F. R. Rosendaal, Departments of Clinical Epide-miology and Hematology, C9-P, Leiden University Medical Center, PO Box 9600, NL-2300 RC, Leiden, The Netherlands. Tel.:+31 71 5264037; fax:+31 71 5266994; e-mail: firstname.lastname@example.org
Summary. While the overall incidence of venous thrombosis is 1–2 per 1000 per year, it is close to 1% per year in the very old. The case–fatality rate of thrombosis is high in the elderly, particularly among those with cancer. The risk of major hemorrhage during anticoagulant treatment is also strongly age-dependent, contributing to the vulnerability of the old patient with thrombosis. From this perspective it is surprising that far fewer studies into the etiology and treatment of venous thrombosis have focused on the elderly than on young and middle-aged patients. In this review we discuss that, while environmental risk factors, such as immobilization and cancer, are important causes of thrombosis in the elderly, abnormalities of the coagulation system are equally, or even more, important than in young individuals. In addition to a review of the literature, new data are presented from the MEGA-study. Thrombosis in the elderly should be a focus of future studies.
As much as children are not small adults, the elderly are not just old adults. Etiology, diagnosis, prognosis and therapy may all be different in the elderly compared with young or middle-aged adults. For several diseases, risk factors have been shown to have reverse effects in old age, e.g. high serum cholesterol and hypertension no longer predict mortality among the very old, and may even have an opposite effect, which has been named reverse epidemiology [1–4]. Not only may risk factors act differently at various ages, but the results of their effects may also differ because of the exponential rise in incidence of venous thrombosis with age. Finally, risk factors may have a different prevalence in the old than in the young (e.g. immobilization, cancer and drug use). It is remarkable that while it can easily be shown that the majority of cases of thrombosis occur in the old, and a substantial number among the very old, studies into etiology and management are almost completely restricted to younger age groups.
In this review we deal with the determinants of thrombosis in general as well as in the elderly. We discuss the effect of the age-specific incidence of venous thrombosis, and finally present new data on the effect of two major risk factor groups, thrombophilia and malignancies, in elderly individuals, by analysis of a substudy of the Multiple Environmental and Genetic Assessment (MEGA) of risk factors for venous thrombosis.
Thrombosis is the formation of obstructive clots in arteries or veins, because of an inbalance of procoagulant, anticoagulant and fibrinolytic factors. In venous thrombosis a clot is formed in reaction to risk factors that are often acute and transient. Then the clot grows and may cause symptoms suddenly or, more often, after a period of several days to weeks.
Probably because of its association with sudden and transient risk factors, rather than insidious factors promoting slowly accumulating atherosclerosis, knowledge of the etiology of venous thrombosis preceded by far that of arterial disease: pregnancy and puerperium were recognized as determinants of venous thrombosis centuries ago, cancer in the mid-19th century, and the hereditary nature of venous thrombosis in the early 1900s [5–8]. The causes of thrombosis were first categorized by Virchow, who in the mid-1800s postulated three general causes: changes in the vessel wall, changes in the blood flow, and changes in the blood composition . Whereas for arterial disease, vessel wall changes (atherosclerosis) are by far the most important, in venous thrombosis stasis and hypercoagulability lead to clot formation and disease.
Venous thrombosis is a common disorder with a frequency of one to three individuals per 1000 per year [10–14]. It commonly appears as superficial or deep vein thrombosis of the leg and pulmonary embolism, and rarely in other locations (upper extremities, liver, cerebral sinus, retina, mesenteric). Venous thrombosis occurs about equally often in men and women, with higher rates among young women than men because of risk factors associated with reproduction (pregnancy, puerperium, oral contraceptives) [13–15]. The post-thrombotic syndrome occurring in 20% of patients with deep vein thrombosis may lead to a chronic handicap . One to ten per cent of venous thromboses proves fatal, with the highest risk of death in those with underlying malignancies [11,13].
Risk of thrombosis in the elderly
The incidence of venous thrombosis is strongly age-dependent: it is extremely uncommon (1 in 100 000 per year) in childhood, and rises to nearly 1% per year in very old age [12–15]. Table 1 shows the incidence of venous thrombosis among those aged 65 years and older as found in three large recent studies: a community-based study from western France, a cohort study from several sites in the USA (Cardiovascular Health Study), and a very large population-based study from North-Trøndelag in Norway [12,14,17]. The incidences are quite similar, at 3–4 per 1000 per year. As all three studies applied established criteria for thrombosis diagnosis, these estimates may be considered reliable, although cases of fatal pulmonary embolism may have been missed. The slightly lower estimates from the Cardiovascular Health Study may be due to self-selection of individuals for a cohort study that includes an intensive baseline examination, which may lead to an under-representation of those with morbid conditions (e.g. malignancies). For studies on a dynamic population, however, it may be difficult to completely exclude cases with a personal history of thrombosis, thus inflating the estimates.
Table 1. Incidence of first venous thrombosis in the elderly (65 + years)
Whereas the French study showed little difference in the incidences between elderly men and women, both the US and Norwegian studies showed a higher risk for elderly men than women, which was pronounced in the US study, but only modest in the much larger study from Norway (Fig. 1).
The incidence of venous thrombosis is < 1 per 1000 per year up to age 50, and then increases rapidly and exponentially (Fig. 1). This has major implications for the age distribution of the patient population with thrombosis, as well as for the effect of risk factors. From the rates of the Norwegian study, the age distribution of patients diagnosed with venous thrombosis can be calculated (Fig. 2)  (http://statline.cbs.nl). This shows that nearly 70% of all patients diagnosed with venous thrombosis are over 60, and nearly 25% are over 80 years old. From this perspective, it is astounding that very few studies have focused on the elderly and that many have specifically focused on young patients, for instance in the context of thrombophilia. It follows from these figures that risk factors may have vastly different effects among the elderly than among the young. For instance, a risk factor that quadruples the risk of venous thrombosis among the young, and doubles the risk among older individuals, causes far more additional cases of thrombosis in the older than in the younger age-groups. These are the relative risks associated with exogenous female hormone use, either as a contraceptive among young women, or as hormonal replacement therapy among older women [18–24]. When we compare two groups of women, aged 20 and aged 60, they have a baseline incidence of thrombosis of 0.25 per 1000 per year, and 1.60 per 1000 per year, respectively. So, when they start using hormones, the postmenopausal hormones will cause an excess of cases (1.6 per 1000) that is twice as large as for oral contraceptive use (0.75 per 1000). This implies that if efforts were aimed at preventing as many events of thrombosis as possible, they should be targeted at risk factors in the elderly.
While age is the strongest determinant of the risk of a first event of thrombosis, it is unclear why it is such a strong risk factor. Explanations that have been offered are more immobility, and presence of other or more risk factors and diseases than in young individuals, as well as decreased muscular tone and aging of the veins themselves, and particularly of the valves. It is likely that all of these contribute.
The case-fatality rate of thrombosis was, over 1 month, 6.4% in the Norwegian study of patients of all ages, and nearly double that rate, at 11% in the US cohort of middle-aged and elderly individuals [13,14]. Treatment of thrombosis is more hazardous in old than in young individuals: the risk of major hemorrhage during anticoagulation increases by nearly 50% for every decade of age .
Genetic risk factors for venous thrombosis
Causes of venous thrombosis are usually divided into environmental and genetic factors, of which the former includes both factors leading to stasis (surgery, immobilization, plaster casts, pregnancy) and to hypercoagulability (female hormones, lupus anticoagulants, malignancies) and the latter almost exclusively refers to a hypercoagulable state because of abnormalities in the coagulation system, such as deficiencies of natural anticoagulants [26,27]. Table 2 lists the risk factors for venous thrombosis.
Table 2. Risk factors for venous thrombosis
TAFI, thrombin activatable fibrinolysis inhibitor; TFPI, tissue factor pathway inhibitor; PCI, protein C inhibitor; PAI-3, plasminogen activator inhibitor-3.
High levels of factor VIII
Protein C deficiency
High levels of factor IX
Protein S deficiency
High levels of factor XI
Factor V Leiden (FVL)
High levels of fibrinogen
High levels of TAFI
Low levels of TFPI
Factor XIII 34Val
APC-resistance in the
Hormonal replacement therapy
absence of FVL
High levels of PCI (PAI-3)
Central venous catheters (thrombosis of arm)
Although familial thrombophilia had been recognized much earlier, the first family with an identified cause (i.e. antithrombin deficiency) was described in 1965 by Egeberg . Deficiencies of protein C and protein S were recognized as causes of heritable venous thrombosis in the 1980s [29,30]. These deficiencies are found in <1% of the population (antithrombin deficiency in only one per 5000) [31–33]. Even though they are strong risk factors for thrombosis, they are so rare that they are only found in a small percentage of patients with venous thrombosis [34–38]. Most of the information on these deficiencies stems from family studies and not from population-based studies. As it has been shown that an individual from a thrombophilic family with an identified thrombophilic defect has a higher thrombotic risk than an individual with the same defect but without a family history of thrombosis, probably because of unknown interacting factors, the data from family studies cannot be extrapolated to all carriers of these deficiencies . In a large multicenter cohort study of familial thrombophilia (European Prospective Cohort on Thrombophilia, EPCOT), in which 575 asymptomatic individuals from such families were followed for nearly 6 years, the annual rate of venous thrombosis was 8 per 1000 (highest, at 17 per 1000, in antithrombin deficiency), without a clear age-effect . In two retrospective family studies of antithrombin deficiency  and protein C deficiency , the incidence appeared to be slightly higher among those aged 45 years and older, at 1–2% per year.
While the deficiencies of natural anticoagulants are rare, factor (F) V Leiden (G1691A) and prothrombin G20210A are common variants with an overall incidence of carriers of 2–5% among Caucasians [41–44]. They are found in 6–20% of patients with deep vein thrombosis [42,45–47]. Both abnormalities increase the risk of thrombosis 3- to 8-fold [42,47], FV Leiden because it leads to APC-resistance and prothrombin 20210A because it is associated with elevated levels of prothrombin [42,48]. In the LITE (Longitudinal Investigation of Thromboembolism Etiology) study, which combines the data from the Atherosclerosis Risk in Communities (ARIC) study and Cardiovascular Health Study (CHS), the absolute risk of thrombosis was investigated for carriers of FV Leiden of different ages. The ARIC included individuals aged 45–64 years at baseline, and the CHS included individuals aged 65 and older. In a nested case–control analysis, FV Leiden led to a 4.6-fold increased risk (vs. non-carriers) in the ARIC study, and a 2.2-fold increased risk in the individuals aged 65 and older from the CHS 49. In a sub-study of the MEGA study, we included 238 patients aged over 70 with a first deep vein thrombosis or pulmonary embolism, and 112 partner controls who provided a DNA sample. Factor V Leiden or prothrombin 20210A was found in 31 patients (13%) and seven controls (6%), yielding an odds ratio of 2.2 (CI95: 0.9–6.2).
Given the exponential increase of the thrombosis incidence by age, these data indicate that FV Leiden will lead to a substantial number of cases of thrombosis in the elderly. A recent prospective follow-up study within families with prothrombin 20210A found a low risk of venous thrombosis in asymptomatic carriers (3.7 per 1000 per year), but three out of four events occurred in individuals aged 50 years and older, suggesting that this mutation does also increase risk in older individuals .
It has been shown that elevated levels of procoagulant factors (i.e. fibrinogen, prothrombin (FII), FVIII, FIX and FXI) as well as of thrombin activatable fibrinolysis inhibitor (TAFI), are associated with the risk of thrombosis [42,51–55]. Levels that are among the highest 10% of the population distribution (90th percentile) lead to a 2- to 3-fold increased risk of venous thrombosis. It is likely that high levels are, at least in part, genetically determined, which has been demonstrated for prothrombin levels (prothrombin 20210A), and FVIII levels (ABO blood group, as well as evidence of unidentified genetic factors) [42,51,56,57]. Mild hyperhomocysteinemia (over 18 μmol L−1), which is found in 5–10% of the population, also leads to a 2-fold increased risk of venous thrombosis, by an unknown mechanism [58–60]. Many coagulation factor levels, as well as plasma concentrations of homocysteine, increase with age . Data from the CHS indicate that elevated levels of coagulation factors remain a determinant of venous thrombosis in the elderly: high levels of FV antigen (> 75th percentile) doubled the risk of thrombosis compared with those in the lowest quintile , individuals with the highest levels of FVIII (> 95th percentile, 195 U dL–1) had a 3-fold increased risk compared with those with the lowest levels (< 25th percentile, 100 U dL−1), while a 1.6-fold increased risk was found for those with FVII levels in the highest 5% of the distribution (170 U dL−1) vs. those with normal levels (< 100 U dL−1) . In the same study, D-dimer levels proved a strong predictor of venous thrombosis, particularly in the oldest individuals . In the Leiden Thrombophilia Study, a high level of fibrinogen was a risk factor for thrombosis mainly in the elderly .
Although data on hypercoagulability in the elderly are scarce, the available reports indicate that abnormalities in the coagulation system increase the risk of venous thrombosis in the elderly as they do in young and middle-aged individuals. Because of the high incidence of high levels of procoagulant factors among the elderly, these are likely to be major contributors to the occurrence of venous thrombosis in these individuals.
Environmental causes of venous thrombosis
Both surgery and major trauma are strong risk factors for thrombosis. In the absence of thromboprophylaxis, orthopedic surgery of the hip and knee may lead to thrombosis in up to 30 to 50% of patients [65,66]. Nearly similar risks are conferred by abdominal surgery, gynecologic surgery and urologic surgery (in particular open prostatectomy) [67–69]. Multiple trauma, particularly when including head trauma, spinal injury, pelvic fractures, femoral fractures and tibial fractures, may cause thrombosis in up to half of the patients [70–72].
Nowadays, these high frequencies of thrombosis in these circumstances no longer occur because of the routine use of anticoagulant prophylaxis. Nevertheless, surgery and trauma are such strong risk factors that even with anticoagulant treatment total hip or knee replacement lead to symptomatic venous thrombosis in 1–3% of patients . Therefore, surgery remains a major cause of venous thrombosis. In the 1990s we observed in the Leiden region, where extended anticoagulant prophylaxis is routinely described for most surgical interventions, that 18% of cases of thrombosis were precipitated by a surgical intervention, and these interventions were associated with a 6-fold increased risk of venous thrombosis . In a recent analysis, over the period 1999–2004, including over 4000 patients with a first venous thrombosis and a similar number of controls, major surgery and orthopedic surgery still led to a 4-fold increased risk of symptomatic thrombosis .
Immobilization, either because of paralysis, bed rest, plaster casts or prolonged travel, leads to an increased risk of thrombosis [76–78]. Immobilization of the lower extremities interferes with the function of the calf musculature, in conjunction with the venous valves, of pumping the blood upstream against gravity. The risk is higher for seated than for supine immobilization. In London during World War II, the risk of pulmonary embolism was increased 6-fold shortly after air raids, during which people sought shelter in the underground where they waited in deckchairs. The risk went down after the chairs had been replaced by beds . Prolonged travel increases the risk of thrombosis 2- to 3-fold, particularly for those with prothrombotic mutations or who are obese . Overall, the risk remains low, at one per 4500 air travellers .
It is likely that many of these factors play a major role in thrombosis in the elderly, as immobilization because of medical conditions is frequent. In the RIETE registry of patients aged over 80 with venous thrombosis it was found that they often had been immobilized for more than 4 days, often had chronic obstructive lung disease and often had heart failure .
In 1865, Armand Trousseau observed the strong association between cancer and venous thrombosis  and lent his name to the Trousseau syndrome (i.e. recurrent thrombophlebitis at various and changing locations (saltans et migrans) indicating occult cancer, notably of the pancreas).
Cancer may cause thrombosis because of thrombogenicity of the tumor itself, for which recent research points to a role of tissue-factor loaded microparticles [83,84]. Large tumors may compress veins and cause thrombosis [85,86]. Furthermore, the disease may lead to weakness and immobility, and the treatment may be thrombogenic, with an overall 2-fold increased risk following chemotherapy vs. other therapies [87–89]. Central venous catheters to administer chemotherapeutics are the most important cause of thrombosis of the arm , and over 10% of patients with a central venous catheter develop symptomatic venous thrombosis of the upper extremity . Prothrombotic abnormalities, such as FV Leiden and prothrombin 20210A enhance this risk .
Adenocarcinomas confer a 3-fold higher risk of thrombosis than other types of cancer . The highest risk of thrombosis is found for pancreatic cancer, with an annualized risk over 10% . Overall, cancer increases the risk of thrombosis seven-fold, with the highest risk in the first months after diagnosis (more than 50-fold increased in first 3 months compared with individuals without cancer), and for those with distant metastases (20-fold increased vs. those without cancer) and for carriers of prothrombotic mutations (12-fold increased compared with those without cancer and such mutations) .
Because cancer is such a strong risk factor for thrombosis, a substantial number of patients who present with thrombosis have cancer, in some cases still undiagnosed. It has been estimated that between 5% and 20% of patients who present with thrombosis have cancer, with the highest prevalence among those with idiopathic thrombosis, of whom a substantial proportion (several per cent) is diagnosed with cancer in the year after the thrombosis [92,96–98]. Cancer is a major contributor to the high case–fatality rate of venous thrombosis [13,14].
In the RIETE registry of very old patients with venous thrombosis, cancer was present in 18% of patients aged over 80, which was not much different from its prevalence in younger patients , although higher than in the MEGA study, where 12% of patients (who were all aged under 70, mean age 50) had ever been diagnosed with cancer prior to the thrombotic event .
When we analyzed the participants in a sub-study of the MEGA study aged 70 years and older on whom we had information on cancer, 36 (12.5%) out of 287 cases and 11 (7.7%) out of 142 controls had been diagnosed with cancer in the 5 years preceding the index date (date of thrombosis diagnosis in cases, and a similar date in controls), yielding an odds ratio of 1.7 (CI95, 0.8–3.7). On a relative scale, this is substantially lower than the relative risk for patients under 70 in the same study (cancer diagnosed in 5 years prior to the index date: OR = 6.6). Even on an absolute scale, when we take into account that the baseline risk of venous thrombosis is around one per 1000 per year in those aged 50, and five per 1000 per year in those aged over 70 (Fig. 1), cancer appears less of a risk factor in old age than at younger ages. Although the prevalence of malignancies is high at advanced age, the odds ratio of 1.7 combined with the prevalence of 7.7% of a recent diagnosis with cancer predicts that only 5% of all venous thrombotic events in the elderly can be attributed to cancer (population attributable risk, PAR).
Hormonal replacement therapy
Estrogens relieve symptoms of menopause, and prolonged treatment reduces the progression of osteoporosis [99–101]. However, hot flushes also often disappear without active treatment , and it is unclear whether hormone therapy prevents fractures [103,104]. Hormone use has no beneficial effect on arterial disease and increases the risk of breast cancer [103,105]. Many studies have shown that hormone replacement therapy increases the risk of venous thrombosis 2- to 4-fold [21–24,106–110]. A recent study has suggested that estrogens administered by transdermal patches, that eliminate a first-pass effect through the liver, do not increase the risk of thrombosis , although previous reports did find an increased risk for patches [106,107].
Most studies on oral preparations concern equine conjugated estrogens. In a recent large case–control study of 586 postmenopausal women with venous thrombosis and 2268 controls, a 1.7-fold increased risk was observed for users of conjugated estrogens, while no excess risk was present for users of esterified estrogens . The risk of venous thrombosis in hormone users is particularly high in those who are obese and have APC-resistance because of FV Leiden [24,113–115]. The absence of an increased risk of users of esterified estrogens was mainly visible in the absence of this synergistic effect with prothrombotic mutations . This was in line with a subsequent study of the effect of various types of postmenopausal hormones on the coagulation system in healthy users, where an acquired APC-resistance was observed in users of conjugated estrogens, but not of esterified estrogens .
Venous thrombosis is a common disorder in the elderly with a substantial morbidity and mortality. Environmental risk factors, particularly immobilization, play an important role in the etiology of thrombosis in the elderly, and cause a substantial proportion of the total number of cases because risk factors associated with disease are more prevalent in old than in young individuals. Abnormalities in the clotting system, either genetic or acquired, appear to be equally and possibly more important in the elderly than in young and middle-aged individuals. Therefore, thrombosis in the elderly should be a focus of future studies.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interests.