Pulmonary contrast enhanced magnetic resonance angiography (CE-MRA) is useful for the primary diagnosis of pulmonary embolism (PE). Many sites have chosen not to use CE-MRA as a first line of diagnostic tool for PE because of the speed and higher efficacy of computerized tomographic angiography (CTA). In this review, we discuss the strengths and weaknesses of CE-MRA and the appropriate imaging scenarios for the primary diagnosis of PE derived from our unique multi-institutional experience in this area. The optimal patient for this test has a low to intermediate suspicion for PE based on clinical decision rules. Patients in extremis are not candidates for this test. Younger women (< 35 years of age) and patients with iodinated contrast allergies are best served by using this modality We discuss the history of the use of this test, recent technical innovations, artifacts, direct and indirect findings for PE, ancillary findings, and the effectiveness (patient outcomes) of CE-MRA for the exclusion of PE. Current outcomes data shows that CE-MRA and NM V/Q scans are effective alternative tests to CTA for the primary diagnosis of PE.

Pulmonary embolism (PE) is a common differential diagnosis in patients with pulmonary fibrosis presenting with a clinical deterioration. Both ventilation/perfusion (V/Q)-single photon emission computed tomography (SPECT) and computed tomographic pulmonary angiography (CTPA) are routinely used to detect PE. However, the value of V/Q-SPECT and CTPA in this scenario has not been studied so far. We aimed to investigate the concordance of V/Q-SPECT and CTPA in patients with pulmonary fibrosis and suspicion of pulmonary embolism.

A total of 22 consecutive patients with pulmonary fibrosis and clinical deterioration who underwent both V/Q-SPECT and CTPA were included in the study and analyzed for the presence of pulmonary embolism.

Nine of 22 patients (41%) had evidence for pulmonary embolism in V/Q-SPECT, and two of these patients had matching evidence for pulmonary embolism in CTPA. In the other seven patients with positive findings in V/Q-SPECT, no evidence of pulmonary embolism was found in CTPA. None of the 13 patients with a negative V/Q-SPECT had evidence for pulmonary embolism in CTPA.

In patients with pulmonary fibrosis and suspected pulmonary embolism, pulmonary embolism is detected more frequently by V/Q-SPECT than by CTPA. Thromboembolic disease is identified on CTPA only in a minority of patients with positive findings on V/Q-SPECT. When making treatment decisions, clinicians should be aware of the high rate of discordant findings in V/Q-SPECT and CTPA in this specific patient population.

Acute pulmonary thromboembolism (PE) is a cardiovascular emergency associated with significant morbidity and a 5-35 % mortality for untreated pulmonary embolism. If promptly diagnosed and treated, the mortality rate can be significantly reduced. Diagnosis of acute PE continues to be a clinical challenge, with diagnostic imaging playing an important role. This review discusses the clinical challenges of diagnosing acute PE, presents an evidence-based review of the current tests and ever-evolving imaging technology, and highlights special considerations related to radiation dose, contrast media use, and pregnant patients.

Current diagnostic management of hemodynamically stable patients with clinically suspected acute pulmonary embolism (PE) consists of the accurate and rapid distinction between the approximate 20-25% of patients who have acute PE and require anticoagulant treatment, and the overall majority of patients who do not have the disease in question. Clinical outcome studies have demonstrated that, using algorithms with sequential diagnostic tests, PE can be safely ruled out in patients with a clinical probability indicating PE to be unlikely and a normal D-dimer test result. This obviates the need for additional radiological imaging tests in 20-40% of patients. CT pulmonary angiography (CTPA) has become the first line tool to confirm or exclude the diagnosis of PE in patients with a likely probability of PE or an elevated D-dimer blood concentration. While single-row-detector technology CTPA has a low sensitivity for PE and bilateral compression ultrasound (CUS) of the lower limbs is considered necessary to rule out PE, multi-row-detector CTPA is safe to exclude PE without the confirmatory use of CUS.

Lung perfusion scintigraphy (LPS) with technetium-99m-labeled macro-aggregates of albumin (Tc-99m-MAA) is well established in the diagnostic of pulmonary embolism (PE). In the last decade, it was shown that single-photon emission computer tomography (SPECT) acquisition of LPS overcame static scintigraphy. Furthermore, there are rare indications for LPS, such as preoperative quantification of regional lung function prior to lung resection or transplantation, optimization of lung cancer radiation therapy, quantification of right-left shunt, planning of intra-arterial chemotherapy, and several rare indications in pediatrics. Moreover, LPS with Tc-99m-MAA is a safe method with low radiation exposure. PE can also be diagnosed by spiral computer tomography (CT), ultrasound, magnetic resonance angiography, or pulmonary angiography (PA, former gold standard). The present review considers all these methods, especially spiral CT, and compares them with LPS with respect to sensitivity and specificity and gives an overview of established and newer publications. It shows that LPS with Tc-99m-MAA represents a diagnostic method of continuing value for PE. In comparison with spiral CT and/or PA, LPS is not to be defeated as mentioned also by the most actual Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) II reports. This applies in particular to chronic or recurring embolisms, whereas currently spiral CT may be of greater value for major or life-threatening embolisms. At present, LPS cannot be replaced by other methods in some applications, such as pediatrics or in the quantification of regional pulmonary function in a preoperative context or prior to radiation therapy. LPS still has a place in the diagnostics of PE and is irreplaceable in several rare indications as described earlier.

The aim of this retrospective study is to illustrate clinical utility and impact of pulmonary embolism (PE) diagnostics of up to date Ventilation/Perfusion SPECT (V/P (SPECT)) applying holistic interpretation criteria.

During a 2-year period 2328 consecutive patients referred to V/P(SPECT) for clinically suspected PE were examined. Final diagnosis was established by physicians clinically responsible for patient care. To establish the performance of V/P(SPECT) negative for PE, patients were followed up by medical records for 6 months.

Ventilation/Perfusion SPECT was feasible in 99% of the patients. Data for follow-up were available in 1785 patients (77%). PE was reported in 607 patients (34%). Normal pattern was described in 420 patients (25%). Pathology other than PE such as a pneumonia, left heart failure, obstructive lung disease, tumour was described in 724 patients (41%). Report was nondiagnostic in 19 patients (1%). Six cases were classified as falsely negative because PE was diagnosed at follow-up and was fatal in one case. Six cases were classified as falsely positive because the clinician decided not to treat. In 608 patients with final PE diagnosis, 601 patients had positive V/P(SPECT) (99%). In 1177 patients without final PE diagnosis 1153 patients had negative V/P(SPECT) (98%).

Holistic interpretation of V/P(SPECT,) yields high negative and positive predictive values and only 1% of nondiagnostic findings and was feasible in 99% of patients. It is a responsibility and a challenge of nuclear medicine to provide optimal care of patients with suspected PE by making V/P(SPECT) available.

Diagnosis of deep vein thrombosis (DVT) and pulmonary embolism (PE) is an important medical problem because of the high fatality rate from PE and the large number of cases not diagnosed before causing death. Over the last decade, there has been considerable research into the diagnostic process. It is widely accepted that venous ultrasound imaging is an accurate test for the diagnosis of DVT and is the imaging test of choice. For PE, computer tomographic pulmonary angiography (CTPA) is replacing ventilation perfusion lung scanning. Technology for CTPA is rapidly evolving and multi-row detector scans have quite reasonable sensitivity and specificity. Despite the accuracy of imaging tests, the post-test probability of disease is highly dependent on pretest probability. Clinical evaluation tools have developed that enable us to accurately categorize patients' risk prior to diagnostic imaging. One advantage of this characterization is an ability to exclude the diagnosis of DVT or PE if clinical probability is sufficiently low and when the D-dimer is negative. There are now a number of D-dimer assays that have well-defined specificities and sensitivities, which enable use in conjunction with clinical probability. A careful combination of clinical assessment, D-dimer and imaging enables safe PE rule out protocols without imaging, an ability to suspect false positive imaging results, and more accurate determination of true positive imaging. These integration strategies result in safer, more convenient and cost-effective care for patients.

This is a retrospective study to determine the accuracy and safety of a negative CT pulmonary angiogram (CTPA) based on clinical outcome and to determine the usefulness of a negative D-dimer assay before CTPA. A total of 483 patients with a negative CTPA study were followed up for 3 months, with the aim of detecting episodes of venous thromboembolism and mortality. Three hundred and forty-nine patients had an immunochromatographic D-dimer assay called 'Simplify', carried out before a CTPA examination. Seventy-eight patients had a negative D-dimer assay and a negative CTPA. Three patients had a negative D-dimer assay and a positive CTPA. All three patients had a moderate pretest clinical probability. Of the 483 patients who had a negative CTPA and a 3-month follow up, 444 (92%) were alive and 39 (8%) had died. Of the 444 patients who were alive, none had any further suspected episode of thromboembolism or had received anticoagulation therapy within the follow-up period. Of those who died, none of the deaths was thought to be as a result of pulmonary embolism (PE). Single-detector helical CT can be used safely as the primary diagnostic test to evaluate PE. Negative Simplify D-dimer assay and low pretest clinical probability exclude CTPA-detectable PE, and a CTPA is unnecessary in this cohort of patients.

The study was prospectively designed to assess the correlation between a new clinical model empirically developed for acute pulmonary embolism (PE) and ventilation/perfusion (V/P) scan results. One hundred sixty consecutive patients with suspected acute PE underwent clinical evaluation before V/P scintigraphy. The clinical probability of PE was categorized according to a structured clinical model empirically developed as low, intermediate, or high, and the results were compared with those of V/P scintigraphy. Forty, 61, and 59 patients were classified as low, intermediate, and high clinical probability, respectively. Seventy-five percent (30/40) of the patients with low clinical probability were also of low scintigraphic probability or had a normal result (r(s): 0.39, p=0.000); 28% (17/61) of the patients with intermediate clinical probability demonstrated intermediate scintigraphic probability (r(s): 0.20, p=0.012); and 68% (40/59) of the patients with high clinical probability were also of high scintigraphic probability (r(s): 0.43, p=0.000). Overall, the correlation of two scoring systems was statistically significant (r(s): 0.39, p=0.000). Unilateral leg swelling (p=0.027), syncope or near syncope (p=0.002), amputation of a hilar artery (p=0.007), and electrocardiographic signs of right ventricular overload (p=0.000) prevailed in patients with high scintigraphic probability. "Syncope-near syncope or hemodynamic collapse" PLUS "electrocardiographic signs of right ventricular overload or hypoxemia" combination had the most significant correlation with a high scintigraphic probability (r(s): 0.31; p=0.000). In conclusion, the new clinical model empirically developed was significantly successful to provide comparable results with V/P scan. This consistency was particularly prominent in patients with low or high clinical probability for PE.

To illustrate the effects of different methods for handling missing data--complete case analysis, missing-indicator method, single imputation of unconditional and conditional mean, and multiple imputation (MI)--in the context of multivariable diagnostic research aiming to identify potential predictors (test results) that independently contribute to the prediction of disease presence or absence.

We used data from 398 subjects from a prospective study on the diagnosis of pulmonary embolism. Various diagnostic predictors or tests had (varying percentages of) missing values. Per method of handling these missing values, we fitted a diagnostic prediction model using multivariable logistic regression analysis.

The receiver operating characteristic curve area for all diagnostic models was above 0.75. The predictors in the final models based on the complete case analysis, and after using the missing-indicator method, were very different compared to the other models. The models based on MI did not differ much from the models derived after using single conditional and unconditional mean imputation.

In multivariable diagnostic research complete case analysis and the use of the missing-indicator method should be avoided, even when data are missing completely at random. MI methods are known to be superior to single imputation methods. For our example study, the single imputation methods performed equally well, but this was most likely because of the low overall number of missing values.

We followed prospectively 834 consecutive patients (70% inpatients), evaluated for suspected pulmonary embolism, for a median time of 2.1 years (range, 0-4.8 yr), and compared the survival rates in patients with proven pulmonary embolism (n=320) with those without (n=514). In multivariate analysis, we modeled the probability of surviving in patients with pulmonary embolism as a function of the extent of pulmonary vascular obstruction at baseline. Among patients with pulmonary embolism, a scintigraphic follow-up was pursued to assess the restoration of pulmonary perfusion over a 1-year period. We found that massive pulmonary embolism (vascular obstruction>or=50%) is a risk factor for mortality within the first few days after onset but, subsequently, has no significant effect on survival. The adjusted risk of death in patients with massive pulmonary embolism was 8-fold higher than in patients without embolism within the first day after the incident event. By contrast, the adjusted risk of death for patients with minor or moderate pulmonary embolism (vascular obstruction<50%) was no higher than in patients without embolism at any time after onset. Most of the patients who survived a year after pulmonary embolism showed a nearly complete restoration of pulmonary perfusion with a considerable improvement in arterial oxygenation. Four (1%) of the 320 patients with pulmonary embolism at presentation developed chronic thromboembolic pulmonary hypertension. These patients featured persistent large perfusion defects in sequential lung scans. Pulmonary embolism with vascular obstruction>or=50% is a strong, independent predictor of reduced short-term survival. This underscores the need for a prompt diagnosis of the disease. Monitoring the resolution of pulmonary embolism by lung scanning may prove useful in identifying patients with persistent perfusion abnormalities who may be at risk of chronic thromboembolic pulmonary hypertension.

Over the last decade, contrast-enhanced spiral CT has been established as a non-invasive alternative to catheter angiography and is now regarded as the first-line imaging investigation for the diagnosis of pulmonary embolism (PE). The reported sensitivities for the diagnosis of PE of spiral CT vary from 45 to 100% and the specificities vary from 78 to 100%. Prospective outcome studies have shown a high negative predictive value for a single-detector spiral CT for PE. Patients' outcomes were not adversely affected in these studies when anticoagulation was withheld after a negative CT pulmonary angiogram. The main limitation of single-detector spiral CT has been its limited ability to detect isolated subsegmental PE. However, multidetector spiral CT allows evaluation of pulmonary vessels down to sixth-order branches and significantly increases the rate of detection of PE in segmental and subsegmental levels. The interobserver correlations for diagnosis of subsegmental PE with multidetector spiral CT exceed the reproducibility of selective pulmonary angiography. If appropriate equipment is available (multidetector CT), then CT pulmonary angiogram is safe to be used as the first-line imaging investigation for the diagnosis of PE.

Epidemiologic studies commonly estimate associations between predictors (risk factors) and outcome. Most software automatically exclude subjects with missing values. This commonly causes bias because missing values seldom occur completely at random (MCAR) but rather selectively based on other (observed) variables, missing at random (MAR). Multiple imputation (MI) of missing predictor values using all observed information including outcome is advocated to deal with selective missing values. This seems a self-fulfilling prophecy.

We tested this hypothesis using data from a study on diagnosis of pulmonary embolism. We selected five predictors of pulmonary embolism without missing values. Their regression coefficients and standard errors (SEs) estimated from the original sample were considered as "true" values. We assigned missing values to these predictors--both MCAR and MAR--and repeated this 1,000 times using simulations. Per simulation we multiple imputed the missing values without and with the outcome, and compared the regression coefficients and SEs to the truth.

Regression coefficients based on MI including outcome were close to the truth. MI without outcome yielded very biased--underestimated--coefficients. SEs and coverage of the 90% confidence intervals were not different between MI with and without outcome. Results were the same for MCAR and MAR.

For all types of missing values, imputation of missing predictor values using the outcome is preferred over imputation without outcome and is no self-fulfilling prophecy.

This review summarizes recent information about the diagnosis of deep venous thrombosis (DVT) and pulmonary embolism (PE) using noninvasive imaging tests, clinical assessment, and D-dimer assays, and describes how these tests can be employed in diagnostic testing algorithms for the investigation of patients with suspected DVT and PE. The clinical diagnosis of deep venous thrombosis is unreliable, but clinical prediction rules based on signs and symptoms do facilitate the categorization of patients into high, low, or medium risk categories. High sensitivity D-dimer assays further help in excluding cases but do not help in ruling in venous thromboembolism. D-dimer assays and clinical prediction rules also help in the diagnosis of pulmonary embolism. These assessments, along with objective imaging studies such as compression ultrasonography for DVT or computerized tomographic pulmonary angiograms for PE can be used in a systematic way to reliably rule in or exclude venous thromboembolism.

While the number of patients with suspected venous thromboembolism (VTE) referred to hospital emergency units increases, the proportion in whom the diagnosis can be confirmed is decreasing. A more efficient but safe diagnostic strategy is needed.

To evaluate the safety of withholding anticoagulant therapy in patients suspected of VTE based on a diagnostic work-up that combines a clinical decision rule (CDR) with a D-dimer test result without performing additional diagnostic tests.

We searched Medline (January 1996-December 2004)-related articles and reference lists of studies in English for prospective clinical studies that managed consecutive patients suspected of VTE and used a D-dimer assay combined with an explicit CDR or implicit clinical judgment.

We identified 11 studies in which 6837 consecutive outpatients suspected of VTE were included. In the combined management studies, the overall rate of thromboembolic events was nine out of 2056 patients (0.44 %, 95% CI 0.2%-0.83%) in whom anticoagulants were withheld based on the D-dimer result and a low clinical score. Similar results were obtained with qualitative and quantitative D-dimer tests and with different decision rules. The rate of exclusion varied between 30% and 50% and was highest with a low incidence of VTE among those referred.

Withholding anticoagulant treatment in patients suspected of VTE on the basis of a work-up consisting of a low clinical probability combined with either a qualitative or quantitative D-dimer test result is safe.

There is growing interest in developing prediction models. The accuracy of such models when applied in new patient samples is commonly lower than estimated from the development sample. This may be because of differences between the samples and/or because the developed model was overfitted (too optimistic). Various methods, including bootstrapping techniques exist for afterwards shrinking the regression coefficients and the model's discrimination and calibration for overoptimism. Penalized maximum likelihood estimation (PMLE) is a more rigorous method because adjustment for overfitting is directly built into the model development, instead of relying on shrinkage afterwards. PMLE has been described mainly in the statistical literature and is rarely applied to empirical data. Using empirical data, we illustrate the use of PMLE to develop a prediction model.

The accuracy of the final PMLE model will be contrasted with the final models derived by ordinary stepwise logistic regression without and with shrinkage afterwards. The potential advantages and disadvantages of PMLE over the other two strategies are discussed.

PMLE leads to smaller prediction errors, provides for model reduction to a user-defined degree, and may differently shrink each predictor for overoptimism without sacrificing much discriminative accuracy of the model.

PMLE is an easily applicable and promising method to directly adjust clinical prediction models for overoptimism.

Objective testing for pulmonary embolism is necessary, because clinical assessment alone is unreliable and the consequences of misdiagnosis are serious. No single test has ideal properties (100% sensitivity and specificity, no risk, low cost). Pulmonary angiography is regarded as the final arbiter but is ill suited for diagnosing a disease present in only a third of patients in whom it is suspected. Some tests are good for confirmation and some for exclusion of embolism; others are able to do both but are often non-diagnostic. For optimal efficiency, choice of the initial test should be guided by clinical assessment of the likelihood of embolism and by patient characteristics that may influence test accuracy. Standardised clinical estimates can be used to give a pre-test probability to assess, after appropriate objective testing, the post-test probability of embolism. Multidetector computed tomography can replace both scintigraphy and angiography for the exclusion and diagnosis of this disease and should now be considered the central imaging investigation in suspected pulmonary embolism.

Clinicians often deviate from the recommended algorithm for the diagnosis of pulmonary embolism consisting of ventilation-perfusion scintigraphy and pulmonary angiography.

To assess the safety and feasibility of a diagnostic algorithm which reduces the need for lung scintigraphy and avoids the use of angiography.

Consecutive patients with a clinical suspicion of pulmonary embolism were prospectively investigated according to an algorithm in which the diagnosis of pulmonary embolism was excluded after a low clinical probability estimate and a normal d-dimer test result, a normal perfusion scintigraphy result, or a non-high probability scintigraphy result in combination with normal serial ultrasonography of the legs. In these patients anticoagulant treatment was withheld and they were followed up for 3 months to record possible thromboembolic events. During the study period, 923 consecutive patients were seen, of whom 292 were excluded because of predefined criteria.

Of the 631 included patients, the diagnosis was refuted on the basis of a low clinical probability estimate and a normal d-dimer test result (95 patients), normal perfusion scintigraphy (161 patients) and non-high probability lung scintigraphy followed by normal serial ultrasonography (210 patients). Of these 466 patients, venous thromboembolic complications during follow-up occurred in six (complication rate 1.3%, 95% confidence interval 0.5, 2.8). The diagnostic protocol was completed in 92% of all included patients.

The diagnosis of pulmonary embolism can be safely ruled out by a non-invasive algorithm consisting of d-dimer testing combined with a clinical probability estimate, lung scintigraphy, or serial ultrasonography of the legs (in case of non-diagnostic lung scintigraphy).

Genetic programming is a search method that can be used to solve complex associations between large numbers of variables. It has been used, for example, for myoelectrical signal recognition, but its value for medical prediction as in diagnostic and prognostic settings, has not been documented.

We compared genetic programming and the commonly used logistic regression technique in the development of a prediction model using empirical data from a study on diagnosis of pulmonary embolism. Using part (67%) of the data, we developed and internally validated (using bootstrapping techniques) a diagnostic prediction model by genetic programming and by logistic regression, and compared both on their predictive ability in the remaining data (validation set).

In the validation set, the area under the ROC curve of the genetic programming model was significantly larger (0.73; 95%CI: 0.64-0.82) than that of the logistic regression model (0.68; 0.59-0.77). The calibration of both models was similar, indicating a similar amount of overoptimism.

Although the interpretation of a genetic programming model is less intuitive and this is the first empirical study quantifying its value for medical prediction, genetic programming seems a promising technique to develop prediction rules for diagnostic and prognostic purposes.

Pulmonary embolism (PE) is a major health concern that affects approximately 600,000 new patients annually. The diagnosis of PE can be difficult to make, and several imaging studies have been developed to aid in this process. Initial evaluation involves the acquisition of a chest radiograph. Findings on radiography, however, are often non-specific. The gold-standard study historically has been pulmonary angiography, with increasing diagnostic yield since the implementation of digital subtraction technology. This is an invasive procedure, however, but the incidence of major complications is low. Less invasive modalities have been developed and include ventilation-perfusion lung scans. These are used as one of the initial screening tests in evaluation of patients with suspected PE. The presence of a high-probability scan usually indicates the presence of a PE, although few patients have high probability scans. The test is significantly affected by underlying pulmonary disease or previous PE. Given this, ventilation-perfusion lung scans are limited as a primary diagnostic tool in the evaluation of suspected PE. Helical computed tomography (CT) is currently under much scrutiny as a diagnostic tool for PE. Currently a prospective, multicenter trial evaluating its efficacy (PIOPED II) has been initiated, but the results are pending. Preliminary reports suggest the helical CT and venous phase CT may become a first line study in patient evaluation. The diagnosis of PE is challenging and several imaging modalities are currently used to assist the clinician. Currently, multiple modalities are often required to make the diagnosis. With the advent of new technology and improved imaging techniques, the diagnosis of PE will become easier.

The assessment of pretest probability (PTP), with stratification into low-, intermediate- and high-risk groups is an essential initial step in the current diagnostic management of patients with suspected venous thromboembolism (VTE). In combination with additional information, it reduces the need for initial and supplementary imaging, and allows considerable refinement of the posterior probability of VTE following non-invasive imaging. PTP may be assessed either empirically or by using various decision rules or scoring systems, the best known of which are the simplified Wells scores for suspected deep vein thrombosis (DVT) and pulmonary embolism (PE), and the Geneva score for suspected PE. Each of these approaches shows similar directional and categorical accuracy, and has been validated as facilitating clinically useful classification of the PTP, although an overview of data suggests that fewer patients tend to be classified as low PTP when assessed empirically. This group is the most important to identify, as several outcome studies have shown that imaging and treatment are safely obviated in outpatients with suspected DVT or PE who have a low PTP in combination with negative d-dimer testing, a subgroup accounting for up to half of all patients studied. Hence, while probably not of critical importance, the explicit approach offered by scoring systems might be preferred over empirical assessment, particularly when used by more junior staff.

The diagnosis of pulmonary embolism is challenging because lung scanning is nondiagnostic in most patients and because pulmonary angiography is invasive. Advancements in computed tomography and magnetic resonance imaging have enabled visualization of pulmonary emboli. Studies using each of these modalities have demonstrated high sensitivity for lobar and segmental emboli and lower sensitivity for subsegmental emboli. A recent study of a management approach using spiral computed tomography combined with testing for deep vein thrombosis using compression ultrasonography was found to have high clinical validity. A number of different magnetic resonance imaging techniques have been used to identify pulmonary emboli, and no single technique has been shown to be superior. Further studies are needed to delineate the role of magnetic resonance imaging in the management of patients with suspected pulmonary embolism.

The diagnosis of pulmonary embolism (PE) is difficult with many patients treated without the disease or left untreated without an adequate diagnostic work up. Recent advances in PE diagnosis are reviewed. The use of risk stratification in PE diagnosis is strongly recommended and evidence on how it can best be performed summarized. The Ginsberg/Wells stratification rule is recommended currently as the best validated rule. Computed tomographic pulmonary angiography (CTPA) was found to have quite poor sensitivity and to be poorly validated. It is recommended as adequate as a positive test in moderate/high risk groups and an exclusionary test in low risk groups or where an adequate alternative diagnosis is found. For D-Dimer tests the only test with adequate sensitivity and validation in management studies is the VIDASCopyright D-Dimer. This is in low/intermediate risk groups in the ED population. The Simpli-RedCopyright test is also reviewed but is too insensitive for most populations. Echocardiography: this is good in compromised patients as it is a bedside test which when negative virtually excludes PE. If positive in the right setting it has a high positive predictive value. A negative echocardiogram predicts a benign clinical course for PE. The rest of the paper details the authors approach to integrating these new techniques with established algorithms and where progress is likely to occur in the next few years. These include improvements in CTPA (plus the addition of CT venography), new point of care D-Dimer tests, better risk stratification rules and integration of new strategies with artificial neural networks or computerized guidelines.

Recent advances in the management of patients with suspected pulmonary embolism have improved diagnostic accuracy and made management algorithms safer and more accessible. Ongoing clinical trials are evaluating whether these diagnostic processes can be made even simpler and less expensive. It is now possible to identify low-risk patients with suspected pulmonary embolism in whom imaging procedures can be avoided altogether.

To develop a structured model to predict the clinical probability of pulmonary embolism.

We studied 1,100 consecutive patients with suspected pulmonary embolism in whom a definite diagnosis had been established. We used logistic regression analysis to estimate the probability of pulmonary embolism based on patients' clinical characteristics; the probability was categorized as low (< or = 10%), intermediate (>10%, < or = 50%), moderately high (>50%, < or = 90%), or high (>90%).

The overall prevalence of pulmonary embolism was 40% (n = 440). Ten characteristics were associated with an increased risk of pulmonary embolism (male sex, older age, history of thrombophlebitis, sudden-onset dyspnea, chest pain, hemoptysis, electrocardiographic signs of acute right ventricular overload, radiographic signs of oligemia, amputation of the hilar artery, and pulmonary consolidation suggestive of infarction), and five were associated with a decreased risk (prior cardiovascular or pulmonary disease, high fever, pulmonary consolidation other than infarction, and pulmonary edema on the chest radiograph). With this model, 432 patients (39%) were rated a low probability, of whom 19 (4%) had pulmonary embolism; 283 (26%) were rated an intermediate probability, of whom 62 (22%) had pulmonary embolism; 72 (7%) were rated a moderately high probability, of whom 53 (74%) had pulmonary embolism; and 313 (28%) were rated a high probability, of whom 306 (98%) had pulmonary embolism.

This prediction model may be useful for estimating the probability of pulmonary embolism before obtaining definitive test results.

This clinical policy focuses on critical issues in the evaluation and management of patients with signs or symptoms of pulmonary embolism (PE). A MEDLINE search for clinical trials published from January 1995 through April 2001 was performed using the key words "pulmonary embolus" with limits of "clinical investigations" and "clinical policies." Subcommittee members and expert peer reviewers also supplied articles with direct bearing on the policy. This policy focuses on 2 major areas of current interest and/or controversy: (1) diagnostic: utility of D -dimer, ventilation-perfusion scanning, and spiral computed tomography angiogram in the evaluation of PE; and (2) therapeutic: indications for fibrinolytic therapy. Recommendations for patient management are provided for each 1 of these topics based on strength of evidence (Level A, B, or C). Level A recommendations represent patient management principles that reflect a high degree of clinical certainty; Level B recommendations represent patient management principles that reflect moderate clinical certainty; and Level C recommendations represent other patient management strategies based on preliminary, inconclusive, or conflicting evidence, or based on panel consensus. This guideline is intended for physicians working in emergency departments or chest pain evaluation units.

The purpose of this study was to assess the clinical follow-up of a negative spiral CT (SCT) angiographic finding after a suspicion of acute pulmonary embolism (PE) in a population of inpatients with cardiac and/or respiratory disease. In this high-risk population, clinical findings suggestive of PE are frequently misleading.

One hundred seventy-five consecutive patients hospitalized in cardiac and pneumology wards underwent SCT angiography for suspected PE over a 30-month period. Angiographic findings were positive in one third. For the 117 patients with negative SCT angiographic findings, a clinical follow-up during a minimum of 6 months was assessed, particularly in relation to recurrent thromboembolism, mortality, and cause of death.

The mean +/- SD follow-up was 21 +/- 11.5 months, and five patients were unavailable for follow-up. Of the 117 patients with negative findings, 81 patients did not receive anticoagulant therapy and 46 patients received anticoagulation for cardiac disease or deep venous thrombosis. Twenty-two patients died during the follow-up period, 3 of them during the first 3 months following the initial event from an undetermined cause. In patients still alive, a new PE occurred in two cases. Patients with a poor cardiopulmonary reserve did not present any recurrent events. In this population, tests other than imaging (d-dimers, cardiac echocardiography, or venous ultrasound) contributed little to eliminate the diagnosis of PE.

Whether or not early deaths are considered or not to be related to a recurrent PE, the rate of recurrence after a negative SCT angiographic finding varied between 1.8% and 4.9%. SCT angiography can be used confidently to rule out significant PE, and may prevent further investigations and unnecessary treatment in an inpatient population with cardiac and/or respiratory diseases.

The diagnosis of deep vein thrombosis (DVT) and pulmonary embolism (PE) has been improved and simplified over the past decade thanks to advances in noninvasive and readily accessible technology. With high degrees of sensitivity and specificity, venous ultrasonography is favored as the initial investigation for DVT. To diagnose PE, most clinicians rely on diagnostic algorithms that combine clinical assessment, noninvasive lung studies, and, if necessary, venous ultrasonography of the legs and D-dimer testing. Substantial progress has also occurred in the treatment of acute venous thromboembolism with the introduction of low-molecular-weight heparins. This class of antithrombotic agents has changed initial therapy from an inpatient, intravenous regimen that required laborious monitoring to an outpatient practice using weight-adjusted doses of once-daily subcutaneous injections. In addition, several new anticoagulants with theoretical advantages over existing agents have entered phase III studies. Aspects of thrombosis treatment that remain controversial include vena caval interruption and the indications for thrombolysis and surgical thromboembolectomy.

Echocardiography is advocated by some as a useful diagnostic test for patients with suspected pulmonary embolism (PE), but its diagnostic accuracy is unknown. The present study was undertaken to determine prospectively the sensitivity and specificity of transthoracic echocardiography in the diagnosis of PE.

We examined 110 consecutive patients with suspected PE. The study protocol included assessment of clinical probability, echocardiography, and perfusion lung scanning. Pulmonary angiography was performed in all patients with abnormal scans. As echocardiographic criteria to diagnose acute PE, we used the presence of any two of the following: right ventricular (RV) hypokinesis, RV end-diastolic diameter >27 mm (without RV wall hypertrophy), or tricuspid regurgitation velocity >2.7 m/sec. Clinical estimates of PE served as pretest probabilities in calculating, after echocardiography, the posttest probabilities of PE.

Pulmonary angiography confirmed PE in 43 (39%) of 110 patients. Echocardiographic diagnostic criteria for PE yielded a sensitivity of 56% and a specificity of 90%. For pretest probabilities of 10%, 50%, and 90%, the posttest probabilities of PE conditioned by a positive echocardiogram were 38%, 85%, and 98%, respectively. The posttest probabilities of PE conditioned by a negative echocardiogram were 5%, 33%, and 81%, respectively.

In unselected patients with suspected PE, transthoracic echocardiography fails to identify some 50% of patients with angiographically proven PE. Although echocardiographic findings of RV strain, paired with a high clinical likelihood, support a diagnosis of PE, the transthoracic echocardiography has to have a better sensitivity to be used as a screening test to rule out PE.

To assess the diagnostic value of lung scintigraphy and helical computed tomography (hCT) in patients with suspected pulmonary embolism (PE), all English-language articles that described lung scintigraphy and hCT in patients with suspected PE were retrieved. Articles were assessed for strength of methodology, based on nine a priori-defined criteria. Parameters of diagnostic accuracy and results of management studies were calculated and evaluated. Lung scintigraphy is diagnostic in approximately 50% of patients with suspected PE. A normal perfusion scan has a chance of recurrent PE in two of 693 patients (0.3%; 95% CI: 0.2-0.4%; fatal in 0.15%). A high-probability lung scan is correlated with angiographically proven PE in 308 of 350 patients (88%; 95% CI: 84-91%). Pulmonary embolism was proven in 385 of 1529 patients (25%; 95% CI: 24-28%) with a nondiagnostic lung scan. Helical CT studies were compared with angiography and lung scintigraphy in 1171 patients, with a prevalence of PE of 39%. The sensitivity and specificity of hCT was 283/320 (88%; 95% CI: 83-91%) and 374/408 (92%; 95% CI: 89-94%), respectively. Only one prospective management study using hCT was available. In patients in whom anticoagulants were withheld based on a normal hCT study, recurrent thromboembolic events occurred in six of 109 patients (5.5%; 95% CI: 2-12%), with one fatality (1%; 95% CI: 0.02-4.3%). Lung scintigraphy is evaluated extensively and yields a diagnostic result in 50% of patients. Helical CT has similar positive predictive value to a high-probability lung scan. However, the exact role of hCT in the management of patients with suspected PE needs to be determined in prospective studies.

We evaluated the utility of venous duplex ultrasonography (VDUS) of the lower extremities in patients with pulmonary embolism (PE) and studied the distribution of venous thrombi in deep vein thrombosis (DVT) patients with and without PE.

We retrospectively reviewed medical records of all inpatients with a final diagnosis of PE or DVT between 1989 and 2000.

Venous thrombosis was detected by VDUS in 229 patients (191 without PE and 38 with PE). The left leg only was involved in 50% of patients (p < 0.05), the right leg only in 33%, and both legs in 17%. The overall distribution of veins affected by DVT was: popliteal vein, 77%; superficial femoral vein, 76%; common femoral vein, 65%; posterior tibial vein, 23%; external iliac vein, 21%; common iliac vein, 9%; great saphenous vein, 7%; and inferior vena cava, 2%. A single venous site was involved in 22% of patients. External iliac vein thrombosis was more frequent in patients with DVT only (24%) than with PE and DVT (5%) (p < 0.05). The venous obstruction was partial in 14% of patients. VDUS of the asymptomatic leg was positive in 14% of patients with unilateral symptoms of DVT, all of whom also had DVT in the symptomatic leg. VDUS was positive for DVT in 90% of patients with PE and concomitant pain or edema of the leg, compared to only 20% of PE patients with no symptoms of DVT.

Sonography should be the first diagnostic test for patients suspected of having PE with any sign or symptom related to DVT. VDUS of the asymptomatic leg is unnecessary in the diagnosis and management of DVT. Omitting the superficial femoral vein examination would lead to some decrease in the sensitivity of VDUS.

We conducted a study to evaluate a noninvasive strategy including spiral computed tomography (CT) in patients with suspected pulmonary embolism (PE). We systematically performed spiral CT, ventilation/perfusion lung scanning, and D-dimer (DD) measurement (VIDAS test), and in some cases (with a normal CT with nondiagnostic lung scan and increased DD) performed venous ultrasonography (US) on 247 consecutive patients with clinically suspected PE in our hospital. Patients in whom PE was deemed absent were not given anticoagulants. All patients were followed for 3 mo. The prevalence of PE in the 228 patients who could be evaluated was 42% (96 of 228). PE was confirmed by spiral CT in 73% of the patients, by a high-probability lung scan in 4%, and by findings on US in 23%. PE was ruled out by a normal lung scan in 14% of the patients, by a normal DD concentration (< 500 ng/ml) in 31%, by an obvious differential diagnosis on spiral CT in 18%, by a similar prior lung scan in 11%, and by the combination of normal spiral CT findings, a nondiagnostic lung scan, a DD concentration > 500 ng/ml, and normal US in 26%. Pulmonary angiography was performed in only two patients, both of whom had a normal spiral CT scan and a high-probability lung scan, and was normal. The 3-mo risk of thromboembolism in patients not given anticoagulants, based on the results of the diagnostic protocol, was 1.7% (95% confidence interval: 1.5 to 2.3%). There were no deaths. The noninvasive strategy of combining spiral CT, lung scanning, DD measurement, and in some cases US, in patients with suspected PE yielded a definite diagnosis in 99% of patients, and appeared to be safe.

To determine whether a helical computed tomographic (CT) scan that is negative for pulmonary embolism (PE) is a sufficiently reliable criterion to safely withhold anticoagulation therapy.

Patients with negative helical CT scans were prospectively compared with patients with negative or low-probability scintigrams. In a 460-bed university hospital and clinic, 1,015 adult patients underwent either scintigraphy or helical CT for possible PE for 25 months. Five hundred forty-eight patients who had negative images and were not receiving anticoagulation therapy were prospectively followed up for 3 months for clinical, new imaging, death certificate, or autopsy evidence of subsequent PE. Ninety-seven patients were lost to follow-up.

Subsequent PE was found in two (1.0%) of 198 patients with negative CT scans, none of 188 patients with negative ventilation-perfusion (V-P) scans, and five (3.1%) of 162 patients with low-probability V-P scans (not statistically significant). Patients in the helical CT group were hospitalized more often, had more severe disease, had more substantial PE risk factors, and had a higher death rate. No deaths were attributed to PE in either group.

The frequency of clinical diagnoses of PE after a negative CT scan was low and similar to that after a negative or low-probability V-P scan. Helical CT is a reliable imaging tool for excluding clinically important PE.

Venous thromboembolic disease has emerged as a significant cause of morbidity and mortality in hospitalized patients. This article reviews the salient features of venous thromboembolism as they pertain to the critically ill. Emphasis is placed on identifying risk factors, diagnostic strategies, prophylaxis, and treatment of this disorder. Deep venous thrombosis and pulmonary embolism, both being manifestations of the same disease processes, are considered together in this discussion of venous thromboembolism.

The accuracy of diagnostic methods for the diagnosis of deep vein thrombosis and pulmonary embolism in symptomatic patients is critically reviewed. In addition, the safety of withholding anticoagulant therapy from patients with suspected deep vein thrombosis or pulmonary embolism in whom the qualified diagnostic strategy was normal is evaluated by determining the frequency of venous thromboembolic complications during 3 months of follow-up. It is shown that the currently used available diagnostic techniques for deep vein thrombosis are all able to identify the majority of patients who indeed have venous thrombosis. However, as result of its accuracy and practical advantages, compression ultrasound is the test of choice in the evaluation of symptomatic patients. Patients with a normal test outcome should be re-tested to detect the small proportion of patients with proximally extending calf vein thrombosis. In the strategy of repeated diagnostic testing, impedance plethysmography could be used as an alternative to ultrasonography. To obtain a reduction in repeat tests various diagnostic strategies have been evaluated and it was shown that these strategies, using non-invasive tests, can be as accurate and safe as the invasive reference strategy. The safeties of the various strategies were very similar; however, important differences were observed with respect to the practical implementation of the various diagnostic strategies. Simplification of the repeated testing strategy by using a D-dimer assay and/or a clinical decision rule seems to be promising. The reference standard for the diagnosis of pulmonary embolism remains pulmonary angiography. Several strategies based on non-invasive diagnostic methods have been evaluated for their safety and complexability. Perfusion-ventilation lung scanning is the most thoroughly evaluated non-invasive technique so far. It seems safe to withhold anticoagulant therapy in patients suspected of pulmonary embolism with a normal perfusion lung scan result; however, further testing is needed in the case of a non-diagnostic perfusion-ventilation lung scan result. At this moment angiography is the method of choice in this category of patients. D-dimer assays, clinical decision rules and ultrasound examinations of the legs seem to have a high potential to limit the need for angiography. Also, spiral computerized tomography and magnetic resonance imaging are promising techniques, but their role in the diagnostic management of pulmonary embolism is still uncertain.

Diagnostic tests are often evaluated by comparison of the areas under receiver operating characteristic (ROC) curves. In this study the authors compared this approach with a more direct method that takes into account consequences of a diagnosis. Data from a prospective study of diagnosis of pulmonary embolism were used for a motivating example. Using multivariable logistic regression analysis, three diagnostic models were built and compared based on their ROC curves. Although model 1 (0.706) and model 2 (0.702) had the same ROC-curve area, they performed differently when risks and benefits of subsequent decisions were considered by applying the treatment probability threshold. Models 1 and 3 (0.611) had substantially different ROC-curve areas but performed similarly taking into account the therapeutic consequences. This demonstrates that comparison of diagnostic tests using the areas under the ROC curves may lead to erroneous conclusions about therapeutic usefulness. To correspond to daily practice, it would be more appropriate to also consider the clinical implications in evaluating diagnostic tests. This is made feasible by explicit definition and application of a treatment threshold.