|
|
| |||||||||||||||||||||||||||||||||||||||
Background Although deep-vein thrombosis and pulmonary embolism are considered common complications after major trauma, their frequency and the associated risk factors have not been carefully quantified.
Methods We performed serial impedance plethysmography and lower-extremity contrast venography to detect deep-vein thrombosis in a cohort of 716 patients admitted to a regional trauma unit. Prophylaxis against thromboembolism was not used.
Results Deep-vein thrombosis in the lower extremities was found in 201 of the 349 patients (58 percent) with adequate venographic studies, and proximal-vein thrombosis was found in 63 (18 percent). Three patients died of massive pulmonary embolism before venography could be performed. Before venography, only three of the patients with deep-vein thrombosis had clinical features suggestive of the condition. Deep-vein thrombosis was found in 65 of the 129 patients with major injuries involving the face, chest, or abdomen (50 percent); in 49 of the 91 patients with major head injuries (54 percent); in 41 of the 66 with spinal injuries (62 percent); and in 126 of the 182 with lower-extremity orthopedic injuries (69 percent). Thrombi were detected in 61 of the 100 patients with pelvic fractures (61 percent), in 59 of the 74 with femoral fractures (80 percent), and in 66 of the 86 with tibial fractures (77 percent). A multivariate analysis identified five independent risk factors for deep-vein thrombosis: older age (odds ratio, 1.05 per year of age; 95 percent confidence interval, 1.03 to 1.06), blood transfusion (odds ratio, 1.74; 95 percent confidence interval, 1.03 to 2.93), surgery (odds ratio, 2.30; 95 percent confidence interval, 1.08 to 4.89), fracture of the femur or tibia (odds ratio, 4.82; 95 percent confidence interval, 2.79 to 8.33), and spinal cord injury (odds ratio, 8.59; 95 percent confidence interval, 2.92 to 25.28).
Conclusions Venous thromboembolism is a common complication in patients with major trauma, and effective, safe prophylactic regimens are needed.
It is generally believed that deep-vein thrombosis and pulmonary embolism are common complications of major trauma5,6. However, there is a paucity of information on the epidemiology of venous thromboembolism in patients with trauma5,6. An autopsy study found deep-vein thrombosis in 65 percent of fatally injured patients, and pulmonary embolism was the cause of death in 20 percent7,8. Several prospective studies have examined the risk of deep-vein thrombosis after trauma,9,10,11,12,13 although these studies have had limitations, including the assessment of highly selected groups of patients with trauma,9,10,11 failure to use a gold-standard diagnostic test for thrombosis,11,12,13 small samples,10,13 and lack of detailed descriptions of patients9,11,13.
In view of the limited information available, our study had three objectives. The first was to determine prospectively with contrast venography the frequency of deep-vein thrombosis in a broad spectrum of patients with trauma. The second objective was to determine the incidence of thrombosis in subgroups of patients according to the site of major injury or the presence of a specific injury. The third objective was to identify the characteristics of patients with trauma that may be associated with a high risk of thromboembolic complications.
Methods
The study was conducted in the Regional Trauma Unit of the Sunnybrook Health Science Centre, the largest level 1 trauma facility for adults in Canada2,14. From January 1989 through April 1991, a cohort of patients with major trauma was prospectively evaluated with objective diagnostic testing for venous thromboembolism. The protocol was approved by the Research Ethics Board of the hospital, and patients or their surrogates were asked to provide consent for participation in the venographic component of the study.
Study Population
Consecutive patients admitted to the trauma unit with an Injury Severity Score15 of at least 9 were assessed for eligibility. A score of 9 was selected as the cutoff point to exclude patients who had only minor injuries or a moderately severe injury of a single system. Other criteria for exclusion were death or discharge within five days after the injury, a history of allergy to radiographic contrast agents, renal failure (a serum creatinine concentration higher than 3.4 mg per deciliter [300 µmol per liter]), foot injuries that would have precluded successful venography, and the need for therapeutic anticoagulation. The patients did not receive mechanical or pharmacologic antithrombotic prophylaxis during the study.
Surveillance for Thromboembolism
The principal procedure used to determine the presence of thrombosis was bilateral contrast venography. Impedance plethysmography was performed every other day in the patients who were able to undergo the procedure16,17. These patients underwent venography 14 to 21 days after admission, or earlier if the hospital stay was shorter than 14 days. For reasons of safety, patients not able to undergo serial plethysmography (because of lower-extremity injuries) underwent venography on day 7 to 10. Hemodynamically stable patients in the intensive care unit were transported to the radiology department for venography. If a patient had a plaster leg cast, it was temporarily removed or a "window" was created in it to allow access to a foot vein.
Venography was performed according to the method of Rabinov and Paulin,18 with the use of iohexol, a non-ionic contrast agent. Since a large proportion of patients with trauma have lower-extremity injuries that can affect the interpretation of venograms, we developed conservative criteria for coding the venograms before the study began. The venogram of each leg was considered adequate if there was complete visualization of at least the proximal 75 percent of both posterior tibial and peroneal veins of the calf, as well as the popliteal, superficial femoral, common femoral, and external iliac veins. Deep-vein thrombosis was defined as a constant intraluminal filling defect in a deep leg vein seen on two or more views. Proximal deep-vein thrombosis was defined as thrombosis involving the popliteal or more proximal veins with or without involvement of the calf veins. Venograms indicating persistent nonfilling of a deep venous segment were considered to be nondiagnostic unless deep-vein thrombosis was identified elsewhere.
Impedance plethysmography was performed in patients who had symptoms or signs suggestive of deep-vein thrombosis16,17. All patients (symptomatic or asymptomatic) with abnormal plethysmograms underwent venography on the same day. Patients with clinical features suggestive of pulmonary embolism underwent ventilation-perfusion lung scanning19,20. A normal perfusion scan was considered to rule out pulmonary embolism, whereas a scan indicating a high probability of embolism, defined as a scan revealing one or more perfusion defects, segmental or larger, with a ventilation mismatch, was considered to confirm the diagnosis. Patients with nondiagnostic lung scans underwent pulmonary angiography, contrast venography, or both within 24 hours after scanning.
Outcome Measures
The results of venography, lung scanning, and pulmonary angiography were evaluated by a panel of experts who were unaware of the clinical details and the original interpretation of the test.
The possible risk factors for venous thromboembolism that we assessed were age, sex, cause of the injury, Injury Severity Score,15 score on the Abbreviated Injury Scale for each of the six body regions,21 major sites of injury, specific injuries (pelvic, femoral, or tibial fractures and spinal cord injuries), blood group, transfusion requirements, need for surgical procedures, cumulative time in the operating room, mobility, and length of hospital stay.
Statistical Analysis
The primary analysis was a determination of the frequency of deep-vein thrombosis in the overall group of patients with trauma and in four clinical subgroups established on the basis of the body regions with predefined major injuries: the head; face, chest, or abdomen; spine; and lower limbs. Rates of thromboembolism were also assessed for several specific injuries, including spinal cord injuries and fractures of the pelvis, femur, tibia, and ankle. Associations between potential risk factors and thrombosis were determined with chi-square statistics for categorical variables and t-tests for continuous variables. All reported P values are two-tailed. The clinically relevant factors that were statistically associated with deep-vein thrombosis (P<0.05) were then assessed by multiple logistic regression, and a prognostic model was developed22,23.
Results
Study Population
During the 28-month study period, there were 1008 admissions to the Sunnybrook Regional Trauma Unit. A total of 292 patients were ineligible for the following reasons: an Injury Severity Score of less than 9 (116 patients); early discharge (89); early death (59); inability to undergo venography because of foot injuries (15), renal failure (4), or allergy to radiographic contrast agents (2); prolonged critical illness precluding transport to the venography suite (6); and the need for therapeutic anticoagulation (1). The remaining 716 patients underwent daily clinical surveillance and, if technically possible, serial impedance plethysmography.
Venography was not attempted in 85 patients because informed consent was not provided (54 patients), death occurred after day 5 but before venography could be performed (29), or symptomatic pulmonary embolism developed before venography could be performed (2). Autopsies performed in 18 of the 29 patients who died after day 5 confirmed the cause of death as head injury in 8, sepsis with multisystem organ failure in 5, massive pulmonary embolism in 3, myocardial infarction in 1, and carcinomatosis in 1. Autopsies were not performed in the other 11 patients, but the clinical cause of death was head injury in 7 and multisystem organ failure associated with sepsis in 4. We were unable to arrange venography for an additional 188 patients before their discharge home or transfer to another facility.
Venography was attempted in the remaining 443 patients. In 56 patients (12.6 percent), venous access was unsuccessful in one or both legs, and in 38 (8.6 percent), venography was judged to be nondiagnostic by the expert panel. The 349 patients in whom contrast venography was adequate constituted the study cohort. The clinical characteristics of the initial cohort of patients who underwent surveillance for deep-vein thrombosis and those in whom venography was adequate are summarized in Table 1.
|
Of the 349 patients with adequate venographic studies, 201 (57.6 percent; 95 percent confidence interval, 52.2 to 62.8) had deep-vein thrombosis. Proximal deep-vein thrombosis was diagnosed in 63 patients (18.1 percent; 95 percent confidence interval, 14.2 to 22.5). Among the 285 patients with adequate venographic studies in both legs, 18 percent had bilateral deep-vein thrombosis, and 3 percent had bilateral proximal deep-vein thrombosis. Fourteen patients had isolated proximal deep-vein thrombosis (calf veins not involved), representing 22 percent of those with proximal deep-vein thrombosis and 4 percent of the overall cohort. Rates of proximal and distal deep-vein thrombosis were virtually identical for the left and right legs. The results in the 36 patients who underwent early venography (because they could not be screened with serial plethysmography) did not differ from the results in the 313 patients who underwent delayed venography. Deep-vein thrombosis was diagnosed in 58.3 percent of those with early studies and in 57.5 percent of those with delayed studies (P = 0.92), and proximal deep-vein thrombosis was diagnosed in 16.7 and 18.2 percent, respectively (P = 0.82).
Only 3 of the 201 patients with deep-vein thrombosis (1.5 percent) had clinical characteristics suggestive of thrombosis before the diagnosis was made by venography. None of the three patients with fatal pulmonary embolism had any clinical features suggestive of venous thromboembolism before their sudden deaths on days 15, 16, and 18 after injury.
The frequencies of deep-vein thrombosis and proximal deep-vein thrombosis according to the site (or sites) of major injury are shown in Table 2. Many of the patients had injuries involving more than one site. The relation between the occurrence of deep-vein thrombosis and each possible pair of injury sites is shown in Figure 1. These conditional probabilities suggest the strong influence of lower-extremity orthopedic injuries on the risk of thrombosis.
|
|
Pulmonary Embolism
During the surveillance period, 39 patients had clinically suspected pulmonary embolism before venography was performed. Pulmonary embolism was confirmed in seven (in three at autopsy, in three by lung scanning, and in one by pulmonary angiography). The diagnosis of pulmonary embolism was ruled out in 22 patients (in 17 by lung scanning, in 4 by angiography, and in 1 at autopsy). The diagnosis of embolism was neither confirmed nor ruled out in the remaining 10 patients with nondiagnostic lung scans, although in 8 of the 10, deep-vein thrombosis was confirmed by venography within 24 hours after scanning, and angiography was therefore not performed. The other two patients refused to undergo pulmonary angiography. Thus, the diagnosis of pulmonary embolism (or deep-vein thrombosis) was confirmed in fewer than half the patients with clinical features suggestive of pulmonary embolism.
Analysis of Risk Factors
The univariate analysis of potential factors for venous thrombosis identified significant associations between thrombosis and the following factors: age, the score on the Abbreviated Injury Scale for the extremities, lower-limb orthopedic injury, spinal cord injury, fracture of the femur or tibia, blood transfusion within the first 24 hours after admission, and surgery (P<0.05 for all these factors) (Table 3). Patients with deep-vein thrombosis were also significantly less mobile and had a longer mean hospital stay than those without thrombosis (36.7 days vs. 28.1 days).
|
Multiple logistic regression identified five statistically significant factors that were independent predictors of deep-vein thrombosis: age, blood transfusion, surgery, fracture of the femur or tibia, and spinal cord injury (Table 4). The goodness-of-fit test of Hosmer and Lemeshow confirmed that the model fit the data set with P = 0.6322.
|
We have confirmed that deep-vein thrombosis is a common complication after major trauma, with an overall incidence of 58 percent in our cohort. Even more important, proximal deep-vein thrombosis was diagnosed in 18 percent of the patients, and three patients died of massive pulmonary embolism despite clinical and plethysmographic surveillance. Given the high rate of proximal deep-vein thrombosis, we expect that the rate of fatal pulmonary embolism (0.4 percent) would have been higher had we not screened the patients with plethysmography and venography and then treated those who had positive tests.
Two previous studies that used routine venography in patients with trauma also reported a high frequency of venous thrombosis9,10. Freeark et al. found deep-vein thrombosis in 35 percent of a highly selected group of 124 patients, more than half of whom had hip fractures alone and all of whom were immobilized for at least three weeks9. A study by Kudsk et al. found deep-vein thrombosis in 63 percent of patients with multiple injuries who were confined to bed for at least 10 days, although only 38 patients were assessed10. Our data are also consistent with the high frequency of thrombosis reported in other, somewhat analogous groups of patients, including those undergoing general surgical procedures24,25 and those with spinal cord injuries26 or fractures of the hip or tibia27,28,29,30.
Fewer than 2 percent of our patients with deep-vein thrombosis had clinical features suggestive of thrombosis before venography was performed, even though many of the asymptomatic patients with deep-vein thrombosis had extensive thrombi. Fatal pulmonary embolism in patients with trauma may occur without warning, in the absence of clinical evidence of deep-vein thrombosis and despite noninvasive venous screening. The diagnosis of pulmonary embolism (or deep-vein thrombosis) was confirmed in only 15 of the 39 patients (38 percent) thought to have pulmonary embolism, although asymptomatic patients were not screened for pulmonary embolism. The poor reliability of clinical screening for thrombosis has also been demonstrated in other studies of patients with trauma7,9,10,12 or other conditions. Among patients with trauma, this problem is compounded by the high frequency of leg pain and swelling due to leg injuries or chest pain and shortness of breath due to chest injuries. We thought it was essential to use venography in this study, because noninvasive tests such as impedance plethysmography and venous ultrasonography have not been adequately validated in studies of patients with trauma and there is considerable evidence demonstrating that both procedures have reduced sensitivity when used to screen patients for deep-vein thrombosis, even proximal deep-vein thrombosis31,32,33,34,35,36,37,38.
We found an unexpectedly high frequency of deep-vein thrombosis among patients whose only major injuries involved the face, chest, or abdomen and among patients with major head injuries. However, the presence of lower-extremity fractures had a strong influence on the risk of deep-vein thrombosis in patients with major injuries in more than one system.
The predictive factors for venous thrombosis in our study were age, the need for surgery or blood transfusion, and the presence of lower-extremity fractures or spinal cord injuries. Although it is well known that elderly patients have an increased risk of thrombosis,39,40,41 age was also a risk factor in our patients, whose mean age was only 39 years. Young patients with trauma, however, remain at high risk. In our study, the incidence of deep-vein thrombosis was 46 percent in those under the age of 30 years.
Patients with fractures of the tibia or femur were almost five times as likely to have deep-vein thrombosis as patients without fractures at these sites. Although this association has been reported for more than 65 years,7,8,9,10,11,12,30,42,43,44 only three other studies have used venography to assess the risk of deep-vein thrombosis in patients with lower-extremity fractures9,10,30. More than 80 percent of our patients with spinal cord trauma had deep-vein thrombosis, and two of the fatal pulmonary emboli were in patients with spinal cord injuries. Previous investigators have also found a strong association between spinal cord injury and the risk of thrombosis26,45,46. Reduced mobility and a longer hospital stay were also associated with an increased risk of thrombosis, whereas the Injury Severity Score was not associated with an increased risk10,11,12. The latter finding suggests that the specific type of injury plays a greater part in the genesis of venous thrombosis than does the severity of the trauma.
Although some groups of patients were found to have a particularly high frequency of thrombosis, only a small proportion of the overall group was at low risk. For example, the risk of deep-vein thrombosis was less than 20 percent in only 8 percent of our patients, and the risk was less than 30 percent in only 20 percent of the patients. Even some of the fully mobile patients with injuries at a single site had extensive proximal deep-vein thrombosis. We therefore consider all patients with major trauma to be at high risk for thromboembolic complications.
The frequencies of deep-vein thrombosis in our study cohort may well be somewhat inflated by the loss of eligible patients who did not undergo venography. Table 1 shows that, as compared with the patients in the initial cohort, those in the final cohort were slightly older, had slightly more severe injuries, and were more likely to have injuries that were subsequently found to be predictive of an increased risk of thrombosis. However, the magnitude of each of these differences is small, and our findings are consistent with those of previous studies of patients with trauma and other groups of patients at high risk for thrombosis.
In summary, our findings demonstrate that patients with major trauma have a very high risk of venous thromboembolism. The consequences of trauma-related venous thrombosis are morbidity due to symptomatic deep-vein thrombosis and pulmonary embolism, mortality due to pulmonary embolism, eventual chronic venous insufficiency, and a substantial expenditure of resources related to the diagnosis and treatment of thromboembolism, a prolonged hospital stay, and delayed rehabilitation. Future studies need to identify safe and effective prophylactic regimens against thrombosis that can be used alone or in combination with reliable screening strategies in patients with trauma.
Supported by grants from the Ontario Ministry of Health, the Rick Hansen Man in Motion Legacy Fund/Canadian Paraplegic Association, and the Sunnybrook Trust for Research.
We are indebted to Dr. G. Pagliarello and the other surgeons at the Sunnybrook Regional Trauma Unit for their encouragement and support; to the trauma nursing staff for their enthusiastic cooperation; to the Radiology Department staff for their assistance in obtaining the venograms; to Sanofi Winthrop (Markham, Ont.) for providing the venographic contrast mediums; to Ms. J. Carter, S. Dier, K. Gammon, C. Rogers, and S. Totton for their technical assistance; and to Drs. J. Hirsh, J. Leclerc, B. McLellan, and D. Naylor for their valuable suggestions.
Source Information
From the Departments of Medicine (W.H.G., K.I.C., R.M.J.) and Research Design and Biostatistics (J.P.S.) and the Clinical Epidemiology Unit (W.H.G., E.C., J.P.S.), Sunnybrook Health Science Centre, University of Toronto, Toronto.
Address reprint requests to Dr. Geerts at Sunnybrook Health Science Centre, Rm. D674, 2075 Bayview Ave., Toronto, ON M4N 3M5, Canada.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Thromboembolism after Major Trauma
Blebea J., Strothman G., Wells P., Anderson D. R., Geerts W. H., Jay R. M.
Extract |
Full Text
N Engl J Med 1995;
332:1448-1450, May 25, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
