This is a guest post by Dr. Lanujan Kaneswaran. Lanujan is a second-year Family Medicine resident at the University of Toronto. He has a background in medical health informatics and machine learning. His areas of interest include artificial intelligence and machine learning in medicine, and health equity through advocacy and technology.
Kaneswaran, L. Can we use D-dimer to assess for left atrial clot in atrial fibrillation?, First10EM, September 11, 2023. Available at:
When managing atrial fibrillation in the emergency department, we typically worry about the presence of left atrial clot and stroke risk associated with rhythm control or cardioversion. We primarily risk stratify these patients through history, specifically around the duration of their symptoms. However, patients often aren’t able to clearly say what time their symptoms started (this is especially muddled if patients wake up with symptoms). In other conditions we use tests, specifically D-dimers, to determine clot presence. Is there utility in using the D-dimer to more objectively detect clot burden or risk stratify atrial fibrillation patients? Having an objective measure could potentially alleviate the uncertainty around history, may cognitively offload this step, and streamline patient flow through the emergency department. A late-2021 systematic review by Diaz-Arocutipa et al. looks at this question.
Diaz-Arocutipa C, Gonzales-Luna AC, Brañez-Condorena A, Hernandez AV. Diagnostic accuracy of D-dimer to detect left atrial thrombus in patients with atrial fibrillation: A systematic review and meta-analysis. Heart Rhythm. 2021 Dec;18(12):2128-2136. doi: 10.1016/j.hrthm.2021.08.027. Epub 2021 Sep 1. PMID: 34481076
This is a systematic review to determine the diagnostic accuracy of D-dimer in detecting left atrial (LA) thrombus in patients with atrial fibrillation. The review searched 4 databases (PubMed, Embase, Scopus, Web of Science) across all dates from inception to December 16, 2020. Studies were assessed with the Quality Assessment of Diagnostic Accuracy Studies 2 tool. The review was reported according to the Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies statement. The review estimates an optimal D-dimer cutoff using a method proposed by Steinhauser, Schumacher and Rücker in their paper “Modelling multiple thresholds in meta-analysis of diagnostic test accuracy studies”. (Steinhauser 2016)
2591 papers were screened, 53 were selected for full-text review, and 11 were included in the final review. Seven studies were performed in Asian countries (mostly in Japan), and 4 studies were performed in European countries. All studies were performed in cardiology clinics.
The range of sample sizes was 40–2494 patients. The mean age ranged from 49.8 to 76 years (median 67.8 years), and 70% of patients were men. The most common comorbidities were hypertension (47%), heart failure (16%), diabetes (13%), and stroke (9%). Of note, 63% of patients were on anticoagulation. 16% of patients were receiving vitamin K antagonists, while 38% were receiving non–vitamin K oral anticoagulants. The mean CHA2DS2-VASc score ranged from 2 to 4.4 (median 2.9) across 5 studies.
All patients appropriately received the gold standard test of transesophageal echocardiography (TEE) in detection of LA thrombus. All D-dimers were collected the same day as TEE, and this was included in their study eligibility criteria. The true positive and negative diagnoses were also determined by TEE. (The use of the same gold standard for everyone is important, as is eliminates partial verification bias.)
In 7 studies (n = 2205), the D-dimer levels were significantly higher in patients with LA thrombus than in patients without thrombus (mean difference 1293 ng/mL; 95% confidence interval 719.9–1866.9 ng/mL).
Below is a summary of the pooled sensitivities and specificities (followed by 95% confidence interval) along with positive and negative predictive values assuming LAT prevalence of 10% (based on Di Minno et al.’s review “Prevalence of left atrial thrombus in patients with non-valvular atrial fibrillation. A systematic review and meta-analysis of the literature”). (Di Minno 2016). Age-adjusted D-dimer were calculated as age times 10 ng/mL in patients older than 50 years. The area under the curve for the optimal cutoff was 0.76.
|Sensitivity||Specificity||Positive predictive value||Negative predictive value|
|D-dimer cutoff||500 ng/mL (7 studies)||50% (26%–74%)||88% (76%–95%)||31.6%||94.1%|
|Age-adjusted (2 studies)||36% (14%–66%)||99% (96%–99%)||80%||93.3%|
|390 ng/mL (AUC 0.76)||68% (44%–85%)||73% (54%–86%)||21.8%||95.4%|
Based on these findings, the review article proposes the following flowchart for D-dimer interpretation:
Although the article text suggests negative D-dimers (measured below the cutoff 390 ng/mL) as an endpoint for ruling out LA thrombi, the decision flowchart mentions suggests “proceed to TEE” as the next step in management among patients with a left atrial thrombus despite a negative D-dimer. As it would be impossible to know which patients were false positives without performing TEEs on everyone, we assume that this was a typo rather than a real recommendation.
Safe miss rate for an LA thrombus?
What is a safe miss rate for an LA thrombus? The negative predictive value of approximately 93.3–95.4% with D-dimer is reasonable, but suggests a 5–7% chance of missed LA thrombus (i.e. false negative).
It is unclear what percentage of LA thrombi lead to post-cardioversion strokes as the research around this is lacking (likely due to obvious ethical concerns). CAEP’s Best Practices algorithm implies a theoretical maximum acceptable stroke risk of approximately 1%. Assuming we are working with the maximum acceptable risk, if more than ~14% of these D-dimer missed LA thrombi led to post-cardioversion strokes, the biomarker would be less safe in risk stratification than the CAEP Atrial Fibrillation Best Practices algorithm. I’m not convinced this review will be practice-changing for most providers; I don’t believe providers would be convinced of these odds and would require more compelling studies comparing it to Canadian best practices.
Based on the pooled data, the authors suggest an optimal D-dimer cutoff of 390 ng/mL. This retrospective threshold is likely overfit to this data set. (It is unclear why we would use a different cut-off for atrial fibrillation than we use in venous thromboembolic disease). Evaluating the true utility of this cutoff (i.e. sensitivity, specificity, positive and negative predictive values, and other ratios) would require external validation in future research. For now, it probably makes more sense to evaluate the performance of D-dimer with conventional cutoffs, which makes the sensitivity look even worse.
Managing false positives
Routine ordering of D-dimers in all patients with atrial fibrillation would yield many false positives, and would invite questions around their interpretation and further management. Should CHADS-CASC type risk factors around safety with cardioversion be assessed simultaneously to D-dimer testing or after a positive D-dimer? What should be done if in the setting of a positive D-dimer but when the patient has less than 12 hours of symptom onset and no absolute contraindications to cardioversion? Should CTPA be considered despite having low suspicion for PE clinically? While the authors offer the flowchart shown above for D-dimer interpretation using their optimal cutoff, it doesn’t address the considerations and steps that need to be taken prior to TEE, and assumes these tests are done in isolation. The potential for cognitive offloading offered by this test in the ideal true negative scenario is undone by the complexity in false positive interpretation and practicalities of ED resource management.
Not an emergency department population
While the prior points suggest this application of D-dimer may be irrelevant to current ED practice, the potential exists if researched in a well-defined, representative population. All the studies referred to in this systematic review were completed in cardiology clinics with long-standing atrial fibrillation. This presumably would lead to a higher rate of LA thrombi in the study population compared to the typical emergency department given the selection bias; cardiology patients are filtered by referring providers and have more risk factors, while ED patients present earlier and are more often self-referred. Hence, the missed LA thrombi rate of 4.6% may be high in this review but could be within an acceptable range in the ED setting. If future ED-focused research can demonstrate lower false negative (missed LA thrombi) and false positive rates, those findings would alleviate prior concerns and could be practice- and potentially guideline-changing (i.e. we could imagine possible utility in patients falling within the 12 to 48 hour window).
The evidence does not currently appear strong enough to support D-dimer’s use in risk stratification of atrial fibrillation patients in the emergency department, but it proposes an interesting approach that warrants further research in a well-defined ED population (such as in patients with 12 to 48 hours of symptoms).
- Almorad A, Ohanyan A, Pintea Bentea G, Wielandts JY, El Haddad M, Lycke M, O’Neill L, Morissens M, De Keyzer E, Nguyen T, Anghel L, Samyn S, Berdaoui B, Tavernier R, Vandekerckhove Y, Duytschaever M, Verbeet T, Knecht S, Castro Rodriguez J. D-dimer blood concentrations to exclude left atrial thrombus in patients with atrial fibrillation. Heart. 2021 Feb;107(3):195-200. doi: 10.1136/heartjnl-2020-317612. Epub 2020 Oct 21. PMID: 33087410
- Andrade, J. G., Aguilar, M., Atzema, C., Bell, A., Cairns, J. A., Cheung, C. C., Cox, J. L., Dorian, P., Gladstone, D. J., Healey, J. S., Khairy, P., Leblanc, K., McMurtry, M. S., Mitchell, L. B., Nair, G. M., Nattel, S., Parkash, R., Pilote, L., Sandhu, R. K., Sarrazin, J. F., … Members of the Secondary Panel (2020). The 2020 Canadian Cardiovascular Society/Canadian Heart Rhythm Society Comprehensive Guidelines for the Management of Atrial Fibrillation. The Canadian journal of cardiology, 36(12), 1847–1948. https://doi.org/10.1016/j.cjca.2020.09.001
- Chen PH, Schwade M, Sharma G, Robinson VJB. Value of D-Dimer in Risk Stratification for Thromboembolism in Patients With Atrial Fibrillation and Low CHA2DS2-VASc Score. J Investig Med High Impact Case Rep. 2023 Jan-Dec;11:23247096231165740. doi: 10.1177/23247096231165740. PMID: 37073479; PMCID: PMC10123902.
- Di Minno MN, Ambrosino P, Dello Russo A, Casella M, Tremoli E, Tondo C. Prevalence of left atrial thrombus in patients with non-valvular atrial fibrillation. A systematic review and meta-analysis of the literature. Thromb Haemost. 2016 Mar;115(3):663-77. doi: 10.1160/TH15-07-0532. Epub 2015 Nov 26. PMID: 26607276.
- Diaz-Arocutipa C, Gonzales-Luna AC, Brañez-Condorena A, Hernandez AV. Diagnostic accuracy of D-dimer to detect left atrial thrombus in patients with atrial fibrillation: A systematic review and meta-analysis. Heart Rhythm. 2021 Dec;18(12):2128-2136. doi: 10.1016/j.hrthm.2021.08.027. Epub 2021 Sep 1. PMID: 34481076.
- Jaakkola, S., Nuotio, I., Hartikainen, J. E. K., & Airaksinen, K. E. J. (2020). Early Cardioversion for Acute Atrial Fibrillation in Low Stroke Risk Patients Is Safe Without Anticoagulation. Journal of the American College of Cardiology, 76(2), 226–227. https://doi.org/10.1016/j.jacc.2020.04.076
- Kida K, Nabeta T, Ishida M, Shiono T, Suzuki N, Doi S, Tsukahara M, Ohta Y, Kimura T, Morishima Y, Takita A, Matsumoto N, Akashi YJ, Ako J, Inomata T. D-dimer levels in patients with nonvalvular atrial fibrillation and acute heart failure treated with edoxaban. J Cardiol. 2022 Jun;79(6):759-767. doi: 10.1016/j.jjcc.2022.01.005. Epub 2022 Feb 9. PMID: 35148920.
- Lin C, Bao Y, Hua W, Zhang N, Jin Q, Xie Y, Wei Y, Luo Q, Liu Z, Ling T, Pan W, Xie Y, Wu L. Differences in D-dimer blood concentration in atrial fibrillation patients with left atrial thrombus or severe left atrial spontaneous echo contrast. J Interv Card Electrophysiol. 2022 Aug;64(2):341-347. doi: 10.1007/s10840-021-00991-3. Epub 2021 May 3. PMID: 33937955.
- Nuotio, I., Hartikainen, J. E., Grönberg, T., Biancari, F., & Airaksinen, K. E. (2014). Time to cardioversion for acute atrial fibrillation and thromboembolic complications. JAMA, 312(6), 647–649. https://doi.org/10.1001/jama.2014.3824
- Ohara T, Farhoudi M, Bang OY, Koga M, Demchuk AM. The emerging value of serum D-dimer measurement in the work-up and management of ischemic stroke. Int J Stroke. 2020 Feb;15(2):122-131. doi: 10.1177/1747493019876538. Epub 2019 Sep 19. PMID: 31537182.
- Steinhauser S, Schumacher M, Rücker G. Modelling multiple thresholds in meta-analysis of diagnostic test accuracy studies. BMC Med Res Methodol. 2016 Aug 12;16(1):97. doi: 10.1186/s12874-016-0196-1. PMID: 27520527; PMCID: PMC4983029.
- Stiell IG, de Wit K, Scheuermeyer FX, Vadeboncoeur A, Angaran P, Eagles D, Graham ID, Atzema CL, Archambault PM, Tebbenham T, McRae AD, Cheung WJ, Parkash R, Deyell MW, Baril G, Mann R, Sahsi R, Upadhye S, Brown E, Brinkhurst J, Chabot C, Skanes A. 2021 CAEP Acute Atrial Fibrillation/Flutter Best Practices Checklist. CJEM. 2021 Sep;23(5):604-610. doi: 10.1007/s43678-021-00167-y. Epub 2021 Aug 12. PMID: 34383280; PMCID: PMC8423652.
- Stiell, I. G., McMurtry, M. S., McRae, A., Parkash, R., Scheuermeyer, F., Atzema, C. L., & Skanes, A. (2019). Safe Cardioversion for Patients With Acute-Onset Atrial Fibrillation and Flutter: Practical Concerns and Considerations. The Canadian journal of cardiology, 35(10), 1296–1300. https://doi.org/10.1016/j.cjca.2019.06.007
- Wu CC, Li XB, Duan JB, He JS, Zhu TG, Yu C, Li D, Ze F, Guo JH, Wang L. [Value of D-dimer and left atrial size combined with CHA2DS2-VASc score in excluding left atrial thrombosis in patients with non-valvular atrial fibrillation]. Zhonghua Yi Xue Za Zhi. 2021 Dec 28;101(48):3938-3943. Chinese. doi: 10.3760/cma.j.cn112137-20210608-01303. PMID: 34954995.