Clinical Risk Factors and Timing of Recurrent Venous Thromboembolism During the Initial 3 Months of Anticoagulant Therapy

Background  In patients with venous thromboembolism (VTE), identifying clinical risk factors for recurrence during the initial 3 months of anticoagulant therapy and knowledge of the time course of recurrence may help clinicians decide about the frequency of clinical surveillance and the appropriateness of outpatient treatment.

Methods  Analysis of a randomized controlled trial database involving 1021 patients with VTE (750 with deep vein thrombosis [DVT] and 271 with pulmonary embolism [PE]) who were followed up for 3 months after the start of anticoagulant therapy. All patients received initial treatment with unfractionated heparin or a low-molecular-weight heparin (reviparin) and a coumarin derivative starting the first or second day of treatment, with a target international normalized ratio of 2.0 to 3.0.

Results  Four independent clinical risk factors for recurrent VTE were identified: (1) cancer (odds ratio [OR], 2.72; 95% confidence interval [CI], 1.39-5.32), (2) chronic cardiovascular disease (OR, 2.27; 95% CI, 1.08-4.97), (3) chronic respiratory disease (OR, 1.91; 95% CI, 0.85-4.26), and (4) other clinically significant medical disease (OR, 1.79; 95% CI, 1.00-3.21). Older age was associated with a decreased risk for recurrent VTE (OR, 0.76; 95% CI, 0.64-0.92). Previous VTE, sex, and idiopathic VTE were not risk factors for recurrence. In patients with DVT or PE, there was no significant difference in the rates of recurrent nonfatal VTE (4.8% vs 4.1%; P = .62), major bleeding (2.9% vs 2.2%; P = .53), and non-VTE death (6.4% vs 7.8%; P = .45), but recurrent fatal PE was more frequent in patients with PE than DVT (2.2% vs 0%; P<.01). There was a clustering of recurrent VTE episodes during the initial 2 to 3 weeks after the start of treatment.

Conclusions  During the initial 3 months of anticoagulant therapy, recurrent VTE is more likely to occur in patients with cancer, chronic cardiovascular disease, chronic respiratory disease, or other clinically significant medical disease. Patients with PE are as likely to develop recurrent VTE as those with DVT; however, recurrence is more likely to be fatal in patients who initially present with PE.

IN PATIENTS with venous thromboembolism (VTE), which includes deep vein thrombosis (DVT) and pulmonary embolism (PE), the 2 most common early sequelae are recurrent VTE and anticoagulant-related bleeding, which occur in about 6% and 3% of patients, respectively, within 3 months of the start of treatment.^1,2 Although risk factors for anticoagulant-related bleeding have been extensively investigated in patients with VTE,^3-8 there is little information about risk factors for recurrent VTE, which can result in significant morbidity and mortality. Recurrent ipsilateral DVT is associated with a 6-fold increased risk for the postthrombotic syndrome, which can result in chronic leg pain and swelling.⁹ In addition, recurrent PE that occurs within 3 months of the start of anticoagulant therapy is fatal in 9% to 26% of patients.¹⁰ Identifying patients at increased risk for recurrent VTE during anticoagulant therapy, and the time course of recurrence, is clinically relevant for 2 reasons: First, this information may help clinicians decide about the frequency of clinical surveillance and the appropriateness of outpatient treatment of VTE.^11-15 Second, early detection and treatment of recurrent DVT, when the size and occlusiveness of the thrombus are less, may result in improved thrombus regression and a decreased risk of the postthrombotic syndrome.¹⁶ Moreover, early detection and prompt treatment of recurrent PE with an alternative antithrombotic strategy (eg, thrombolysis or extended heparin therapy) and supportive measures (ie, oxygen and intravenous fluids) may reduce mortality.¹⁷

Most studies that have investigated risk factors for recurrent VTE, such as the extent of lower limb DVT,¹⁸ previous VTE,^19,20 and idiopathic VTE,^21-24 are based mainly on patients who had stopped taking anticoagulant drugs. It is possible that risk factors for recurrence differ in patients who are receiving anticoagulant therapy. In 2 cohort studies^25,26 that investigated risk factors for recurrent VTE during anticoagulant therapy, the results are limited because the duration of follow-up was only 8 days, clinical and radiologic outcomes were combined, and the main outcome was recurrent PE but not DVT. In another cohort study⁵ that investigated risk factors for recurrent thromboembolism in patients with VTE, stroke, or cardiovascular disease, the results were not reported separately for patients with VTE. Although patients with VTE who initially present with PE might be considered to be a high-risk group,^27,28 no study, to our knowledge, has compared the risk for sequelae in patients with PE or DVT.

To identify patients at increased risk for recurrent VTE during the initial 3 months of anticoagulant therapy, we analyzed a database of 1021 patients with VTE. The objectives of this study are (1) to identify clinical risk factors for recurrent VTE during anticoagulant therapy, (2) to determine whether the clinical presentation of VTE (ie, as DVT or PE) affects the risk for recurrence, and (3) to determine the time course of recurrence within 3 months of the start of treatment.

The study population consisted of 1021 consecutive patients with objectively confirmed symptomatic VTE who were participating in a clinical trial investigating a low-molecular-weight heparin, reviparin sodium (Clivarin; Knoll Pharmaceuticals, Lugwigshafen, Germany), for the initial treatment of VTE.²⁹ For this analysis, patients were classified as having DVT or PE based on their clinical presentation. Patients were classified as having DVT alone if they had symptomatic lower limb DVT without symptoms of PE, and DVT was confirmed by venous ultrasonography or venography. Patients were classified as having PE if they had symptomatic PE, with or without symptoms of DVT, and PE was confirmed by a high-probability ventilation-perfusion lung scan, pulmonary angiography, or, if the lung scan was nondiagnostic, demonstration of DVT by venous ultrasonography or venography.

Patients were randomly allocated to receive initial treatment with (1) intravenous unfractionated heparin sodium, with dose adjustments according to a nomogram,³⁰ to achieve a target activated partial thromboplastin time of 1.5 to 2.5 times the control value or (2) a weight-adjusted, twice-daily dose of subcutaneous reviparin sodium (ie, 3500 anti–factor Xa units for patients weighing 35-45 kg, 4200 units for patients weighing 46-60 kg, and 6300 units for patients weighing >60 kg). The mean + SD duration of treatment with unfractionated heparin and reviparin was 5.8 + 2.1 days and 6.3 + 2.4 days, respectively. All patients received a coumarin derivative, starting the first or second day of treatment, for at least 3 months, with a target international normalized ratio (INR) of 2.0 to 3.0. In terms of the adequacy of oral anticoagulant therapy for all patients during the 3-month follow-up, the INR was less than 2.0, 2.0 to 3.0, and greater than 3.0 in 20%, 57%, and 23% of measurements, respectively.

The following clinical characteristics were identified a priori determine whether they were independent risk factors for recurrent VTE: (1) age, (2) sex, (3) the etiology of VTE, (4) previous VTE, (5) clinical presentation (ie, DVT or PE), (6) cardiovascular disease, (7) respiratory disease, (8) the presence of 1 or more clinically significant medical diseases (ie, renal, hepatic, gastrointestinal, neurologic, hematologic, and multisystem), and (9) the presence of cancer. The etiology of VTE was classified according to whether VTE occurred after exposure to a transient risk factor (ie, recent [<4 weeks] surgery, trauma, or immobility for 3 consecutive days), whether VTE occurred in the presence of a permanent risk factor (ie, previous VTE or cancer), or whether VTE was idiopathic (occurring in the absence of a risk factor). The presence of cardiovascular or respiratory disease was considered separately from other diseases because we postulated that such patients would be more likely to develop clinically overt recurrent VTE than patients without cardiorespiratory disease. In addition, patients with cardiovascular or respiratory disease were classified as having acute or chronic disease. Definitions of concurrent medical diseases and cancer are available from the authors on request.

The clinical outcomes during the 3-month follow-up were symptomatic recurrent nonfatal VTE, recurrent fatal PE, major bleeding, and non-VTE death. All patients with suspected recurrent VTE underwent objective diagnostic tests. Recurrent DVT was diagnosed by the appearance of a new intraluminal filling defect or an extension of a previous filling defect on a venogram, a new noncompressible vein segment, or a 4-mm or more increase in the diameter of a thrombus on venous ultrasound.^31,32 Recurrent nonfatal PE was diagnosed by the appearance of a new intraluminal filling defect, an extension of a previous filling defect, or a sudden cutoff of vessels greater than 2.5 mm in diameter on pulmonary angiogram; a new perfusion scan defect involving 75% or more of a lung segment; or a nondiagnostic lung scan and confirmed recurrent DVT.³³ Recurrent PE that was fatal was confirmed by autopsy or clinical findings. Major bleeding was defined as bleeding that was clinically overt and was associated with a decrease in the hemoglobin level greater than 20 g/L within 24 hours, a need for blood transfusion, or permanent discontinuation of anticoagulant therapy.³⁴ All outcomes were evaluated by an independent adjudication committee that was unaware of patients' treatment allocation. No patient was lost to clinical follow-up. All patient information obtained from the study database was based on a dichotomous (yes/no) result because there were no baseline clinical characteristics or outcomes that were documented as unknown.

Univariate and multivariate stepwise forward logistic regression analysis was used to identify independent predictors of recurrent VTE. The odds ratio, and corresponding 95% confidence interval, was reported for each variable in the model. Variables were included in the final (multivariate) model based on their level of significance in the univariate model. For each variable that was considered but not included in the model, these statistics were reported as if that variable had been included in the model. Because there was no statistically significant difference in the rates of recurrent VTE in patients who received unfractionated heparin or reviparin (ie, 4.9% vs 5.3%; P = .77 by χ²₁), a treatment variable was not included in the model.

To determine whether there were any associations between baseline clinical characteristics and the clinical presentation of VTE, univariate analysis was used to compare the means or proportions of clinical factors between patients with DVT or PE. When means were compared, the significance level was based on the t statistic; when proportions were compared, the significance level was based on the χ² statistic. The Kaplan-Meier method of survival analysis was used to analyze the time course of recurrent VTE, which was compared, in a subjective manner, with the time course of major bleeding episodes and non-VTE deaths.

The clinical characteristics of the study population are summarized in Table 1. Of 1021 patients, 750 presented with lower limb DVT (648 proximal and 102 distal) and 271 presented with PE.

The effect of clinical characteristics on the risk for recurrent VTE using a univariate model is presented in Table 2. With the multivariate model (Table 3), 4 clinical characteristics were independent risk factors for recurrent VTE during anticoagulant therapy: (1) cancer, (2) chronic cardiovascular disease, (3) chronic respiratory disease, and (4) the presence of 1 or more other clinically significant medical diseases. Although chronic respiratory disease did not attain conventional statistical significance as a risk factor for recurrence (P = .11), it was considered clinically significant and was included in the multivariable model because of the magnitude of the odds ratio of 1.91 and because of a consistent association between other chronic diseases and recurrent VTE. Previous VTE, idiopathic VTE, sex, and the clinical presentation of VTE (ie, as DVT or PE) were not risk factors for recurrence. Older age, based on 10-year increments, was associated with decreased risk for recurrence.

The baseline clinical characteristics of patients with DVT alone were similar to those of patients with PE (with or without symptomatic DVT) according to age, sex, and the prevalence of most concurrent medical diseases (Table 4). However, patients with PE had a significantly higher prevalence of chronic cardiovascular disease and acute or chronic respiratory disease (P<.01). In terms of clinical outcomes (Table 5), there was no significant difference in the rates of recurrent nonfatal VTE, major bleeding, and non-VTE deaths in patients with DVT or PE. However, the rate of recurrent fatal PE was significantly higher in patients who initially presented with PE (P<.01).

During the 3-month follow-up, recurrent VTE occurred in 58 patients (6%); of these episodes, 52 were nonfatal and 6 were fatal (Figure 1). Major bleeding and non-VTE deaths occurred in 28 and 72 patients (3% and 7%), respectively. In terms of the time course of recurrent VTE, there was a clustering of episodes after the start of treatment: 26% (15/58) occurred within 7 days, 57% (33/58) occurred within 14 days, and 72% (42/58) occurred within 21 days. By comparison, 46% (13/28) of major bleeding episodes occurred within 7 days, 64% (18/28) occurred within 14 days, and 75% (21/28) occurred within 21 days of the start of treatment. Non-VTE deaths were uniformly distributed during 3-month follow-up.

In this analysis of a large, diverse patient population with symptomatic VTE, the presence of cancer, chronic cardiovascular disease, chronic respiratory disease, and other clinically significant medical disease were independent risk factors for recurrent VTE during the initial 3 months of anticoagulant therapy. Of these clinical characteristics, cancer was the strongest risk factor for recurrent VTE. This finding supports the results of 2 smaller cohort studies^9,35 in which patients with VTE and cancer had a 2- to 5-fold higher rate of recurrent VTE than patients without cancer. The mechanism of warfarin-resistant recurrent VTE observed in patients with VTE and cancer is not known but might be related to the generation of tumor-related procoagulant factors, such as tissue factor and cysteine protease factor X activators.^36,37 Our finding that chronic cardiovascular disease, chronic respiratory disease, and other clinically significant medical disease were independent risk factors for recurrent VTE might reflect decreased mobility in these patients, resulting in venous stasis and an increased risk for recurrent VTE. The explanation for our finding that older age is associated with decreased risk for recurrent VTE is not clear but is in agreement with a population-based cohort study²⁴ of patients with VTE that reported a decreased risk for recurrence with older age (odds ratio, 0.85; 95% confidence interval, 0.84-0.86).

In contrast to findings of other studies^19-23 in which patients with previous VTE or idiopathic VTE were at increased risk for recurrence, in the present study these clinical characteristics were not predictors of recurrent VTE. It is possible that a history of previous VTE and the occurrence of idiopathic VTE are not important predictors of disease recurrence while patients are receiving anticoagulant therapy. Consistent with this hypothesis, in 2 clinical trials^38,39 of anticoagulant therapy involving patients with VTE, a history of previous VTE and idiopathic VTE were risk factors for recurrence only after anticoagulant drug use was discontinued.

Compared with patients who initially presented with DVT alone, patients presenting with PE were not more likely to develop recurrent VTE but were more likely to die of recurrent PE and had a higher prevalence of cardiovascular or respiratory disease. Thus, patients with PE might not have a more aggressive form of VTE that is prone to recurrent emboli. Instead, the development of symptomatic PE and the increased risk of death from recurrent PE might reflect the presence of impaired cardiorespiratory reserve. Consistent with this hypothesis, in a clinical trial⁴⁰ involving patients with symptomatic lower limb DVT who underwent baseline ventilation-perfusion lung scan testing, the prevalence of cardiorespiratory disease was higher in patients with symptomatic PE than in those with asymptomatic PE (ie, 22% [32/145] vs 15% [8/52]; odds ratio, 1.57; 95% confidence interval, 0.64-4.25). Furthermore, in a cohort study⁴¹ of 140 patients with symptomatic PE, 10 of 12 deaths due to initial or recurrent PE occurred in patients with cardiac or respiratory insufficiency. In our patient population, 6 of 6 PE deaths occurred in patients with cardiovascular or respiratory disease.

There are potential limitations of this study that should be addressed. First, because the study population was derived from a clinical trial, several patient groups were excluded: patients with clinically massive VTE who received thrombolytic drug therapy, those with a life expectancy of less than 3 months, and pregnant women. Consequently, our findings may not apply to these patients. However, these patients composed only 16% (273/1724) of all patients screened for this study. Other patients were excluded for reasons that are unlikely to affect our analysis of risk factors for recurrent VTE (ie, patient received >48 hours of anticoagulation before screening, contraindication to anticoagulation, lack of informed consent, and geographic inaccessibility). Thus, our findings are likely to be generalizable to most patients with VTE. Second, we did not evaluate the effect of prothrombotic abnormalities (eg, factor V Leiden) on the risk for recurrent VTE because screening for such abnormalities was not routinely performed. To our knowledge, no study has investigated the effect of prothrombotic abnormalities on the risk for recurrent VTE during anticoagulant therapy. In patients who have completed a 3- or 6-month course of anticoagulant therapy, some abnormalities (ie, homozygous factor V Leiden, antiphospholipid antibodies, combined factor V Leiden and prothrombin gene mutations) but not others (ie, heterozygous factor V Leiden)) seem to be associated with an increased risk of recurrent VTE.^39,42,43 Third, we acknowledge that identifying risk factors for recurrent VTE was not an a priori objective of the clinical trial that formed the basis of our analysis. Thus, it is possible that the clinical characteristics used in the multivariate model (eg, concurrent medical diseases) were not accurately documented in the patient database. However, because this clinical trial was evaluating an investigational drug (reviparin) for the treatment of VTE, it is likely that the database was accurate. Furthermore, because there was no previous knowledge of this study's objectives, it is unlikely that there was bias in the data collection. Finally, we acknowledge that because the adequacy of anticoagulant therapy at the time of VTE recurrence was not part of the study database, it is possible that some recurrent VTE episodes were related to inadequate anticoagulation. However, this should not affect the study results relating to risk factors for recurrent VTE because the presence of such factors (eg, cancer and concurrent diseases) is unlikely to have been associated with less frequent INR or activated partial thromboplastin time testing and inadequate anticoagulation.

The clinical implications of this study are 2-fold and can be applied to most patients with symptomatic VTE who are starting anticoagulant therapy. First, in patients with VTE who are at increased risk for recurrence (ie, with cancer, chronic cardiovascular disease, chronic respiratory disease, or other comorbidity), increased clinical surveillance may be warranted during the initial 2 to 3 weeks after the start of treatment, when recurrent VTE is most likely to occur. In these patients, closer clinical surveillance would allow earlier recognition and expedited treatment of recurrent VTE. Second, because the risk for recurrent VTE is similar in patients who initially present with PE or DVT alone, outpatient treatment with low-molecular-weight heparin may be appropriate for selected patients with PE. However, before outpatient treatment for PE is widely implemented, additional studies are required to evaluate the safety of this approach compared with in-hospital treatment and to identify patient groups, such as those with cardiorespiratory disease, in whom in-hospital treatment may be preferable.

To summarize, in patients with symptomatic VTE who receiving anticoagulant drug therapy, recurrent VTE is more likely to occur in patients with cancer, chronic cardiovascular disease, chronic respiratory disease, or other clinically significant medical disease. Patients presenting with PE are as likely to develop recurrent VTE as patients presenting with DVT alone; however, recurrent VTE was more likely to be fatal in patients who initially present with PE.

Accepted for publication June 30, 2000.

This study was supported in part by a research scholarship from the Medical Research Council of Canada, Ottawa (Dr Crowther) and a Career Investigator Award from the Heart and Stroke Foundation of Ontario, Toronto (Dr Ginsberg).

We thank Clive Kearon, MD, PhD, Hamilton Civic Hospitals Research Centre, Hamilton, Ontario, for his helpful comments and Knoll Pharmaceuticals, Lugwigshafen, Germany, for allowing us access to the Columbus Study patient database.

Reprints: James D. Douketis, MD, FRCPC, St Joseph's Hospital, Room F-538, 50 Charlton Ave E, Hamilton, Ontario, Canada L8N 4A6 (e-mail: jdouket@fhs.mcmaster.ca).