| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a The Pennsylvania State University, University Park
b Institute of Gerontology, Jönköping, Sweden
c University of Southern California, Los Angeles
Dee A. Haynie, 22400 Skyline Blvd., #33, La Honda, CA 94020.
Decision Editor: Toni C. Antonucci, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
DEPRESSIVE symptoms in very old people—age 80 and over—were explored in this study. Depression is a major health and common mental health problem among the elderly population, yet few studies of depression focus on the very old, despite the fact that the oldest old make up the fastest growing age group in the United States and many European countries (
Treas 1995
).
The relationship between aging and depression is ambiguous. When studies focus on clinically diagnosed depression, using DSM-III-R or DSM-IV criteria (
American Psychiatric Association 1987,
American Psychiatric Association 1994), prevalence rates are usually lower for major depressive disorder in older populations (
Girling et al. 1995). By contrast, when depressive symptoms are assessed as opposed to diagnoses, some studies report a curvilinear relationship with age.
Newman 1989 found that symptom scales yield a negative linear and a positive curvilinear trend, with higher depression scores in young adulthood, lower in middle age, and higher scores in the oldest age cohorts.
Kessler, Foster, Webster, and House 1992 replicated the curvilinear age curves with two large national surveys in the United States using an abbreviated form of the Center for Epidemiologic Studies–Depression (CES–D) scale (
Radloff 1977).
Gatz, Johansson, Pedersen, Berg, and Reynolds 1993 found both linear and curvilinear relations of depressive symptoms and age using the CES–D in a Swedish sample aged 29 to 95. There was some elevation of symptoms among adults in their 30s, lower scores in the 40s, and then a gradual increase decade by decade, until there was a marked elevation in very old age.
Gatz and colleagues 1993, however, found that age differences were less pronounced when examining the percentage of individuals scoring above the clinical cutoff score of 16 on the CES–D. In other words, the proportion of people meeting the CES–D cutoff criterion does not rise as rapidly with age as the overall symptom score.
The association between depressive symptoms and age in older adults has been explored in a number of studies.
Mirowsky and Ross 1992 examined the influence of decreased survival among the most disadvantaged individuals, increased physical dysfunction, erosion of personal control, and changes in social status over the life cycle (i.e., gains and losses of marital partners, employment, and income) as possibly accounting for age differences in symptoms of depression. Similarly,
Blazer, Burchett, Service, and George 1991 controlled for biological factors and stressful life events, including chronic illness, physical disability, reduced income, and loss of close relatives. Both studies reported that the covariates could largely explain cross-sectional age differences in depressive symptoms. In addition,
Mirowsky and Ross 1992 inferred an underlying benefit of increased maturity to explain why average levels of depression were not more responsive to accelerating declines and losses.
Factors identified in the etiology of depression are not necessarily inherent to the aging process (
Roberts, Kaplan, Shema, and Strawbridge 1997). In other words, chronological age per se is not the critical variable for depressive symptoms in later life. When the aging pattern was defined primarily by negative changes in health and psychological functioning, these changes were associated with depression. Risk factors such as poor physical health, disability, and social isolation account for higher depression scores among older persons. In turn, high depressive symptoms may be associated with an increased risk of subsequent onset of disability in activities of daily living (
Bruce, Seeman, Merrill, and Blazer 1994). This implies that healthy, normally functioning, older adults may be at no greater risk of depression than younger adults. It also suggests that intervention strategies that modify the risk factors could lower rates of depression (
Roberts et al. 1997).
Along with examining the rates of depression in later life, an important question is whether or not there are qualitative differences in the experience of depressive symptoms in very old age. In their study of Swedish young and old adults,
Gatz and coworkers 1993 found that four items of the CES–D indicating lack of well-being were endorsed to a similar degree by every age group except the oldest. These four items include the statements: I felt that I was just as good as other people; I felt hopeful about the future; I was happy; and, I enjoyed life (during the past week). The oldest age group endorsed fewer statements of happiness or well-being.
Gatz and colleagues 1993 hypothesized that this pattern of response might indicate that depression represents a different kind of experience for older adults. Lack of well-being may become a more salient feature of depression for older adults rather than negative symptoms such as sadness and tearfulness.
A particular issue when studying the oldest old is high rates of attrition, which could affect estimates of rates of depression.
Blazer and Koenig 1996 cite various studies indicating that there are higher rates of mortality for depressed elderly people (men in particular) with concurrent physical health problems. This relationship is found in clinical samples after important covariates have been controlled. Thus, people surviving into late life may be less likely to be depressed than other age groups, when morbidity and mortality are controlled. There may be other sources of selective attrition from longitudinal samples, besides mortality, that are related to depression. People having higher depressive symptoms may be more likely to refuse to continue in longitudinal samples. Another possible influence on changes in depressive symptoms over time is comorbidity with dementia. In a review of several longitudinal studies,
Nussbaum 1997 found evidence that depression in some older adults might represent an early marker for development of progressive dementia. As dementia progresses, however, reports of depressive symptoms decrease.
This study examined rates of depressive symptoms across three waves in a study of the very old (aged 80 and above), both initially and over time. Sociodemographic characteristics, activities of daily living (ADLs), subjective health, loneliness, and cognition were examined as potential correlates of depressive symptomatology. We also investigated sample attrition and comorbidity of dementia and depression for their possible relation to depressive symptoms. It was hypothesized that lack of well-being as opposed to negative affect would be a more salient feature of the depressive symptomatology expressed by these older people.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Information on depressive symptoms was available for 275 (78%) of the 351 people interviewed at baseline (Table 1 ). Failure to complete this part of the interview was mainly due to the frailty of respondents. A comparison of people who completed the CES–D at baseline and those who did not shows that the two groups differed on measures of age, ability to perform ADL tasks, and having a diagnosis of dementia. Participants completing the CES–D were younger, had better ADL performance, and were less likely to have a diagnosis of dementia (Table 2 ).
|
|
Demographic characteristics of the sample at baseline (N = 275) are shown in Table 3 . The average age of the participants was 83.18 years (SD = 2.91). Nearly two thirds (65.8%) of the participants were women. Eighty-nine participants (32.4%) were married at baseline, 37 (13.5%) were unmarried, 141 (51.3%) were widowed, and 8 (2.9%) were divorced. Educational achievement was on average 7.12 years (SD = 2.14), which is typical of older cohorts in Sweden. Most of the participants lived independently (90.2%, n = 248) and alone (58.5%, n = 161).
|
Depressive symptoms.
Depressive symptoms were measured using the 20-item CES–D scale (
Radloff 1977). Developed from samples in the United States, the CES–D has been found to have similar psychometric properties when used in translation into Swedish (
Gatz et al. 1993). The CES–D is not intended for use in making clinical diagnoses, but rather as a way to identify groups that are at risk for depression and to study the relationship between depressive symptoms and other variables.
Participants are asked about the frequency with which they experienced the 20 items during the past week. The items are rated on a 4-point scale, ranging from "rarely or none of the time" to "most or all of the time." Four positive items are reverse scored, so that higher scores (lack of well-being) indicate more depressive symptomatology.
Three scores are reported: (a) the total score of depressive symptoms, a sum of all 20 items; (b) the sums of the 16 negative items and 4 positive items, reflecting depressed mood and lack of well-being, respectively; and (c) the proportion of the sample that scores 16 and above, indicating a possible case of depression in a community population of older adults (
Lewinsohn, Seeley, Roberts, and Allen 1997).
There was high internal consistency of the CES–D in the present sample, as measured by Cronbach's alpha for all three waves (range 0.87–0.90). Construct validity is good in both clinical and community samples (
Avlund, Kreiner, and Schultz-Larsen 1996;
Heikkinen 1982).
Sociodemographic characteristics
The following sociodemographic characteristics were selected for their possible relationship to depression: age, gender, marital status, education, and living arrangement. Marital status was treated as a dichotomous variable in the analyses; (1) currently married at Wave 1, and (2) not married, which included widowed and divorced.
Dementia diagnosis
Dementia may have a relation to depression scores. People with early, mild dementia may have elevated depression symptoms, whereas in later stages depressive symptoms are generally lower (
Nussbaum 1997). For that reason, we considered whether participants met criteria for a diagnosis of dementia. A panel of three clinicians made independent diagnoses using DSM-III-R criteria. Diagnosis was based on review of results of a battery of neuropsychological tests, a neuropsychological interview with an informant, and medical records. Disagreements were addressed through a consensus conference. Diagnosis of dementia was made independently of information from the CES–D.
Activities of daily living
ADLs were evaluated along two dimensions, personal activities of daily living (PADLs) and instrumental activities of daily living (IADLs). These two dimensions are aspects of disability (i.e., difficulty doing activities to meet daily environmental demands) as opposed to functional limitations (
Verbrugge and Jette 1994).
(a) Personal activities of daily living. Participants rated their ability to perform the following six self-care tasks: getting up from and going to bed; taking a bath or shower; keeping clean and tidy, combing, shaving, etc.; dressing and undressing; going to the toilet; and eating. Ratings were on a 4-point scale ranging from being able to perform the task independently to not being able to do the task at all.
(b) Instrumental activities of daily living. Participants rated their performance on seven IADLs: housework, including cleaning and laundry; making the bed; cooking; shopping for food, etc.; going to places that can't be reached by foot; handling economy, postal and bank business; and using the telephone. Ratings were on a 4-point scale ranging from being able to perform the task independently to not being able to do the task at all.
These two scales are highly correlated and were summed to create a single ADL score. Cronbach's alpha for the three waves ranged from 0.92 to 0.95. Higher scores indicate better functioning.
Subjective health
Self-rated health was assessed with four questions: How do you appraise your general health condition? How do you rate your health compared to what it was two years ago? How do you rate your health compared to others your own age? and, Do you think that your health condition is preventing you from doing the things you would like to do? The ratings of the items were on a 3-point scale: good, about average, and poor. Cronbach's alpha for the three waves ranged from 0.60 to 0.67.
Loneliness
A 5-item loneliness scale was used. Participants were asked the following questions: (1) if they were troubled by feelings of loneliness; (2) whether they had friends with whom they could talk; (3) whether they feel a part of a set of friends; (4) whether they lack company; and (5) whether they feel abandoned. Response options consisted of a 4-point scale that ranged from nearly always/to a high degree to hardly ever (never)/not at all.
Cognition
Cognitive functioning was assessed using the Information subtest (verbal) and the Block Design subtest (timed performance) of the Wechsler Adult Intelligence Scale—Revised (WAIS–R). We selected these two cognitive measures to better differentiate people who are at the high end of functioning.
Analyses
The findings on depressive symptoms are analyzed in 5 steps:
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In order to examine changes over time, we next divided the sample according to whether they had Wave 1, 2, or 3 of data on the CES–D. Table 4 shows the prevalence of depressive symptoms at Wave 1 for those participants with CES–D scores at all three waves (n = 164). These participants showed a mean score of 7.46 (SD = 7.03) at Wave 1, 5.81 (SD = 7.65) at Wave 2, and 5.75 (SD = 8.10) at Wave 3. Depression symptoms decreased significantly from Wave 1 to Wave 2 (paired t (163) = 3.04, p < .01), but remained the same from Wave 2 to Wave 3 (paired t (163) = .098, p = .922). The group with two waves of data also showed a decrease from a mean of 10.00 (SD = 11.05) at Wave 1 to a mean of 8.01 (SD = 10.08) at Wave 2. This difference was not significant (paired t (53) = 1.68, p = .100). These results were replicated when the analyses were conducted using data from the other twin of the twin pair.
|
|
|
Role of Negative Affect and Lack of Well-Being
We next examined the scores for the negative symptoms and the items reflecting a lack of well-being on the CES–D. Because there were more negative than lack of well-being items, the mean strength of response per symptom was calculated by dividing the total scores for negative and lack of well-being items by the number of items. Scores were coded on both subscales so that higher scores would always indicate more depression. The results are shown in Table 7 for participants with three waves of data. There was a greater lack of well-being at Wave 1 compared to the negative symptoms (mean item responses of 0.74 and 0.26, respectively, paired t [163] = –9.35, p < .001). These results indicate that a lack of well-being, as opposed to negative affect, was a relatively bigger contributor to the overall depression score. Lack of well-being was also higher at Wave 2 (mean item response = 0.46 for lack of well-being and 0.23 for negative items, paired t [163] = –5.98, p < .001), and the same pattern was found for Wave 3 (mean item response = 0.46 for lack of well-being and 0.22 for negative items, paired t [163] = –5.87, p < .001).
|
Predictors of Negative Items and Lack of Well-Being Items of the CES–D
Multiple regression was used to examine factors associated with depressive symptoms. Because the two factors that make up the CES–D for this sample (negative symptoms and lack of well-being) were endorsed differently by participants, we decided to conduct separate regressions for these two scales. Variables were entered into the regressions in two blocks. Sociodemographic characteristics were entered first, as they are often considered strong predictors of depression (
Heikkinen, Berg, and Avlund 1995). The predictor variables were age, gender, marital status, and education. The second block of variables reflected current functioning, including ADLs, subjective health, cognition, and loneliness.
The bivariate correlations of the negative items criterion variable and predictors can be seen in Table 8 . The results of the multiple regression analysis are shown in Table 9 . The set of sociodemographic characteristics alone were not significant predictors (F [4,222] = .37, p = .83) of depression and accounted for less than 1% of the variance. When all of the predictors were entered, the result was significant (F [9,217] = 5.89, p < .001) and accounted for 19% of the variance. Activities of daily living, subjective health, and block design were significantly related to negative affect. Subjective health was reverse scored so that a lower score indicated higher perceived health. Thus, the positive coefficient indicates that better subjective health is related to lower negative affect. Higher scores on the block design were related to lower negative affect. When the criterion variable was lack of well-being items, none of the predictor variables were significant.
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
;
Kiljunen et al. 1997). The rates of depressive symptoms in this sample of oldest old were relatively low at baseline and decreased somewhat from Wave 1 to Wave 2. Most participants stayed below the clinical cutoff of 16 for the CES–D from Wave 1 to Wave 2 (82.9%), and the majority stayed below the cutoff from Wave 2 to Wave 3 (80.5%). The proportion of the participants scoring above the cut-off did not differ from one wave to the next. These results do not replicate findings from other studies of Swedish twins, where depressive symptoms increase in very late life (
Gatz et al. 1993), but are consistent with other Scandinavian findings on the oldest old, suggesting relatively low rates of depressive symptoms (
Kiljunen et al. 1997).
The Wave 1 baseline mean score on the CES–D across all subjects was 8.80, and 7.46 for participants with three waves of CES–D data. These means are similar to the
Berkman and coworkers 1986 study in which the mean scores on the CES–D ranged from 4.86 for people aged 65 and older without any physical disabilities to 13.51 for those with major functional disabilities. Those participants aged 75 years and older had a mean score of 9.01.
Our findings also suggest that the low rates of depressive symptoms may, in part, reflect selective attrition. At baseline, people with greater disabilities and/or dementia did not complete the CES–D. Attrition at subsequent waves was largely due to mortality, with people who died before the third wave having higher initial scores on the CES–D. Participants who received a probable dementia diagnosis at some point in the study did not differ significantly on initial CES–D scores when compared to those participants who never received such a diagnosis. This result was found not only for those participants who completed the CES–D at all three waves, but also for those who had at least completed the CES–D at baseline. Thus, the findings on attrition raise the possibility that depressive symptoms were underestimated to some extent by the exclusion of some people with severe disability at Wave 1 and higher mortality of those with initially more depressive symptoms.
Gatz and colleagues 1993 emphasize the importance of recognizing that the presence of symptoms of depression may represent a different kind of experience for older adults. Lack of well-being as proposed by those researchers was in fact more salient for this sample of the oldest old. Interestingly, the results in the current study indicate that a significant decrease in depressive symptoms over time was due primarily to a decrease in symptomatology, indicating a lack of well-being. One possible explanation for the role of lack of well-being in contributing to depressive symptomatology is that this pattern of findings is consistent with what is seen in two-factor theories of well-being, where both positive affect and negative affect drop in intensity (
Stallings, Dunham, Gatz, Baker, and Bengtson 1997). Stallings and colleagues found that, consistent with a two-factor theory of well-being, desirable life events predicted change in positive affect and undesirable events predicted change in negative affect. Results from their study suggest that the expectation of life events might be more important than whether they are desirable or undesirable.
Another explanation for the decrease in lack of well-being involves the possibility that being a participant in the study introduced a positive effect. In other words, there might have been a reactive component in terms of participation. The participants were interviewed in depth by nurses, and these experiences may have provided the participants with some degree of social involvement and a sense that their psychological and physical well-being were being monitored. While the cause of the decrease in lack of well-being cannot be pinpointed, it is clear that depression was not increasing throughout the three waves.
A finding of the study indicated that when negative affect and lack of well-being were used as criterion variables, only the model for negative affect was significant. Variables were entered in two blocks. Sociodemographic characteristics (age, gender, marital status, and education) alone were not significant predictors of negative affect. Three additional predictors, namely ADLs, subjective health, and block design, were significantly related to negative affect. This finding indicates that negative symptoms were more related to ADL deficits, suggesting that in a more disabled sample, depressive scores might be higher. These are predictors that typically have been found in studies of depressive symptoms (
Zarit, Femia, Gatz, and Johansson 1999). There has been less effort to examine negative affect and lack of well-being separately. The findings raise the possibility that lack of well-being has a different set of predictors, at least in very late life.
The results of a study by
Lewinsohn and colleagues 1997 found that functional impairment, physical disease, and the tendency to answer items in a socially desirable way were correlated with the CES–D and with age. In the present study the participants were relatively healthy and high functioning. Therefore, some of the expected relationships, such as low functioning on ADLs and depression, were not found for those participants who completed the CES–D on at least one occasion of measurement. A comparison of people who completed the CES–D at baseline and those who did not shows that the two groups differed on measures of severity and age. Participants completing the CES–D were younger, had better ADL performance, and were less likely to have a diagnosis of dementia. In fact, the best predictors of negative affect for this sample included subjective health and performance on the block design.
One potential limitation of this study is the generalizability of findings from a sample of twins. Among samples of older twins, two particular problems might limit generalizability of findings. Twin studies, including OCTO–Twin, typically require that both twins be alive and willing to participate. These requirements could bias the sample toward better functioning individuals. The question of the representativeness of twin samples was investigated by
Simmons and coworkers 1997. They compared a population-based sample in Sweden of non-twins in their 80s to a random selection of one member of each twin pair from the current sample across the domains of vitality, well-being, physical and cognitive functioning, and health utilization to evaluate possible selection bias. Relatively small differences were revealed after controlling for age, education, and gender.
The authors concluded that, although a selection bias may exist in twin samples, the extent of bias appears to be small. A selection effect is observable in participation rates, however. If twins are treated as individuals, the rates of refusal between twin and singleton samples are similar; however, because both twins must agree to participate, the result is a lower overall participation rate at Wave 1 (64% compared to 86%). Given the availability of comprehensive and longitudinal data on a sample aged 80 and older, we believed that the potential limitations of using twin data were outweighed by the benefits.
Kiljunen and colleagues 1997 offer alternative explanations for the low prevalence of depressive symptoms and depression in very old people. They suggest that those who survive into very old age may have physical and psychiatric advantages as well as coping abilities that have allowed them to handle their life circumstances. Therefore, depression may become less prevalent. It can also be assumed that the most depressed individuals may have died before they reached the age of 85 years, and individuals with optimistic mood are still alive. This perspective meshes well with the description of people in their late 80s and early 90s as having an "aura of survivorship," as characterized by
Johnson and Barer 1996.
|
Acknowledgments |
|---|
The OCTO–Twin Study (Origins of Variance In the Old-Old: Octogenarian Twins) is an ongoing longitudinal study that is being conducted at the Institute of Gerontology (IFG), University College of Health Sciences, Jönköping, Sweden, in collaboration with the Center for Developmental and Health Genetics at the Pennsylvania State University, University Park, PA, and the Division of Genetic Epidemiology, Karolinska Institute, Stockholm, Sweden.
We are greatly indebted to project assistants Lene Ahlbäck (1991–present), Agneta Carlholt (1993–present), Gunilla Hjalmarsson (1991–1993), Eva Georgsson (1993–1995), and Anna-Lena Wetterholm (1993–1994), who traveled throughout the country and examined twins. Invaluable support in coordinating the study (1991–present) is provided by Inger Cronholm at IFG and Elana Pyle at Penn State.
Received for publication September 16, 1999. Accepted for publication July 6, 2000.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  R. S. Wilson, S. E. Arnold, T. L. Beck, J. L. Bienias, and D. A. Bennett Change in Depressive Symptoms During the Prodromal Phase of Alzheimer Disease Arch Gen Psychiatry, April 1, 2008; 65(4): 439 - 445. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T.-M. Lyyra, T. M. Tormakangas, S. Read, T. Rantanen, and S. Berg Satisfaction with present life predicts survival in octogenarians. J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2006; 61(6): P319 - P326. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Fiske, M. Gatz, and N. L. Pedersen Depressive Symptoms and Aging: The Effects of Illness and Non-Health-Related Events J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2003; 58(6): P320 - 328. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
