Global Mortality Rates: Variations and their Consequences for the Experience of Dying

Clive Seale. Handbook of Death and Dying. Editor: Clifton D Bryant. Volume 1. Thousand Oaks, CA: Sage Reference, 2003.

The statistical distribution of mortality varies considerably, both historically and across regions of the world. Most obviously, life expectancy has risen over time, yet people in some countries can expect to have much longer lives than people in others. Within countries there are marked differences in the age distribution of death for different groups; for example, life expectancy is influenced by gender and by socioeconomic differences. Additionally, there is variability among regions of the world in different causes of death, and there are great differences among countries in the availability of formal health care. Given the combination of these factors with cultural variations that affect matters as diverse as family size, gender and filial roles, beliefs about how health professionals should behave, and religious customs, it is clear that there is potentially a great deal of variation in people’s experiences of dying and of care before death.

My purpose in this chapter is to map out some of this variation, initially by reviewing available statistical data about the matters I have mentioned and then by addressing the consequences of these patterns for the experience of dying in different parts of the world. I draw also on some qualitative data about the experience of dying. Toward the end of the chapter, I consider future prospects, given the continuation of present trends.

Longevity

In 1995, average life expectancy at birth worldwide was 65 years, having risen from just 48 years in 1955. The World Health Organization (WHO 1998a) has predicted that average life expectancy will rise to 73 years by 2025; by that date, WHO estimates, no country will have an average life expectancy of less than 50 years. Reductions in infant mortality and in early childhood deaths are responsible for most of this increase. In 1955, the infant mortality rate (IMR) worldwide is estimated to have been 148 per 1,000 live births; by 1995 the IMR was 59, and WHO (1998a) predicts that it will reach 29 by 2025. Mortality rates for children under 5 years old have shown similar declines (WHO 1998a).

Underlying these global figures is considerable diversity. Table 1 displays life expectancy and infant mortality rates for the major regions of the world from 1995 through 2000. The country with the highest life expectancy and lowest IMR during this period was Japan (80 years and 4.3), and the country with the lowest life expectancy and highest IMR was Ethiopia (37.5 years and 169.3) (WHO 1998b). Table 1 shows the impact of these changes (as well as the impact of reductions in fertility rates that tend to follow reductions in mortality rates) on the age structures of populations. Broadly speaking, under the circumstances of such a demographic transition, the proportion of elderly people in populations rises, and the experience of dying is increasingly a feature of old age.

This chapter is reprinted from Social Science and Medicine, 51, “Changing Patterns of Death and Dying,” Clive Seale, pp. 917-930, copyright 2000. Used with permission from Elsevier Science.

However, global mortality figures mask the fact that the picture of improvement is not universal. In 16 countries, average life expectancy decreased in the period from 1975 through 1995 (WHO 1998a). In particular, the trend in certain Eastern European countries stands in sharp contrast to that in other developed countries, showing a considerable divide between the newly independent states of the former Soviet Union and the rest of Europe. In the Russian Federation itself, life expectancy for males dropped from 64.3 in 1985-90 to 57.5 in 1994; for females, the figures went from 74.4 to 71.3 (Pearce et al. 1997). These declines reflect increases in particular causes of death, influenced by deterioration in the public health services as well as worsening material and social conditions, with high alcohol consumption playing a major role in mediating the stress of these conditions. Middle-aged males seem to have borne the brunt of this worsening mortality picture, with increases in deaths from cardiovascular disease, accidental poisonings, and suicide, and, most strikingly, large increases in death by homicide, so that by 1993-94 Russia overtook the United States as the country with the highest homicide rate (Whitehead and Diderischen 1997; Okolski 1986). Since 1994, however, mortality rates have improved in Russia, and by 1998 they were back at the levels they had been in the early 1980s (Shkolnikov, McKee, and Leon 2001). A similar decline followed by a rise to previous levels has been experienced in Romania, and for quite similar reasons, with the exception that an epidemic of pediatric AIDS (stemming from some disastrous practices by health care providers specific to the 1980s) played a part, adding yet more premature death to that caused in middle-aged adults by cirrhosis and cardiovascular disease (Dolea, Nolte, and McKee 2002).

Additionally, AIDS is having a serious impact on overall life expectancy in some African countries, where the spread of this disease has been greatest so far. In Uganda, for example, where AIDS is the leading cause of death for young adults, life expectancy at birth in 1995-2000 was 41.4 (WHO 1998b), whereas once it was 56.5 years (Gilks et al. 1998). In the nine African countries with adult HIV prevalence of 10% or more, life expectancy by 2010-15 will be 47 years on average, compared with the 62.4 years that would have been the case without AIDS. Because of transmission to infants, there will also be rises in infant mortality in such countries (UNAIDS 1998a). Figure 1 shows changing life expectancies in African countries, illustrating the impact of AIDS. UNAIDS (2002) has expressed concerns that this experience in Africa may be repeated over the next few years in countries in South and East Asia, where the epidemic is at an earlier stage.

In spite of these reversals, the overall trend worldwide is toward longer life and the aging of populations. This has particular consequences for the experience of dying, as this experience becomes increasingly merged with the general problems of old age. The trend toward longer life also has consequences for the experience of bereavement, as death occurs to people who have often been excluded already from participation in mainstream social and family life, so that they are to a degree “socially dead” and already mourned for before they are biologically dead. Additionally, the aging of populations has consequences for the sources of care on which dying people can draw. Other factors, however, also play a part in influencing the experience of dying. These include cultural differences in the living arrangements and social status of elderly people, the roles played by gender and social inequalities, the nature of health services available to elderly and dying people, and the changes in the causes of death that have led to the increases in longevity already reviewed. In all of these areas there are marked differences between “developed” and “developing” countries.

Causes of Death

As is well-known, the demographic transition that most developing countries undergo involves a shift in causes of death, broadly speaking, from a predominance of infectious diseases in pretransition societies to degenerative conditions in countries that have experienced the transition. Historically, in European countries, typhoid, cholera, typhus, measles, diphtheria, whooping cough, and smallpox were major causes of death. Epidemics caused surges of mortality compared with present-day patterns, which are less volatile. Nonepidemic infectious diseases, such as pneumonia, bronchitis, tuberculosis, and enteritis, then became more important, before the present time, when the degenerative diseases of cancer, heart disease, and stroke are the major causes of death in developed European societies (McNicoll 1986).

China in the early 1980s presents an example of a country midway through the epidemiological transition (Jamison et al. 1984). This country gained 27 years in overall life expectancy in the years between 1960 and 1980 (this figure controls for the huge surge in mortality caused by the famine period around 1960). IMR dropped from 290 in 1940-45 to 65 in 1975-80. Rural areas lagged behind urban areas in the transition from a predominance of parasitic and infectious diseases to degenerative diseases. One example of a disease that was once an important cause of mortality but that had declined in importance by the early 1980s is neonatal tetanus, which is estimated to have killed up to 10% of infants in pre-1949 China but was a rarity by 1981. Malaria control measures had reduced the proportion of the population living in areas endemic for the disease from 70% in the early 1950s to around one-third by 1979. Childhood pneumonia was estimated to be at around one-fifth its former level in the early 1960s, once the control of acute respiratory diseases became a priority for health services. Tuberculosis, although still an important public health concern, had declined considerably as a cause of death, being one of several diseases in which mass immunization campaigns had made an impact. Table 2 shows the distribution of deaths by major causes in selected rural and urban areas of China in 1982. The data displayed in the table demonstrate the difference made by the urban factor in the transition to degenerative diseases as causes of death, as well as the continuation of infectious diseases, TB and respiratory disease, as smaller but still important causes of death, particularly in rural areas. Chinese life expectancy at birth in 1996 stood at 69.9; IMR was 37.6 (WHO 1998b), suggesting further movement along the epidemiological and demographic transitions since the early 1980s.

By way of contrast, Tanzania in the 1990s presents an example of a country somewhat further back in the demographic transition whose progress in reducing mortality has been set back considerably by AIDS. In 1996, Tanzanian life expectancy stood at 51.4; IMR was 80.1 (WHO 1998b). According to a 1997 report from the Tanzania Ministry of Health concerning selected areas of the country, HIV/AIDS was the leading cause of death among adults ages 15-59, with the spread of pulmonary tuberculosis having increased dramatically in the previous decade due to the effects of HIV infection on immunity. Infectious disease accounted for between half and twothirds of deaths occurring to 15- to 59-year-old men, whose death rates were now as important in causing public health concern as infant mortality had been in previous decades. Maternal mortality (which still accounts for more than half a million deaths each year worldwide, particularly in Asia and Africa; WHO 1998a) accounted for 8-10% of deaths of women ages 15-49 in these areas of Tanzania. It was only among the elderly—who were of course a lower proportion of the population than in post-transition countries—that stroke, cancers, and chronic heart failure were predominant causes of death, although acute febrile illness and pneumonia still caused death in significant proportions of elderly Tanzanians.

In spite of the continuing progress in most societies toward and through the demographic transition, the picture worldwide in 1997 showed that infectious and parasitic diseases remained the cause of about one-third of all deaths (WHO 1998a). The most common infectious diseases causing death were acute lower respiratory infections (7.1% of deaths worldwide), tuberculosis (5.6%), diarrhea (4.8%), HIV/AIDS (4.4%), and malaria (between 2.9% and 5.2%). The diseases more commonly associated with the degenerative conditions prevalent in post-transition societies included 29% of deaths worldwide caused by circulatory diseases. Of these, the most common were coronary heart disease (13.8% of all deaths), cerebrovascular disease (8.8%), and other heart diseases (5.7%). Cancers accounted for 12% of deaths worldwide in 1997, with lung cancer being the most common, followed by colon and rectum, liver, and breast cancers (WHO 1998a).

Optimistic predictions suggest that, in spite of the impact of AIDS, there will be continuing progress at a global level away from infectious disease as a cause of death and toward degenerative disease as more countries undergo the transitions characteristic of developing countries. Nevertheless, infectious disease will be a major cause of mortality for many decades to come, particularly in the world’s poorest countries (Heuveline, Guilot, and Gwatkin 2002). Additionally, the prospect of unforeseen reversals to the pattern of transition (demonstrated, for example, by the impact of AIDS and by the surges of mortality in Eastern Europe since the fall of communism) should not be discounted. As I will show below, we know a considerable amount about the experience of dying from degenerative conditions in Western societies, but less has been recorded about the needs and dying trajectories of those whose deaths are caused by infectious diseases.

Population Structures

Although the effects of AIDS and the changes experienced in Eastern Europe suggest modifications to the picture of worldwide improvements in longevity, it is nevertheless the case that the aging of populations is the norm in most countries around the world. This trend has implications for the sources of care on which people can draw as they approach the ends of their lives. The pattern of demographic transition undergone by most developing countries involves a decline in fertility as well as in mortality rates, as the increased chances of surviving childhood are taken into account by potential parents. This creates a population structure in which there is a low proportion of younger members compared with elderly people. The age structures in different regions of the world are displayed in Table 1, which shows that Europe has the highest proportion of elderly people in its population, and Africa has the lowest. Predictions for the future draw particular attention to Latin America and Southeast Asia, which in 1998 were expected to see increases of up to 300% in the older population by 2025 (WHO 1998a). Globally, WHO (1998a) estimates that the proportion of people age 65 and older in relation to people age 20 and younger will shift from 16/100 in 1998 to 31/100 by 2025.

Because old age is associated with increased levels of disability (see below), these changes in population structures have prompted concern among some observers about societies’ ability to meet the health and social care needs of elderly people (e.g., Ogawa 1986). Insofar as these needs merge with needs for care of the dying, similar considerations apply. As a reflection of these forces, research has emerged in developed countries that has demonstrated an exceedingly high cost of health care for elderly people in their last year of life (Culler, Callahan, and Wolinsky 1995; Experton et al. 1996). Some 30% of the health care expenses of retired people in Western industrialized countries are incurred in the last year of life (Felder 1997). WHO (1998a) has expressed particular concern about the impacts of these changes in developing countries, where health and social care budgets already face difficulties. In relation to informal family care, demographic changes may be accompanied by a loosening of filial ties, as Hossain (1999) notes in a review of population aging in Bangladesh; Hossain expresses particular concerns about the situation of elderly Bangladeshi women. Other commentators have suggested that the general social status and power of elderly people decline in these circumstances (McNicoll 1986; Hull and Jones 1986). As McNicoll (1986) puts it, “Gerontocratic control in a family system, to take one instance especially important in many African societies, is inconsistent with too many gerontocrats” (p. 18). The relevance of these concerns for the “slippery slope” argument within debates about euthanasia as a social policy should not be ignored, and the relatively disadvantaged position of elderly women is of relevance here (Seale and Addington-Hall 1995; Markson 1995).

Gender Differences in Longevity

Gender differences in longevity, which are particularly evident in developed countries, mean that females come to outnumber males as age increases. In pretransition societies, apart from the fact that there are fewer elderly people anyway, this gender difference is less marked, with some societies showing higher rates of female mortality, particularly for the very young and at childbearing ages. Ware (1986) has argued that in such societies decisions about the allocation of scarce resources, such as food and medicine, often discriminate against the survival of female infants, although outright infanticide is probably rare. The effects of multiple childbirths on those women who survive their own childhoods also contribute to high female mortality. But as societies become relatively more affluent and fertility decreases, women are increasingly likely to survive to old age, and they commonly outlive men. In their historical and cross-national review of the ratios of women to men, Guralnik, Balfour, and Volpato (2000) largely rule out the role played by improvements in children’s health as a causative factor, instead claiming that changes in the probability of survival in middle age and late life explain many of the gender differentials that exist. Because of the common marriage pattern in most societies that involves older men marrying younger women, women are also more likely to experience widowhood in old age, and so are likely to approach death alone.

Sri Lanka is an example of a society where mortality rates for women in relation to men have reversed over time. Life expectancy for Sri Lankan females in 1945-47 was 43.1 years; for men it was 44.8. Mortality in childhood was higher for girls than for boys, with the exception of the first year of life, during which male mortality was greater. In the main childbearing years of early adulthood (15-24), women were at their greatest disadvantage, with this continuing at a lesser level for older childbearing years (25-44). However, for the ages beyond 45, men were more likely to die than were women. By 1970-72 a gender transition had occurred, with life expectancy for women being 66.7 and that for men being 63.8. Although mortality at ages 1-4 was still greater for girls than for boys, for all other age groups it was either the same for females and males or much improved for females (Ware 1986).

In the United Kingdom, by contrast, life expectancy has been greater for females than for males at least since 1871, which is the earliest point at which official statistics enable calculation of gender differences in life expectancy (Gray 1993). By that time, death from infectious diseases had declined considerably from earlier levels, and birthrates had been falling steeply since the 1860s, indicating a society going through the latter stages of demographic transition. By 1995, life expectancy in the United Kingdom was 74 for males and 79.2 for females (Eurostat 1997). This difference, of 5.2 years, is exceeded in some other developed countries, but the most dramatic sex differences in favor of women are to be found in some Eastern European countries, due to the excessive impact on male mortality of the social changes reviewed above. In Estonia, for example, in 1995 life expectancy was 61.7 years for men, compared with 74.3 for women (Eurostat 1997). In 1994, these figures for the Russian Federation were 57.5 and 71.3, respectively, a difference of 13.8 years (Pearce et al. 1997). Although gender differences of this magnitude may be only a temporary phenomenon in these countries of the former Soviet Union, their example suggests that there are some circumstances that justify a reversal of the public health concern over excessive female mortality applicable to developing countries.

Socioeconomic Differences

Socioeconomic variations in mortality within countries (bearing in mind that some really large inequalities occur between countries), with the richer and more educated possessing a variety of advantages that translate into longer lives, are also relevant to any consideration of the experience of dying. The adequacy with which different countries document the benefits of high socioeconomic status varies, but there are good reasons to believe that it is a universal phenomenon. Where material conditions are poor, differential access to food, hygienic living conditions, and health care are major influences. Wilkinson (1996) suggests that the persistence of inequalities in richer societies, where a basic level of access to these things is reasonably well established for all, is due to psychosocial factors, excess mortality in the worse-off being caused by physiological effects of stress arising from social exclusion. A more conventional explanation stresses the contributory role of class-related patterns of health behavior, as well as some continuing effects of material deprivation (Townsend and Davidson 1982). In the United States, racial differences in mortality that run in parallel with material deprivation and social exclusion have been particularly evident (Whitehead and Diderischen 1997; U.S. Congress 1993).

Although this means that age at death is likely to be differentiated systematically within societies, the roles of cultural differences among socioeconomic groups in attitudes toward the care of elderly people, or toward dying, have been less well documented. However, it is likely that in some developed countries there is variation in how people from different socioeconomic backgrounds feel about the extent to which they can control dying and plan for death. People’s attitudes about such matters may influence their views concerning the desirability of euthanasia or hospice care, for example, as well as the desirability of an open awareness about the presence terminal disease (Seale 1998).

Implications of Demographic Changes

Mortality differences, then, are considerable, both within and between societies. The key features of these for an understanding of the experience of dying are rising life expectancy, causing an aging of the population, coupled with the changing pattern of diseases causing death, within which there are considerable gender and class differences. The demographic and epidemiological transitions mean that dying comes increasingly to be a part of the more general experience of old age and its concomitant disabilities. There is large variation, however, in the extent of this transition worldwide, perhaps limiting the applicability of Western models of terminal or palliative care. A consideration of some of the differences in the dying trajectories involved in different diseases may help to illustrate this point.

Dying Trajectories and Needs

Some scholars have stated that the shift from infectious to degenerative disease involves a lengthening of dying trajectories (e.g., Lofland 1978; Hull and Jones 1986). Coupled with advances in medical science’s capacity to predict death at an early stage in some diseases, this has led to the emergence of a particular category of experience, that of “terminal illness,” around which has developed the expertise of hospice and palliative care in some developed countries. This simple story requires some modification, which can begin with an assessment of what is known about the prevalence of disability and symptoms, as well as the length of dying trajectories, in the time before death.

Surveys of nationally representative samples of elderly people and of people in the last year of life in the United Kingdom are a useful starting point. British government surveys have shown, broadly speaking, that the prevalence of disability reflects the aging of the population over time. Since the General Household Survey (GHS) began in 1972, the proportion of adults reporting a long-standing illness that limited their daily activities rose from 15% of people living in private households to 22% in 1996. At each time point, women have been somewhat more likely than men to report this, and the prevalence of limiting long-standing illness, unsurprisingly, shows a sharp rise with age, so that among people age 75 and over, 52% report such illness (Office for National Statistics 1998a). Studies in the United States have found similar patterns (Feldman 1986).

The perception that increased life expectancy may bring with it a greater burden of disability toward the end of life led some researchers to calculate a new statistic of “healthy life expectancy” (HLE) to modify the traditional life expectancy statistic. Between 1976 and 1994 in the United Kingdom, when life expectancy rose by more than 4 years for males and more than 3 for females, HLE showed almost no change (Office for National Statistics 1998b). As a result of this increased level of disability, 30% of people age 85 or older surveyed in the 1996 GHS needed help at home in climbing the stairs, 24% needed help with bathing or showering, 8% needed help with dressing and undressing, and smaller proportions needed assistance with other self-care activities (Office for National Statistics 1998a).

According to the World Health Organization’s (2000) report on world rates of healthy life expectancy (which WHO calls “disability adjusted life expectancy,” or DALE) in 1999, people in poorer countries “lost” some 14% of their lives to disability, compared with 9% in richer countries. The top 5 nations in terms of DALE were Japan (74.5 years), Australia (73.2), France (73.1), Sweden (73.0), and Spain (72.8). The United States ranked 24th (70.0 years) for a variety of reasons, including the poor health of some members of minority groups and high levels of violence. All of the bottom 10 countries were in sub-Saharan Africa, reflecting the effects of HIV/AIDS.

The symptoms and restrictions experienced by people in the last year of life have been recorded in three U.K. surveys that have described people dying in 1969, 1987, and 1990 by means of interviews with surviving relatives and others who knew the deceased. The first two of these studies, conducted by Cartwright, Hockey, and Anderson (1973) and Seale and Cartwright (1994), used nationally representative samples, permitting comparisons over time. On both occasions, respondents were asked to say which of a number of areas of restriction had been experienced by the people who died. These included such activities as getting in and out of the bath, dressing and undressing, and washing. The major changes concerned the length of time that such restrictions were experienced. In 1969, 30% had needed help with at least one of these activities for a year or more. By 1987, the proportion had risen to 52%. In this respect, the figures support those from the GHS.

Respondents were also asked, on both occasions, to report whether the deceased experienced certain symptoms at all during the last 12 months of life, and, for symptoms reported at this stage, how long they were experienced. The major changes since 1969 again concerned the duration of some of the symptoms: Mental confusion, depression, and incontinence were all experienced over longer time periods by people in the later study. When they had controlled for age, the researchers found that these increases were all related to the greater proportion of people age 75 or older in the 1987 study.

Recalling that cancer, heart disease, and stroke have become increasingly prevalent as causes of death as infectious diseases have declined in importance, the analyses conducted for these separate groups in the third survey (Addington-Hall 1996) are of interest. This 1990 survey, the Regional Study of Care for the Dying (Addington-Hall and McCarthy 1995), was not nationally representative, but its large size permits comparison of these different leading causes of death, as shown in Table 3. Pain, nausea and vomiting, difficulty swallowing, constipation, and pressure sores are more prevalent among people dying from cancer. Breathlessness is a particular problem for people with heart disease, whereas mental confusion and incontinence affect a high proportion of people dying from strokes. Overall, cancer caused, on average, a larger number of symptoms, and respondents considered a larger proportion of these to have been “very distressing” for the dying person. However, the duration of symptoms in cancer was less than for other conditions (Addington-Hall, Altmann, and McCarthy 1998). This pattern was also found in the earlier study of deaths in 1986 (Seale and Cartwright 1994), which suggests that an experience of longer-term disability was more typical in people not dying from cancer, who also tended to be a little older, on average, than those dying from cancer.

Researchers have conducted studies of dependency, symptoms, and needs for care in the period before death in the United States (Hunt et al. 1991; Kai et al. 1993; Dudgeon and Kristjanson 1995; Andershed and Ternestedt 1997), Germany (Bickel 1998), Finland (Hunt, Bonett, and Roder 1993), and Australia (Karlsen and Addington-Hall 1998). Equivalent information is lacking, however, for patterns of dying in countries at earlier stages of the epidemiological transition. This means that death from infectious diseases and other causes prevalent in developing countries, as well as death in younger groups, is not well described. An exception is AIDS, for which there are descriptions of Western populations before the advent of effective antiretroviral therapies (Sims and Moss 1991). However, it is likely that these Western AIDS deaths were dissimilar from typical patterns in, say, African countries, where people progress more rapidly to death due to the presence of other uncontrolled diseases such as TB, pneumonia, and salmonellosis infections, without living long enough to experience the pattern of coinfections experienced in the West (Gilks et al. 1998; UNAIDS 1998b; Nunn et al. 1997). It is also possible that, in the future, increased availability of life-preserving drugs will change the experience of this disease in poorer countries, although there remain a variety of obstacles to this. Additionally, studies of populations in developed countries rely on the contrast between the less healthy and the average, healthy adulthood that is the norm in richer countries. Morbidity data, reflecting the prevalence of debilitating but not life-threatening disease or malnutrition, which is the norm for many in developing countries (Hull and Jones 1986), might reduce this sense of contrast between states of health and illness.

Finally, the assumption that degenerative disease creates longer dying trajectories might be questioned if comparative data were available. This may be obviously true if the point of comparison is with cholera, pneumonia, or trauma, but tuberculosis and AIDS both cause considerable long-term debilitation, dependency, and symptoms, even in an environment that precipitates an earlier death from these causes in some countries. For example, Gilks et al. (1998) report a World Bank estimate that, on average, a Tanzanian adult with AIDS has 17 episodes of illness requiring more than 280 days of care; they point out the particular strain that such chronic illness places on poor families when the sufferer may be a parent with dependent children or other dependent relatives.

Care of Dying People

Households and Family Structure

On the whole, sick people turn first to their families for help, so it is important to know about any factors that change the availability of such informal care. In developed countries, the aging of populations is generally accompanied by decreasing family size and a growing propensity of elderly people to live in households separate from their children. Because of gender differences in longevity and marriage patterns, this commonly leaves many elderly widows living alone toward the end of life, dependent on nonresident caregivers for assistance if they get sick. A notable exception to the trend toward living alone in old age is Japan, where, in spite of the high proportion of elderly people in the population, the proportion of elders who live alone is small. In the 1980s, for example, only 10% of Japanese age 65 and older lived alone in private households, compared with 30% in the United Kingdom, 31% in the United States, and 40% in Sweden (U.S. Congress 1993). Cultural preferences regarding appropriate family relationships lie behind the Japanese pattern and may well apply also in many less developed countries.

On the whole, data on the household structures of elderly people are unavailable for developing countries, but data on the extent of single-person households, a rough proxy indicator of the proportion of elderly living alone, are available for some. In Bangladesh, the Philippines, and Thailand, for example, single-person households are rare (generally less than 3% of all households) and their proportion has not increased over time (Young 1986). Remember, however, that in developing countries death is less confined to the older years, so a profile of elderly households is relevant to a smaller proportion of those needing care when dying. In developing countries, concerns about family care for the dying may be offset by concerns about care for the dependents of people who have died, something that has become less relevant in developed countries because of the demographic factors outlined, but also because of relatively sophisticated systems of social insurance that are unaffordable in developing countries. With the pattern of mortality from AIDS in Africa, this has been a particularly pressing concern; as the number of orphans increases, these children are experiencing various forms of social and educational deprivation and abuse as a result of the lack of adults to care for them (UNAIDS 1998a, 2002).

U.K. data on elderly households show trends over time that follow a pattern similar to that seen in the few other countries where such data are available (for example, Australia, Canada, Japan, and the United States; see Young 1986). In 1996-97, 15% of households in the United Kingdom consisted of a single person above pensionable age living alone; in 1961, this figure was only 7% (Office for National Statistics 1998b). At the more recent date, approximately four times as many elderly women as elderly men lived alone. The 1996 GHS found that 87% of people age 65 and over were living either alone or with only a spouse (Office for National Statistics 1998a); this proportion has also been steadily increasing over time (Grundy 1996). Grundy (1996) also reports surveys showing reductions since 1962 in the proportion of elderly parents with at least one child living within 10-15 minutes’ travel. These changes have had significant consequences in terms of the sources of informal help and care that people can draw upon as they approach death. A 1987 survey of the last year of life found that people living alone were in a particularly unfortunate situation concerning potential sources of help. They were the least likely to have any children or siblings alive and were most likely to be widowed or divorced and old; they were also the group most likely to progress to institutional care (Seale and Cartwright 1994).

Institutional Care and Place of Death

The widespread trend toward institutional care for elderly people is very much a phenomenon of developed countries, adopted as a solution to the shortage of informal care available in families. This shortage of available family care has been caused both by demographic factors and by features of the social organization of advanced industrialized societies that often separate elderly people from mainstream social and family life. The Japanese example suggests that demographic pressures alone are inadequate to explain the growth of institutional care, for in Japan relatively low proportions of the elderly are in institutions in spite of the country’s large numbers of elderly people. In the early 1980s, for example, 4% of the Japanese population age 65 and over lived in institutions, compared with 6% in the United States and 11% in the Netherlands (U.S. Congress 1993). Data from Australia, Canada, the United Kingdom, and the United States show that the elderly living in institutions are predominantly female (Arber and Ginn 1991; Young 1986).

Although most people who enter residential institutions for the elderly will eventually die there, these are not generally perceived as places primarily devoted to the care of “dying” people. That is more normally the perception of hospices, although caring for “the dying” is also seen as a legitimate part of general hospital care. A large proportion of people in developed societies die in hospitals, rather than at home or, indeed, in hospices. Table 4 shows the proportions of deaths that occur in hospitals for selected countries. Data on place of death are available only for countries where statistical surveillance is well developed, which means that countries with high proportions of such deaths tend to keep track of the figures. Nationally representative figures for African countries, for example, are unavailable. Such countries are likely to show marked rural/urban differences in the proportions of hospital deaths, and overall the proportion is likely to be low.

In countries with high rates of deaths occurring in hospitals, the rates have been rising steadily for many years. In 1960, for example, only 50% of people in England and Wales died in hospitals (General Register Office 1962), compared with the 66% shown in Table 4 for the United Kingdom as a whole in 1996. However, there are indications that the trend toward hospital deaths tends to level off once figures of 60-70% are reached (Brameld et al. 1998; Hunt et al. 1991). This is partly because there is a residual core of sudden deaths, but also because deaths in other institutions, such as residential homes and hospices, increase; in addition, there is often an increase in the successful provision of supportive community care for those who wish to die at home, something that in some countries may be determined by the ability to pay for such care (Dudgeon and Kristjanson 1995).

It has become part of the professional ethic of specialists in terminal care that a supported death at home is generally preferable to a death in hospital, and much of the effort of community hospice services is devoted to achieving this outcome. Researchers have noted the effects of community care and specialist hospice provision on place-of-death figures in the United States (Pritchard et al. 1998) and in that country specifically in relation to AIDS deaths, where hospital deaths have declined for whites, gay men, and men in general, although not for injecting drug users or children (Kelly et al. 1993). Studies in Sweden (Andershed and Ternestedt 1997) and Australia (Hunt et al. 1993; Hunt and McCaul 1998) have recorded shifts toward hospice as a place of death in recent years. Research in the United States (Stearns et al. 1996) and Australia (Brameld et al. 1998; Hunt et al. 1993) has found that people in their 80s and above have lower rates of hospital care, or of death in hospital, than do the “younger” elderly. This is largely due to alternative institutional provision in these countries rather than a greater proportion of home deaths. A pattern of frequent, short, nonterminal stays in the hospital in the last year of life may also develop in such countries (Brameld et al. 1998).

In a review article, Grande, Addington-Hall, and Todd (1998) note many of these features and make the additional point that gender makes a difference in all of this, given that, for the reasons outlined earlier, women in these countries tend to be disadvantaged in their ability to draw on informal family care as they approach death and are therefore less likely to die at home. Significantly, Grande et al.’s review covers 12 U.S. papers, 14 Australian, 10 Italian, 3 Swedish, and 1 each from Switzerland, Israel, and Canada. Place of death is not a topic that has been studied systematically outside developed countries.

In several Anglophone countries, but particularly the United Kingdom and the United States, a critique of the quality of care for the terminally ill arose in the 1950s and 1960s, fueled by a general readiness to question scientific and professional authority and by widespread concerns about rights to individual autonomy in the face of institutional power (reflected also in critiques of institutions for the mentally ill, for example). The institutionalization and apparent medicalization of care for the dying were criticized for their dehumanizing emphasis on curative efforts at the expense of palliative care, and place-of-death statistics became a symbol around which these dissatisfactions coalesced.

Health Care Systems

Care specifically aimed at dying people and their families in developed countries has been marked in recent decades by the rise of the hospice movement and, most recently, the development of palliative care as a medical and nursing specialty (for details on the spread and character of this movement, see Seale 1998; Clark, Hockley, and Ahmedzai 1997). In addition to imparting a new vigor to strictly medical efforts to palliate the symptoms of terminal illnesses (chiefly cancer, but also motor neurone disease and AIDS), this movement has stimulated a concern with the psychosocial well-being of patients and their families that has led to an extension of traditional medical expertise. Nurses, by successfully claiming particular psychosocial skills, have gained a significant degree of autonomy from medical dominance in this arena of health care, assisted also by the formation of community support teams to advise lay caregivers in their homes. More recently, some have asserted that the palliative care approach is relevant to preterminal phases of terminal disease (Doyle 1996) or to additional diseases such as stroke, heart disease, and dementia, or could be applied in the context of nursing home care (Field and Addington-Hall 1999; Clark et al. 1997). To express such ambitions, however, is not to underestimate the palliative components that already exist in the health care specialties that serve these groups.

The appeal of hospice and palliative care in developed countries can be understood within a broader cultural context. The tendency to plan for and control major life events is already an important feature of self-identity in late-modern societies (Giddens 1991), although it is particularly concentrated among more educated groups and may be stronger in Anglophone countries than elsewhere (Seale 1998). Life planning, saving, taking out insurance against sickness as well as death (Benjamin 1986), and investing in schooling or training can all be engaged in with more confidence if mortality is predictably placed at the end of the life course (Chernichovsky 1986). Gradually, dying too becomes subject to this wish for control, and hospice and palliative care practitioners provide a relevant expertise to assist this. Comparison of cancer deaths in hospitals and hospices has revealed the relative success of hospice patients in planning the manner of their deaths, with hospitals showing higher proportions of deaths resulting from unplanned, emergency admissions (Seale and Kelly 1997). The shift in professional attitudes and practices toward open disclosure of prognosis that has occurred in developed Anglophone countries (Novack et al. 1979; Seale and Cartwright 1994) has served to promote patients’ desire for control while additionally opening up a new arena for psychosocial expertise (Kübler-Ross 1969).

The rising support in many developed countries for euthanasia, as an alternative means to relieve suffering and dependency toward the end of life and to control the manner and timing of death, is a reflection of similar forces (for a review of the literature on this subject, see Seale 1997). Although the hospice and euthanasia movements clash at the level of public debate because of religious differences, for individual patients they offer very similar opportunities for control and self-direction near death (Seale, Addington-Hall, and McCarthy 1997). Significantly, calls for medically assisted euthanasia are not heard in developing countries, where the suffering of dying people is of a different nature, but nevertheless considerable and in some cases occurs over lengthy periods (as noted earlier). The desire for medical assistance in this may also, paradoxically, reflect a general dependency on medically delivered solutions to suffering, which are not shared in developing countries. Justice (1995), for example, describes a culturally sanctioned method of fasting to death in Banares, India, where there is no medical involvement.

There is evidence to suggest that the largely Anglophone phenomena of hospice or palliative care and support for euthanasia are somewhat alien to the cultures of some developed countries. The cases of Japan and, to a lesser extent, Italy are relevant here. Important underlying factors seem to include respect for the traditional authority of the medical profession coupled with relatively strong religious observance, as well as a lesser emphasis on individual autonomy and greater reliance on intrafamilial support during disruptive life events. Studies suggest that the Anglo-American practices of informing most patients with cancer of their disease, of stressing the benefits and opportunities of open awareness of dying, and of involving patients rather than their families in decision making would be culturally inappropriate in countries such as Japan and Italy (Long and Long 1982; Kai et al. 1993; Surbone 1992; Gordon 1990). Japanese prohibitions against organ transplantation (Lock 1995) arise from religious considerations that may also explain the absence of a significant movement to support euthanasia. Nevertheless, Anglophone models of palliative care are increasingly spreading to European and other non-Anglophone developed countries. This is particularly marked in Eastern Europe, where the changes of political climate have led to a host of initiatives to relieve terminal suffering (Luczak 1997).

Where resources support the provision of specialist services for dying people, then, awareness of cultural differences can help practitioners to determine the appropriate form of such services. In developing countries there is the additional factor of scarce resources to consider. In practice, most debate about health care in developing countries is confined to concerns with the appropriate balance between prevention and cure, or between hospital and community services (Okolski 1986; Northrup 1986; Hull and Jones 1986); issues of palliative and terminal care, which might be considered luxuries affordable by richer countries, are rarely addressed. People in developing countries may see the involvement of religious authorities in issues of dying as more appropriate than the involvement of health services. Nevertheless, examples do exist. Western-style hospice care has developed in certain cities in India, for example, although access to such care tends to be confined to wealthier clients, and such initiatives have had to struggle against entrenched professional attitudes and working practices that are reminiscent of the situation 50 years ago in U.K. and U.S. health care (Burn 1997). The 11 cancer centers in India reach less than 10% of terminally ill cancer patients, and only 16,000 of the estimated 350,000 people with cancer pain are treated for their pain each year (Stjernsward 1997). Medical initiatives coexist with models of palliative care for the poor that draw more on religious traditions than on medicine, as seen in Mother Teresa’s work.

Community palliative care initiatives in developing countries are likely to be funded at low levels and may focus on relieving the social care needs that arise from extreme poverty exacerbated by illness rather than on relieving medical conditions. This has been the experience of community initiatives in relation to AIDS care in some African countries (WHO 1994) and particularly Uganda (Gilks et al. 1998), which have attended to medical and nursing aspects of terminal AIDS care only insofar as this has been affordable. In relation to palliative cancer care, sub-Saharan Africa “remains isolated from hospice knowledge” according to Hockley (1997), with the exception of South Africa and Zimbabwe, where there are nascent hospice movements. Where palliative care initiatives exist in developing countries, they must deal with a variety of problems, including “an inadequate public health infrastructure, poor administrative systems, the lack of oral morphine and restrictions on opioid prescribing, the general poverty of patients, and poor educational opportunities for health professionals” (Haber 1999:1303).

The World Health Organization’s Cancer Pain and Palliative Care Program represents a pragmatic approach to the difficulties of providing palliative care in developing countries, where a combination of poverty and cultural differences militate against the wholesale application of Western models. By prioritizing pain relief as an essential core component of a global campaign, WHO initiatives are able to concentrate on removing obstacles to the availability of opioids and promoting a low-cost approach to relieving this core distressing symptom of terminal disease (Stjernsward 1997).

Conclusion

A number of issues arise from this review that are important if we are to understand the implications of changing causes of death and life expectancies for the future experience of dying. I have described the aging of populations worldwide and have pointed out the consequences this can have for availability of care for elderly people toward the end of life, in both developed and developing countries. Gender differences in longevity, social status, and living arrangements have implications for the experience of old age and the availability of care toward the end of life, and these differ internationally. The experiences of dying people merge with the more general experience of old age in countries that have experienced the demographic transition. This may have consequences for the relationship between services specializing in the care of dying people and services specializing in the care of elderly people.

Different patterns of life expectancy and disease burden around the world require us, too, to question the extent to which Western models of terminal and palliative care are applicable in developing countries. The care needs and dying trajectories of diseases commonly causing death in developing countries may be rare in other countries and may not be of the sort that are appropriately met through existing models of palliative care. Even if applicable, should such services attract resources in poorer countries with high levels of preventable disease? Considering richer, developed countries, there is evidence that in some there are cultures and communication practices between health care workers and their clients that are at variance with those in which palliative care originally developed.

The sudden, perhaps temporary, declines in life expectancy that have been seen in some Eastern European countries suggest that we should exercise caution in assuming that in the future there will be a general passage of all countries through the demographic and epidemiological transitions experienced by richer countries of the world. In particular, the spread of AIDS and its impact on populations in Africa and some other regions where governments have been slow to implement preventive measures and are too poor to afford drug therapies will continue for some decades to come. This disease has a somewhat unpredictable future trajectory, but it is already clear that it has a very considerable impact on the experience of dying, on the lives of survivors, and on the economies of the countries worst affected.