Relation of psychological distress to the international normalized ratio in patients with venous thromboembolism with and without oral anticoagulant therapy

Authors

  • R. VON KÄNEL,

    1. Division of Psychosomatic Medicine, Department of General Internal Medicine, Inselspital, Bern University Hospital, University of Bern, Bern
    2. Department of Clinical Research, University of Bern, Bern
    Search for more papers by this author
  • F. VÖKT,

    1. Division of Psychosomatic Medicine, Department of General Internal Medicine, Inselspital, Bern University Hospital, University of Bern, Bern
    Search for more papers by this author
  • F. DEMARMELS BIASIUTTI,

    1. University Clinic of Hematology and Central Hematology Laboratory, Inselspital, Bern University Hospital, and University of Bern, Bern
    Search for more papers by this author
  • S. STAUBER,

    1. Division of Psychosomatic Medicine, Department of General Internal Medicine, Inselspital, Bern University Hospital, University of Bern, Bern
    Search for more papers by this author
  • W. A. WUILLEMIN,

    1. Division of Hematology and Central Hematology Laboratory, Luzerner Kantonsspital
    Search for more papers by this author
  • P. S. LUKAS

    1. Division of Psychosomatic Medicine, Department of General Internal Medicine, Inselspital, Bern University Hospital, University of Bern, Bern
    2. Zentrum für Suchtmedizin, Basel, Switzerland
    Search for more papers by this author

Roland von Känel, Division of Psychosomatic Medicine, Department of General Internal Medicine, Inselspital, Bern University Hospital, CH-3010 Bern, Switzerland.
Tel.: +41 31 632 20 19; fax: +41 31 382 11 84.
E-mail: roland.vonkaenel@insel.ch

Abstract

Summary.  Background:  Psychological distress might affect the international normalized ratio (INR), but effects might vary depending on oral anticoagulant (OAC) therapy.

Objectives:  To investigate the association of psychological distress with INR and clotting factors of the extrinsic pathway in patients with and without OAC therapy.

Patients and methods:  We studied 190 patients with a previous venous thromboembolism (VTE); 148 had discontinued OAC therapy and 42 had ongoing OAC therapy. To assess psychological distress, all patients completed validated questionnaires to measure symptoms of depression, anxiety, worrying, anger and hostility. INR, fibrinogen, factor (F)II:C, FV:C, FVII:C and FX:C were measured as part of outpatient thrombophilia work-up.

Results:  In VTE patients without OAC therapy, the odds of a reduced INR (< 1.00) were significantly increased from 1.5 to 1.8 times for an increase of 1 standard deviation (SD) in symptoms of depression, anxiety, worrying and anger, respectively, after adjusting for gender, age, body mass index, socioeconomic status, hematocrit and C-reactive protein. Worrying, anger and hostility also showed significant direct associations with FVII:C. In patients with OAC therapy, INR was unrelated to a negative affect; however, lower FVII:C related to anxiety and worrying as well as lower FX:C related to anger and hostility were observed in patients with OAC therapy compared with those without OAC therapy.

Conclusions:  Psychological distress was associated with a reduced INR in VTE patients without OAC therapy. The direction of the association between psychological distress and activity in some clotting factors of the extrinsic coagulation pathway might differ depending on whether VTE patients are under OAC therapy or not.

Introduction

The prothrombin time (PT)-derived international normalized ratio (INR) measures fibrinogen, clotting factors (F)II, V, VII and X of the extrinsic pathway of coagulation. The INR is used to monitor chronic oral anticoagulant (OAC) therapy and is responsive to the three vitamin K-dependent procoagulant clotting factors II, VII and X [1]. In patients with OAC therapy, fluctuations of the INR bear an increased risk of both thromboembolic and bleeding complications [1].

Psychological distress in the form of a negative affect (e.g. depression and anxiety) may ensue if a person cannot resolve environmental demands (‘stressors’) within a reasonable time [2]. In healthy individuals, acute psychosocial stress as well as chronic stress and distress, particularly depression, anxiety and anger/hostility, may all elicit hypercoagulability through various neuroendocrine pathways and hemoconcentration [3,4; for review]. Accordingly, mental stress is now considered an important pre-analytical confound of coagulation tests [5].

Frequent reasons for unexpected INR fluctuations in otherwise stable patients with OAC therapy include changes in diet, long-term alcohol consumption, poor compliance with OAC therapy and undisclosed use of drugs affecting the oral anticoagulant metabolism [6]. Comparably little awareness has been paid to psychological stress and distress as a cause of INR fluctuations in patients with OAC therapy. Moreover, psychobiologic effects of stress on INR might vary depending on whether study participants are on OAC therapy or not. Three previous studies showed that giving a speech and/or a mental arithmetic shortened PT in healthy men [7–9], although significantly so in only one of these studies [9]. In another study, no significant change in PT was observed during a period of increased work stress in healthy accountants [10]. While the individuals from these studies were not on OAC therapy, a case report reasoned that life event stress had contributed to a clinically relevant increase in INR in a patient under OAC therapy [11]. As stress triggers an innate immune response [12], the authors argued that cytokine-induced suppression of CYP450 isoenzymes reduced warfarin metabolism during the stress period [11]. In stressed patients with OAC therapy, increased warfarin availability thus might outweigh coagulation activation to result in a net INR increase.

To our knowledge, an association between a negative affect and INR has not been investigated in individuals both with and without OAC therapy. The primary aim of this study was to examine the association of symptoms of depression, anxiety, worrying, anger and hostility with the INR in patients with and without OAC therapy after they had experienced a venous thromboembolism (VTE). We further explored whether fibrinogen, FII:C, FV:C, FVII:C and FX:C in their relation with a negative affect would contribute to INR. To minimize confounding, we only included patients with normal liver and renal function and without any indication for acute inflammation and apparent hemoconcentration. We hypothesized that a more negative affect would be associated: (i) with a reduced INR and higher levels of clotting factors in patients with no OAC therapy as well as (ii) with higher INR values and lower levels of clotting factors in patients with OAC therapy.

Patients and methods

Study participants and recruitment

The 190 participants of this study were consecutively enrolled between March 2006 and April 2011 from a total of all 2204 patients scheduled for thrombophilia work-up at the outpatient clinic at the Central Hematology Laboratory, Inselspital, Bern University Hospital, Bern, Switzerland [13]. The local ethical committee approved the protocol and all participants provided written consent. Inclusion criteria were a previous objectively diagnosed venous thromboembolic event, either deep venous thrombosis (DVT), pulmonary embolism (PE) or both, and being 18 years or older. Patients with a secondary cause for VTE were included if they had a temporary thrombotic risk situation (e.g. oral contraceptive use and surgery) but no persistent causes. Specific exclusion criteria were persistent risk situations for VTE and diseases possibly affecting prothrombotic measures at the time when VTE occurred. These were autoimmune diseases, major endocrine disorders (e.g. diabetes), rheumatic diseases, active cancer, coronary artery disease, renal and liver diseases, infectious diseases (e.g. human immunodeficiency virus disease) and permanent immobilization.

The recruitment procedure of initially 324 patients with a DVT and/or PE out of 2204 patients has been detailed elsewhere [13]. For the present study, we additionally excluded four patients not having passed their 18th birthday at the time when completing questionnaires, as well as 66 patients with missing information for socioeconomic status (SES) (n = 3), laboratory (n = 29), clinical (n = 24) and psychometric (n = 10) data to allow a full linear regression approach. Further exclusion criteria applying to 64 patients were < 3 months between VTE and thrombophilia work-up (n = 24); OAC therapy stopped less than 1 month before thrombophilia work-up; C-reactive protein (CRP) ≥ 10 mg L−1 suggesting an active inflammatory process (n = 13); plasma creatinine above the upper norm value (> 104 μmol L−1 for men, > 85 μmol L−1 for women) (n = 4); alanine transaminase above three times the upper norm value (> 147 I/U for men, > 102 I/U for women) (n = 2); and hematocrit outside the normal reference range (< 40%/> 50% for men, < 36%/> 44% for women) (n = 14). Six patients without OAC therapy with abnormally low levels for FII:C (< 33%), FVII:C (< 38%) and FX:C (< 19%) and corresponding high INR (> 1.72) as well as one patient under OAC therapy with an INR of 1.07 and values for FII:C of 99%, FVII:C of 89% and FX:C of 115% were excluded as possible single factor deficiency or as outliers, respectively. This procedure yielded a final study sample of 148 patients with no OAC therapy and 42 patients under OAC therapy at the time of assessment.

A questionnaire for the self-assessment of sociodemographic and psychosocial factors was sent by regular mail to all patients 10 days before their clinical visit. Upon arrival at the clinic, patients handed in the completed questionnaires. The treating physician obtained demographic factors, a personal history (including self-report on weight and height to compute the body mass index [BMI]) and performed a thrombophilia work-up. The highest educational level attained was used to define SES (the few patients with no graduation were grouped along with those indicating vocational education). Additional information was abstracted from the patients’ medical records. The decision about the assessment of prothrombotic measures as part of the thrombophilia work-up was based on clinical grounds and unrelated to study objectives.

Assessment of psychological distress

Symptoms of depression and anxiety during the previous week were assessed using the validated German version [14] of the Hospital Anxiety and Depression scale (HADS) [15]. The HADS does not consider somatic indicators of depression and anxiety as these might directly originate from an underlying medical condition. The depression and anxiety subscales comprise seven items each to be rated on a four-point Likert scale (0 = not at all, 3 = mostly; sum score 0–21). Typical items are ‘I feel as if I am slowed down’ (depression) and ‘I get sudden feelings of panic’ (anxiety). Worrying was assessed through a validated German version [16] of the 16-item Penn State Worry Questionnaire (PSWQ) [17]. Typical items are ‘Many situations made me worry’ and ‘Once I started worrying, I couldn’t stop’. All items are rated on a five-point Likert scale (0 = not at all typical for me, 4 = very typical for me; sum score 0–64). Feelings of anger in the previous week were measured using the validated German version [18] of the seven-item Profile of Mood States (POMS) anger subscale [19]. Typical items are ‘angry’ and ‘bad-tempered’, all rated on a seven-point Likert scale (0 = not at all, 6 = extremely; sum score 0–42). Cynical hostility was assessed using the German 23-item cynicism subscale [20] that was derived from the Minnesota Multiphasic Personality Inventory (MMPI)-2 [21]. Typical items are ‘I think nearly anyone would tell a lie to keep out of trouble’ and ‘It is safer to trust nobody’. Items are rated as 0 = false or 1 = true (sum score 0–23); the MMPI-2 cynicism subscale is strongly related to the Cook–Medley Hostility Scale that is part of the MMPI [21]. Cronbach’s alpha indicated reasonably good to very good reliability in the present patient sample for all scales (HADS depression: 0.82; HADS anxiety: 0.74; PSWQ worrying: 0.91; POMS anger: 0.92; and MMPI-2 cynical hostility: 0.85).

Laboratory assessment

For the assessment of prothrombotic factors, blood was collected from an antecubital vein into plastic syringes (Monovette, Sarstedt, Nümbrecht, Germany) with 1 mL of 0.106 mol L−1 trisodium citrate. Samples were centrifuged twice at 1500 × g for 10 min each at 15–18 °C; plasma aliquots were either analyzed right away or stored in polypropylene tubes at −70 °C. All coagulation assays were performed on a Behring Coagulation System automated analyzer (Dade Behring, Marburg, Germany), using commercial reagents and standard human plasma (SHP; Dade Behring AG, Düdingen, Switzerland) as previously detailed [13,22]. Assays of INR, fibrinogen, FII:C, FV:C, FVII:C and FX:C belong to the standard thrombophilia laboratory work up at our institution. The reason is to exclude liver synthetic insufficiency and/or vitamin K deficiency or antagonism in case low values of protein C, protein S, and antithrombin are found. For clinical routine, INR values of 1.00 or higher are indicated in precise units, whereas INR values below 1.00 are collectively expressed as ‘INR < 1.00’. Thus, patients without OAC therapy were categorized into those with an INR < 1 (‘reduced INR’) and those with an INR ≥ 1.00 (‘normal/elevated INR’). Between-run coefficients of variation are < 10% for all measures. For biochemical analyzes, venous blood was collected in appropriate tubes and immediately analyzed within the clinical routine at the Institute of Clinical Chemistry, University Hospital Bern, Switzerland (certified ISO/ICE). A Modular P800 system with standard test kits (Roche Diagnostics, Rotkreuz, Switzerland) was used to determine CRP (immunoturbidimetric assay), creatinine (creatinase enzymatic/photometric method) and ALAT (alpha-ketoglutaratic enzymatic method). Hematocrit was determined from EDTA blood samples using an automatic cell counter (ABS Micros CRP; Polymed, Glattbrugg, Switzerland). Hematocrit was automatically calculated from the erythrocyte concentration and the impedance-determined mean corpuscular volume of erythrocytes.

Statistical analysis

Data were analyzed using pasw version 18.0 (SPSS Inc., Chicago, IL, USA). Level of significance was set at P < 0.05 (two-tailed). Values of INR and clotting factors showed a normal distribution in both patient groups as verified by the Kolmogorov–Smirnov test. Mann–Whitney U-test and Pearson’s chi-square test were used to test for group differences in continuous and categorical variables, respectively. Spearman’s rank correlation coefficients were calculated to estimate the relationship between two variables. We did not adjust P-values for multiple comparisons because the primary outcome variable was INR, whereas all clotting factor levels were tested as secondary outcomes. Rather than testing associations between INR/clotting factors and individual domains of psychological distress, we were primarily interested in the robustness of an association between psychological distress and INR/clotting across several domains of negative affect [23].

We applied logistic linear regression analysis (Wald test) to calculate the odds (95% CI) for patients with no OAC therapy to have a reduced INR (i.e. INR values smaller than 1.00) with an increase of 1 standard deviation (SD) in negative affect. Linear regression analysis was employed to identify which negative affect scales were significantly linked with INR (patients with OAC therapy) and the various clotting factors. We computed three models: Model 1 did not adjust for covariates showing zero-order correlations; Model 2 adjusted for a priori selected covariates gender, age, BMI and SES because these might affect coagulation activity, warfarin metabolism and adherence [4,6,24]; Model 3 made further adjustments for hematocrit and CRP in order to account for potential stress-triggered hemoconcentration and inflammatory responses that both might affect the levels of clotting factors [9,25]. All covariates were forced into models in one block. Because patients with no OAC therapy contributed a larger sample (n = 148) than patients under OAC therapy (n = 42), the mere significance level of a correlation coefficient is misleading. Therefore, we also used Fisher r-to-z transformation [26] to examine whether the correlations between negative affect scales and clotting factors would significantly differ between the two patient groups.

Results

Patient characteristics

Table 1 shows the characteristics of the 190 patients with a previous VTE stratified by current OAC therapy (yes/no). The proportion of men and the BMI were both higher, whereas the time elapsed since the last VTE was shorter in patients with OAC therapy than in those without. Expectedly, the INR was higher and clotting activities of FII, FV, FVII and FX were all lower in patients with vs. those without OAC therapy. In addition, hematocrit was relatively higher in patients with OAC therapy, whereas ALAT, creatinine and CRP levels as well as negative affect measures were all similar in the two patient groups.

Table 1.   Characteristics of 190 patients with a previous venous thromboembolism (VTE)
VariableOral anticoagulant therapy P-value
  1. ALAT, alanin transaminase; DVT, deep venous thrombosis; INR, international normalized ratio; PE, pulmonary embolism; VTE, venous thromboembolic event. Values are given as percentages or mean ± SD with range in parentheses. Statistical analysis used the chi-square test and Mann–Whitney U-test.

 No (n = 148)Yes (n = 42) 
Male gender (%)54.773.80.027
Age (years)47.4 ± 12.9 (18–80)44.8 ± 14.8 (18–75)0.282
Body mass index (kg m−2)26.7 ± 4.7 (17.7–47.2)28.1 ± 4.9 (19.8–45.5)0.042
Time since last VTE (months)20.0 ± 38.7 (3–328)10.8 ± 17.2 (3–85)< 0.001
Diagnoses (%)
 DVT60.842.90.112
 PE31.847.6
 DVT plus PE7.49.5
Educational level (%)
 Graduate school (university)43.238.10.820
 Tertiary education42.647.6
 Vocational education14.214.3
Depression (score)3.5 ± 3.2 (0–16)3.7 ± 2.9 (0–12)0.559
Anxiety (score)5.2 ± 3.5 (0–16)6.1 ± 3.5 (1–17)0.168
Worrying (score)24.3 ± 10.1 (4–51)27.3 ± 10.9 (12–51)0.051
Anger (score)7.3 ± 7.3 (0–30)8.4 ± 8.5 (0–32)0.662
Hostility (score)7.4 ± 4.9 (0–20)8.5 ± 5.0 (1–20)0.203
INR < 1.00 (%), ≥ 1.00 (%)41.2, 58.82.58 ± 0.64 (1.52–4.25) 
Fibrinogen (g L−1)3.17 ± 0.73 (1.70–5.21)3.30 ± 0.81 (1.49–5.31)0.262
FII:C (%)103.2 ± 15.2 (68–147)27.1 ± 8.2 (14–48)< 0.001
FV:C (%)109.1 ± 19.4 (60–174)96.4 ± 17.0 (65–137)< 0.001
FVII:C (%)102.5 ± 23.3 (35–163)27.6 ± 11.4 (11–65)< 0.001
FX:C (%)104.4 ± 16.6 (64–151)15.0 ± 5.2 (8–32)< 0.001
Hematocrit (%)42.9 ± 3.1 (36–50)44.1 ± 2.7 (37–50)0.019
ALAT (U L)25.1 ± 13.4 (7–116)34.1 ± 23.9 (6–109)0.078
Creatinine (μ mol L−1)73.2 ± 10.9 (53–104)75.4 ± 12.3 (53–95)0.241
C-reactive protein (mg L−1)1.91 ± 1.69 (1–9)2.10 ± 2.00 (1–9)0.806

Associations amongst sociodemographic and negative affect measures

In the entire sample of 190 patients, all negative affect scales showed significant direct correlations with each other (r-values between 0.31 and 0.70, all P-values < 0.001). The smallest correlation coefficients emerged between hostility and the other four negative affect measures (all r-values < 0.39), whereas correlation coefficients amongst depression, anxiety, worrying and anger scales were all > 0.44.

Lower SES was associated with greater levels of worrying (r = −0.15, P = 0.038) and hostility (r = −0.21, P = 0.004), but not with the other domains of negative affect (all P-values > 0.10). Women reported worrying more than men (26 ± 11 vs. 23 ± 10, P = 0.036), younger patients had more anxiety symptoms (r = −0.23, P = 0.001) and those with a higher BMI had more depressive symptoms (r = 0.18, = 0.015). There were no significant associations of hematocrit (all P-values > 0.06), CRP (all P-values > 0.32), time since last VTE (all P-values > 0.31) and diagnosis of VTE (all P-values > 0.28) with the negative affect scales.

Association between negative affect and prothrombotic measures in patients without oral anticoagulant therapy

International normalized ratio  Out of a total of 148 patients without OAC therapy, there were 61 patients with an INR < 1.00 and 87 patients with an INR ≥ 1.00. The time since last VTE did not significantly differ between these two patient groups (P = 0.88). Table 2 shows three logistic regression models for the association between an increase of 1 SD in measures of negative affect and the odds of a patient with no OAC therapy to have an INR < 1. Irrespective of covariate adjustment, the odds for a reduced INR increased significantly between 44% and 78% for an increase of 1 SD in symptom levels of depression, anxiety and worrying, respectively (Models 1–3). Greater hostility was significantly associated with increased odds of a reduced INR in the unadjusted Model 1, but not in the adjusted Models 2 and 3. The association between higher anger and the odds for reduced INR reached significance in the fully adjusted Model 3. Lower SES and in the majority of equations also greater age were significantly associated with increased odds of a reduced INR; gender, BMI, hematocrit and CRP were not significantly related to the odds of a reduced INR.

Table 2.   Odds for reduced international normalized ratio (INR) with increased negative affect in 148 patients with no oral anticoagulant therapy
Variables enteredDepressionAnxietyWorryingAngerHostility
  1. The columns show the odds ratio (95% CI) for a patient to have an international normalized ratio (INR) below 1.00 vs. an INR equal to or above 1.00 with 1 SD increase in negative affect with no adjustment for covariates (Model 1), with adjustments for gender, age (in 5 year steps), body mass index, socioeconomic status (Model 2) as well as with additional adjustments for hematocrit and C-reactive protein (Model 3). The totally explained variance of the models is expressed as Nagelkerke R2. Significance level: *P < 0.05, **P < 0.01, ***P < 0.001.

Model 1
 Affect measure (1 SD)1.68 (1.18–2.40)**1.44 (1.03–2.02)*1.64 (1.15–2.32)**1.28 (0.92–1.78)1.42 (1.01–1.99)*
 Model coefficients
Model summary
χ2 = 9.01, d.f. = 1**
R2 = 0.080
χ2 = 4.61, d.f. = 1*
R2 = 0.041
χ2 = 8.17, d.f. = 1**
R2 = 0.072
χ2 = 2.16, d.f. = 1
R2 = 0.020
χ2 = 4.27, d.f. = 1*
R2 = 0.038
Model 2
 Male gender0.45 (0.21–1.00)0.52 (0.25–1.11)0.56 (0.26–1.19)0.47 (0.22–1.01)0.51 (0.24–1.08)
 Age1.18 (1.01–1.37)*1.17 (1.01–1.36)*1.16 (1.00–1.35)*1.21 (1.03–1.42)*1.14 (0.98–1.32)
 Body mass index0.96 (0.88–1.04)0.98 (0.90–1.05)0.97 (0.90–1.05)0.98 (0.90–1.05)0.98 (0.91–1.05)
 Socioeconomic status2.05 (1.20–3.48)**2.13 (1.27–3.59)**2.00 (1.19–3.37)**2.05 (1.22–3.45)**1.98 (1.16–3.35)*
 Affect measure (1 SD)1.77 (1.22–2.58)**1.49 (1.05–2.11)*1.56 (1.08–2.24)*1.46 (0.99–2.13)1.24 (0.86–1.79)
 Model coefficients
Model summary
χ2 = 25.93, df = 5***
R2 = 0.217
χ2 = 21.16, df = 5**
R2 = 0.180
χ2 = 22.09, df = 5**
R2 = 0.187
χ2 = 20.02, df = 5**
R2 = 0.170
χ2 = 17.52, df = 5**
R2 = 0.150
Model 3
 Male gender0.42 (0.14–1.24)0.42 (0.14–1.21)0.45 (0.15–1.32)0.37 (0.12–1.09)0.42 (0.14–1.22)
 Age1.17 (1.00–1.37)1.17 (1.00–1.36)*1.16 (1.00–1.35)1.21 (1.03–1.43)*1.03 (1.00–1.06)
 Body mass index0.95 (0.83–1.03)0.96 (0.88–1.05)0.96 (0.88–1.05)0.96 (0.88–1.05)0.97 (0.89–1.05)
 Socioeconomic status2.09 (1.23–3.58)**2.18 (1.29–3.69)**2.04 (1.21–3.45)**2.09 (1.24–3.54)**2.01 (1.12–3.42)*
 Hematocrit1.03 (0.87–1.23)1.06 (0.90–1.26)1.06 (0.89–1.25)1.07 (0.90–1.26)1.05 (0.89–1.24)
 C-reactive protein1.10 (0.86–1.40)1.08 (0.85–1.36)1.06 (0.84–1.34)1.08 (0.86–1.37)1.06 (0.84–1.33)
 Affect measure (1 SD)1.78 (1.22–2.59)**1.51 (1.06–2.16)*1.57 (1.09-2.26)*1.49 (1.01–2.20)*1.25 (0.86–1.80)
 Model statistics
Model summary
χ2 = 26.60, df = 7***
R2 = 0.222
χ2 = 21.99, df = 7**
R2 = 0.186
χ2 = 22.71, df = 7**
R2 = 0.192
χ2 = 20.94, df = 7**
R2 = 0.178
χ2 = 18.06, df = 7*
R2 = 0.155

Clotting factors  Patients with a reduced INR had, compared with those with a normal/elevated INR, significantly higher levels of fibrinogen (3.46 ± 0.74 g L−1 vs. 2.96 ± 0.65 g L−1, P < 0.001), FV:C (119 ± 19% vs. 102 ± 17%, < 0.001), FVII:C (117 ± 20% vs. 92 ± 20%, P < 0.001) and FX:C (108 ± 16% vs. 102 ± 16%, = 0.026), but similar levels of FII:C (105 ± 14% vs. 102 ± 16%, = 0.17). Women had higher fibrinogen levels than men (3.31 ± 0.78 g L−1 vs. 3.05 ± 0.67 g L−1, P = 0.049). Age showed direct associations with fibrinogen (r = 0.35, P < 0.001), FV:C (r = 0.27, = 0.001) and FVII:C (r = 0.23, P = 0.006). Lower SES was associated with higher levels of fibrinogen (r = −0.17, P = 0.035), FV:C (r = −0.21, P = 0.010), FVII:C (r = −0.20, P = 0.016) and FX:C (r = −0.18, P = 0.026). Those with a higher BMI had higher levels of fibrinogen (r = 0.24, P = 0.003), FVII:C (r = 0.19, P = 0.022) and FX:C (r = 0.20, P = 0.015). Higher CRP levels were associated with higher levels of fibrinogen (r = 0.46, P < 0.001), FII:C (r = 0.33, P < 0.001), FV:C (r = 0.18, P = 0.026), FVII:C (r = 0.29, P < 0.001) and FX:C (r = 0.25, P = 0.002). The time since last VTE (all P-values > 0.15) and hematocrit (all P-values > 0.12) were not significantly associated with any clotting factor.

Table 3 shows the results of the unadjusted (Model 1) and multivariate adjusted (Models 2 and 3) linear regression analysis for the association between the negative affect scales and individual clotting factors in the 148 patients with no OAC therapy (cf. columns labeled with ‘OAC−’). Levels of depression, worrying, anger and hostility, but not of anxiety, all showed a significant direct association with FVII:C after taking into account covariates (Models 2 and 3). Depression, worrying and hostility, but not anxiety and anger, also showed significant direct relations with FVII:C in the unadjusted Model 1.

Table 3.   Associations between negative affect domains and prothrombotic measures within and between the two patient groups
Affect domainModelINRFibrinogenFII:CFV:CFVII:CFX:C
OAC+OAC+OAC−OAC+OAC−OAK+OAC−OAC+OAC−OAC+OAC−
  1. INR, International Normalized Ratio; OAC+, 42 patients with oral anticoagulant (OAC) therapy; OAC−, 148 patients with no OAC therapy. For the within-group comparison between psychometric and prothrombotic measures, partial correlation coefficients (r) are given without adjustment for covariates (Model 1), after adjusting for gender, age, body mass index and socioeconomic status (Model 2), and with additional adjustments for hematocrit and C-reactive protein (Model 3); significance level: *P < 0.05, **P < 0.01, ***P < 0.001. The significance of the between-group comparison for the associations between psychometric and prothrombotic measures is given by the z-scores and exact P-values from the Fisher r-to-z transformation.

Depression1 r = −0.10 r = 0.22 r = 0.02 r = 0.07 r = 0.09 r = 0.05 r = 0.10 r = −0.12 r = 0.17* r < 0.01 r = 0.15
  z = 1.129, P = 0.259 z = −0.112, P = 0.911 z = −0.279, P = 0.780 z = −0.162, P = 0.109 z = −0.816, P = 0.415
2 r = −0.18 r = 0.06 r = 0.01 r = 0.11 r = 0.08 r = −0.09 r = 0.09 r = −0.02 r = 0.17* r = 0.02 r = 0.12
  z = 0.278, P = 0.781 z = 0.168, P = 0.867 z = −1.001, P = 0.317 z = −1.063, P = 0.288 z = −0.558, P = 0.577
3 r = −0.20 r = 0.10 r = 0.05 r = 0.14 r = 0.10 r = −0.08 r = 0.10 r < 0.01 r = 0.18* r = 0.03 r = 0.12
  z = 0.279, P = 0.780 z = 0.225, P = 0.822 z = −1.001, P = 0.317 z = −0.998, P = 0.318 z = −0.502, P = 0.616
Anxiety1 r = 0.10 r = −0.05 r < 0.01 r = −0.12 r = 0.12 r = 0.05 r = 0.04 r = −0.10 r = 0.10 r = −0.26 r = 0.14
  z = −0.278, P = 0.781 z = −1.337, P = 0.181 z = 0.056, P = 0.956 z = −1.112, P = 0.266 z = −2.257, P = 0.024
2 r = 0.07 r = −0.09 r = 0.03 r = −0.08 r = 0.12 r = 0.09 = 0.07 r = −0.02 r = 0.12 r = −0.24 r = 0.13
  z = −0.667, P = 0.505 z = −1.113, P = 0.266 z = 0.112, P = 0.911 z = −0.779, P = 0.436 z = −2.082, P = 0.037
3 r = 0.04 r = −0.02 r = 0.03 r = −0.03 r = 0.14 r = 0.14 r = 0.08 r = 0.04 r = 0.14 r = −0.22 r = 0.15
  z = −0.277, P = 0.782 z = −0.948, P = 0.343 z = 0.337, P = 0.736 z = −0.559, P = 0.576 z = −2.078, P = 0.038
Worrying1 r = 0.08 r = 0.08 r = 0.03 r = −0.20 r = 0.09 r = −0.02 r = 0.10 r = −0.12 r = 0.25** r = −0.32* r = 0.13
  z = 0.278, P = 0.781 z = −1.624, P = 0.104 z = −0.667, P = 0.505 z = −2.084, P = 0.037 z = −2.339, P = 0.019
2 r = 0.04 r = −0.04 < −0.01 r = −0.17 r = 0.05 r = −0.05 r = 0.09 r = −0.02 r = 0.23** r = −0.32* r = 0.09
  z = −0.220, P = 0.826 z = −1.229, P = 0.219 z = −0.778, P = 0.437 z = −1.409, P = 0.159 z = −2.339, P = 0.019
3 r = 0.02 r = 0.04 r = −0.01 = −0.12 = 0.05 r = −0.01 r = 0.09 r = 0.05 r = 0.24** r = −0.31 r = 0.10
  z = 0.277, P = 0.782 z = −0.946, P = 0.344 z = −0.556, P = 0.578 z = −1.080, P = 0.280 z = −2.333, P = 0.020
Anger1 r = 0.04 r = 0.10 r = −0.08 r = −0.10 r = 0.04 r = 0.06 r = −0.03 r = −0.19 r = 0.09 r = −0.12 r = 0.15
  z = 1.001, P = 0.317 z = −0.778, P = 0.437 z = 0.499, P = 0.618 z = −1.567, P = 0.117 z = −1.506, P = 0.132
2 r = 0.01 r = 0.17 r = 0.02 r = −0.08 r = 0.06 r = 0.10 r = 0.03 r = −0.17 r = 0.18* r = −0.10 r = 0.13
  z = 0.841, P = 0.400 z = −0.777, P = 0.437 z = 0.390, P = 0.697 z = −1.938, P = 0.053 z = −1.281, P = 0.200
3 r = 0.10 r = 0.09 r = 0.05 r = −0.10 r = 0.10 r = 0.10 r = 0.04 r = −0.22 r = 0.21* r = −0.13 r = 0.16
  z = 0.223, P = 0.824 z = −0.112, P = 0.266 z = 0.344, P = 0.738 z = −2.422, P = 0.015 z = −1.619, P = 0.105
Hostility1 r = 0.10 r = 0.25 r = 0.13 r = −0.07 r = 0.16 r = 0.02 r = 0.16* r = −0.16 r = 0.35*** r = 0.06 r = 0.15
  z = 0.691, P = 0.489 z = −1.283, P = 0.200 z = −0.784, P = 0.433 z = −2.921, P = 0.003 z = −0.505, P = 0.614
2 r = 0.08 r = 0.21 r = 0.05 r = −0.09 r = 0.12 r = −0.09 r = 0.07 r = −0.17 r = 0.30*** r = 0.05 r = 0.10
  z = 0.904, P = 0.366 z = −1.169, P = 0.242 z = −0.889, P = 0.374 z = −2.668, P = 0.008 z = −0.279, P = 0.780
3 r = 0.11 r = 0.26 r = 0.04 r = −0.05 r = 0.13 r = −0.06 r = 0.07 r = −0.15 r = 0.31*** r = 0.06 r = 0.11
  z = 1.253, P = 0.210 z = −1.002, P = 0.316 z = −0.772, P = 0.470 z = −2.615, P = 0.009 z = −0.279, P = 0.780

Association between negative affect and prothrombotic measures in patients with oral anticoagulant therapy

In the 42 patients under OAC therapy, continuous measures of INR showed a significant inverse association with FII:C (r = −0.84, P < 0.001), FVII:C (r = −0.82, < 0.001) and FX:C (r = −0.82, P < 0.001), but not with fibrinogen (r = −0.04, = 0.82) and FV:C (r = 0.27, P > 0.08). Age was significantly associated with FV:C (r = 0.43, = 0.005) and BMI was associated with fibrinogen (r = 0.36, p = 0.021). CRP showed direct relationships with fibrinogen (r = 0.52, P < 0.001), FII:C (r = 0.33, = 0.033) and FVII:C (r = 0.36, P = 0.021). Gender (all P-values > 0.28), hematocrit (all P-values > 0.25) and time since last VTE (all P-values > 0.39) were all not significantly associated with any prothrombotic measure.

Table 3 (cf. columns labeled with ‘OAC+’) shows that there was an inverse relationship between worrying and FX:C without covariate adjustment (Model 1) and also after adjusting for gender, age, BMI and SES (Model 2); however, additional adjustment for hematocrit and CRP yielded this relationship non-significant (Model 3). There were no significant associations between negative affect scales and INR (continuous measures), fibrinogen, FII:C, FV:C and FVII:C with and without covariate adjustment.

Comparison of associations between negative affect and clotting factors in patients with vs. without oral anticoagulant therapy

Several associations between the negative affect scales and clotting factors turned out to be significantly different between patients with OAC therapy and those without OAC therapy (z-scores > 1.960, P-values < 0.050). As can be seen in Table 3, the relationship of anger and hostility with FVII:C showed a group difference in the fully adjusted Model 3 and, for hostility, also in Models 2 and 3. The relationship between worrying and FVII:C was different between groups in the unadjusted Model 1. Moreover, anxiety and worrying showed group differences in their relation with FX:C in the unadjusted and adjusted Models 1–3.

Discussion

We found that in patients who had discontinued OAC after having experienced VTE, depression, anxiety, worrying and anger were associated with a 50%–80% increased relative risk of having an INR below 1.00. A similar finding emerged for hostility, although that association did not withhold adjustment for covariates. In terms of individual clotting factors, negative affect scales were significantly correlated with FVII:C, but not with the other assessed clotting factors; this may not seem surprising as FVII is the clotting factor that has the profoundest impact on PT [1]. Interestingly enough, low SES was associated with both reduced INR and elevated clotting factor activities, a finding that is consistent with the literature showing that low social position is related to a prothrombotic state [27]. Taken together, our results concur with the notion that a negative affect across a range of domains is associated with higher coagulation activity irrespective of important confounds, thereby potentially increasing the risk of incident and recurrent thrombosis [4].

While coagulation changes with negative affect constructs investigated in the present study have extensively been discussed to be involved in atherothrombotic cardiovascular disease [2,27,28], evidence for psychological distress to be an important contributing factor to VTE is only emerging. Perceived stress [29], depressive mood and a low positive affect [30] as well as traumatic stress experiences [31] may all increase the risk of incident VTE. Although not studied here, epidemiologic and experimental work clearly shows that altered autonomic nervous system function, including increased sympathetic activation and decreased vagal tone, might link both psychological stress and negative affect (‘distress’) with enhanced clotting [4]. Our findings thus may fill a gap in a proposed cascade leading from challenging events to the perception of these as distressful and autonomic responses resulting downstream in activation of the extrinsic coagulation pathway (i.e. reduced INR) in individuals without OAC therapy.

It is a strength of this study that we investigated the association between psychological distress and INR in patients with previous VTE, whereas the aforementioned studies assessed incident VTE in relation to stressful life events and distress [29–31]. Whether the relationship between increased psychological distress and reduced INR will ultimately increase the risk of recurrent VTE remains to be seen. Our data further suggest that, except from hostility, the assessed negative affect constructs were of equal importance in their relation to reduced INR.

In patients under OAC therapy, we did not find a significant association between negative affect and INR both in the uncontrolled and controlled analyzes. We cannot exclude the possibility of limited statistical power to explain this. Specifically, in fully adjusted models, depressive symptoms explained 4.0% of the variance in the INR in patients with OAC therapy and 3.6% of variance in FVII:C in those without OAC therapy, but only the latter result turned out to be statistically significant. However, in terms of individual clotting factors, we found that patients under OAC therapy with greater levels of worrying had significantly lower FX:C, but only when CRP was not controlled for. This observation would concur with the notion that the innate immune response accompanying psychological distress attenuates the metabolism of oral anticoagulant drugs so to increase the inhibition of vitamin K-dependent clotting factors [11]. Moreover, several of the associations of negative affect with FVII:C and FX:C were significantly different between patients with OAC therapy and those without, even after taking CRP levels into account.

Our findings yield evidence for a differential relationship between psychological distress and measures of the extrinsic coagulation pathway depending on whether VTE patients were on OAC therapy or not. We found that clotting factor activity in relation to negative affect was comparably lower in patients under OAC therapy compared with those who had discontinued OAC therapy. In the former group, stressors severe enough to elicit substantial amounts of negative affect might be suspected to contribute to clinically relevant fluctuations in the INR. From a clinical perspective, patients could be advised to seek psychological support from their primary care provider should they observe unexpected fluctuations, particularly so an increase in their INR at times of feeling ‘stressed out’. Also, health care providers may want to ask patients actively about stress and negative feelings as part of the clinical management of unexpected INR fluctuations.

The study has several limitations inherent in its clinical, observational and cross-sectional design. The results might not be generalizable to individuals without a previous VTE. Patients were not randomized to discontinuation of OAC therapy and the trend for higher worrying in those with OAC therapy might relate to the ongoing requirement for being medicated, but might also exert greater effects on coagulation. We did not assess adherence to OAC therapy which might have been lower in participants with elevated levels of depressive symptoms [32] and also in relation to illness perceptions and beliefs [33], thereby influencing vitamin K-dependent prothrombotic measures in those under OAC therapy. The performance of this study within the routine of a tertiary clinical setting did not allow us to systematically assess various health behaviors (e.g. smoking, physical activity and alcohol consumption) that might affect coagulation [27]. Nonetheless, based on laboratory work-up and history taking, we were able to partially account for such confounding. Our data cannot establish a causal relationship leading from negative affect to alterations in the INR and clotting factors. There is a bidirectional relationship between inflammation and psychological distress [34,35] and, as was the case in this sample in which CRP correlated with several clotting factors, inflammation also interacts with coagulation [36]. Various psychometric instruments are available to assess negative affect all of which have their pros and cons. As an example, its shortness makes the HADS a convenient and widely used measure to assess anxiety and depression in clinical populations, but the usefulness of the HADS for this purpose has recently been challenged [37].

To sum up, we found that psychological distress is associated with reduced INR and increased FVII:C in VTE patients who have discontinued OAC therapy. In contrast, VTE patients under OAC therapy had comparably lower activity in some vitamin K-dependent clotting factors of the extrinsic coagulation pathway in relation to elevated psychological distress. The question whether these relations may contribute to the risk of recurrent VTE after patients have discontinued OAC therapy as well as INR fluctuations in those under OAC therapy warrants further studies.

Acknowledgements

The authors thank Shekibe Sakiri, Roswitha Krogull, Jessica Thomet, Kristina Joder, Ursula Wipf, Silvia Andrey, Melanie Imbach, Jerry van Mook and Roberto Rizzi for technical assistance during the conduct of the study and Annette Kocher for editorial support.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Ancillary