Abstract
Simone De Beauvoir [1] recorded her negative views about old age. She shared a widely held opinion that late life is accompanied by an increased likelihood of depression, which was regarded as a predictable, understandable response to the losses and declines of the last period of life. ‘Old people are physically fragile. Socially they are outcasts, and this has serious effects upon their mental state. Both their existential situation and their sexual state are favourable to the development of neuroses and psychoses’ (p. 548 [1]). Jarvik [2], after referring to the deprivations and losses associated with late life, asked rhetorically why old persons are not all in a profound state of depression.
Numerous researchers have confirmed that depression is, indeed, a common problem in old age, though reported prevalence rates vary, depending to a large extent on the criteria used to define depression and ‘caseness’. Copeland et al. [3] described depression as a ‘highly prevalent condition among older people in the European Community, and at a level of severity selected… as meriting intervention’. Findings from large community surveys have been summarized [4–7]. Studies selected by the author as being representative, and which confirm that depression is common in late life, are listed in Table 1 (DSM-defined depressive disorders) and Table 2 (depression identified by use of structured interview schedules). A more complete list is provided by Beekman et al. [7], who referred to 34 epidemiological studies of old age depression, and the author is aware of others from around the world.
The prevalence in old age of major depression and other depressive disorders (fulfilling DSM criteria): results from representative studies.
The prevalence of depressive symptoms and clinically significant depression (AGECAT, etc.)
Studies that used DSM criteria to diagnose depressive disorder recorded prevalence rates of major depression in old age, and of DSM depressive disorders other than major depression, that varied considerably [8–19,Table 1. In Canberra, the prevalence of depressive disorders was reported to be 1%% [14], but in Hobart it was much higher [9]. The study by Forsell et al. [18] of a Swedish sample with mean age 84.6 years included only major depression and dysthymia as depressive disorders, the combined rate being 10.7%%, whereas the Epidemiologic Catchment Area (ECA) study [10] reported the combined rate among persons aged over 75 years was only 2.4%%. There was less inconsistency between the findings of researchers who reported rates of clinically significant depression identified in community studies of older people, using structured interview schedules [3,19–29, Table 2. The studies were conducted in various countries, and no doubt demographic differences could explain some of the variation in rates.
The study by Newman et al. [19] is of interest because, having reported the combined rate of major and minor depressions among Edmonton older persons as 4.5%%, they noted that the prevalence according to AGECAT (a computerized diagnostic algorithm developed by Copeland et al. [30]) was 11.4%%. They commented that AGECAT status is determined mainly by the number of identified dysphoric symptoms.
Do depression rates vary with age?
Relatively few studies allow comparison of rates of depression in different age-groups and in those that do, the findings have been inconsistent [31]. Of 14 studies that examined age-associated variations in scores on depressive symptom scales, five showed an increase with age, four showed no age-group differences, three showed a decrease, and two reported a fall followed by a rise [31]. A further study [32] reported a decrease with age, but noted that certain depression items (including hopelessness and loss of interest) were more likely to be endorsed by elderly people. The authors concluded that the nature of depression experienced by younger and older people may differ qualitatively, and referred to agerelated changes in the somatic and psychological components of depression.
Among studies that reported the prevalence of depressive disorders defined by DSM criteria, several reported no age difference, but the ECA study showed a peak depression rate at age 25–44 years [33], two studies [34], [35] reported a peak at age 55–64 years, and another [36] showed a peak at 60–69 years. In some the pattern differed between genders. The Mental Health and Wellbeing study in Australia [37] (which did not use DSM criteria) found that peak rates among men and women were at 35–44 years and 18–24 years, respectively. Romanovski et al. [38] provided further details about the ECA study's Baltimore site, where the peak prevalence rate of major depression was at age 45–64 years (2.0%%), but the peak prevalence of DSM-III depressive disorders other than major depression was among persons older than 65 years (5.0%%).
Those reporting lower rates of major depression in old age have provided various theories to explain their findings [39]:
1. Mortality rates are higher among those with major depression, so that they die before reaching old age; episodes of major depression in old age are briefer than at earlier ages, possibly because they become ‘subsyndromal’ more quickly.
2. Depression presents differently in old age, so that criteria for DSM-IV major depression are not fulfilled even in cases of severe mood disturbance. Age-related changes can lead to altered manifestation of symptoms.
3. Those with depression are selectively removed from the community to residential care; there is a well–recognized association between depression and disability [40].
4. Factors associated with development of major depression in younger age-groups are less commonly encountered in old age.
5. A cohort effect may have led to a higher prevalence of major depression in younger generations.
However, the apparent decrease in prevalence in old age could have been due to errors in case ascertainment. This does not apply just to recognition of major depression. There may be reasons why depressed older people may respond differently compared with depressed young people when interviewed by researchers. They may have more difficulty (because of age-associated brain changes) perceiving or remembering details about their symptoms. Other reasons are discussed below.
However, whatever the reasons for the inconsistencies, there is reason to question whether it is appropriate for those examining the epidemiology of depression to concentrate attention on major depression and dysthymia (as some have done), without reporting the prevalence and course of other types of depression, and factors associated with their development and outcome. This applies particularly to late life depression, because of differences in the way depression presents in old age [41]. Symptom patterns among elderly people, who are depressed in relation to physical ill-health or disability or brain changes, may not correspond to those considered necessary for a DSM diagnosis of major depression or dysthymia, even though their distress is just at persistent and severe, and resultant impairments in functional abilities may have devastating consequences. Lyness et al. [42] showed that patients with ‘subsyndromal’ depressive symptoms were functionally disabled to a degree comparable to patients suffering major depression or dysthymia. It was shown that such patients experienced significantly more social dysfunction and disability than those who were symptom free, and their psychosocial functioning improved when they were treated with an antidepressant [43]. As stated by Copeland [44] it did not make sense that many depressions identified by clinicians have to be described as ‘subsyndromal’. It would be more relevant to identify and classify cases that would benefit from clinical intervention. Blazer [45] commented that the debate over the relative frequency of depression in different age-groups is often meaningless because of varied definitions of ‘caseness’.
Many cases of treatable depression (especially in old age) do not fulfil criteria for major depression or dysthymia. It follows that epidemiological studies that limited themselves to major depression and dysthymia give only limited information about how the precipitants and consequences of treatable depression may differ between age-groups.
Jorm [31] rejected from inclusion in his review those studies that covered only one end of the adult life-span. He therefore omitted the study by Roberts et al. [46] that showed an increase in the prevalence of major depressive episode from 8.1%% at age 50–59 years, through 6.9%% at 60–69, to 10.4%% at 70–79 and 12.7%% at 80 years and more. Beekman et al. [16] showed a progressive rise, but from only 1.3%% at 55–59 years to 2.7%% at 80–85 years, with a corresponding rise in rates of minor depression from 9.4%% to 16.7%%. Kramer et al. [10] reported a rise in major depression rate from 0.7%% at 65–74 to 1.3%% at 75 years or more, while Kay et al. [9] reported a rise from 6.3%% at 70–79 years to 15.5%% at 80 years or more. Prince et al. [47] referred to data from 14 centres in Europe (21 724 subjects aged 65 years or more) when noting a trend for an increase in scores on a depression scale with increasing age, but an analysis of data from nine of the centres [3] showed no overall tendency for the prevalence of depressive ‘caseness’ to rise or fall with age. The rates vary considerably, as discussed, though a majority of studies have shown that, among those aged over 65 years, the prevalence of depression increases with age. This suggests that even if rates of depression really are higher in youth and middle age than in old age, there may be a second peak in late old age. The studies demonstrate the importance of examining rates for the ‘young-old’ and ‘old-old’ separately.
It is relevant to note that among males in Australia there are two peaks in the suicide rate, one in young adulthood, the other in late old age, the lowest rate being at about age 60 years [48]. The rate of female suicide is much lower and remains much the same across different adult age groups. It is also relevant to note that (if latest Australian prevalence studies are correct regarding depressive disorders) a far higher percentage of older males with major depression than of young males with this diagnosis die by suicide. This is further evidence that the diagnosis itself may not be as important (in relation to management and outcome) as formulating the relevance and meaning of symptoms.
Errors of ascertainment
It should be recognized that older persons may respond differently from younger people to being questioned about symptoms and feelings. Older people with depressive disorders are less likely to acknowledge being sad, down or depressed in mood [39]. They are less likely to admit to feelings of hopelessness or worthlessness or anhedonia [49]. Difficulties with hearing or understanding questions may be more frequent in old age, but it could be that elderly people are more inclined to misrepresent their feelings in order to fend off perceived threats to their self-esteem. Studies that rely on subject-reported symptoms may underestimate the prevalence of disorders that fulfil DSM criteria.
Diagnostic interview schedules used in cross-age studies have been those primarily developed for assessing mental disorders among physically well, young or middle-aged adults. They were not designed to facilitate recognition of depression precipitated by or associated with physical, cognitive and environmental changes that become more common in late life. Jorm [31] noted that the Diagnostic Interview Schedule and the Composite International Diagnostic Interview (CIDI) may discount symptoms that may be attributable to physical illness. Unless interviewers are enabled to use clinical judgement (which is not feasible if they are lay interviewers) somatic symptoms may be mistakenly attributed to physical disorders rather than depression. It is desirable to give consideration to the clinical context of all symptoms that could be depressive, even if feelings of depression are denied, and to seek additional information from those who know the subjects well even if they too believe the symptoms have a physical cause.
Another consideration before accepting evidence from cross-age studies that prevalence and presentation of depression vary with age is that older persons might be more likely than younger people to decline involvement in surveys. This could lead to bias. There is certainly evidence that persons with certain age-associated organic disorders may be excluded from such surveys: the Mental Health and Wellbeing study [37] excluded persons with moderate or severe dementia and those living in residential care. It may well be that people with disabilities (including deafness) are less able and less willing to be involved in community surveys. Response tendencies can be influenced by cognitive status and the presence of chronic disease [50]. Older people with disabilities are known to be more at risk of developing depression [40], but may selectively exclude themselves from epidemiologic studies.
Most cross-age studies have not reported whether response rates differed between age-groups. The recent big Australian study [37] did not report differential response rates. Following the ECA study, Kramer et al. [10] recorded that interviews were completed in only 60%% of those with people aged over 65 years, but from over 80%% of those under 65 years. Henderson et al. [51] reported a response rate of 54%% among people aged 60–79 years but 70%% among those aged 18–59 years.
Given the inconsistencies and potential for bias discussed above, it cannot be stated unequivocally that major depression is less common in old age than at younger ages. Some of the above-mentioned evidence indicates that the prevalence is lower in early old age and escalates in late old age. Most cross-age studies included too few very elderly persons to allow comment on this issue.
Conclusion
Findings concerning age-related differences in the prevalence of depression have been inconsistent. Errors in case ascertainment could explain why some cross-age studies have found a much lower prevalence of major depression in old age. Such studies commonly do not provide comparisons of the prevalence of ‘subsyndromal’ depressions in different age-groups, even though they may be just as distressing and disabling as major depression.
Data from studies concerning the varying prevalence of major depression in different age-groups should not determine allocation of mental health care resources for management of depression. What matters is whether those resources, together with other services available in the community, can help reduce distress, improve quality of life, and have a beneficial effect on functional abilities of depressed people. It may be that, per dollar spent, effects on subsyndromal depression may be more costbeneficial than on those with major depression. But this is not an either/or situation. Research interest and allocation of health care resources should not be limited to major depression and dysthyma.
There is evidence that depression (including major, dysthymia, organic and ‘not otherwise specified’) is as common in old age as in earlier life. Yet data from the Mental Health and Wellbeing study [35] have been quoted in various policy documents to imply that older people are less likely than young people to need psychiatric treatment for depression. Further, the implication to general practitioners has been that depression is far less common in old age, and therefore there is less need to look out for it. This is a misinterpretation of the data.
Age itself does not predispose to or protect from depression, although the prevalence of different types of depression appears to be related to age-associated factors, including disability. There is a need for recognition that severe, distressing but treatable depression in old age commonly does not fulfil DSM-IV criteria for major depression or dysthymia.