Article

The VIDAS D-dimer test for venous thromboembolism: a prospective surveillance study shows maintenance of sensitivity and specificity when used in normal clinical practice

Diagnostics

The VIDAS D-dimer test for venous thromboembolism: a prospective surveillance study shows maintenance of sensitivity and specificity when used in normal clinical practice

David Mountain FACEM*, Ian Jacobs PhD, Andrew Haig FACEM

Department of Emergency Medicine, Sir Charles Gairdner Hospital, Perth, Australia Department of Academic Emergency Medicine, University of Western Australia, Perth, Australia Department of Emergency Medicine, Royal Perth Hospital, Perth, Australia

Received 26 June 2006; received in revised form 18 September 2006; accepted 25 September 2006

Abstract

Background: As a result of a number of clinical management studies, D-dimer (DD) tests such as VIDAS (BioMe’rieux Australia P/L-Sydney, NSW) have been recommended to reduce venous thromboembolism investigations. Surveillance studies for new tests are recommended. We prospectively assessed VIDAS DD in normal practice.

Methods: Consecutive emergency patients and inpatients (IPs) with DD or VTE investigations were prospectively identified. Investigation results and early chart review including predefined factors reducing specificity were documented. A latex DD was also performed. Patients were followed for at least 3 months for recurrent VTE.

Results: Four hundred three patients (emergency, 64%; VTE-positive, 12%; 95% followed up) were analyzed. VIDAS sensitivity was 96% (95% confidence interval 86%-99%), specificity 38% (confidence interval, 34%-44%; negative likelihood ratio, 0.11), and emergency specificity 51%. Latex sensitivity was 76%. Cancer, trauma, recent operations, IP status, and advanced age were associated with markedly reduced specificity. Specificity in older emergency patients (N70 years old) and younger IPs (b70) without comorbidities was 20% to 30%, but sensitivity was maintained at 100%.

Conclusions: VIDAS DD probably maintains adequate sensitivity in normal clinical practice for low- or even intermediate-risk patients. Latex agglutination had poor sensitivity. Specificity is best in younger low- morbidity emergency patients. These findings need validation in larger multicenter surveillance studies. D 2007

Data in this study have been presented as: Mountain D, Haig A. Presentation. A prospective observational study in an Unselected population of the VIDAS D-dimer (VIDAS DD) test for excluding acute venous thrombo-embolic (VTE) disease shows retention of sensitivity and specificity in a real environment. ACEM/ASEM joint scientific meeting, November 2000, Canberra, Australia.

* Corresponding author. Department of Emergency Medicine, Sir Charles Gairdner Hospital, Hospital Ave, Nedlands, Perth WA, Australia 6009. Tel.: +61 8 93463333; fax: +61 8 93463516.

E-mail address: [email protected] (D. Mountain).

0735-6757/$ – see front matter D 2007 doi:10.1016/j.ajem.2006.09.004

Introduction

Providing a diagnosis of pulmonary embolism (PE) and Deep venous thrombosis clinically is difficult [1,2]. Definitive diagnosis normally relies on radiologic inves- tigations. At the time of this study, recommended testing often required pulmonary angiography (PA) or serial ultra- sounds (USs). Clinical audit confirmed that compliance to recommended diagnostic regimens was poor at our institu- tion. Studies have confirmed low rates of appropriate testing for PE in normal clinical practice [3,4]. Many clinicians are averse to PA because of its perceived morbidity, mortality, and limited availability. In current practice, multislice computed tomographic pulmonary angiography (CTPA) is often used as the default investigation, but radiation dosing, problems with readings (particularly by junior staff), dye issues, and suboptimal views in up to 15% of patients make universal use of this investigation problematic, particularly in low-prevalence emergency department (ED) populations [5].

Sensitive blood tests for excluding venous thromboem- bolism (VTE) would be useful. Most target markers of thrombus generation and breakdown. D-dimer (DD) tests

have been the most promising of these products. Standard enzyme-linked immunosorbent assays (ELISAs) have high sensitivity but are logistically unsuited to clinical practice. Newer rapid ELISA tests have been introduced and show similar results to standard ELISA testing in study con- ditions [5].

Few DD tests have been adequately tested against gold- standard diagnostic testing or in management studies. A study of 1000 outpatients used the VIDAS DD test (a modified nonbatched ELISA) to manage patients [6]. A negative test result (confirmed by negative US finding if DVT was clinically suspected) meant the patient had no further investigation or therapy for VTE. Using 3-month follow-up as part of the gold standard, VIDAS DD was 99% sensitive. Based on this (and other studies demonstrating good sensitivity against gold-standard tests), our institution changed from using a latex agglutination (LA) DD (Agen) to VIDAS [5,7].

A prospective audit was performed to demonstrate adequate clinical test performance and allay clinician concerns. Surveillance and clinical management studies (prospective audit) have been suggested for monitoring the performance and utility of Diagnostic tests.in normal clinical

Fig. 1 Eligible patients–outcomes using STARD criteria [14].

Table 1 Excluded patients–reasons for exclusion and

numbers of samples excluded from analysis

Reasons for exclusions

No samples

Repeated samples–similar results Tests (DD/VQ) done for non-VTE

diagnosis

Patient not managed at primary institution

Patients not available to follow-up Total excluded

No. of samples/patient

episodes excluded

16

30

11

15

16

88

use, similar to type IV surveillance studies for therapeutic drugs [8,9].

Aims

The primary hypothesis for this study was that the VIDAS DD would maintain sensitivity for proven PE/DVT of more than 95% with adequate specificity when compared with a gold standard. Secondary aims were to examine factors affecting specificity and to compare performance with the Agen LA test.

Methods

Background

The setting was a 600-bed adult teaching hospital serving a population of approximately 400000 people. Emergency department attendances were 39 000 yearly, with 45% admitted. Approximately 400 ventilation perfusion lung scans (VQ) were performed, with 150 VTEs diagnosed annually. At the time, a single-slice helical CTPA was only being used occasionally and was of marginal significance to PE diagnosis at this center before the study ended. Investigations ordered were purely at the discretion of treating clinicians. The VIDAS DD test was introduced in July 1999. Any clinician could order DD, and a negative result was considered to exclude significant VTE. Staff members were aware of the prospective audit but did not collect data.

Population

All patients who had either DD or a VQ scan ordered

from July 1999 to December 1999 were included. This included those patients with DD but only investigated for DVT. Coagulation laboratory and nuclear medicine depart- ments provided lists of patients on a daily basis (except

Table 2 Type of VTE suspected and prevalence of disease Type of VTE suspected No. of VTE-positive/No.

investigated (%)

PE 30/353 (8%)

weekends). Patients were excluded for the following

DVT

12/33 (36%)

reasons: DD was done for a diagnosis other than VTE, for

PE + DVT

7/17 (41%)

example, disseminated intravascular coagulation; patients

were not managed at this institution; follow-up was impossible; or no suitable blood samples were available.

Investigations/follow-up

All patients had a VIDAS DD test performed either clinically or on stored samples taken (within 48 hours of the index symptoms) for other tests. VIDAS is normally performed on citrated samples, giving quantitative results. EDTA samples were used if citrate was unavailable. Samples were stored as per standard laboratory protocols. Latex agglutination (Agen) DD was performed for comparison on almost all samples (6 patients were missed because of lost samples), but the results were unavailable to clinicians. All patients were followed up until they reached defined end points. Positive VTE was confirmed by either a high- probability VQ scan, positive CTPA, PA, or echocardiogram; deep venous thrombosis was confirmed by positive US, venogram, computed tomography, or any VTE at postmortem (PM). Pulmonary embolism was excluded by a negative PA result PM or a normal VQ scan. Deep venous thrombosis was excluded by negative venogram or US finding (for low- or intermediate-risk DVT patients). Patients with a definitive alternative diagnosis clearly explaining all symptoms (eg, Chest x-ray showing pneumothorax, US showing ruptured Baker cyst) were considered negative for VTE. Any negative CTPA required clinical follow-up because CTPA was not considered to be adequately validated at the time. Patients with an indeterminate workup but clinically treated as VTE were included as positive for VTE even if DD was negative to avoid overestimating sensitivity.

All other patients were followed up for a minimum of 3 months by a combination of chart review (performed on all patients) and contacting general practitioners (GPs) or the patients/relatives for telephone interview. Patients who died had their notes reviewed (by DM) and were classified according to whether PE was likely or unlikely to have contributed to death. Comorbidities expected to affect specificity, such as recent surgery or major trauma, active cancer, or severe concurrent disease (eg, sepsis or recent intensive care unit [ICU] admission) were documented for all patients [10-12]. The ED population was compared to inpatients (IPs) for demographics, outcomes, test perfor- mance, and factors affecting specificity.

Data retrieval

All definitions (including end points) were confirmed between the investigators before data retrieval. Data were

Total VTE- VTE- ED

IP

VIDAS-

VIDAS-

Latex-

Latex-

positive

negative

positive

negative

positive

negative

n (% of total population)

403 (100)

49 (12)

354 (88)

257 (64)

143 (36)

265 (66)

138 (34)

151 (38)

246 (62)

Male (% of subgroup)

190 (47)

25 (51)

166 (47)

119 (46)

69 (48)

135 (51)

55 (40)

81 (54)

105 (43)

Female (%)

213 (53)

24 (49 )

188 (53)

138 (54)

74 (52)

130 (49)

83 (60)

70 (46)

141 (57)

Age, mean (median)

62 (64)

61 (63)

62 (64)

60 (61)

66 (68)

67 (69)

52 (55)

66 (69)

60 (62)

retrieved by chart review directly into a labeled Excel spreadsheet by 2 authors (DM, AH). Data included demo- graphics; tests ordered and their results; if symptoms of VTE were present in ED; risk factors for VTE and coexistent active conditions or previous VTE; presenting features; whether pretest probability was documented; and results of CXR, arterial blood gas, and pulse oximetry. The authors performed follow-up without blinding to clinical details.

Table 3 Population characteristics–total and for selected subgroups

Statistics

Data were analyzed using SPSS (version 11.5; SPSS, Chicago, Ill). Frequency counts and proportions were calculated for all categorical data and primary end points. Because of the skewed nature of the distribution, medians were calculated for age. Sensitivity, specificity, and positive and negative Likelihood ratios were all calculated with 95% confidence intervals (CIs) derived for these values.

Ethics

The study was presented to the chair of the ethics committee. It was exempted from ethics review because it was performed as a prospective clinical audit and had no interventions apart from patient contact for follow-up.

Results

There were 491 consecutive episodes (see Fig. 1 for Standards for Reporting of Diagnostic Accuracy [STARD] outcome details) reviewed, with 88 excluded for the reasons shown in Table 1 [13]. Of the 16 lost to follow-up, 8 were subsequently found at a later review (N2 years) to have been free of VTE. However, these additional negative follow-ups were not included in the statistical analysis.

The population studied had a VTE prevalence of 12%. Prevalence was higher in the IP group (17% vs 9%) and in patients investigated for DVT (Tables 2-4). Patients with VTE were not significantly older than the whole population, and there was no sex bias (Table 3). Characteristics for total population and population subgroups are outlined in Table 3. Among the subjects, 47% were male, and mean age was

62 years. Inpatients were significantly older than ED patients (66 vs 60 years; P b .001; 95% CI for difference in means, –10.42 to –3.25 years). It was also noted that patients with a positive VIDAS DD were much older than those with a negative result (67 vs 52 years). The VIDAS test had a sensitivity of 96% and a specificity of 38%, giving a negative LR of 0.11 (Table 5). Agen LA test showed a sensitivity of 79% and a specificity of 66% in the 397 (of 403) patients who had latex performed as well (2 x 2 data withheld for brevity–available on request from corresponding author). When death or VTE was used as the end point of interest, sensitivity of VIDAS was maintained at 97% (95% CI, 95%-99%). This was true even if patients with other causes for death or minimal risk of VTE death were excluded (see Table 6).

The overall specificity of the VIDAS test was 38% (Table 5), 51% in ED and 15% for IP (Table 4). Prevalence of factors suspected to affect specificity and the likelihood of a positive VIDAS test are detailed in Table 7. All of these factors, including age of older than 70 years, cancer, recent operations or trauma, or recent ICU admission, were associated with poor specificity of the VIDAS test. In patients older than 70 years, 16/84 (19%) of ED patients had a negative VIDAS compared with 1/65 (2%) of IPs. In older ED patients, 16/74 (22%) had negative VIDAS if other factors associated with raised DD were removed (eg, 10 were postoperative or had cancer). In IPs younger than

Table 4 Emergency vs IP VIDAS test performance

ED (prevalence–9%)

Condition present (24)

Absent (235)

IP (prevalence–17%)

Condition present (25)

Absent (119)

Test-positive

22

116

25

102

Test-negative

2

119

0

17

Sensitivity (95% CI)

92% (74%-98%)

100% (84%-100%)

Specificity (95% CI)

51% (45%-57%)

15% (9%-22%)

Negative LR (95% CI)

0.16 (0.04-0.62)

not applicable

Positive LR (95% CI)

1.86 (1.56-2.23)

not applicable

VTE-positive (49)

VTE-negative (354)

Test-positive (265)

TP

47

FP

218

Test-negative (138)

FN

2

TN

136

Sensitivity (95% CI)

96% (86%-99%)

Specificity (95% CI)

38% (34%-44%)

Negative LR (95% CI)

0.11 (0.03-0.42)

Positive LR (95% CI)

1.56 (1.4-1.7)

70 years and without raised specificity associated comor- bidities, 16/54 (30%) had negative test results.

Table 5 VIDAS test performance–2 x 2 table

TP indicates true positives; TN, true negatives; FP, false positives; and

FN, false negatives.

Follow-up and VTE testing after DD

Of 354 patients without confirmed VTE, 47 died (Table 6). Of the 47 deaths, 45 had a positive VIDAS. Of the 2 patients with a negative DD, one patient was definitively without PE at PM (died of acute myocardial infarction secondary to gastrointestinal bleeding). The second patient had known pulmonary hypertension due to Atrial septal defect (ASD) shunt and died during admission. Pulmonary embolism could not be definitively excluded in this case (clinical risk low/VQ intermediate).

There were 76 patients who had a positive DD test but no further investigations for PE. All of these patients were followed up clinically without a VTE event. Of 138 patients with a negative DD, 28 patients had other investigations for PE. Ten had definitive testing excluding PE (8 normal VQ, 1 alternative diagnosis on CTPA, and 1 negative PA), 2 with indeterminate investigations were deemed positive for VTE (clinical details of cases follow), and the rest had indeter- minate testing with negative follow-up. Patients positive for PE had the following positive confirmatory tests: 19 high- probability VQ, 7 CTPA, 3 PA, 14 US, and 4 VTE picked on other imaging (echocardiogram, computed tomography of the thorax/abdomen).

Both cases assigned as false-negative diagnosis of VTE by VIDAS did not have a definitive diagnosis of PE. The first patient presented dyspneic with Chest tightness and had a history of previous PE but normal arterial blood gases (Po2, 95; Pco2, 35) and CXR. A DD was negative, and the patient was discharged from ED. She continued to have symptoms and was sent for a VQ scan by her GP. This showed a segmental perfusion-ventilation mismatch (intermediate by

prospective investigation of pulmonary embolism diagnosis [PIOPED] criteria) and was treated as PE. The treating respiratory consultant was not certain of the diagnosis of acute PE when contacted for follow-up. She probably had intermediate risk of PE before VQ and had a posttest probability (by PIOPED criteria) of 30% to 50% [1].

The second patient had had a previous PE/DVT and was a smoker on the pill with 7 days of leg pain and 1 day of pleuritic chest pain, and dyspnea with oxygen saturation of 99% on room air. She was assigned a high pretest probability of PE. D-dimer was negative and VQ intermediate. No other tests were performed, and she was anticoagulated. By PIOPED criteria, she had a 55% chance of PE [1].

Discussion

This study looks at some key issues for real-life implementation of a new DD test (eg, in unselected patients for whom any clinician, of any experience, can order the test). Firstly, does that test retain sensitivity when performed outside of trial situations? Secondly, for which populations does the test have adequate specificity (utility)? Finally, how does LA perform against ELISA (VIDAS) in these conditions?

We have found that the sensitivity was maintained at 96%, although with a lower CI of 86%. Negative LR (0.11) suggests a moderate-to-large reduction in pretest probability for a negative VIDAS. Confidence intervals are wide because of the low numbers of confirmed VTE. This is a feature of populations weighted to outpatients [14,15]. In addition, this was not a study protocol, and, therefore, ordering was allowed for any patient with any concern for VTE. In many studies, there is a selection bias to high- prevalence populations, probably related to strict inclusion criteria and the reluctance of physicians to enter patients with low risk of VTE into trials with invasive diagnostic tests [16].

The sensitivity was not affected if patients who died during follow-up (1/47 negative VIDAS without PM sensitivity of 97%; 95% CI, 95%-99%) were included as VTE. Even if only those deemed possible or probable VTE- associated deaths (1/26 negative VIDAS) are included, there is no change to the initial sensitivities for VIDAS. The same was true if patients treated clinically as PE without an adequate diagnostic workup were included because we had included the 2 of 7 with negative VIDAS and therapy for PE as VTE-positive in the original calculations.

Table 6 Deaths in patients with an incomplete or negative definitive initial workup

Risk group for PE at death

Very low PE risk (b10%)

Definitive–PE-negative workup

palliative care–expected death

Death–PE possible (10%-30%)

Unexpected death–PE likely (N30%)

n (47)

6

8

9

19

7

DD-positive (45)

6

7

9

18

7

Table 7 Effect of comorbidities on DD specificity

Risk factors for reduced DD specificity

No. with risk factor

VIDAS-positive (%)

Active cancer

58 (12 also postoperative, 2 in ICU)

54 (93)

Postoperative/major trauma or delivery within 4/52

48 (12 also had ca, 6 in ICU)

47 (98)

ICU admission within 2/52

9 (6 postoperative [2 with Ca])

9 (100)

No. of IP with at least 1 risk factor from above (total IP no.)

54 (145)

54 (100)

Age N70 yr

149

131 (88)

Ca, cancer.

Because of the wide CIs for sensitivity and negative LR, the data from this study cannot by itself justify using this test in normal clinical practice. However, because the sensitivity (and negative LR) in this unselected population is similar to that achieved in previous management studies, it suggests that this test can safely be used in low- and probably intermediate-risk outpatients to exclude VTE [5,6,17].

Specificity for VIDAS overall was 38%, with outpatient (ED) specificity at 51% and IPs at 15%. This difference in specificity is similar to findings for this and other similar DD tests [6,17]. We looked at some common conditions and patient characteristics (see Table 6) that have been associ- ated with high false-positive rates. It seems that these conditions (recent operations or major trauma, active cancer, and possibly age N70 years) make specificity too low for DD to be a useful screening test. However, there are some caveats to this. We found that in outpatients older than 70 years, 16/84 (19%) had negative VIDAS. This percent- age was not changed if other risk factors were used to stop testing (14/73 [19%]). If age of older than 70 years and prothrombotic risk factors were used to stop VIDAS testing in the 144 IP, 16/54 (30%) would have negative test results. This suggests that a significant number of younger IP (b70 years) without obvious prothrombotic conditions could usefully have DD testing. Because this is subgroup analysis with only modest numbers, these findings should be confirmed in larger prospective cohort studies.

We compared VIDAS to the LA Agen that was previously used and had been freely available to clinicians. The LA Agen had poor sensitivity (78%) but better specificity (67%). This is similar to previous findings for commercial LA tests and confirms their unsuitability in real practice unless a very low-prevalence VTE population can be accurately identified [2,5,7]. The latex test (Agen) had not been used very frequently for VTE diagnosis in our institution before changing to VIDAS probably because of justifiable concerns about the test’s lack of sensitivity (Mr Nick Micholopoulos, coagulation laboratory senior scientist, personal communication).

Methodological concerns for this study

This population was unselected, and the whole range of patients with VTE would be expected (excepting massive PE) with a smaller proportion of the larger VTEs seen in clinical trials[18,19]. There were no exclusions on the basis of delay from symptom onset in this study (as would be

expected in most trials). Therefore, we would expect that this unselected population would have decreased sensitivity for DD tests. This has been demonstrated by the SimpliRed test where initial sensitivities were reported as more than 95% in gold-standard trials, dropping to approximately 85% in management studies, and have been reported as low as 65% in observational studies[5,7,20]. Finally, it is probable that some patients with VTE were missed because the minority of this population had a gold-standard diagnostic workup. However, it is the ability to safely discharge patients and know that they will remain clinically disease- free that clinicians want from a test [21]. In this study, a negative DD had a 99% negative predictive value for excluding VTE based on follow-up or standard diagnosis. Because not all patients had DD performed as part of their diagnostic workup, some tests were performed on stored samples or EDTA tubes. However, if there were any effect, we would expect this to be reduced specificity (more false positives) from activated samples.

The small numbers of positive patients and wide CI for sensitivity were inevitable because of the large numbers of patients being tested and relatively low prevalence rate. Logistically, we were unable to collect data for longer than 6 months, and this was the available population. Clearly, other similar observational studies or larger studies would clarify further the real sensitivity of this test in normal practice.

Was our follow-up adequate? Follow-up was achieved in a high number of cases (N96%). The follow-up was via verbal contact with GP/patient or by review of further IP admissions after the 3-month period. Although this is not as good as direct patient contact (eg, clinic review), we felt that patients (or GPs) were unlikely to forget being diagnosed with DVT or PE and being started on anticoagulants. Chart review was used to extract data on readmission, symptoms, and coexistent disease. Only 2 investigators reviewed the charts and used preagreed criteria for the major findings looked for. A formal interrater reliability was not calculated or compared to an independent third party. The investigators were not blinded to results when extracting data. Formal interrater testing, blinding to results, and follow-up by an independent reviewer would strengthen further studies in this area [22].

Finally, we did not control ordering practices, so that variations in practice and test ordering were almost certainly present between various units and seniority of staff.

However, this is how tests are ordered in reality, and it is the ability of a test to maintain performance in variable conditions that is important for demonstration of real clinical safety and utility.

Implications for current practice

This study suggests (but cannot, in isolation, confirm) that VIDAS maintains sensitivity when used in standard clinical conditions. Given that VIDAS (and other tests without management study tests) is already being used in clinical practice, these results improve our knowledge of real-life performance. We can certainly suggest use in low- risk patients (b10% risk of VTE) where a posttest probability of disease would be less than 2%. Its use in intermediate-risk patients could not be recommended on the basis of this study, but in association with evidence from previous management studies, its use is probably reason- able. We have found that a number of conditions or patient characteristics reduce specificity to the point where testing is unlikely to be useful. In particular, patients who are within

2 weeks of surgery, have major trauma, or have active cancer have very low rates of negative VIDAS. In addition, IP as a group has poor specificity (15%). Education or ordering control would need to be rigorous to select the group of younger IPs without obvious prothrombotic confounding conditions who could benefit from VIDAS testing. Older patients (older than 70 years) also show poor specificity (particularly IP). Older patients in ED popula- tions may have marginal utility for testing (20% negative test results). These findings would need to be validated prospectively in other populations. Finally, there seems little place for older latex tests in excluding VTE unless used in well-selected, very low-risk populations.

Our study suggests that this test may be overused. In addition to the low prevalence of disease (particularly in outpatients ([9%]), there was evidence of indiscriminate testing where a positive DD did not lead to other testing (78 patients). This probably suggests that the clinical risk of PE was very low before the test was ordered or another condition was diagnosed soon after. If ordering is indis- criminate, there would be concerns with the Cost benefit of the test (laboratory charges are approximately AUS $30) and the potential for overuse of more expensive radiologic investigations for VTE. This has been noted in other studies, although the overall effect was to increase VTE diagnosis and numbers of investigations [15,23].

Further studies

Prospective observational studies in multiple centers or over longer time frames would improve our knowledge of these tests’ safety in real life. Follow-up could be improved by formal clinic review at 3 months and better controls for interrater reliability. In addition, larger prospective multi- center trials are needed to clarify subgroups where VTE can be clinically ruled out without a DD test or where testing is

very unlikely to be negative because of comorbidities [15]. They would assist in clarifying guidelines for reducing inappropriate testing particularly relating to underlying illness, age, and IP status. Studies are also needed to further investigate how DD testing is integrated into clinical practice and how its use affects investigation patterns and diagnosis rates for VTE.

Conclusion

We have found that in normal practice, the VIDAS test probably maintains adequate sensitivity for VTE with a negative LR of 0.11. Specificity in outpatients was reasonable but is low in IPs. A latex DD did not perform adequately for routine clinical usage. Other prothrombotic conditions (cancer, postoperative state and need for ICU, advanced age, and IPs) have been identified as probably reducing specificity to the point of clinical futility. In combination with evidence from other studies, we suggest that negative VIDAS DD is a suitable exclusionary test for ED patients with a low (and possibly intermediate) probability of VTE.

Acknowledgments

The authors are indebted to Dr Nick Michalopoulos, senior scientist in the coagulation laboratory, and his staff, for performance of the DD tests and the very generous donation of time and information provided to the authors.

References

  1. Value of the Ventilation/perfusion scan in acute pulmonary embolism. Results of the prospective investigation of pulmonary embolism diagnosis (PIOPED). The PIOPED Investigators. JAMA 1990;263: 2753 – 9.
  2. Manganelli D, Palla A, Donnamaria V, Giuntini C. Clinical features of pulmonary embolism. Doubts and certainties. Chest 1995;107: 25S- 32S.
  3. Egermayer P, Town GI. Compliance with guidelines for the investigation and management of patients with suspected pulmonary embolism at Christchurch Hospital. N Z Med J 1998;111:70 – 3.
  4. Hagen PJ, van Strijen MJ, Kieft GJ, et al. The application of a Dutch consensus diagnostic strategy for pulmonary embolism in clinical practice. Neth J Med 2001;59:161 – 9.
  5. Stein PD, Hull RD, Patel KC, et al. D-dimer for the exclusion of acute venous thrombosis and pulmonary embolism: a systematic review. Ann Intern Med 2004;140:589 – 602.
  6. Perrier A, Desmarais S, Miron MJ, et al. Non-invasive diagnosis of venous thromboembolism in outpatients. Lancet 1999;353:190 – 5.
  7. Legnani C, Pancani C, Palareti G, et al. Contribution of a new, rapid, quantitative and automated method for D-dimer measurement to exclude deep vein thrombosis in symptomatic outpatients. Blood Coagul Fibrinolysis 1999;10:69 – 74.
  8. Buller HR, Lensing AW, Hirsh J, ten Cate JW. Deep vein throm- bosis: new non-invasive diagnostic tests. Thromb Haemost 1991;66: 133 – 7.
  9. Sackett DL, Haynes RB. The architecture of diagnostic research. Br Med J 2002;324:539 – 41.
  10. Bounameaux H, Miron MJ, Blanchard J, et al. Measurement of plasma D-dimer is not useful in the prediction or diagnosis of postoperative deep vein thrombosis in patients undergoing total knee arthroplasty. Blood Coagul Fibrinolysis 1998;9:749 – 52.
  11. Owings JT, Gosselin RC, Anderson JT, et al. Practical utility of the D-dimer assay for excluding thromboembolism in severely injured trauma patients. J Trauma 2001;51:425 – 9 [discussion 29-30].
  12. Lee AY, Julian JA, Levine MN, et al. Clinical utility of a rapid whole- blood D-dimer assay in patients with cancer who present with suspected acute deep venous thrombosis. Ann Intern Med 1999;131: 417 – 23.
  13. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 2003;326:41 – 4.
  14. Bounameaux H, Perrier A, Wells PS. Diagnostic strategies for suspected pulmonary embolism among outpatients. Semin Vasc Med 2001;1:189 – 94.
  15. Kline JA, Mitchell AM, Kabrhel C, et al. Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost 2004;2:1247 – 55.
  16. Hartmann IJ, Prins MH, Buller HR, Banga JD. Acute pulmonary embolism: impact of selection bias in prospective diagnostic studies. ANTELOPE Study Group. Advances in New Technologies Evaluat-

ing the Localization of Pulmonary Embolism. Thromb Haemost 2001; 85:604 – 8.

  1. Miron MJ, Perrier A, Bounameaux H, et al. Contribution of noninvasive evaluation to the diagnosis of pulmonary embolism in hospitalized patients. Eur Respir J 1999;13:1365 – 70.
  2. Heit JA, Minor TA, Andrews JC, et al. Determinants of plasma fibrin D-dimer sensitivity for acute pulmonary embolism as defined by pulmonary angiography. Arch Pathol Lab Med 1999;123:235 – 40.
  3. De Monye W, Sanson BJ, MacGillavry MR, et al. Embolus location affects the sensitivity of a rapid quantitative D-dimer assay in the diagnosis of pulmonary embolism. Am J Respir Crit Care Med 2002; 165:345 – 8.
  4. 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. Ann Emerg Med 2000;35:121 – 5.
  5. Perrier A, Bounameaux H. Validation of helical computed tomogra- phy for suspected pulmonary embolism: a near miss? J Thromb Haemost 2005;3:14 – 6.
  6. Gilbert EH, Lowenstein SR, Koziol-McLain J, et al. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996;27:305 – 8.
  7. Goldstein NM, Kollef MH, Ward S, Gage BF. The impact of the introduction of a rapid D-dimer assay on the diagnostic evaluation of suspected pulmonary embolism. Arch Intern Med 2001;161:567 – 71.

Leave a Reply

Your email address will not be published. Required fields are marked *