The epidemiology of the end-of-life experience

The epidemiology of the end-of-life experience

Epidemiology has been defined as ‘the study of the distribution and determinants of disease frequency’^{(1 )}. It is the study of how often, and in which populations, disease occurs. Epidemiology is central to the development of strategies for the prevention and management of disease in populations and for the planning of health services. Epidemiological data can also provide information about the nature of the progression of specific diseases and the effect of treatment options. In this chapter, we will take a broad view of the epidemiology of ‘disease’ and will discuss epidemiology as it relates to the human experience towards the end of life with an emphasis on diseases, symptoms, and psychosocial experiences.

Three major questions relate to the epidemiology of palliative medicine and must be addressed in relation to specific diseases.

• ♦ Which diseases do people die from, in particular when death is not unexpected, sudden, or immediate?

• ♦ What is the incidence and prevalence of symptoms and distress in individuals living with these diseases?

• ♦ What is the trajectory and time course of life threatening illness?

The answers to these questions provide an evidence base for the management of symptoms that are common in specific diseases and care settings, and for the development of services for those people living with these diseases and for their caregivers. This type of information can also provide evidence that can assist in the design of palliative care education programmes which aim to target the professional needs of those engaged in the provision of care towards the end of life.

The palliative care population base

The World Health Organization (WHO) defines palliative care as an ‘approach which improves the quality of life of patients and their families facing life-threatening illness, through the prevention and relief of suffering by means of early identification and impeccable assessment and treatment of pain and other problems, physical, psychological, and spiritual’^{(2 )}. Definition of the palliative care population is important as it helps to articulate what palliative care is, who needs it, and who should provide it^{(3 )} (see Section 2). The type of palliative care services in demand and the needs for particular clinical skills and knowledge required by a patient living with a life-threatening disease are largely affected by the type of disease as well as the socioeconomic, cultural, home, and natural environments that the patient inhabits.

Incidence

‘Incidence quantifies the number of new events or cases of disease that develop in a population of individuals at risk during a specified time interval’.^{(1 )} This is an estimate of the probability that a patient in a given population at risk, will develop the condition during a specified period of time. For example, the incidence of bowel obstruction in a population of patients with colorectal cancer followed for 30 months from the date of diagnosis. Incidence can be summarized as: $$\frac{\text{Number}\text{of}\text{new}\text{cases}}{\text{Total}\text{population}\text{at}\text{rist}}$$

Prevalence

‘Prevalence quantifies the proportion of individuals in a population who have the disease at a specific instant and provides an estimate of the probability (risk) that an individual will be ill at a point in time’.^{(1 )} For example, the point prevalence of patients meeting DSM-IV criteria for depression at admission to a palliative care unit. Prevalence can be summarized as: $$\frac{\text{Number}\text{of}\text{cases}}{\text{Total}\text{population}\text{at}\text{a}\text{given}\text{point}\text{in}\text{time}}$$

Limitations of mortality statistics

In reviewing mortality statistics, it is important to have an understanding of the limits of the existing data. Mortality statistics provide information on death rates and causes of death in populations. To obtain this data, epidemiologists must rely on diverse sources of information. The Global Burden of Disease (GBD) study initiated by the WHO provides a comprehensive set of mortality and morbidity statistics by age, sex, and region. The GBD study is based on four sources of information^{(4 )}:

1. 1 Death registration systems. These provide information, not always complete^{(5 )}, on the causes of death for most high-income countries as well as many countries in Eastern Europe, Central Asia, Latin America, and the Caribbean.

2. 2 Sample death registration systems. These systems are used in China and India where the deaths of a large proportion of the population are not registered, in particular deaths in rural areas. To supplement death registration systems, a sample of the population in rural areas is registered and the death rate established. This rate is then extrapolated to the broader population.

3. 3 Epidemiological assessments. These assessments provide estimates of deaths for major diseases, such as HIV/AIDS, malaria, and tuberculosis (TB), for countries in the regions most affected by these conditions. Epidemiological assessments deduce case fatality rates (i.e. people who have a specified disease who die as a result of that disease within a given period of time) from surveys on the incidence or prevalence of a specific disease over a specific period of time combined with knowledge of the usual mortality for that condition.

4. 4 Cause of death models. These are used in regions (including most of sub-Saharan Africa) with non-existent or incomplete mortality data to estimate deaths according to broad cause groups.

Only a third of the world’s population resides in regions where complete civil registration systems that provide adequate, cause- specific mortality data exist. In most of Africa, South East Asia, the Middle East, and parts of the Pacific, where over one-quarter of the world population resides, there is little or no mortality monitoring^{( 4 , 6 )}.

In countries such as China and India, which rely on a relatively small sample of the population to determine national mortality statistics (called sample vital registration systems), death data may be unrepresentative. In China for example, death registration systems cover less than 10 per cent of the population. Furthermore, deaths are under-reported, even in regions with such registration systems in place. An assessment of China’s mortality reporting system suggested that in regions covered by that system, adult deaths were under reported by 30 per cent^{(6 )}.

Reporting errors relating to cause of death are a worldwide problem. The uncertainty for ‘all cause’ mortality is reported to range from ±1 per cent in high-income countries to around ±20 per cent for sub-Saharan Africa. Uncertainty intervals are even greater in estimates of mortality by specific cause. For example, the uncertainty intervals for deaths from ischaemic heart disease were estimated to range from ±12 per cent for high-income countries to ±30 per cent for sub-Saharan Africa^{(7 )}. Even in countries where deaths are reasonably consistently reported, for example, in high-income countries, significant proportions of reported deaths have been evaluated as containing errors ranging from ‘unacceptable’ or ‘inaccurate statements’ through to ‘major errors’^{(5 )}. Reporting errors are more frequent in those regions where deaths occur without the involvement of medical practitioners. In these regions, mostly low- and middle-income countries, mortality data collection relies on family members for information on the cause of death^{(6 )}.

It has also been suggested that regions with sub-optimal mortality monitoring systems are at added risk of reporting bias^{(8 )}. One example is where death rate estimates are drawn from data provided by ‘groups in competition for scarce resources that are acting as advocates for affected populations’^{(7 )} or groups that are responsible for both—for data collection and reporting to funding bodies^{(7 )}. Potential conflicts of interest, financial incentives, and pressures must therefore also be considered when assessing the validity of epidemiological data^{(9 )}. Along with the issues stated earlier, there is the problem that mortality data is reported by single cause of death (for reasons of complexity) despite the fact that co-morbidities and health risks may significantly contribute to cause of death. This may introduce further biases.

The comprehensive nature of the death and morbidity information captured is also a product of the coding and reporting systems in use. The electronic recording of death data and the increasing adoption of reporting systems such as the International Classification of Diseases version 10 (ICD 10) by the majority of countries (from four countries in 1994 to 75 countries in 2003) have brought about improvements in ‘real-time’ availability and interpretation of death data^{(7 )}. There are, however, limitations with regard to the comprehensiveness of the information gathered.

In many instances, the principal causes of death included in mortality reports exist within a spectrum of conditions that vary in severity—from conditions where disease-modifying therapy can provide a cure or substantially improved survival–to those with a universally poor prognosis^{(9 )}.

TB, one of the leading causes of death globally provides an example of the complexity that occurs where multiple co-morbidities contribute to a death. This disease can present in various entities including highly morbid stains such as, multidrug-resistant tuberculosis (MDR-TB). TB may also present in a variety of clinical contexts (e.g. with no co-morbidities vs. with significant co-morbidities e.g. HIV/AIDS). The documented mortality rates due to MDR-TB in studies from high-income nations range between 50 to 100 per cent in HIV infected men to between 31 to 44 per cent in HIV-negative men^{(10 )}. Most global mortality data do not therefore report type and clinical context of TB contributing to the overall TB death rate. Disease-reporting systems such as the ICD 10 provide the opportunity to overcome these kind of limitations.

In summary, when interpreting mortality data, it is important to be aware of its limitations. Despite these limitations, however, the increasing availability of population data from different sources and regions provides the opportunity to reduce sampling biases and to gain a more complete representation of the causes and experiences of death in different parts of the world. Over time, developments in information systems will provide the potential for further improvements in the comprehensiveness and accuracy of mortality statistics.

Life expectancy

Life expectancy and cause of death vary greatly worldwide and may be associated with demographic characteristics and other factors such as socioeconomic status^{( 11 )( 12 )}. Differences in life expectancy are greatest when the lowest income countries are compared with the highest income countries (Fig. 3.2.1 and Fig. 3.2.2). Factors such as occupational, political, cultural, and lifestyle risks as well as ethnicity, gender, and genetics also influence these data. For example, worldwide females have a predicted additional 5 years in life expectancy, in comparison to males (70 years in women compared to 65 in men, for the period 2005 to 2010)^{(13 )}.

• View full-sized figure
• Download figure as PowerPoint slide

Figure 3.2.1

Countries of the world grouped by Gross National Income (GNI), 2003.

Source: reprinted with permission from the World Bank Group, 2009 World Bank Atlas: Income Per Person. http://go.worldbank.org/7ETAD6CKO0 ^{(11 )}.

• View full-sized figure
• Download figure as PowerPoint slide

Figure 3.2.2

Population life expectancy worldwide, 2006.

Source: United Nations, World Population Prospects, 2007. Reprinted with permission from Institute National d’Etudes Demographique. http://www.ined.fr/en/teaching kits/length of life death mortality/world life expectancy/ ^{(12 )}.

Leading causes of death

The leading reported causes of death vary between regions at different levels of economic development. The World Bank classifies the countries of the world into three income groups: low, middle, and high. In 2002, an estimated 57 million people died in the world and, of these deaths, 85 per cent occurred in low- and middle-income countries^{(14 )}. Table 3.2.1 presents the 10 leading causes of death worldwide. Among these, three major categories emerge: cardiovascular diseases (ischaemic heart and cerebrovascular diseases are the most common in this category), infectious and parasitic diseases (lower respiratory tract infections, AIDS, diarhoeal diseases, TB, and malaria are the most common), and cancers (cancers of trachea/lung are the most common cause of cancer death)^{(14 )}. Table 3.2.2 describes the causes of death in relation to the income group of countries.

In high-income countries the leading 10 causes of death by disease group are cardiovascular diseases, followed by all cancers (lung, colorectal, breast, and stomach are included in the top 10) and other chronic diseases which include chronic obstructive pulmonary disease (COPD), dementias and diabetes (Table 3.2.2). Death rates from most infectious and parasitic diseases in high-income countries are low most (<5 per cent) and except for lower respiratory infections, these diseases do not feature as leading causes of death in this income group.

The leading cause of death in high-income countries varies between demographic groups (socioeconomic class, gender, age, and ethnicity). For example, more males than females die of cancer and injuries and more females die from cardiovascular conditions and other chronic diseases such as dementias and diabetes (Table 3.2.3).

There are key differences observed between the different age groups. Among people aged 15 to 59 from high-income countries, cancers are the leading cause of death, (32.3 per cent)^{(7 )}. Injuries (21.3 per cent) and cardiovascular diseases (21.6 per cent) rank second and third in this age group^{(7 )}. In the 60-and-over age group, cardiovascular diseases (42.1 per cent) are by far the most common cause of death followed by cancers (25.3 per cent)^{(7 )}. Deaths due to injuries (2.9 per cent) are significantly lower in this group, compared to the younger age groups. Deaths from lower respiratory infection (5.0 per cent) and chronic diseases such as COPD (4.3 per cent), and dementias including Alzheimer’s (3.7 per cent) are all higher in people aged 60 and over.

In children (0–14 age group) from high-income countries, perinatal conditions and congenital anomalies are the leading causes of death (Table 3.2.3). Perinatal conditions include diseases related to low birth-weight and prematurity (31 per cent of perinatal deaths) as well as birth asphyxia and trauma (34 per cent of perinatal deaths). Injury is the third leading cause of death among children under 14 in this income group (17 per cent of all deaths). The most common injuries result from road traffic accidents, which accounted for 5.9 per cent of deaths in this income group in 2001.

Marked differences in health status, life expectancy, and cause of death also exist between ethnic groups in some high-income countries; in particular between indigenous and non-indigenous people. For example, data from Australia for the period 2005–2007, shows that male indigenous Australians had a life expectancy at birth 11.5 years below that of the non-indigenous Australian population; for female indigenous Australians, the difference was 9.7 years^{(15 )}. Indigenous groups from other high-income countries also have comparatively higher rates of death and morbidity from cancer, respiratory disease, stroke, injury, and diabetes^{( 16 , 17 )}.

The leading causes of death in middle-income countries are cardiovascular diseases (37.0 per cent) followed by cancers (16.0 per cent), and chronic diseases (>8 per cent; in particular COPD). However, in contrast to high-income countries, HIV/AIDS (3.0 per cent) and perinatal conditions (2.9 per cent) feature among the leading causes of death (Table 3.2.2).

Populations from low-income countries have not experienced the rising life expectancy observed in the rest of the world, and in several countries of sub-Saharan Africa, life expectancy has declined to 40 years or below^{(18 )}. The leading causes of death among populations from low-income countries are infectious and parasitic diseases (>45 per cent; lower respiratory infections, HIV/AIDS, diarrhoeal diseases, malaria, and TB are included in the top 10). Infections and parasitic disease account for about half of all causes of deaths, therefore constitute a significant proportion of the total deaths each year for the whole world. See Table 3.2.2 for more detail about causes of death in low-income countries.

In both low- and middle-income countries there is little variation in the leading causes of death between males and females, with the exception of all types of injury and TB both of which are higher among males (Table 3.2.4). Amongst children (0–14 years) in low- and middle-income countries, infectious and parasitic disease followed by perinatal conditions are the most common causes of death^{(7 )}. With regard to infectious causes of death in these children, lower respiratory infections (17 per cent), diarrhoeal diseases (13.4 per cent), malaria (9.2 per cent), measles (6.2 per cent), AIDS (3.7 per cent), whooping cough (2.5 per cent) and tetanus (1.9 per cent) are the most common infections accounting for these deaths (Table 3.2.4).

The median neonatal mortality rate (NMR) in low- and middle-income countries combined is 33 per 1000 live births accounting for almost 4 million deaths. This is several times higher than the NMR in high-income countries (4 per 1000 live births accounting for 42 000 deaths). The highest NMRs are in countries of sub-Saharan Africa (in some the NMR rate is up to 65 deaths per 1000 live births)^{(19 )}. In these regions, death statistics for people aged 60 and over are not dissimilar to those in high-income countries. Death in those people surviving to the sixth decade in low- and middle-income countries is predominantly due to cardiac and vascular diseases (48.7 per cent), followed by cancer (14.5 per cent, the leading tumour types are gastro-oesophageal and lung)^{(7 )}. In this income group, death from chronic diseases (especially COPD and diabetes) is significantly more frequent and death from injuries are less frequent in the over 60-year-olds than among younger age groups.

Projections for the future: leading causes of death

Mortality projections represent potential future occurrences. For these projections current data are used to estimate future events to provide a ‘forecast’. Projections do however provide indicators useful in the planning of health services at both the global and local levels^{(20 )}. These projections also furnish palliative care providers with some insight into future palliative care needs of specific populations.

Worldwide projections for 2030, based on 2002 data, predict an overall increase in life expectancy (from 65 to 70) and a decreasing rate of infant deaths (Table 3.2.5). In some regions, however, life expectancy will continue to remain far behind the rest of the world. This is particularly the case for sub-Saharan Africa (remaining below 55 in 2030) largely as a consequence of AIDS, war, and poverty^{(21 )}. The rate of global death from communicable disease is expected to decrease from 41 per cent in 2002 to 31 per cent; however, deaths due to AIDS are expected, by contrast, to continue to rise.

Other causes of death that are expected to increase further by 2030 worldwide are: circulatory diseases, COPD, diabetes mellitus, and lung, stomach, liver, and colorectal cancers^{(22 )}. It is expected that potentially-modifiable lifestyle risks such as smoking, will continue to be a major contributor to premature deaths globally^{(20 )}. By 2015, deaths attributed to smoking will account for 10 per cent of all global deaths (6.4 million)^{(22 )}. This projected rise in smoking related death assumes a predicted increase in smoking rates in low- and middle-income countries as well as in younger women across the world^{(22 )}.

For high-income countries, mortality projections for 2030 predict that the cardiovascular diseases will be the most common cause of death (Table 3.2.5). Cancers (with lung cancer expected to remain the tumour type responsible for the largest proportion of deaths) and chronic diseases including COPD, diabetes mellitus and dementias are the predicted leading causes of death in this income group^{(22 )}.

In middle-income countries, cardiovascular disease is predicted to continue as the leading cause of death into 2030. Death from chronic non-cancer and non-cardiovascular disease is also expected to increase; AIDS, COPD, and diabetes are amongst the causes of death expected to rise in this income group. Death from AIDS is expected to increase dramatically to become the fourth most common cause of death in middle-income countries. Deaths from cancers, particularly cancer of the lung, stomach, and liver, are also expected to increase^{(22 )}. Population increases in regions with high prevalence of diseases such as gastric cancer, and the growing prevalence of lifestyle risks such as smoking, may contribute to an increase in deaths from these cancers.

In low-income countries, infectious and parasitic diseases (HIV, lower respiratory infections, diarrhoeal diseases, and malaria) are expected to continue to be among the leading causes of mortality in 2030. Deaths from AIDS, estimated to increase by 76 per cent are expected to contribute significantly to this increase^{(22 )}. Cardiac, vascular and other chronic diseases are also projected to be among the leading causes of death in this income group^{(22 )}. The cardiovascular group of diseases is expected to continue as the most common cause of death. Of all cardiac and vascular diseases in this income group, ischaemic heart disease is projected to be the leading cause of death.^{(22 )} Diabetes is expected to appear for the first time among the top 10 causes of death in low-income countries, while deaths from COPD also are expected to increase (by 77 per cent) (Table 3.2.5). By 2030 TB (ranked eighth in low-income countries in 2002) is expected to drop out of the top 10 causes of death in this income group.

Projections like these are useful for the purpose of long-term planning for nationally-based palliative care services. It should be acknowledged that such projections are based on predictive models. Although acknowledged important variables that impact mortality are factored into these models, there is nonetheless uncertainity related to this process. That is; projections of future mortality may not match actual future mortality if, for example important changes in lifestyle, economic, and social factors evolve in a way that has not been adequately accounted for. Three such factors that are currently included in mortality models are; the relationship between economic and social development, known disease patterns, and tobacco and obesity risks. These may change in unpredictable ways. Similarly, unexpected health improvements related to treatment and disease prevention strategies, may also affect future mortality trends.

Place of care: where are palliative care services and support needed?

One aim of palliative care is to provide services, where possible, in a location of patient choice.^{(23 )} Preferred place of death (often the wish to die at home) has been an item solicited in clinical settings and collected in research and clinical audit. Preferred ‘place of care’ may however be a more appropriate focus of enquiry.^{( 25 , 26 )} Increasingly, preferred place of care is being studied and there is an acknowledgement that this preference may change along the illness experience^{(27 )}. Preferred ‘place of care’ is expected to vary over an individual’s disease trajectory and should therefore be ascertained longitudinally over the course of the illness experience. Data about changes in preferred place of care over time help inform planning and support flexibility in service provision up until and including death^{(24 )}.

Data collected on ‘place of death’ are for the most part, from only high-income nations. In general, hospital care is scarce in low-income countries, which would suggest that the vast majority of people in these regions die outside the hospital setting^{(28 )}. The opposite is true in some high-income countries. The available data suggest that more than 50 per cent of deaths in England, the United States, Germany, Switzerland, and France take place in the hospital^{(23 )}. Significant variations in place of death exist among high-income countries, with lower rates of hospital death reported in the Netherlands (35 per cent), Ireland (30 per cent), and Italy (35 percent)^{(23 )}. When asked to project a location of choice, the overwhelming majority of patients indicate a preference for care at home up until, and including, the time of death^{( 29 – 31 )}.

Reasons why many people die in the hospital despite a projected desire to be cared for at home until death, and why preferred place of death may vary over time, are complex and potentially interdependent (see Fig. 3.2.3)^{( 32 , 33 )}. With disease progression, priorities often change and care planning at the end of life must accommodate these changes^{(34 )}. Also, as diseases progress, there are differences between patients and their carers regarding the experience of an illness and the expectations of care^{( 35 , 36 )}. Generally, palliative care services attempt to support care in the location of choice and provide flexibility to accommodate changes overtime. Still, for many patients at the end of life ‘care at home’ is not a practical option and admission to inpatient care ought not to be viewed as ‘failure’^{(37 )}.

• View full-sized figure
• Download figure as PowerPoint slide

Figure 3.2.3

Factors influencing place of death.

Source: redrawn from Higginson IJ, Costantini M (2008). Dying with cancer, living well with advanced cancer. The European Journal of Cancer, 44(10), 1414–24, with permission from Elsevier^{(32 )}.

Gomes and Higginson carried out a systematic review of the risk factors influencing place of death^{(38 )}. The authors report data drawn from 58 original studies from 13 countries, totalling over 1.5 million patients. In this population more than 80 per cent had a cancer diagnosis. Factors found to be strongly associated with patients dying at home included; poor functional status, extended family support, patient preference, living with relatives, and availability and intensity of home care. Factors associated with dying in the hospital included: previous admission to hospital, suffering from non-solid tumours such as leukaemia and lymphoma, residing in an area with available hospital beds, and being from an ethnic minority (Fig. 3.2.3). Others have identified the availability of hospital care as a principal determinant of place of death^{( 39 , 40 )}. It has also been suggested that the capacity for inpatient palliative care services to provide care at short notice, often without the need to attend an emergency department, may allow patients to stay at home longer and facilitate shorter duration hospitalizations. Further research may shed more light on this matter.

The few studies that provide projections of where people from high-income populations are likely to die, suggest the rate of death at home will decline further over time. One study, using data collected by the British Office of National Statistics, found that the proportion of deaths at home in England and Wales had fallen from 31.1 per cent in 1974 to 18.1 per cent in 2003. Epidemiological models predict that deaths at home in the same region, will fall further, to 9.6 per cent by 2030^{(27 )}. These predictions should be considered in context of the more immediate factors that influence place of death including patients’ wishes and the availability of home and caregiver support.

The availability of, and access to, palliative care expertise: are palliative care experts available if needed?

Appropriate assessment and symptom management is indicated for all dying people, regardless of setting^{( 41 , 42 )}. Competencies in symptom management at the end of life and knowledge about the referral process to appropriate palliative care services are required to some degree from all generalist and most specialist health practitioners. A core aspect of end-of-life care involves facilitating a patient’s transition across settings (home, hospital or inpatient setting) and across levels of specialist care according to the need of individual patients^{(43 )}. In addition to providing consultation or direct care for patients at the end of life, specialist palliative care professionals can play a significant role in terms of education, policy, and planning for generalist and non-palliative care specialist health professionals^{(44 )}.

Palliative care consultation and referral is available in most high-income countries through mainstream health services in community, inpatient, and acute hospital settings but availability is much more limited in middle- and low-income countries (Fig. 3.2.4)^{(45 )}. In high-income countries, many tertiary-referral centres accommodate integrated consultative, specialist palliative care services within acute and sub-acute settings^{( 46 – 50 )}. Despite the availability of these services, a significant proportion of patients with far advanced disease in these countries, still have limits on their access to symptom management and end-of-life care^{(51 )}.

• View full-sized figure
• Download figure as PowerPoint slide

Figure 3.2.4

Levels of palliative care development: all countries.

Source: reprinted from the International Observatory on End of Life Care. Wright, M., Wood, J., Lynch, T., et al. Mapping levels of palliative care development: A global view (2006). Lancaster University. http://www.eolc-observatory.net/global/pdf/world map.pdf ^{(45 )}.

An Italian longitudinal survey of 2000 cancer patients, with follow up until death, found that of those who died who were admitted to hospital, 20 per cent received palliative care support. While among those who died cared for at home, the rate was even lower—14 per cent^{(29 )}. In the United Kingdom, estimates from national and regional data suggest that between 25 per cent and 65 per cent of those dying from cancer in one year received specialist palliative care, and 15–25 per cent received inpatient hospice care^{(52 )}. A comprehensive population based study from Western Australia, with data from 27 971 deaths, identified that 68 per cent of individuals dying from cancer received specialist palliative care^{(53 )}.

There are epidemiological data to support the claim that patients dying from illnesses other than cancer are less likely than those with cancer to receive specialist palliative care^{( 54 , 55 )}. This disparity is well illustrated in the Western Australian population-based study mentioned earlier. In that study, only 8 per cent of patients dying from selected non-malignant conditions (heart failure, renal failure, COPD, Alzheimer’s disease, liver failure, Parkinson’s disease, motor neuron disease, HIV/AIDS, and Huntington’s disease) were shown to have received specialist palliative care in the 12 months prior to death^{(53 )}. That stated, not receiving specialist palliative care cannot be assumed in every case to represent inadequate care, but where, disparities exist and/or access is not clearly defined as needs based, it does raise the possibility of the presence of unmet needs. Local needs assessments, using defined criteria may allow further comparison and benchmarking in relation to palliative care access^{( 56 , 57 )}.

As shown in Fig. 3.2.4 ^{(45 )}, some middle-income countries (e.g. Argentina, Chile, Cuba, South Africa, and Malaysia), have specialist palliative care provided through mainstream health services^{(45 )}. It has however been suggested that palliative care services in these countries are inequitably distributed, and are mostly limited to individuals from higher-income groups living in urban areas^{(42 )}. It is estimated that only 5–10 per cent of the population of South America in need of palliative care, receive it^{(58 )}. Furthermore, palliative care services in middle-income regions, like their high-income region counterparts, have largely evolved in the context of providing care for people with specific diseases, in particular AIDS and cancer^{(45 )}. In summary, palliative care services in most middle-income countries are available for only some of the patients in need, and are frequently unavailable for the poor and those living in rural and remote regions.

In low-income countries, three main areas of palliative care need have been identified—symptom management, counselling, and financial assistance^{(59 )}. The majority of dying people in low-income countries are cared for at home and in communities by family and/or neighbours. In general terms, there are few palliative care services and those that exist reach only a small proportion of the people in need.

Uganda is a case in point where, despite attempts to integrate palliative care within public-health services, access is still a problem due to the fact that only an estimated 41 per cent of the population has access to any basic health care^{(59 )}. The disparities in palliative care service provision in India illustrate the disparities that exist in most low and many middle-income countries. For example, a study from 2001 reports that 16 states or territories had some hospice or palliative care service provision, including one with up to 90 per cent coverage. Another 19 states or union territories, however, had no specialist or planned palliative care services^{(60 )}. Many palliative care services in low-income countries (e.g. Guatemala) focus on the care of people with specific conditions, especially HIV/AIDS^{(60 )}. Even if these services are taken into account, half of the deaths in low-income countries are due to advanced, progressive, chronic diseases such as cardiovascular disease and cancer, and while these diseases have a known symptom burden amenable to palliative intervention, there is little specialist palliative care available.

Whether a region has a high, middle, or low income, palliative care services are not always available to meet the needs of all dying patients. An emphasis on needs-based, rather than diagnosis or prognosis-based, provision of care has been advocated by many^{( 23 , 61 )}. Accurate and comprehensive population-based data are fundamental to identifying the needs of patients with far advanced disease, so that the available general and specialist palliative care services can be appropriately and equitably managed.

Despite significant advances, the data suggest that unmet health-care needs amenable to palliative care interventions persist. This will be further explored in the following paragraphs, but, for example, The Study to Understand Prognosis and Preferences for the Outcomes and Risks of Treatment (SUPPORT), which followed 9105 adults hospitalized in the United States with at least one of nine life-threatening diagnoses, found that proxies reported inadequate pain control in 50 per cent of conscious patients during the last 3 days of life^{(51 )}. Epidemiological studies of cancer patients from other countries reveal similar rates of inadequate pain control^{( 63 , 64 )} and unmet needs are also evident for patients with non-malignant, life-threatening conditions^{(65 )}.

Symptom data and health-care needs in palliative care: which health-care needs are relevant to palliative care?

Healthcare needs at the end of life, and not only at the end of life, have a direct relationship with symptom experience during that time. Debate about the definition of what constitutes a health-care need is not new, it echoes centuries of Western philosophical discourse which has attempted, in one way or another, to address the question—what life befits a human being? Contemporary liberal philosophers such as Rawls and Sen, draw on the traditions of Aristotle as well as Kant in considering a healthcare need to be; that which is required for a person to fulfil his or her potential with vitality while preserving dignity, justice, and equity.^{( 66 , 67 )}

This philosophical perspective sits well with the principles of biomedical ethics (autonomy, beneficence, non-malfeasance and justice; see Section 5). Importantly this philosophical perspective carries with it not only justification for symptom management at the end of life, but an ethical imperative for it.^{(41 )} Clinical epidemiology, the focus of this chapter, is indispensable to this task, for in order to meet individual and group needs equitably across a population, these needs must first be systematically identified. Epidemiology is a tool that can provide the data to fulfill this task; it can identify the spectrum of needs that exist and assess the efficacy of interventions on a population basis.

The definition of a health-care need in the context of palliative care generally^{( 57 , 68 , 69 )}, and in relation to many specific diseases at the end of life, has been addressed in the peer reviewed literature^{( 70 , 71 )}. Recently, clinicians and researchers have focused on developing and describing appropriate epidemiological processes that can be used to identify, compare, and contrast the palliative care needs of different people and populations.^{( 57 , 73 )}

Barriers to defining population needs and identifying appropriate intervention and management strategies exist. Barriers include the increased focus on curative treatments due to medical advances in the modern era and the increasing information load associated with contemporary medicine. These factors, among others, have been associated with diminished focus on approaches essential for palliative care assessment and management. This is seen also in the comparatively sparce attention paid to palliative, compared to curative measures in the modern medical literature^{(74 )}. Some quality epidemiological studies, however, have begun to address the symptom experience of patients at the end of life. This type of data can be used to identify needs that may then be addressed through evidence-based management strategies, inform service planning and provide general hypotheses for research. Importantly, these data are also needed to provide adequate answers to questions that come from individuals in clinical settings such as ‘am I likely to have pain that cannot be managed?’ Finally, such data help to establish a health-care system’s success in addressing symptom-related needs.

Methodological issues and limitations of epidemiological symptom data: what does the data mean for those within the palliative care population?

When reviewing and interpreting symptom-related epidemiological data there are a number of important aspects that must be considered. For a detailed overview readers are referred to the review of symptom assessment methodology that is available in Chapter 7.7. A summary of the key points in symptom assessment, as they relate to the epidemiology of symptoms at the end of life follows.

1. Defining the population from which data are obtained is of utmost importance. Care must be exercised when interpreting findings and translating the results of research into clinical practice^{(75 )}. For the most part, extrapolation beyond the source population is best avoided. Heterogeneity in patient characteristics such as primary disease, disease stage, and access to care, may render generalizations about the symptom-experience itself or the factors contributing to the nature of the experience inappropriate^{( 75 , 76 )}. For example, some studies of patients with far-advanced disease have been conducted in the last year of life, others in the last days of life, and in others, ‘time to death’ is not presented. Some symptom reports may reflect pooled data from patients at various disease stages, for example, a cancer report may include patients undergoing adjuvant treatment for cancer, and/or those with early- and late-stage metastatic disease. Lastly, a major problem in the symptom-related literature is that ‘cancer’ without further detail (e.g. lung, colon etc.) has been the unifying ‘diagnosis’ that has identified the subjects in many symptom-related studies. These studies, therefore, include diverse populations and interpreting data in relation to a particular malignancy can therefore present significant problems.

In the setting of non-malignant disease, it is also common to find wide variability in the reporting of data related to stage of illness. The palliative care literature has evolved largely in the absence of standardized data points. Across local and national borders studies, instruments for data collection, reporting standards approaches (subjective, observer-related etc), and populations vary. Although for each symptom a number of studies can be identified, the heterogenous target populations and a poverty of characteristics defining populations, can mean that it is very difficult to compare studies or pool data in meta-analyses.

2. The patient experience is personal and subjective: a patient’s lived experience is, by definition, subjective and cannot be wholly known by another, it is the complex interaction of sense perception, cognitive functions, and the unique domain of each individual. Accurate symptom data can nevertheless be recorded and relies on disclosure and faithful recording of subjective information. A common concern regarding data relating to the experience of individuals at the very end of life is that patients may be unable to report on the symptom experience due to the incapacity of illness (due to, e.g. delirium or unconsciousness) in the last days to hours of life. Other concerns that may arise, especially when chart reviews are the source of data, are that a patient may have withheld, or simply not reported, information about his or her symptom experience.

3. The accuracy of data is dependent on the efficacy of communication between subject and researcher. The factors mentioned earlier can influence this and, clearly, in studies where the medical record is reviewed for the collection of data, the quality of documentation in the clinical record will influence the validity of the report. There are validated tools to facilitate the collection and recording of symptom data directly from patients and also tools to record proxy symptom ratings^{( 77 – 79 )} (see Chapter 7.7). No matter which tool is used within a study, when reviewing the validity of the epidemiological data being reported, the validity of the tool must be determined along with whether a particular tool was chosen with the target population in mind, and whether a report documents a tool’s limitations^{( 79 , 80 )}.

4. The symptom experience changes over time and the burden imposed by a particular symptom may change over time^{( 81 , 82 )}. The isolated nature of prevalence data does not reflect the dynamic changes of the symptom experience over time. Point prevalence data reports on the symptom status at a specific point in time. Period prevalence reports on symptom status during a period of time in the recent past (e.g. ‘right now’, ‘in the last 2 weeks’, or ‘in the last year’). Other theories describe the complex changes in symptom experience over time including, for example, ‘coping’^{( 83 – 85 )}, adaptation^{(86 )}, resilience^{( 32 , 87 )}, and ‘response shift’^{(82 )}. Each of these serve to help shape our understanding of the relationship of epidemiological data to symptom experience^{(82 )}.

5. The symptom experience is multidimensional, inter-relates with the bio-psychosocial spiritual and cultural domains and may have characteristics linked with particular populations. Given the limits of prevalence and incidence data in illuminating the symptom experience towards the end of life, investigations that explore symptom burden are important in the provision of robust insights into the epidemiology of the end-of-life experience. An approach to this on an individual level is found in studies that address symptom burden or distress in addition to symptom prevalence and/or incidence^{( 81 , 88 – 91 )}.

The overall effects of symptom burden may be characterized using tools designed to capture the multidimensional nature of symptoms and the inter-relationship of different aspects of symptom experience^{( 81 , 91 – 93 )}. Through the use of these types of tools, a symptom may be studied in relation to its phenomenology and its impact as well as in relation to domains of function including a variety of family, social, financial, spiritual, and existential issues.

Finally, it has increasingly been recognized that the study of the distress generated by the symptom experience is as important, if not more, than prevalence. Unfortunately, although data in relation to all of these constructs would contribute to a more complete picture of the epidemiology of the end-of-life experience, for many conditions, such epidemiological data are sparse or unavailable.

The characterization of the patient population, together with information about tools used, is very important in allowing for the interpretation of data and for understanding the limits of the generalizability of a study. A good example of key elements of comprehensive characterization is the following:

Inpatients and outpatients with prostate, colon, breast or ovarian cancer were evaluated using the Memorial Symptom Assessment Scale and other measures of psychological condition, performance status, symptom distress and overall quality of life. The mean age of the 243 evaluable patients was 55.5 years (range 23–86 years); over 60% were women and almost two-thirds had metastatic disease. The Karnofsky Performance Status (KPS) score was < or = 80 in 49.8% and 123 were inpatients at the time of assessment.^{(91 )}

6. In considering the overall experience of symptom burden towards the end of life, questions still exist as to which symptoms are the most common and/or most burdensome in the context of particular conditions or within particular health systems, and whether the common symptoms were appropriately assessed in a given study. Certainly, more studies exist for pain, fatigue, and nausea than for other symptoms, but investigations are often limited by loose or varied use of definitions for the symptoms themselves. Approaches to this problem include using validated tools to capture the full array of symptoms experienced^{(94 )}. Meta-analyses have also been used to address the problem of small sample size. However, the heterogeneity of studies included in meta-analyses is also problematic^{(76 )}.

Collaborative^{(56 )}, multi-centre studies with attention to inclusion criteria that carefully define the population reported, can go a long way towards improving case recruitment as well as maximizing the homogeneity of data^{(95 )}. In addition, electronic data linkage has proven to be a powerful tool in palliative care services research in areas such as estimating patient needs, service utilization, cost, and place of care^{( 53 , 96 , 97 )}.

7. Finally, it is crucial to consider symptom-related epidemiological data in the context of the availability of effective symptom management. This is especially important for patients and caregivers. For example, a prevalence figure of 70 per cent for severe pain may reflect true point prevalence but may also reflect the under-treatment of a very manageable condition. Such data may reflect a ‘moment in time’ and not a prolonged experience. Such figures can be alarming for patients and caregivers unless linked with explanations and evidence about the potential for palliative treatments to alleviate distress. Such figures, stated alone, do not serve to clarify that most pain can be treated, nor clarify that, of the 10–20 per cent of patients, who respond poorly to initial pain management, standard multi-disciplinary approaches are available to improve even refractory pain, suffering, and symptom burden^{(98 )}.

The aim of this section has been to illustrate the relationship of empirical data and the symptom experience, and the uses, limitations, and challenges inherent in the interpretation of symptom-based epidemiological data. The following sections review symptoms at the end of life in the light of their incidence and prevalence, severity, frequency, associated distress; and in relation to impact on function and global burden for patients as well as caregivers. The important relationship between symptom burden and outcomes related to patient experience (such as distress) will also be expanded^{(81 )}.

Symptom occurrence by cause of death: what symptoms can be expected over time?

Until recently, symptom prevalence studies in the palliative care setting have focused predominantly on patients with cancer diagnoses. There are now a number of good studies that have explored the prevalence of symptoms in patients with life-threatening and far-advanced chronic lung disease^{( 70 , 89 )} and cardiovascular disease^{( 55 , 61 , 88 , 99 )}. Two recent meta-analyses have also provided excellent overviews of this area of study^{( 55 , 100 )}.

Although generalizing is problematic, the available evidence suggests there is a core group of symptoms experienced across disease states in the last days, and probably the last year of life. In the last year of life, for example, symptoms reported to have high prevalence in cancer include fatigue (including lack of energy and weakness), pain, depression, anxiety, and loss of appetite^{(55 )}. Table 3.2.6 represents the data from a meta-analysis, which included 64 studies across progressive cancer and non-cancer illnesses^{(55 )}. One can see from this table that the ranges for symptoms are very variable, largely due to the heterogeneous populations and varying methodologies that have been used in the studies included in meta-analysis^{(55 )}. In this table, Solano et al. illustrate that a similar spectrum of symptoms has been identified as prevalent in cancer, heart disease (with fatigue, dyspnoea, anxiety, pain, and insomnia, appearing among the top six symptoms), and COPD (with dyspnoea, fatigue, pain, insomnia, and anxiety among the top six symptoms)^{(55 )}.

Another extensive recent meta-analysis studied symptoms in cancer alone and included 46 studies including reports from 26 223 patients. In this meta-analysis, Teunissen et al. analysed separately those studies conducted ‘in the last 1–2 weeks of life’ and ‘other’ studies^{(100 )}. Table 3.2.7 presents the information from this study that summarizes the presence of symptoms in advanced cancer from studies that did not focus solely on the last 1–2 weeks of life. In this investigation, five symptoms (fatigue, pain, lack of energy, weakness, and appetite loss) occurred in more than 50 per cent of the patients in the pooled group of studies that was studied in the period before the last 2 weeks of life^{(100 )}.

It should be noted that, despite these two meta-analyses, studies that focus on symptoms experience within the context of specific diseases in the palliative care setting remain somewhat scarce, and samples sizes are frequently small. Another frequent limitation is the common absence of the identification of the ‘time interval between study and death’, performance status, and disease stage. In addition, a major concern relating to these studies is that ‘cancer’ reflects many diagnoses and more data are needed in relation to specific cancers.

While common symptoms occur across different diseases, the nature of the symptom experience may vary across and among disease states. For example, dyspnoea in lung cancer versus dyspnoea in congestive cardiac failure or dyspnoea in chronic lung disease, may have different levels of intensity and burden as well as different time courses. These conditions also may be accompanied by differing co-morbidities. All of these variables clearly may result in differing lived experiences in relation to specific symptoms. For example, data comparing dyspnoea in chronic lung disease with dyspnoea in lung cancer suggests that, although the number of symptoms experienced in the last year and week of life are similar in the two groups the duration of dyspnoea was longer in patients with chronic lung disease^{(71 )}. Studies such as these help to define symptom experience over time and illustrate the comparative symptom burden in patients with cancer and non-cancer diagnoses.

For all disease states, significant population-based studies of symptoms with excellent methodology can provide examples of the potential for epidemiology to inform service planning and clinical management^{( 53 , 101 – 103 )}. Also, studies that use population databases that may be linked state- or region-wide may help to provide an accurate denominator for prevalence studies and for characterizing population-based needs.

Physical symptoms during the last year of life ‘what will it be like?’

Psychological symptoms and neuropsychiatric disorders during the last year of life

Distress in the setting of advanced illness

Distress, as described in relation to the ‘bothersomeness’ associated with a specific symptom or experience has been discussed earlier. Another global concept of distress exists and has been defined as:

‘a multi-factorial, unpleasant emotional experience of a psychological (cognitive, behavioural, emotional), social, and/or spiritual nature that may interfere with the ability to cope effectively with cancer, its physical symptoms and its treatment. Distress extends along a continuum, ranging from common, normal feelings of vulnerability, sadness, and fears to problems that can become disabling, such as depression, anxiety, panic, social isolation, and existential and spiritual crisis’^{(141 )}.

Attempts to measure this construct have been undertaken by some investigators, and the ‘routine’ use of the ‘distress thermometer’ and an associated checklist to detect distress in cancer patients in the clinical setting has also been encouraged, particularly by the National Comprehensive Cancer Network in the United States^{( 141 , 142 )}. Over time it will be most interesting to see whether epidemiological data about the distress thermometer or checklist further illuminate the advanced cancer experience, particularly in relation to the items on the checklist. At this time it appears that the thermometer scores themselves correlate with psychological distress^{(142 )}. Of note, data related to the distress checklist may be useful for identifying common concerns and needs, rather than for identifying specific diagnoses.

Symptom occurrence in the last days of life: what symptoms can be expected in the very last days of life?

Much has been written in the recent decade about the mandate for impeccable care for all dying patients at the end of life. Recently, health-funding bodies, locally and nationally, have invested in comprehensive programmes to assist generalist and specialist clinicians in caring for patients in the last days of life, regardless of setting or diagnosis^{( 58 , 143 , 144 )} (see Section 19). The problems relating to epidemiologic reports about the last days of life have both similarities and differences to those discussed earlier in relation to reports about advanced, progressive disease. Fatigue/weakness/lack of energy are high on the prevalence list for symptoms in the last days of life with dyspnoea and pain also highly prevalent^{(100 )}.

Table 3.2.8 provides pooled prevalence data reviewed in a comprehensive meta-analysis of cancer-related symptoms in the last 1–2 weeks of life^{(100 )}. Symptoms during the last days or weeks of life have also been captured by other studies. Data for this time, especially the very final days of life, are however limited, and in addition, in these settings, proxy reports become more common in the methodology of studies.

A major study—The SUPPORT study from the United States—reported that in interviews conducted after a patient died, surrogates indicated that 50 per cent of all conscious patients who died in the hospital experienced moderate or severe pain during their last 3 days of life^{(51 )}. Others have reported similar high-prevalence figures for the presence of transient proxy-reported pain and shortness of breath ^{(145 )}. These data exist in the context of other studies which report that the dying process is most commonly peaceful although with the possibility, indeed likelihood, of the presence of some transient, treatable distress. That stated, these studies and others highlight the importance of investigation at this time of life by addressing and defining distress with specific individual and system based assessments.

Reports from hospice programmes and pain studies suggest that most deaths can be peaceful^{( 146 – 148 )}. For example, the late Dame Cicely Saunders recorded data regarding 100 consecutive deaths at St Christopher’s Hospice in the United Kingdom and described 98 out of the 100 patients as dying peacefully. In this population, 60 patients were reported as peaceful for the whole of the last 24 hours and transient distress was experienced by the rest of this group. Unfortunately, the details of the duration of transient distress are unclear from the report^{(146 )}.

More work is needed on this subject but of note, one of the first studies in this area was a study that utilized nurse reports of patient comfort and was conducted by Osler^{( 149 , 150 )}. This survey included 486 deaths occurring between 1900 and 1904 and was not confined to cancer patients. Most patients were described as dying comfortably with ‘no sign of death one way or another, … like their birth, death was a sleep and a forgetting’. That stated, transient distress was reported in approximately one-fifth of patients with ‘bodily pain or distress’ reported in 90 patients, ‘mental apprehension’ in 11, ‘positive terror’ in two and ‘bitter remorse’ in one. Unfortunately, not unlike present-day studies, the study did not describe the palliative interventions used to treat this distress, the impact of those interventions, or the duration of distress.

Further illuminating the experience of patients in the last hours of life are studies that relate to the management of symptom distress and the indications for sedation at the end of life. Several studies have addressed the rate of sedation in the last days of life. Barriers to interpretation of these data exist because of different impressions of what constitutes ‘sedation’ versus the use of medications with some sedative effects for the management of a symptom. Ventafridda and Kohara report palliative sedation was required for 52.5 per cent and 50.3 per cent of patients in Italy and Japan, respectively^{( 151 , 152 )}. Elsayem and Fassinger report rates of 15 per cent and 18 per cent from M.D. Anderson in the United States and Edmonton in Canada, respectively^{( 153 , 154 )}. Notably, the M.D. Anderson study reports 41 per cent of dying patients received ‘sedation’^{(153 )}. The variability of rates of sedation may be related to sample bias or different definitions of what constitutes ‘sedation’. Fainsinger^{(154 )}, reported ‘physician intention to sedate’ and the Kohara^{(152 )} study included ‘mean sedative dose’, both of which can help to characterize the clinical situation. This important area requires more investigation.

Finally, with regard to the last days of life, some symptoms (such as dyspnoea, anorexia, difficulty in swallowing) have been noted to be associated with a poor prognosis^{(155 )}. Readers are referred elsewhere in this text for a discussion of the relevance of symptoms and functional status to prognosis (see Chapter 3.3).

Communication, consciousness, and mental acuity have been rated as highly important by cancer patients at the end of life^{(156 )}. Level of consciousness has been reported in few studies; however, the National Mortality Followback Study, a large epidemiological study in the United States, addressed many aspects of health care including the end-of-life experience and presented reports of consciousness at the end of life ^{(157 )}. In the 1986 investigation, the next of kin of 18 733 deceased patients were surveyed with an impressive 87.3 per cent response. Proxies reported on the deceased persons’ ‘trouble understanding where he or she was during the last year of life’. Over the last year of life, the proxy reports indicated that 15 per cent of decedents had this difficulty for the ‘last few hours or days’, 13 per cent for ‘some of the time’, and 8 per cent for ‘all or most of the time’. The data that pertain to cancer suggest that, in the very last days and hours of life, most people are able to communicate until close to the end of life. A survey of 100 cancer patients who died at St. Christopher’s Hospice in the United Kingdom described 10 per cent as alert, 67 per cent as drowsy or semiconscious, and23 per cent as unarousable or unconscious during the 24 hours before death^{(179 )}.

The prevalence of delirium also can be viewed as a marker for consciousness being ‘at risk’ towards the end of life. Delirium is characterized in part by a disturbed level of consciousness and the prevalence, up to 85 per cent in the end-of-life setting, has been discussed above. Other observations that may be relevant to this aspect of the experience include reports of ‘confusion’ which, in the Teunisen meta-analysis, was found to have a prevalence of 24 per cent in the last 1–2 weeks of life^{(100 )}. In the same sample, ‘drowsiness’, was reported in 38 per cent (see Table 3.2.8)^{(100 )}. Conill reports a prevalence of confusion of 68 per cent among 176 patients in the last week of life^{(158 )}. Delirium is also reported to accompany the last hours of life for patients with non-cancer-related illnesses, but data on the non-cancer illness experience in the very last hours or days of life are patchy.

Cultural experiences and the existential context

Death is highly laden with emotional, social, and cultural significance. There are two aphorism in the English vernacular that serve to illustrate one cultural perspective on this subject ‘Only two things are certain in life: death and taxes’^{(159 )} and ‘there are only two themes in literature sex and death’. A diverse spectrum of beliefs about the spiritual/existential and cultural aspects of death and the period prior to, and after, it exist. Despite the fact that much is published in popular media and the fact that many hold strong beliefs relating to this time of life, and the time before and after death, there exists extremely little in terms of epidemiological data concerning the ‘metaphysical’ domains of experience at the end of life. As is the case within western philosophical debate generally, public debate frequently focuses on whether experiences (for example ‘the near death experience’)can be accounted for by metaphysical explanations alone, whether there are metaphysical explanations for phenomena, and the relationship between brain, mind, and consciousness. Occasionally scientific reports about ‘metaphysical’ aspects of experience at the end of life are published in the peer reviewed literature. When this happens they are frequently reported by the popular media. The near death experience is an example of the fascination which surrounds limited empirical information^{( 160 , 161 )}.

Cultural factors are also important in relation to symptom experience, distress, and communication at the end of life (see Chapter 3.7). More well-designed epidemiological studies looking at culture as a variable in symptom experience are needed.