PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Arch Gen Psychiatry. Author manuscript; available in PMC 2013 October 1.
Published in final edited form as:
PMCID: PMC3646479
NIHMSID: NIHMS447711

20-Year Depressive Trajectories among Older Women

Amy L. Byers, PhD, MPH,1,2 Eric Vittinghoff, PhD, MPH,3 Li-Yung Lui, MS, MS,4 Tina Hoang, MPH,2 Dan G. Blazer, MD,5 Kenneth E. Covinsky, MD, MPH,2,6 Kristine E. Ensrud, MD, MPH,7 Jane A. Cauley, DrPH,8 Teresa A. Hillier, MD, MS,9 Lisa Fredman, PhD,10 and Kristine Yaffe, MD1,2,3,11

Abstract

Context

Despite the high occurrence of depressive symptoms in older adults, especially women, little is known about the long-term course of late-life depressive symptoms.

Objective

To characterize the natural course of depressive symptoms among elderly women followed for nearly 20 years, going from young old to oldest old.

Design

Using a latent class growth-curve analysis, we analyzed women enrolled in an ongoing prospective cohort study (1988–2009).

Setting

Clinic sites in Baltimore, MD, Minneapolis, MN, the Monongahela Valley near Pittsburgh, PA, and Portland, OR.

Participants

We studied 7240 community-dwelling women age 65 years or older.

Main Outcome Measure

The Geriatric Depression Scale (GDS) short form (range: 0–15) was used to assess depressive symptoms repeatedly over follow-up.

Results

We identified four latent classes over 20 years, comprising an expected 28% of women with minimal depressive symptoms, 54% with persistently low symptoms, 15% with increasing symptoms, and 3% with persistently high symptoms. In an adjusted model for latent class membership, odds ratios (ORs) for belonging in the increasing and persistently high depressive symptom classes, respectively, compared with minimal symptom group were substantially and significantly (P < .05) elevated for baseline smoking (ORs, 4.69 and 7.97), physical inactivity (ORs, 2.11 and 2.78), small social network (ORs, 3.24 and 6.75), physical impairment (ORs, 8.11 and 16.43), myocardial infarction (ORs, 2.09 and 2.41), diabetes (ORs, 2.98 and 3.03), and obesity (ORs, 1.86 and 2.90).

Conclusions

Over 20 years, approximately 20% of older women experienced persistently high or increasing depressive symptoms. In addition, these women had more comorbidities, physical impairment, and negative lifestyle factors at baseline. These associations support the need for intervention and prevention strategies to reduce depressive symptoms into oldest-old years.

INTRODUCTION

While studies suggest that as many as 27% of community-dwelling older adults (aged 65 years and older) experience depressive symptoms,1 little is known about the long-term course and heterogeneity of late-life depression. Given the projected expansion of the older population and the health and economic costs of depression,2, 3 greater understanding of the chronic course of depressive symptoms into oldest-old years and the typologies that describe this chronic course will inform intervention and prevention strategies.

Most prior studies have assessed depressive symptoms at only one time point.4 However, while depression peaks in young adulthood, it also tends to occur at a high and varying rate in late life.5 Because of the association of depression and morbidity, poor health outcomes and mortality,610 a persistent increase in depressive symptoms over time may have major health implications. Only a few studies have assessed trajectories of depressive symptoms in older participants and these have been limited by short follow-up.1116 In addition, since the oldest old (those age 80 years and older) are the fastest growing age group in the United States and a high incidence of depressive symptoms have been found among the oldest old,17 long term investigation of depressive symptoms is needed in this group.

The primary objective of our study was to characterize the natural course of depressive symptoms among older women followed for nearly 20 years into their ninth and tenth decades of life. A secondary objective was to examine if lifestyle, function, and comorbidity factors known to be associated with late-life depression differentially predict more severe depressive symptom trajectories.

METHODS

Participants

We studied 7240 elderly community-dwelling women from the Study of Osteoporotic Fractures (SOF), a prospective cohort study of 9704 women age 65 years and older, originally recruited between September 1986 and October 1988 from population-based listings in 4 areas of the United States: Baltimore, MD; Minneapolis, MN; the Monongahela Valley near Pittsburgh, PA; and Portland, OR. Women were excluded if they were unable to walk without help or had had a bilateral hip replacement. At years 2 (1988–1990), 6 (1992–1994), 10 (1996–1998), 16 (2002–2004), and 20 (2007–2009) the subjects were administered a depression scale. Our study included 7240 women who had at least two depression scale measures over the nearly 20 study years. The sample was 99.7% Caucasian, with mean (SD) age of 72.8 (4.7) years. During the study, 3654 women died and 622 ended the study early. The average number of depression measurements was 3.4 and the median length of follow up was 12.2 years. All women provided written informed consent, and the study was approved by the committees on human research at each site. In addition, current study analyses were approved by the institutional review board of the University of California, San Francisco.

Measures

Information on participants’ age, education, marital status and living arrangements was collected at baseline. The Geriatric Depression Scale (GDS) short form, a 15-item questionnaire with scores ranging from 0 to 15 assessing the number of depressive symptoms based on binary item responses (i.e., ‘yes’ or ‘no’), was used to determine the trajectories of depressive symptoms over follow-up. A final GDS score was determined for women with no more than 5 missing items (≤ 33%). This score was calculated by first taking the average of the non-missing items and then multiplying the average by 15. The GDS was initially administered during year 2 (defined as the baseline examination in this study) and repeated at follow-up years 6, 10, 16, and 20. It is a validated and reliable self-report scale used to detect depressive symptoms in older participants,18 as the GDS scale is structured to minimize the measurement of common but nonspecific factors in older adults (e.g., fatigue, sleep disturbance, and poor concentration).19 A GDS cut-off of ≥ 6 indicates depression.

Lifestyle factors were measured at baseline and included current smoking, number of drinks of alcohol, physical activity, and social networks. Alcohol use was assessed as the current number of drinks consumed per week with ≥ 7 drinks/week defined as frequent consumption.20 Physical inactivity was defined as no low-, moderate-, or high-intensity physical activities as measured with a modified Paffenbarger scale.21, 22 The Lubben Social Network Scale (LSNS),23 a validated 10-item self-report inventory assessing type of relationships and size and frequency of contact, was used to determine an averaged social network score for each participant ranging from 0 to 5 (higher scores equated to larger networks). We defined participants as having a small social network if they fell below the median of the averaged LSNS (< 3.3). A similar cutoff defining small networks has been used in other SOF studies.24

Physical function was assessed at baseline for each of 5 activities (i.e., walking 2–3 blocks, climbing 10 steps without resting, preparing meals, shopping, and doing heavy housework). Global cognitive function was assessed by the modified version of the Mini-Mental State Examination (mMMSE)25 with a maximum score of 26. Scores of less than 23 indicated cognitive impairment (equivalent to at least 1.5 standard deviations below the mean).22 At baseline, medical history was obtained including myocardial infarction (MI), stroke, diabetes, and breast cancer and clinical measures of hypertension and obesity (body mass index ≥ 30 kg/m2). Finally, participants were asked about current use (within the past 30 days) of medications including antidepressants; reports of current use were checked by examining the labels of drugs.

Statistical Analyses

Latent class growth-curve analysis (LCGA), as implemented in the Proc Traj procedure in SAS version 9.1.3 (SAS Institute Inc., Cary, NC),26 was used to estimate mean trajectories of the GDS scores as a function of current age at each visit. In contrast to standard growth-curve analysis, LCGA estimates mean trajectories for two or more unobserved or latent classes, in combination with probability of membership in each latent class for every observation in the sample. Using Proc Traj, the repeated GDS scores were modeled as censored normal. In addition to demographic variables, included for face validity, the multivariable Proc Traj sub-model used to estimate probabilities of latent class membership included covariates that were statistically significant (P < .05) in bivariate analyses. To describe bivariate associations, we estimated prevalence of baseline characteristics within latent classes and assessed differences in characteristics across latent classes using weighted repeated measures analyses. In these analyses, each participant contributed to each latent class in proportion to her predicted probability of class membership, based on the Proc Traj results. Robust standard errors were used to account for lack of independence of the weighted data across the four groups.

We hypothesized a priori that there would be five latent classes, four closely resembling the classes estimated by Proc Traj in the four-class model we selected, and an additional class with declining GDS scores. However, a five-class model did not include a declining trajectory, instead splitting one of the four other hypothesized classes into two similar subgroups. Moreover, the Bayesian Information Criteria (BIC), conventionally used to determine the number of latent classes, pointed to a large number of barely distinguishable classes. Thus, we selected the four-class model as a parsimonious description of the sample in best accord with our prior hypothesis. Statistical tests for models were two-tailed with P < .05 defining statistical significance.

RESULTS

Figure 1 presents the four latent class GDS mean trajectories estimated by the selected LCGA model, plotted by current age at each visit. The predicted probabilities of group membership totaled 28% with minimal depressive symptoms, 54% with persistently low symptoms, 15% with increasing symptoms, and 3% with persistently high symptoms.

Figure 1
Mean trajectories of depressive symptoms by increasing age among older women

Bivariate Analyses

Table 1, and Figures 2a, 2b, and 2c show the results for our weighted bivariate analyses characterizing the four latent classes. Women with the highest versus lowest depression trajectory over time were slightly younger at baseline (71.9 vs 73.1 years) and had less years of education (11.8 vs 13.1 years) (Table 1; P < .01 for trend). Current use of antidepressants was particularly low for all groups, but increased from respondents with minimal to persistently high symptoms (i.e., 0.6%, 1.5%, 3.5%, and 7.1%, respectively); ever use of antidepressants over the entire study period included 12% of women with minimal symptoms, 21% with persistently low symptoms, 31% with increasing symptoms, and 41% with persistently high symptoms.

Figure 2Figure 2Figure 2
a. Prevalence of comorbidities across trajectory groups of depressive symptoms
Table 1
Baseline Demographic Characteristics of 7240 Older Women According to Depressive Symptom Trajectories

We found that the highest disease burden was in the group with persistently high depressive symptoms (Figure 2a; MI = 13.6%, Stroke = 6.5%, Diabetes = 14.2%, Hypertension = 43.2%, and Obesity = 32.5%) and the lowest disease burden was in the minimal symptom group (MI = 3.8%, Stroke = 1.5%, Diabetes = 3.4%, Hypertension = 32.0%, and Obesity = 13.6%) (all P < .001 for trend). We also examined the prevalence of breast cancer, and in going from minimal to high symptoms the trend was statistically significant but less remarkable (i.e., 4 to 7%; P = .01 for trend). For “negative” lifestyle factors, prevalence of smoking, physical inactivity, and small social network increased with severity of trajectory (Figure 2b; P < .001 for trend), with physical inactivity and small social network having the largest differences (persistently high vs minimal symptoms: 34.5% vs 12% and 70.5% vs 41.3%, respectively). Frequent alcohol consumption (≥ 7 drinks/week) appeared to have a reverse relationship, where the persistently high group had the lowest prevalence (8.1%) compared with the other groups (10–13%) (P = .02 for trend). In addition, physical and cognitive impairment increased from minimal to persistently high groups (Figure 2c; P < .001 for trend).

Multivariable Model

For the multivariable model, the strongest association was seen with physical impairment, where the odds of increasing and persistently high symptoms over 20 years was eight times (odds ratio [OR], 8.11; 95% confidence interval [CI], 5.98–11.02) and sixteen times (OR, 16.43; 95% CI, 10.29–26.23) higher for those with any physical (IADL) difficulty compared to those with no physical impairment (Table 2). In addition, “negative” lifestyle factors were strong predictors of increasing and persistently high trajectories. Those who smoked were 5- and 8-fold more likely to have increasing and persistently high symptoms (ORs, 4.69 and 7.97; 95% CIs, 2.96–7.43 and 4.37–14.54, respectively). Women who were physically inactive were over twice as likely to have increasing symptoms and almost three times more likely to have persistently high symptoms than women who were active (ORs, 2.11 and 2.78; 95% CIs, 1.52–2.93 and 1.77–4.36, respectively). Women who had a small social network had nearly a 7-fold increased odds of persistently high symptoms (OR, 6.75; 95% CI, 4.33–10.53).

Table 2
Multivariable Model: Factors Independently Associated with Trajectories of Depressive Symptoms Over 20 Years Compared with Minimal Symptomsa

We found a 2 to 5-fold increased odds of persistently high trajectory of depressive symptoms for those women who had myocardial infarction (OR, 2.41; 95% CI, 1.18–4.90), stroke (OR, 5.12; 95% CI, 1.63–16.08), diabetes (OR, 3.03; 95% CI, 1.49–6.14), obesity (OR, 2.90; 95% CI, 1.88–4.48), and breast cancer (OR, 2.51; 95% CI, 1.24–5.09) (Table 2). Myocardial infarction, diabetes, and obesity were also significantly (P < .05) and independently associated with increasing symptoms. However, stroke and breast cancer were only associated with persistently high symptoms, while hypertension was only significantly associated with increasing symptoms (OR, 1.63; 95% CI, 1.29–2.07), independent of other covariates.

COMMENT

We identified four trajectories of depressive symptoms among older women followed for nearly 20 years (minimal depressive symptoms, persistently low symptoms, increasing symptoms, and persistently high symptoms). The prominent characteristics describing increasing and persistently high depressive trajectories compared with minimal symptoms were negative life style factors (i.e., smoking, physical inactivity, and small social network), physical impairment, and comorbidities (i.e., myocardial infarction, diabetes, and obesity). This study confirms that high and increasing depressive symptoms occur with concerning frequency over the course of elderly years with survival to oldest old having increased symptom burden.

Few prior studies have considered the heterogeneity and natural course of depressive symptomatology in older adults. Our findings are supported by a 6-year study from the Longitudinal Aging Study Amsterdam in which four depression trajectory patterns were identified: remission, remission plus recurrence, chronic-intermittent, and chronic.27 By utilizing a latent class cluster approach, we were able to characterize individuals by subtypes; identifying trajectories that may more realistically reflect the long-term course of late-life depression. Most previous latent class analyses describing heterogeneity in depressive symptoms have examined one point in time,2833 while the few studies that have investigated trajectories over multiple years have used one or two years of data,14, 15 used a one-item question to assess depression,12 or had limited sample size to accurately assess follow up.16

Two studies that assessed 10 and 12 years of depressive symptoms had inconsistent results. Our findings provide evidence supporting the 10-year study of 3922 older community-dwelling adults aged 60 years and older, where the study identified four distinct trajectories (i.e., persistent low, persistent mild, late peak, and high-chronic).13 In contrast, the 12-year study identified six latent trajectories in 1260 community-dwelling adults aged 65 years and older.11 Unlike our results, this study found that two of the trajectories started high and declined with time. Finally, unlike this previous research, we had a very large sample size (over 7000 women) which provided information on a substantial number of oldest-old women, despite attrition due to dropouts and death.

Given our large sample size, we were able to study a comprehensive list of covariates known to be associated with late-life depression as potential predictors of high and increasing depressive trajectories. Studies have consistently found that various medical conditions, lifestyle factors, and functional impairment are associated with late-life depression or increased depressive symptoms;4, 6, 3437 however, most of these studies are either cross-sectional or describe factors related to a single mean trajectory of depression or depressive symptoms over time. In contrast, we described multiple trajectories and generated a set of characteristics, in combination, that best predict belonging to more severe depressive trajectories over the long term. To this end, we determined that physical impairment, myocardial infarction, diabetes, obesity, smoking, physical inactivity, and small social networks differentially predict high and increasing trajectories over two decades, going from young old to oldest old. Although some previous latent trajectory studies have examined the association between depressive trajectories and related factors, results have been mixed with most not considering these and other important covariates in combination.11, 12, 14, 16

The identification of four distinct trajectories of depressive symptoms over the course of 20 years has important implications. First, these trajectories provide new information concerning the heterogeneity of late-life depression. Second, the findings suggest that if left unresolved the natural course of depressive symptoms in older women will become more burdensome, where approximately 20% of women are expected to have high or increasing symptoms over the long term. Since the trajectories were determined by the latent class model, there were no specific cutoffs assigned to the groups. However, the upper two trajectory groups (increasing and persistently high symptoms) suggest a trend toward clinically significant depression (i.e., GDS ≥ 6).

Third, by identifying important predictors of these trajectories, we are able to determine risk factors that may be targets for prevention and intervention. In particular, our study suggests that modifiable risk factors, such as increased physical activity or exercise, increased social engagement or connectedness, or reduction of obesity may help to reduce the progression of symptom burden. Although an association between chronic conditions such as myocardial infarction and diabetes and risk of depression has been supported by prior research,34, 38, 39 few studies have considered the importance of lifestyle factors as predictors of depression above and beyond morbidity. In our study, lifestyle factors (i.e., smoking, physical inactivity, and small social networks) had an impressive impact on increasing and persistently high depressive symptoms. Future studies on the impact of changes in lifestyle factors associated with changes in depressive symptoms are needed in order to target lifestyle factors for prevention and intervention of late-life depression. In general, research investigating positive lifestyle factors as a modality for combating depression or depressive symptoms in older adults has been modest.40 Furthermore, current antidepressant use was low in our study and ‘ever use’ was less than 50% for those in the highest trajectory group. Although the literature has shown that the use of antidepressants has increased since our study baseline (1988),41 our results suggest greater need for early recognition and treatment of depression symptoms in late life. Moreover, our results have implications for increasing the use of maintenance regimens of antidepressant pharmacotherapy in older adults in order to reduce depressive symptoms over the long term. Such maintenance treatment is also strongly supported by the maintenance trials from Pittsburgh and elsewhere.42, 43 In summary, investigation into combating symptom burden through managing cardiovascular conditions, improving physical function, and promoting positive lifestyle factors either alone or in combination with antidepressant use is suggested by our findings.

Of note, our results support evidence linked to the ‘vascular depression hypothesis,’44 showing that vascular disorders and vascular risk factors may precipitate and perpetuate the severity and chronicity of depression into late life. Because the risk of vascular disease increases in post-menopausal women,45 our findings may suggest an increased risk of late onset depression or change in the course of pre-menopause depression after the onset of vascular disease. This implicates medications besides or in addition to antidepressants (e.g., drugs used for prevention and treatment of cerebrovascular disease), as well as lifestyle changes, to prevent and intervene on vascular depression in older women.

The strengths of this study include a large sample of community-dwelling older women, information on depressive symptoms over nearly two decades, and carefully measured potential predictors. To our knowledge, our study is the first to characterize the natural course of depressive symptoms over two decades in elderly women—the group at greatest risk of living with high rates of late-life depression and related illness and disability.46, 47

There are some limitations of this study. Our sample was restricted to women. Although women are at much higher risk of developing depressive disorders and comprise a larger proportion of the oldest old, gender differences in trajectories of depressive symptoms may be important, especially considering potentially different targets of prevention and intervention. One prior study found that depressive trajectories in men were affected by perceived health and disability, while trajectories in women were influenced by disability and social support.16 In addition, the results of this study are not generalizable to nonwhite older adults or less healthy women. Moreover, we did not have a complete set of covariates available to investigate other potentially important baseline predictors of high and emerging depressive symptoms. This included incomplete information on socioeconomic status (e.g., income), sleep disturbance or sleep disorders, and medical history of cancer. Although the SOF study collected information on breast cancer at baseline, information specific to other cancers that have been found to be associated with geriatric depression, such as colon cancer,48 was not available. Also, we had limited information on depression treatment and, therefore, we were unable to investigate the impact of treatment on changing trajectories. For example, we did not have information available on when the subjects initiated antidepressant treatment (only that they were taking antidepressants in the prior 30 days from the interview date) nor did we have information available on their use of non-pharmacological treatment for depression. Furthermore, we did not have information on prior history of depression and, thus, could not determine whether the baseline covariates were consequences of previous depression. Finally, because data on the GDS was collected at visits that were approximately 4 years apart, variation of depressive symptomatology during these gaps in time is unknown.

Understanding the long term course of depressive symptoms among those who live into their ninth and tenth decades of life is imperative. The results of this study help to describe these patterns by identifying four trajectories of depressive symptomatology; highlighting the persistence of symptoms in the oldest old. Although we determined key predictors of more severe depressive trajectories (i.e., comorbidities, physical impairment, and negative lifestyle factors at baseline), further investigation into combating increased burden over the long term is needed. In addition, these associations may be bidirectional, which suggests further investigation in longitudinal analyses. It is important that future research tests the bidirectionality of such associations as physical/cognitive impairment and depressive symptoms, elucidating the etiological relationship and where to intervene. Furthermore, future research needs to examine other important covariates not accounted for in the current study (e.g., socioeconomic gradient) and investigate whether associations differ by gender and racial/ethnic diversity. Finally, because the number of trajectory classes selected was driven by prior hypothesis, our findings need to be carefully interpreted and confirmed by future work.

Given the projected expansion of the elderly population, increased life expectancy, and the health and economic costs of depression, the potential public health burden of late-life depressive disorders implicated by this study are concerning. These findings emphasize the importance of improving the recognition, monitoring and treatment of depressive symptoms over the long term, and of developing better interventions to forestall their development.

ACKNOWLEDGMENTS

Funding/Support and Role of Sponsor: This work was supported by a National Institute of Mental Health grant (R01 MH086498; PI: Dr. Yaffe), which was administered by the Northern California Institute for Research and Education, and with resources of the Veterans Affairs Medical Center, San Francisco, California. Dr. Byers is supported by a K01 Career Development Award (MH079093) from the National Institute of Mental Health. Dr. Yaffe is supported, in part, by a K24 Midcareer Investigator Award (AG031155) from the National Institute on Aging. The Study of Osteoporotic Fractures (SOF) was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases and the National Institute on Aging (Public Health Service grants: 2 R01 AG027574-22A1, R01 AG005407, R01 AG027576-22, 2 R01 AG005394-22A1, AG05407, AG05394, AR35582, AR35583, AR35584, AG026720, R01 AG18037, and R01 AG028144-01A1). The sponsors had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or the preparation, review, or approval of the manuscript.

Disclaimer: We acknowledge that the original collector of the data, sponsoring organizations, agencies, or the U.S. government bear no responsibility for use of the data or for interpretations or inferences based upon such uses. The views and opinions expressed in this report are those of the authors and should not be construed otherwise.

Footnotes

Portions of this manuscript were presented at the American Association for Geriatric Psychiatry’s 2011 Annual Meeting, San Antonio, Texas, March 18–21, 2011 and the Gerontological Society of America’s 2011 Annual Meeting, Boston, Massachusetts, November 18–22, 2011.

Author Contributions: Dr. Vittinghoff and Ms. Lui conducted the statistical analyses for this study. Dr. Byers had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Conflicts of interest: The authors have no competing interests, including specific financial interests or relationships or affiliations relevant to the subject of this manuscript.

REFERENCES

1. Koenig HG, Blazer DG. Epidemiology of geriatric affective disorders. Clin Geriatr Med. 1992;8(2):235–251. [PubMed]
2. U.S. Census Bureau. An Older and More Diverse Nation by Midcentury. [Accessed February 22, 2011];U.S. Census Bureau Newsroom. http://www.census.gov/newsroom/releases/archives/population/cb08-123html Released August 14, 2008.
3. Cassano P, Fava M. Depression and public health: an overview. J Psychosom Res. 2002;53(4):849–857. [PubMed]
4. Hybels CF, Blazer DG. Epidemiology of late-life mental disorders. Clin Geriatr Med. 2003;19(4):663–696. [PubMed]
5. Byers AL, Yaffe K, Covinsky KE, Friedman MB, Bruce ML. High occurrence of mood and anxiety disorders among older adults: the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2010;67(5):489–496. [PMC free article] [PubMed]
6. Blazer DG, Hybels CF. Origins of depression in later life. Psychol Med. 2005;35(9):1241–1252. [PubMed]
7. Walke LM, Byers AL, Tinetti ME, Dubin JA, McCorkle R, Fried TR. Range and severity of symptoms over time among older adults with chronic obstructive pulmonary disease and heart failure. Arch Intern Med. 2007;167(22):2503–2508. [PMC free article] [PubMed]
8. Bruce ML. Depression and disability in late life: directions for future research. Am J Geriatr Psychiatry. 2001;9(2):102–112. [PubMed]
9. Yaffe K, Edwards ER, Covinsky KE, Lui LY, Eng C. Depressive symptoms and risk of mortality in frail, community-living elderly persons. Am J Geriatr Psychiatry. 2003;11(5):561–567. [PubMed]
10. Lenze EJ, Rogers JC, Martire LM, Mulsant BH, Rollman BL, Dew MA, Schulz R, Reynolds CF. The association of late-life depression and anxiety with physical disability: a review of the literature and prospectus for future research. Am J Geriatr Psychiatry. 2001;9(2):113–135. [PubMed]
11. Andreescu C, Chang CC, Mulsant BH, Ganguli M. Twelve-year depressive symptom trajectories and their predictors in a community sample of older adults. Int Psychogeriatr. 2008;20(2):221–236. [PMC free article] [PubMed]
12. Hong SI, Hasche L, Bowland S. Structural relationships between social activities and longitudinal trajectories of depression among older adults. Gerontologist. 2009;49(1):1–11. [PubMed]
13. Kuo SY, Lin KM, Chen CY, Chuang YL, Chen WJ. Depression trajectories and obesity among the elderly in Taiwan. Psychol Med. 2011;41(8):1665–1676. [PubMed]
14. Cui X, Lyness JM, Tang W, Tu X, Conwell Y. Outcomes and predictors of late-life depression trajectories in older primary care patients. Am J Geriatr Psychiatry. 2008;16(5):406–415. [PubMed]
15. Doyle F, McGee H, Delaney M, Motterlini N, Conroy R. Depressive vulnerabilities predict depression status and trajectories of depression over 1 year in persons with acute coronary syndrome. Gen Hosp Psychiatry. 2011;33(3):224–231. [PubMed]
16. Huang JF, Wong RH, Chen CC, Mao IF, Huang CC, Chang WH, Wang L. Trajectory of depression symptoms and related factors in later life--a population based study. J Affect Disord. 2011;133(3):499–508. [PubMed]
17. Blazer DG. Psychiatry and the oldest old. Am J Psychiatry. 2000;157(12):1915–1924. [PubMed]
18. Sheikh J, Yesavage J. Geriatric Depression Scale: recent evidence and development of a shorter version. Clin Gerontol. 1986;5:165–173.
19. Yesavage JA. Geriatric Depression Scale. Psychopharmacol Bull. 1988;24(4):709–711. [PubMed]
20. Bartels SJ, Coakley E, Oxman TE, Constantino G, Oslin D, Chen H, Zubritsky C, Cheal K, Durai UN, Gallo JJ, Llorente M, Sanchez H. Suicidal and death ideation in older primary care patients with depression, anxiety, and at-risk alcohol use. Am J Geriatr Psychiatry. 2002;10(4):417–427. [PubMed]
21. Paffenbarger RS, Jr, Wing AL, Hyde RT. Physical activity as an index of heart attack risk in college alumni. Am J Epidemiol. 1978;108(3):161–175. [PubMed]
22. Middleton LE, Barnes DE, Lui LY, Yaffe K. Physical activity over the life course and its association with cognitive performance and impairment in old age. J Am Geriatr Soc. 2010;58(7):1322–1326. [PMC free article] [PubMed]
23. Lubben JE. Assessing social networks among elderly populations. Fam Community Health. 1988;11:42–52.
24. Rutledge T, Matthews K, Lui LY, Stone KL, Cauley JA. Social networks and marital status predict mortality in older women: prospective evidence from the Study of Osteoporotic Fractures (SOF) Psychosom Med. 2003;65(4):688–694. [PubMed]
25. Teng EL, Chui HC. The modified Mini-Mental State (3MS) Examination. J Clin Psychiatry. 1987;48(8):314–318. [PubMed]
26. Jones BL, Nagin DS, Roeder K. A SAS procedure based on mixture models for estimating developmental trajectories. Sociological Methods Research. 2001;29:374–393.
27. Beekman AT, Geerlings SW, Deeg DJ, Smit JH, Schoevers RS, de Beurs E, Braam AW, Penninx BW, van Tilburg W. The natural history of late-life depression: a 6-year prospective study in the community. Arch Gen Psychiatry. 2002;59(7):605–611. [PubMed]
28. Hybels CF, Blazer DG, Pieper CF, Landerman LR, Steffens DC. Profiles of depressive symptoms in older adults diagnosed with major depression: latent cluster analysis. Am J Geriatr Psychiatry. 2009;17(5):387–396. [PMC free article] [PubMed]
29. Bogner HR, Richie MB, de Vries HF, Morales KH. Depression, cognition, apolipoprotein e genotype: latent class approach to identifying subtype. Am J Geriatr Psychiatry. 2009;17(4):344–352. [PMC free article] [PubMed]
30. Eaton WW, Dryman A, Sorenson A, McCutcheon A. DSM-III major depressive disorder in the community. A latent class analysis of data from the NIMH epidemiologic catchment area programme. Br J Psychiatry. 1989;155:48–54. [PubMed]
31. Sullivan PF, Kessler RC, Kendler KS. Latent class analysis of lifetime depressive symptoms in the national comorbidity survey. Am J Psychiatry. 1998;155(10):1398–1406. [PubMed]
32. Carragher N, Adamson G, Bunting B, McCann S. Subtypes of depression in a nationally representative sample. J Affect Disord. 2009;113(1–2):88–99. [PubMed]
33. Chen L, Eaton WW, Gallo JJ, Nestadt G. Understanding the heterogeneity of depression through the triad of symptoms, course and risk factors: a longitudinal, population-based study. J Affect Disord. 2000;59(1):1–11. [PubMed]
34. Blazer DG. Depression in late life: review and commentary. J Gerontol A Biol Sci Med Sci. 2003;58(3):249–265. [PubMed]
35. Krishnan KR, Delong M, Kraemer H, Carney R, Spiegel D, Gordon C, McDonald W, Dew M, Alexopoulos G, Buckwalter K, Cohen PD, Evans D, Kaufmann PG, Olin J, Otey E, Wainscott C. Comobidity of depression with other medical diseases in the elderly. Biol Psychiatry. 2002;52(6):559–588. [PubMed]
36. Pan A, Sun Q, Czernichow S, Kivimaki M, Okereke OI, Lucas M, Manson JE, Ascherio A, Hu FB. Bidirectional association between depression and obesity in middle-aged and older women. Int J Obes (Lond) 2012;36(4):595–602. [PMC free article] [PubMed]
37. Almeida OP, Alfonso H, Pirkis J, Kerse N, Sim M, Flicker L, Snowdon J, Draper B, Byrne G, Goldney R, Lautenschlager NT, Stocks N, Scazufca M, Huisman M, Araya R, Pfaff J. A practical approach to assess depression risk and to guide risk reduction strategies in later life. Int Psychogeriatr. 2011;23(2):280–291. [PubMed]
38. Frasure-Smith N, Lesperance F, Juneau M, Talajic M, Bourassa M. Gender, depression, and one-year prognosis after myocardial infarction. Psychosom Med. 1999;61(1):26–37. [PubMed]
39. Anderson RJ, Freedland KE, Clouse RE, Lustman PJ. The prevalence of comorbid depression in adults with diabetes: a meta-analysis. Diabetes Care. 2001;24(6):1069–1078. [PubMed]
40. Teri L, McCurry SM, Logsdon RG, Gibbons LE, Buchner DM, Larson EB. A randomized controlled clinical trial of the Seattle Protocol for Activity in older adults. J Am Geriatr Soc. 2011;59(7):1188–1196. [PMC free article] [PubMed]
41. Blazer DG, Hybels CF, Simonsick EM, Hanlon JT. Marked differences in antidepressant use by race in an elderly community sample: 1986–1996. Am J Psychiatry. 2000;157(7):1089–1094. [PubMed]
42. Reynolds CF, 3rd, Frank E, Perel JM, Mazumdar S, Kupfer DJ. Maintenance therapies for late-life recurrent major depression: research and review circa 1995. Int Psychogeriatr. 1995;7(Suppl):27–39. [PubMed]
43. Reynolds CF, 3rd, Dew MA, Pollock BG, Mulsant BH, Frank E, Miller MD, Houck PR, Mazumdar S, Butters MA, Stack JA, Schlernitzauer MA, Whyte EM, Gildengers A, Karp J, Lenze E, Szanto K, Bensasi S, Kupfer DJ. Maintenance treatment of major depression in old age. N Engl J Med. 2006;354(11):1130–1138. [PubMed]
44. Alexopoulos GS, Meyers BS, Young RC, Campbell S, Silbersweig D, Charlson M. The ‘vascular depression’ hypothesis. Arch Gen Psychiatry. 1997;54:915–922. [PubMed]
45. Mosca L, Benjamin EJ, Berra K, Bezanson JL, Dolor RJ, Lloyd-Jones DM, Newby LK, Piña IL, Roger VL, Shaw LJ, Zhao D, Beckie TM, Bushnell C, D'Armiento J, Kris-Etherton PM, Fang J, Ganiats TG, Gomes AS, Gracia CR, Haan CK, Jackson EA, Judelson DR, Kelepouris E, Lavie CJ, Moore A, Nussmeier NA, Ofili E, Oparil S, Ouyang P, Pinn VW, Sherif K, Smith SC, Jr, Sopko G, Chandra-Strobos N, Urbina EM, Vaccarino V, Wenger NK. Effectiveness-based guidelines for the prevention of cardiovascular disease in women--2011 update: a guideline from the American Heart Association. Circulation. 2011;123(11):1243–1262. [PMC free article] [PubMed]
46. Barry LC, Allore HG, Guo Z, Bruce ML, Gill TM. Higher burden of depression among older women: the effect of onset, persistence, and mortality over time. Arch Gen Psychiatry. 2008;65(2):172–178. [PMC free article] [PubMed]
47. Barry LC, Allore HG, Bruce ML, Gill TM. Longitudinal association between depressive symptoms and disability burden among older persons. J Gerontol A Biol Sci Med Sci. 2009;64(12):1325–1332. [PMC free article] [PubMed]
48. Mohile SG, Fan L, Reeve E, Jean-Pierre P, Mustian K, Peppone L, Janelsins M, Morrow G, Hall W, Dale W. Association of cancer with geriatric syndromes in older Medicare beneficiaries. J Clin Oncol. 2011;29(11):1458–1464. [PMC free article] [PubMed]