| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Health Services Research and Development (L.S.K., K.A.B., M.B.M., C.L.B.), Veterans Affairs Puget Sound Health Care System, Seattle, Washington; Department of Health Services (L.S.K., K.A.B.), University of Washington, Seattle, Washington; Department of Medicine (K.A.B., C.L.B.), University of Washington, Seattle, Washington; Department of Psychiatry and Behavioral Sciences (W.J.K., E.L.), University of Washington, Seattle, Washington; Center for Health Studies (E.L.), Group Health Cooperative of Puget Sound, Seattle, Washington; VA Center of Excellence for Substance Abuse Treatment and Education (K.A.B.), Seattle, Washington; Primary and Specialty Medical Care (K.A.B.), VA Medical Center, Seattle, Washington.
Address correspondence and reprint requests to Leslie S. Kinder, 8506 SE 72nd Street, Mercer Island, WA 98040. E-mail: lskinder{at}stanfordalumni.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: Patients from seven Department of Veterans Affairs medical centers completed mailed questionnaires. Depression and PTSD status were determined from patient self-report of a prior diagnosis and/or electronic administrative data. Date of death was ascertained from Veterans Health Information Systems and Technology Architecture and the Department of Veterans Affairs' Beneficiary Identification and Records Locator System.
Results: Among 35,715 primary care patients, those with a history of depression without a history of PTSD (n = 6876) were at increased risk of death over an average of 2 years compared with patients with neither depression nor PTSD after adjustment for demographic variables, health behaviors, and medical comorbidity (hazard ratio (HR) = 1.17; 95% Confidence Interval (CI) = 1.06–1.28). However, patients with a history of PTSD without a history of depression (n = 748) were not at increased risk of death compared with patients with neither depression nor PTSD (HR = 0.84; 95% CI = 0.63–1.13). Patients with a history of both (n = 3762) were at increased risk of death after adjustment for demographic factors, although not after additional adjustment for health behaviors and medical comorbidity (HR = 0.90; 95% CI = 0.78–1.04).
Conclusions: In a large sample of veterans, a prior diagnosis of depression, but not PTSD, was associated with an increased risk of death over an average of 2 years after adjusting for age, demographic variables, health behaviors, and medical comorbidity.
Key Words: depression • posttraumatic stress disorder • mortality • veterans
Abbreviations: PTSD = posttraumatic stress disorder; VA= Department of Veterans Affairs; ACQUIP = Ambulatory Care Quality Improvement Project; MHI-5 = Mental Health Inventory; CHF = congestive heart failure; MI = heart attack/myocardial infarction.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Both MDD and PTSD are common responses to traumatic events and frequently co-occur. In addition, prolonged PTSD may predispose individuals to the development of MDD, and at the same time, a history of MDD may predispose an individual to the development of PTSD after a traumatic event (7). Further, MDD and PTSD share many features, such as sleep disturbance, decreased concentration, anhedonia, and withdrawal, making differentiation difficult. Previous findings suggest that MDD may be present in close to 50% and dysthymia in close to 20% of patients with PTSD among the US population (7,8), and symptoms of depression may be present in close to 80% of those who screen positive for PTSD among VA outpatients (6,9).
Little is known about the long-term health consequences of MDD, PTSD, and their co-occurrence in veterans. Studies conducted in community and patient samples have generally supported a relationship between depression and the subsequent development of medical comorbidity and mortality, although some experts have been concerned that these findings reflect inadequate adjustment for baseline health status (10–14). PTSD has also been linked to medical comorbidity and mortality (15); however, the role that baseline health status, and specifically depression, may play in the relationship between PTSD and medical outcomes has not been established. No studies which have examined the relationship between PTSD and mortality have reported the prevalence of comorbid depression, nor the association between PTSD and mortality independent of depression. In this report, we examined whether a diagnosis of depression alone, a diagnosis of PTSD alone, or a diagnosis of both depression and PTSD, as determined from electronic administrative data and/or self-report, were associated with all-cause mortality in VA primary care patients over an average of 2 years. We conducted analyses adjusting for demographic variables, health behaviors, and medical comorbidity. We hypothesized that depression and PTSD would each be independently associated with mortality, and that patients with both depression and PTSD would be at greatest risk of death.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Measures
History of Depression and PTSD
Depression status was determined from electronic administrative data or patient self-report of a prior diagnosis. International Classification of Diseases, Ninth Revision (ICD-9) codes associated with inpatient and outpatient visits at any of the seven VA facilities were obtained beginning October 1, 1995. We included any inpatient and outpatient visits with codes for depression or dysthymia (296.2, 296.3, 300.4, 311) that occurred between October 1, 1995 and study entry to identify diagnoses of depression from the administrative record. Patients also completed the ACQUIP Health Checklist at study entry, which included a list of 24 common medical conditions for which patients were asked to indicate if their doctor or nurse had ever told them that they had any of the problems listed, including "depression or other emotional troubles." The self-report question added additional cases to those reflected by ICD-9 codes alone; it captured those patients with diagnoses and/or treatment that may have preceded patient record abstraction, patients who remained untreated, or patients treated outside the VA system, in addition to those in current treatment. Given the problems with the underdiagnosis of and failure to treat depression (18), a composite measure of patient records and/or self-report was selected as the most inclusive categorization of patients with a history of depression possible with the data available.
As with depression, PTSD status was determined from electronic administrative records or patient self-report of a prior diagnosis. The composite measure included any inpatient or outpatient visits with ICD-9 codes for PTSD (309.81) that occurred between October 1, 1995 and study entry and/or endorsement on the Health Checklist that their doctor or nurse had ever told them that they had "PTSD (posttraumatic stress)." For analyses, patients were divided into four mutually exclusive groups: those with no history of depression or PTSD, those with a history of depression alone, those with a history of PTSD alone, and those with a history of both depression and PTSD.
Mortality
Vital status and date of death were available through June 30, 2000 in the ACQUIP data repository. Death was determined by 1) weekly extraction of Veterans Health Information Systems and Technology Architecture (VISTA), and 2) VA's Beneficiary Identification and Records Locator System, which records dates of death for veterans whose families file for the veteran's death benefit but does not include cause of death. A combination of these sources has been found to accurately identify 97.9% of deaths among VA patients (negative predictive value 97%) (19).
Covariates
Demographics
Age at the start of the study was calculated as the difference between date of birth, extracted from VISTA, and date of study entry. Race was based on a combination of self-report from the Health Checklist and the electronic record. Gender was extracted from VISTA. Education and marital status were determined from the Health Checklist.
Health Behaviors
Alcohol consumption, smoking status, and drug use were derived from the Health Checklist. The Alcohol Use Disorders Identification Test, Consumption Questions is a three-item validated alcohol screening questionnaire (with scores ranging from 0–12; 0–4 points for each question) (20,21) that assesses frequency and typical quantity of drinking, as well as the frequency of episodic heavy drinking. Smoking status was classified as never, past, or current. Drug abuse history was based on a single-item question about whether a doctor or nurse ever told the patient he or she had drug abuse.
Medical Conditions
Seven diseases from the Health Checklist, based on single-item questions about prior diagnoses, were included as markers of medical status: cancer; chronic lung disease, emphysema, asthma, or bronchitis; congestive heart failure (CHF); diabetes; heart attack (MI); pneumonia; and stroke. Each was previously shown to be independently associated with mortality in this sample (22).
Current Depressive Symptomatology
Depressive symptoms during the 4 weeks before completion of the Health Checklist were assessed with the Mental Health Inventory (MHI-5), a 5-item self-administered questionnaire, with raw scores ranging from 5 to 30 (23). Scores of 
17 are considered a positive screen for depression (23), and are reported for descriptive purposes. Continuous MHI-5 scores were used in all exploratory analyses undertaken to better characterize the relationship between depression, PTSD, and mortality.
Statistical Analyses
Descriptive Baseline Analyses
We performed descriptive analyses to compare the characteristics of patients with neither depression nor PTSD, patients with depression alone, patients with PTSD alone, and patients with both depression and PTSD. Because age is strongly associated with depression, PTSD, and mortality in this population, we computed age-adjusted prevalence estimates of key covariates by direct standardization using the entire group of respondents as the reference population. Logistic regression models were used to determine statistically significant differences, comparing patients with no depression and no PTSD with patients with either depression, PTSD, or both. Patient characteristics, both categorical and continuous, were entered into separate models as independent variables with age.
Survival Analyses
To examine the relationship that depression and PTSD have with mortality, we constructed Cox proportional hazards models with depression alone, PTSD alone, and depression plus PTSD groups simultaneously entered as separate independent variables into a series of models, which effectively allowed for a different association between each category and mortality. Patients were followed until June 30, 2000. Age (in quintiles) was included as a covariate in all analyses because it was expected a priori to be a strong confounder due to its strong association with depression, PTSD, and mortality. The demographics model included gender, race, education, and marital status in addition to age. The health behaviors model included smoking, alcohol and drug use in addition to age and demographic variables. The complete model included medical comorbidities (cancer, lung disease, CHF, diabetes, MI, pneumonia, and stroke) in addition to age, demographic variables, and health behaviors.
Post Hoc Analyses
We conducted a variety of exploratory analyses. First, to better understand the relationships that depression and PTSD have with mortality, we added a measure of current depressive symptoms (MHI-5 modeled as a continuous variable) as an additional covariate to the complete model described above. The MHI-5 was not included in the main models. Second, we also examined the association that depression and PTSD status had with the number of outpatient health care visits. Using linear regression, depression alone, PTSD alone, and depression plus PTSD groups were entered as independent variables, with number of visits as the dependent variable. Finally, to better understand how our findings compare with the existing literature on PTSD and mortality, we rescored PTSD to include those with and without a history of depression. This new variable, along with age, was entered as an independent variable into a Cox model. Statistical analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, NC) and STATA version 9.1 (StataCorp, College Station, TX).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
Compared with patients with no history of depression or PTSD, patients with a history of depression alone as well as patients with a history of depression and PTSD were at increased risk of death over the follow-up period, although the latter association did not remain after adjusting for medical comorbidity (Table 2). Patients with a history of PTSD alone were not at increased risk of death. Moreover, patients with PTSD in addition to depression had no greater risk of dying over 2 years than those with depression alone (Table 2).
|
To determine whether the association that history of depression and/or PTSD had with mortality persisted after adjusting for current depressive symptoms or distress, we added continuous MHI-5 scores to the model. Current depressive symptoms with the MHI-5 were associated with an increased risk of death (after adjusting for depression history, PTSD history and all covariates: hazard ratio (HR) = 1.02, 95% Confidence Interval (CI) = 1.01–1.03). We found that history of depression did not remain significantly associated with mortality after adjusting for current depressive symptoms, age, demographic variables, health behaviors, and medical comorbidities (Table 2). We also found that a history of both depression and PTSD was not associated with an increased risk of death after adjusting for current depressive symptoms. After adjusting for current depressive symptoms and other covariates, a history of both depression and PTSD was associated with a lower risk of death compared to those with neither depression nor PTSD.
The observed lower mortality among patients with both PTSD and depression compared to patients with neither might reflect greater intensity of mental health treatment among these patients with secondary general health benefits (24,25). Post hoc analyses to explore this hypothesis revealed that patients with depression and PTSD had higher outpatient utilization than patients with either depression alone or PTSD alone (Table 3). After adjusting for number of outpatient visits and the interaction between number of visits and depression-PTSD group, the negative association depression and PTSD had with mortality was no longer significant (HR = 0.87, 95% CI = 0.70–1.07, p = .19).
|
To better understand how these findings relate to prior studies which reported increased mortality among patients with PTSD without adjustment for depression, we conducted additional exploratory analyses. In age-adjusted analyses, we found that patients with a history of PTSD, redefined to include patients either with or without a history of depression, were at increased risk of death (PTSD HR = 1.13 (1.00–1.28), 
2 = 4.03, p = .04). However, as expected, this association did not remain after adjusting for history of depression (PTSD HR = 0.92 (0.80–1.04), 2 = 1.78, p = .18; Depression HR = 1.44 (1.32–1.57), 2 = 71.33, p < .0001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
As is consistent with the body of literature conducted in general community samples (14,26,27), the data presented here highlight the importance of depression as a predictor of health outcomes and mortality among veterans. Patients with a history of depression were more likely to engage in deleterious health behaviors, such as smoking and drug use, compared with patients with neither depression nor PTSD. In addition, these patients were more likely to have reported a range of medical comorbidities associated with increased mortality, including cancer, lung disease, CHF, MI, pneumonia, and stroke. Although we cannot conclude from these data that depression preceded the development of medical comorbidities, our findings are consistent with prior research showing that depressed patients are more likely to engage in deleterious health behaviors, which could lead to the development of comorbidities and premature mortality.
Similar to patients with a history of depression, patients with a history of PTSD were more likely than patients with neither depression nor PTSD to smoke, have a history of drug use, cancer, lung disease, CHF, MI, pneumonia, or stroke. However, a history of PTSD, in the absence of a history of depression, was not associated with increased mortality in this sample of primary care patients. Only a few prior studies have examined the relationship between PTSD and mortality; and of those studies, most reported that patients with PTSD had increased mortality (28–31). However, none of these studies attempted to examine the relationship between PTSD and mortality independent of depression. When we examined PTSD, without accounting for depression, we did observe a modest age-adjusted association between PTSD and mortality, which is consistent with this literature. Our findings suggest that depression should not be ignored in studies evaluating the health consequences of PTSD. In this sample, 77% of patients with a history of PTSD also reported a history of depression, and 58% reported significant depressive symptomatology at the time they completed the Health Checklist. The co-occurrence of depression and PTSD may reflect symptom overlap and discrimination difficulties, trauma as a shared etiology, or the influence of one disorder on the development of the other disorder. Therefore, PTSD may typically exist with comorbid depression, and as a result, it may be difficult to disentangle the effects of PTSD from the effects of depression. The absence of an independent association between PTSD and mortality may be due to several additional factors. First, we may not have observed an association because of the limitations of our measure of PTSD in this sample. PTSD often remains undiagnosed and untreated, and our measure of PTSD therefore may misclassify some patients with undiagnosed and untreated PTSD, potentially biasing results toward the null. We also did not have a PTSD symptom measure, and like depression, it may be only those with current symptoms, or severe symptoms, who are at greater risk of dying within 2 years. Second, this study assessed PTSD, in some cases, decades after active duty. Prior studies have noted that the majority of deaths associated with PTSD have been due to "external" causes (i.e., accidents and suicides) (29,30), which may be particularly relevant in the first few years post combat exposure (32). Thus, patients with a prior diagnosis of PTSD in this study could include relatively healthy survivors—those who were not as vulnerable to death from accidents or suicides.
We found that patients with a history of depression were no longer at increased risk of death after adjusting for current depressive symptoms with the MHI-5, suggesting that current depressive symptoms may explain the association between history of depression and mortality. Each 1-point increase in MHI-5 score was associated with a 2% increase in the risk of death in these analyses. Patients who reported current depressive symptoms may have been more vulnerable to, or may have been more likely to behave in ways that could have adversely affected relatively short-term health outcomes, such as 2-year mortality. Also, patients who reported current depressive symptoms may have experienced more severe and/or persistent symptoms, which could have influenced health outcomes. Alternatively, underlying disease may have led to both depressive symptoms and death. However, in fully adjusted analyses that account for degree of comorbidity, current symptoms of depression remained a significant independent predictor of death.
Surprisingly, after adjusting for current depressive symptoms, patients with a history of both depression and PTSD were at a lower risk of dying compared with patients without evidence of depression or PTSD. In post hoc analyses, we found that patients with a history of both depression and PTSD had nearly twice the number of outpatient visits, including primary care visits, than those with neither depression nor PTSD during the year after completion of the Health Checklist. After adjusting for number of outpatient visits, a history of depression and PTSD was no longer associated with lower mortality. This observation supports the hypothesis that VA patients in this dual diagnosis group were actively engaged in health care, and that heath care utilization may help explain some of the differences in survival between groups.
Strengths and Limitations
This study has a number of strengths and limitations. Our study included >35,000 primary care patients from seven geographically diverse VA medical centers, and used questionnaires that uniformly assessed psychological and medical status, including both depression and PTSD, and electronic administrative data. The unique dataset enabled us to examine the independent relationship that depression and PTSD have with mortality after adjusting for a number of demographic characteristics, health behaviors, and medical comorbidities. At the same time, we are cautious about drawing conclusions from these results alone, given that these are secondary analyses based on a dataset not originally designed to study the relationship between depression, PTSD, and mortality. Our measures of depression and PTSD may be an underestimate of cases, given that depression and PTSD often remain undiagnosed and untreated. However, our composite measure, based on both self-report and administrative medical records, captures more cases than would be recognized through either assessment approach alone. Validation of these measures was not possible in ACQUIP or the present study, although we observed that our measures of depression and PTSD were associated with adverse health behaviors, as well as psychiatric and medical comorbidity in expected directions. This was not an inception cohort and this study did not have a measure of the duration of either depression or PTSD. We also did not have data about PTSD symptomatology or nature of the traumatic event. In addition, we were unable to verify whether depression and PTSD preceded the onset of medical problems or resulted from those medical problems, an important issue to evaluate in future studies. At the same time, the literature demonstrating a strong association between prior military combat exposure and PTSD in veteran populations indicates that, for many of the patients in this study, PTSD was likely linked to military service (5). The 2-year duration of follow-up is short; however, a mean age of 58 among those with either depression, PTSD, or both suggests that many of these veterans may have initially developed depression and PTSD decades before the study, and the results could partially reflect a cumulative exposure to psychological problems over those years. Although we did have a measure of depressive symptoms, the MHI-5, it is not a diagnostic measure used to assess major depression and may better reflect general distress. Our single-item measure of self-report depression history is nonspecific and may have been endorsed by patients with emotional problems other than depression. However, even with this measure, we observed that depression history was associated with other variables in the expected direction. Our comparison group of patients without a known history of depression or PTSD is limited to other veterans using the VA for health care who were largely older and had more coexisting disease, which may have biased our results for the association between depression, PTSD, and death toward null. Although we attempted to control for the confounding effects of age and coexisting diseases by adjustment, it is possible that some degree of residual confounding persisted. Finally, we did not have data about cause of death. Understanding cause of death in future studies (e.g., suicide) could have important intervention implications.
In summary, among veterans engaged in primary care in the VA who participated in a quality improvement study, a prior diagnosis of depression, but not PTSD, was associated with an increased risk of death over an average of 2 years after adjusting for age, demographic variables, health behaviors, and medical comorbidity. Patients with prior diagnoses of both depression and PTSD seem to be at no greater risk than those with depression alone.
We gratefully acknowledge Dr. Stephan Fihn, Jorja Henikoff, and the ACQUIP investigators for their guidance with the project.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication November 17, 2006; revision received August 21, 2007.
Dr. Kinder was supported by the VA Health Services Research and Development Fellowship and Dr. Bryson was supported by a VA Career Development Award (RCD00-018). The Ambulatory Care Quality Improvement Project (ACQUIP) was funded by Grants #SDR96-002 and IIR99-376 from the VA Health Services Research and Development Service.
DOI:10.1097/PSY.0b013e31815aac93
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. C. Phillips, G. D. Batty, C. R. Gale, I. J. Deary, D. Osborn, K. MacIntyre, and D. Carroll Generalized Anxiety Disorder, Major Depressive Disorder, and Their Comorbidity as Predictors of All-Cause and Cardiovascular Mortality: The Vietnam Experience Study Psychosom Med, May 1, 2009; 71(4): 395 - 403. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Bryson, D. H. Au, H. Sun, E. C. Williams, D. R. Kivlahan, and K. A. Bradley Alcohol Screening Scores and Medication Nonadherence Ann Intern Med, December 2, 2008; 149(11): 795 - 803. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. W. Bradford, M. M. Kim, L. E. Braxton, C. E. Marx, M. Butterfield, and E. B. Elbogen Access to Medical Care Among Persons With Psychotic and Major Affective Disorders Psychiatr Serv, August 1, 2008; 59(8): 847 - 852. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
