Epidemiology is the study of disease and its determinants in populations. Epidemiological studies investigate what the patterns of diseases are and why they develop in particular populations. Epidemiology is important in improving our understanding of what causes disease. Similarly, understanding epidemiology and how to interpret observational data is crucial to avoid inaccurate causal inferences. Unfortunately, critically evaluating all of the data on mental health problems in old age was outside the scope of this chapter, but we try to highlight some important limitations and gaps in the literature. There are exciting developments in epidemiology that will guide our knowledge about the determinants of psychiatric disease. For example, genome-wide association studies (GWAS) are increasing our understanding of genetic associations with psychiatric diseases.
Epidemiology is the study of disease and its determinants in populations. Epidemiological studies investigate what the patterns of diseases are and why they develop in particular populations.1 Epidemiology is important in improving our understanding of what causes disease. Similarly, understanding epidemiology and how to interpret observational data is crucial to avoid inaccurate causal inferences. Unfortunately, critically evaluating all of the data on mental health problems in old age was outside the scope of this chapter, but we try to highlight some important limitations and gaps in the literature. There are exciting developments in epidemiology which will guide our knowledge about the determinants of psychiatric disease. For example, genome-wide association studies (GWAS) are increasing our understanding of genetic associations with psychiatric diseases. In Alzheimer’s disease (AD), epidemiology has revealed genetic susceptibilities and identified modifiable environmental risk factors, as well as leading to improved diagnostic criteria through the use of biomarkers (see Chapter 10). It has been estimated that modifying risk factors could ‘delay or prevent a third of dementia cases’,2 which has enormous implications for the information that clinicians give to their patients, as well as for public health policies.
Descriptive epidemiology describes the spread of diseases and their severity in populations. In descriptive epidemiology the disease pattern is established with regard to time, place and person (see Box 1.1). Thus, for example, we might consider the population affected, the characteristics of the disease and how it changes over time. Analytical epidemiology seeks to explain why people get the disease.
|What||= Disease||Descriptive Epidemiology|
|Why||= Cause, risk factors||Analytical Epidemiology|
In analytical epidemiology, the association between the exposure and the outcome is studied to try to establish whether certain exposures are risk factors for the development of disease.
Observational studies allow associations to be derived from observing populations, rather than from performing experiments. This means it is harder to establish whether associations are causal. One of the most significant difficulties in interpreting observational data is determining whether the exposure (A) caused the outcome (B). Without experimental conditions, it is difficult to reach this conclusion since B may have caused A (reverse causation), or C may have caused A and/or B (confounding factors).
Basic measurements in epidemiology include prevalence and incidence rates (see Box 1.2).
Incidence rates ‘express the rate of occurrence of new cases of a disease and may be defined as the frequency of some event related to a disease (e.g. onset of symptoms, hospital admission) related to the size of the population and a specified time period’.
The gold standard experimental design to try to establish causality is a randomized controlled trial; however, this is not always practical or ethical. You cannot randomize people to a treatment thought to increase the risk of disease. Therefore, careful observational studies with appreciation of their limitations are an important source for understanding diseases and their determinants.
There are different types of observational studies which we shall briefly discuss here. The main study designs are descriptive studies of population prevalence, cross-sectional surveys, case–control studies and cohort studies.
Prevalence is the proportion of a population who have a specific disease. The point prevalence is the prevalence of the disease measured at a particular time point. The period prevalence is the prevalence during a time interval. The prevalence of dementia increases significantly with age; therefore, prevalence for this condition is often expressed for age brackets. Prevalence of dementia according to five-year age groups can be reviewed in Table 1.1. Roughly, the prevalence of dementia doubles every five years after 65 years of age.3
|Age||Population prevalence (%) of late-onset dementia|
The prevalence is roughly the same as the incidence multiplied by the disease duration; diseases with short durations, therefore, have similar prevalence and incidence rates. For dementia, a chronic condition, the prevalence is far greater than the incidence. Incidence is not always a useful way to describe mental illnesses. Most mental illnesses, for example, schizophrenia and bipolar affective disorder, have relapsing and remitting courses which can be hard to capture in a single incidence study. In any case, both prevalence and incidence figures depend on the definition of a case and its operational criteria.4
A cross-sectional survey is a type of observational study that measures relevant data on a sample population at a specific time point.1 In a cross-sectional study a sample population is chosen and then a survey is conducted to determine whether the population has the disease or the risk factor of interest. Cross-sectional surveys are quick and cheap to perform. However, they may be prone to selection bias if those responding to the survey are not representative of the general public. They are an inefficient method of determining prevalence of rare diseases and it is hard to establish cause and effect. The sample population that is surveyed must be considered carefully because certain diseases will be over-represented in certain populations. For example, the prevalence of late-onset schizophrenia is higher in inpatient settings and nursing homes than in the community.5
Case–control studies are more useful than cross-sectional surveys in studying rare diseases. Rather than studying a whole representative population, cases are chosen as individuals who have the disease of interest. Controls are then chosen as individuals who are disease-free. The risk factors between the cases and controls are then examined to determine whether a particular risk factor is associated with developing the disease. In this way an odds ratio (OR) can be calculated. An OR determines the odds of the outcome developing in the exposure group, as compared with the odds of the outcome developing in the control group (see Box 1.3).
Relative risk (RR) is the ratio of the probability of developing the disease in the exposure group, compared to the non-exposure group
The odds ratio is the ratio of the odds of the disease in the exposure group compared to the odds of the disease in the control group
In old age psychiatry, case–control studies have been most commonly used to investigate risk factors for the development of dementia. There are disadvantages to this type of study, for example, they are prone to biases. Individuals in the case group may have been more extensively investigated or they may be more likely to recall exposures to risk factors than those in the control group. In addition, reverse causality can be problematic. For example, depression has been identified as a risk factor for dementia; contrariwise, perhaps dementia causes depression; or there may be a common pathophysiological pathway (e.g. cerebrovascular disease or inflammation) to cause both.6, 7 Given these limitations, cohort studies where participants are followed up over time may be used to help ascertain causality.
Cohort studies are a form of longitudinal study where individuals are chosen based on whether they have had exposure to a risk factor of interest, and then followed up to ascertain whether they develop the disease.1 The ratio of the probability of developing the outcome in the exposure group compared to the non-exposure group is expressed as the RR (see Box 1.3). RR = 1 suggests no difference in risk between the two groups; RR >1 suggests an increased risk in the exposed group, whereas RR <1 suggests a reduced risk in the exposed group.
Cohort studies are more helpful in determining causality than other types of observational study. The individuals chosen are disease free to begin with, which makes it more likely that the exposure caused the outcome. However, cohort studies are costly because they require a large number of individuals and long follow-up periods.1
Epidemiology of Dementia
Dementia is a syndrome caused by a range of diseases that cause progressive cognitive decline, severe enough to affect functioning.8 There are many different causes of dementia. In this chapter, we shall only discuss the epidemiology of the three leading causes of dementia: Alzheimer’s disease, vascular dementia and dementia with Lewy bodies (DLB). Although there are some forms of dementia that affect younger people (see Chapter 3), dementia is in general a disease of older people, with age being the biggest risk factor. Consequently, with an ageing global population, the prevalence of dementia worldwide has been increasing. However, a review of world literature found stable or decreasing incidence and prevalence for dementia in high-income countries.9 In the United Kingdom, for instance, two consecutive Cognitive Function and Ageing Studies (CFAS I and CFAS II) showed that over the course of two decades there was a 23% reduction in the prevalence of dementia in three areas of England (Cambridgeshire, Newcastle and Nottingham): on the basis of the data from CFAS I, the researchers would have expected 8.3% (884,000 people) to have dementia, whereas the figure was 6.5% (670,000 people), representing a decrease of 1.8%.10 An important limitation to consider in light of these findings is the difference in response rates between the two studies. CFAS I had a response rate of 80%, compared to 56% in CFAS II, which makes interpretation of the findings difficult.
However, the CFAS results have been replicated in other countries9 and they provide a rich source of further information. For example, first, when combined with neuropathological findings, the CFAS raise fundamental questions about the basis of cognitive impairment. Wharton et al. have shown: a high prevalence of mixed Alzheimer’s and vascular pathology; a significant population of people who die with dementia, but whose brains show less of the pathology associated with dementia; and a group with a lot of pathology, but little dementia.11 This reflects the earlier findings of the Nun Study,12 which focused on nuns in the Order of Notre Dame in the United States of America (USA). Second, the studies have been used to show that people living with dementia are now relying much more on unpaid care than are people with other long-term conditions.13 Third, there has been much speculation around the reasons for the decrease in the prevalence of dementia in higher-income countries. They probably reflect changes in lifestyle. For instance, there is evidence that reduced salt intake has led to decreases in blood pressure, which may have contributed to decreases in mortality from stroke and ischaemic heart disease.14 Such changes would presumably have an impact on dementia pathology too. Meanwhile, the CFAS have shown that higher educational attainment, a more complex mid-life occupation and late-life social engagement are all associated with preserved cognitive function in later life.15
There is a lack of epidemiological data from low- and middle-income countries, but it is likely that, with their populations ageing, the prevalence of dementia is increasing. There are an estimated 47 million people living with dementia worldwide with estimates that this will rise to 48.1 million people by 2020 and to 90.3 million by 2040.16 For some years, the 10/66 Dementia Research Group has looked at ageing in low- and middle-income countries.17 In the regions covered by the 10/66 investigators, the prevalence of dementia has been estimated to be between 5.6% and 11.7% in those aged 65 and over.18 Their research shows great variation in prevalence rates, especially in comparison with European studies.4
The leading cause of dementia is AD. Alois Alzheimer first described the pathological hallmarks of AD-plaques and neurofibrillary tangles-in 1907, but even then there were vascular components too.19 There is an overlap between the risk factors for AD and other forms of dementia, and these are laid out in Table 1.2. Identifying these risk factors has led to recommendations to delay and prevent the development of dementia. These include active treatment of hypertension and diabetes, alongside smoking cessation advice and management of hearing loss.2
|Age and sex||The largest risk factor for the development of dementia and AD is advancing age. Women of all ages have higher prevalence and incidence of dementia than men.|
|Education and intellectual stimulation||There is an association between lower education levels and higher incidence of dementia and AD. The cognitive reserve hypothesis is a possible explanation for this finding.74 This theory suggests that education and intellectual stimulation extend neural reserve; therefore, the brain can withstand more insults before developing clinical symptoms of dementia. The ‘use it or lose it’ theory has also been used to explain the finding that lower levels of cognitive stimulation in early and middle life are associated with a higher prevalence of AD.75|
|Psychosocial factors||Longitudinal studies have shown that increased social engagement is associated with lower rates of cognitive decline and grand-parenting (up to a certain duration) is also associated with a reduced incidence of dementia.76 As discussed earlier, depression has been identified as a possible risk factor for developing dementia.7|
|Vascular risk factors||Smoking, high cholesterol, high blood pressure and diabetes mellitus have all been associated with an increased risk of AD. 78|
Vascular dementia is a form of dementia which is caused by diminished cerebral blood flow (see Chapter 2). It often presents as cognitive impairment following a stroke, although it can present as cognitive impairment with established cerebrovascular disease and no history of stroke.20
The classification of dementia is debated – how helpful are such classifications? – and we know that often we are not describing discrete entities.21 Generally, people presenting with cognitive impairment in their 60s and 70s are more likely to have a single aetiology underlying their dementia. Older people, however, as we have already seen, are more likely to have mixed pictures.11
Vascular dementia is the second most common cause of dementia after AD. Some degree of cerebrovascular disease is seen in 80% of dementias.20 Summing up a variety of studies, Kalaria suggested that ‘the worldwide frequency of [vascular dementia] in autopsy-verified cases is determined to be 10–15%, being marginally less than when clinical criteria alone are used’.22 Again, the prevalence of vascular dementia rises with age: among people 65 years and over the prevalence is 1.6%, but it rises from 0.1% in females and 0.5% in males between 65 and 69 years to 5.8% in females and 3.6% in males who are 90 years or above.23
Dementia with Lewy bodies is the most common cause of dementia after AD and vascular dementia. Lewy bodies are neuronal inclusion bodies and the location in the brain where they aggregate determines the clinical syndrome that predominates. Lewy bodies largely located in the basal ganglia lead to Parkinson’s disease. Cerebral Lewy bodies are associated with DLB, although they are not specific to it and are also found at autopsy in patients with AD.24 Like other dementias, the prevalence of DLB increases with age. The average age at presentation is 75 years.25 Lewy body pathology is identified more commonly than Lewy body disease is diagnosed clinically. In one systematic review where dementia classification was in accordance with the criteria of the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) and DLB subtype was diagnosed using the original (1996) or revised (2005) International Consensus Criteria, DLB accounted for 4.2% of all diagnosed dementias in the community, a figure that rose to 7.5% in secondary care.26 The incidence of DLB was 3.8% of new dementia cases.26 The exact prevalence of DLB is debated. Elsewhere it has been estimated to be 10–22% of the dementias in people 65 years or over, suggesting that 1% of this population have DLB.4 But the diagnostic rate depends on the diagnostic criteria used.26 Men have a higher prevalence of DLB than women, with one study showing a male-to-female preponderance in the ratio of 4:1.27 Possible reasons for this male preponderance include environmental risk factors and male sex hormones. Alternatively, women may be more likely to develop Alzheimer’s pathology and, therefore, have their cognitive decline attributed to AD.28
Behavioural and psychological symptoms of dementia (BPSD) include irritability, apathy, psychotic symptoms, agitation, ‘wandering’ (although it must be acknowledged that this term is unhelpful given that it covers a variety of behaviours, many of which are not purposeless as might be suggested by ‘wandering’), disinhibition, aggression (which, like ‘wandering’, might from a different viewpoint represent a perfectly reasonable response to a frustrating situation), restlessness, depression, sleep and appetite changes. A systematic review of longitudinal studies of BPSD, the majority from the USA, Canada and Europe, showed a high prevalence of apathy, depression, agitation and ‘wandering’, and a low prevalence of anxiety and hallucinations.29 Affective symptoms such as apathy and depression tend to persist rather than occur episodically. Included studies used a variety of assessment measures. For example, studies that used the informant-based Behavioural Pathology in Alzheimer’s Disease Rating Scale (BEHAVE-AD) to measure BPSD demonstrated a higher prevalence than studies using the Neuropsychiatric Inventory (NPI).29
Of course, dementia tends not to come alone. Older people are prone to a variety of comorbid physical diseases. In the Newcastle 85+ cohort study, people aged 85 years or older had on average five diseases, the most prevalent being hypertension (57.5%) and osteoarthritis (51.8%), with moderate to severe cognitive impairment featuring in 11.7%.30 Data from CFAS I and CFAS II, mentioned earlier, demonstrated that in the 10 years between the two studies the prevalence of diabetes and visual impairment in diabetes increased, which provides evidence that can be correlated with other possible comorbidities.13 But it is also possible that more than one mental illness might be present at one time. This is a complicated area because of difficulties ensuring that the same diagnostic criteria have been applied across studies and because of the possibility that one condition leads to or is part of the other. In one systematic review and meta-analysis of comorbid mental illnesses in people with dementia compared with their healthy peers, the mean prevalence of depression ranged from 20% in AD, to 37% in frontotemporal degeneration and to 13% in healthy controls.31 When all the different types of dementia were combined, the prevalence of depression increased to 25%, whereas in the controls it was 13%. In the same study, the prevalence of anxiety in all types of dementia was 14%, but this was not statistically different from the rate in the healthy controls (3%), albeit the anxiety analyses were thought to be underpowered (i.e. the sample sizes were not sufficient to show statistical differences). The authors of the review also considered the literature on post-traumatic stress disorder (PTSD) and dementia. They did not find enough (only two studies) to perform a meaningful statistical analysis, but the overall prevalence of PTSD in dementia was 4.4%.31
In a study in the north of England looking at people who had been diagnosed with early-onset dementia (called ‘pre-senile dementia’ in the paper, which referred to people between the ages of 45 and 64 years identified from records between 1985 and 1989) who had died by the end of 1998, 19.3% died at home, 24.5% in nursing or residential homes and 56.3% in hospital.32 In people aged over 65 years, dementia is one of the leading causes of death, and for those who have died from another cause with dementia as a contributory factor, stroke, Parkinson’s disease and ischaemic heart disease are the most common causes.33
Indeed, cognitive impairment is strongly associated with an increased mortality; a UK population study gives a median survival time from diagnosis of dementia to death as 4.1 years.34 A large cohort study of nursing-home residents in the USA revealed that, compared with residents with terminal cancer, residents with advanced dementia received more active and non-palliative interventions, less advanced care planning and palliative symptoms were not as well controlled.35 Similar patterns are seen in the UK. For example, in a study in London between 2012 and 2014, of 85 people living with advanced dementia in nursing homes or their own homes, pain (11% at rest, 61% on movement) and significant agitation (54%) were common and persistent.36 Aspiration, dyspnoea, septicaemia and pneumonia were more frequent in those who had died by the end of the study (mortality at 9 months was 37%). The researchers found that 76% had ‘do not resuscitate’ orders, 30% had an advance care plan, 40% had a documented preferred place of death and 40% had a lasting power of attorney.36
The Epidemiology of Functional Illnesses and Suicide in the Elderly
Depression is one of the most common mental health conditions to affect older people, although it is not a normal part of ageing. Depression in older people is associated with impaired functioning,37 increased use of healthcare services38 and an increase in mortality.39 Given that depression is a readily treatable disorder, this is a potential target for a significant improvement in morbidity and mortality in older people.38 Older people suffer bereavements and social role transitions such as retirement or loss of independent living. Despite this, major depression is not common in older people, with a review of the literature finding a prevalence of 1–5% (most studies report prevalence figures closer to 1%) in large studies in the USA of community-dwelling older adults.40 These studies tended to use DSM-IV criteria to define major depressive disorder. Minor depression, however, is common. The prevalence of depression was 8.7% (95% confidence interval (CI): 7.3–10.2) in those living in the community in England and Wales,41 and it was as high as 27.1% in a large UK-based cohort study of those living in institutions.42 The definition of minor depression varies between studies. In some it refers to clinically significant depression that does not meet criteria for major depressive disorder. In others, however, it denotes subclinical depressive symptoms. Which of these definitions is adopted will have a significant effect on prevalence and highlights the importance of the diagnostic criteria adopted by any particular study. Similarly, studies that use depression symptom checklists rather than diagnostic criteria can overestimate rates of depression in older people because positive responses can relate to physical symptoms or bereavement rather than depression.40 Data from 47 studies of older adults in acute hospitals show that the prevalence of depression is 29%.43