- Split View
-
Views
-
Cite
Cite
S. Goodacre, F.C. Sampson, A.J. Sutton, S. Mason, F. Morris, Variation in the diagnostic performance of D-dimer for suspected deep vein thrombosis, QJM: An International Journal of Medicine, Volume 98, Issue 7, July 2005, Pages 513–527, https://doi.org/10.1093/qjmed/hci085
- Share Icon Share
Abstract
Background: Numerous studies have evaluated the accuracy of D-dimer in diagnosing suspected deep vein thrombosis (DVT), but results are conflicting.
Aim: To overview estimates of the diagnostic accuracy of D-dimer and identify causes of variation.
Design: Systematic review, meta-analysis and meta-regression.
Methods: We searched Medline, EMBASE, CINAHL, Web of Science, Cochrane Database of Systematic Reviews, Cochrane Controlled Trials Register, Database of Reviews of Effectiveness, the ACP Journal Club, citation lists, and contacted manufacturers. We selected studies that compared D-dimer to a reference standard in patients with suspected DVT. Data were analysed by random effects meta-analysis and meta-regression.
Results: We included 97 studies reporting 198 assays in 99 different patient groups. Overall estimated sensitivity and specificity of D-dimer were 90.5% and 54.7%, but both estimates were subject to significant heterogeneity (p < 0.001). Meta-regression identified that some heterogeneity was explained by study setting, exclusion criteria, whether recruitment was consecutive or the study prospective, whether D-dimer and the reference standard were measured blind, and whether the D-dimer threshold was determined a priori. Sensitivity and specificity also varied between ELISA (94% and 45% respectively), latex (89% and 55%) and whole blood agglutination assays (87% and 68%). Sensitivity was higher for proximal than distal DVT. Specificity was dependent upon whether clinical probability of DVT was high (specificity 51%), intermediate (67%) or low (78%).
Discussion: D-dimer has good sensitivity, but poor specificity, for DVT. Estimates are subject to substantial heterogeneity from various sources. D-dimer specificity appears to be strongly dependent upon the pre-test clinical probability of DVT.
Introduction
Clinical suspicion of deep vein thrombosis (DVT) is a frequent cause for emergency diagnostic assessment.1 The D-dimer assay is a simple, cheap test that may be useful in DVT diagnosis. Numerous research studies have been published evaluating the diagnostic accuracy of D-dimer for DVT, and attempts have been made to systematically review these studies and draw broad conclusions.2–4 The most recently published review4 examined the use of D-dimer for both DVT and pulmonary embolus (PE), and concluded that an enzyme-linked immunoassay (ELISA) D-dimer had a negative predictive value for DVT that was as diagnostically useful as negative duplex ultrasonography findings. However this analysis also reported that the ELISA assay has poor specificity, a finding that may limit its usefulness in practice.
Several issues were not explored in this review,4 or others,2,3 and merit detailed analysis. Although there is substantial evidence of variation in the results of different studies, only a limited search for causes of heterogeneity was undertaken. Identifying sources of heterogeneity by using techniques such as meta-regression may produce valuable information with which to interpret results, extrapolate findings to different settings, and guide future research. Two specific issues warrant particular attention. Firstly, D-dimer may have greater sensitivity for proximal (rather than distal) DVT. Proximal DVT carries a higher risk of propagation than distal DVT,5 so missed distal DVT is not as potentially serious. Ultrasonography has become a widely accepted diagnostic test for DVT despite limited sensitivity for distal DVT, because its sensitivity for proximal DVT is high.6 Secondly, the diagnostic parameters of D-dimer may vary with the pre-test clinical probability of DVT. This has important implications if D-dimer is used in a Bayesian framework (i.e. by applying the likelihood ratio for a D-dimer result to an estimate of pre-test probability, to derive an estimate of post-test probability), because Bayes' Theorem assumes that diagnostic test performance is independent of pre-test probability.
We aimed to: (i) undertake a systematic review of the literature and meta-analysis to estimate the sensitivity and specificity of D-dimer in the diagnosis of patients presenting with clinically suspected DVT; (ii) undertake meta-regression to identify causes of heterogeneity between studies; and (iii) undertake meta-analysis limited to studies that report proximal and distal DVT separately, or report results stratified by pre-test clinical probability.
Methods
Search strategy
We sought to identify all diagnostic cohort studies of patients with clinically suspected DVT who underwent testing with D-dimer followed by a reference standard investigation. We searched Medline, EMBASE, CINAHL, Web of Science, Cochrane Database of Systematic Reviews, Cochrane Controlled Trials Register, Database of Reviews of Effectiveness, and ACP Journal Club using the search strategy outlined in the Appendix. The bibliographies of all articles selected for the review were scanned for potentially relevant articles that were not identified by the original search. Manufacturers of D-dimer assays were contacted and asked to supply any unpublished studies.
Selection of articles
Two reviewers (FS and SG) screened the titles and abstracts of all articles to independently identify potentially relevant articles. Full copies of all selected articles were retrieved. These were then reviewed by the same two reviewers, who independently selected articles that measured the diagnostic performance of D-dimer, compared to a reference standard of venography, ultrasound or plethysmography, in patients with suspected DVT. At both stages of selection, a kappa score was calculated and disagreements resolved by discussion. Studies published in English, French, Spanish, Italian or German were included. Studies published in other languages were excluded. Abstracts and letters were included if they reported data in sufficient detail to allow inclusion in the analysis. If not, the authors were contacted and asked to provide details of the data or any full publications.
We specifically excluded: studies that measured the risk of DVT developing after D-dimer testing, rather than the probability of DVT being present at the time of testing; case-control studies, in which D-dimer measurements in a group of patients with DVT were compared to a control group of patients without DVT; management studies that ruled out DVT on the basis of a negative D-dimer alone; studies that used a reference standard other than venography, ultrasound or impedance plethysmography; studies with <10 patients; and studies of patients with suspected pulmonary embolus.
Data extraction
Two independent reviewers (SG and FM or SM) extracted the following data from the selected studies onto a standardized pro forma: the setting for patient recruitment, any exclusion criteria that might influence D-dimer performance (malignancy, post-operative patients, pregnant, anticoagulated, trauma, sepsis, or prolonged history), population demographics, whether recruitment was consecutive and/or data collection prospective, which D-dimer assay(s) were used, the threshold value(s) used, the reference standard used, and the number of true positives (proximal and distal), true negatives, false positives and false negatives (proximal and distal), either as reported or calculated from the reported data. Discrepancies were checked and resolved by an independent reviewer (FS). If it was not possible to extract the necessary data from the published report, we contacted the authors for clarification. We reviewed the data reported by each study, and removed studies that contained duplicated data. Although we collected data from groups of asymptomatic patients receiving D-dimer testing, we have only reported data here from patients with clinically suspected DVT.
Quality assessment
The same two reviewers independently assessed the quality of each study according to the following criteria, with disagreements being resolved by an independent reviewer (FS): (i) was application of the reference standard independent of the D-dimer result? (ii) was D-dimer measured by operators who were blind to the reference standard result? (iii) was the reference standard performed by operators who were blind to the D-dimer result? (iv) was a pre-determined threshold value for D-dimer used (a priori threshold value), or was the threshold derived from the data (post hoc threshold)?
Statistical analysis
Analyses used MetaDiSc statistical software.7 Each analysis used a random effects model to estimate overall sensitivity and specificity, and a χ2 test for heterogeneity between studies. Where zero counts occurred for study data, a continuity correction of 0.5 was added to the raw data for that study in order to make the calculation of sensitivity and specificity defined. Initially all studies were analysed together and random effects meta-regression undertaken to identify potential causes of heterogeneity for sensitivity and specificity separately. Any covariate that showed an association with sensitivity or specificity (p < 0.1) was selected, and subgroups of studies identified by such covariates were meta-analysed separately. We also undertook the following separate meta-analyses: studies of ELISA, latex and whole-blood agglutination assays; proximal and distal DVT (where studies reported them separately); studies of each individual D-dimer assay; studies reporting results by Wells clinical risk stratification;8 and groups of patients with malignancy.
Funnel plots for sensitivity and specificity were constructed to assess the possibility of publication bias. Such plots are constructed by plotting the log odds of sensitivity and specificity against their corresponding standard errors. This transformed scale should produce a funnel that is approximately symmetrical if no publication bias is present. Fixed-effect pooled estimates, together with guidelines in which 95% of studies should lie if no heterogeneity is present, are overlaid on the plots to aid interpretation of heterogeneity and bias.
Results
The flow of articles considered for the review is outlined in Figure 1. We scanned the titles or abstracts of 1140 articles and selected 247 potentially relevant articles for retrieval (kappa 0.87). Review of the full articles identified 131 that met the inclusion criteria (kappa 0.94). Six of these contained duplicated data9–14 and were excluded. We were unable to extract or analyse appropriate data from a further 15 articles.15–29 Thirteen articles reported groups of asymptomatic patients and are not reported here.30–42 Thus 97 articles were included in the meta-analysis:43–139 90 were published in English, three in French, two in Spanish, and one each in Italian and German. Two articles111,136 reported two groups of patients, so a total of 99 groups were analysed. The number of assays tested on each groups varied from one to 13. Overall, 198 analyses of D-dimer assays were reported.
The patients were recruited from the following settings: emergency department 16, out-patient referrals 31, in-patients 9, mixed setting 29, not reported 14. Exclusions were reported for 49 groups: 10 excluded post-operative patients, 19 excluded pregnant patients, 33 excluded anticoagulated patients, 23 excluded those with previous thromboembolism, three excluded those with recent trauma, four excluded those with sepsis, and 18 excluded patients with a prolonged history.
The mean or median age of the sample varied from 51 to 69 years (median 59), with the exception of one study that recruited exclusively over 70s. The proportion of males ranged from 17% to 62% (median 42%). Overall prevalence of DVT varied from 2% to 78% (median 36%). Prevalence of proximal and distal DVT was reported for 51 groups. The proportion of proximal DVT (of all DVT detected) varied from 27% to 100% (median 77%).
The reference standard consisted of venography (34 studies), ultrasound alone (28), ultrasound with clinical follow-up (10), serial ultrasound (6), either ultrasound or venography (13), or a combination of ultrasound and plethysmography (8). The reference standard was definitely independent of the results of D-dimer testing in 86, D-dimer measurement was definitely blind to the reference standard in 43, the reference standard measurement was definitely blind to D-dimer in 50, and the threshold value for D-dimer was defined before analysis in 82.
Figure 2 shows the results for all studies plotted on an ROC plane. The true positive rate (sensitivity) is plotted against the false positive rate (1-specificity) for each study included in the meta-analysis. A study of a perfect test (100% sensitivity and specificity) would be plotted in the top left hand corner. The ROC curve thus demonstrates the overall diagnostic performance of D-dimer assays and the variation between the results of individual studies. The pooled estimates (95%CI) for sensitivity and specificity were 90.5% (90.0–91.1) and 54.7% (54.0–55.4) respectively, but the results show substantial heterogeneity, particularly for specificity. Sensitivity ranged from 48% to 100% (χ2 test for heterogeneity p < 0.001). Specificity ranged from 5% to 100% (χ2 test for heterogeneity p < 0.001).
Table 1 shows the results of meta-regression. A number of covariates were associated with systematic variation in sensitivity and specificity. However the significant associations for age and D-dimer threshold were due to one group in each analysis having an extreme value for the covariate that exerted substantial leverage upon the regression line: Le Blanche et al. selected patients aged over 70 to recruit a group with a mean age of 86 years,78 while Crippa et al. used a threshold of 2300 ng/ml for positivity.128 Neither association was significant when the group with an extreme value was removed.
Variable . | . | Sensitivity p . | Specificity p . |
---|---|---|---|
Patients recruited | Mixed | 0.030 | 0.010 |
ED only | 0.32 | 0.065 | |
Out-patient only | 0.010 | 0.022 | |
In-patient only | 0.94 | 0.24 | |
Patients excluded | Post-operative | 0.85 | 0.92 |
Pregnant | 0.049 | 0.079 | |
Anticoagulated | 0.011 | 0.007 | |
Past history of DVT | 0.38 | <0.001 | |
Trauma | 0.47 | 0.21 | |
Sepsis | 0.76 | 0.89 | |
Long history | 0.061 | 0.82 | |
Mean age | 0.046 (0.11)* | 0.01 (0.21)* | |
% Males | 0.30 | 0.13 | |
Prevalence of DVT | 0.27 | 0.38 | |
Consecutive patients recruited | 0.47 | 0.016 | |
Prospective study | 0.026 | 0.047 | |
D-dimer threshold used | 0.090 (0.97)** | 0.68 (0.45)** | |
Reference standard used (venography vs. other) | 0.37 | <0.001 | |
Reference standard independent of D-dimer result | 0.11 | 0.21 | |
D-dimer measured blind to reference standard | 0.40 | 0.054 | |
Reference standard measured blind to D-dimer | 0.98 | 0.035 | |
D-dimer threshold defined before the study | 0.002 | 0.34 |
Variable . | . | Sensitivity p . | Specificity p . |
---|---|---|---|
Patients recruited | Mixed | 0.030 | 0.010 |
ED only | 0.32 | 0.065 | |
Out-patient only | 0.010 | 0.022 | |
In-patient only | 0.94 | 0.24 | |
Patients excluded | Post-operative | 0.85 | 0.92 |
Pregnant | 0.049 | 0.079 | |
Anticoagulated | 0.011 | 0.007 | |
Past history of DVT | 0.38 | <0.001 | |
Trauma | 0.47 | 0.21 | |
Sepsis | 0.76 | 0.89 | |
Long history | 0.061 | 0.82 | |
Mean age | 0.046 (0.11)* | 0.01 (0.21)* | |
% Males | 0.30 | 0.13 | |
Prevalence of DVT | 0.27 | 0.38 | |
Consecutive patients recruited | 0.47 | 0.016 | |
Prospective study | 0.026 | 0.047 | |
D-dimer threshold used | 0.090 (0.97)** | 0.68 (0.45)** | |
Reference standard used (venography vs. other) | 0.37 | <0.001 | |
Reference standard independent of D-dimer result | 0.11 | 0.21 | |
D-dimer measured blind to reference standard | 0.40 | 0.054 | |
Reference standard measured blind to D-dimer | 0.98 | 0.035 | |
D-dimer threshold defined before the study | 0.002 | 0.34 |
p values when included in random effect weighted meta-regression. *The revised value excluded one study with an extreme value for mean age. **The revised value excluded one study that used an extreme threshold value.
Variable . | . | Sensitivity p . | Specificity p . |
---|---|---|---|
Patients recruited | Mixed | 0.030 | 0.010 |
ED only | 0.32 | 0.065 | |
Out-patient only | 0.010 | 0.022 | |
In-patient only | 0.94 | 0.24 | |
Patients excluded | Post-operative | 0.85 | 0.92 |
Pregnant | 0.049 | 0.079 | |
Anticoagulated | 0.011 | 0.007 | |
Past history of DVT | 0.38 | <0.001 | |
Trauma | 0.47 | 0.21 | |
Sepsis | 0.76 | 0.89 | |
Long history | 0.061 | 0.82 | |
Mean age | 0.046 (0.11)* | 0.01 (0.21)* | |
% Males | 0.30 | 0.13 | |
Prevalence of DVT | 0.27 | 0.38 | |
Consecutive patients recruited | 0.47 | 0.016 | |
Prospective study | 0.026 | 0.047 | |
D-dimer threshold used | 0.090 (0.97)** | 0.68 (0.45)** | |
Reference standard used (venography vs. other) | 0.37 | <0.001 | |
Reference standard independent of D-dimer result | 0.11 | 0.21 | |
D-dimer measured blind to reference standard | 0.40 | 0.054 | |
Reference standard measured blind to D-dimer | 0.98 | 0.035 | |
D-dimer threshold defined before the study | 0.002 | 0.34 |
Variable . | . | Sensitivity p . | Specificity p . |
---|---|---|---|
Patients recruited | Mixed | 0.030 | 0.010 |
ED only | 0.32 | 0.065 | |
Out-patient only | 0.010 | 0.022 | |
In-patient only | 0.94 | 0.24 | |
Patients excluded | Post-operative | 0.85 | 0.92 |
Pregnant | 0.049 | 0.079 | |
Anticoagulated | 0.011 | 0.007 | |
Past history of DVT | 0.38 | <0.001 | |
Trauma | 0.47 | 0.21 | |
Sepsis | 0.76 | 0.89 | |
Long history | 0.061 | 0.82 | |
Mean age | 0.046 (0.11)* | 0.01 (0.21)* | |
% Males | 0.30 | 0.13 | |
Prevalence of DVT | 0.27 | 0.38 | |
Consecutive patients recruited | 0.47 | 0.016 | |
Prospective study | 0.026 | 0.047 | |
D-dimer threshold used | 0.090 (0.97)** | 0.68 (0.45)** | |
Reference standard used (venography vs. other) | 0.37 | <0.001 | |
Reference standard independent of D-dimer result | 0.11 | 0.21 | |
D-dimer measured blind to reference standard | 0.40 | 0.054 | |
Reference standard measured blind to D-dimer | 0.98 | 0.035 | |
D-dimer threshold defined before the study | 0.002 | 0.34 |
p values when included in random effect weighted meta-regression. *The revised value excluded one study with an extreme value for mean age. **The revised value excluded one study that used an extreme threshold value.
Random effects meta-analysis was repeated, stratified by each significant covariate, to estimate sensitivity and/or specificity for studies with, or without, the covariate. The results are shown in Table 2 More selective groups (i.e. recruited from out-patient or ED only, or reporting exclusion criteria) tended to have higher sensitivity and specificity. Higher quality studies (prospective studies, those recruiting consecutive patients, venographic reference standard, D-dimer and reference standard measured blind) tended to have higher specificity. Studies that determined the D-dimer threshold after data analysis had higher sensitivity. However, there was still substantial heterogeneity among results, even when stratified by these criteria.
Variable . | Sensitivity . | Heterogeneity p . | Specificity . | Heterogeneity p . |
---|---|---|---|---|
Mixed patient selection | 87% (86–88) | <0.001 | 50% (49–51) | <0.001 |
ED patients only | 89% (87–91) | <0.001 | 62% (60–64) | <0.001 |
Out-patients only | 94% (93–95) | <0.001 | 59% (58–60) | <0.001 |
Studies excluding pregnant patients | 95% (93–97) | <0.001 | 57% (56–58) | <0.001 |
Studies excluding anticoagulated patients | 93% (92–94) | <0.001 | 61% (60–62) | <0.001 |
Studies excluding patient with past history of thromboembolism | 91% (90–93) | <0.001 | 65% (63–66) | <0.001 |
Studies excluding patients with a prolonged history | 93% (92–94) | <0.001 | 54% (52–56) | <0.001 |
Consecutive patients recruited | 91% (90–92) | <0.001 | 57% (56–58) | <0.001 |
Prospective study | 90% (89–91) | <0.001 | 58% (57–59) | <0.001 |
Venographic reference standard | 91% (90–92) | <0.001 | 62% (61–64) | <0.001 |
D-dimer measured blind to reference standard | 90% (89–91) | <0.001 | 56% (55–57) | <0.001 |
Reference standard measured blind | 90% (89–91) | <0.001 | 57% (56–58) | <0.001 |
D-dimer threshold derived from data | 95% (94–96) | 0.245 | 41% (39–44) | <0.001 |
Variable . | Sensitivity . | Heterogeneity p . | Specificity . | Heterogeneity p . |
---|---|---|---|---|
Mixed patient selection | 87% (86–88) | <0.001 | 50% (49–51) | <0.001 |
ED patients only | 89% (87–91) | <0.001 | 62% (60–64) | <0.001 |
Out-patients only | 94% (93–95) | <0.001 | 59% (58–60) | <0.001 |
Studies excluding pregnant patients | 95% (93–97) | <0.001 | 57% (56–58) | <0.001 |
Studies excluding anticoagulated patients | 93% (92–94) | <0.001 | 61% (60–62) | <0.001 |
Studies excluding patient with past history of thromboembolism | 91% (90–93) | <0.001 | 65% (63–66) | <0.001 |
Studies excluding patients with a prolonged history | 93% (92–94) | <0.001 | 54% (52–56) | <0.001 |
Consecutive patients recruited | 91% (90–92) | <0.001 | 57% (56–58) | <0.001 |
Prospective study | 90% (89–91) | <0.001 | 58% (57–59) | <0.001 |
Venographic reference standard | 91% (90–92) | <0.001 | 62% (61–64) | <0.001 |
D-dimer measured blind to reference standard | 90% (89–91) | <0.001 | 56% (55–57) | <0.001 |
Reference standard measured blind | 90% (89–91) | <0.001 | 57% (56–58) | <0.001 |
D-dimer threshold derived from data | 95% (94–96) | 0.245 | 41% (39–44) | <0.001 |
Variable . | Sensitivity . | Heterogeneity p . | Specificity . | Heterogeneity p . |
---|---|---|---|---|
Mixed patient selection | 87% (86–88) | <0.001 | 50% (49–51) | <0.001 |
ED patients only | 89% (87–91) | <0.001 | 62% (60–64) | <0.001 |
Out-patients only | 94% (93–95) | <0.001 | 59% (58–60) | <0.001 |
Studies excluding pregnant patients | 95% (93–97) | <0.001 | 57% (56–58) | <0.001 |
Studies excluding anticoagulated patients | 93% (92–94) | <0.001 | 61% (60–62) | <0.001 |
Studies excluding patient with past history of thromboembolism | 91% (90–93) | <0.001 | 65% (63–66) | <0.001 |
Studies excluding patients with a prolonged history | 93% (92–94) | <0.001 | 54% (52–56) | <0.001 |
Consecutive patients recruited | 91% (90–92) | <0.001 | 57% (56–58) | <0.001 |
Prospective study | 90% (89–91) | <0.001 | 58% (57–59) | <0.001 |
Venographic reference standard | 91% (90–92) | <0.001 | 62% (61–64) | <0.001 |
D-dimer measured blind to reference standard | 90% (89–91) | <0.001 | 56% (55–57) | <0.001 |
Reference standard measured blind | 90% (89–91) | <0.001 | 57% (56–58) | <0.001 |
D-dimer threshold derived from data | 95% (94–96) | 0.245 | 41% (39–44) | <0.001 |
Variable . | Sensitivity . | Heterogeneity p . | Specificity . | Heterogeneity p . |
---|---|---|---|---|
Mixed patient selection | 87% (86–88) | <0.001 | 50% (49–51) | <0.001 |
ED patients only | 89% (87–91) | <0.001 | 62% (60–64) | <0.001 |
Out-patients only | 94% (93–95) | <0.001 | 59% (58–60) | <0.001 |
Studies excluding pregnant patients | 95% (93–97) | <0.001 | 57% (56–58) | <0.001 |
Studies excluding anticoagulated patients | 93% (92–94) | <0.001 | 61% (60–62) | <0.001 |
Studies excluding patient with past history of thromboembolism | 91% (90–93) | <0.001 | 65% (63–66) | <0.001 |
Studies excluding patients with a prolonged history | 93% (92–94) | <0.001 | 54% (52–56) | <0.001 |
Consecutive patients recruited | 91% (90–92) | <0.001 | 57% (56–58) | <0.001 |
Prospective study | 90% (89–91) | <0.001 | 58% (57–59) | <0.001 |
Venographic reference standard | 91% (90–92) | <0.001 | 62% (61–64) | <0.001 |
D-dimer measured blind to reference standard | 90% (89–91) | <0.001 | 56% (55–57) | <0.001 |
Reference standard measured blind | 90% (89–91) | <0.001 | 57% (56–58) | <0.001 |
D-dimer threshold derived from data | 95% (94–96) | 0.245 | 41% (39–44) | <0.001 |
ELISA assays were reported by 91 analyses in 58 cohorts (35 analyses reported proximal and distal DVT separately). Pooled sensitivity was 94% (95%CI 93–95) and specificity 45% (95%CI 44–46). Latex assays were reported by 74 analyses in 52 cohorts (22 analyses reported proximal and distal DVT separately). Pooled sensitivity was 89% (95%CI 88–90) and specificity 55% (95%CI 54–56). Whole-blood agglutination assays were reported by 29 analyses in 29 cohorts (10 analyses reported proximal and distal DVT separately). Pooled sensitivity was 87% (95%CI 85–88) and specificity 68% (95%CI 67–69). Analysis of studies that reported proximal and distal DVT separately showed that all assays had higher sensitivity for proximal than distal DVT: 98% (95%CI 97–99) vs. 86% (95%CI 84–88) for ELISA, 94% (95%CI 92–95) vs. 79% (95%CI 75–83) for latex, and 84% (95%CI 80–88) vs. 64% (95%CI 55–73) for whole-blood agglutination.
Table 3 shows the pooled estimates of sensitivity and specificity for individual assays that were investigated by three or more eligible studies, and are thus suitable for meta-analysis. Although differences are apparent between the individual assays, the confidence intervals for estimates are relatively wide, suggesting that differences may not be significant. Furthermore, even these estimates for individual assays are subject to significant heterogeneity, suggesting that heterogeneity in estimates of D-dimer sensitivity and specificity cannot be explained simply differences between assays.
Test . | Assay type . | Manufacturer . | Sensitivity . | Specificity . |
---|---|---|---|---|
Asserachrom (n = 19)58,62,67,69,75,81,84,90,93,98,100,106,111,116,123,124,132,138 | Classical ELISA | Diagnostica Stago | 95% (93–96) p = 0.005 | 36% (34–38) p < 0.001 |
Enzygnost (n = 5)69,73,95,100,123 | Classical ELISA | Dade Behring | 96% (92–98) p = 0.046 | 62% (57–67) p < 0.001 |
Fibrinostika (n = 3)69,100,102 | Classical ELISA | Organon Teknika | 95% (90–98) p = 0.063 | 37% (30–45) p = 0.761 |
Dimertest EIA Gold (n = 6)62,77,91,103,109,129 | Classical ELISA | Agen | 90% (85–93) p = 0.023 | 67% (62–71) p < 0.001 |
VIDAS (n = 17)43,46,48,58,59,69,71,77,78,81,91,98,99,100,101,138,139 | Rapid ELISA | Biomerieux | 96% (95–98) p = 0.008 | 39% (36–42) p < 0.001 |
Nycocard (n = 14)48,63,66,69,70,75,82,86,91,95,97,100,107,136 | Rapid ELISA | Nycomed | 89% (87–91) p < 0.001 | 50% (47–54) p < 0.001 |
Instant IA (n = 8)69,75,81,89,91,95,96,100 | Rapid ELISA | Diagnostica Stago | 92% (90–94) p = 0.478 | 54% (49–58) p < 0.001 |
Minutex (n = 8)52,59,69,75,82,91,99,123 | Latex | Biopool | 92% (90–94) p < 0.001 | 45% (42–49) p < 0.001 |
D-Di latex (n = 4)75,106,132,116 | Latex | Diagnostica Stago | 80% (72–86) p < 0.001 | 58% (51–64) p < 0.001 |
LPIA (n = 3)91,128,129 | Latex | Mitsubishi Kasei | 98% (94–99) p = 0.482 | 58% (51–65) p = 0.244 |
BC (n = 4)69,73,77,123 | Latex | Dade Behring | 90% (84–94) p < 0.001 | 53% (47–60) p < 0.001 |
IL test (n = 6)46,59,60,68,69,126 | Latex | Instrumentation Laboratories, Milano | 94% (91–97) p = 0.077 | 52% (47–56) p < 0.001 |
Miniquant (n = 4)49,59,62,138 | Latex | Biopool | 96% (91–99) p = 0.528 | 36% (31–40) p < 0.001 |
STA-liatest (n = 8) 44,45,46,47,54,57,69,84 | Latex | Diagnostica Stago | 94% (92–96) p = 0.883 | 46% (43–50) p = 0.002 |
Tinaquant (n = 7)57,58,69,71,88,99,125 | Latex | Roche | 96% (93–97) p = 0.023 | 52% (47–56) p = 0.014 |
Turbiquant (n = 3)46,69,71 | Latex | Dade Behring | 88% (81–93) p = 0.439 | 46% (41–52) p = 0.042 |
SimpliRED (n = 29)50,51,55,60,62,64,65,69,72,74,76,79,80,83,85,87,92,94,99,103,105,123,124,127,131,133,136,138 | Whole blood agglutination | Agen | 87% (85–88) p < 0.001 | 68% (67–69) p < 0.001 |
Test . | Assay type . | Manufacturer . | Sensitivity . | Specificity . |
---|---|---|---|---|
Asserachrom (n = 19)58,62,67,69,75,81,84,90,93,98,100,106,111,116,123,124,132,138 | Classical ELISA | Diagnostica Stago | 95% (93–96) p = 0.005 | 36% (34–38) p < 0.001 |
Enzygnost (n = 5)69,73,95,100,123 | Classical ELISA | Dade Behring | 96% (92–98) p = 0.046 | 62% (57–67) p < 0.001 |
Fibrinostika (n = 3)69,100,102 | Classical ELISA | Organon Teknika | 95% (90–98) p = 0.063 | 37% (30–45) p = 0.761 |
Dimertest EIA Gold (n = 6)62,77,91,103,109,129 | Classical ELISA | Agen | 90% (85–93) p = 0.023 | 67% (62–71) p < 0.001 |
VIDAS (n = 17)43,46,48,58,59,69,71,77,78,81,91,98,99,100,101,138,139 | Rapid ELISA | Biomerieux | 96% (95–98) p = 0.008 | 39% (36–42) p < 0.001 |
Nycocard (n = 14)48,63,66,69,70,75,82,86,91,95,97,100,107,136 | Rapid ELISA | Nycomed | 89% (87–91) p < 0.001 | 50% (47–54) p < 0.001 |
Instant IA (n = 8)69,75,81,89,91,95,96,100 | Rapid ELISA | Diagnostica Stago | 92% (90–94) p = 0.478 | 54% (49–58) p < 0.001 |
Minutex (n = 8)52,59,69,75,82,91,99,123 | Latex | Biopool | 92% (90–94) p < 0.001 | 45% (42–49) p < 0.001 |
D-Di latex (n = 4)75,106,132,116 | Latex | Diagnostica Stago | 80% (72–86) p < 0.001 | 58% (51–64) p < 0.001 |
LPIA (n = 3)91,128,129 | Latex | Mitsubishi Kasei | 98% (94–99) p = 0.482 | 58% (51–65) p = 0.244 |
BC (n = 4)69,73,77,123 | Latex | Dade Behring | 90% (84–94) p < 0.001 | 53% (47–60) p < 0.001 |
IL test (n = 6)46,59,60,68,69,126 | Latex | Instrumentation Laboratories, Milano | 94% (91–97) p = 0.077 | 52% (47–56) p < 0.001 |
Miniquant (n = 4)49,59,62,138 | Latex | Biopool | 96% (91–99) p = 0.528 | 36% (31–40) p < 0.001 |
STA-liatest (n = 8) 44,45,46,47,54,57,69,84 | Latex | Diagnostica Stago | 94% (92–96) p = 0.883 | 46% (43–50) p = 0.002 |
Tinaquant (n = 7)57,58,69,71,88,99,125 | Latex | Roche | 96% (93–97) p = 0.023 | 52% (47–56) p = 0.014 |
Turbiquant (n = 3)46,69,71 | Latex | Dade Behring | 88% (81–93) p = 0.439 | 46% (41–52) p = 0.042 |
SimpliRED (n = 29)50,51,55,60,62,64,65,69,72,74,76,79,80,83,85,87,92,94,99,103,105,123,124,127,131,133,136,138 | Whole blood agglutination | Agen | 87% (85–88) p < 0.001 | 68% (67–69) p < 0.001 |
Test . | Assay type . | Manufacturer . | Sensitivity . | Specificity . |
---|---|---|---|---|
Asserachrom (n = 19)58,62,67,69,75,81,84,90,93,98,100,106,111,116,123,124,132,138 | Classical ELISA | Diagnostica Stago | 95% (93–96) p = 0.005 | 36% (34–38) p < 0.001 |
Enzygnost (n = 5)69,73,95,100,123 | Classical ELISA | Dade Behring | 96% (92–98) p = 0.046 | 62% (57–67) p < 0.001 |
Fibrinostika (n = 3)69,100,102 | Classical ELISA | Organon Teknika | 95% (90–98) p = 0.063 | 37% (30–45) p = 0.761 |
Dimertest EIA Gold (n = 6)62,77,91,103,109,129 | Classical ELISA | Agen | 90% (85–93) p = 0.023 | 67% (62–71) p < 0.001 |
VIDAS (n = 17)43,46,48,58,59,69,71,77,78,81,91,98,99,100,101,138,139 | Rapid ELISA | Biomerieux | 96% (95–98) p = 0.008 | 39% (36–42) p < 0.001 |
Nycocard (n = 14)48,63,66,69,70,75,82,86,91,95,97,100,107,136 | Rapid ELISA | Nycomed | 89% (87–91) p < 0.001 | 50% (47–54) p < 0.001 |
Instant IA (n = 8)69,75,81,89,91,95,96,100 | Rapid ELISA | Diagnostica Stago | 92% (90–94) p = 0.478 | 54% (49–58) p < 0.001 |
Minutex (n = 8)52,59,69,75,82,91,99,123 | Latex | Biopool | 92% (90–94) p < 0.001 | 45% (42–49) p < 0.001 |
D-Di latex (n = 4)75,106,132,116 | Latex | Diagnostica Stago | 80% (72–86) p < 0.001 | 58% (51–64) p < 0.001 |
LPIA (n = 3)91,128,129 | Latex | Mitsubishi Kasei | 98% (94–99) p = 0.482 | 58% (51–65) p = 0.244 |
BC (n = 4)69,73,77,123 | Latex | Dade Behring | 90% (84–94) p < 0.001 | 53% (47–60) p < 0.001 |
IL test (n = 6)46,59,60,68,69,126 | Latex | Instrumentation Laboratories, Milano | 94% (91–97) p = 0.077 | 52% (47–56) p < 0.001 |
Miniquant (n = 4)49,59,62,138 | Latex | Biopool | 96% (91–99) p = 0.528 | 36% (31–40) p < 0.001 |
STA-liatest (n = 8) 44,45,46,47,54,57,69,84 | Latex | Diagnostica Stago | 94% (92–96) p = 0.883 | 46% (43–50) p = 0.002 |
Tinaquant (n = 7)57,58,69,71,88,99,125 | Latex | Roche | 96% (93–97) p = 0.023 | 52% (47–56) p = 0.014 |
Turbiquant (n = 3)46,69,71 | Latex | Dade Behring | 88% (81–93) p = 0.439 | 46% (41–52) p = 0.042 |
SimpliRED (n = 29)50,51,55,60,62,64,65,69,72,74,76,79,80,83,85,87,92,94,99,103,105,123,124,127,131,133,136,138 | Whole blood agglutination | Agen | 87% (85–88) p < 0.001 | 68% (67–69) p < 0.001 |
Test . | Assay type . | Manufacturer . | Sensitivity . | Specificity . |
---|---|---|---|---|
Asserachrom (n = 19)58,62,67,69,75,81,84,90,93,98,100,106,111,116,123,124,132,138 | Classical ELISA | Diagnostica Stago | 95% (93–96) p = 0.005 | 36% (34–38) p < 0.001 |
Enzygnost (n = 5)69,73,95,100,123 | Classical ELISA | Dade Behring | 96% (92–98) p = 0.046 | 62% (57–67) p < 0.001 |
Fibrinostika (n = 3)69,100,102 | Classical ELISA | Organon Teknika | 95% (90–98) p = 0.063 | 37% (30–45) p = 0.761 |
Dimertest EIA Gold (n = 6)62,77,91,103,109,129 | Classical ELISA | Agen | 90% (85–93) p = 0.023 | 67% (62–71) p < 0.001 |
VIDAS (n = 17)43,46,48,58,59,69,71,77,78,81,91,98,99,100,101,138,139 | Rapid ELISA | Biomerieux | 96% (95–98) p = 0.008 | 39% (36–42) p < 0.001 |
Nycocard (n = 14)48,63,66,69,70,75,82,86,91,95,97,100,107,136 | Rapid ELISA | Nycomed | 89% (87–91) p < 0.001 | 50% (47–54) p < 0.001 |
Instant IA (n = 8)69,75,81,89,91,95,96,100 | Rapid ELISA | Diagnostica Stago | 92% (90–94) p = 0.478 | 54% (49–58) p < 0.001 |
Minutex (n = 8)52,59,69,75,82,91,99,123 | Latex | Biopool | 92% (90–94) p < 0.001 | 45% (42–49) p < 0.001 |
D-Di latex (n = 4)75,106,132,116 | Latex | Diagnostica Stago | 80% (72–86) p < 0.001 | 58% (51–64) p < 0.001 |
LPIA (n = 3)91,128,129 | Latex | Mitsubishi Kasei | 98% (94–99) p = 0.482 | 58% (51–65) p = 0.244 |
BC (n = 4)69,73,77,123 | Latex | Dade Behring | 90% (84–94) p < 0.001 | 53% (47–60) p < 0.001 |
IL test (n = 6)46,59,60,68,69,126 | Latex | Instrumentation Laboratories, Milano | 94% (91–97) p = 0.077 | 52% (47–56) p < 0.001 |
Miniquant (n = 4)49,59,62,138 | Latex | Biopool | 96% (91–99) p = 0.528 | 36% (31–40) p < 0.001 |
STA-liatest (n = 8) 44,45,46,47,54,57,69,84 | Latex | Diagnostica Stago | 94% (92–96) p = 0.883 | 46% (43–50) p = 0.002 |
Tinaquant (n = 7)57,58,69,71,88,99,125 | Latex | Roche | 96% (93–97) p = 0.023 | 52% (47–56) p = 0.014 |
Turbiquant (n = 3)46,69,71 | Latex | Dade Behring | 88% (81–93) p = 0.439 | 46% (41–52) p = 0.042 |
SimpliRED (n = 29)50,51,55,60,62,64,65,69,72,74,76,79,80,83,85,87,92,94,99,103,105,123,124,127,131,133,136,138 | Whole blood agglutination | Agen | 87% (85–88) p < 0.001 | 68% (67–69) p < 0.001 |
Five studies,51,64,80,85,123 using six assays (five whole blood agglutination and one latex), reported results stratified by Wells clinical risk score. Meta-analysis of these studies showed that D-dimer had a sensitivity (95% CI, p for heterogeneity) of 96% (92–99, p = 0.047) and specificity of 51% (38–53, p = 0.001) in patients with a high Wells score; a sensitivity of 88% (79–94, p = 0.18) and specificity of 67% (62–72, p = 0.003) in patients with an intermediate score; and a sensitivity of 91% (72–99, p = 0.69) and specificity of 78% (74–81, p < 0.001) in patients with a low Wells score. Thus the specificity of D-dimer appears to be dependent upon the clinical probability of DVT. Five studies reported patients with malignancy separately.49,50,52,80,127 Meta-analysis of these studies showed that D-dimer had a sensitivity of 95% (90–97, p = 0.026) and a specificity of 46% (39–52, p = 0.0026) in patients with malignancy.
Funnel plots of the studies are shown by sensitivity in Figure 3 and specificity in Figure 4. The funnel plot for specificity appears to be symmetrical, whereas the funnel plot for sensitivity is markedly asymmetrical. Although there are many small studies reporting high sensitivity, there appear to be very few small studies reporting low sensitivity. One possible explanation for this is publication bias: small studies have only been published if they reported high sensitivity.
Discussion
Our meta-analysis confirms that D-dimer has good sensitivity for DVT but poor specificity. ELISA assays are more sensitive, but less specific, than latex and whole blood agglutination assays. However, sensitivity and specificity vary between individual assays, so that some latex assays may have better sensitivity than some ELISA assays. Since D-dimer is principally used to rule out DVT, clinicians will usually be more concerned that the test should have optimal sensitivity, although the need for rapid results may favour latex, whole blood agglutination and rapid ELISA assays over the classical ELISA assays. Table 3 may provide some guidance in comparing assay types, but care should be taken to consider confidence intervals, heterogeneity among individual study results, and the effect of varying the threshold for positivity upon sensitivity and specificity.
There was substantial heterogeneity among results of almost all analyses. We found that patient selection and (to a lesser extent) quality criteria may explain some heterogeneity. Studies that recruited from a single setting or reported using exclusion criteria had higher sensitivity and specificity. This is probably because factors that may lead to exclusion from the studies may be associated with false positive D-dimer results (e.g. pregnancy) or false negative results (e.g. prolonged history). Studies that recruited consecutively, prospective studies, and those with blinded measurement of D-dimer and the reference standard had slightly higher specificity. This was an unexpected finding, as failure to adhere to quality criteria is usually associated with higher estimates of sensitivity and specificity.140 It is possible that this finding is confounded by another unreported factor. For example, high-quality studies may also have more rigorous procedures for measuring D-dimer or selecting patients for testing that result in improved diagnostic performance. Blinded measurement of D-dimer in particular would be unlikely to influence results, unless a qualitative assay was used. However, reporting that D-dimer was measured blind may reflect a more rigorous approach to measurement that is in turn associated with fewer false positive results. Finally, we found that using a threshold value that was derived from the study data led to higher estimates of sensitivity. This suggests that when a threshold is derived post hoc, it tends to be chosen at a level that maximizes sensitivity.
Although some covariates were highly statistically significant predictors, analysis stratified by these covariates did not dramatically affect estimates of sensitivity or specificity, and estimates remained subject to substantial heterogeneity. Much of the observed heterogeneity therefore remains unexplained and presumably due to unreported factors. Our analysis provides a clue to the cause of this unexplained heterogeneity. We found that specificity was strongly dependent upon clinical probability of DVT. This suggests that differences in clinical probability can have a powerful effect upon specificity. It also has important practical consequences. Firstly, using D-dimer in patients with a low clinical probability of DVT is likely to generate fewer false positives than one might expect from estimates of specificity derived from an unselected population. Secondly, if D-dimer specificity, and thus likelihood ratios, are dependent upon pre-test probability then a fundamental assumption of Bayes theorem (that test performance is independent of pre-test probability) is not fulfilled. This means that if we use D-dimer in a Bayesian framework (i.e. applying the likelihood ratios of D-dimer results to the pre-test odds of DVT to determine the post-test odds of DVT in an individual patient) then our estimates of post-test odds, or probability, may be misleading. Finally, it may be appropriate to use different thresholds for positivity according to pre-test probability. For example, since D-dimer is principally used to rule out DVT in low risk patients, it may be appropriate to use a lower threshold for positivity to minimize the risk of inadvertent discharge with DVT. Our data suggest that, although using a lower threshold may reduce specificity, it may still be acceptable in patients with a low clinical probability.
Analysis of studies reporting proximal and distal DVT separately showed that all three assays had higher sensitivity for proximal DVT. The pooled sensitivities of ELISA and latex assays were comparable to the estimated sensitivity of ultrasound for proximal DVT. This suggests that if we are not concerned about detecting and treating distal DVT then D-dimer may have sufficient sensitivity to rule out DVT. However, the clinical importance of distal DVT is controversial, and many clinicians would want to detect and treat this condition.141,142
Limitations
This is the most comprehensive review yet undertaken of the diagnostic performance of D-dimer for suspected DVT, but it has some limitations that must be appreciated. We undertook only limited efforts to identify unpublished data. Although publication bias is a potentially important factor in meta-analysis of diagnostic tests, there is little empirical data to determine how this might influence findings. Most published meta-analyses of diagnostic tests have only searched electronic databases and reference lists, and have not addressed the possibility of publication bias.143 This is perhaps not surprising. Registries of diagnostic test studies have not yet been established and, because diagnostic test studies are relatively easy to undertake, unpublished data could reside in a wide variety of potential locations. Tracing such data would be arduous and potentially unproductive.
The funnel plots showed that small studies reporting high sensitivity are more numerous than small studies reporting low sensitivity. It is highly plausible that this could be due to publication bias. D-dimer is likely to be most clinically useful as a rule-out test for patients at low risk of DVT. Authors may therefore be more eager to submit their work, and journals to publish it, if it reports high sensitivity. Specificity would not be expected to influence publication to a similar degree: since D-dimer is not likely to be used to definitively diagnose DVT, poor specificity would not be a barrier to publication. If this meta-analysis has been subject to publication bias, then our estimate of sensitivity is likely to represent an over-estimate.
Reporting of the studies was frequently poor and many of the details we sought for meta-regression were missing. We wanted to explore a number of other covariates, such as the proportion of the study population with co-morbidities, but were unable to because these variables were rarely reported. Even where potential sources of heterogeneity were identified it is not clear what the ‘true’ values for diagnostic parameters are. Selective groups may produce higher estimates, but these may not be accurate if they do not reflect the population in which the test will be used. Readers need to consider how the groups reported here compare to their patient population.
Implications for future research
Future studies of D-dimer should collect and report accurate data on patient selection, application of exclusion criteria, and the presence of co-morbidities. We recommend that the proportion of the study population with the following characteristics should be reported: malignancy, anticoagulant treatment, past history of DVT, pregnancy, sepsis, recent trauma or surgery. Most importantly, patients should be classified according to their clinical probability of DVT, and results reported separately for different risk groups. The Wells score is a widely used, validated score that is ideal for this purpose. Finally, in view of the evidence of publication bias, there is perhaps a need for registers of diagnostic studies, similar to those recently developed for clinical trials.
Conclusions
D-dimer has good sensitivity, but poor specificity, for diagnosing suspected DVT. All analyses showed substantial heterogeneity, which was partially explained by differences in patient selection and methodology in the individual studies. D-dimer sensitivity is higher for proximal than distal DVT, while specificity appears to be dependent upon the pre-test clinical probability of DVT, being higher in patients with a low clinical probability of DVT.
Appendix: Search strategy and years searched on each database
Strategy
Key terms and mesh headings may vary between databases.
exp ‘Sensitivity and Specificity’/
exp Diagnostic Errors/
Reference Values/
Reproducibility of Results/
likelihood functions/
1 or 2 or 3 or 4 or 5
specificity.mp. [mp=title, abstract, cas registry/ec number word, mesh subject heading]
sensitivity.mp.
false negative$.mp.
false positive$.mp.
true negative$.mp.
true positive$.mp.
predictive value$.mp.
reproducibility.mp.
ROC curve.mp.
Diagnos$.mp.
reference value$.mp.
likelihood function$.mp.
likelihood ratio$.mp.
6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19
exp Venous Thrombosis/
deep vein thrombosis.tw.
phlebothrombosis.tw.
venous thrombosis.tw.
venous thromboembolism.tw.
deep venous thrombosis.tw.
DVT.tw.
21 or 22 or 23 or 24 or 25 or 26 or 27
20 and 28
Fibrin Fibrinogen Degradation Products/ or Enzyme-Linked Immunosorbent Assay/
Reagent Kits, Diagnostic/
d-dimer.tw.
simplired.tw.
30 or 31 or 32 or 33
29 and 34
Years searched
Medline 1996–2004
CINAHL 1982–2004
Embase 1980–2004
Web of science 1985–2004
CCTR 1898–2004
CDSR 1990–2004
DARE 1995–2004
The United Kingdom Health Technology Assessment R&D Programme funded this project (reference number 02/03/01). The views and opinions expressed therein are those of the authors, and do not necessarily reflect those of the UK Department of Health. We thank Kathryn Paulucy for clerical assistance in preparing this article, Angie Ryan for assistance with the literature searches and Eddie Hampton for constructive comments during development of this project.
References
The Verity Steering Committee.
Becker DM, Philbrick JT, Bachhuber TL, Humphries JE. D-dimer testing and acute venous thromboembolism: A shortcut to accurate diagnosis?
Kelly J, Rudd A, Lewis RR, Hunt BJ. Plasma D-dimers in the diagnosis of venous thromboembolism.
Stein PD, Hull RD, Patel KC, Olson RE, Ghali WA, Brant R, et al. D-dimer for the exclusion of acute venous thrombosis and pulmonary embolism: a systematic review.
Kearon C, Julian JA, Newman TE, Ginsberg JS. Noninvasive diagnosis of deep venous thrombosis. McMaster Diagnostic Imaging Practice Guidelines Initiative.
Zamora J, Muriel A, Abrair V.
Wells PS, Hirsh J, Anderson DR, Lensing AWA, Foster G, Kearon C, et al. Accuracy of clinical assessment of deep-vein thrombosis.
Aschwanden M, Labs K, Jeanneret C, Gehrig A, Jaeger K. Can a D-dimer assay, alone or combined with structured clinical risk assessment, rule out deep venous thrombosis in symptomatic patients?
Caliezi C, Funfsinn N, Mauron T, Baumgartner I, Sulzer I, Ulrich M, et al. Performance of a new fibrin monomer assay to exclude deep vein thrombosis in symptomatic outpatients.
Mayer W, Hirschwehr R, Hippmann G, Odpadlik H, Bayer P, Partsch H. Whole-blood immunoassay (SimpliRED) versus plasma immunoassay (NycoCard) for the diagnosis of clinically suspected deep vein thrombosis.
Shields P, Turnipseed S, Panacek E, White R. Validation of the Canadian clinical probability rule and use of bedside D-dimer testing in the diagnosis of acute venous thrombosis in an emergency room setting.
D-dimer testing and impedance plethysmography were effective for the exclusion of deep venous thrombosis.
SimpliRED D-dimer assay did not rule out venous thromboembolism in the emergency department.
Kraaijenhagen RA, Piovella F, Bernardi E, Verlato F, Beckers EA, Koopman MM, et al. Simplification of the diagnostic management of suspected deep vein thrombosis.
Lukaschek J, Schmid HR, Vorburger C, Beer JH. Critical appraisal of the diagnostics of 270 consecutive cases of suspected venous thromboembolism and established consequences at a non-university center.
Strandberg K, Astermark J, Bjorgell O, Becker C, Nilsson PE, Stenflo J. Complexes between activated protein C and protein C inhibitor measured with a new method: comparison of performance with other markers of hypercoagulability in the diagnosis of deep vein thrombosis.
Owings JT, Gosselin RC, Anderson JT, Battistella FD, Bagley M, Larkin EC. Practical utility of the D-dimer assay for excluding thromboembolism in severely injured trauma patients.
Lowe GD, Haverkate F, Thompson SG, Turner RM, Bertina RM, Turpie AG, et al. Prediction of deep vein thrombosis after elective hip replacement surgery by preoperative clinical and haemostatic variables: the ECAT DVT Study. European Concerted Action on Thrombosis.
Dunn ID, Hui AC, Triffitt PD, Crozier AE, Gregg PJ, Sinclair ME, et al. Plasma D-dimer as a marker for postoperative deep venous thrombosis. A study after total hip or knee arthroplasty.
Wattrisse G, Lecoutre D, Dufossez F, Ramon C, Groux-Pante C. The detection of thromboembolism in orthopedic traumatology: the value and limitations of D-dimer determination by the latex method.
Borris LC, Sorensen JV, Lassen MR, Walenga JM, Fareed J, Jorgensen LN, et al. Components of coagulation and fibrinolysis during thrombosis prophylaxis with a low molecular weight heparin (Enoxaparin) versus Dextran 70 in hip arthroplasty.
Jorgensen LN, Lind B, Hauch O, Leffers A, Albrecht-Beste E, Konradsen LA. Thrombin-antithrombin III-complex & fibrin degradation products in plasma: surgery and postoperative deep venous thrombosis.
Siragusa S. Plasma D-dimer test accuracy can be affected by heparin administration.
Kraaijenhagen RA, Koopman MMW, Piovella F, Siragusa S, Rossi L, Verlato F, et al. Simplification of the diagnostic management of outpatients with symptomatic deep vein thrombosis with D-dimer measurements.
Ville D, Clermont N, Trzeciak MC, Dechavanne M. Contribution of D-dimer levels in the diagnosis of post operative deep vein thrombosis (DVT) after total hip replacement (THR).
Loreth RM, Berger HH, Albert FW. The Diagnostic-Value Of D-Dimer (DD) And Thrombin Antithrombin- Iii Complex (TAT) In Suspected Deep-Vein Thrombosis.
Dale S, Gogstand GO, Brosstad F, Godal HC, Holtlund J, Mork E, et al. The Value Of C-Reactive Protein (Crp) And D-Dimer In The Diagnosis Of Deep Venous Thrombosis (DVT).
Holtzmann M. The D-Dimer Test ‘SimpliREDR’ in ambulatory therapy of deep leg/iliac venous thrombosis.
Wahl WL, Brandt MM, Ahrns K, Corpron CA, Franklin GA. The utility of D-dimer levels in screening for thromboembolic complications in burn patients.
Shiota N, Sato T, Nishida K, Matsuo M, Takahara Y, Mitani S, et al. Changes in LPIA D-dimer levels after total hip or knee arthroplasty relevant to deep-vein thrombosis diagnosed by bilateral ascending venography.
Roussi J, Bentolila S, Boudaoud L, Casadevall N, Vallee C, Carlier R, et al. Contribution of D-Dimer determination in the exclusion of deep venous thrombosis in spinal cord injury patients.
Abraham P, Ternisien C, Hubert L, Pidhorz L, Saumet JL. Does venous microemboli detection add to the interpretation of D-dimer values following orthopedic surgery?
Bounameaux H, Miron MJ, Blanchard J, de Moerloose P, Hoffmeyer P, Leyvraz PF. Measurement of plasma D-dimer is not useful in the prediction or diagnosis of postoperative deep vein thrombosis in patients undergoing total knee arthroplasty.
Levy G, Levy PY, Hessmann J, Monin P. [Diagnosis of post-operative venous thrombosis using determination of plasma D-dimer]. [French].
Cofrancesco E, Cortellaro M, Corradi A, Ravasi F, Bertocchi F. Clinical utility of prothrombin fragment 1+2, thrombin antithrombin III complexes and D-dimer measurements in the diagnosis of deep vein thrombosis following total hip replacement.
Douketis JD, McGinnis J, Ginsberg JS. The clinical utility of a rapid bedside D-dimer assay for screening of deep vein thrombosis following orthopaedic surgery.
Cofrancesco E, Cortellaro M, Corradi A, Ravasi F, Bertocchi F. Coagulation activation markers in the prediction of venous thrombosis after elective hip surgery.
Harvey RL, Roth EJ, Yarnold PR, Durham JR, Green D. Deep vein thrombosis in stroke. The use of plasma D-dimer level as a screening test in the rehabilitation setting.
Crippa L, Ravasi F, D'Angelo SV, Varagona R, Milani E, Safa O, et al. Diagnostic value of compression ultrasonography and fibrinogenrelated parameters for the detection of postoperative deep vein thrombosis following elective hip replacement: a pilot study.
Bongard O, Wicky J, Peter R, Simonovska S, Vogel JJ, de Moerloose P, et al. D-dimer plasma measurement in patients undergoing major hip surgery: use in the prediction and diagnosis of postoperative proximal vein thrombosis.
Bounameaux H, Khabiri E, Huber O, Schneider PA, Didier D, de Moerloose P, et al. Value of liquid crystal contact thermography and plasma level of D-dimer for screening of deep venous thrombosis following general abdominal surgery.
Walsh K, Kelaher N, Long K, Cervi P. An algorithm for the investigation and management of patients with suspected deep venous thrombosis at a district general hospital.
Bucek RA, Reiter M, Quehenberger P, Weltermann A, Kyrle PA, Minar E. Thrombus precursor protein, endogenous thrombin potential, vonWillebrand factor and activated factor VII in suspected deep vein thrombosis: is there a place for new parameters?
Bucek RA, Reiter M, Quehenberger P, Minar E. C-reactive protein in the diagnosis of deep vein thrombosis.
Arrive L, Monnier-Cholley L, Serru V, Carrat F, Vassal T, Dahan H, et al. D-Dimer determination combined with clinical probability for the diagnosis of leg venous thrombosis.
Johanning JM, Franklin DP, Thomas DD, Elmore JR. D-dimer and calf circumference in the evaluation of outpatient deep venous thrombosis.
Larsen TB, Stoffersen E, Christensen CS, Laursen B. Validity of D-dimer tests in the diagnosis of deep vein thrombosis: a prospective comparative study of three quantitative assays.
Shitrit D, Levi H, Huerta M, Rudensky B, Bargil-Shitrit A, Gutterer N. Appropriate indications for venous duplex scanning based on Ddimer assay.
ten Wolde M, Kraaijenhagen RA, Prins MH, Buller HR. The clinical usefulness of D-dimer testing in cancer patients with suspected deep venous thrombosis.
Shields GP, Turnipseed S, Panacek EA, Melnikoff N, Gosselin R, White RH. Validation of the Canadian clinical probability model for acute venous thrombosis.
Schutgens RE, Esseboom EU, Haas FJ, Nieuwenhuis HK, Biesma DH. Usefulness of a semiquantitative D-dimer test for the exclusion of deep venous thrombosis in outpatients.
Cornuz J, Ghali WA, Hayoz D, Stoianov R, Depairon M, Yersin B. Clinical prediction of deep venous thrombosis using two risk assessment methods in combination with rapid quantitative D-dimer testing.
Bucek RA, Koca N, Reiter M, Haumer M, Zontsich T, Minar E. Algorithms for the diagnosis of deep-vein thrombosis in patients with low clinical pretest probability.
Dryjski M, O'Brien-Irr MS, Harris LM, Assett J, Anicke D. Evaluation of a screening protocol to exclude the diagnosis of deep venous thrombosis among emergency department patients.
Siragusa S, Terulla V, Pirrelli S, Porta C, Falaschi F, Anastasio R, et al. A rapid D-dimer assay in patients presenting at the emergency room with suspected acute venous thrombosis: accuracy and relation to clinical variables.
Bucek RA, Quehenberger P, Feliks I, Handler S, Reiter M, Minar E. Results of a new rapid d-dimer assay (cardiac d-dimer) in the diagnosis of deep vein thrombosis.
Funfsinn N, Caliezi C, Biasiutti FD, Korte W, Z'Brun A, Baumgartner I, et al. Rapid D-dimer testing and pre-test clinical probability in the exclusion of deep venous thrombosis in symptomatic outpatients.
Shitrit D, Heyd J, Raveh D, Rudensky B. Diagnostic value of the D-dimer test in deep vein thrombosis: improved results by a new assay method and by using discriminate levels.
Harper P, Marson C, Grimmer A, Monahan K, Humm G, Baker B. The rapid whole blood agglutination d-dimer assay has poor sensitivity for use as an exclusion test in suspected deep vein thrombosis.
Bates SM, Grand'Maison A, Johnston M, Naguit I, Kovacs MJ, Ginsberg JS. A latex D-dimer reliably excludes venous thromboembolism.
Gosselin RC, Owings JT, Utter GH, Jacoby RC, Larkin EC. A new method for measuring D-dimer using immunoturbidometry: a study of 255 patients with suspected pulmonary embolism and deep vein thrombosis.
Hein-Rasmussen R, Tuxen CD, Wiinberg N. Diagnostic value of the Nycocard, Nycomed D-dimer assay for the diagnosis of deep venous thrombosis and pulmonary embolism: a retrospective study.
Anderson DR, Wells PS, Stiell I, MacLeod B, Simms M, Gray L, et al. Management of patients with suspected deep vein thrombosis in the emergency department: combining use of a clinical diagnosis model with D-dimer testing.
Permpikul C, Chantavatcharakorn K, Bouranasompop C, Chaiyasoot W. Whole blood agglutination D-dimer test for the diagnosis of deep vein thrombosis.
Bradley M, Bladon J, Barker H. D-dimer assay for deep vein thrombosis: its role with colour Doppler sonography.
LaCapra S, Arkel YS, Ku DH, Gibson D, Lake C, Lam X. The use of thrombus precursor protein, D-dimer, prothrombin fragment 1.2, and thrombin antithrombin in the exclusion of proximal deep vein thrombosis and pulmonary embolism.
Villa P, Ferrando F, Serra J, Faus H, Mira Y, Vaya A, et al. Quantification of D-dimer using a new fully automated assay: its application for the diagnosis of deep vein thrombosis.
Van Der Graaf F, van den Borne H, van der Kolk M, de Wild PJ, Janssen GW, van Uum SH. Exclusion of deep venous thrombosis with D-dimer testing-comparison of 13 D-dimer methods in 99 outpatients suspected of deep venous thrombosis using venography as reference standard.
Trujillo-Santos AJ, Garcia de Lucas MD, Rios-Tamayo R, Jimenez-Puente A, Garcia-Sanchez JE. [Clinical and analytic diagnostic evaluation of deep venous thrombosis of the lower limbs]. [Spanish].
Sadouk M, Desmarais S, Patenaude JV, Lepage R. Comparison of diagnostic performance of three new fast D-dimer assays in the exclusion of deep vein thrombosis.
Farrell S, Hayes T, Shaw M. A negative SimpliRED D-dimer assay result does not exclude the diagnosis of deep vein thrombosis or pulmonary embolus in emergency department patients.
Scarano L, Prandoni P, Gavasso S, Gomiero W, Carraro G, Girolami A. Failure of soluble fibrin polymers in the diagnosis of clinically suspected deep venous thrombosis.
Carter CJ, Serrano K, Breen DJ, Buckley AR, Devine DV. Rapid fibrin D-dimer tests for deep venous thrombosis: factors affecting diagnostic utility.
Wahlander K, Tengborn L, Hellstrom M, Olmarker AH, Peterson LE, Stigendal L, et al. Comparison of various D-dimer tests for the diagnosis of deep venous thrombosis.
Wells PS, Anderson DR, Bormanis J, Guy F, Mitchell M, Gray L, et al. Application of a diagnostic clinical model for the management of hospitalized patients with suspected deep-vein thrombosis.
Legnani C, Pancani C, Palareti G, Guazzaloca G, Coccheri S. Contribution of a new, rapid, quantitative and automated method for D-dimer measurement to exclude deep vein thrombosis in symptomatic outpatients.
Le Blanche AF, Siguret V, Settegrana C, Bohus S, Le Masne de Chermont E, Andreux JP, et al. Ruling out acute deep vein thrombosis by ELISA plasma D-dimer assay versus ultrasound in inpatients more than 70 years old.
Aschwanden M, Labs KH, Jeanneret C, Gehrig A, Jaeger KA. The value of rapid D-dimer testing combined with structured clinical evaluation for the diagnosis of deep vein thrombosis.
Lennox AF, Delis KT, Serunkuma S, Zarka ZA, Daskalopoulou SE, Nicolaides AN. Combination of a clinical risk assessment score and rapid whole blood D-dimer testing in the diagnosis of deep vein thrombosis in symptomatic patients.
Wijns W, Daoud N, Droeshout I, Pradier O, Wautrecht JC, Golzarian J, et al. Evaluation of two D-Dimer assays in the diagnosis of venous thromboembolism.
Lindahl TL, Lundahl TH, Ranby M, Fransson SG. Clinical evaluation of a diagnostic strategy for deep venous thrombosis with exclusion by low plasma levels of fibrin degradation product D-dimer.
Mauron T, Baumgartner I, Z'Brun A, Demarmels Biasiutti F, Redondo M, Do DD, et al. SimpliRED D-dimer assay: comparability of capillary and citrated venous whole blood, between-assay variability, and performance of the test for exclusion of deep vein thrombosis in symptomatic outpatients.
Escoffre-Barbe M, Oger E, Leroyer C, Grimaux M, Le Moigne E, Nonent M, et al. Evaluation of a new rapid D-dimer assay for clinically suspected deep venous thrombosis (Liatest D-dimer).
Wells PS, Anderson DR, Bormanis J, Guy F, Mitchell M, Lewandowski B. SimpliRED D-dimer can reduce the diagnostic tests in suspected deep vein thrombosis.
Khaira HS, Mann J. Plasma D-dimer measurement in patients with suspected DVT–a means of avoiding unnecessary venography.
Wildberger JE, Vorwerk D, Kilbinger M, Piroth W, Hunter DW, Wienert V, et al. Bedside testing (SimpliRED) in the diagnosis of deep vein thrombosis. Evaluation of 250 patients.
Knecht MF, Heinrich F. Clinical evaluation of an immunoturbidimetric D-dimer assay in the diagnostic procedure of deep vein thrombosis and pulmonary embolism.
Guazzaloca G, Palareti G, Legnani C, Fortunato G, Grauso F, Rodorigo G, et al. [Deep venous thrombosis. Evaluation of a non-invasive diagnostic procedure based on compression ultrasonography and measurement of D-dimer plasma levels]. [Italian].
Kozman H, Flemmer MC, Rahnama M. Deep venous thrombosis: prediction by D-dimer?
Legnani C, Pancani C, Palareti G, Guazzaloca G, Fortunato G, Grauso F, et al. Comparison of new rapid methods for D-dimer measurement to exclude deep vein thrombosis in symptomatic outpatients.
Jacq F, Heron E, Rance A, Cesarini ML, Emmerich J, Fiessinger JN. [Evaluation of a test for rapid detection of D-dimers for the exclusion of the diagnosis of venous thrombosis]. [French].
Janssen MC, Verbruggen BW, ter Hark PJ, Novakova IR. Factor VIIA determination compared to D-Dimer in diagnosis of deep venous thrombosis.
Fiessinger JN, Heron E, Jacq F, Rance A, Emmerich J. Rapid blood test for the exclusion of venous thromboembolism in symptomatic outpatients.
Scarano L, Bernardi E, Prandoni P, Sardella C, Rossi L, Carraro P, et al. Accuracy of two newly described D-dimer tests in patients with suspected deep venous thrombosis.
Leroyer C, Escoffre M, Le Moigne E, Grimaux M, Cagnioncle O, Oger E, et al. Diagnostic value of a new sensitive membrane based technique for instantaneous D-dimer evaluation in patients with clinically suspected deep venous thrombosis.
Killick SB, Pentek PG, Mercieca JE, Clarke MF, Bevan DH. Comparison of immunofiltration assay of plasma D-dimer with diagnostic imaging in deep vein thrombosis.
Borg JY, Levesque H, Cailleux N, Franc C, Hellot MF, Courtois H. Rapid quantitative D-dimer assay and clinical evaluation for the diagnosis of clinically suspected deep vein thrombosis.
Janssen MC, Heebels AE, de Metz M, Verbruggen H, Wollersheim H, Janssen S, et al. Reliability of five rapid D-dimer assays compared to ELISA in the exclusion of deep venous thrombosis.
Elias A, Aptel I, Huc B, Chale JJ, Nguyen F, Cambus JP, et al. D-dimer test and diagnosis of deep vein thrombosis: a comparative study of 7 assays.
D'Angelo A, D'Alessandro G, Tomassini L, Pittet JL, Dupuy G, Crippa L. Evaluation of a new rapid quantitative D-dimer assay in patients with clinically suspected deep vein thrombosis.
Gavaud C, Ninet J, Ville D, Coppere B, Hanss M, Bureau Du Colombier P, et al. [Diagnosis of venous thrombosis and/or pulmonary embolism by determination of d-dimers using ELISA. Review based on a study of 80 consecutive patients hospitalized in an emergency unit]. [French].
Brenner B, Pery M, Lanir N, Jabareen A, Markel A, Kaftori JK, et al. Application of a bedside whole blood D-dimer assay in the diagnosis of deep vein thrombosis.
Bouman CS, Ypma ST, Sybesma JP. Comparison of the efficacy of D-dimer, fibrin degradation products and prothrombin fragment 1+2 in clinically suspected deep venous thrombosis.
Wells PS, Brill-Edwards P, Stevens P, Panju A, Patel A, Douketis J, et al. A novel and rapid whole-blood assay for D-dimer in patients with clinically suspected deep vein thrombosis.
Tengborn L, Palmblad S, Wojciechowski J, Peterson LE, Stigendal L. D-dimer and thrombin/antithrombin III complex–diagnostic tools in deep venous thrombosis?
Dale S, Gogstad GO, Brosstad F, Godal HC, Holtlund J, Mork E, et al. Comparison of three D-dimer assays for the diagnosis of DVT: ELISA, latex and an immunofiltration assay (NycoCard D-Dimer).
Hansson PO, Eriksson H, Eriksson E, Jagenburg R, Lukes P, Risberg B. Can laboratory testing improve screening strategies for deep vein thrombosis at an emergency unit?
Carter CJ, Doyle DL, Dawson N, Fowler S, Devine DV. Investigations into the clinical utility of latex D-dimer in the diagnosis of deep venous thrombosis.
Ibrahim KM, O'Neill AI, Parkin DJ. D-dimer (latex) assay in the diagnosis of deep venous thrombosis.
Heijboer H, Ginsberg JS, Buller HR, Lensing AW, Colly LP, Wouter ten Cate J. The use of the D-dimer test in combination with non-invasive testing versus serial non-invasive testing alone for the diagnosis of deep-vein thrombosis.
Kroneman H, Van Bergen PF, Knot EA, Jonker JJ, de Maat MP. Diagnostic value of D-dimer for deep venous thrombosis in outpatients.
Grau E, Linares M, Estany A, Martin F. Utility of D dimer in the diagnosis of deep venous thrombosis in outpatients.
Chang-Liem GS, Lustermans FA, van Wersch JW. Comparison of the appropriateness of the latex and Elisa plasma D-dimer determination for the diagnosis of deep venous thrombosis.
de Boer WA, de Haan MA, Huisman JW, Klaassen CH. D-Dimer latex assay as screening method in suspected deep venous thrombosis of the leg. A clinical study and review of the literature.
Boneu B, Bes G, Pelzer H, Sie P, Boccalon H. D-Dimers, thrombin antithrombin III complexes and prothrombin fragments 1+2: diagnostic value in clinically suspected deep vein thrombosis.
Mossaz A, Gandrille S, Vitoux JF, Abdoucheli-Baudot N, Aiach M, Fiessinger JN. The value of D-dimer assays in the emergency diagnosis of venous thrombosis. [French].
Chapman CS, Akhtar N, Campbell S, Miles K, O'Connor J, Mitchell VE. The use of D-Dimer assay by enzyme immunoassay and latex agglutination techniques in the diagnosis of deep vein thrombosis.
Bounameaux H, Schneider PA, Reber G, de Moerloose P, Krahenbuhl B. Measurement of plasma D-dimer for diagnosis of deep venous thrombosis.
Ott P, Astrup L, Jensen RH, Nyeland B, Pedersen B. Assessment of D-dimer in plasma: diagnostic value in suspected deep venous thrombosis of the leg.
Heaton DC, Billings JD, Hickton CM. Assessment of D dimer assays for the diagnosis of deep vein thrombosis.
Rowbotham BJ, Carroll P, Whitaker AN, Bunce IH, Cobcroft RG, Elms MJ, et al. Measurement of crosslinked fibrin derivatives–use in the diagnosis of venous thrombosis.
Bozic M, Blinc A, Stegnar M. D-dimer, other markers of haemostasis activation and soluble adhesion molecules in patients with different clinical probabilities of deep vein thrombosis.
Gosselin RC, Owings JT, Jacoby RC, Larkin EC. Evaluation of a new automated quantitative D-dimer, Advanced D-Dimer, in patients suspected of venous thromboembolism.
Lindahl TL, Lundahl TH, Fransson SG. Evaluation of an automated micro-latex D-dimer assay (Tina- quant (R) on Hitachi 911 analyser) in symptomatic outpatients with suspected DVT.
Legnani C, Pancani C, Palareti G, Guazzaloca G, Coccheri S. Performance of a new, fast D-dimer test (IL Test (TM) D-Dimer) for the management of outpatients with suspected deep vein thrombosis in emergency situations.
Lee AYY, Julian JA, Kearon C, Wells PS, Levine M, Ginsberg JS. Diagnostic accuracy of SimpliFED (R) D-dimer testing in cancer patients with clinically suspected deep vein thrombosis.
Crippa L, Tomassini L, Rizzi B, Pattarini E, D'Alessandro G, Vigano S. Clinical pre test probability and D-dimer in the diagnosis of deep vein thrombosis.
Legnani C, Palareti G, Guazzaloca G, Poggi M, Coccheri S, Golfieri R, et al. Validation of a new quantitative automated latex photometric immunoassay of plasma D-dimer (LPIA-100) for the diagnosis of symptomatic DVT.
Pini M, Quintavalla R, Pattacini C, Manotti C, Polti T, Larini P, et al. Combined use of the strain-guage plethysmography and latex D-Dimer test in clinically suspected deep venous thrombosis.
Brenner B, Pery M, Lanir N, Jabbarin A, Kaftori JK, Gaitini D, et al. Application of SimpliRed D-Dimer in the diagnosis of deep-vein thrombosis.
Elias A, Aillaud MF, Roul C, Monteil O, Villain P, Serradimigni A, et al. Assessment of D-Dimer measurement by ELISA or latex methods in deep-vein thrombosis diagnosed by ultrasonic duplex scanning.
Kilroy DA, Ireland S, Reid P, Goodacre S, Morris F. Emergency department investigation of deep vein thrombosis.
Del Rio SM, Fajardo JAG, Pedrosa MM, Gutierrez V, Carrera S, Puerta CV. Clinical evaluation of D-Dimer in the diagnosis of thromboembolic disease.
Leroyer C, Bressollette L, Oger E, Le Ber C, Nonent M, Boulch MT, et al. Diagnostic value of plasma D-Dimer measurement, using ELISA test, in front of a suspected deep venous thrombosis.
Mayer W, Hirschwehr R, Partsch H. The D-dimer bedside test ‘SimpliREDR’ as diagnostic aid in suspected thrombosis.
Miranda C, Harden M. Comparison of D-dimer findings with venous duplex examinations: A work in progress.
Gosselin RC, Owings JT, Kehoe J, Anderson JT, Dwyre DM, Jacoby RC, et al. Comparison of six D-dimer methods in patients suspected of deep vein thrombosis.
Cini M, Legnani C, Cavallaroni K, Bettini F, Palareti G. A new rapid bedside assay for D-dimer measurement (Simplify D-dimer) in the diagnostic work-up for deep vein thrombosis.
Lijmer JG, Mol JJ, Heisterkamp S, Bonsel GJ, van der Meulen JHP, Bossuyt PMM. Empirical evidence of design-related bias in studies of diagnostic tests.
Giannoukas AD, Labropoulos N, Burke P, Katsamouris A, Nicolaides AN. Calf deep venous thrombosis: A review of the literature.
Lohr JM, James KV, Deshmukh RM, Hasselfeld KA. Calf vein thrombi are not a benign finding.
Author notes
From the 1Medical Care Research Unit, University of Sheffield, Sheffield, 2Centre for Biostatistics and Genetic Epidemiology, Department of Health Sciences, University of Leicester, Leicester, and 3Department of Emergency Care, Sheffield Teaching Hospitals, Sheffield, UK