Risk of Fatal Pulmonary Embolism in Patients With Treated Venous Thromboembolism FREE

VENOUS thromboembolism (VTE), which presents clinically as deep vein thrombosis (DVT), pulmonary embolism (PE), or both, is associated with 3 main sequelae: recurrent nonfatal VTE, the postthrombotic syndrome, and fatal PE. In patients with symptomatic DVT or PE, the risk of recurrent nonfatal VTE is estimated at 5% to 10% during the first year after diagnosis,¹ and about 2% to 3% per year thereafter.² The postthrombotic syndrome, which is associated with chronic venous insufficiency, leg swelling, and pain, is reported to occur in 20% to 30% of patients within 5 years after DVT.² Reliable estimates of the risk of fatal PE, which is the most feared complication of DVT and PE, are lacking. In retrospective studies, the risk of fatal PE is reported to be 5% in patients with DVT,³ and 8% to 23% in patients with PE.^3- 5 These estimates might be exaggerated because of overrepresentation of patients with severe disease who are likely to be at higher risk of fatal PE (patient selection bias), and overreporting of PE as the cause of death in patients with previous VTE (diagnostic suspicion bias). Prospective studies, which are less likely to have these methodologic limitations, have not been large enough to provide reliable estimates of the risk of fatal PE. Reliable information about the risk of fatal PE in patients with DVT or PE would be valuable when discussing the safety of discontinuing anticoagulant therapy, and when discussing disease prognosis. To provide reliable estimates of the risk of fatal PE, we combined the findings of methodologically rigorous prospective studies that reported fatal PE outcomes in patients with symptomatic DVT or PE who were treated with anticoagulants for 3 months. In addition, we determined how often recurrent VTE is fatal (case-fatality rate), and if the prognosis differed in patients presenting with DVT or PE.

The English-language MEDLINE database was searched for articles published between 1966 and September 1997 to identify all prospective studies of patients with DVT or PE. The key words used for the search were thrombophlebitis, diagnosis, treatment, and prognosis.Current Contents and bibliographies of relevant articles were scanned to identify additional articles. To limit the analysis to studies with the highest methodologic quality, study selection criteria were developed a priori. Inclusion criteria were as follows: (1) Patients with symptomatic DVT (proximal or distal) or PE who were treated for at least 5 days with heparin (intravenous or adjusted-dose subcutaneous unfractionated heparin [UH], to maintain the activated partial thromboplastin time between 1.5 and 2.0 times the baseline value, or a low-molecular-weight heparin [LMWH]) and 3 months of oral anticoagulants (with a targeted international normalized ratio of 2.0 to 4.5) or 3 months of LMWH or adjusted-dose subcutaneous UH. Patients receiving additional treatments (eg, thrombolysis, inferior vena cava filter) were not eligible for the analysis. (2) The qualifying event (DVT or PE) was confirmed with accurate diagnostic tests; the diagnosis of DVT required the presence of an intraluminal filling defect on venography⁶ or venous noncompressibility on Duplex ultrasound⁷; the diagnosis of PE required the presence of an intraluminal filling defect on pulmonary angiography or a high-probability ventilation-perfusion (V/Q) lung scan.⁸ (3) All patients were followed prospectively and outcome events (symptomatic recurrent nonfatal VTE and fatal PE) were documented. Exclusion criteria were as follows: (1) Consecutive patients were not recruited. (2) Among patients with recurrent nonfatal VTE or fatal PE, it could not be determined if the qualifying event was DVT or PE. (3) The timing of fatal PE (ie, whether the patient was taking or not taking anticoagulants) was not reported or could not be obtained through correspondence with the study authors. (4) Reliable methods were not used to diagnose symptomatic recurrent VTE and fatal PE: recurrent DVT had to be diagnosed by a new intraluminal filling defect on venography,⁹ a new noncompressible vein segment on Duplex ultrasound,¹⁰ or a newly abnormal impedance plethysmography test,¹¹ with or without an abnormal ¹²⁵ I-fibrinogen leg scan⁹; recurrent PE had to be diagnosed by a new intraluminal filling defect on pulmonary angiography or a new high-probability V/Q lung scan⁸; fatal PE that was confirmed or highly probable was diagnosed by autopsy, a high-probability V/Q scan prior to death, or a high clinical suspicion of fatal PE. All eligible studies were reviewed independently by 2 authors (J.D. and S.B.) to assess suitability for inclusion, and all included studies were reviewed to assess agreement in outcome event reporting. Agreement was evaluated using the κ statistic,¹² and disagreements were resolved by consensus.

The results of individual studies were combined to determine 3 outcome event rates: (1) fatal PE, (2) recurrent symptomatic nonfatal VTE (including rates of recurrent DVT and PE), and (3) case-fatality rate of recurrent VTE, defined as the proportion of all recurrent VTE events (fatal and nonfatal) resulting in death due to PE. Patients presenting with both DVT and PE were classified as having PE as the primary manifestation of recurrent nonfatal VTE. Patients who developed sudden death of undetermined cause, which might have been due to fatal PE, were classified as having possible fatal PE. The analysis was performed separately in (1) patients presenting with DVT, (2) patients presenting with PE, and (3) patient subgroups with DVT or PE followed for 3 or 12 months. Whenever possible, event rates were determined separately for the periods during anticoagulant therapy (initial 3 months after diagnosis) and following anticoagulant therapy. To determine the rates of fatal PE and recurrent nonfatal VTE during anticoagulant therapy, we summed the outcome events occurring in the DVT and PE patient populations during the initial 3-month follow-up period. The rates of fatal PE and recurrent nonfatal VTE following anticoagulant therapy were expressed as events per 100 patient-years to standardize for different follow-up durations across studies. The case-fatality rate of recurrent VTE was expressed as the proportion of all recurrent VTE events (nonfatal and fatal) that were fatal. Rates of fatal PE and case fatality were also determined after combining possible fatal PE events with confirmed or highly probable fatal PE events. In addition, we determined the classification of recurrent nonfatal events (DVT or PE) in patients who presented initially with DVT or PE.

The Pearson χ² test was used to compare outcome event rates in patients with DVT and PE. Two-sided P values and exact 95% confidence intervals were reported for all proportions. All reported rates were based on events that occurred in patients who had completed a study; patients who were lost to follow-up were excluded from the analysis.

Of 137 retrieved studies, 52 satisfied the inclusion criteria, of which 27 were not eligible because of 1 or more exclusion criteria, leaving 25 studies for the analysis.^13- 37 There were 21 studies of 4221 patients presenting with DVT (19 randomized controlled trials,^13- 30,33 2 prospective cohort studies^31- 32), and 5 studies of 1302 patients presenting with PE (4 randomized controlled trials,^33- 36 1 prospective cohort study³⁷). One study³³ included patients presenting with either DVT or PE (Table 1). Follow-up was completed in 99.3% of patients with DVT (4190/4221), and 100% of patients with PE. There was high inter-rater agreement with documentation of the number of outcome events per study (κ=0.70). A list of excluded studies, and the reason(s) for their exclusion, is available on request.

Baseline patient characteristics that could influence the risk of recurrent VTE are outlined in Table 1. In patients with DVT, 16% had a history of previous VTE, 15% had cancer at the time of diagnosis, and 33% had VTE in association with a transient VTE risk factor (ie, surgery, trauma, immobilization). Information on baseline patient characteristics was not available for 53% of patients with PE,^33,35,37 thereby limiting the analysis of baseline characteristics in patients with PE. In both groups of patients presenting with DVT or PE, 34% received LMWH and 66% received UH.

During anticoagulant therapy, the rate of fatal PE was 0.4% (17/4221 patients); if possible fatal PE events were included, the rate of fatal PE was 0.5% (23/4221 patients). Following anticoagulant therapy, there were 4 PE deaths during 1320 patient-years of follow-up (0.3/100 patient-years); there were no possible fatal PE events following anticoagulant therapy. During anticoagulant therapy, the rate of recurrent nonfatal VTE was 3.8% (156/4104 patients); following anticoagulant therapy, there were 75 recurrent nonfatal VTE events during 1190 patient-years of follow-up (6.3/100 patient-years). The case-fatality rate of recurrent VTE was 8.8% (15/171) during anticoagulant therapy; if possible fatal PE events were included, it was 11.9% (21/177). The case-fatality rate was 5.1% (4/79) following anticoagulant therapy (Table 2).

During anticoagulant therapy, the rate of fatal PE was 1.5% (19/1302 patients); if possible fatal PE deaths were included, the rate of fatal PE was 2.3% (30/1302 patients). Following anticoagulant therapy, there were no confirmed or possible fatal PE events during 256 patient-years of follow-up. All PE deaths occurred during the first 2 weeks of anticoagulant therapy. We were unable to determine the rate of recurrent nonfatal VTE (and therefore the case-fatality rate) separately for the periods during and following anticoagulant therapy because the timing of recurrent nonfatal VTE events (whether the patient was taking or not taking anticoagulants) was not reported in 1 large PE study.³⁷ For the combined follow-up period, during and following anticoagulant therapy, there were 53 recurrent nonfatal VTE events during 590 patient-years of follow-up (9.0/100 patient-years). The case-fatality rate of recurrent VTE for the combined follow-up period was 26.4% (19/72); if possible PE deaths were included, it was 36.1% (30/83) (Table 2).

Patients presenting with DVT were more likely to have DVT (78.6%) than PE (21.4%) as a manifestation of recurrent VTE, whereas patients presenting with PE were more likely to have PE (81.1%) than DVT (18.9%) as a manifestation of recurrent VTE. In a patient subgroup with DVT or PE who were followed for 3 months, the rate of fatal PE was higher in patients with PE than those with DVT (1.4% and 0.3%, respectively, P<.001); the case-fatality rate of recurrent VTE was higher in patients with PE than those with DVT (29.5% and 5.6%, respectively, P<.001). In a patient subgroup that was followed for 12 months, the rate of fatal PE was higher in patients presenting with PE than those presenting with DVT (1.7% and 0.4%, respectively, P=.04); the case-fatality rate of recurrent VTE was higher in patients with PE than those with DVT (18.7% and 6.7%), but this difference did not attain statistical significance (P=.1) (Table 3).

We investigated the risk of fatal PE in patients with symptomatic DVT or PE who were treated with 3 months of anticoagulant therapy by pooling findings from methodologically rigorous randomized controlled trials and prospective cohort studies. There are 2 conclusions that follow from our analysis of this patient population: (1) Fatal PE is a rare event in patients with treated DVT or PE both during and following anticoagulant therapy, and (2) patients who present with PE are more likely to die of recurrent VTE than patients who present with DVT.

Our first conclusion is based on a follow-up of 4221 patients with DVT and 1302 patients with PE. We estimate that 1 in 250 patients presenting with DVT will develop fatal PE during a 3-month course of anticoagulant therapy; after anticoagulant therapy is discontinued, the risk of fatal PE is estimated at 0.3 per 100 patient-years of follow-up (0.3% per year). We estimate that 1 in 68 patients presenting with PE will develop fatal PE during a 3-month course of anticoagulant therapy; there were no fatal PE events following 3 months of anticoagulant therapy. These estimates are lower than those in retrospective studies, in which the risk of fatal PE in patients with DVT or PE were reported to be 5% to 23%.^3- 5 Our estimates are likely to be valid because we included only prospective studies of patients with confirmed DVT or PE who underwent careful clinical follow-up. Further, recurrent nonfatal VTE and fatal PE were primary outcome events in these studies, and therefore, it is likely that these events were reliably documented. As a test of the robustness of our estimates, when possible fatal PE events, reported as sudden death of undetermined cause, were combined with confirmed or highly probable fatal PE events, the adjusted rates of fatal PE remained low.

In patients presenting with DVT or PE, the case-fatality rate of recurrent VTE was higher during anticoagulant therapy than after anticoagulant therapy was stopped. One possible explanation for this finding is that patients with a first episode of DVT or PE have more extensive thrombosis, and therefore are at higher risk of having larger emboli, compared with patients with recurrent DVT or PE, who might have less extensive thrombosis because of earlier recognition and treatment. Another possible explanation is that patients who survive the initial 3-month period after diagnosis represent a healthier patient subgroup who would be more likely to survive recurrent thrombotic events that occur after anticoagulant therapy is stopped, resulting in a lower rate of fatal PE during this period.

Our second conclusion is based on findings in 2 patient subgroups presenting with DVT or PE who were followed for 3 or 12 months. Patients presenting with PE had 3 to 5 times higher case fatality with recurrent VTE than patients presenting with DVT. This may be explained by our finding that patients with PE were 4 times more likely to develop PE than DVT as a manifestation of recurrent VTE, and therefore, would be at higher risk for fatal PE. It is possible that patients with PE have a more aggressive form of VTE that is more prone to embolic events. Another possibility is that compared with patients presenting with DVT alone, patients presenting with PE might be a population with greater comorbidity (eg, limited cardiorespiratory reserve) who might be more likely to develop initial symptoms from small emboli, and who might be at higher risk of death from recurrent embolic events. In one study included in this analysis that reported on comorbidity in patients with fatal PE, 7 of 7 patients who died of PE had major comorbid conditions.³⁷ In a retrospective study of 140 patients with PE, 8 of 8 PE deaths occurred in patients with cardiorespiratory disease, advanced cancer, or life-threatening sepsis.⁵ However, because information was not available to compare comorbidity in patients with DVT or PE, further study is required to investigate factors that determine prognosis in patients with DVT or PE.

There are 2 limitations of our analysis that should be addressed. First, we acknowledge that because our estimates are derived mainly from randomized controlled trials, which tend to exclude certain patients, our findings are not generalizable to all patients with DVT or PE. In the studies included in our analysis, several patient groups were excluded: patients with massive VTE, who were eligible to receive thrombolytic therapy; patients in whom anticoagulant therapy was contraindicated, some of whom received an inferior vena cava filter; patients with hypercoagulable diseases who required long-term anticoagulant therapy; patients with a limited life expectancy (ie, <3 months); and patients who were pregnant. Consequently, our findings would not apply to these patients, some of whom are likely to be at higher risk for recurrent nonfatal VTE and fatal PE. However, these patients make up a minority of all patients who present with DVT or PE. In those studies that reported the number of patients excluded and the reason(s) for exclusion,^{23,25- 27,29,33- 34} 3% of patients were excluded because of massive VTE, 5% were excluded because of a contraindication to anticoagulants, 7% were excluded because of hypercoagulable diseases or a need for long-term anticoagulant therapy, and 7% were excluded because of a limited life expectancy. It is probable that a similar proportion of patients were excluded from the other studies in this analysis because the patient inclusion criteria were similar across studies. Thus, our findings are likely to be generalizable to most patients with symptomatic DVT or PE who are treated with 3 months of anticoagulants. Although we combined patients in whom DVT or PE was treated with different heparin regimens (ie, UH, LMWH), it is unlikely that this would have resulted in a differential treatment effect, since these heparin regimens have been found to have similar efficacy in preventing recurrent VTE.^38- 39 Second, the outcome event rates that we calculated are based on an analysis of a large, heterogeneous population, but there was inadequate information to determine event rates in patient subgroups with idiopathic VTE, secondary VTE, previous VTE, and malignant disease. There is evidence that many factors, including the extent of lower limb DVT (proximal or distal),¹ a history of previous VTE,⁴⁰ the etiology of VTE,¹ and the presence of a hypercoagulable disease,^1,41 influence the risk of recurrent VTE, and therefore, the risk of fatal PE. These factors should be considered when assessing the risk of recurrent VTE, and therefore, fatal PE in individual patients.

The clinical implications of our analysis are 2-fold, and can be applied to most patients with symptomatic DVT or PE. First, our finding that fatal PE is a rare event in patients who have completed 3 months of anticoagulant therapy suggests that extending the duration of treatment is unlikely to prevent many deaths due to PE. However, in patients with idiopathic DVT or PE that occurs in the absence of any predisposing VTE risk factors, extending the duration of anticoagulant therapy beyond 3 months is likely to reduce the risk of recurrent nonfatal VTE.^42- 43 Second, although VTE is often considered a life-threatening disease, our findings suggest that, once it is diagnosed and treated, the risk of fatal PE is low, particularly in patients who have completed 3 months of anticoagulant therapy. Our findings provide a basis to counsel patients with DVT or PE about their prognosis. This analysis suggests that in patients with DVT or PE who receive conventional anticoagulant therapy (ie, heparin and warfarin), and who do not have comorbid diseases associated with a limited life expectancy, fatal PE is rare. In summary, less than 1% of patients with DVT and less than 2% of patients with PE will die of fatal PE in the year after diagnosis.