Exploring Assistance in Sweden and the United States

Abstract

Purpose:Few international comparisons of health services are performed using microlevel data. Using such data, this paper compares the need for and receipt of assistance with activities of daily living (ADLs) in comparable samples in the United States and Sweden, a country with a universal system of community-based services.Design and Methods: Data from national surveys of community residents completed at approximately the same time in each nation are used to create comparable measures of need and assistance. Descriptive and logistic regression analyses compare need and assistance patterns across the nations and identify individual factors that explain receipt of assistance and unmet needs.Results:Our results indicate that a simple story of greater use of paid formal services in Sweden and more unpaid informal use in the United States masks a more complex relationship. Assistance with ADLs seems to be more targeted in Sweden; narrow differences in assistance widen considerably when the analysis is limited to those reporting need. Implications:Although these two different health systems result in similar levels of overall ADL assistance, a detailed microlevel comparison reveals key distinctions. Further microlevel comparisons of access, cost, and quality in cross-national data can further aid our understanding of the consequences of health policy.

The aging of the population that now characterizes most industrialized societies has been marked by an increase in the number of healthy and independent older people, but also by growing numbers of people with disabilities who require regular, ongoing assistance with one or more instrumental and personal activities of daily living. The impact of a growing frail population of elders has led to concern variously for the emotional and physical burden faced by family members and the financial burden to society of increased health and long-term care costs. Countries have differed historically in how they meet these long-term care needs. The United States has emphasized an individual's responsibility in arranging and paying for care. Although a patchwork set of community programs assist disabled elders and their families, there is no universal system that helps individuals or their families purchase services or that provides those services directly. Other countries have developed extensive systems of publicly funded, community-based care. The explicit goals of these programs include reducing the burden of care on families while maximizing the older person's possibility of remaining in the community.

The present study is a comparison of care provided by family and formal providers to the old-old (75 years and older) in two countries, the United States and Sweden. These countries differ considerably in historical and current patterns for funding long-term care, thereby providing a very different context for care for disabled adults and their families. In the more individually oriented system in the United States, the lack of a well-developed and affordable formal care system would likely mean that families would have to play a greater role in providing care to disabled elders. By comparison, Sweden has a well-developed system of long-term care with an emphasis on universal access and affordability and the prevention of institutionalization (e.g., Care of the Elderly in Sweden, 1992; Coleman, 1994; Sundström & Thorslund, 1994). Various types of care are provided in the home, such as nursing services and assistance with activities of daily living, for little or no cost to the older person. Adult day care and overnight respite are also widely available (Jarrott, Leitsch, Zarit, Berg, & Johansson, 1998). These programs are designed to support an older person at home without placing extensive and unmanageable demands on the family. Most of the key elements of the system have been in place for over 20 years and have wide public acceptance.

Although these differences in approach suggest different outcomes for families and their elders, only a few studies have examined how much families and formal services contribute to care of disabled elders in countries with different policy structures (e.g., Davey & Patsios, 1999; Habib, Sundström, & Windmiller, 1993). This type of descriptive comparison can be useful as a first step in understanding how families respond to the needs of the older generation in different social and policy contexts and also how much formal care is used where cost has largely been removed as a barrier.

Despite changes in the modern family such as decreased family size, increased geographic dispersion, and extensive employment of women, families remain the most important source of help to disabled elders. Concern that formal services might substitute for informal care is not borne out by the available evidence (Denton, 1997; Doty, 1992; Penning, 2002; Stoller, 1989). Even in the most extensive welfare systems, older people receive more help from their families than from formal services (Sundström, 1986). Relatively few people rely only on formal help. Rather, formal help often comes into effect when family care is unavailable or insufficient. It complements and supports, but does not replace what families provide (Davey & Patsios, 1999; Sundström, 1986). Perhaps the key question is not whether formal care substitutes for informal, but how well the available help from both sources can meet the needs of an expanding population of disabled elders. There is mounting evidence that even in a welfare state of the Swedish type, public services are not keeping up with growing numbers of elders, and recent cutbacks in services have prompted more help from families (Sundström, Johansson, & Hassing, 2002).

The policy context is an important consideration in the decisions families make in arranging care for an older relative, but this decision is shaped by several other factors as well (Davey et al., 1999). Sociodemographic characteristics such as age, gender, and marital status are associated both with need and the availability of informal helpers. It can generally be assumed that the need for assistance increases with advancing age. Older men are more likely to be married and to have a spouse who can assist them in the event of disability, whereas older women are less likely to have an available spouse and more likely to depend on children or other relatives for assistance. Availability of a daughter or daughter-in-law may be a key factor in receiving help, although sons often provide some assistance, such as help with managing finances, and even personal care if no one else is available to do so.

Cultural traditions may also affect patterns of care. One of the key influences of culture is on patterns of coresidence of elderly parents and their children. Countries vary historically in the proportion of elderly parents who share the same household with their children (Sundström, 1993). Although the number of shared households has been decreasing in virtually every country that has been studied, the Scandinavian countries started at a much lower level of coresidence and remain lower than other countries. In Sweden, approximately 5% of elderly people live with an adult child, compared with approximately 15% in the United States (Sundström, 1993). In these multigenerational households, there is typically someone available who can provide assistance to elderly parents if it is needed, usually a daughter or daughter-in-law, but sometimes other family members.

A study by Habib and colleagues (1993), which compared rates of formal care in Sweden and Israel, illustrates the importance of shared versus separate households. The amount of formal care was higher in Sweden than Israel, which has a much more limited system of services. This difference, however, was due mainly to the fact that a greater proportion of older Swedes lived alone, whereas coresidence with a child who provided assistance was more common in Israel. In both countries, formal services were directed toward older people who lived alone.

These findings are consistent with research from the United States on predictors of formal service use. People who receive formal services are more likely to have greater levels of disability and to have fewer or no informal helpers to provide assistance (Bass & Noelker, 1987; Branch & Jette, 1983; Cantor, 1979; Chappell, 1992; Chappell & Guse, 1989). Kemper (1992), however, found that the presence of family help also increased the use of formal services.

Whereas several models have been proposed to explain older adults' preferences for informal support from a variety of sources (e.g., Chappell & Guse, 1989; Penning, 1990; Shanas, 1979; Silverstein & Litwak, 1993; Tennstedt, Crawford, & McKinlay, 1993), we are only beginning to extend these empirical and theoretical models to the investigation of care mix (e.g., Denton, 1997). To date, several models have been presented to account for the interface between formal and informal care systems. The chief focus for much of that research has been on the issue of whether or not the public provision of formal services reduces the use of informal help.

It is not entirely clear how policy and other factors result in the mix of formal and informal services used. There is relatively little empirical support in any country for the notion that formal services cause family members to provide less support for their relatives (Denton, 1997; Femia et al., 1997; Kemper, 1992; Noelker & Bass, 1989; Penning, 2002; Pezzin, Kemper, & Reschovsky, 1996; Stoller, 1989; Stoller & Pugliesi, 1991). Negative support for such a perspective is also seen in research on the institutionalization process, wherein families typically continue to provide high levels of assistance following institutionalization (e.g., Zarit & Whitlatch, 1992). A closer investigation of studies reporting a negative correlation between formal and informal support suggests that these studies often report negative correlations within assistance for a particular activity of daily living (ADL) domain (e.g., Denton, 1997). The balance of previous research appears to support either the supplementarity or complementarity perspectives. Data currently available to social scientists, however, generally remain insufficient to disentangle these competing explanations or to further reveal how these and other factors interact to determine the patterns of formal and informal assistance used.

Beyond the question of who provides help, the bottom line is whether older adults receive sufficient amounts of assistance to support them in the community. Unmet need can be defined as needing help to carry out an ADL, but not receiving that help (Allen, 1994; Allen & Mor, 1997; Branch, 2000). A study of disabled adults found that unmet need for assistance varied, depending on the specific ADL (Allen & Mor, 1997). Rates of unmet needs for instrumental ADLs (IADLs) were generally higher than for personal ADLs. Unmet needs ranged from a low of 4% (eating) to a high of 35% (heavy housework). People with a greater number of unmet needs were more likely to be under the age of 65, to have fewer or no potential helpers, and to have more limited financial resources. In another examination of unmet need that focused on married couples, wives were found to have more unmet need, particularly for IADL activities, than husbands (Allen, 1994). Husbands caring for their wives tended to provide less help, particularly when their wives' disability was relatively mild.

To date, few comparative studies of assistance and unmet need have been conducted (Davey & Patsios, 1999; Rowland, 1992). Furthermore, most of the available comparative studies on long-term care have involved descriptions of programs or of aggregate data on service utilization (e.g., Schieber, Poullier, & Greenwald, 1993), rather than comparisons at the individual level. There are certainly pitfalls involved in cross-national comparisons, notably that they inevitably confound policy and culture. The cultural norms and values of people in a given society certainly affect service use independent of policies that determine cost and access. Cultural factors may also influence development of policies and services. Although little can be done to control for these inherent problems, cross-national comparisons can contribute useful empirical data to policy discussions, when they are interpreted with appropriate caution.

The present investigation examines family and formal help to the old-old (75 and older) in two different social and policy contexts, Sweden and the United States, and addresses the following questions:

1. How does need for care compare between the two systems?

2. How do the two systems, one with extensive publicly supported services like Sweden and one without such services, like the United States, compare in receipt of care for the whole population, as well as for those in need of assistance?

3. How is receipt of care related to personal characteristics of the individuals, and do cross-national differences in assistance persist once these characteristics are controlled?

4. Are the rates of unmet need lower in Sweden, where the more extensive formal system may be able to compensate for inadequate or unavailable family help?

Data and Methods

This study is a two-group comparison (Sweden and the United States) of rates of utilization of informal and formal community-based services. Public policies have a major influence on access, availability, and affordability of services, but other factors are likely to play a role as well. In our comparisons of Sweden and the United States, we examine patterns of assistance in the context of several key variables. We do so for two reasons. First, if the two countries differ on a variable associated with service use (e.g., if a higher proportion of elderly people live alone in one country compared with the other), that could account for some of the differences found in service use. Second, it is possible that formal and informal help will differ between the two countries for some particular subgroups (e.g., formal services augmenting the help to a spouse more in one country than the other). The factors we consider are as follows: (a) potential availability of informal help, (b) current functioning, (c) age, (d) gender, and (e) socioeconomic status. These variables are described in the paragraphs that follow.

This study involves a secondary analysis of existing data sets. For the United States, the source of data is the 1992 Medicare Current Beneficiary Survey (MCBS). The MCBS is a nationally representative longitudinal panel survey of Medicare beneficiaries conducted by the Center for Medicare and Medicaid Services (CMS). As the health insurance program for aged adults in the United States with strong incentives for participation, Medicare beneficiaries represent almost the complete population of persons over the age of 65; a comparison of CMS enrollment figures with Census Bureau population estimates indicates that approximately 98% of persons over the age of 65 participate in either or both of Medicare's programs, Part A (Hospital Insurance) and Part B (Supplemental Medical Insurance). Nonparticipants include some who are not eligible for the program (often recent immigrants who do not meet program qualifications) and those eligible persons who choose not to participate.

The MCBS includes information on use of community-based long-term care services for a representative sample of older people in the United States. MCBS also collects extensive information on individuals' use and expenditures for health services, sources of payment, type of health insurance, access to care, health and functional status, and socioeconomic and demographic characteristics. The MCBS represents an ideal data set for the proposed research. No other public use file can match it for depth or breadth of information about health services utilization patterns among elderly individuals. More information on the MCBS can be found at http://www.cms.hhs.gov/mcbs/default.asp

The study population for this investigation is 4,583 community-dwelling Medicare beneficiaries aged 75 and older who were interviewed for the 1992 MCBS. Longitudinal sampling weights were applied to the data to ensure that they were representative of U.S. Medicare beneficiaries aged 75 and older. The sum of the sampling weights was set equal to the number of respondents in order to give an accurate reflection of the statistical power of our analyses (i.e., the weights were normalized prior to analysis).

Data for Sweden were obtained from the 1994 survey, Hemma På Äldre D'ar—Aging at Home (AH; National Board of Health and Welfare, 1994). The survey investigated health, functional status, utilization of formal, community-based services, assistance from informal providers, and sociodemographic characteristics. AH is a national, representative sample of 1,378 people aged 75 and older living in community housing. The sampling frame was the total registry of the population. The sample was stratified to yield approximately equal numbers of men and women in three age groups: 75–79, 80–84, and 85 and older. As a result, men and the oldest-old were oversampled. Sampling within each combination of age group and gender was random. Normalized sampling weights were again applied to the data to ensure that the data were representative of the Swedish population aged 75 and older. The two surveys include comparably worded items for the critical domains of demographics, perceived health, limitations in ADLs and IADLs, formal service use, assistance from informal caregivers, and satisfaction with health care services. Using these survey questions, we have constructed equivalent measures from each data set.

Demographic Information

Information was obtained on age, gender, and socioeconomic status. For age, it was possible to classify respondents into three categories: (a) 75–79, (b) 80–84, and (c) 85 and older. Sufficient numbers of respondents at these ages are available in each survey. We examined several indicators for comparing socioeconomic status between the two samples. In the end, the variable that seemed most comparable between the two countries was formal education. It was possible to categorize people in the samples as having (a) some grade school education (grundskola in Sweden), (b) high school education (gymnasium and equivalents in Sweden), and (c) college level education. Although education does not tell us how many economic resources are available to a person, it is correlated with income. Education also affects service use, as people with more education are often more informed consumers.

Functional Disability

Current functioning is measured by the amount of assistance someone needs with IADLs and personal activities of daily living (PADLs). We also sum these two to examine assistance with both sets of ADLs. IADL and PADL disabilities indicate the need for assistance, and they are important to consider when patterns of service use are examined. IADLs common to both surveys were housework, meal preparation, and shopping. PADLs common to both surveys were receipt of assistance with bathing, dressing, transferring from a bed or chair, walking, and toileting.

Both the MCBS and the AH ask a short series of questions (two to four questions) about each PADL and IADL. The translation of the wording for the individual activity questions that we were provided varies slightly in the Swedish survey, but each question generally begins by asking respondents whether they were able to manage the activity or needed help with it. A follow-up question for the IADLs alone asks whether the subjects were unable to manage the activity without assistance. We define a limitation for the Swedish sample as needing help or being unable to manage the activity without assistance.

The MCBS asked respondents whether they had any difficulty performing the activity or did not do it. If they did not do it, they were asked whether that was because of a health reason. Individuals in the MCBS were considered to have a limitation if they had any difficulty or did not do the activity because of a health reason.

Prior to conducting the analysis, we compared different ways of constructing comparable measures from the MCBS to the AH. The measures we use in the analysis, already described, were those that appeared to be most comparable in need for assistance, given the language differences. Because so few MCBS respondents reported not doing PADL tasks for health reasons, biases introduced by the difference in survey format are likely to be quite small.

We looked at the number of IADL and PADL items a respondent needed help with and a total ADL score representing the sum of IADL and PADL items. We also created a summary variable that classifies people as follows: (a) needing no help, (b) needing help with IADLs only, (c) needing help with PADLs only, and (d) needing help with both IADLs and PADLs. Given the hierarchical relation of IADLs and PADLs (Lawton, 1971; Spector, Katz, Murphy, & Fulton, 1987), there were relatively few people who needed help only with PADLs. A shortcoming of these data is that, on one hand, the sets of three IADL and five PADL items do not correspond to standard scales. On the other hand, we have data on most important functions.

Potential Availability of Informal Help

Following Habib and associates (1993), we constructed an index of potential availability of help as follows: (a) elder lives with a spouse and has living children, (b) elder lives with spouse and has no living children, (c) elder lives with a child, (d) elder lives alone but has children, and (e) elder lives alone and has no children. This categorization reflects normative expectations in both countries for providing help. There was a small number of persons who were living with both a spouse and adult children. These persons are included in group (a). We also found small groups of elders in both countries with other living arrangements (e.g., living with a boarder who provides some help), but these groups were too small and heterogeneous to be included in the analysis. We also could not take account of whether or not living children were located nearby their elderly parents.

Informal and Formal Assistance

In each survey it was determined if a respondent received help with a PADL or IADL and who provided that help. Helpers were classified as informal (from family, friends, or volunteers) or formal (provided by paid helpers or agencies). We classified people into 4 groups: (a) receiving no help, (b) receiving informal help only, (c) receiving formal help only, and (d) receiving both informal and formal help. The last two categories were combined because of small numbers of persons who only received formal help. It was also possible to determine unmet needs. Respondents who reported a disability in a PADL or IADL but did not receive help with that function were considered to have an unmet need. It should be noted that unmet need here is conceptualized as having no help for an ADL. This underestimates unmet need, because it does not include situations in which the individual receives insufficient help.

After the data were edited for comparability, an analysis of the data was completed by using the statistical package LIMDEP. Descriptive statistics, weighted by using population weights from each survey, were computed. A regression analysis of the type of assistance received (no assistance, informal help only, and formal help only or formal and informal help) was completed by using multinomial logistic methods. Results reported in the text are the marginal probabilities of the effect of each independent variable on the likelihood that the individual falls into the assistance category. A binomial logistic regression analysis on the presence of any unmet ADL needs (defined as presence of one or more ADL limitations for which help was not received) was also completed. Both regressions were carried out on a pooled sample of individuals who had at least one ADL need, with a dummy variable indicating country (1 = Sweden, 0 = United States) included in the regression.

Results

To address our first research question, Table 1 presents information on the PADL, IADL, and ADL (the sum of PADL and IADL) needs of the persons in the U.S. and Sweden samples. Consistent with international data on the health of the U.S. and Swedish populations, this table indicates that the sample of community-dwelling elderly persons from the United States are more likely to need assistance with more ADLs (except for a slightly greater level of IADL need among Swedish persons with at least one need). The difference between the two nations is most dramatic in the area of PADLs, where the percentage of Americans needing assistance is nearly twice as large and more heavily weighted toward those with multiple PADL needs.

Our second research question is addressed in Table 2, which summarizes receipt of assistance across two dimensions: those who receive assistance with different types of ADLs and those who receive different types of assistance, that is, help from the formal and informal sectors. Results indicate that among all persons, U.S. residents are slightly less likely to get help with both IADLs and PADLs (10.5% vs. 13.6%). This narrow difference, however, widens considerably if we restrict our analysis to those who have at least one ADL need (20.7% vs. 32.3%). In the Swedish survey, under 2% of those with an ADL need receive no help (have an unmet need), compared with more than one fifth of Americans.

Table 2 details the type of assistance people receive in each country. The top set of rows indicates whether the individual receives assistance with PADLs, IADLs, both, or neither. The bottom set of rows shows whether or not the person receives unpaid informal help, paid formal help alone or in combination with informal help, or no help at all (because the formal help information in the MCBS does not tell which ADLs are assisted, the numbers receiving no help differ slightly in the United States when these two measures are used). For each country, the chart shows the type of help received for all persons, and only for those reporting at least one ADL need. Thus, if we examine all persons, Swedish residents are somewhat more likely to be getting paid formal help alone or in combination with informal help (18.9% vs. 17.0%), whereas U.S. residents are relatively more likely to be getting informal help alone (i.e., without formal help; 26.7% vs. 22.7%). If we examine only those persons who have at least one need, however, differences in receipt of informal care alone are reversed, with Swedish persons being slightly more likely to get only informal help (53.6% vs. 52.6%). Among those with an ADL need, the difference in receipt of formal help is much more dramatic, with 45.0% receiving formal help in Sweden, compared with 27.4% in the United States.

Figure 1 further details this difference, showing the levels of unmet need for PADLs, IADLs, and all ADLs. This chart shows the percentage of persons in each country who have at least one PADL, IADL, or ADL need for which they report receiving no assistance. More than one fifth of Americans have at least one IADL need for which they receive no help; more than two fifths of Americans have at least one PADL need for which they receive no help; and almost three fifths of Americans have at least one IADL or PADL need for which they receive no help. In all cases, the level of unmet need among Swedish elders is less than 4%.

These differences in receipt of care are determined, of course, in part by the characteristics of individuals. To begin considering our third research question, Table 3 summarizes the sociodemographic characteristics of respondents in the two samples. Several interesting patterns emerge. In the full sample, the Swedish sample is slightly younger and has more women. Educational levels are also lower in the Swedish sample. The most important contrast in the full sample is in the potential availability of help; U.S. residents are much more likely to live with children than those in Sweden.

There are some important differences in the samples by types of assistance received. In Sweden, informal care tends to be much more heavily used by men and people with a spouse, whereas in the United States informal care is a bit more heavily directed toward women and those living with children. Formal care or a combination of formal and informal care in Sweden is likely to be received by women, those over the age of 80, and those living alone. In the United States, formal care is also directed more toward the oldest-old, women, and those living alone or living with children.

Tables 4 and 5 test these univariate findings by using multivariate regressions to address our third and fourth research questions, respectively. First, we estimated a multinomial regression on type of assistance received by using the sample of persons with at least one ADL need. The dependent variable was a three-category classification of no assistance, informal assistance alone, and formal assistance (either with or without informal assistance). Table 4 reports the marginal probabilities for the effects of the independent variables on the probability of falling into each of the three classes.

In general, age, living arrangements, and the level of ADL need (in addition to country) predict type of assistance received. Gender and education rarely contribute to the explanation of the types of assistance received. In comparison to those aged 75 to 79, persons over the age of 80 have a slightly lower probability (between 1.9% and 4.1% lower) of receiving no assistance and a higher probability (between 3.7% and 8.0% higher) of receiving formal assistance (alone or with informal assistance). In comparison with those living with a spouse and having children, those who live alone (either with or without children) are more likely to receive no assistance or informal assistance alone and less likely to receive any formal assistance. Finally, the number of IADL needs, but not the number of PADL needs, helps explain receipt of no assistance, whereas both types of need increase the probability of formal help. PADL need has a much larger effect on formal assistance than does IADL need. The two need variables have contrasting effects on the receipt of informal assistance alone; greater PADL need is associated with a lower probability of receiving informal support alone, whereas greater IADL need is associated with a higher probability of receiving this pattern of support.

After the effects of these other variables are removed, the country variable still contributes to explaining the differences in types of assistance. Persons in Sweden are much less likely (probability lower by almost 23%) to receive no assistance and much more likely to receive some formal help (probability higher by 30%). Although the effect is small, Swedish residents are less likely to receive informal assistance alone, with the probability of such assistance lower by approximately 7%.

Table 5 shows the results from the logistic regression on the presence of unmet need for any ADL. Persons over the age of 80 are approximately 5% less likely to have an unmet need, with a slightly larger effect above the age of 85. Those who live with their children are also less likely to have an unmet ADL need, with the probability of having an unmet need reduced by approximately 12%. Living alone, either with or without living children, increases the probability of unmet need. There is no significant effect of gender or education on unmet need.

After these individual level factors are controlled for, the descriptive data on unmet needs in the two countries are supported. Persons in Sweden are only 10% as likely as Americans with similar characteristics to have an unmet need. This factor is clearly the most important determinant of unmet need among the independent variables that could be included in this study.

Discussion

We set out to examine whether older adults in the United States and Sweden differed in need for, and receipt of, formal and informal assistance with ADLs. We found that in almost all respects older persons in the United States report greater need for ADL assistance than do older persons in Sweden. Whether this is because of objective factors such as age or poorer health or subjective, cultural factors such as differences in expectations of American and Swedish older persons, other factors such as socially structured inequalities by race, economic, or other differences cannot be unequivocally determined through our analysis. Previous studies do indicate differences in cultural or national expectations in ratings of need (e.g., Donelan et al., 1999). Measurement differences across surveys are also likely to play a role. Specifically, the MCBS asks whether the respondent has difficulty performing an activity, whereas the AH asks about need for help. As a result, the reported differences in levels of need between the U.S. and Swedish population is likely to be overestimated because levels of need for help in the United States are likely to be too high. Likewise, differences in estimates of unmet need are also likely to be too high in the United States because some of the U.S. respondents with no help may not consider themselves to have need for assistance, despite experiencing difficulty performing ADLs.

Despite the greater reported need for assistance among Americans, the descriptive and multivariate regression results suggest that Swedish residents receive more help overall. When the entire sample is compared, differences in the types of ADLs for which a person receives assistance and differences in the types of assistance provided are narrow, though largely suggestive of greater assistance for Swedish residents. More important, however, if the analysis is restricted to those who have one or more ADL needs, there are dramatic differences in assistance patterns.

One of the clear patterns is that few people in Sweden who report a specific ADL need fail to receive assistance for it; in the United States, a substantial portion of people with needs receive no assistance and many people who report no ADL need do receive assistance. Although there are too many missing factors in this analysis to draw a definitive conclusion, this may be driven in part by the structure of formal care in the two systems. Formal care in the United States largely targets those with a short-term postacute care need. Thus, it may be that a substantial portion of the U.S. sample had no long-term ADL need, but did have a short-term need for formal services that was met. In contrast, the Swedish system targets formal services to those with long-term ADL needs. The United States, without a well-coordinated system of long-term care, fails to meet the needs of its residents in this area.

These findings also suggest that cultural differences play an important role in determining how much informal help is given. Although gender showed some interesting relationships in descriptive data, the effects of gender in the multivariate model were not significant. However, these may have to be explored more carefully. These differences may, in part, be due to the effect of gender on perceived need for help, with older men viewed as needing more help from family in Sweden and older women getting more help from family in the United States.

Another consideration is that older Swedes are more likely to live alone and get somewhat less help from informal sources than people in the United States. As this preference antedates the development of old-age care programs (Sundström, 1993), it may be that informal help is lower for cultural reasons, rather than as a result of substitution of formal for family help. Davey and colleagues (1999) found that rates of informal helping were actually relatively higher when elders lived with adult children in Sweden compared with the United States. Of course, in all likelihood, policy reflects the cultural preferences for independence, supporting a context in which people can live alone, which in turn lowers the amount of help these people receive from informal sources.

Our results clearly indicate that cross-national microdata studies of health care offer a promising path for future research. Simple comparisons of the aggregate data across these two nations would hide some of the interesting links discovered in this study. Much fruitful research on international aspects of income support has benefited from recent developments that matched microdata efforts in developed nations (Palmer, Smeeding, & Torrey, 1988). Extending this research into studies of microdata on health systems is a great need. Our project also offers insights into the limitations facing current research in this area. Difficulty in matching questions, particularly on the need and use of services and on areas of income and socioeconomic status, is natural. Two final issues that cannot be addressed in the current framework, either analytically or theoretically, is the role played by ethnocultural variability, especially within the United States, and the importance of residential propinquity in care mix. Work on survey developments in this field has to stress collaboration for improved data comparability at the levels of language, content, and the meaning of key constructs themselves.

References