The epidemiology of depression in the elderly can be approached at three levels: its occurrence in the elderly living in the community, in those reaching primary care, and in the residents of hostels and nursing homes.

For many decades across the world, the traditional pattern has been for the highest rates of suicide to be in elderly men. This has now changed. In over a third of countries, both developed and less developed, it is younger people who have come to carry the highest rates. The World Health Organization provides a valuable resource for such data, showing rates by age and gender for nearly all countries from 1950 onwards.⁽⁶⁾ The pattern varies considerably between countries. For example, in the United Kingdom in 2002, men aged 75 and over had a rate of 10 per 100 000, whereas the highest rate was 18 per 100 000, in men aged 35-44 years. In the United States and in the Russian Federation, the rates for men aged 75 and over were 41 and 89 per 100 000, respectively.

The main risk factors for suicide in the elderly are a past history of an attempt, depressive disorder, physical illness or disability, chronic pain, recent losses, social isolation, and access to lethal means. While universal interventions are more powerful than selected factors in prevention,⁽⁷⁾ these attributes can be used in selective intervention to identify groups at increased risk. Furthermore, being multiplicative, these markers are of great value in individual cases by alerting the clinician to a person needing particular attention. Here is another example of the use of epidemiology for prevention. A systematic review of suicide prevention strategies for all age groups concluded that two interventions did reduce rates: physician education in recognizing and treating depression; and restricting access to lethal means.⁽⁸⁾ Both of these interventions have close relevance to the elderly.

The subject matter here refers to older people who have enduring attitudes and behaviour that bring difficulties for themselves or for others.⁽⁹⁾ There is only sparse information on the prevalence of personality disorders in the general population, let alone specifically in the elderly. One exception, based on a national survey of mental health, found a lifetime prevalence of 6.5 per cent across all age groups with a trend towards lower rates with increasing age.

In clinical practice, it has long been suggested that traits such as impulsivity and externalizing behaviours tend to become less frequent in later life, whereas anxiety-prone, dependent, schizoid, paranoid, or obsessional persons are likely to change little as they age, or to become more so. Bergmann’s pioneering enquiries among the elderly of Newcastle upon Tyne found that it was the anxiety-prone and insecure types that had late-onset neurotic disorders. A more recent study of late-life depression found an overall prevalence of comorbid personality disorder of 10-30 per cent. The group formerly known as neurotic and more recently as Cluster C in the DSM classification, had the higher prevalence. The Cluster B group, those with borderline, narcissistic, histrionic, and antisocial traits, were rare. What is not yet established, however, is if this lower prevalence also exists in the general population of the elderly, not just among cases with depressive disorder who have reached treatment in specialist services.

The epidemiology of personality disorders in later life is therefore significant for two reasons. First, some types are associated with increased risk of anxiety, depression, or paranoid states (vide infra). Second, there remains much yet to understand about the natural history of the personality disorders across the lifespan.

For the functional psychoses of late life, epidemiological information comes from two sources: studies of persons who have reached psychiatric services; and surveys of elderly persons living in the general community.⁽¹⁰⁾ Psychotic symptoms probably exist as a continuum of severity, with only the more developed cases meeting diagnostic criteria. These often, but not always, reach psychiatric services, not uncommonly through being brought to the attention of the police. States phenomenologically similar to those found in clinics do occur in the community in non-trivial numbers. For cases that reach the threshold for a diagnosis by virtue of the range and severity of symptoms and behaviour, it has been proposed that cases with onset after the age of 60 years be called ‘very-late-onset schizophrenia-like psychosis’. The syndrome has a 1-year prevalence
of 0.1 to 0.5 per cent. For advancing knowledge about the aetiology of schizophrenia, any information on it might be useful in explaining why people with this syndrome have reached the seventh decade or later in life without becoming psychotic, and only then develop it. It is more common in women. This is unlikely to be due to different social visibility or access to services. It is associated with a better premorbid level of social and occupational functioning. Premorbid paranoid or schizoid traits have been implicated and both clinical and community-based studies have found an association with sensory impairment such as deafness or poor eyesight. Personal and environmental factors associated with ageing have been considered, such as physical ill health, bereavement, loss of friends, and loss of income, but these have not been shown to contribute significantly. Genetic factors appear to be less important than in earlier onset schizophrenia.

It is generally believed that the prevalence of alcohol abuse and dependence declines during adult life and that the elderly have low rates in most communities. This may well be the case, but some other factors have to be considered. Whatever the prevalence, the absolute numbers will rise in the future because of the unprecedented growth in the elderly population. Next, the assumption may be false. In community surveys, errors in the ascertainment of alcohol abuse may lead to an underestimate for older persons. Most screening instruments were developed for use on younger adults, so their validity in the elderly is largely undetermined. Measures of the quantity drunk may mislead because smaller amounts may have an intoxicating effect in persons whose body fat, lean tissue, cerebral reserve, and metabolic function have declined. So the usual cut-off for problem drinking may be set too high for the elderly. One review of screening instruments concluded that the CAGE and MAST-G scales were appropriate, whereas other widely used instruments were not.⁽¹¹⁾ Next, all the studies have been cross-sectional. The elderly may have lower rates because of a cohort effect, whereby people born in the first half of the twentieth century may have been more moderate drinkers for all their life, compared to the high levels of consumption that are now found in the young of both sexes.

The actual values for prevalence are dependent on the instrument used and the definition used to define problem drinking, alcohol abuse, or dependence.⁽¹²⁾ One review of community studies gives a figure of 5.1 per cent using various definitions. Invariably, men have higher rates than women. There is also considerable variation between countries and across different cultures. In identifying cases, a distinction of clinical significance needs to be made between late-onset and long-standing alcohol abuse. In primary care, accident and emergency departments, hospital in-patients, and nursing homes, the prevalence is much higher, yet cases are consistently under-recognized. The use of screening instruments in all of these settings has been advocated to improve this.

Alcohol abuse carries important comorbidity. In addition to all the established medical complications, it is associated with falls, subclinical delirium, cognitive decline, and depression. One study demonstrated a five-fold increase in the risk of developing a psychiatric disorder, especially depression and dementia. Simultaneous use of benzodiazepines, itself common in older persons, is clearly an additional and important factor. Against all this, it should be recalled that moderate alcohol use has been found in population studies to be associated with better mental and cardiovascular health, as well as being subjectively enjoyable.

In the last two decades the dementia field has registered a tremendous scientific progression in many research areas including aetiology, pathogenesis, clinical aspects, treatment, and prevention. These advances have opened new perspectives, especially concerning definitions and diagnostic criteria, which have a relevant impact on epidemiological research.

Dementia is still defined as a syndrome which includes memory deficits and disturbance of other higher cortical functions; these major symptoms are commonly accompanied, and occasionally preceded, by deterioration in emotional control, social behaviour or motivation. However, it has become apparent that memory impairment may not necessarily be the major or first symptom for dementia subtypes such as frontotemporal dementia (FTD) and vascular dementia (VaD). Furthermore, as the current definition requires impairment severe enough to interfere with daily functioning, in several cases a delay of the diagnosis occurs. For that reason, a new research line has emerged with the aim to detect early Alzheimer’s disease (AD) and other dementias, and the terms mild cognitive impairment (MCI) and cognitive impairment no-dementia (CIND) have been proposed to identify those subjects that show a clear cognitive deficit but do not fulfil diagnostic criteria for dementia. Finally, it is well known that dementia syndrome can be induced by many different underlying diseases, and that a differential diagnosis may be difficult for several reasons. AD as well as other dementia subtypes shows heterogeneity with distinct clinical and pathological characteristics; many different dementing disorders overlap in clinical and pathologic features; and different dementing disorders may make a common contribution or interact in causing dementia symptoms. Thus, rather than viewing, for example, AD and VaD as dichotomous entities, it may be more relevant to consider the role of their additive or synergistic interactions in producing a dementia syndrome.^(13,14)

Following these new perspectives, in this chapter we will summarize the major findings from the most recent epidemiological research according to three major topics: early detection of AD and other dementias, incidence and risk factors for AD and dementia, and prevalence and impact of the dementing disorders at the individual and societal levels.