Association of Posttraumatic Stress and Depressive Symptoms With Mortality in Women

This cohort study examines the association between posttraumatic stress disorder, with and without comorbid depression, and mortality in women.


Introduction
Posttraumatic stress disorder (PTSD) has been associated with increased risk of chronic disease, including hypertension, cardiovascular disease, and type 2 diabetes, [1][2][3][4][5] and with greater prevalence of health risk factors, such as obesity and smoking. 6,7 Furthermore, PTSD has been associated with biological changes involved in several disease processes, 8 including hypothalamic-pituitary-adrenalaxis alterations and related inflammation and immune dysregulation, [9][10][11] oxidative stress, 12 poor sleep, 13 and indicators of accelerated aging. 5,14 These diseases, health risk factors, and biological changes are associated with increased mortality. 15 However, the association between PTSD and all-cause mortality has been investigated almost exclusively in samples of male military veterans. In these studies, [16][17][18][19][20][21][22] PTSD has been associated with increased risk of all-cause mortality as well as mortality from cardiovascular disease, cancer, and external causes, such as unintentional and intentional injury, with few exceptions. 23 However, PTSD occurs at higher rates among women. In the United States, lifetime prevalence of PTSD in women is more than 2-fold that of men (9.7% vs 3.6%). 24 Risk of PTSD following a trauma exposure is similarly 2-fold higher in women compared with men. 25,26 Only 1 large study, 27 using Danish medical records, has investigated the association of PTSD with mortality in women or civilians, finding more than 2-fold the risk of death in individuals with PTSD vs those without PTSD.
Depression often co-occurs with PTSD 28 and, like PTSD, it is far more prevalent in women than men. 29,30 Depression has been independently associated with greater prevalence of health risk factors 31,32 and risk of mortality. [33][34][35][36] Evidence also suggests that PTSD with depression may constitute a particularly severe subtype of posttraumatic response, with unique biological outcomes important for physical health. 37 Thus, PTSD with depression may be associated with even greater risk of mortality compared with PTSD alone. However, to our knowledge, only 3 relatively short-term studies have examined this possibility. One study, 38 among Japanese earthquake survivors ages 65 years and older, found increased risk of all-cause mortality over 3.3 years of follow-up among individuals who had depression at baseline, regardless of whether they had high PTSD symptoms, compared with individuals who did not have depression at baseline. Individuals with PTSD and depression were not at significantly increased risk of dying during the follow-up period compared with those with depression only. A 2013 study 39 of 391 patients with end-stage kidney disease found increased risk of death among patients with both depression and PTSD and patients with depression alone but not among those with PTSD alone, compared with patients with neither disorder, in 3.5 years of follow-up. A 2010 study 40 of in-hospital mortality after a coronary artery bypass grafting surgical treatment found that patients with comorbid depression and PTSD were at increased risk of death compared with patients with neither disorder. Patients also had an increased risk of death if they had depression alone or PTSD alone.
Several studies 6,31,32,41 have found that individuals with PTSD or depression, compared with individuals without these disorders, have increased prevalence of health risk factors, such as smoking and obesity, which may contribute to increased mortality. Thus, higher prevalence of health risk factors in individuals with co-occurring PTSD and depression may account for possible increased mortality. It remains largely unknown whether PTSD is associated with increased mortality among civilians and women, whether co-occurring depression is associated with further increased risk, and whether health-risk factors are associated with these increased risks of death. In the present study, we examined the association of PTSD and depression symptoms with risk of death in a large prospective cohort of women, the Nurses' Health Study II. 42 We further examined whether healthrelated factors, including body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), smoking, and exercise, were associated with differences in mortality among individuals with PTSD or depression.

Methods
This cohort study examined women in the Nurses' Health Study II. The institutional review board of Brigham and Women's Hospital approved that study's protocol. Return of the questionnaire by the respondent via US mail constituted implied informed consent. The Nurses' Health Study II is an ongoing cohort study of 116 429 women, enrolled in 1989 at ages 25 to 42 years (median age, 34.0 years) and followed biennially. In 2008, 60 804 women who completed the most recent biennial questionnaire and an earlier supplemental questionnaire were mailed a supplemental PTSD questionnaire, with 54 687 women responding. As the Nurses' Health Study II was initially formed to study the health effects of oral contraceptive use, only women were enrolled.

Measures Trauma, PTSD, and Depression
For each of 15 potentially traumatic events (eg, serious motor vehicle crash) and an additional other event, women in the study reported in 2008 whether they had ever experienced the event. They were asked which event they considered their worst or most distressing event. Seven PTSD symptoms in relation to this worst event were queried with the Short Screening Scale for Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) 43 PTSD. 44 Trauma exposure and PTSD symptoms were jointly coded as no trauma exposure (reference), trauma and no PTSD symptoms, 1 to 3 PTSD symptoms (subclinical), 4 to 5 PTSD symptoms (moderate), and 6 to 7 PTSD symptoms (high). In a representative sample of Detroit residents ages 18 to 45 years, 44 a cutoff of 4 or more identified PTSD cases with sensitivity of 80%, specificity of 97%, positive predictive value of 71%, and negative predictive value of 98%, and a cutoff of 6 or more identified cases with sensitivity of 38%, specificity of 99%, positive predictive value of 87%, and negative predictive value of 95%. We additionally coded PTSD symptoms as a continuous variable (range, 0-7). Past-week depressive symptoms were assessed in 2008 using the Center for Epidemiologic Studies Depression Scale-10 (CESD-10) 45 and dichotomized at 10 or more to indicate probable depression. 45 We additionally coded depressive symptoms as a continuous variable (range, 0-30). The CESD-10 has been validated against the highly validated longer form, the Center for Epidemiologic Studies Depression Scale-20, in a sample of older adults in a US health-maintenance organization (Cohen κ, 0.97) 45 and against clinical evaluations with good psychometrics. 46,47 To examine the co-occurrence of PTSD and depression with mortality, we also characterized PTSD and depression with an interaction term, using indicator variables as follows: no depression or of death and ascertainment and coding of death record data, cause of death was available for 384 of 555 women (69.1%) who died during follow-up.

Health-Related Factors and Covariates
Health-related factors included BMI, smoking status, physical activity, and marital status at or before study baseline in 2008. We did not time-update these factors, as illness preceding death could lead to weight loss, smoking cessation, and reduced physical activity. Self-reported height in 1989 and weight in 2007 were used to calculate BMI, coded with continuous and squared terms, as this Highest occupation (ie, jobs that usually have higher status and pay) of either parent during the respondent's infancy was reported as farmer, laborer, blue-collar (eg, mechanic or bus driver) or lower white-collar worker (ie, secretarial or clerical work), or managerial or professional. Parental education was reported as high school or above, some college, or college graduate or above, and parental home ownership in respondent's infancy was coded as yes or no. In 2005, respondents indicated their race/ethnicity by selecting 1 or more of the following: White, Black or African American, American Indian or Alaska Native, Native Hawaiian or Pacific Islander, or other. For analyses, race was coded White or non-White, as 50 137 individuals in the sample (97.2%) selected only White. Age was measured in months.

Statistical Analysis
We examined the distribution of health-related factors and covariates by PTSD and depression status in 2008. We then calculated the association of health-related factors with mortality by fitting a single Cox proportional hazard model with all factors included as independent variables, adjusted for age and race/ethnicity. To investigate the association of PTSD and depression with mortality, we first examined the association of PTSD with depression. We then ascertained the best-fitting model, using the Akaike information criterion 48 to compare 4 models: PTSD alone; depression alone; PTSD and depression; and PTSD, depression, and a PTSD-depression interaction term, using indicator variables as previously described. Finally, we fit 2 Cox proportional hazard models using the best-fitting model, adjusted for age, race/ethnicity, and childhood socioeconomic status and further adjusted for health-related factors, including BMI, smoking status, physical activity, and marital status. In additional analyses, we examined the association of mortality with depression and PTSD symptoms coded continuously, with an interaction term calculated by multiplying the 2 continuous variables, among women exposed to a traumatic event.
To reduce concerns that illness caused both PTSD or depressive symptoms and death, we excluded 3026 women who reported serious illness as their worst trauma and excluded the first year of person-time after the 2008 PTSD questionnaire, meaning we excluded 52 women who died during that year. We excluded an additional 7 women who did not respond to questionnaires between 2009 and 2017, leaving 51 602 women in the sample. For all models, hazard ratios (HRs) were estimated using the phreg procedure in SAS statistical software version 9.4 (SAS Institute). A 2-sided P < .05 was considered significant in statistical tests. Data analysis was performed from September 2018 to November 2020.
To improve power, we examined cause of death in 3 aggregated groups: women with no depression or PTSD symptoms, women with any (1-7) PTSD symptoms or with depression but not both, and women with any PTSD symptoms and depression. We tested differences between the

JAMA Network Open | Psychiatry
Association of Posttraumatic Stress and Depression With Mortality in Women reference group of women with no depression or PTSD symptoms and each of the other 2 groups using χ 2 tests.

Results
At baseline in 2008, this study included 51 602 women, with mean (range) age 53 95% CI, 1.02-1.65), and low physical activity (ie, less than 3 metabolic equivalents/wk) was associated with nearly 50% increased risk of death compared with high physical activity (ie, 42 or more metabolic equivalents/wk) (HR, 1.49; 95% CI, 1.09-2.04). Childhood socioeconomic status and participant's race/ethnicity were not associated with increased mortality.
We found that PTSD was associated with depression. While probable depression was reported probable depression, and a PTSD-depression interaction term (P for interaction = .06). In models adjusted for age, race/ethnicity, and childhood socioeconomic indicators, co-occurring PTSD and depression were associated with mortality. Women with high PTSD symptoms and depression were at nearly 4-fold increased risk of death compared with women with no trauma exposure or depression (HR, 3.80; 95% CI, 2.65-5.45; P < .001). Women with depression and moderate PTSD symptoms (HR, 2.03; 95% CI, 1.35-3.03; P < .001) and depression and subclinical PTSD symptoms (HR, 2.85; 95% CI, 1.99-4.07; P < .001) were also at increased risk of death compared with women with no trauma or depression ( Table 2). With further adjustment for health factors, women with high PTSD symptoms and depression remained at increased risk of death (HR, 3.11; 95% C, 2.16-4.47; P < .001) ( Table 2). Women with subclinical PTSD symptoms without probable depression had increased risk of death compared with women with no trauma or depression (HR, 1.43; 95% CI, 1.06-1.93; P = .02) ( Table 2). Depression in women without trauma exposure was associated with more than 2-fold increased risk of death (HR, 2.39; 95% CI, 1.44-3.95; P < .001). However, risk of mortality among women with depression and trauma exposure who did not develop PTSD symptoms was not increased compared with the reference group (HR, 1.28; 95% CI, 0.74-2.21; P = .39) ( Table 2). To further explore which potentially modifiable health factors might account for increased risk of death among women with PTSD and depression, we fit 3 additional models separately adjusted for BMI, smoking, and physical activity. The association of PTSD and depression with mortality remained significant in models adjusted for BMI or smoking. For example, among women with depression and high PTSD symptoms, the HR adjusted for BMI was 3.57 (95% CI, 2.49-5.13; P < .001) and the HR adjusted for smoking was 3.40 (95% CI, 2.67-4.89; P < .001) ( Table 2).

JAMA Network Open | Psychiatry
In analyses among women who had trauma exposure, with PTSD and depression symptoms coded continuously, the best-fitting model included PTSD, depression, and an interaction term. In women without depression symptoms, increased number of PTSD symptoms was not associated with increased mortality (HR per PTSD symptom, 0.97; 95% CI, 0.91-1.04; P = .42). Depression symptoms and depression-PTSD interaction were associated with increased risk of mortality (HR per

Discussion
This cohort study, to our knowledge the first large study of co-occurring PTSD and depression, found that women with co-occurring high PTSD symptoms and probable depression had nearly 4-fold  Among women with depression in our study, those who were exposed to a traumatic event but did not develop PTSD symptoms were not at increased risk of death compared with women with no trauma exposure or depression, while other women with depression, including those with no exposure to trauma, were at increased risk. It may be that not developing PTSD symptoms after experiencing trauma is an indicator associated with psychological resilience 50 and this resilience may be protective against the physical health effects of depression.
Symptoms of PTSD and depression overlap, with dysphoria and numbing common to both disorders. 51

Limitations
Our study has several limitations. Our sample included predominantly White women ages 43 to 64 years, which may limit generalizability. The sample also included only respondents who survived until the PTSD questionnaire was administered, which may have attenuated associations. 66 In addition, our measures captured symptoms of both disorders rather than clinical diagnoses, lifetime PTSD symptoms were queried retrospectively, and only past-week depressive symptoms were queried, which may have resulted in misclassification. We lacked information on illicit substance use and abuse, which have been associated with increased risk of death in veterans with PTSD. 19,20 Illicit substance use may have accounted for an additional part of the association of PTSD and depression with mortality beyond the health factors we examined.