationally representative federal surveys that use state-of-the-art data collection methods: (1) The Survey of Income and Program Participation, a periodic longitudinal survey of 96,000 household respondents conducted in person by the Census Bureau in 2005. 2 (2) The 2007 National Health Interview Survey, an annual in-person survey of 95,000 household respondents sponsored by H. Stephen Kaye (steve.kaye@ucsf.edu) is an associate adjunct professor in the Institute for Health and Aging at the University of California, San Francisco, and co principal investigator of thecenteronpersonal Assistance Services. Charlene Harrington is a professor of sociology at the University of California, San Francisco, and principal investigator of the Center on Personal Assistance Services. Mitchell P. LaPlante is an adjunct professor in the Department of Social and Behavioral Sciences at the University of California, San Francisco, and co principal investigator of the Center on Personal Assistance Services. JANUARY 2010 29:1 HEALTH AFFAIRS 11

the National Center for Health Statistics. 3 (3) The 2007 American Community Survey, the Census Bureau s annual mail-in survey of three million respondents living in all residential settings. 4 (4) The 2004 National Nursing Home Survey, a representative survey of 13,500 residents living in 1,500 nursing homes, conducted in person by the National Center for Health Statistics. 5 (5) The Medical Expenditure Panel Survey, from the Agency for Healthcare Research and Quality, providing monthly spending data on 34,000 household respondents. We analyzed 5,703 home health spending records from 2005 06. 6 All five surveys contain questions regarding the performance of such self-care activities as bathing and dressing, generally known as activities of daily living (ADLs), and other routine activities, such as shopping and doing housework, known as instrumental activities of daily living (IADLs). Four of the surveys ask about the need for help from other people, a standard measure of long-term care need. The exception is the American Community Survey, which asks about difficulty performing the activities. For consistency across surveys, ADLs comprise bathing, dressing, transferring from bed or chair, eating, using the toilet, and getting around inside the home or nursing home room; we excluded activities for which only supervision is provided. Population numbers, proportions, and spending were calculated using sampling weights to obtain nationally representative estimates. Comparisons mentioned in the text have been tested for statistical significance, using tests appropriate for complex sampling designs, and were found to be significant at the 95 percent confidence level or greater. Findings ESTIMATES OF LONG-TERM CARE POPULATION SIZE Exhibit 1 shows estimates from four surveys of the population needing long-term care services. We identify two main population groups by residential setting: (1) Community residents, comprising households and noninstitutional group quarters such as group homes, dormitories, and homeless shelters. The National Health Interview Survey and Survey of Income and Program Participation are household surveys, but the American Community Survey includes noninstitutional group quarters as well. (2) Institutional residents, comprising nursing homes, facilities for people with intellectual and developmental disabilities, other residential health care facilities, and also prisons and jails. The National Nursing Home Survey targets nursing homes only, but the American Community Survey covers residents of all types of institutions. Focusing first on community residents, we present three tiers of population estimates based on the level of identified need. BROADLY DEFINED POPULATION: The broadly defined long-term care population needs help with one or more ADLs or IADLs. The ten or eleven million people, or about 4 percent of community residents, in this category may get assistance from family members, friends, or paid helpers, and might also rely on meal delivery, transportation, or homemaker services provided by community organizations or government programs. This broad definition is particularly relevant to policymakers concerned about the continued availability of family helpers in the face of the aging of the population, the impact of such help on families, and the ability of people with disabilities to fully participate in society. The Survey of Income and Program Participation yields the largest and probably most accurate estimate, because it asks about help in each activity separately, rather than asking about multiple activities at once, as in the National Health Interview Survey. The somewhat narrower American Community Survey measure, consisting of a single question about difficulty with certain ADL tasks, is included here because it captures a far broader population than questions about needing ADL help; furthermore, our analysis of a similar measure in the Survey of Income and Program Participation indicates that the vast majority of people reporting ADL difficulty need help with either ADLs or IADLs and that most people needing help in multiple ADLs/IADLs report ADL difficulty. INTERMEDIATE POPULATION: The intermediate long-term care population is composed of people needing ADL help. Both the Survey of Income and Program Participation and National Health Interview Survey yield estimates of roughly 4.7 million, or just under 2 percent of the population. The assistance that such people receive is essential for their health, functioning, personal dignity, and very survival. 7 For this reason, need at the ADL level is often seen as an indicator of potential usage of paid help, especially for people without live-in family helpers. In addition, it is often treated as a minimum eligibility requirement for publicly funded community-based long-term care services. NARROWLY DEFINED POPULATION: The narrowly defined long-term care population includes people needing help with two or more ADLs (for example, bathing and dressing together, but not bathing alone). Its members are often said to have an institutional level of need. This group is of particular policy rele- 12 HEALTH AFFAIRS JANUARY 2010 29:1

EXHIBIT 1 Population Needing Long-Term Services In The United States, By Age, Residential Setting, Measure Used, And Data Source, Selected Years Age group (thousands of people, percent of population) All ages a <18 a 18 64 65+ Measure, data source No. % No. % No. % No. % % <65 Community residents Broadly defined LTC population Gets ADL/IADL help, SIPP 10,887 4.1 393 0.8 5,073 2.8 5,421 15.5 50.2 Gets ADL/IADL help, NHIS 9,613 3.5 281 0.5 4,409 2.4 4,923 13.6 48.8 ADL difficulty, ACS 8,382 3.0 460 0.9 4,154 2.2 3,769 10.4 55.0 Intermediate LTC population Gets ADL help, SIPP 4,774 1.8 299 0.6 1,972 1.1 2,503 7.2 47.6 Gets ADL help, NHIS 4,673 1.7 281 0.5 1,899 1.0 2,493 6.9 46.7 Narrowly defined LTC population Gets help with 2+ ADLs, SIPP 3,143 1.2 193 0.4 1,303 0.7 1,647 4.7 47.6 Gets help with 2+ ADLs, NHIS 3,169 1.2 219 0.4 1,205 0.6 1,746 4.8 44.9 Gets help with 3+ ADLs, SIPP 2,301 0.9 154 0.3 924 0.5 1,223 3.5 46.8 Gets help with 3+ ADLs, NHIS 2,305 0.8 179 0.3 862 0.5 1,265 3.5 45.2 Institutional residents Broadly defined LTC population ADL difficulty (any inst.), ACS 1,575 19 268 1,288 18.2 Nursing home residents, ACS 1,788 250 1,538 14.0 Nursing home residents, NNHS 1,492 3 172 1,317 11.7 Narrowly defined LTC population, nursing homes only Gets help w/ 2+ ADLs, NNHS 1,277 3 133 1,141 10.6 Gets help w/ 3+ ADLs, NNHS 1,184 3 121 1,060 10.5 All residential settings Broadly defined LTC population ADL difficulty, ACS 9,957 3.5 479 0.9 4,422 2.3 5,056 13.4 49.2 SOURCE Authors tabulations of 2005 data from the Survey of Income and Program Participation (SIPP) and from the 2007 National Health Interview Survey (NHIS), the 2007 American Community Survey (ACS), and the 2004 National Nursing Home Survey (NNHS); 2007 ACS nursing home resident data from Table S2601B at http://factfinder.census.gov NOTES ADL is activities of daily living. IADL is instrumental activities of daily living. a Beginning at age 5 for NHIS and ACS, age 6 for SIPP, and age 18 for ACS nursing home population; IADL measure asked of ages 15+ in SIPP and 18+ in NHIS. vance because eligibility for many federal and state programs, especially those entailing institutional services, as well as for long-term care insurance benefits, hinges on requiring help with multiple ADLs. The narrowly defined longterm care population numbers about 3.2 million, including 2.3 million requiring help with three or more ADLs. Its members are at high risk for institutionalization when needs go unmet. ESTIMATES: Although older adults are much more likely than younger people to need longterm care, approximately half of the broadly defined long-term care population living in the community is nonelderly. Even among the narrowly defined long-term care population, whose disabilities are more significant, more than 45 percent are under age sixty-five. With respect to the institutional long-term care population, estimates of the number of nursing home residents vary from 1.5 million to 1.8 million, the vast majority of whom need help with multiple ADLs. (The larger, more recent estimate from the American Community Survey is probably more accurate.) It should be noted that these are point-in-time estimates rather than annual totals, a common feature of administrative data. Unlike the community-dwelling long-term care population, the nursing home population is predominantly (more than fourfifths) elderly. In all, 10.0 million Americans, living either in the community or in institutions, report ADL difficulty. About half of this population is under age sixty-five. CHARACTERISTICS OF THE LONG-TERM CARE POPU- LATION Demographic, economic, and functional characteristics of the broadly defined long-term care population, based on data from the 2007 JANUARY 2010 29:1 HEALTH AFFAIRS 13

American Community Survey, are shown in Exhibit 2. Institutional and community residents are shown separately, with the latter divided into two categories: 8.0 million people living in households, and 366,000 people living in noninstitutional group quarters, such as group homes. Institutional residents are both far older and much more likely to be female than all of the other groups. Compared to people without long-term care needs, there is a higher proportion of whites and African Americans, and a lower proportion of Latinos, Asians, and Pacific Islanders, among people with long-term care needs in all settings. There is a lower proportion of Native Americans among institutional longterm care recipients, but a higher proportion in the household long-term care population, than among people without long-term care needs. Adults with long-term care needs who live in institutions are less than half as likely to be married as those living in households, who are themselves much less likely to be married than adults without long-term care needs. Very few adults living in noninstitutional group quarters are married, and most have never been married. The high proportion of unmarried people in the community-resident long-term care group translates to a much greater likelihood of either living alone or sharing a residence with nonrelatives. People in such circumstances are of particular policy interest, because they often lack a ready supply of unpaid helpers and therefore have a greater need for paid services. Regardless of residence, people with longterm care needs tend to be less educated, and EXHIBIT 2 Demographic, Economic, And Functional Characteristics Of The Broadly Defined Long-Term Care (LTC) Population, By Residential Status, 2007 LTC population a Community residents Institutional residents Households Group quarters People without LTC needs Population (thousands) 1,575 8,016 366 270,968 Age and sex Percent age 65 81.8% 45.2% 40.0% 12.1% Median age (years) 82 62 57 38 Percent female 66.8% 59.5% 52.7% 50.5% Race/ethnicity White 82.2% 76.4% 79.5% 76.1% African American 14.3 16.0 15.8 12.8 Asian/Pacific Islander 1.5 3.1 2.0 5.2 American Indian/Alaska Native 0.9 2.4 1.2 1.4 Latino/Hispanic 5.1 10.4 6.3 14.6 Marital status (age 18+) Married 18.0% 40.7% 6.5% 53.8% Widowed 51.7 25.0 21.4 5.7 Otherwise unmarried 30.3 34.3 72.1 40.5 Living arrangements Household with family/relatives 0.0% 73.8% 0.0% 82.3% Alone/other 100.0 26.2 100.0 17.7 Educational attainment (age 18+) High school graduate 59.8% 66.5% 47.4% 85.2% College graduate 9.5 11.5 7.9 25.6 Income Family income <100% FPL 22.1% 62.4% 12.0% Family income <200% FPL 49.3% 91.1% 29.0% Median household income $32,400 $60,000 Median individual income (ages 18+) $9,200 $10,800 $8,000 $23,900 Functional status Mobility impairment 92.8% 89.4% 75.0% 7.0% Cognitive impairment 75.8 55.3 84.8 4.3 Sensory impairment 37.2 30.7 31.6 3.5 SOURCE Authors tabulations of public use data from the American Community Survey. NOTES Tabulations exclude children younger than age 5. Poverty status (percent of federal poverty level, or FPL) is not determined for institutional residents. a People with difficulty bathing, dressing, or getting around inside the home. 14 HEALTH AFFAIRS JANUARY 2010 29:1

to have much lower personal incomes, than people without long-term care needs. Among people living in households, median household income for those with long-term care needs is little more than half that of people with no such needs. Just under half of the household population with long-term care needs lives in or near poverty (<200 percent of the federal poverty level), as do more than nine-tenths of those in noninstitutional group quarters. Finally, although the vast majority of long-term care recipients in all settings experience mobility impairments, cognitive impairments (broadly defined) affect only 55.3 percent of household residents, 75.8 percent of those in institutions, and 84.8 percent of those in noninstitutional group quarters. SOURCES OF HELP FOR THE COMMUNITY-RESIDENT LONG-TERM CARE POPULATION Family members are far and away the principal providers of assistance to the long-term care population living in households. Data from the Survey of Income and Program Participation, which asks long-term care recipients about their main (primary) helper and a possible additional (secondary) helper, are shown in Appendix Exhibit 1. 8 Despite the ever-increasing attention and resources devoted to paid, noninstitutional long-term care services, only 13.0 percent of the broadly defined long-term care population (or 22.5 percent of the narrowly defined population) use paid helpers in either a primary or secondary role. Elderly people with broadly defined long-term care needs use paid help at more than twice the rate of nonelderly people (18.0 percent versus 7.8 percent), and people living alone are nearly four times as likely to have paid helpers as those living with relatives (26.2 percent versus 7.1 percent). Nearly half of the narrowly defined long-term care population living alone gets paid help. Use of a paid secondary helper (such as for respite care) is rare across all groups. The 13.0 percent of the broadly defined longterm care population receiving paid help translates into approximately 1.4 million U.S. adults. The survey probably underestimates the usage of paid help, because only information for the two main helpers is recorded, and the questions assume that a paid helper is not a family member who gets paid for his or her time. Receipt of help from a paid relative is not measured in any ongoing federal survey. Despite the greater reliance upon paid helpers among elderly people with long-term care needs, their usage of unpaid help is about the same as that of working-age adults, just above 90 percent. Nearly all people with long-term care needs who live with family get unpaid help, compared to 81 percent of people living alone or with nonrelatives. Among the narrowly defined long-term care population, only 70.4 percent of those living alone get unpaid help. Principal sources of help vary markedly with age (Exhibit 3). Among the narrowly defined long-term care population, help from parents dominates for people under age thirty but then falls sharply at higher ages. Between ages thirty and seventy-four, the spouse is the dominant source of help, followed by an offspring more likely a daughter than a son. For ages seventy-five and older, when the spouse may have died or become a less effective helper, daughters and sons become the principal helpers. Usage of paid helpers hovers at roughly 15 percent below age sixty, after which it begins to rise once parents are no longer available and, at higher ages, spouses decline in prevalence as helpers. The pattern by age, coupled with the much higher usage of paid helpers among people living alone, seems to imply that people generally get help from any available relative (or nonrelative), and only in the absence of such helpers are paid workers sought out. Per Appendix Exhibit 1, 8 relatively few people with long-term care needs receive no help at all. Even among people living alone, only 4.7 percent lack personal assistance entirely. However, this measure captures only a small fraction of the total unmet need for long-term care, which is far more often experienced as a lack of sufficient help than as a complete absence of help. 7 SOURCES OF PAYMENT FOR LONG-TERM CARE IN THE COMMUNITY AND IN NURSING HOMES According to our analysis of Medical Expenditure Panel Survey data, 1.6 million community residents of all ages receive paid long-term care each month (see Exhibit 4 and Appendix Exhibit 2), 8 consistent with the 1.4 million figure from the Survey of Income and Program Participation for adults. This total excludes 0.9 million people who receive home health services that do not involve ADL or IADL help. Medicare and Medicaid are the primary payers: Each program pays for all or part of the services received by about one-third of community residents. Nearly one-quarter of recipients pay out of pocket: 18.6 percent pay all or most of the charges, and 5.4 percent pay less than half. Private insurance rarely pays for such services. Some 19.3 percent of recipients get their help paid for, in whole or in part, by some other source, generally a state or local program. A majority of the community-resident longterm care population receiving paid help are reported as needing ADL help. This population is more likely to receive federally funded services, and much less likely to pay for the bulk of their services themselves, than people needing only IADL help. JANUARY 2010 29:1 HEALTH AFFAIRS 15

EXHIBIT 3 Major Sources Of Help With Daily Activities Among Community Residents With Two Or More Activities Of Daily Living (ADL) Needs, By Age 80% 70 60 50 Daughter/son 40 Spouse 30 20 Paid helper 10 0 Parent 15 29 30 44 45 59 60 74 75 84 85+ Age of person receiving services (years) SOURCE Authors tabulations of 2005 data from the Survey of Income and Program Participation. Once again we find that a much smaller number of nonelderly than elderly people receive paid help. Services provided to people under age sixty-five are far more likely to be paid for by Medicaid, and far less likely to be paid for either by Medicare or out of pocket, than are services provided to their older counterparts. People receiving noninstitutional long-term care often get additional services, generally delivered by professionals such as nurses or physical or occupational therapists, which are intended to treat a health condition or restore functioning. This fact complicates analysis of long-term care spending, because charges for these professional services, which are often substantial, cannot generally be separated from those for personal assistance. A majority of community residents receiving paid long-term care, however, get only personal assistance. For the 829,000 people in this category, Medicaid is a much more prominent payer than Medicare, and self-pay contributes substantially. In contrast, the 794,000 people who receive personal assistance plus some type of medical services at home are more than twice as likely to have their services paid for by Medicare, and only half as likely to be paying out of pocket. Among people receiving personal assistance without professional services, most obtain services through agencies rather than from selfemployed, independent providers. The top payers for agency-provided services are Medicaid and Medicare, and very little is paid for out of pocket. However, the opposite is true of independent providers, the vast majority of whom (85.1 percent) are paid primarily by the recipient or the family. Despite the existence of consumerdirected, independent-provider options in some states, Medicaid pays for only 10.1 percent of independent providers. Of all consumers of Medicaid-paid personal assistance with no medical component, only 5.9 percent use independent providers; Medicare pays for agencyprovided services almost exclusively. In contrast, among those paying for services out of pocket, most use independent providers, perhaps based on lower rates or a preference for greater consumer control. During the initial period of service delivery, the main payer is Medicare, which often covers rehabilitation and restorative services following hospitalization ( postacute care ). After six 16 HEALTH AFFAIRS JANUARY 2010 29:1

EXHIBIT 4 Payments And National Expenditures For Paid Long-Term Care (LTC) Services, By Setting, Level Of Need, Age Group, Type Of Services Received, And Duration Of Receipt Of Services, Selected Years Number of recipients (thousands) Percent of recipients making an out-of-pocket payment Median monthly payment From all sources a Out of pocket b Community residents (2005 06) Any LTC services at home c 1,623 24.0% $795 $214 $33.7 Level of need Needs ADL help 942 17.8 926 480 25.3 Needs only IADL help 682 32.5 545 120 8.5 Age group Under 65 542 16.1 773 167 15.3 65+ 1,081 27.9 806 280 18.4 Type of services received Personal assistance only 829 30.6 550 120 10.7 Needs ADL help 379 19.7 810 400 6.7 Needs only IADL help 450 39.7 400 100 4.0 Agency provider 650 15.3 703 90 9.5 Independent provider 180 86.0 152 120 1.2 Personal assistance plus professional 794 17.1 1,075 500 23.1 services Needs ADL help 562 16.5 1,162 500 18.6 Needs only IADL help 232 18.4 954 530 4.5 Duration of receipt of services (prior to interview) Initial 3 months 319 29.2 772 220 5.7 Months 4 6 194 29.5 611 300 3.0 Month 7 and beyond 1,116 22.7 768 264 24.2 Other home health services d 872 13.6 571 72 11.8 Nursing home residents (2004) All residents 1,492 71.5 4,230 923 113.7 Excluding 3 mos. after hospitalization 1,321 76.2 4,170 916 93.9 Age group Under 65 175 49.2 3,990 677 13.5 65+ 1,317 74.5 4,260 960 100.2 Length of stay (at time of interview) 30 days 156 26.3 8,160 1,883 19.9 31 90 days 144 54.9 4,980 1,271 13.5 91 days 1 year 359 76.9 4,170 1,080 25.1 13 months 3 years 451 79.3 4,080 926 29.7 >3 years 382 82.0 4,080 792 25.5 Annual expenditure (billions of 2009 dollars) SOURCE Authors tabulations of public use data from the 2005 and 2006 Medical Expenditure Panel Survey (for community residents) and the 2004 National Nursing Home Survey (for nursing home residents). a Detailed data on the contributions of various sources of payment (self/family as primary or secondary payer, Medicaid, Medicare, privateinsurance,andother)maybefoundinthetechnicalappendix,onlineasinnote8. b Median includes only recipients making an out-of-pocket payment. c Personal assistance and other home health services provided to people receiving paid personal assistance at home. d Home health services provided to people not receiving paid personal assistance at home. months, Medicaid becomes the most prominent payer. Medicaid and the consumer are the two major payers for nursing home stays, with out-ofpocket payments generally secondary to Medicaid. Nonetheless, one-fifth of nursing home bills are paid out of pocket entirely or primarily, at a considerable financial burden. Medicare is the third most prominent payer, followed by private insurance. Although Medicare covers payments for only 17.7 percent of nursing home residents, it is a major payer for the first three months of a nursing home stay. The proportion of residents whose charges are paid primarily out of pocket increases from 10.7 percent during the first thirty days to 28.3 percent during the fourth through twelfth months and then declines to 15.9 percent after three years. Partly because of eligibility rules requiring that a person s assets be spent down, Medicaid pays for only about one-fifth of residents during the first month, and that proportion grows to reach just over fourfifths for stays of greater than three years. JANUARY 2010 29:1 HEALTH AFFAIRS 17

PER RECIPIENT LONG-TERM CARE SPENDING The monthly payment data in Exhibit 4 reveal that typical per recipient spending on noninstitutional and institutional long-term services is on entirely different scales. The median monthly payment for community residents is $795, while the median for nursing homes is $4,230; inflation-adjusting both figures to 2009 yields $928 and $5,243, respectively. An even starker comparison results if we consider only home-based personal assistance services, excluding professional services, for which the median expenditure is $550 per month ($635 in 2009 dollars). A fairer comparison would be to consider only people with ADL needs, whose median communitybased spending is $1,069 in 2009 dollars, about one-fifth of the nursing home expenditure. One might reasonably object that even the last comparison does not adequately take into account differences in the levels of need between the community-resident and institutional population. Because the Medical Expenditure Panel Survey lacks information on specific ADL limitations, we cannot directly compare spending on people with the same level of need, but we can use the Survey of Income and Program Participation to compare the extent of need among people with ADL needs who get paid, community-based long-term care services with that of the institutional long-term care population. Despite a somewhat lower level of need among community than institutional residents (mean number of ADL needs is 3.5 versus 3.9), however, a substantial minority of community residents with ADL needs report a very high level of need (36.5 percent with five or six ADL needs, compared to 48.1 percent of the institutional population). These data suggest that differences in level of need could account for some, but by no means all, of the differences in spending between institutional and noninstitutional services. Indeed, the distribution of institutional and noninstitutional long-term care expenditures (Exhibit 5), with the latter limited to the intermediate long-term care population, shows little overlap between the two residential settings, despite the substantial overlap in levels of need. Nearly all nursing home stays cost at least $3,500 per month, in 2009 dollars. But most noninstitutional expenditures are under $1,500 per month, in 2009 dollars, and 87.4 percent are under $3,500 per month. Out-of-pocket spending is also much greater for institutional than for noninstitutional services (Exhibit 4). Among people with ADL needs, the median nursing home out-of-pocket expense is nearly twice as high as that for noninstitutional services ($554 versus $1,065, in 2009 dollars). The first month of a nursing home stay typically costs twice as much as any month after the first year. The beginning of a stay typically involves additional services, often because the person has just been discharged from a hospital. Because such services fall under the Medicare postacute budget category, they are not always classified under the long-term care umbrella; Exhibit 4 shows expenditures for all residents except those who were hospitalized prior to admission and are in their first ninety days of a stay. ESTIMATED NATIONAL LONG-TERM CARE SPENDING We estimate the total annual spending on paid long-term care services, delivered either in a recipient s home or in a nursing home, as $147.4 billion, adjusted for inflation to 2009 dollars. The figure is an approximate estimate of total nursing home spending plus spending for community residents receiving assistance with daily activities, and including other home health services delivered to those individuals. Our estimate of $113.7 billion in annual nursing home spending would increase to $136.2 billion if we were to use the American Community Survey s higher estimate of the nursing home population. Our estimate of $33.7 billion in noninstitutional long-term care spending excludes home health services provided to people not receiving assistance with daily activities, estimated at $11.8 billion. Noninstitutional services account for only 22.9 percent of the $147.4 billion total. About 13 percent of that total is for services delivered to people within three months of admission to a nursing home following hospitalization; a good part of that spending is probably for postacute services that some readers might not consider long-term care. Some 80.5 percent of total expenditures, or $118.6 billion in 2009 dollars, goes to people age sixty-five or older. Elderly recipients account for a majority of both community-based and nursing home based expenditures (54.6 percent and 88.1 percent, respectively). On the noninstitutional side, three-quarters of the total is for people with an ADL level of need. More than half is for those with ADL needs who are also getting medical care at home. Among people getting only personal assistance at home, the vast majority (88.7 percent) of funds go to agencies. Only the remaining 11.3 percent go to independent providers, whose typical monthly charges are much less than those of agency providers. Only 0.8 percent of total national longterm care spending goes to independent providers of personal assistance alone. Estimating the enormous economic value of unpaid long-term care is beyond the scope of this paper, but it has been addressed by others. 9,10 18 HEALTH AFFAIRS JANUARY 2010 29:1

EXHIBIT 5 Distribution Of Monthly Long-Term Care Spending, By Residential Setting, Inflation-Adjusted To 2009 Home/community services for people with ADL needs Nursing home services Percent of recipients SOURCE Authors tabulations from the 2005 and 2006 Medical Expenditure Panel Survey and the 2004 National Nursing Home Survey. Discussion Our analyses of data from five national surveys paint a portrait of long-term care in the United States that is sometimes surprising. Some ten to eleven million community-dwelling Americans need help with daily activities. Adding the approximately 1.5 million people receiving longterm care services in nursing homes yields an overall long-term care population of about 12 million, or roughly 4 percent of the total population. The proportion of the population needing long-term care rises dramatically with age, a fact that leads many to assume that most of the longterm care population is elderly. Not so: About half of community-dwelling Americans needing long-term care are younger than age sixty-five. Even when the much older nursing home population is added in, only a slight majority of the entire long-term care population is elderly. Research and data collection focusing solely on older adults with long-term care needs misses half of the story. A more important imbalance is in the amounts spent: Four-fifths of national long-term care spending goes to the half of the long-term care population who are elderly. Most of that is public spending, but there is a substantial out-of-pocket component, and most of it is spent on institutional services. Is this imbalance in spending the result of different circumstances (for example, greater availability of unpaid helpers for the nonelderly or greater health care needs among the elderly) or of public programs that serve only or primarily older people, or offer only institutional services, which younger people do not want? Further research is needed, but we suspect that public programs foster age inequities in the availability of paid services. Aside from age itself, another key difference between the institutional and noninstitutional populations is the much higher rate of cogni- JANUARY 2010 29:1 HEALTH AFFAIRS 19

tive impairment among those living in nursing homes. Age differences also contribute to a greater likelihood of nursing home residents being widowed or otherwise unmarried, compared to their community-dwelling counterparts. Indeed, a large majority of household residents with long-term care needs live with a spouse, family, or other relatives, who typically serve as ready sources of unpaid help. The critical importance of unpaid help is made clear by data showing that about nine-tenths of the community-dwelling long-term care population relies on a family member, relative, friend, or volunteer as the primary source of help with daily activities. Only among the narrowly defined long-term care population living alone does the prevalence of a paid primary helper (44 percent) exceed one-quarter of recipients, and it is only 7 percent for those living with family. The vast majority of the long-term care population with access to unpaid help appear to use it, and to use it almost exclusively a finding that should help allay policymakers fears that greater access to publicly financed long-term care would result in a high rate of substitution of paid services for unpaid services. Furthermore, usage of secondary paid helpers is tiny across all groups, which indicates that paid help is rarely used to supplement unpaid help. One crucial piece of missing information is the extent to which long-term care needs are met or unmet. Unlike unmet need for health care, unmet need for long-term care is not routinely measured in any federal survey of people of all ages. Data from the mid-1990s indicate that unmet long-term care needs affected about one-fifth of the long-term care population 7,11 but without recent, recurring, and geographically detailed estimates, it is difficult to assess the extent to which low rates of use of paid help are a matter of preference versus lack of access, and to what degree increased availability of paid long-term care could reduce unmet need. The long-term care population living alone, without spouses or other family members conveniently situated to offer help, is particularly vulnerable. Because income levels among the community-resident long-term care population are so low, it is doubtful that many such people could afford to pay for services entirely or mostly on their own. A major payer for long-term care among community residents is Medicaid, whose programs offering noninstitutional long-term care vary widely from state to state and even from one disability group to another. As a consequence, there are undoubtedly access disparities resulting in greater unmet needs in some places than others, and for some population groups than others. There is an urgent need for populationbased data that could facilitate assessing the scope of the problem, identifying policy solutions, and estimating their potential costs. The typical monthly nursing home bill is far greater than that for noninstitutional services, and national expenditures on nursing homes are more than three times those for noninstitutional long-term care services. Greater per recipient nursing home costs can be only partly attributed to a greater need for services. Whether institutional service delivery can be justified despite its expense, based on services needed and desired by the recipient, depends on individual circumstances, including the availability of adequate personal assistance and appropriate living arrangements in the community. Further research is desperately needed to better assess the appropriateness and cost-effectiveness of different long-term care settings based on the recipient s functional and personal characteristics. Another potential means of reducing expenditures would be for government programs to use independent providers of noninstitutional long-term care, an approach that would also afford recipients greater control over their services than through agency providers. A redistribution in long-term care spending from institutional to noninstitutional settings, and from agency to independent providers, appears to offer the potential for a sizable reduction in spending or for an expansion of services to a broader population for the same expenditure. We hope that our findings, along with the additional research we recommend, will help public programs make the most effective use of long-term care dollars. This research was conducted at the Center for Personal Assistance Services with funding from the National Institute on Disability and Rehabilitation Research (Grant no. H133B080002). 20 HEALTH AFFAIRS JANUARY 2010 29:1

NOTES 1 U.S. Department of Health and Human Services. The future supply of long-term care workers in relation to the aging baby boom generation: report to Congress. Washington (DC): DHHS; 2003. 2 U.S. Census Bureau. Survey of Income and Program Participation 2004 panel wave 5 topical module microdata file. Washington (DC): U.S. Census Bureau; 2009. 3 National Center for Health Statistics. National Health Interview Survey 2007 microdata file. Hyattsville (MD): NCHS; 2008. 4 U.S. Census Bureau. American Community Survey 2007 public use microdata file. Washington (DC): U.S. Census Bureau; 2008. 5 National Center for Health Statistics. National Nursing Home Survey 2004 microdata file, revised. Hyattsville (MD): NCHS; 2009. 6 Agency for Healthcare Research and Quality. Medical Expenditure Panel Survey 2005 and 2006 home health visits microdata files. Rockville (MD): AHRQ; 2008. 7 LaPlante MP, Kaye HS, Kang T, Harrington C. Unmet need for personal assistance services: estimating the shortfall in hours of help and adverse consequences. J Gerontol B Psychol Sci Soc Sci. 2004;59(2): S98 108. 8 The appendix is available online at http://content.healthaffairs.org/ cgi/content/full/29/1/ hlthaff.2009.0535/dc1 9 LaPlante MP, Harrington C, Kang T. Estimating paid and unpaid hours of personal assistance services in activities of daily living provided to adults living at home. Health Serv Res. 2002;37(2):397 415. 10 Arno PS, Levine C, Memmott MM. The economic value of informal caregiving. Health Aff (Millwood). 1999;18(2):182 8. 11 Desai MM, Lentzner HR, Weeks JD. Unmet need for personal assistance with activities of daily living among older adults. Gerontologist. 2001;41 (1):82 8. JANUARY 2010 29:1 HEALTH AFFAIRS 21