DECISION RULES AND CLINICAL ASSESSMENT
Estimating the pretest probability for VTE in a patient is the first step in selecting a diagnostic pathway. Figure 56-3 shows one diagnostic algorithm; no singular diagnostic test or algorithm perfectly excludes or diagnoses VTE. Aggressive diagnostic searches can cause harm disproportionate to benefit from hemorrhage associated with anticoagulation for a false-positive result or self-limited small clot diagnosis or from contrast nephropathy. One approach is to test further only in those with pretest probabilities of >2.5%26; those with a pretest probability of PE <2.5% are more likely to be harmed than helped by a diagnostic test, even a d-dimer assay.
Pulmonary embolism rule-out criteria (PERC) rule—diagnostic algorithm for pulmonary embolism (PE). *Some physicians prefer to start with a clinical decision rule such as the Wells' score (where <2, 2–6, and >6 are used instead of <15%, 15%–40%, and >40%, respectively). Note: Determine renal function by clinical picture (healthy, no risk factors for reduced glomerular filtration rate [GFR]) or calculated GFR. Nondiagnostic ventilation–perfusion (V̇/Q̇) scan findings require confirmation from results of another test, such as CT pulmonary angiography (CTPA), if benefits outweigh risks. + = positive for PE; – = negative for PE; Cr = creatinine; High = high probability scan findings; LMWH = low-molecular-weight heparin; Nl = normal; Nondx = nondiagnostic (any reading other than normal or high probability); quant = quantitative.
The PE rule-out criteria (Table 56-4) reliably forecast a probability of PE that is below the 2.0% test threshold in patients with a gestalt low clinical suspicion.27,28 The American College of Emergency Physicians' Clinical Guidelines committee provided a 2b recommendation for the PE rule-out criteria rule.29 The PE rule-out criteria rule had an apparently high failure rate in one secondary analysis done in a European population with a prevalence of disease of 27% in conjunction with a low-risk Geneva score.30 However, the subsequent published erratum to the work showed that the PE rule-out criteria rule had 100% sensitivity in the same population when combined with low gestalt pretest probability as designed, which was also the case in other studies.28,31 Thus, taken together, the weight of the available information indicates that PE can be reliably excluded by the combination of a low gestalt pretest probability plus a negative PE rule-out criteria rule. However, not all patients who have any positive PE rule-out criteria must undergo an objective test for PE, since anyone over 50 years old would be tested with any finding even if suspicion was low. The key is generating a clinical gestalt first and, if low, using the PE rule-out criteria to guide further testing.
TABLE 56-4Pulmonary Embolism Rule-Out Criteria Rule (all nine factors must be present to exclude pulmonary embolism)32 ||Download (.pdf) TABLE 56-4 Pulmonary Embolism Rule-Out Criteria Rule (all nine factors must be present to exclude pulmonary embolism)32
Clinical low probability (<15% probability of pulmonary embolism based on gestalt assessment)
Age <50 years
Pulse <100 beats/min during entire stay in ED
Pulse oximetry >94% at near sea level (>92% at altitudes near 5000 feet above sea level)
No prior venous thromboembolism history
No surgery or trauma requiring endotracheal or epidural anesthesia within the last 4 weeks
No estrogen use
No unilateral leg swelling, defined as asymmetrical calves on visual inspection with patient's heels raised off the bed
Most validated PE prediction systems categorize the patient into one of two (low or above low probability) or three (low, moderate, or higher probability) categories.33 One method uses a computerized database-derived method based on the method of attribute matching to estimate a discrete numerical percentage probability of PE.26 The Wells' score is the most robust scoring system for categorizing the pretest probability for both PE (Table 56-5) and DVT (Table 56-6).
TABLE 56-5Wells' Score for Pulmonary Embolism ||Download (.pdf) TABLE 56-5 Wells' Score for Pulmonary Embolism
|Factor ||Points* |
|Suspected deep venous thrombosis ||3 |
|Alternative diagnosis less likely than PE ||3 |
|Heart rate >100 beats/min ||1.5 |
|Prior venous thromboembolism ||1.5 |
|Immobilization within prior 4 wk ||1.5 |
|Active malignancy ||1 |
|Hemoptysis ||1 |
TABLE 56-6Wells' Score for Deep Vein Thrombosis ||Download (.pdf) TABLE 56-6 Wells' Score for Deep Vein Thrombosis
|Clinical Feature ||Points* |
|Active cancer (treatment within 6 mo, or palliation) ||1 |
|Paralysis, paresis, or immobilization of lower extremity ||1 |
|Bedridden for >3 d because of surgery (within 12 wk) ||1 |
|Localized tenderness along distribution of deep veins ||1 |
|Entire leg swollen ||1 |
|Unilateral calf swelling of >3 cm (below tibial tuberosity) ||1 |
|Unilateral pitting edema ||1 |
|Collateral superficial veins ||1 |
|Alternative diagnosis as likely as or more likely than deep venous thrombosis ||–2 |
|Prior history of DVT or PE† ||1 |
The Wells' original rules separate patients into low-, moderate-, and high-probability groups. The Wells' DVT and PE scoring systems both reliably produce a stepwise increase in probability of clot presence with higher scores. The Wells' scoring system has been modified to classify patients being evaluated for possible PE into a low-risk (score ≤4) or a non–low-risk group (score >4), which has become a mainstream method of use.34 The Charlotte rule is designed to categorize patients into either a low-risk or non–low-risk group (designated "safe" or "unsafe" in Figure 56-4).
Charlotte rule for patients with suspected pulmonary embolism (PE). Patients for whom the decision tree leads to the "safe" boxes have a sufficiently low pretest probability to permit PE to be ruled out solely on the basis of a normal quantitative d-dimer test result. COPD = chronic obstructive pulmonary disease; HR = heart rate; Sao2 = arterial oxygen saturation; SBP = systolic blood pressure.
The Wells' rule has a subjective component—clinical judgment of the likelihood of an alternative diagnosis. Both the Charlotte rule and the revised Geneva score (Table 56-7) exclude subjective assessments.
TABLE 56-7Revised Geneva Score for Pulmonary Embolism ||Download (.pdf) TABLE 56-7 Revised Geneva Score for Pulmonary Embolism
|Clinical Variable ||Points* |
|Age >65 y ||1 |
|Previous venous thromboembolism ||3 |
|Surgery requiring anesthesia or fracture of lower limb in past month ||2 |
|Active malignancy ||2 |
|Unilateral leg pain ||3 |
|Hemoptysis ||2 |
|Unilateral leg edema ||4 |
|Heart rate || |
| 75–94 beats/min ||3 |
| >95 beats/min ||5 |
Runyon et al35 studied the accuracy of unstructured, gestalt estimation of pretest probability for PE, grouping patients into low-, moderate-, and high-risk categories (where the cutoffs suggested to the clinicians were <15%, 15% to 40%, and >40%); the measured frequencies of PE were similar when stratified by Wells' or Geneva scores.33 Moreover, the interobserver agreement for gestalt classification was good (κ = 0.60), and gestalt estimation did not show a decrease in sensitivity based on training level.35 Gestalt estimation is a viable method to estimate pretest probability and is preferred over no estimation.
The d-dimer assay is the best blood test to exclude VTE, working on the principle that clots contain fibrin that is degraded naturally through the action of plasmin. Fibrin breakdown liberates the d-dimer protein into the blood. Multiple manufacturers each produce a slightly different assay, but the d-dimer test can be broadly classified as either qualitative or quantitative. Qualitative tests generally have lower diagnostic sensitivity but higher specificity compared with quantitative tests. Quantitative tests are usually done in the central hospital laboratory and require a turnaround time of at least 1 hour. Different d-dimer assays have different thresholds for normal because of different capture antibodies and optical methods of detection. Some laboratories report results in d-dimer mass concentration (e.g., nanograms per milliliter or micrograms per milliliter), and others report fibrinogen equivalent units. Two fibrinogen molecules produce one d-dimer unit, so that the fibrinogen equivalent unit number is twice the d-dimer number for the same measurement. For PE and DVT, the diagnostic sensitivity of automated quantitative d-dimer assays ranges from 94% to 98% and the specificity ranges from 50% to 60%. The d-dimer has a half-life of approximately 8 hours and can be elevated for at least 3 days after symptomatic VTE.
The most common causes of a false-negative or positive d-dimer result are listed in Table 56-8; notably, all risk factors for VTE may elevate the d-dimer level. d-dimer increases with age and should be adjusted upward for age to maintain adequate exclusionary ability. The most common formula studied is age × 10 nanograms/mL (e.g., an 80-year-old patient would have an adjusted threshold for abnormal at 800 nanograms/mL).34 This assumes the conventional d-dimer cutoff of 500 nanograms/mL; in a large multicenter study, this adjusted approach resulted in a very low false-negative rate (0.3%) when used in conjunction with a Wells' score ≤4 or a simplified revised Geneva score <5.34
TABLE 56-8Factors Known to Alter the d-Dimer Level from Expected Values ||Download (.pdf) TABLE 56-8 Factors Known to Alter the d-Dimer Level from Expected Values
|Potential False-Positive Levels ||Potential False-Negative Levels |
|Age >70 y ||Warfarin treatment |
|Pregnancy ||Symptoms lasting over 5 d |
|Active malignancy or metastasis ||Presence of small clots |
|Surgical procedure in previous week ||Isolated small pulmonary infarction |
|Liver disease ||Isolated calf vein thrombosis |
|Rheumatoid arthritis ||Lipemia |
|Infections || |
|Trauma || |
Chest CT angiography is the most common imaging modality for PE, identifying a clot as a filling defect in contrast-enhanced pulmonary arteries (Figure 56-5). Equipment with more detector heads (e.g., 64- or 128-head scanners) allows better resolution and observation of filling defects in subsegmental pulmonary arteries. The diagnostic sensitivity and specificity of a technically adequate mutidetector CT scan are about 90%.36 Most CT pulmonary angiography protocols require the patient to lie supine and hold their breath for a few seconds, and the scan requires injection of approximately 120 mL of contrast by a computer-controlled injection device. In most centers, the patient must have a peripheral IV catheter (20 gauge or larger) or an approved indwelling line to allow injection of the contrast, and a central catheter cannot be used for injection.
Axial image from a chest CT angiogram demonstrating a filling defect consistent with acute pulmonary embolism. Two white arrowheads outline a circular filling defect in the right middle lobar pulmonary artery. The long white arrow projecting in the left lung points to a filling defect in a segmental artery in the posterior medial segmental artery.
In addition to clot recognition, a CT scan can detect alternative diagnoses, often pneumonia (8% to 22% of cases).37 Interobserver agreement in identifying segmental or larger filling defects is very good, but interobserver agreement for subsegmental clots is poor. In practice, approximately 10% of CT scans are inadequate from secondary motion artifact or poor pulmonary artery opacification, commonly in obese or very tachypneic patients. Acute life-threatening contrast-triggered anaphylaxis or pulmonary edema is very rare, occurring in about 1 in 1000 patients.38 About 15% of patients undergoing contrast-enhanced chest CT scan develop contrast nephropathy, defined as a 25% or greater increase in the serum creatinine concentration within 2 to 7 days of the examination.39,40 At present, the only clearly helpful prophylactic measure to reduce contrast nephropathy is hydration with intravenous balanced crystalloid solutions.41 Other complications from CT scanning include contrast extravasation into a limb that can cause pain or compartment syndrome, and creation of a secondary thrombophlebitis.
Ventilation-Perfusion Lung Scanning
Ventilation–perfusion (V̇ /Q̇) lung scanning can identify a perfusion defect when ventilation is normal. V̇ /Q̇ scanning measures scintillation produced from a gamma ray–emitting atom and yields an image that plots the density of scintillations emitted from the chest using two phases. The perfusion images are usually obtained first and require a peripheral IV catheter for injection and the ability of the patient to sit up and lie down during the procedure. The ventilation component requires the patient to breathe into a nebulizer to inhale an aerosol that contains the isotope. A V̇ /Q̇ scan with homogeneous scintillation throughout the lung in the perfusion portion has nearly 100% sensitivity in excluding PE, regardless of the appearance of the ventilation portion. A V̇ /Q̇ scan with two or more apex central wedge-shaped defects in the perfusion phase (Figure 56-6) with normal ventilation in these regions indicates >80% probability of PE. All other V̇ /Q̇ scan findings are nondiagnostic; taken alone, these cannot exclude or diagnose PE.
Ventilation–perfusion lung scan series consistent with a high probability of acute pulmonary embolism (PE) using standard criteria. The first and third rows project the perfusion phases of the examination, and the second and fourth rows show the ventilation phases. The black arrowheads point to wedge-shaped defects in the perfusion images. Comparison with the corresponding ventilation view immediately below shows relatively homogeneous scintillation activity in the anatomic segments that lack perfusion. These defects are consistent with an acute PE.
Direct pulmonary angiography identifies a clot as a filling defect within the pulmonary artery. This test requires placement of a catheter into the pulmonary artery, usually through the femoral vein, followed by injection of 150 to 300 mL of contrast material. Advantages of direct pulmonary angiography include the ability to demonstrate filling defects in 3-mm or smaller vessels; the ability to measure pulmonary artery pressures; and the potential to treat PE with intrapulmonary catheter–directed modalities or to deploy a vena caval filter. Disadvantages of pulmonary angiography include lack of availability, radiation exposure, and the possibility of contrast-related complications, cardiac dysrhythmias, and rare cardiac or pulmonary arterial perforation.
Venous US is the imaging test of choice in DVT; it can be done quickly and does not use ionizing radiation. When performed by experienced ultrasonographers, it has a diagnostic sensitivity of 90% to 95% and a specificity of 95% for DVT, and venous US has a sensitivity of about 40% as a surrogate method to diagnose PE. The mean sensitivity for DVT of US performed by a trained emergency physician compared with the reference imaging test is 96.1% (95% CI, 90.6% to 98.5%), with a weighted mean specificity of 96.8% (95% CI. 94.6% to 98.1%).42 However, the details of test performance and training remain important, as two-point compression US may miss some clots and new users require experience with at least 10 examinations before gaining adequate skill.
Compression ultrasonography operates on the principle that normal veins compress but thrombosed veins do not. A 7.5-MHz probe is used to visualize the common femoral, superficial femoral, and popliteal veins for comparison with the adjacent femoral and popliteal arteries. The sonographer manually compresses the probe and compares the flattening of the vein with that of the artery. If the vein compresses completely whereas the artery remains patent, this finding indicates "normal compressibility," and the absence of a thrombus is inferred (Figure 56-7). Conversely, if the vein does not compress, the examination is considered positive for a thrombus. The practice of examining the saphenous, tibial, and peroneal veins varies among centers.
Compression venous US images showing normal findings and findings indicating deep venous thrombosis. A. Compression venous US of the common femoral vein and femoral artery. The left view shows a sonographic image of the right femoral artery (A) and common femoral vein (V) obtained immediately inferior to the inguinal ligament. The image on the right shows the same view after manual compression by the operator. The image demonstrates obliteration of the vein while the artery remains open. This is a normal US finding for the vein. B. Venous US image showing evidence of common left femoral vein thrombosis after compression (right panel). The common femoral vein (V) does not compress. Echogenic thrombolytic material can be observed within the vein.
The disadvantages of formal US include the need for specialized equipment and a qualified sonographer and radiologist. The examination can be difficult to perform in obese patients, and the probability of an indeterminate result increases with higher patient body mass index. The compression component of the examination causes pain in some patients and could promote clot embolization. Prior history of DVT can make it difficult to determine if venous noncompressibility represents an old or new clot. Color flow Doppler US can help identify recannulization, suggesting a chronic clot, which may not need anticoagulation.
Indirect CT venography acquires axial images of the leg veins during the venous return phase minutes after the contrast is injected for chest CT angiography. At present, the low rate of clinical utility, increased gonadal radiation, poor technical resolution, and low interobserver reliability for distal DVT all point away from the value of routine CT venography in the workup of an ED patient with suspected PE.43
Planar venography requires injection of contrast material into a small vein in the foot with a proximal venous tourniquet. Subsequent images detail the entire leg venous system, identifying filling defects anywhere from the distal calf up to the iliac vein. Because alternative diagnostic imaging is readily available and a contrast-induced leg clot can result from the test in 10% to 15% of uses, venography is now rarely performed.