- Top of page
- Patients and methods
We compared three rapid d-dimer methods for the diagnosis of venous thromboembolism. Patients presenting to four teaching hospitals with the possible diagnosis of deep vein thrombosis or pulmonary embolism were investigated with a combination of clinical likelihood, d-dimer (SimpliRED) and initial non-invasive testing. Patients were assigned as being positive or negative for deep vein thrombosis or pulmonary embolism based on their three-month outcome and initial test results. The three d-dimer methods compared were: (a) Accuclot d-dimer (b) IL-Test d-dimer (c) SimpliRED d-dimer. Of 993 patients, 141 had objectively confirmed deep vein thrombosis or pulmonary embolism. The sensitivity of SimpliRED, Accuclot and IL-Test were 79, 90 and 87% respectively. All three d-dimer tests gave similar negative predictive values. The SimpliRED d-dimer was found to be less sensitive than the Accuclot or IL-Test. When combined with pre-test probability all three methods are probably acceptable for use in the diagnosis of venous thromboembolism.
The diagnosis of deep vein thrombosis or pulmonary embolism (venous thromboembolism) is problematic because clinical diagnosis alone is unreliable and the gold standard tests (venography or pulmonary angiography) are invasive and expensive (Kearon et al, 1998a). Consequently, non-invasive imaging such as lung scanning or compression ultrasonography is usually performed first. However, a normal compression ultrasound of the proximal veins or a non-diagnostic ventilation perfusion (V/Q) scan does not rule out a diagnosis of clinically significant deep vein thrombosis or pulmonary embolism respectively (Birdwell et al, 1998; Kearon et al, 1998a,b; Wells et al, 1998a; Anderson & Wells, 1999; Perrier et al, 1999). Management studies have demonstrated that further testing is required, but this results in many normal tests, high expense, and inconvenience for the patient because of the low frequency of pulmonary embolism and deep vein thrombosis (Turkstra et al, 1997; Kearon et al, 1998b). A normal d-dimer test can help to exclude venous thromboembolism in many such patients (Wells et al, 1995; Freyburger et al, 1998; Ginsberg et al, 1998; Kahn, 1998; Wells et al, 1998a; Anderson et al, 1999; Perrier et al, 1999).
The d-dimer tests used in most diagnostic laboratories to screen for disseminated intravascular coagulation are latex agglutination assays which are not sufficiently sensitive to exclude venous thromboembolism (Carter et al, 1993). Initial studies using enzyme-linked immunoabsorption assay (ELISA) d-dimer testing showed sensitivities and negative predictive values approaching 100% (Bounameaux et al, 1994; vanBeek et al, 1996; Janssen et al, 1997). However, ELISA testing is not practical for use on a routine basis, particularly in an Emergency Room setting because batch testing is recommended and, consequently, results are not available in a timely fashion. Moreover, ELISA testing is expensive and has a low specificity.
Consequently, more sensitive and rapid latex agglutination tests and point of care tests for d-dimer have been evaluated (Dale et al, 1994; Ginsberg et al, 1995, 1998; Elias et al, 1996; Turkstra et al, 1996; Bernardi et al, 1998; Wells et al, 1998b). Depending on the population studied and the test used, the sensitivities vary from 80% to 100% with negative predictive values generally greater than 90% (Dale et al, 1994; Ginsberg et al, 1995, 1998; Elias et al, 1996; Turkstra et al, 1996; Bernardi et al, 1998; Wells et al, 1998b). Problems with several of these methods, however, include a greater expense compared with standard latex agglutination d-dimer tests, a need for dedicated instrumentation, a lack of suitability of the test for other d-dimer testing indications (e.g. as a screen for disseminated intravascular coagulation) and the lack of automation leading to subjective interpretation of test results and poorer quality control.
In this study, we compare the accuracy of three rapid d-dimer tests for the diagnosis of acute venous thromboembolism. One of these assays, SimpliRED, is a point of care manual method that has only been used for the diagnosis of venous thromboembolism. The other two, Accuclot and IL-Test d-Dimer, are both suitable to other indications for d-dimer testing and, until now, have not been well studied in deep vein thrombosis and pulmonary embolism. Accuclot is a semiquantitative latex-agglutination manual method, whereas the IL-Test d-dimer is a fully quantitative and automated method.
- Top of page
- Patients and methods
There were a total of 993 consecutive patients (468 for possible deep vein thrombosis and 525 for possible pulmonary embolism) included in the study and 141 (14%) had objectively documented deep vein thrombosis or pulmonary embolism.
Sensitivity for venous thromboembolism was statistically significantly lower for the SimpliRED d-dimer test 79% (95% C.I. 77–82) than for the Accuclot 90% (88–92, P = 0·0035) and the IL-Test 87% (84–89, P = 0·0339) assays (Tables I and II). However, the specificity of SimpliRED d-dimer test 76% (73–79), was higher than for the other two assays 72% (69–75, P = 0·0088) and 71% (68–73, P = 0·0003) (Tables I and II). Although there was a trend towards a higher negative predictive value with the Accuclot assay compared with the SimpliRED test [98% (97–99) versus 96% (94–97)], this difference was not statistically significant (Tables I and II). Of note, SimpliRED had 29 false negative results compared with 14 for Accuclot and 19 for the IL-Test. Only seven patients were diagnosed with venous thromboembolism in the 3 month follow-up period after the initial investigations were negative. Finally, Table III gives a subanalysis of deep vein thrombosis and pulmonary embolism for sensitivity, specificity and negative predictive value.
Table I. Comparison of D-dimer tests (n = 993).
|True positive (n)||112||127||122|
|False positive (n)||205||237||251|
|True negative (n)||647||615||601|
|False negative (n)||29||14||19|
|Sensitivity percentage (95% C.I.)||79 (77–82)||90 (88–92)||87 (84–89)|
|Specificity percentage (95% C.I.)||76 (73–79)||72 (69–75)||71 (68–73)|
|Negative predictive value percentage (95% C.I.)||96 (94–97)||98 (97–99)||97 (96–98)|
|Positive predictive value percentage (95% C.I.)||35 (32–38)||35 (32–38)||33 (30–36)|
Table II. Comparative accuracy of D-dimer tests.
|Method comparison||P-value for sensitivity||P-value for specificity ||P-value for negative predictive value|
|SimpliRED vs. Accuclot||0·0035||0·0088||0·0834|
|SimpliRED vs. IL||0·0339||0·0003||0·4269|
|Accuclot vs. IL||0·2278||0·0637||0·0614|
Table III. Subanalysis of D-dimer results for DVT (n = 468) and PE (n = 525).
|Sensitivity (%) for DVT||78||91||85|
|Sensitivity (%) for PE||80||91||91|
|Specificity (%) for DVT||67||63||61|
|Specificity (%) for PE||79||76||74|
|Negative predictive value (%) for DVT||91||96||93|
|Negative predictive value (%) for PE||97||99||99|
- Top of page
- Patients and methods
The diagnosis of deep vein thrombosis and pulmonary embolism remains problematic (Kearon et al, 1998a). Negative compression ultrasound testing and non-diagnostic V/Q scans do not necessarily rule out clinically significant deep vein thrombosis or pulmonary embolism (Kearon et al, 1998a). Recently the use of non-invasive testing, d-dimer testing and clinical pre-test probability has simplified the diagnosis of venous thromboembolism and dramatically reduced the need for more invasive testing.
This study, which is the largest study of its kind to date, shows that both the Accuclot and the IL-Test d-dimer methods have a significantly better sensitivity than SimpliRED for venous thromboembolism with sensitivities of 90%, 87% and 79% respectively. However, consistent with other results is a trade-off with lower specificity.
Several previous studies have illustrated the usefulness of sensitive D-dimer testing with the most widely studied of these methods being SimpliRED (Ginsberg et al, 1995; Turkstra et al, 1996; Kearon et al, 1998b; Wells et al, 1998b). The main advantages of SimpliRED are that to date it is one of the most widely studied d-dimer methods and that it can be used as a point-of-care test. When used as a point-of-care test the main potential disadvantage is that it is operator-dependent, which could decrease the accuracy of the test if performed by untrained personnel. This test is also non-quantitative and so it cannot be used as a screening test for disseminated intravascular coagulation or other conditions that can cause elevation of d-dimer levels.
Initial studies using SimpliRED reported sensitivities for the presence of deep vein thrombosis or pulmonary embolism > 90% (Ginsberg et al, 1995, 1998; Turkstra et al, 1996; Wells et al, 1998b). More recent studies, which have used larger patient populations (Janssen et al, 1997; Freyburger et al, 1998; Ginsberg et al, 1998), suggest that the true sensitivity of SimpliRED for venous thromboembolism is < 90% (Janssen et al, 1997; Freyburger et al, 1998; Ginsberg, 1998). This is probably because the gold standard tests were not used in all the earlier studies and so some smaller venous clots went undetected. Other sensitive d-dimer tests have since been shown to have higher sensitivities than the SimpliRED (Janssen et al, 1997; Freyburger et al, 1998; Ginsberg, 1998). Recent work has illustrated that despite reasonable negative predictive values for all patients, a negative SimpliRED result is only able to safely exclude deep vein thrombosis or pulmonary embolism in patients with low pre-test clinical likelihood. This means that d-dimer testing is not an appropriate stand-alone test (Ginsberg et al, 1998; Wells et al, 1998b; Farrel et al, 2000).
Accuclot is a manual semiquantitative latex agglutination method that is potentially susceptible to observer bias. It does have the advantage compared with SimpliRED that it could be used as a screening test for disseminated intravascular coagulation if performed in dilutions to create a higher cut-off for positive results. Hence, it would probably be appropriate as a sole d-dimer method for coagulation laboratories that are investigating for both venous thromboembolism and disseminated intravascular coagulation.
IL-Test is also appropriate for the investigation of both venous thromboembolism and disseminated intravascular coagulation because a quantitative result is obtained. It has the distinct advantage that it is fully automated and may be more appropriate for busy coagulation laboratories that are investigating both of these conditions.
Limitations of our results are the fact we batch tested Accuclot and IL-Test methods by the same technologist, whereas SimpliRED was performed by several different technologists at the time of initial patient presentation. All testing was, however, carried out without knowledge of either the patient's pre-test clinical likelihood or subsequent results of objective tests to avoid bias. Our study may have an incorporation bias potentially favouring SimpliRED because the investigation for venous thromboembolism was influenced by the SimpliRED test result and not the other two.
On the basis of this study, we conclude that the SimpliRED d-dimer is less sensitive than Accuclot or IL-Test methods when used as a screening test for venous thromboembolism. None is sufficiently sensitive to use a negative result in isolation to rule out deep vein thrombosis; that is, it must be used together with pre-test clinical likelihood. However, as has been previously demonstrated for SimpliRED and others, when combined with clinical assessment a negative result will exclude deep vein thrombosis or pulmonary embolism if pre-test clinical probability is low (Ginsberg et al, 1998; Anderson & Wells, 1999). Negative results in patients with moderate or high pre-test clinical probability has the possible potential to exclude deep vein thrombosis if the initial ultrasound is negative or V/Q scan non-diagnostic (Ginsberg et al, 1998; Wells et al, 1999b; Farrel et al, 2000). Further assessment by large management studies is appropriate.