| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, North Carolina
b Center for the Study of Aging, Duke University Medical Center, Durham, North Carolina
Dan G. Blazer, Department of Psychiatry and Behavioral Sciences, Center for the Study of Aging, Box 3003, Duke University Medical Center, Durham, NC 27710 E-mail: blaze0001{at}mc.duke.edu.
Decision Editor: John E. Morley, MB, BCh
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Methods. Information on depression (CES-D scores), mortality, demographics, body mass index, chronic disease, smoking history, cognitive impairment, functional impairment, self-rated health, and social support was obtained from a stratified probability-based sample of community-dwelling elderly persons, with equal distribution between African Americans and whites in the Piedmont of North Carolina. Descriptive statistics were calculated, and logistic regression was used for a series of models with progressively more control variables.
Results. The unadjusted relative odds of mortality among depressed subjects at baseline was 1.98 over 3 years of follow-up. Inclusion of age, gender, and race into the model did not reduce the relative odds. When chronic disease and health habits, cognitive impairment, functional impairment, and social support were added to the model, the odds ratios for mortality with depression were 1.74, 1.69, 1.29, and 1.21, respectively. This decrease in odds ratios was not observed for other variables in the model when additional variables were added.
Conclusions. The estimated odds of dying if depressed moved toward unity as other risk factors for mortality were controlled. Unlike other known risk factors for mortality in the elderly population, depression appears to be associated with mortality through a number of independent mechanisms, perhaps through complex feedback loops.
THE association of depression and mortality among the general population has been well established in the literature (1)(2)(3)(4)(5). The evidence for such an association in the elderly population, however, remains inconclusive, especially in well-controlled studies. Bruce and Leaf found a fourfold increase in the odds of dying over a follow-up of 15 months if persons 55 years of age or older experienced a mood disorder in a community sample (6). Henderson and colleagues confirmed an association between depression and mortality in Australia (7), and Takeida and colleagues found such an association in Japan (8). Fredman and colleagues, in contrast, found that neither major depression nor significant depressive symptoms were associated with higher mortality when individuals 60 years of age or older in a community sample from North Carolina were followed for 24 months (9). Thomas and colleagues also failed to find an association in controlled analyses of a community sample from New York City (10).
One reason for these conflicting results is the issue of which variables are controlled in the analysis. Many potential control variables are associated with both depression and mortality, the primary ones being age, medical illness, cognitive impairment, and functional impairment. Depression has been shown in numerous studies to be associated with medical illness (11)(12). Depression is also known to be associated with cognitive impairment, especially via comorbidity with Alzheimer's disease and vascular dementia (13). The relationship between depression and functional impairment has also been well established in the literature, with the association suggesting a complementary feedback mechanism (i.e., in longitudinal studies, depression can precede functional impairment or functional impairment can precede depression) (14). Functional impairment, in turn, is a known risk factor for increased mortality (15). Depression is also known to be associated with smoking, impaired social support, unmarried status, and poor self-rated health (16)(17)(18)(19). These factors, in turn, have been associated with increased mortality (19)(20).
Furthermore, many investigations that controlled for functional impairment did not find a relationship between depression and mortality (9)(21). One interesting variant of this finding is that depression may be a risk for mortality in persons with physical illness (22). Inouye and colleagues found that functional measures were strong predictors of mortality after hospitalization and that these measures contributed substantially to the prognostic ability of five burden-of-illness indices, such as depression and cognitive impairment (15). Whooley and colleagues found that depressive symptoms were a significant risk factor for cardiovascular and noncancer, noncardiovascular mortality but not cancer mortality in older women. Therefore, the association of depression and mortality in elderly persons is complex, probably involving many potential pathways. Nevertheless, few community studies have adequately adjusted for social variables, comorbidity, and functional status (22).
We therefore explored the association of depressive symptoms and mortality over 3 years in a large, racially diverse community sample of older adults, the Established Populations for Epidemiologic Studies of the Elderly project at Duke (also known as the Piedmont Health Survey of the Elderly) (23)(24). Over 4000 older adults were evaluated in 1986/87 for chronic illnesses and functional status and were followed for 10 years. We hypothesized that (i) depression (as assessed by the Center for Epidemiologic Studies Depression scale) is associated with 3-year mortality in this sample; (ii) mortality is also associated with age, male gender, lower body mass index (BMI), chronic disease, cognitive impairment, smoking history, impaired activities of daily living (ADL) (physical function impairment), impaired social support, and poor self-rated health; and (iii) the odds for mortality with depression moves toward unity as covariates known to be associated with both depression and mortality are controlled.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
The study design is presented elsewhere (23)(24). Briefly, a four-stage stratified household sampling design was used to select a probability sample of 4162 community residents 65 years of age or older. The age range at baseline was 65 to 105 years. Two thirds of the participants were women. To enhance statistical precision for estimating racial differences, African Americans were over-sampled and constituted 53.5% of the study respondents. Fewer than 15 of the persons interviewed were neither African American nor white and were categorized as white in the analyses. Of the respondents, nearly one half lived in the U.S. Census Bureau-defined rural area. All levels of socioeconomic status were represented among both African-American and white study subjects. The manner of sample selection allowed the development of weights that take into account race, gender, age group, number of older persons in the household, location of residence, and lack of response. The weights also allowed projection of the sample data to reflect the status of the same age population (approximately 28,000 in 1986) in the geographic region used as the sampling frame. This study was approved by the Duke University Medical Center Institutional Review Board and the Office of Management and Budget. Informed consent was obtained from each participant before data collection.
Data Collection
At baseline and again at each of the three in-person follow-up interviews, trained interviewers visited the homes of the study subjects and used comprehensive structured questionnaires to gather information. For participants too ill or unable to provide information, an appropriate proxy (the person most familiar with the participant) was asked to respond to a subset of the objective interview questions. Demographic factors were represented by dichotomous variables for age (65–74 vs 75 and older), race, gender, and education (
11 years vs <11 years). Self-reported morbidity data (not available from proxy interviews) included multiple aspects of the participants' physical, social, and mental function, including the Center for Epidemiologic Studies Depression Scale (CES-D) (26). All questions from the original CES-D were included verbatim; however, response options were collapsed to a yes/no format for reporting the presence or absence of a symptom during the week preceding the interview. The CES-D score determined by the usual method corresponds highly to our modified approach (24). Perceived social support was assessed by two questions: "In times of trouble can you count on at least some of your family or friends most of the time, some of the time, or hardly ever?" and "Can you talk about your deepest problems with at least some of your family or friends most of the time, some of the time, or hardly ever?" The responses were summed for an overall measure.
Cognitive status was determined by the Short Portable Mental Status Questionnaire (27). Health status was determined by an index of chronic illnesses (28). Perceived health was assessed by self-report (excellent, good, fair, or poor). Physical disability was assessed using items from a modified Katz ADL Scale and from the Rosow-Breslau physical health scale and dichotomized by no impairments versus one or more impairments for the Katz Scale and difficulty with two or more activities versus one or fewer activities on the Rosow-Breslau scale (29)(30). Smoking history was obtained by self-report of current and lifetime reports of pack years. BMI was calculated as weight (kg)/height (m)2.
Death was determined by information from a family member, other information source, and/or death certificate throughout the 10 years of the survey. For the first analyses, we divided deaths into died between baseline and P2, died between P2 and P3, and died between P3 and P4. The second analyses focused on death between baseline and P2. Complete data were available for 3664 (88%) of the baseline respondents. A total of 486 (13.3%) died between P1 and P2.
Statistical Analysis
All data were weighted to adjust for the sampling design and to allow inference to the five-county area. The analyses proceeded in three phases. In the first phase, the data were summarized by means of percentages for all covariates for the baseline interview. In the second phase, bivariate odds for mortality between baseline and P2, between P2 and P3, and between P3 and P4 were calculated. Each wave utilized all those present at the beginning of the period. Relative odds of mortality secondary to depression, uncontrolled and then controlled, were calculated. In the third phase, relative odds of mortality between baseline and P2 secondary to depression were calculated, controlling for progressively more control variables. Control variables were entered individually. The logistic regression models were run using SUDAAN (Research Triangle Institute, Research Triangle Park, NC) to adjust for design effects (31).
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
|
Table 2 presents the uncontrolled odds of mortality for the depressed subjects during three intervals of the survey (baseline to P2, P2 to P3, and P3 to P4). In uncontrolled analyses, subjects who were depressed at baseline were nearly twice as likely to be deceased 3 years later. Similarly, the odds of mortality at P3 for those depressed at P2 were 1.83, and the odds of mortality at P4 for those depressed at P3 were 1.76. In follow-up analyses controlled for age, gender, and race, this pattern persisted, as did the pattern when education, income, BMI, chronic health status, and smoking history were added to the model. Therefore, these analyses suggest that the risk for mortality among depressed elderly persons is about twofold over 3 to 4 years of follow-up regardless of whether depression is assessed at baseline, P2, or P3.
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
When progressively more inclusive models of the association of depression and mortality were developed, the relative odds of dying if depressed moved toward unity. Adding BMI, chronic diseases, and smoking history to the model decreased the relative odds to 1.74. When cognitive impairment was added, the odds decreased to 1.69. When Katz ADL impairment was added, the odds decreased to 1.45 with a further drop to 1.29 when Rosow-Breslau ADL impairment was added. Therefore, our three hypotheses were substantiated. In addition, the interaction of depression with these covariates on mortality was not significant in these analyses, suggesting that although depression works indirectly through these covariates to increase the risk of mortality, it does not act differentially across values of these covariates (e.g., depression is not a greater risk at high values of BMI compared with low values of BMI).
As in previous studies, when functional status was controlled in these analyses, the significant association between depression and mortality approached unity (9)(21). Nevertheless, the trend toward unity began by adding other variables known to be associated with mortality, such as smoking, chronic illness, and lower BMI. In contrast, the other risks for mortality did not move toward unity as additional variables were added to the models, with the exception of cognitive impairment. It is possible that depression may act on mortality through multiple relatively independent pathways. These pathways may lead to double feedback loops. For example, depression is known to increase the likelihood of functional impairment, and functional impairment, in turn, is a risk for depression (32)(33). Functional impairment is also known to be a predictor of mortality (21). A double feedback loop may explain this association. That is, depression affects ADL, ADL affects depression, and the influence of depression on mortality is effected partially through this double feedback loop. This is likely happening with other significant variables as well, such as smoking (depressed persons are more likely to smoke, and smokers are more likely to be depressed) and cognitive impairment (16).
The evidence presented in this study documents the association of depression and mortality, an association that appears to be mediated primarily through covariates. However, the evidence is indirect. Nevertheless, the conflicting findings from studies of depression and mortality in late life may be explained in part by the inferences from our findings and the findings of others (9)(21). Depression, as assessed by the CES-D, as an isolated construct may not directly increase the risk of mortality in late life but rather works through intermediate variables. These findings must be considered with caution, however. For example, the threshold of depression is set much lower using the CES-D than the threshold for major depression in the Diagnostic and Statistical Manual of Mental Disorders–Fourth Edition (34). In addition, the cases of depression identified are community based. Older adults experiencing major depression and seeking medical care from a psychiatrist may very well be at risk for increased mortality (such as mortality secondary to suicide) even when the control variables used in this study are taken into account.
|
Acknowledgments |
|---|
Received June 30, 2000
Accepted July 6, 2000
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  F. D. Wolinsky, H. W. Mahncke, M. W. V. Weg, R. Martin, F. W. Unverzagt, K. K. Ball, R. N. Jones, and S. L. Tennstedt The ACTIVE Cognitive Training Interventions and the Onset of and Recovery from Suspected Clinical Depression J Gerontol B Psychol Sci Soc Sci, September 1, 2009; 64B(5): 577 - 585. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. G. Blazer Depression in Late Life: Review and Commentary Focus, January 1, 2009; 7(1): 118 - 136. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. G. Blazer How Do You Feel About...? Health Outcomes in Late Life and Self-Perceptions of Health and Well-Being Gerontologist, August 1, 2008; 48(4): 415 - 422. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. C. Barry, H. G. Allore, Z. Guo, M. L. Bruce, and T. M. Gill Higher Burden of Depression Among Older Women: The Effect of Onset, Persistence, and Mortality Over Time Arch Gen Psychiatry, February 1, 2008; 65(2): 172 - 178. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Ryan, I. Carriere, K. Ritchie, R. Stewart, G. Toulemonde, J.-F. Dartigues, C. Tzourio, and M.-L. Ancelin Late-life depression and mortality: influence of gender and antidepressant use The British Journal of Psychiatry, January 1, 2008; 192(1): 12 - 18. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Rozzini, T. Sabatini, A. H. Ranhoff, and M. Trabucchi DEPRESSION IN OLDER MEDICAL INPATIENTS J. Gerontol. A Biol. Sci. Med. Sci., July 1, 2007; 62(7): 796 - 797. [Full Text] [PDF]
|
|
|
|
|
|
J. McCusker, M. Cole, A. Ciampi, E. Latimer, S. Windholz, and E. Belzile Does Depression in Older Medical Inpatients Predict Mortality? J. Gerontol. A Biol. Sci. Med. Sci., September 1, 2006; 61(9): 975 - 981. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. van den Brink, G. A. M. van den Bos, M. Tijhuis, S. Aijanseppa, A. Nissinen, S. Giampaoli, and D. Kromhout Hierarchy Levels, Sum Score, and Worsening of Disability Are Related to Depressive Symptoms in Elderly Men From Three European Countries J Aging Health, February 1, 2006; 18(1): 125 - 141. [PDF]
|
|
|
|
|
|
M. Ganguli, Y. Du, H. H. Dodge, G. G. Ratcliff, and C.-C. H. Chang Depressive Symptoms and Cognitive Decline in Late Life: A Prospective Epidemiological Study Arch Gen Psychiatry, February 1, 2006; 63(2): 153 - 160. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. E. Freeman, B. L. Egleston, S. K. West, K. Bandeen-Roche, and G. Rubin Visual Acuity Change and Mortality in Older Adults Invest. Ophthalmol. Vis. Sci., November 1, 2005; 46(11): 4040 - 4045. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Onder, B. W. J. H. Penninx, M. Cesari, S. Bandinelli, F. Lauretani, B. Bartali, A. M. Gori, M. Pahor, and L. Ferrucci Anemia Is Associated With Depression in Older Adults: Results From the InCHIANTI Study J. Gerontol. A Biol. Sci. Med. Sci., September 1, 2005; 60(9): 1168 - 1172. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Matthews, R. A. Nelesen, and J. E. Dimsdale Depressive Symptoms Are Associated With Increased Systemic Vascular Resistance to Stress Psychosom Med, July 1, 2005; 67(4): 509 - 513. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Gatz, S. L. Tyas, P. St. John, and P. Montgomery Do Depressive Symptoms Predict Alzheimer's Disease and Dementia? J. Gerontol. A Biol. Sci. Med. Sci., June 1, 2005; 60(6): 744 - 747. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Everson-Rose, J. S. House, and R. P. Mero Depressive Symptoms and Mortality Risk in a National Sample: Confounding Effects of Health Status Psychosom Med, November 1, 2004; 66(6): 823 - 830. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. S. Wilson, J. L. Bienias, C. F. Mendes de Leon, D. A. Evans, and D. A. Bennett Negative Affect and Mortality in Older Persons Am. J. Epidemiol., November 1, 2003; 158(9): 827 - 835. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley, J. H. Flaherty, and D. R. Thomas Editorial: Geriatricians, Continuous Quality Improvement, and Improved Care for Older Persons J. Gerontol. A Biol. Sci. Med. Sci., September 1, 2003; 58(9): M809 - 812. [Full Text] [PDF]
|
|
|
|
|
|
D. S. Charney, C. F. Reynolds III, L. Lewis, B. D. Lebowitz, T. Sunderland, G. S. Alexopoulos, D. G. Blazer, I. R. Katz, B. S. Meyers, P. A. Arean, et al. Depression and Bipolar Support Alliance Consensus Statement on the Unmet Needs in Diagnosis and Treatment of Mood Disorders in Late Life Arch Gen Psychiatry, July 1, 2003; 60(7): 664 - 672. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. G. Blazer Depression in Late Life: Review and Commentary J. Gerontol. A Biol. Sci. Med. Sci., March 1, 2003; 58(3): M249 - 265. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley Editorial: Hot Topics in Geriatrics J. Gerontol. A Biol. Sci. Med. Sci., January 1, 2003; 58(1): M30 - 36. [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley Editorial: Citations, Impact Factor, and the Journal J. Gerontol. A Biol. Sci. Med. Sci., December 1, 2002; 57(12): M765 - 769. [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley, H. M. Perry III, and D. K. Miller Editorial: Something About Frailty J. Gerontol. A Biol. Sci. Med. Sci., November 1, 2002; 57(11): M698 - 704. [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley and J. H. Flaherty Editorial It's Never Too Late: Health Promotion and Illness Prevention in Older Persons J. Gerontol. A Biol. Sci. Med. Sci., June 1, 2002; 57(6): M338 - 342. [Full Text]
|
|
|
|
|
|
N. Minicuci, S. Maggi, M. Pavan, G. Enzi, and G. Crepaldi Prevalence Rate and Correlates of Depressive Symptoms in Older Individuals: The Veneto Study J. Gerontol. A Biol. Sci. Med. Sci., March 1, 2002; 57(3): M155 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley Editorial: Drugs, Aging, and the Future J. Gerontol. A Biol. Sci. Med. Sci., January 1, 2002; 57(1): M2 - 6. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
