Depression and Risk of Venous Thromboembolism: A Population-Based Retrospective Cohort Study.

Depression and Risk of Venous

Thromboembolism: A

Population-Based

Retrospective Cohort Study

Cynthia Wei-Sheng Lee, PhD, Chun-Hui Liao, MD, Cheng-Li Lin, MSc,

Ji-An Liang, MD,

Fung-Chang Sung, PhD, and Chia-Hung Kao, MD

INTRODUCTION

V

which the blood clots inappropriately, resulting in considerable morbidity and mortality. VTE comprises pulmonary

embolism (PE), which occurs when a clot breaks free and enters the arteries of the lungs, and deep vein thrombosis (DVT), which refers to clots in the deep veins of the body (1). The incidence of VTE is 15.9 and 71 to 117 per 100,000 personyears in Taiwanese (2) and white (3–5) populations, respectively. Survival after VTE is lower than expected, especially

for patients with PE, of whom 30% experience VTE recurrence and develop venous stasis syndrome (6). The scores

for known risk factors such as advanced age, immobility, operations, cancer, and obesity are increasing in society; thus,

VTE is a growing public health problem (1).

The risk of VTE is increased in patients with psychiatric disorder (7,8), especially those with schizophrenia and bipolar disorder (9). A specific algorithm for VTE prevention in hospitalized patients with psychiatric disorder has

been proposed to manage this problem (10). In addition,

an increased risk of VTE is correlated with self-reported depressive feelings (11), and there is a link between psychosocial

risk factors and incident VTE (12–14). Depression has been linked to an increased risk of arterial cardiovascular diseases (15). The current status of depression in Taiwan was investigated in a community-based walk-in screening, using the Center for Epidemiological Studies Depression

Scale, and the results demonstrated that the ratio of suspected depressed patients was 16.4% of all recruited

participants (16). Although the etiology of cardiovascular diseases probably differs from that of VTE (17), both diseases involve coagulation processes and the risk factors

for cardiovascular diseases and incident VTE are quite similar (18). Therefore, we postulate that depression contributes to VTE development.

Because VTE is a debilitating disease, understanding its relationship with depression is crucial. According to our review of relevant research, few prospective epidemiological

studies have investigated this relationship. By comparing the incidence of VTE in depressed and nondepressed patients by using data from the National Health Insurance Research Database (NHIRD) of Taiwan, we evaluated the

probability of developing VTE after depression. METHODS

Data Sources

To perform a nationwide, retrospective, population-based analysis, we obtained insurance claims data on 23 million people insured by Taiwan's National Health Insurance (NHI) program from the National

Health Research Institutes. The NHI program in Taiwan, a universal insurance system established by the Bureau of NHI, Department of

Health, was implemented in March 1995 and covered more than 99% of the population by 2007. The patient information we acquired included data on all medical services received between 1996 and 2011,

the personal characteristics of patients, and the characteristics of physicians and hospitals. For privacy protection, the patient identification

numbers necessary to link files with identities was scrambled by the National Health Research Institutes to ensure patient confidentiality. Our

study was approved by the Ethics Review Board of China Medical University (CMU-REC-101-012).

We selected 35,274 newly diagnosed depressed patients (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 296.2–296.3, 300.4, and 311) without previous diagnoses of DVT or PE in 2000 to 2011 from a subset of the patients enrolled in NHI program, the Longitudinal Health Insurance Database 2000 of Taiwan. The Longitudinal Health Insurance Database 2000, a subset of NHIRD, contains all original claim data of 1,000,000 beneficiaries enrolled in year 2010 randomly sampled from the year 2010 Registry for Beneficiaries of the NHIRD, where registration data of every beneficiary of the NHI program during the period of January 1, 2010, to December 31, 2010, were

drawn for random sampling. There are approximately 27.38 million individuals in this registry (NHIRD).All the registration and claimdata of these

1,000,000 individuals collected by the NHI program constitute the LHID2010. There was no significant difference in the sex distribution (χ2 = 0.067, df = 1, p = .796) between the patients in the LHID2010 and the patients in the NHIRD (http://nhird.nhri.org.tw/en/Data_Subsets. html#S3); the depression diagnosis date was defined as the index date. For each identified depression patient, two insured people without depression or a history of DVT or PE were randomly selected from the same index period and frequency-matched according to sex, age, and index year. The depression cohort comprised 35,274 patients and the nondepression cohort comprised 70,548 patients. The cumulative

censoring rate for 12 years (2000–2011) was 16.7% in the depression cohort, which was slightly higher than that in the

nondepression cohort (15.2%) (Fig. 1). The possible reasons for the discontinuity of NHI include death, withdrawal of insurance, immigration,

prison sentence, and so on.

We considered atrial fibrillation (AF; ICD-9 code 427.3), hypertension (ICD-9 codes 401–405), diabetes (ICD-9-CM code 250), cerebral vascular accident (CVA; ICD-9-CM codes 430–438), heart failure (ICD-9 code 428), lower leg fracture or operations (ICD-9 codes 820– 823; ICD-9-OP codes 81.51–81.54), cancers (ICD-9 codes 140–208), smoking (ICD-9 code 305.1), and obesity (ICD-9 code 278) to be

preexisting comorbidities that potentially confound the association between depression and VTE.

Statistical Analysis

The primary outcome was the occurrence of VTE (ICD-9 codes 415.1 and 453.8, excluding iatrogenic PE, ICD-9 code 415.11), which was determined

by linking patient records with ambulatory and inpatient care data from the NHIRD.

We compared the distributions of categorical demographic variables and comorbidities between the depression and nondepression cohorts, and examined the differences, using a χ2 test. To estimate the cumulative incidence of VTE risk in the two cohorts, we performed survival analysis by using the Kaplan-Meier method and determined significance by using the log-rank test. Multivariate Cox proportional hazards regression was used to examine the effect of depression on the risk of VTE development, which was determined according to the adjusted hazard ratio (HR) with a 95% confidence interval (CI). A multivariable model was used to control for age, sex, and the comorbidities, namely, AF, hypertension, diabetes, CVA, heart failure, lower leg fracture or operations, and cancers. We performed further analysis to assess whether medication with antidepressants

played a role in the VTE outcome. Available antidepressants were divided into three groups, selective serotonin reuptake inhibitors (SSRIs), non-SSRIs, and atypical antipsychotics. A two-tailed p value of .05 was considered statistically significant. All statistical analyses were performed using

SAS statistical software (Version 9.3 for Windows; SAS Institute, Inc, Cary, NC).

RESULTS

Themean age of the patients in the two cohorts was approximately 48 years and 61.6% of the patients were women

(Table 1). Compared with the nondepression cohort, the depression cohort was more likely to have AF (0.98% versus 0.60%), hypertension (33.6% versus 23.7%), diabetes (15.0% versus 1.04%), CVA (15.9% versus 6.77%), heart failure (3.33% versus 2.08%), lower leg fracture or operations (2.14% versus 1.49%), cancer (5.68% versus

3.46%), smoking (1.43% versus 0.59%), and obesity (1.90% versus 1.12%; all p < .05).

The Kaplan-Meier analysis revealed that the cumulative incidence curves of VTE were significantly higher for the depression cohort than for the nondepression cohort (logrank test, p < .001; Fig. 2). In the depression cohort, the rate of VTE incidence was 6.19 per 10,000 person-years (6.68

and 5.90 per 10,000 person-years for men and women, respectively; Table 2). Overall, the risk of VTE was 1.38-fold

greater in the depression cohort than in the nondepression cohort (95% CI = 1.09–1.73), after we adjusted for age, sex, and comorbidities. Compared with patients without depression, patients with depression younger than 49 years exhibited the HR with the greatest magnitude

(1.73, 95% CI = 1.06–2.84). Among patients without comorbidities, the risk of VTE was 2.09-fold higher in the

depression cohort than in the nondepression cohort (95% CI = 1.37–3.21).

The results of univariable and multivariable Cox proportional hazard regression analyses for the association between VTE and depression are listed in Table 3. The

adjusted HR of VTE development was a 1.04-fold increase (95% CI = 1.03–1.05) with age (every 1 year). The risk of developing VTE was greater for patients with the comorbidities of hypertension (adjusted HR = 1.53, 95% CI = 1.02–

1.73), diabetes (adjusted HR = 1.33, 95% CI = 1.02–1.73), heart failure (adjusted HR = 1.95, 95% CI =1.33–2.85), and cancer (adjusted HR = 2.37, 95% CI = 1.71–3.29).

Table 4 shows that patients with depression and comorbidities had a significantly higher risk of VTE (adjusted

HR = 3.26, 95% CI = 2.29–4.63) than did patients without depression and comorbidities. We observed that the HRs of VTE for patients with depression and hypertension,

diabetes, heart failures, and cancers were of a higher magnitude than those of patients without depression or the

aforementioned comorbidities (adjusted HR = 2.11 95% CI = 1.48–3.03; adjusted HR = 1.72, 95% CI = 1.18–2.50; adjusted HR = 2.17, 95% CI = 1.24–3.78; and adjusted HR = 2.86, 95% CI = 1.78–4.60, respectively). Furthermore, to examine the combined effects of depression and

treatment, we divided the depression cohort into three subgroups according to the treatment received (Table 5).We observed

that patients with depression who received treatment had a lower VTE risk than did those who did not. DISCUSSION

Our results indicated that depression is associated with the incidence of subsequent VTE development. The risk of

VTE increased considerably in male and younger (≤49 years) depressed patients. In addition, patients with comorbidities such as hypertension, diabetes, heart failure, and cancer had a higher risk of depression-associated VTE, although this risk could be insignificantly reduced slightly through antidepressant use.

Doctors should consider depression as a risk factor for VTE to facilitate the early diagnosis of VTE in depressed patients. The adverse effects of psychotropic drugs have been determined responsible for increased VTE risk in

depressed patients (19–22). In addition, several case reports have indicated that antidepressant use is associated with the onset of VTE (23–25). However, our results indicated that antidepressants exert an insignificant protective influence against future VTE development in depressed patients.

The adverse effects of antidepressants addressed by previous studies might reflect the relationship between depression and VTE rather than the drug effect of the antidepressants.

Further investigation is required to enhance our understanding of the relationship between depression and VTE

at the pathological level. Depression is a chronic state, and chronic stress is reported to elicit homeostatic changes on blood coagulation and fibrinolysis (26,27). Acute psychological stress activates coagulation and fibrinolytic systems

and platelets (28), disturbing the delicate balance of these hemostatic components and possibly enhancing the risk of VTE. Depressed patients exhibited significantly

higher levels of plasma haptoglobin, fibrinogen, complement components 3, C4, and α1-acid-glycoprotein than

did the healthy controls; however, these acute phase proteins returned to normal levels in depressed patients who

underwent chronic antidepressant treatment (29). Therefore, changes in these plasma proteins might be reliable indicators for future VTE development in depressed patients.

A reason for why SSRIs were not associated with reduced VTE risk could be because SSRIs inhibit platelet activity, whereas platelets do not play an important role in incident VTE (as opposed to elevated levels of D-dimer, factor VIII,

and activated partial thromboplastin time).

The strengths of our study include the use of populationbased

and NHIRD data rather than data obtained from self-reported drug use surveys. Many previous studies have

verified the high accuracy and validity of cardiologyrelated and autoimmune disease diagnosis in the NHIRD

(30), thus indicating the high validity and accuracy of the ICD-9 codes in the database. However, this study has the following limitations. First, the NHIRD lacks crucial data such as detailed demographic information regarding smoking habits, alcohol consumption, body mass index,

socioeconomic status, and family history of systemic diseases. All of these factors are indirectly associated with

depression and increase the potential risk of developing VTE. In addition, because the NHIRD does not contain

life-style data, we were unable to correlate VTE with inactivity or malnutrition. However, because the NHIRD

covers a highly representative sample of Taiwan's general population and the insurance reimbursement policy is universal, these demographic factors are unlikely to have affected antidepressant prescription allotment in the sample group.

Second, because depression might influence mobility and muscle strength, we performed a cohort analysis to eliminate confounding factors caused by VTE. However, evidence derived from a cohort study is generally of lower quality than that obtained from randomized trials; a cohort study can have several biases stemming from adjustments made for confounding variables. Despite our meticulous

study design that adequately controlled for known confounding factors, a potential key limitation is that a bias

caused by unknown confounders might exist. Third, the

diagnoses recorded in the NHI claims are primarily for administrative billing purposes and have not been verified

for scientific use. We were unable to contact patients directly about their depression severity because all beneficiaries listed in the NHIRD are anonymous. Fourth, there was no screening test for depression when participants entered

the NHI program. All the participants were enrolled and diagnosed by the physician because of associated complaints

with depression. It is possible that some depressive patients were undetected. Fifth, the different censoring rates in depressed and nondepressed individualsmight result in potential

deviation. For example, if patients die early in life of a

“competing” disease (e.g., cardiac death) and VTE is a disease that increases with age, then these competing end

points may reduce the magnitude of the association between depression and VTE. In addition, we were unable

to include depression diagnosed before 1996 in our calculations. This omission may have caused an underestimation

of depression incidence that subsequentlyweakened the observed association between depression and VTE. However,

the data we obtained regarding depression and VTE diagnosis were reliable.

Our results indicated that VTE incidence is associated with depression. The risk of VTE is significant among men, people 49 years or younger, and depressed patients with comorbidities. Antidepressant treatment is not significantly associated with the risk of depression-associated

VTE. However, larger population-based studies or largescale randomized clinical trials are required to confirmthese findings before any definitive conclusions can be drawn.