Pulmonary Embolism
DEFINITION AND ETIOLOGY
Pulmonary embolism (PE) refers to the presence of endogenous or exogenous substances within the pulmonary vascular bed that result in partial or complete hindrance to blood flow. Among all causes for PE, deep venous thrombosis is the most common (Box 1).
PREVALENCE AND RISK FACTORS
Pulmonary embolism is, in the overwhelming majority, a consequence of deep venous thrombosis (DVT); the two together constitute venous thromboembolism (VTE). The incidence of PE thus reflects the presence of risk factors for VTE in a given population (Box 2).
Box 2 Risk Factors for Venous Thromboembolism
Adapted from Geerts WH, Pineo GF, Heit JA, et al: Prevention of venous thromboembolism: The Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126:338S-400S; and Anderson FA Jr, Spencer FA: Risk factors for venous thromboembolism. Circulation 2003;107:I9-I16.
There are many known risk factors for VTE and PE (see Box 2). These seemingly disparate risk factors may be easier to comprehend if Virchow’s triad (venous stasis, hypercoagulability, and endothelial injury) is remembered. The pathogenic mechanisms underlying the risk factors for VTE are either the de novo development or an accentuation of pre-existing elements of Virchow’s triad. Thus, trauma can result in immobility (venous stasis), tissue injury (leading to hypercoagulability), and endothelial injury. Patients who have pre-existing risk factors for VTE (e.g., hypercoagulability) may be more at risk for developing VTE after a predisposing event (e.g., surgery or trauma) than another person who experiences similar events but is free from risk factors. The presence of several risk factors in a patient results in a synergistic increase in the risk for VTE.
A subset of patients with DVT will develop pulmonary embolism. The location of the thrombus within the deep venous system is a key factor in this progression. Proximal (knee and above) deep vein involvement portends an increased risk of pulmonary embolism. A high probability ventilation/perfusion (
) scan (suggestive of PE) was seen in 40% to 50% of patients with symptomatic proximal DVT but with no symptoms of PE.1 The real incidence of PE is probably higher, even when false-positive high-probability 
scans are accounted for, because the sensitivity of a high-probability scan is only about 50%. Quantifying symptomatic PE in untreated proximal DVT is understandably difficult; one review suggested an incidence of 50% over a 3-month period in this group.1
In contrast, progression with distal (calf) DVT is less common. Asymptomatic distal (calf) DVT progresses in only about one sixth of patients to involve more-proximal veins.1 Symptomatic distal DVT extends proximally in up to one third of such episodes.
DIAGNOSIS
There is no clinical feature and no single diagnostic test that reliably distinguishes patients with and without PE. In any patient, the presence of multiple signs, symptoms, and historical features that are known to occur with a higher frequency in patients with PE increases the likelihood of PE. This rationale underlies the development of clinical prediction models for PE (Tables 1 and 2) and, in turn, implies that excessive reliance on any one element for diagnosing PE can be misleading. This approach has been formally evaluated for clinical prediction models for DVT where individual elements of the Wells clinical prediction rule for DVT were less helpful for diagnosing DVT than when combined in a prediction model.2 It is likely that a similar situation exists for clinical prediction rules for PE as well.
Table 1 Clinical Prediction Rules (Wells Score)
| Variable | Points |
|---|---|
| Clinical signs and symptoms of DVT (minimum of leg swelling and pain with palpation of the deep veins) | 3 |
| Alternative diagnosis is less likely than PE | 3 |
| Heart rate >100 bpm | 1.5 |
| Immobilization or surgery in the previous 4 wk | 1.5 |
| Previous DVT or PE | 1.5 |
| Hemoptysis | 1 |
| Malignancy (on treatment, treated in the last 6 mo or palliative) | 1 |
DVT, deep venous thrombosis; PE, pulmonary embolism.
Table 2 Risk Categories Based on Wells Score
| Cumulative Score | Risk Category |
|---|---|
| Version Used in PIOPED II | |
| <2 | Low |
| 2-6 | Intermediate |
| >6 | High |
| Version Used in CHRISTOPHER Study | |
| ≤4 | PE unlikely |
| >4 | PE likely |
PE, pulmonary embolism; PIOPED, Prospective Investigation of Pulmonary Embolism Diagnosis.
Adapted from Wells PS, Anderson DR, Rodger M, et al: Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: Increasing the model’s utility with the SimpliRED D-dimer. Thromb Haemost 2000;83:416-420.
Efforts to identify modalities (e.g., helical computed tomography [CT]) that have a high negative predictive value independent of clinical assessment have not been entirely successful. In particular, there are concerns about selection bias in studies on helical CT, the modality on which most attention has been focused in the search for a context-independent test.3 At present, diagnostic algorithms that include clinical prediction rules, laboratory tests, and imaging are the most reliable ways to diagnose PE.
CLINICAL FEATURES
Symptoms commonly associated with pulmonary embolism include dyspnea, chest pain, cough, hemoptysis, apprehension, palpitations, syncope, and sweating. Pleuritic chest pain and sudden dyspnea are significantly more common in patients with pulmonary embolism but are not pathognomonic.4 Signs commonly associated with PE include tachypnea, tachycardia, neck vein distention, a fourth heart sound, a loud pulmonary component of the second heart sound, inspiratory crackles, pleural rub, and low-grade fever (<38.0° C). Tachypnea and tachycardia are more common in patients with PE, but the other signs are not helpful in distinguishing between patients with and without PE.4
Clinical Prediction Rules
The Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study results suggested that the clinical impression of an experienced physician, arrived at by the synthesis of clinical symptoms, signs, laboratory data, and imaging, had reasonable correlation with the actual incidence of pulmonary embolism in each of the incidence categories (low, intermediate, and high).5 Clinical prediction rules thus evolved in a milieu of accumulating data that underscored the unreliability of individual signs or symptoms or a syndromic approach to the diagnosis of PE and the possibility of improved diagnostic performance from a systematic evaluation of a group of key factors (risk factors, signs, symptoms, imaging) that were helpful in discriminating between patients with and without PE.
Many different clinical prediction rules have been proposed.6 The simplified (or modified) Wells prediction rule7 has been used in most prospective validation studies (see Table 1). It evaluates seven factors and categorizes patients into three risk categories: low, intermediate, and high. In a prospective study that used the simplified Wells prediction rule, the incidence of PE was 1.3%, 16.2%, and 40.6% in the low-, intermediate-, and high-risk categories, respectively.8
Wells and colleagues,7 in the same study, included alternative cutoff values for the clinical prediction rule that yielded a dichotomous outcome (“PE likely” and “PE unlikely”) instead of the three risk categories (see Table 1). This was done to help dichotomize the patient population into those who did not need further investigations (those who were in the “PE unlikely” group and had negative D-dimer tests) and all others who would need further workup. With this dichotomous model and additional D-dimer testing, 46% of the patients in their validation cohort were categorized as “PE unlikely” and had negative D dimers.7 In this subgroup (PE unlikely and negative D dimer), the rate of PE was only 1.7%. In comparison, only 27% of patients had a combination of low risk and negative D dimer as assessed by the three-category model in the same study.7 The dichotomous model was thus able to limit further investigations in 46% of patients, with no increase in the incidence of PE on follow-up. A prospective study has confirmed the safety of the two-category approach along with CT angiography and D-dimer testing.9
At present, sufficient data support the use of the Wells prediction rule (either the two- or three-category modification) in conjunction with a sensitive D-dimer test in patient populations known to have a relatively low incidence of PE.7,9
Any score not validated prospectively or developed in a patient population that is distinctly different (ethnicity, encounter setting, high vs. low incidence of PE, pre-screened patient population with a low incidence of alternative diagnosis) from the one the score is being applied to might not categorize patients as well as in the original study. In a retrospective study of PIOPED patients, the Wells score performed poorly in specific subgroups (surgical, ICU and CCU groups), but its diagnostic accuracy was acceptable for outpatients.10
It is unlikely that a particular schema will suffice for all clinical situations, and therefore it is difficult to recommend a particular prediction rule for universal use. The rigid application of clinical scores is not advocated; in the modified Wells score, this is explicitly avoided by the inclusion of a high scoring criterion that reminds the clinician to judge whether an alternative diagnosis is less likely than PE (see Table 1).4 Thus, with due diligence, clinical assessment models (modified Wells score, Geneva score) can be used by physicians of varying levels of experience to classify patients into low-, intermediate-, and high-risk categories as accurately as experienced physicians can by their clinical gestalt.3
D Dimer
A highly sensitive D-dimer test would ensure levels above the chosen threshold for nearly all patients with PE. However, such high sensitivity often comes at the cost of low specificity (high false-positive rates). Even if elevated levels are not taken as evidence of PE, positive D-dimer tests usually result in further investigations, each with its adverse effects and inherent false-positive and false-negative rates. If more patients are directed toward further investigations after a highly sensitive D-dimer test, the advantage of detecting a higher number of true positives will be offset to a greater or lesser degree by the cost and adverse effects associated with investigating a greater absolute number of patients. As is evident, a balance between high sensitivities (ensuring that no patients with PE are missed) and high specificity (ensuring that no normal patient receives anticoagulation for an erroneous diagnosis of PE) is essential. A false-negative rate of 1% to 2% is considered acceptable for a diagnostic protocol (not each individual test) in view of similar numbers encountered with pulmonary angiography, which is considered the gold standard.4
An appropriate protocol categorizes a patient into differing risk categories with a clinical prediction rule (e.g., low, intermediate, and high risk for the Wells score), thus estimating the pretest probability based on incidence rates for similar patients in historical validation studies (e.g., 1%-3% for the low-risk category in the Wells score). An appropriate protocol also selects a test with an appropriate likelihood ratio (e.g., negative likelihood ratio of 0.13 for the ELISA assay) and yields a post-test probability (<1% for this example). Thus, an adequate match between the patient population and the selected clinical prediction rule is necessary to ensure a reasonably accurate estimation of the pretest probability. Similarly, a test with an appropriate likelihood ratio is necessary to ensure that it significantly changes the pretest probability.4
These considerations have a direct bearing on D-dimer tests used in diagnosing PE. Many different assays are used for D-dimer assessment. A meta-analysis by Stein11 demonstrated better sensitivity and likelihood ratios for enzyme-linked immunosorbent assay (ELISA)-based methods than for other assays. The sensitivity of the quantitative rapid ELISA assay was 95%, with a negative likelihood ratio of 0.13 (cutoff value <500 ng/mL). A negative test thus effectively excludes patients who have a low pretest probability. These variations underscore the importance of knowing the particular assay used locally.
The results of a D-dimer test, irrespective of the sensitivity of the assay, cannot be interpreted in isolation. The diagnostic accuracy of D-dimer test is less in hospitalized patients than in outpatients.12 The incidence of a false-negative D-dimer result in patients categorized as “PE likely” or high probability for PE was considerable (10.3% and 20%, respectively) in one study, although a less-sensitive whole-blood assay was used.7 Even when sensitive assays are used, the false-negative rate rises as the pretest probability rises. These considerations reiterate the necessity for interpreting results from diagnostic studies as well from clinical prediction rules in the context of the various factors that can influence the manifestations and sequelae of PE.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
