Ingrid Pabinger, Clinical Division of Haematology and Haemostaseology, Department of Medicine I, Medical University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria. Tel.: +43 1 40400 4448; fax: +43 1 40400 4030. E-mail: email@example.com
Summary. Background: Data on the survival of individuals with hereditary thrombophilia are rare and only come from retrospective studies. Objective: The aim of the present study was to assess mortality in individuals with known thrombophilia with and without a history of thrombosis in comparison to a control group. Patients/methods: The European Prospective Cohort on Thrombophilia (EPCOT) study is a prospective, multi-center observational study performed to assess the risk of thrombosis in persons with inherited thrombophilia. In an extension of the present study, the vital status was assessed in 1240 individuals with thrombophilia (mean age 40.9 years, 59% women, 196 with antithrombin, 341 with protein C, 276 with protein S-deficiency, 330 with factor (F)V Leiden and 97 with combined defects, and 62% with a history of venous thrombosis [VT]) and 875 controls (mean age 42.5 years, 48% women, 7% with a history of VT). Results: Seventy-two individuals with thrombophilia and 45 controls died during follow-up. The risk of death, adjusted for gender, thrombosis history and center, was not associated with thrombophilia (hazard ratio [HR] thrombophilia individuals vs. controls: 1.09, 95% confidence interval [CI] 0.66–1.78). When individuals with thrombophilia were evaluated separately, a history of thrombosis was not associated with mortality: the risk of death after adjustment for gender, anticoagulation and center was HR 0.79 (95% CI, 0.41–1.54). Conclusions: No increased risk of death in individuals with thrombophilia, not even in those with a history of thrombosis, was observed.
Acute venous thrombosis (VT) is a potentially life-threatening disease with an incidence of around 1–2 per year in 1000 in the general population . Death may occur as a result of a massive pulmonary embolism (PE) or because of recurrent emboli in the pulmonary vasculature leading to heart failure . In some patients death occurs as a result of certain atypical sites of thrombosis, such as thrombosis of the cerebral or the mesenteric veins.
Mortality within 7 days after VT has been reported to be as high as 25% . In the International Cooperative Pulmonary Embolism Registry (ICOPER) the outcome in patients with a PE was studied . Among 2454 patients a mortality rate of 11.4% after 2 weeks and 17.4% after 3 months was found. The most frequently reported causes of death were a recurrent PE, cancer, sudden cardiac death and respiratory failure. In a cohort of patients that had been included in an interventional trial comparing 6 weeks to 6 months of anticoagulation  their 10-year mortality risk was 28.5%. This cohort also included patients with cancer and with other chronic diseases and the mean age was relatively high. In another cohort, the risk of a fatal PE was 0.19–0.49 events per 100 person years after discontinuation of anticoagulation .
Studying mortality in thrombophilia is difficult as a consequence of the rarity of these genetic variations, particularly antithrombin, protein C and protein S deficiency. A first approach was to develop a new method, the ‘Family Tree Mortality Ratio’, in which historical cohorts were constructed, based on pedigrees and derived genotypes. This method allowed observations to go back to the early 1800s and compared mortality in carriers with population figures. Such studies found no excess mortality for antithrombin deficiency and protein C deficiency [7–9]. A second approach, which we present in the present study, is to construct a cohort from patients registered at many centers.
The European Prospective Cohort on Thrombophilia (EPCOT) study was initiated in 1994 to prospectively study the risk of (recurrent) venous thrombosis in a large European cohort with a deficiency in antithrombin, protein C or S or the factor (F)V Leiden. The aim of the present study was to analyze mortality in individuals with thrombophilia included in EPCOT compared with EPCOT control individuals.
The design and aims of the EPCOT study have been described previously . Briefly, individuals with a deficiency in antithrombin, protein C or protein S, or FV Leiden were recruited in nine European centers between July 1994 and January 1997. After its discovery in 1996, the prothrombin G20210A variation was determined in the EPCOT participants. In addition, a group of controls consisting of partners, friends or acquaintances of our participants were enrolled. These controls did not undergo thrombophilia testing. Controls were excluded, if they were known to have heritable thrombophilia or if they were related to a participant with an inherited thrombotic defect. There were no exclusion criteria for individuals with inherited thrombophilia: we included those who were symptomatic as well as those who were asymptomatic, and those who received anticoagulant treatment as well as those who did not. The present study was approved by the local Ethics Committees for the various centers, and all individuals with thrombophilia and controls gave their written informed consent. All nine EPCOT-centers were asked to participate in the present study, and seven consented. These seven centers had previously included 2233 participants in EPCOT, 1289 with thrombophilia and 944 controls. Of these, the vital status could be ascertained in the EPCOT extension study in 1.240 individuals with thrombophilia (96%) and 875 controls (93%) by personally contacting the study participants and/or their relatives. Observation started with inclusion in the EPCOT between March 1994 and September 1997, continued after the official end of EPCOT study in January 2001 and ended at death or on 31 December 31 2006, whichever came first. In addition to the vital status, we inquired whether an individual with thrombophilia or a control had experienced a thrombotic event after the end of the initial EPCOT follow-up period. Thrombotic events were defined as follows: the term ‘major’ summarized objectively a confirmed deep vein thrombosis (DVT) and PE, a ‘minor’ event was superficial thrombophlebitis (STP). Information on long-term anticoagulation had been collected during the EPCOT observation period. We defined a patient as being on long-term anticoagulation when this patient was on continuous anticoagulation at the end of the original EPCOT observation period. Furthermore, we gathered information on the causes of death. Data were analyzed by I.P., J.L. and A.K.; all authors had access to primary clinical data.
The participants’ characteristics were evaluated using descriptive statistics in SPSS (SPSS 16.0; SPSS Inc., Chicago, IL, USA). For numeric variables mean value and ranges were given. Survival analyzes were performed in SAS (Version 9.2; SAS Institute Inc., 2002–2008, Cary, NC, USA) to evaluate mortality in individuals with thrombophilia and controls. If still alive, the participant was censored at the 31 of December 2006 – otherwise, a subject’s age at death was considered the relevant end point. As individuals with thrombophilia and controls were only observable after study entry a so-called ‘delayed entry model’ was employed . This enabled estimation of survival probabilities using the Kaplan–Meier method and comparison via Cox regression models to be restricted to subjects who had already entered the study. Univariate and multivariate Cox regression models were used to estimate the ratio of death rates in individuals with thrombophilia and controls, and to evaluate the effect of gender and thrombosis history on survival. In defining an individual as having a thrombosis history, we just included those with a history of a major thrombotic event. Furthermore, individuals with thrombophilia were analyzed in a separate Cox regression model to assess the effects of the different types of thrombophilia (deficiencies of protein C, protein S or antithrombin, FV Leiden and combined defects). In all Cox regression models, the study center was entered as a stratification factor. Results of Cox regression analyzes are described by estimates of the risk ratios with 95% confidence intervals (CIs). In an additionally performed sensitivity analysis, using the block re-sampling method as described by Davison and Hinkley , the bootstrap percentile method was used to calculate 95% CIs for the rate ratios in order to allow for the dependencies of participants from the same family.
Data from 2115 participants, 1240 thrombophilic individuals (59%) out of 469 families and 875 controls (41%), were available for analyzes. The mean age at entry into EPCOT was 40.9 years (range 1.9–91.4) in individuals with thrombophilia and 42.5 years (range 2.9–86.8) in controls. Table 1 shows the general characteristics of study participants, the distribution with regard to the type of thrombophilia and their thrombosis history. Thrombosis history (major and minor) was present in 210/340 (59%) of subjects with protein C deficiency, in 192/330 (58%) with FV Leiden, in 172/275 (63%) of those with protein S deficiency, in 131/194 (68%) of those with antithrombin deficiency and in 67/96 (70%) of subjects with a combination of thrombophilia. Of all our participants, 417 individuals (34%) with thrombophilia were on long-term anticoagulation. The median observation period was 12.30 years (interquartile range [IQR]: 11.77–12.62) in patients and 12.30 years (IQR: 11.81–12.59) in controls.
Table 1. Characteristics of study participants
Individuals with thrombophilia
Data on numbers and percent are given, if not stated differently. FVL, factor V Leiden.
Age, mean (range), years
Type of thrombophilia
Protein C-deficiency + FVL
Protein S-deficiency + FVL
Antithrombin-deficiency + FVL
Protein C + Protein S-deficiency
Protein S + Antithrombin-deficiency
Risk of death during follow-up
In all, 117 participants (5.5%) died during follow-up. In Table 2 the number of deaths among individuals with thrombophilia and controls is given in total and stratified for the presence or absence of a history of thrombosis and the various deficiency states. There were 72 (5.8%) deaths among thrombophilic individuals and 45 (5.1%) among controls. In both groups, more men than women died, slightly more persons with a history of VT died in comparison to those without.
Table 2. Numbers of deaths in individuals with thrombophilia and controls
Individuals with thrombophilia (N = 1240) (%)
Controls (N = 875) (%)
Data on numbers and percent are given. FVL, factor V Leiden; N, number.
Deceased with a positive history of thrombosis
Deceased with a negative history of thrombosis
Deceased per type of thrombophilia
Protein C-deficiency + FVL
Protein S-deficiency + FVL
Antithrombin-deficiency + FVL
The probability of survival of individuals with thrombophilia compared with controls is shown in Fig. 1. The probability of being alive at 70 years of age was 85% (95% CI, 81–90) in individuals with thrombophilia and 88% (95% CI, 83–93) in controls.
In a further analysis thrombophilia patients with and without a positive history of thrombosis were studied. Kaplan–Meier analyzes in all individuals with thrombophilia, and in male and female subjects separately, are shown in Fig. 2A–C, respectively. There was no difference in survival when individuals with thrombophilia were stratified into groups with a positive and negative thrombotic history: the probability of being alive at 70 years of age was 84% (95% CI: 79–90) in individuals with and 88% (95% CI, 80–96) in those without a history of thrombosis (Fig. 2A). Neither was there a difference when male and female participants with thrombophilia were analyzed separately. For men, the probability of being alive at 70 years of age was 80% (95% CI, 71–88) in those with and 77% (95% CI, 62–95) in those without a history of VT (Fig. 2B). The corresponding probabilities and 95% CI for women at 70 years of age with and without a history of thrombosis are 89% (83%–96%) and 95% (88–100) (Fig. 2C).
According to Cox regression analyzes the risk of death was not increased for individuals with thrombophilia compared with controls, the univariate hazard ratio (HR) was 1.14, 95% CI 0.78–1.67 and the multivariate HR was 1.09 (95% CI, 0.66–1.78, adjustment for gender and thrombosis history) (Table 3). Men clearly had a higher fatal risk in comparison to women (HR 1.77, 95% CI, 1.19–2.62).
Table 3. Cox-regression analysis, stratified by center for survival in individuals with thrombophilia vs. controls
HR, hazard ratio; CI, confidence interval. *Adjusted for gender and a history of thrombosis.
Individuals with thrombophilia vs. controls
In a separate analysis we compared thrombophilic patients with certain characteristics (Table 4). Thrombosis history was not a risk factor for increased mortality, the HR in multivariate analysis including gender and long-term anticoagulation was 0.79 with a 95% CI, of 0.41–1.54. Long-term anticoagulation did not decrease the risk of death (HR 1.44, 95% CI, 0.81–2.57). Again, also in the group of thrombophilic patients, men had a higher risk of death than women. Results of the additionally performed sensitivity analysis, which allows for the dependencies of participants from the same family by calculating bootstrap CIs, were comparable and revealed very similar 95% CIs for the HRs.
Table 4. Univariate and multivariate Cox regression in individuals with thrombophilia only, stratified by center
Five observations with missing thrombosis status and missing information on anticoagulation, respectively. HR, hazard ratio; CI, confidence interval.
Gender (male vs. female)
The various genetic defects in the group of participants with thrombophilia were analyzed separately. Each population with a specific defect was compared with the entire group of individuals with the other defects. The HR for participants with protein C deficiency was 0.70 (95% CI, 0.40–1.23), for those with FV Leiden 0.58 (0.32–1.06), with protein S deficiency 1.9 (1.08–3.34) and with antithrombin deficiency 1.65 (0.91–2.94).
Causes of death
In both groups, a large number of causes of death, including individuals in whom ‘senility’ was stated, were not specified (36 among individuals with thrombophilia and 30 among controls).
In the thrombophilia group, the known causes of death (n = 36) were an arterial thromboembolism (13 cases), cancer (12), a venous thromboembolism (one), an intracerebral hemorrhage (one), infection (six), kidney failure (two) and accident (one).
Among controls, known causes of death (n = 15) were an arterial thromboembolism (five cases), cancer (seven cases), infection, accident and liver cirrhosis (one case each). Neither a venous thromboembolism nor a hemorrhage appeared as causes of death in any control person.
Our prospective, controlled, multicenter and multinational study revealed that familial thrombophilia does not increase overall mortality, not even in the presence of a history of VT. These data confirm earlier historical studies [7,9,13,14] in individuals and families with antithrombin deficiency, protein C deficiency and FV Leiden. We found that women have a better prognosis for survival than men, irrespective of having thrombophilia or not. This difference is widely known in developed countries .
The results from this present study are relevant to clinical care. First of all, when individuals and family members with thrombophilia are seeking genetic counseling, they can be informed that a normal life expectancy is highly probable and that there are no indications for an impaired life expectancy, which may very likely have a positive influence on their quality of life by reducing anxiety. Second, a normal life expectancy strongly argues against primary prophylaxis in individuals with thrombophilia – a controversial issue that may again emerge, when new anticoagulants will become available in the future. Third, as a previous VT does not seem to influence long-term survival, this information has to be considered when patients are counseled on the duration of anticoagulation after the initial phase, i.e. after 3 months. Long-term anticoagulation is recommended in the most recent guidelines of the American College of Chest Physicians (ACCP) with a grade 1A recommendation in all patients with proximal VT or PE , independent of the presence of thrombophilia. It is crucial that, when deciding on long-term anticoagulation, which often means indefinite, even life-long anticoagulation, the risk of cerebral bleeding resulting in death or major impairment and the impaired quality of life are always balanced against the risk of a fatal VT.
Interestingly, persons from the control group with a history of VT had a higher mortality than those without. As it was not the aim of the present study to separately investigate the control group, we did not perform a separate statistical analysis. But it could be hypothesized that persons from the general population more often have an underlying disease that is associated with a shorter survival. It could also be the case that individuals with thrombophilia are better cared for, when thrombosis occurs and/or are better educated by their physicians on the symptoms of VT or PE, so diagnosis is not delayed, neither by the patient nor the physician. This might consequently contribute to the low risk of death in individuals with thrombophilia and VT or PE.
When individuals with different states of thrombophilia were investigated separately, those with protein S deficiency seemed to have a higher mortality rate than those with other types of thrombophilia. Whether this was just by chance or whether these persons definitely have a higher risk of mortality needs to be addressed in large future studies.
The data derived from the present study clearly have to be differentiated from survival analyzes in patients with acute venous thrombosis. We know from prospective studies and registries that these patients’ mortality is increased, specifically in the first weeks to months after the event as a result of recurrent VT or underlying diseases, such as cancer, respiratory or heart failure [2,4]. Individuals with thrombophilia frequently have manifestations of VT early in life without having additional diseases.
The present study has several limitations. One major limitation is that some participants were tested for thrombophilia about 30 years before the EPCOT study started. Those who died in that same period thus were not included in the present study. The treatment policy of long-term anticoagulation could have influenced the survival rate of our thrombophilia group. However, as we corrected for anticoagulation, we adjusted for a major influence of anticoagulant treatment on life expectancy. We did not test the control group for thrombophilia. A percentage of 5%–7% is expected to have thrombophilia, mainly FV Leiden or the prothrombin variation, so we cannot exclude a minor influence on the mortality data.
The results of the present study cannot be extrapolated to patients with severe underlying diseases, such as those with heart or respiratory failure or cancer. A decreased survival was noted in cancer patients with a history of VT in comparison to those without . The members of our thrombophilia arm were included because they had been diagnosed with thrombophilia only and their frequency of other major diseases therefore was likely to be identical to that of the general population.
A strong point of the present study is that our loss of follow-up was only around 6% and was well below 20% in each of the centers. Unfortunately, causes of death were unknown or not well characterized for 56% of the deceased persons, so maybe the number of deaths caused by venous thrombosis or by hemorrhage was slightly higher than shown in our results. This, however, cannot be assumed to influence the validity of our results, as mortality in the thrombophilia group was not elevated in comparison to the control group.
In conclusion, we found that survival was not decreased in individuals with inherited thrombophilia included in the EPCOT cohort, not even in the case of a history of VT.
I. Pabinger, C.Y. Vossen, F.R. Rosendaal: designed study, included patients, evaluated data, wrote manuscript and approved last version of manuscript; J. Lang: designed study, data management, statistical data analysis, evaluated data, wrote manuscript and approved last version of manuscript, J. Conard, M.C. García-Dabrio, W. Miesbach, C. Legnani, P. Svensson: included patients, evaluated data, corrected manuscript and approved last version of manuscript; A. Kaider: statistical data analysis, evaluated data, wrote manuscript and approved last version of manuscript.
We thank E. Briët, J. Fontcuberta, G. Palareti and I. Scharrer, who were principal investigators of EPCOT in Leiden/The Netherlands, Barcelona/Spain, Bologna/Italy and Frankfurt/Germany. We further thank S. Koder (Vienna), M. H. Horellou (Paris), L. Velmans (Leiden), I. de Jonge (Leiden) and J. C. Souto (Spain) for their contribution to the study. We would also like to thank T. Altreiter (Vienna) for proof reading the manuscript. The EPCOT-Study was supported by the European Commission, by a Biomed II grant no BMHI-CT94-1565 (coordinator F. R. Rosendaal).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.