Three Part Question
In [adult patients presenting with suspected VTE] does [the use of clinical probability adjusted d-dimer thresholds compared to the standard threshold of <500ng/ml FEU] result in [safe exclusion of VTE and reduction in referral for imaging investigations]
Clinical Scenario
A fit and well 48 year old woman attends your department with pleuritic chest pain and new shortness of breath on exertion. Her heart rate is 105. Several family members have had a recent viral infection. There are no clinical features suggestive of pneumonia. You are keen to exclude pulmonary embolism, so calculate the Wells score as 1.5 (low risk) and subsequently order a D-dimer. This comes back at 505ng/mL FEU. Your lab threshold for a positive test is 500ng/mL FEU. It always makes you cross for some reason when the D-dimer is so close to being ‘negative’. You are complaining about this at the workstation when a colleague overhears, and asks if you have considered applying a clinical probability adjusted (CPA) threshold to the D-dimer test. Your colleague explains that this approach allows you to adjust the D-dimer threshold according to pre-test probability as defined by the Wells score – accepting <1000ng/mL FEU as a negative result in patients at low risk. This is the first you have heard of this approach – you resolve to look it up immediately and see if it could help you safely avoid further imaging and interim anticoagulation.
Search Strategy
Medline (1946 to Jan Week 3 2025) and Embase (1980 to week 3 2025) databases were searched via the OVID interface using the following search criteria:
[Pulmonary embolism.ab,ti. OR Deep Vein Thrombosis ab,ti. OR Venous Thromboembolism/ OR VTE/] AND [D-dimer.ab,ti. OR D-dimer/ OR FDP OR Fibrinogen degradation products/] AND [clinical probability adjusted.ab,ti. OR CPA/ OR probability.mp OR adjusted thresholds/.]
Search Outcome
407 papers were initially identified. We discounted narrative review articles, case reports, cohort studies of <50 patients and articles limited to highly specific patient subgroups (cancer). The search was limited to English language only. After abstract sifting, 24 original research papers were selected for further review, 3 of which were directly relevant to the original three-part question. Two further papers were identified and included following review of references and the grey literature. All papers are summarised in Table 1.
Relevant Paper(s)
Author, date and country |
Patient group |
Study type (level of evidence) |
Outcomes |
Key results |
Study Weaknesses |
Robert- Ebadi et al 2023 Switzerland | 3308 outpatients with suspected PE, taken from 3 cohort studies. Prevalence of PE was 22.1% | Post hoc analysis of CPA thresholds to prospectively collected dataset | Failure rate (% with VTE diagnosis at 90 days despite negative initial work up) | 2.3% (95% CI 1.7 to 3.2) for CPA strategy 0.0% (95% CI 0.0 to 0.4) for Wells Strategy | Missing data.
Population almost 20 years old (from 2006).
High prevalence of VTE at baseline.
Comparison of new strategy against group with high imaging (and potential overdiagnosis) rates.
|
Efficiency (% where VTE excluded without imaging) | 49% for CPA strategy 29.8% (95% CI, 29.8 to 31.4) for Wells strategy |
Kearon et al 2022 Canada | 1508 outpatients with suspected DVT.
Prevalence of DVT was 11.3%
| Prospective diagnostic management (implementation) study of CPA threshold | Failure rate (% with VTE diagnosis at 90 days despite negative initial work up) | 0.6% (95% CI 0.3 to 1.2) for CPA strategy | Multiple exclusions, including prior DVT, pregnancy, suspected PE.
Limited numbers in moderate risk category.
Potential for selective enrolment.
|
Efficiency (% where VTE excluded without imaging) | 38% for CPA strategy 19% for Wells strategy |
Kearon et al 2019 Canada | 2017 patients with suspected PE.
PE prevalence of 7.4%
| Prospective diagnostic management (implementation) study of CPA threshold | Failure rate (% with VTE diagnosis at 90 days despite negative initial work up) | 0.0% (95% CI 0.0 to 0.3) for CPA strategy | Limited numbers in moderate risk category.
Potential for selective enrolment.
Lower prevalence of PE than other studies.
|
Efficiency (% where VTE excluded without imaging) | 65.7% for CPA strategy 48.1% for Wells strategy |
Gerber et al 2023 Switzerland | 68 studies involving 141,880 patients with suspected VTE.
Prevalence was 7.8%
| Systematic review and cohort meta-analysis. Results reported with assumed prevalence of 5% and 20%. | Failure rate (% with VTE diagnosis at 90 days despite negative initial work up) | 0.42 to 1.95% for CPA strategy 0.23 to 1.08% for Wells strategy. | Significant heterogeneity of included trials.
Mainly observational studies with inherent bias.
Variable reporting of pre-test probability with limited assurance on prospective accuracy.
|
Efficiency (% where VTE excluded without imaging) | Additional 34 to 40% potential imaging studies saved using CPA strategy. |
Comment(s)
The D-Dimer blood assay is a marker of fibrinogen degradation products, therefore often used in evaluation of suspected thrombotic disease. D-Dimer is a continuous (not a dichotomous) variable.
Although previous studies have identified a lower threshold of below 500ng/mL Fibrinogen Equivalent Units (FEU) as highly sensitive for negative testing, there is potential to use D-dimer values in a more informed manner.
Previous research has evaluated the role of adjusting D-dimer thresholds to improve the specificity of the test in older people. This method is now supported by international guidelines and appears to improve test efficiency without compromising sensitivity.[6] Further adjusting the d-dimer threshold to clinical pretest probability (accepting a higher threshold value in patients at low clinical risk of VTE) has the potential to further improve test efficiency and avoid the complications of invasive investigations in both younger and older people with suspected VTE.
The evidence reviewed suggests that use of a CPA d-dimer threshold can significantly improve test efficiency with only a small potential impact on sensitivity/failure rate. This impact is potentially influenced by VTE prevalence and exaggerated by use of historical data, overdiagnosis and de novo events (post assessment) within the literature reviewed. In the only large prospective implementation study of CPA d-dimer thresholds to exclude PE without recourse to imaging, no clinically relevant cases of VTE were detected up to 90 days following discharge. Overall, these data suggest that CPA D-dimer thresholds are appropriate for clinical use in appropriate low risk patients with suspected PE.
We found only a single study on the use of CPA thresholds in suspected DVT. Further work is required to evaluate the generalisability of this method across international populations. CPA thresholds have not been studied extensively in patients at moderate risk of PE or in pregnant people, so should not be applied to these groups. CPA thresholds have been directly compared to age adjusted thresholds with inconclusive results – several retrospective studies report an increase in test efficiency with CPA thresholds but a subsequent decrease in negative predictive value.[7, 8] The decision to use a CPA (in preference to age adjusted) threshold should therefore be a shared decision informed by both clinician and patient preference.
Further clinical models have been developed which incorporate CPA thresholds alongside novel measures of gestalt.[9] Such methods have performed well in validation studies, but require further work to determine reliability and external validity across variable populations with variable prevalence.[10, 11]
Clinical Bottom Line
Recent evidence supports the use of CPA D-dimer thresholds to exclude PE without recourse to imaging, in people at low clinical pretest probability. CPA thresholds should not currently be applied in complex subgroups, those at moderate to high pretest probability of PE or in cases of suspected DVT.
References
- Stals et al Safety and Efficiency of Diagnostic Strategies for Ruling Out Pulmonary Embolism in Clinically Relevant Patient Subgroups : A Systematic Review and Individual-Patient Data Meta-analysis Annals of Internal Medicine 2022 Feb;175(2):244-255
- Robert-Ebadi et al External validation of the PEGeD diagnostic algorithm for suspected pulmonary embolism in an independent cohort
- Kearon et al Diagnosis of deep vein thrombosis with D-dimer adjusted to clinical probability: prospective diagnostic management study BMJ 2022; 376
- Kearon et al Diagnosis of Pulmonary Embolism with d-Dimer Adjusted to Clinical Probability New England Journal of Medicine 2019 Nov 28;381(22):2125-2134
- Gerber et al Utility and limitations of patient-adjusted D-dimer cut-off levels for diagnosis of venous thromboembolism—A systematic review and meta-analysis Journal of Internal Medicine May 2023