Not a shift goes by in which EPs don't think about pulmonary embolism. Chest pain, shortness of breath, unexplained hypoxia, syncope, or tachycardia. You name it, PE enters the differential, and EPs routinely rely on well-studied, validated decision instruments to make the diagnosis.
These diagnostic tools are based on the formulation of the pretest probability, the application of tests (D-dimer, PERC, or advanced imaging), and the resulting post-test probability. (Emerg Med Australas 2005;17:322; JAMA 2006;295:172.) Despite having well-established standards for PE workup, current practice patterns vary.
Conventional high-sensitivity D-dimer testing (<500ug/L) is the backbone of the modern PE workup. It has been extensively studied and carries an overall sensitivity of 99.5% when applied to patients with low pretest probability of PE. (Am J Respir Crit Care Med 1997;156[2 Pt 1]:492.) The corresponding specificity of 41%, however, leaves much to be desired. Similarly, the well-known PE rule-out criteria (PERC) has been validated with 97.4% sensitivity, but has a poor specificity of 21.9%. (J Thromb Haemost 2008;6:772.) The overall low specificity using a conventional D-dimer cutoff has led to a significant burden of overtesting with advanced imaging. I think we can do better for our patients.
The American College of Emergency Physicians updated its clinical policy in 2011 to simplify the Wells' criteria into a dichotomous score, which has been widely accepted into clinical practice. (Ann Emerg Med 2011;57:628.) In the two-tier system, “PE likely” patients (Wells' score >4) go straight to advanced imaging with CTPA or VQ scanning. “PE unlikely” results from a score of 4 or less, and further testing (with PERC and subsequently D-dimer if PERC is positive) is suggested for patients in that category.
Table 1 illustrates the posttest probabilities for a patient with a pretest probability for PE of less than 12 percent and various combinations of testing. A patient with a Wells' score of less than or equal to 4 and a negative PERC score is considered ruled out with an accepted miss rate of 1.6 percent. This is standard of care, but it is also the least sensitive approach to a PE rule-out strategy.
A More Sensitive D-Dimer
Baseline D-dimer levels have been proven to increase with age. (Gerontology 1995;41:159.) Age-adjusted D-dimer (age x 10 ug/L for patients over 50) has been studied and validated with an overall pooled sensitivity of 97%. (BMJ 2013;346:f2492.) The ADJUST-PE trial published in 2014 was a prospective validation study across 19 European hospitals, evaluating 3,346 patients with suspected PE with an age-adjusted D-dimer model. (JAMA 2014;311:1117.) The PE miss rate at three months using an age-adjusted versus conventional D-dimer cutoff was 0.3 percent and 0.1 percent, respectively. This study has been met with criticism for using six different D-dimer assays and for using the Wells' score and revised Geneva score for determining pretest probability. Regardless, the age-adjusted D-dimer in this study performed better than our current standard of care using Wells ≤4 and PERC.
Another study showed that this finding still holds up when excluding the pretest probability. Sharp, et al., conducted a retrospective review of 31,094 patients evaluated for PE in 2015, and found a 93% sensitivity using an age-adjusted approach. (Ann Emerg Med 2016;67:249.) If all previous studies were ignored and these data were used alone, application of an age-adjusted D-dimer to a Wells' score ≤4 would result in a 1.3 percent false-negative rate of PE, which is still better than the current model (1.6%).
These authors also calculated potential harm reduction resulting from unnecessary imaging. Using an age-adjusted model instead of the standard model, we could prevent 322 cases of contrast-induced nephropathy, 29 cases of severe renal failure, and 19 deaths related to contrast-induced nephropathy per 10,000 suspected PE encounters.
Table 2 lists the posttest probabilities using the original “low risk” Wells' criteria and various testing modalities. Age-adjusted D-dimer produces a remarkably low miss rate of 0.12 percent in this low-risk population, typically patients with no risk factors for PE, but fall out of PERC because of age over 50. With this degree of sensitivity, a modified approach could be incorporated by EPs who are uncomfortable with applying the age-adjusted D-dimer to patients with Wells' scores ≤4.
The use of an age-adjusted D-dimer threshold was endorsed by the American College of Physicians in 2015. (Ann Intern Med 2015;163:701.) Emergency physicians undeniably evaluate more undifferentiated patients than internists, and should widely endorse this policy. The evidence is sufficient to do this safely without missing additional significant PE. It also allows for reduced testing for PE in elderly populations prone to complications of contrast-induced nephropathy.
Instituting a change of practice will be challenging. It will require modification of years of routine clinical practice for some. It may be uncomfortable and resisted by some, but EPs should embrace the new evidence and update their practice. This is an opportunity to provider better, safer care to our patients.Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.