Introduction

Between now and 2030, the older population will increase sharply throughout Europe. And as we know, the risk of physical and psychological impairment increases with advancing age, often making day-to-day care indispensable. So as the European population ages, we may expect a major increase in care demand. However, this demand closely depends on the type of household in which the older person is living and their marital and family environment more broadly. Disabled people living alone more often receive professional care than those living with a partner or others (Arber et al. 1988; Breuil-Genier 1998; Grundy 2006; Martel and Legaré 2001; Pickard et al. 2000). Partner and family are the primary carers in the event of disability (Chappell 1991; Walker et al. 1993). Changes in the living arrangements of older people therefore have strong policy implications, because they change the balance between formal and informal careFootnote 1. But equally important are their consequences for individual wellbeing. Thus for the disabled older people, day-to-day care provided by a partner makes it possible to postpone or avoid institutionalization (Carrière and Pelletier 1995; Freedman 1996), which is generally regarded as a last resort (Oldman and Quilgars 1999). Generally speaking, couples are in a better economic situation than people living alone and those living in a couple are healthier (Glaser et al. 1997). Finally, living in a couple favours social integration. Many studies have shown that older people living alone more often suffer from social isolation, loneliness and boredom than those living in a couple (Delbès and Gaymu 2003a; De Jong Gierveld et al. 1997).

In recent decades substantial changes have affected the living arrangements of older people in Europe and in most other developed countries: while the proportions of older people living alone or in a couple have risen, the proportion of those living with othersFootnote 2 outside the nuclear family has decreased (Michael et al. 1980; Murphy and Grundy 1994; Ruggles 2001; United Nations 2005; Van Solinge and Esveldt 1991; Wolf 1995). Although these trends are similar in all countries, clear country group profiles emerge, based on observed levels. The proportions of older people living with persons other than a partner are in general much lower in northern than in southern and eastern Europe, while the opposite has been observed for levels of institutionalization (Pampel 1992; Tomassini et al. 2004). To explain these contrasts between the different European regions, some researchers have taken inspiration from the theory of the “three worlds of welfare capitalism” (Esping-Andersen 1990). For others, the current diversity of living arrangements of older people can be attributed to the survival of ancestral family systems (Reher 1998). More recent research attempts to combine these two approaches (Iacovou 2000; Ogg and Renaut 2005). This results in more complex classifications, in which certain European countries do not fit smoothly into the north-south gradient (Glaser et al. 2004).

Although the various socio-cultural contexts in which the older persons live influence their living arrangements differently, many studies have shown that certain individual characteristics play the same role in all countries. These analyses all show that, for older people everywhere, increasing age, being single, lack of surviving offspring, poor health and low educational level or low income all reduce the likelihood of residential independence. Differences between married, single, widowed and divorced persons are well documented: single persons more often live in institutions, married persons more rarely (Dolinsky and Rosenwaike 1988; Gaymu et al. 2006; Ricci 1991). Having poor health and having no surviving offspring also have a strong effect on living arrangements: they reduce residential independence and favour institutionalization (Angel and Himes 1992; Desesquelles and Brouard 2003; Soldo et al. 1990; Stinner et al. 1990). Lastly, research into the links between socio-economic conditions and living arrangements shows that older people who are better off financially are less likely to cohabit with other kin and more likely to live alone (Hoyert 1991; Pampel 1992; Van Solinge and Esveldt 1991).

Due to cohort and period effects, older people’s socio-demographic characteristics will change, thereby affecting their likelihood of living in a couple, alone, with others or in an institution. This article presents the results of demographic projections of the living arrangements of older persons in 2030 in nine European countries and illustrates the wide range of socio-political situations among the major regions (Belgium, France, Germany and the Netherlands for Western Europe; Finland and England and Wales for Northern Europe, Italy and Portugal for Southern Europe; Czech Republic for Eastern Europe). These countries are included in a European Community funded project on Future Elderly Living Conditions in Europe (FELICIE). Some authors have used this type of demographic projection (Comas-Herrera et al. 2006; Duée et al. 2005; Keefe et al. 2005; Pickard et al. 2000) to assess trends in the cost of informal care. Here we view the issue from the older person’s standpoint. After a brief analysis of their situation in 2000, we analyse how their living arrangements are likely to change in the future under the impact of changes in their marital status and health, which are among the factors with the strongest impact on living arrangements.

This study focuses on the population aged 75 and over, since the age of 75 is often a watershed in terms of risks of widowhood and disability.

Data and methods

This article presents the results of cohort component projections, mixing a pure cohort component transition model and derived projections. Projections by age, gender and marital status are made using a dynamic multi-state population projection model (Kalogirou and Murphy 2006). Disability and living arrangements are determined through derived projections.

The method employed can be summarized in three steps.

The first step is to prepare projections by age, gender and marital status. These projections are so-called multi-state population projections using a transition matrix, calculated with the LIPRO Model (Van Imhoff and Keilman 1991). The trends observed in each country over the past ten years for mortality, marriage rates and divorce rates, taken from vital records, have been prolonged. In all countries, mortality gains have been greater among married people and divorce rates have increased at these ages. Marriage rates have fallen everywhere except in Finland, France and the Netherlands. Mortality hypotheses have been adjusted to those adopted in the projections established by various national Offices of Statistics at the turn of the twenty first century. In all cases it is assumed that the life expectancy of women will increase less than that of men. Table 1 summarizes the results of a publication by Kalogirou and Murphy (2006) detailing these hypotheses. It shows that, except for men aged 75–84, the proportion of older persons with a spouse will rise despite the increasing proportion of divorcees.

Table 1 Distribution of older persons by marital status, gender and age in 2000 and 2030 (all felicie countries, weighted by country population)

The second step is to extend these projections by marital status with indicators on health, applying a prevalence ratio model to the marital status projection results. The prevalence ratio method is commonly known as the participation ratio method and also known as the incidence rate method (George et al. 2004; Siegel 2002). The outcome is a projection by age, gender, marital status and health status. Superimposing a health distribution on the projected population by marital status makes strong assumptions on causality between health and mortality. It is clear that health and mortality are interrelated. However, this interrelationship is too complex to be implemented in a macro level two stage projection as applied here. As dealing with this problem creates many technical problems two scenarios were developed (see below) in order to at least perceive what the consequences of health status improvements are, given the mortality assumptions from Kalogirou and Murphy (2006). To assess health status, three different data sources were used: the European Community Household Panel (ECHP), national health surveys, and national statistics on the populations living in institutions. The prevalence of health and disease observed in a specific population depends largely on the health question that is asked. We therefore opted for an international data source—the ECHP—that asks the same questions in all countries. The ECHP question we used was: Are you hampered in daily activities by any physical or mental health problem, illness or disability? (severely/to some extent/no). Those who answer “severely” are assumed to be in need of care. But the ECHP is restricted to private households, and living in institutions is related to health and marital status. On the basis of the ECHP alone the prevalence of disability would be seriously underestimated, and differently so, depending on the country.

Since the sample sizes in the ECHP are generally small, particularly at older ages, it does not give reliable age-specific prevalence of disability for the unmarried population. The analysis was therefore restricted to the married population only and calculates age-specific profiles of disability for the married. Then, using the proportion of married people living in institutions from national statistics, the age-profile was adjusted in order to reflect the prevalence in the total married population. Finally, national health surveys were used to estimate odds ratios of disability for widowed, single and divorced persons compared to married persons, and to apply these odds ratios to the age-specific prevalence of married persons observed in the ECHP.

For the future two scenarios were developed (Table 2):

  • In the Constant Disability Share (C) scenario the proportions of disabled by age and gender remain constant (= as observed over the time period 1995–1999).

  • In the Healthy Life Gain (H) scenario it is assumed that the additional years gained in life expectancy will be healthy years. In other words the total number of years lived in disability will remain constant. In this scenario, it is assumed that the proportions of disabledFootnote 3 by gender and marital status decrease at the same speed (by about 20%, see Table 2).

The various sub-populations obtained in this way (by age, gender, marital status and health status) are then broken down according to whether or not surviving offspring are available. In accordance with other studies (Murphy et al. 2006), the results (not shown here) show that in 2030, a higher proportion of older people will have surviving offspring than is the case today.

Table 2 Percentage of people in poor health, by age, gender and marital status in 2000 and 2030, under two health trend scenarios (all felicie countries, weighted by country population)

The third step is to combine the previous set of projections (by age, gender, marital status and health) with estimations on the relationship between health and living arrangement—living alone, with partner, with other people or in an institution.

Definitions of the four categories of living arrangements

  • Living alone: living in a one-person household.

  • Living in a couple: living with a spouse or a cohabitee and possibly with others.

  • Living with others: combines all other categories of private households (one-parent family or other forms of co-residence).

  • Living in a collective household: all persons not living in a private household, mainly in an “institution”. This category is not fully comparable across countries. Persons living in sheltered housing units are sometimes included in this category and sometimes counted in the population of private households. This bias could not be eliminated in this study.

The distribution by living arrangement is available only for the base year. This distribution is assumed to remain constant for the entire projection period. By multiplying the estimated data matrix “age by gender by marital status” with the (fixed base year) distribution by age by gender by marital status by living arrangement we get to the living arrangement projection.

Next the living arrangement projection has to be integrated with the previous set of projections by age, gender, marital status and health. The key technique used in this integrated projection is the entropy maximization technique (Willekens 1999). This technique involves using iterative proportional fitting to estimate missing data from available data. The Expectation–Maximization (EM) algorithm is a common method for solving log-likelihood equations when closed-form solutions are not readily available, as is the case when data are incomplete (in the ECHP for example). The basic principle is to specify a probability model, the parameters of which are estimated from the available data, and to apply the model to estimate the missing data (Willekens 1999). The method is applied to both the base year and the out-year projections.

The projection by age, gender, marital status and health status (in two scenarios) and the projection by age, gender, marital status and living arrangement are considered to be fixed. If these projections are combined the relationship between health status and living arrangement (given age, gender and marital status) is unknown. To estimate the full matrix “age by gender by marital status by health status by living arrangement” for the entire projection period with the entropy maximization technique an initial guess of this matrix (and thus the relationship) is needed.

The initial full matrix for the base year is derived from several sources. In Belgium, France, Germany and England and Wales the national sources (Handicap Incapacité, Dépendance survey in France (1999); Individual Samples of Anonymised Records in England and Wales (2001); Micro census in Germany (1999); census in Belgium (2001)) make it possible to use the distribution of living arrangement by all the other variables. For the remaining countries (Czech Republic, Finland, Italy, Portugal and the Netherlands) the ECHP distribution of living arrangements (living alone or not) by age, gender, marital status (married or not) and health was used. Persons living in institutions were added on the basis of census information concerning their number, age, gender and marital status; their health status is assumed to be similar to that observed in the French dataFootnote 4.

The final result is a single integrated projection by age, gender, marital status, health status and living arrangement. The marginal totals of the final projection are exactly in line with the projections from steps 1 and 2.

Results

Effects of marital status and health on the structure of living arrangements in 2000

Older people in good health are more likely to be living in a couple than are those in poor health (taking all the FELICIE countries together, weighted by country population, among persons aged 75 and over, these figures are 70 and 60% respectively for men and 29 and 19% for women (Fig. 1)). Among the married, living in a couple is scarcely at all linked to health status: they almost always live with their partnerFootnote 5 (97% for those in good health against 93% for those in poor health). The same remark applies to unmarried people but, regardless of health, they almost never have this living arrangement (except divorced men). On the other hand, good health favours residential independence among people not living in a couple. Widowed and divorced persons—whose living arrangements are very similar for a given health status—generally live alone (about 80%) if they are in good health. If they are in poor health, the proportion living at home is virtually halved, in favour of living with others or living in an institution (each of these situations accounting for about 25% of individuals). Lastly, never married people are always distinguished by a higher propensity to live in institutions and this living arrangement becomes preponderant (40%) for single people in poor health.

Fig. 1
figure 1

Structure of living arrangements of older persons aged 75 and over in 2000, by health status and marital status (all felicie countries, weighted by country population)

The greater residential independence of married and/or healthy people is observed in all countries, and the Europe-wide uniformity of living arrangements of the married population with no health problems is striking. But while poor health and/or unmarried status limit residential independence throughout Europe, there is more geographical variation on this point and living arrangements vary widely from country to country. It is mainly the relative frequency of intergenerational co-residence versus institutionalization that creates a distinction between countries. An explicit example is that of disabled unmarried women aged 75 and over (Fig. 2). It appears that in some countries their situation is determined more by family solidarity than social solidarity, while in others it is the reverse. Co-residence is favoured in Portugal (56%), the Czech Republic (42%) and Italy (44%), whereas institutionalization is favoured in the Netherlands (41%), Belgium (40%) and, to a lesser extent, in Germany (32%) and France (28%). In England and Wales and in Finland, unmarried dependent persons opt for both types of living arrangement with equal frequency (about 25% in each case), while half of them are (still) living alone.

Fig. 2
figure 2

Structure of living arrangements of unmarried women aged 75 and over, in poor health in 2000, in the separate felicie countries

Effects of trends in marital status on the future structure of living arrangement

These results, which are in perfect agreement with many other studies, combined with the expected growth in the proportion of married older people (Kalogirou and Murphy 2006) and the likely improvement in health, suggest that older people in the future will more often enjoy residential independence. The projections made under the FELICIE project distinguish between the effects of trends in marital status and those of health status on the structure of living arrangements for future cohorts of older adults. From the scenario in which the health status at a given age does not change, we can learn that it is only changes in marital status that determine the future living arrangements. In this regard, the basic trends expected over the coming decades are a decline in widowhood and, as a result of the generation effect, a higher proportion of divorced people and a constant proportion of single people. Except in the case of men in certain countries such as Finland, Germany and the Netherlands where the proportion of being single will (almost) double, while nevertheless remaining below 12%. But the extent of these changes varies by gender and by age (Table 1).

For women, widowhood will decline sharply because of the drop in male mortality and the reduction in mortality differential between the sexes. This mortality trend is more than compensating for the increase in divorce. Consequently, throughout Europe, in the future, women will more often grow old in the company of a partner. Today, between 21% (Czech Republic) and 37% (Italy) of women aged 75–84 live in a couple. In thirty years’ time this current maximum will be the minimum (in the Czech Republic, Finland and Portugal) whereas 58% of Italian women will be living their old age in a coupleFootnote 6 (Table 3). The trend is even stronger for women aged 85 and over: the proportions of those living in a couple will triple almost everywhere. In 2030, women in this age group will more often be living in a couple than women aged 75 and over today.

Table 3 Structure of older people’s living arrangements in 2000 and 2030, by age and gender, in the separate felicie countries

While men of 85 and over like their female counterparts, will less often be widowed and hence more often be living in a couple (60% in 2030 compared to 48% in 2000, on average), this will not be the case for men aged 75–84. In this age bracket, few men are widowed (19% on average) and the decline in the risk of widowhood (to only 12% in 2030) will scarcely or not at all compensate for the rise in the proportion of divorced men (increasing from 2 to 10% between now and 2030). These trends concur with those observed in recent decades for the youngest old (Delbès and Gaymu 2003b). The proportion of men aged 75–84 living in a couple will fall everywhere in the next 30 years except in England and Wales and in Italy, where it will remain unchanged.

The composition of the unmarried population will undergo profound changes in the future. Today, for example, 4% of unmarried women aged 75 and over are divorced and 84% are widows (for men, these figures are 6 and 75% respectively); these proportions will reach 18 and 72% for women in 2030 (22 and 50% for men). But as widows and divorced people have very similar living arrangements this redistribution has almost no effect on the trend for the total population in the proportions of persons living alone, with other people or in institutions. These groups will all shrink to a comparable degree, because of the lower risk of being unmarried. In 2000, taking all the countries together, nearly half of all women aged 75 and over lived alone, but in 30 years time only 38% will do so (Table 3). A lower percentage of them will be living with other people or in institutions (13 and 8% respectively compared to 16 and 10% in 2000). The same trends can be seen for men aged 85 and over; but the structure of living arrangements for men aged 75–84 will remain remarkably stable.

These trends can be seen among persons in good and poor health alike, and in all countries (Fig. 3). Everywhere, disabled persons, especially women and the oldest age groups, will more likely have the benefit of living in a couple. In 2000, 25% of disabled women aged 75–84 and 9% of disabled women aged 85 and over live in a couple. In 2030, these proportions will reach 40 and 23% respectively—higher than the proportions for women in good health today (34 and 12% respectively, on average). The same observation is valid, to a lesser extent, for men aged 85 and over: in 2000, 52% of those in good health live in a couple (43% of those in poor health); in 2030 an equivalent proportion (54%) will be found for the disabled male group.

Fig. 3
figure 3

Structure of older people’s living arrangements in 2000 and 2030, by age and health status (all felicie countries, weighted by country population)

The structure of living arrangements for older Europeans will thus change considerably in the future, owing to changes in marital status. The hypotheses on which these evaluations of future marital structures are based have been shown to be robust. They are shown to depend primarily on the current proportion of married persons and changes in mortality, with trends in marriage and divorce rates and migration playing only marginal roles. Different mortality hypotheses have been tested successively but they produce only minor variations in the estimations of future marital structures (Kalogirou and Murphy 2006).

Effects of trends in health status on future living arrangement structures

While the future of marital status and its consequences for the future structure of living arrangements is to a large extent predictable, trends in health status remain uncertain, as no clear trend emerges from the past. However, in overviewing health trends in low-mortality countries over a 25-year period, Robine et al. (2003) support the dynamic equilibrium scenario. The authors conclude that “disability-free life expectancy has evolved very differently depending on the severity level of disability: a decrease for the most severe levels of disability (institutionalisation and/or bed confinement) and an increase for the less severe levels of disability (no ADL dependency)”. The conclusion from past studies is that “at worst the increase in life expectancy is accompanied by a pandemic of light and moderate but not severe disability”. Although suggesting that loss of health scenarios should also be examined, even Parker et al. (2005) conclude: “whether these findings are indicative of a new trend, and emergence of a frail population or a minor fluctuation in an otherwise positive development of compression remains to be seen in future studies”. Furthermore, some changes in the structure of the older population (higher level of education, lower paid workload) or behavioural changes (for example, attitudes to prevention) suggest that the older people of tomorrow will enjoy a better health status than those of today. Pursuing this logic, we have drawn up an “improvement in health status” scenario in which we assume that all additional years of life expectancy will be years spent in good health and a loss of health scenario has not been consideredFootnote 7.

Over the next 30 years, we can expect an increase in the proportion of people living in couples but, as we have seen, health status has no impact on this living arrangement: whatever their state of health, the married older people almost always live in a couple and the unmarried ones almost never do. So an improvement in health will not affect the future proportion of older people living in a couple (Table 4). And the result would be similar under the hypothesis of much faster progress will be made in health. On the other hand, as a natural consequence of an improvement in health, a higher proportion of people will both have good health and be living in a couple. The main beneficiaries of this trend are those aged 85 and over: in 2000, 31% of men in this age group were both in good health and living in a couple. By 2030 this figure will reach 39% solely due to changes in marital structures, and 44% if health status improves as well.

Table 4 Structure of older people’s living arrangements by age, gender and health status in 2000 and 2030, under two health trend scenarios (all felicie countries, weighted by country population)

Further, although an improvement in health has no impact on the proportion of older persons living in a couple, there is a slight redistribution among those not living in a couple: they are slightly more likely to live alone (proportions increase by one or two percentage points) and a little less likely to live with other people or in an institution. Furthermore, of the increased proportion of people living independently, fewer will be in poor health. Thus today, for example, 18% of women aged 85 and over live alone and are in poor health. But taking into account a lower rate of widowhood, they will represent no more than 15% in 2030 and only 12% assuming health improvement. In this case, the effect of the improvement in health is as strong as that of changes in marital status. This similarity of effect applies to other population groups in poor health: women aged 75–84 living alone and men aged 85 and over living “not alone” (i.e. with others or in an institution). An improvement in health will further reduce the proportion of these particularly vulnerable sub-populations.

Comparing these two scenarios shows the strong influence of marital status on overall trends in living arrangement structure. By comparison, an improvement in health status, which in fact is not certain to occur, plays only a marginal role. But it has more impact, indeed as much impact as marital structures, on reducing the size of some population categories in poor health. Further, the effect of an improvement in health will be crucial for trends in total numbers: if health status remains unchanged, the population aged 75 and over in poor health will increase by 71% between now and 2030 (and by nearly 90% for those aged 85 and over); but this growth rate is only 41% (55%) under the “improvement in health” scenario considered here. The difference made by improvement versus stability in health is most important for certain population groups, for example those living in institutions. If health does not improve, demand for institutional care will increase by 70–80% in the Czech Republic, Finland and the Netherlands, by 40–50% in Belgium, France and Italy and by 35–40% in Germany, the United Kingdom and Portugal. Improvement in health would divide the above growth rates in the institutionalised populations by 2 or 2.5 in all countries under research (Gaymu et al. 2007).

Discussion

Future changes in marital status among older people will enable a larger proportion of those among them who are disabled enjoy residential independence: a higher proportion of women in all age groups will grow old living in a couple, and to a lesser extent this will also be true for men aged 85 and over. Both men and women will be less likely to live with other persons or in an institution. On the other hand, for men aged 75–84, the probability of opting for a specific living arrangement will remain remarkably stable. So these trends will result everywhere in men and women having rather more similar situations in their old age. Further, an improvement in health status will lead to a slightly higher proportion of older persons living alone, and those enjoying such residential independence will more often be in good health.

The down-side of the decline in widowhood will be that increasing proportions of older persons will have to cope with their partner’s disability, even with the improvement in health status envisaged here. At these advanced ages, there is a high risk that the care-giving partners will themselves have disability problemsFootnote 8 and many studies have shown that the carer role is itself damaging to health. In particular, the work-load and stress involved increase the risk of psychological disorders, particularly depression (Schulz et al. 1995). And whichever scenario we take for health trends, the proportion of men having to cope with a dependent partner will increase sharply in the future, whereas for women this proportion may remain stable. In 2000, 17% of women compared to only 4% of men had a partner aged 85 or over in poor health. As couples survive longer, these percentages will be 21 and 12% in 2030 or, if health status improves, 16 and 10%. Today, men find it more difficult to cope with their partner’s dependency than women do: they more often rely on professional carers (Martel and Legaré 2001) and more often put their partners in an institution (Delbès et al. 2006). The decreased risk for women of living alone will therefore only be combined with better care when their health fails if the men of tomorrow provide this better care. This probably means a fundamental change in men’s attitudes and behaviour. The same applies to all tomorrow’s potential carers. In 2030, a higher proportion of older persons than today will have surviving offspring (Murphy et al. 2006), but we cannot tell to what extent these children will be available to provide care. Today, throughout Europe, families and essentially female relatives provide most of the solidarity for the frailest older people (Mestheneos and Triantafillou 2005). Those of tomorrow, having found their identity in areas of life other than their family roles, mainly in their working careers, may be less willing to accept the constraints imposed by the health problems of their partners, parents or parents-in-law. One of the limitations of this research is that it does not attempt to take account of behavioural changes (in the nature of parents’ and children’s obligations, different generations’ desires for residential independence, social norms etc.) or possible changes in policies regarding care of the disabled, whereas trends in these factors could transform the structure of living arrangements for older people not living in a couple far more radically than the changes described here.

The same applies to the economic situation of the older adults, which is another fundamental determinant of their living arrangements. In recent decades, throughout Europe, older persons have turned away from intergenerational co-residence in favour of staying on in their own home, and improvements in their economic situation have greatly contributed to this change (Michael et al. 1980; Mickus et al. 1997; McGarry and Schoeni 2000). Future trends in older people’s purchasing power will therefore be an important factor in any continuation of this trend. And everywhere in Europe there is uncertainty as to the future financing of old-age pensions and the financial capacity of tomorrow’s older people to call in professional carers for day-to-day support. Furthermore, although many countries in recent years have introduced specific policies to provide support in this area, might not this collective solidarity gradually decline in the face of ever increasing expenditure on social protection? Future policy orientations may also affect other areas that have not been included in this study, such as residential accommodation, house prices, and migration. Each of these could influence the price of formal care and could completely change the quality of life for older people still living at home.

Although the changes in the socio-demographic characteristics of the older population analysed here will not substantially alter the contrasts in older people’s living arrangements in Europe, some other contextual factors might have a much greater impact. Today, older people’s living arrangements differ radically between southern and northern Europe; co-residence of generations and family care are favoured in the south, while collective care and institutionalization are favoured in the north. However, family behaviour patterns and government policy (particularly policies on the funding of care for the dependent) are inseparable and interdependent. For example, is the continued preponderance of intergenerational co-residence in southern Europe not partly due to the fact that there are no alternatives? And will the trend towards reducing institutional care—as is now the case in some countries like the Netherlands (Jacobzone et al. 2000)—not in the long run shape new forms of family solidarity? The extent to which families will be involved in future informal care and next to that the (changing) economic resources of older persons and possibly changing future policy orientations within each country, will play important roles in directing the trends towards greater uniformity in living arrangements for the older Europeans of tomorrow. In all countries, these trends will have a major impact on the balance between formal and informal care. Today, even in countries with an extensive formal care structure, the majority of care is provided by family members. Nevertheless informal care cannot provide all the answers and in anticipation of foreseeable changes in older people’s living arrangements the expansion of formal support for family carers has also to be considered in the future.