SEARCH

SEARCH BY CITATION

Keywords:

  • epidemiology;
  • risk factor;
  • venous thrombosis

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Summary.  Cost-effective strategies for the identification of subjects at risk of venous thromboembolism (VTE) in the active population are still lacking. Our objectives were to identify risk factors for venous thromboembolism in active subjects. We analyzed data from a population-based sample of 15 055 Caucasians aged 18–65 years randomly selected from the census list of the township of Vicenza, Italy. A validated methodology was used to retrospectively identify subjects with previous VTE. Body mass index (BMI), smoking, oral contraceptive use, previous superficial vein thrombophlebitis (SVT) and familial history of VTE, all at the age of first thrombosis, were ascertained by direct interview and by review of available medical records. Ninety-two deep vein thromboses [prevalence: 61.1/10 000, 95% confidence interval (CI) 49.2–74.9], three upper deep vein thrombosis (prevalence: 1.9/10 000, 95% CI 0.4–5.8) and 21 pulmonary embolism (prevalence: 13.9/10 000, 95% CI 8.6–21.3) were identified. After age and sex adjustment, clinically identifiable risk factors were: history of SVT [odds ratio (OR) = 6.8], oral contraceptive use (OR = 4.7), family history of VTE (OR = 4.5), smoking (OR = 1.7) and BMI above the third tertile (OR vs. mid-tertile 2.9). While previous SVT and BMI were associated with VTE in all circumstantial situations (surgery/trauma, pregnancy or idiopathic VTE), for oral contraceptive use, positive family history and smoking the degree of association varied significantly depending on the situation. Non-fatal VTE affects 0.7% of the subjects belonging to an active population, 56% of cases being potentially preventable. In 30% of VTE cases, at least two easily recognizable risk factors are present. Clinical assessment of risk factors remains the mainstay of VTE prevention.

About 30% of cases of venous thromboembolism (VTE) are diagnosed in the active people aged 60 years or less [1,2]. Although VTE in non-hospitalized patients has a low mortality rate [3], it is frequently complicated by long-term sequelae, with an estimated added cost of US$6000 for each case [3–6]. Long-term morbidity after VTE is even more relevant in young patients, and therefore the identification of subjects among active people who are at higher risk of VTE is of primary importance. Unfortunately, the importance of clinical determinants on VTE in the active population is largely unknown, since most available studies are based either on hospital discharge diagnoses or on cohorts of patients referred to specialized clinics [1,7,8]. Screening of asymptomatic subjects for genetic or acquired abnormalities which may increase the risk of VTE [9,10], such as factor (F)V Leiden, prothrombin G20210A or hyperhomocysteinemia, is not recommended on a cost-benefit basis [11]. In this study, we analyzed a large cohort of active subjects aged 18–65 of the Vicenza Thrombophilia and Atherosclerosis (VITA) Project to determine the prevalence of first non-fatal VTE and to identify the clinical determinants of VTE in the active population.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Study cohort

The VITA Project is a cross-sectional epidemiological investigation that collected clinical data and a blood sample from 15 055 white subjects aged 18–65 years and born on an even date, from May 1993 to June 1997 [12]. All participants were randomly selected from the census list of the district of Vicenza, Italy; subjects with severe physical or mental disease or with a history of active cancer in the last year were excluded. All subjects underwent a clinical examination before blood sampling [12]. A trained nurse measured weight, height, waist and hip circumferences on lightly dressed subjects. Body mass index (BMI) was computed as the ratio between weight and the square of height and expressed as kg m−2. A validated questionnaire investigating personal and familial history of VTE, history of use of heparin or warfarin for therapy of VTE, presence of circumstantial situations (surgery or trauma requiring at least 5 days of immobilization, pregnancy or puerperium) was administered to all subjects [13]. Previous VTE was diagnosed in subjects referring VTE symptoms with subsequent documented use of anticoagulant therapy (heparin or warfarin) or, in lack of the latter, venous reflux at Doppler ultrasound. This approach had a sensitivity of 71.3% and a specificity of 98.9% [13]. In subjects diagnosed as having suffered previous VTE, according to the above mentioned criteria, medical records regarding the episode were specifically sought. Subjects diagnosed as previous VTE were asked if their weight at time of VTE was greater or less than the present weight by 5 kg classes (e.g. did you weigh 5, 10, 15 or 20 kg more than now?) and the midpoint of the class of body change was added or subtracted to the actual weight to compute BMI at time of thrombosis. We considered as idiopathic VTE those cases occurring without exposure to the previously mentioned circumstantial situations. History of oral contraceptive intake during lifetime and at time of VTE, smoking habits, use of anticoagulant drugs and reproductive history (number of pregnancies, abortions and menopausal status) was collected by a direct interview. History of superficial vein thrombophlebitis (SVT) was collected with a validated questionnaire [13], followed by personal interview by a trained physician. Family history of VTE was considered positive when at least an additional first-degree relative had documented VTE [14]. The incidence of VTE for each episode of trauma or surgery was computed by dividing the number of VTE episodes occurred in our cohort after trauma or surgery by the expected number of hospitalization for trauma or surgery in the population from which the cohort was derived. This value was computed from the annual rate of discharges from our hospital, which provides over 95% of all primary care admissions in the area, for trauma (ICD codes: 805–828; 860–869; 890–897; 928) or surgery, stratified by age and sex. This indirect measurement was preferred to the direct estimation of the number of surgeries or traumas in each single subject to avoid a differential recall in those with adverse events after trauma or surgery. Only major surgical procedures (thoracic, abdominal, gynecologic, orthopedic and neurological surgery) were considered; surgeries related to cancer were excluded. The study protocol was approved by the local ethic committee, and all the participants gave their informed consent.

Statistical methods

Only the first episodes of VTE were considered. By using logistic regression analysis [15], we contrasted previous history of familial thromboembolism and occurrence of SVT, smoking, use of oral contraceptives and BMI at time of first VTE in thrombotic subjects vs. those values collected at time of enrollment into the study in asymptomatic subjects. As VTE cases were ascertained using a methodology with sensitivity and specificity below 100%, subjects incorrectly classified (either false positives or false negatives) tend to reduce the estimates of risk, an effect known as classification bias [16]. Since the classification bias may be considered to be non-differential for the considered variables (e.g. it is unlikely that the sensitivity and specificity of our method vary depending on smoking status), corrected odds ratio (OR) may be obtained by incorporating the sensitivity and specificity into the logistic regression [17,18]. Using the logistic regression model, we subsequently evaluated the thrombotic risk according to the number of risk factors present in each subject.

We subsequently tested the hypothesis that different VTE subtypes may have a different degree of association with recorded clinical characteristics. For this analysis, subjects were classified within four outcome categories: asymptomatic subjects, VTE in pregnancy or puerperium, VTE after surgery and/or trauma and idiopathic VTE. These latter were entered as three separate outcomes of the same dependent variables in a polytomous logistic regression model [19]. The attributable fraction for each variable was estimated from the prevalence of that exposure variable in the population and its odds ratio, as computed by the multivariate logistic regression coefficients to adjust for the effect of the other risk factors [20]. All computations were carried out using the Stata software package [21].

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Prevalence and type of non-fatal VTE in the active population

We identified 116 subjects with at least one episode of non-fatal VTE among 15 055 subjects for whom a detailed clinical history and a blood sample were available. Ninety-two cases were lower deep vein thrombosis (DVT) [prevalence: 61.1/10 000, 95% confidence interval (CI) 49.2–74.9], three upper DVT (prevalence: 1.9/10 000, 95% CI 0.4–5.8) and 21 pulmonary embolism (PE) (prevalence: 13.9/10 000, 95% CI 8.6–21.3) (Table 1). Recurrent VTE was present in three subjects, each one having suffered from two episodes; in these subjects, we retained for analysis only the first episode. Six subjects with previous VTE were still taking oral anticoagulants at time of study. Medical records were available for 78 subjects (67.2% of all subjects). Sixty-four percent of VTE cases occurred in females (P = 0.02, Fisher's exact test); after exclusion of cases occurred during pregnancy or puerperium and during oral contraceptives intake there was no difference in sex distribution (P = 0.31; Fig. 1). Prevalence of VTE related to trauma, surgery or idiopathic VTE was similar in males and females (57.1/10 000, 95% CI 45.6–70.5). PE was slightly more frequent in males (57.1%), while DVT was more frequent in females (69.5%, overall Fisher's exact test P= 0.005). Gender was the only factor associated with a different distribution of DVT vs. PE; all other characteristics shown in Table 1 were not differently distributed in subjects with DVT or PE (P > 0.1 for all variables, Fisher's exact test). Based on the number of observed person-years, the expected annual incidence of non-fatal VTE could be roughly estimated as 0.25 per thousand from age 18–39 years and 0.46 per thousand from age 40–65 years.

Table 1.  Characteristics of subjects evaluated by the VITA Project
 Thromboembolism (n = 116)Asymptomatic (n = 14 939)
Upper DVT (n = 3)Lower DVT (n = 92)Pulmonary embolism (n = 21)All VTE
  1. SVT, superficial vein thrombophlebitis; VTE, venous thromboembolism. *Percentage of female subjects.

Age, years (median)5054515343
Female/male1/264/289/1274/427932/7007
Use of oral contraceptives (%)*0 (–)10 (16)1 (11)11 (14)993 (13)
Smoking (%)1 (33)29 (31)8 (40)38 (33)3954 (26)
Family history of VTE (%)0 (–)15 (16)1 (5)16 (13)553 (4)
Previous SVT (%)0 (–)20 (22)4 (20)24 (20)546 (3)
O blood group (%)0 (–)34 (37)9 (45)41 (35)6536 (44)
Body mass index
 Lower-tertile2 (67)18 (20)4 (19)24 (21)4997 (33)
 Mid-tertile0 (0)29 (31)5 (24)34 (29)5002 (34)
 Upper-tertile1 (33)45 (49)12 (57)58 (50)4940 (33)
image

Figure 1. Risk of non-fatal venous thromboembolism in the VITA cohort.

Download figure to PowerPoint

Risk factors for non-fatal VTE in the active population

After age and sex adjustment, previous episodes of SVT, family history of VTE, use of oral contraceptives, smoking and BMI above the third tertile (corresponding to a BMI greater than 25.1 kg m−2, a cut-off corresponding to the World Health Organization criteria for being overweight [22]) at time of thrombosis were all associated with an increased risk of VTE (Table 2). Logistic regression coefficients did not differ if analysis was restricted to the 78 subjects for whom medical records were available (data not shown). Subjects were subsequently classified into six categories of risk, according to the number of the above-mentioned risk factors present. Taking subjects with none of these risk factors as the baseline, a steady increase of the risk of VTE was apparent among strata (Table 3). At least two risk factors for VTE were found in about 12% of the general population, with a risk of VTE in these subjects 5-fold higher than that found in those without clinically detectable risk factors.

Table 2.  Relative risk and population attributable fraction of non-fatal venous thromboembolism in subjects of the VITA Project, as estimated from a multivariate model
 Prevalence (%)Odds ratio (95% CI)Corrected odds ratio (95% CI)Population attributable fraction (%)
  1. SVT, superficial vein thrombophlebitis; VTE, venous thromboembolism. *Use of oral contraceptives at time of VTE or at time of investigation; percentages are referred to the female population. †The mid-tertile was taken as the baseline for risk estimation.

Previous SVT570 (3.8)4.9 (3.0–7.8)6.8 (3.9–12.0)12.9
Oral contraceptives use*1005 (6.6)3.9 (1.9–8.0)4.7 (2.0–10.8)16.0
Positive family history569 (3.8)3.5 (2.0–6.1)4.5 (2.4–8.5)0.7
Smoking3992 (26.5)1.6 (1.1–2.3)1.7 (1.0–2.7)13.7
Body mass index
 Lower-tertile5021 (33.3)0.7 (0.4–1.2)0.5 (0.3–0.9)
 Upper-tertile4998 (33.2)1.7 (1.1–2.6)2.9 (1.4–6.2)16.6
Table 3.  Relative risk of non-fatal venous thromboembolism in subjects of the VITA Project, by number of risk factors present
Number of risk factors at time of VTE*n (% of the cohort)AsymptomaticVTE (% of thrombotic subjects)Relative risk (95% CI)
  1. SVT, superficial vein thrombophlebitis; VTE, venous thromboembolism. *Use of oral contraceptives, smoking, body mass index above the third tertile, previous history of SVT, positive family history of VTE.

None6720 (44.6)669921 (18.1)
One6376 (42.3)632551 (43.9)2.5 (1.5–4.2)
Two1818 (12.1)178335 (30.2)6.2 (3.6–10.7)
Three135 (0.9)1278 (6.9)20.1 (8.7–46.2)
Four6 (0.04)51 (0.9)63.8 (7.1–569.6)

Risk factors for non-fatal VTE in circumstantial situations

Surgery or trauma and pregnancy were circumstantial situations present in 65 cases of VTE (56% of all VTE cases). In women, 22 cases of VTE during pregnancy were recorded among 12 212 pregnancies, corresponding to one VTE for every 555 pregnancies; the 22 cases occurred in a total of 5270 women with at least one pregnancy, corresponding to a prevalence of one VTE during pregnancy in every 239 women with at least one pregnancy.

Forty-three subjects suffered from VTE after surgery or trauma. Based on the rate of hospital discharges for trauma or surgery unrelated to neoplasia (263/10 000 person-years), we estimated that the age- and sex-adjusted rate of VTE after trauma or surgery was 25.3 cases per 10 000 episodes of trauma or surgery not related to neoplasia (0.2%, or one case for every 385).

Using polytomous logistic regression, we were able to test the hypothesis that different risk factors show a different degree of association with VTE depending on three circumstantial situations: idiopathic VTE, surgery or trauma and pregnancy (Table 4). Among all considered risk factors, only previous SVT was associated with the development of VTE in circumstantial situations. In contrast, use of oral contraceptives was associated with idiopathic VTE but not with VTE occurring after surgery or trauma (OR = 8.7 vs. 1.08, respectively). Similarly, a family history of VTE was associated with VTE in pregnancy and idiopathic VTE (OR = 8.5 and 4.6, respectively) but not with VTE after surgery or trauma. A similar heterogeneity of the magnitude of risk was also observed for smoking and a BMI above the upper tertile.

Table 4.  Relative risk of non-fatal venous thromboembolism in subjects of the VITA Project, by type of inciting factor. Odds ratio refer to relative risks within each inciting factor
 Odds ratio and 95% confidence intervals
Pregnancy (n = 22)Surgery/trauma (n = 42)Idiopathic (n = 52)
  1. SVT, superficial vein thrombophlebitis; VTE, venous thromboembolism. *The lowest tertile was taken as baseline.

Previous SVT3.7 (1.2–11.1)4.6 (2.1–10.1)5.8 (3.0–11.4)
Use of oral contraceptives at time of VTE1.08 (0.1–9.5)8.7 (3.3–23.0)
Positive family history8.5 (3.2–21.9)0.5 (0.07–4.0)4.6 (2.2–9.6)
Smoking0.9 (0.3–2.6)1.9 (1.0–3.6)1.5 (0.9–2.8)
Body mass index*
 Mid-tertile0.8 (0.2–3.1)0.9 (0.3–2.4)1.9 (0.9–4.2)
 Upper-tertile2.3 (0.8–6.9)2.2 (1.0–4.9)2.2 (1.0–4.8)

As only previous SVT and positive family history were associated with VTE in pregnancy, we recomputed the risk of VTE during pregnancy for each woman associated with these two risk factors. The risk of VTE in pregnancy was one case in every 332 women with no risk factors, one in every 97 women with one risk factor and one in every 12 women with two risk factors. We similarly recomputed the risk of VTE after surgery or trauma for each subject according to presence of previous SVT, smoking or BMI above the third tertile. The risk of VTE after surgery was one case in every 769 subjects with no risk factors, one in every 333 subjects with one risk factor, one in every 104 subjects with two risk factors and one in every 45 subjects with three risk factors.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In this study, we performed a cross-sectional evaluation of VTE in 15 055 subjects to investigate the prevalence, distribution and risk factors of non-fatal VTE in the active population aged 18–65 years.

Given the cross-sectional design of this study and the consequent retrospective evaluation of thrombotic episodes and risk factor exposure, a preliminary discussion of the potential limitations of our findings seems appropriate. First, since all cases of VTE associated with a subsequently fatal illness (e.g. cancer) were lost, our analyses are valid only for cases of non-fatal VTE. Furthermore, other important risk factors, such as paresis, hospital confinement, active cancer or use of central venous catheter [23] could not be assessed. Second, since the causes of VTE might change with age, our findings apply only to subjects below 65 years [24]. Third, a potential bias may arise from the retrospective assessment of cases and exposures to risk factors, an effect known, respectively, as disease or exposure misclassification bias. Regarding disease misclassification, our methodology allowed ascertainment of VTE cases with an incomplete sensitivity (78.3%). Thus, since about 20% of cases of non-fatal VTE may have been missed, the true prevalence of previous non-fatal VTE may be as high as one case in every 100 subjects in the active population. Nevertheless, a recent prospective study, in which the cumulative prevalence of VTE at age 50 was 0.5 case in every 100 subjects [24] and the agreement of our estimated annual incidence of VTE with that reported by series from hospital records [7,25,26], suggest that our prevalence estimates are affected by a negligible error. Moreover, all logistic regression estimates were corrected for disease misclassification [17]. Regarding exposure misclassification, it should be remembered that a spurious association could arise only for a differential misclassification, particularly if referral of risk factors is increased in cases, whereas otherwise a retrospective evaluation tend to diminish the degree of the association [18]. We were able to validate diagnosis and exposure to circumstantial situations and risk factors in 66% of subjects with VTE and found no substantial change of risk estimates by limiting the analyses to these subjects. This essentially rules out the possibility of a differential exposure misclassification bias in our study. Moreover, recall of estrogen use and body weight has been demonstrated to be sufficiently accurate for case control studies [27–30]. Nonetheless, we cannot exclude that our estimates could be biased toward more conservative estimates.

A first finding of the study is that we observed that non-fatal VTE affect about one subject in every 130. Most importantly, our study shows that about half (56%) of the first episodes of VTE recorded were related to circumstantial situations (pregnancy, surgery or trauma) and were thus potentially preventable. The observed incidence of pregnancy-related VTE in our cohort (one case in every 555 pregnancies, 0.18%) is similar to those previously reported, ranging from 0.02 to 0.2% [31–35]. In contrast, we observed a slightly lower incidence of VTE due to trauma or surgery, with an estimated incidence of one case in every 385 exposures to trauma or surgery. For comparison, the reported incidence of symptomatic VTE ranges from one case in every 47 to one in every 167 depending on the type of surgery [36,37]. Possible reasons for this discrepancy are directly related to the study methodology. First, the young age of investigated subjects and the exclusion of subjects with known cancer resulted in a low number of high-risk procedures such as hip, knee and cancer surgeries. Second, we could not detect those asymptomatic VTE cases that are diagnosed in clinical trials by phlebography. An alternative explanation for the observed lower number of VTE cases after surgery or trauma is that a more widespread use of heparin prophylaxis may have actually reduced the incidence of VTE in this setting.

As a large proportion of cases of VTE occurring in the active population may be preventable, the identification of risk factors is particularly important. Recently, much emphasis has been given to inherited or acquired abnormalities predisposing to VTE, such as FV Leiden, prothrombin G20210A or hyperhomocysteinemia [9]. These abnormalities are common in Caucasians and may interact with circumstantial risk factors, but laboratory screening of asymptomatic subjects is not recommended on a cost-benefit basis [11,38]. In contrast, some factors that may be ascertained with a simple clinical examination (SVT, history of VTE in a first-degree relative, smoking and a BMI above the second tertile) emerged as important determinants of VTE in the active population. An increased risk of VTE in patients with history of SVT has been also recently reported [23], while increased BMI is associated also with recurrence of VTE [39,40]. Although the effect of smoking and obesity is low in terms of relative risk, it should be noted that they exert a significant impact at a population level because of their prevalence, as shown by their population attributable fraction. As reported in Table 2, while a BMI above the second tertile is associated with only a 2.9-fold increased risk of VTE, about 16% of cases of VTE in the population may be attributed to a BMI above 25 kg m−2. For comparison, only 7% of VTE cases in our cohort could be attributed to FV Leiden [11]. In addition, since risk factors interact with a multiplicative effect, association of smoking with use of oral contraceptives (a behavior found in about 25% of the women aged below 45 years in our cohort) may for instance result in an 8-fold risk of VTE. As shown in Table 3, about 12% of subjects in the general population had at least two risk factors, with a relative risk of VTE in these subjects fivefold higher than that found in subjects with no risk factors. Therefore, clinically identifiable risk factors are not only more prevalent but may also result in a risk higher than that of inherited abnormalities of clotting factors.

A further finding of the study is that risk factors for VTE (such as use of oral contraceptives or family history of thrombosis) show a different association with VTE depending on the particular circumstantial situation. As shown in Table 4, family history of VTE was apparently not associated with VTE related to surgery or trauma, whereas it was strongly associated with idiopathic or pregnancy-related VTE. It is therefore likely that surgery or trauma may represent a sufficient cause for venous thromboembolism, whereas inherited risk factors may be more common in idiopathic or pregnancy-related VTE. The relevance of these findings is, however, limited by the small number of events recorded in this study and the consequent overlap of the risk estimates in different inciting circumstances.

In conclusion, about one subject in every 130 in the active population aged 18–65 years may be conservatively estimated to have suffered VTE. More than half of the cases of VTE in this age range may be considered to be preventable, and in more than one-third of the VTE cases, at least two easily recognizable risk factors may be present. Although these findings need to be confirmed by prospective studies, this study suggests that clinical assessment of risk factors should remain the mainstay for the identification of the high-risk patient and for the primary prophylaxis of VTE.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank Drs E. Boscaro, R. Trombini, D. Zenere for clinical data collection; all the secretaries, nurses and laboratory technicians. A special thank to all the family doctors of the town of Vicenza for their enthusiastic collaboration. Mario Cleves of Stata Corp. is especially acknowledged for his invaluable help in writing the routine for logistic regression adjusted for misclassification bias.This study was supported by grants from Fondazione Cassa di Risparmio di Verona, Vicenza, Belluno ed Ancona, Italy; ALT, Associazione per la Lotta alla Trombosi, Milan, Italy and AVEC, Associazione Veneta per l'Emofilia e le Coagulopatie.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Nordstrom M, Lindblad B, Bergqvist D, Kjellstrom T. A prospective study of the incidence of deep-vein thrombosis within a defined urban population. J Intern Med 1992; 232: 15560.
  • 2
    Coon WW. Epidemiology of venous thromboembolism. Ann Surg 1977; 186: 14964.
  • 3
    Prandoni P, Lensing AW, Cogo A, Cuppini S, Villalta S, Carta M, Cattelan AM, Polistena P, Bernardi E, Prins MH. The long-term clinical course of acute deep venous thrombosis. Ann Int Med 1996; 125: 17.
  • 4
    Bergqvist D, Jendteg S, Johansen L, Persson U, Odegaard K. Cost of long-term complications of deep venous thrombosis of the lower extremities: an analysis of a defined patient population in Sweden. Ann Int Med 1997; 126: 4547.
  • 5
    McColl MD, Ellison J, Greer IA, Tait RC, Walker ID. Prevalence of the post–thrombotic syndrome in young women with previous venous thromboembolism. Br J Haematol 2000; 108: 2724.
  • 6
    Hirsh J. Epidemiology and pathogenesis of venous thrombosis. J Am Coll Cardiol 1986; 8: 104B113B.
  • 7
    Anderson FAJ, Wheeler HB, Goldberg RJ, Hosmer DW, Patwardhan NA, Jovanovic B, Forcier A, Dalen JE. A population-based perspective of the hospital incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch Intern Med 1991; 151: 9338.
  • 8
    Heijboer H, Brandjes DP, Buller HR, Sturk A, Ten Cate JW. Deficiencies of coagulation-inhibiting and fibrinolytic proteins in outpatients with deep-vein thrombosis. N Engl J Med 1990; 323: 15126.
  • 9
    Rosendaal FR. Venous thrombosis: a multicausal disease. Lancet 1999; 353: 116773.
  • 10
    Miletich JP. Thrombophilia as a multigenic disorder. Semin Thromb Hemost 1998; 24 (Suppl. 1): 1320.
  • 11
    Rodeghiero F, Tosetto A. Activated protein C resistance and factor V gene mutation are independent risk factors for venous thromboembolism. Ann Int Med 1999; 130: 64350.
  • 12
    Rodeghiero F, Tosetto A. The epidemiology of inherited thrombophilia: the VITA Project. Thromb Haemost 1997; 78: 63640.
  • 13
    Frezzato M, Tosetto A, Rodeghiero F. Validated questionnaire for the identification of previous personal or familial venous thromboembolism. Am J Epidemiol 1996; 143: 125765.
  • 14
    Tosetto A, Frezzato M, Rodeghiero F. Family history and inherited thrombophilia. Br J Haematol 1995; 89: 2278.
  • 15
    Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research. Belmont, CA: Lifetime Learning Publishers, 1982.
  • 16
    Hsieh CC. The effect of non-differential outcome misclassification on estimates of the attributable and prevented fraction. Stat Med 1991; 10: 36173.
  • 17
    Magder LS, Hughes JP. Logistic regression when the outcome is measured with uncertainty. Am J Epidemiol 1997; 146: 195203.
  • 18
    Schlesselman JJ. Case-Control Studies. New York: Oxford University Press, 1982.
  • 19
    Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: John Wiley & Sons, 1989.
  • 20
    Lachin JM. Biostatistical Methods. New York: John Wiley & Sons, 2000.
  • 21
    StataCorp. Stata Statistical Software: release 6.0. College Station Stata Corporation, 1999.
  • 22
    Anonymous. Physical status. the use and interpretation of anthropometry: report of a WHO expert committee. WHO Techn Report Series 1995; 854: 1452.
  • 23
    Heit JA, Silverstein MD, Mohr DN, Petterson TM, O'Fallon WM, Melton LJ. Risk factors for deep vein thrombosis and pulmonary embolism: a population-based case-control study. Arch Intern Med 2000; 160: 80915.
  • 24
    Hansson PO, Welin L, Tibblin G, Eriksson H. Deep vein thrombosis and pulmonary embolism in the general population: ‘the study of men born in 1913’. Arch Intern Med 1997; 157: 166570.
  • 25
    Oger E. Incidence of venous thromboembolism: a community-based study in Western France. EPI-GETBP Study Group. Group d'Etude de la Thrombose de Bretagne Occidentale. Thromb Haemost 2000; 83: 65760.
  • 26
    Silverstein MD, Heit JA, Mohr DN, Petterson TM, O'Fallon WM, Melton LJ. Trends in the incidence of deep vein thrombosis and pulmonary embolism: a 25-year population-based study. Arch Intern Med 2000; 158: 58593.
  • 27
    Greendale GA, James MK, Espeland MA, Barrett-Connor E. Can we measure prior postmenopausal estrogen/progestin use? The Postmenopausal Estrogen/Progestin Interventions Trial. The PEPI Investigators. Am J Epidemiol 1997; 146: 76370.
  • 28
    Goodman MT, Nomura AM, Wilkens LR, Kolonel LN. Agreement between interview information and physician records on history of menopausal estrogen use. Am J Epidemiol 1990; 131: 81525.
  • 29
    Norgan NG, Cameron N. The accuracy of body weight and height recall in middle-aged men. Int J Obes Relat Metab Disord 2000; 24: 16958.
  • 30
    Must A, Willett WC, Dietz WH. Remote recall of childhood height, weight, and body build by elderly subjects. Am J Epidemiol 1993; 138: 5664.
  • 31
    Bergqvist A, Bergqvist D, Hallböök T. Deep vein thrombosis during pregnancy. Acta Obstet Gynecol Scand 1983; 62: 448.
  • 32
    McColl MD, Ramsay JE, Tait RC, Walker ID, McCall F, Conkie JA, Carty MJ, Greer IA. Risk factors for pregnancy associated venous thromboembolism. Thromb Haemost 1997; 78: 11838.
  • 33
    Lindqvist P, Dahlback B, Marsal K. Thrombotic risk during pregnancy: a population study. Obstet Gynecol 1999; 94: 5959.
  • 34
    Andersen BS, Steffensen FH, Sorensen HT, Nielsen GL, Olsen J. The cumulative incidence of venous thromboembolism during pregnancy and puerperium – an 11 year Danish population-based study of 63,300 pregnancies. Acta Obstet Gynecol Scand 1998; 77: 1703.
  • 35
    Gherman RB, Goodwin TM, Leung B, Byrne JD, Hethumumi R, Montoro M. Incidence, clinical characteristics, and timing of objectively diagnosed venous thromboembolism during pregnancy. Obstet Gynecol 1999; 94: 7304.
  • 36
    Dahl OE, Gudmundsen TE, Haukeland L. Late occurring clinical deep vein thrombosis in joint-operated patients. Acta Orthop Scand 2000; 71: 4750.
  • 37
    Sigel B, Ipsen J, Felix R. The epidemiology of lower extremity deep venous thrombosis in surgical patients. Ann Surg 1974; 179: 27890.
  • 38
    Helmerhorst FM, Bloemenkamp KW, Rosendaal FR, Vandenbroucke JP. Oral contraceptives and thrombotic disease: risk of venous thromboembolism. Thromb Haemost 1997; 78: 32733.
  • 39
    Heit JA, Mohr DN, Silverstein MD, Petterson TM, O'Fallon WM, Melton LJ. Predictors of recurrence after deep vein thrombosis and pulmonary embolism: a population-based cohort study. Arch Intern Med 2000; 160: 7618.
  • 40
    National Task Force on the Prevention and Treatment of Obesity. Overweight, obesity, and health risk. Arch Intern Med 2000; 160: 898904.