1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

ACADEMIC EMERGENCY MEDICINE 2011; 18:22–31 © 2011 by the Society for Academic Emergency Medicine


Background:  Computed tomography angiograms (CTAs) for patients with suspected pulmonary embolism (PE) are being ordered with increasing frequency from the emergency department (ED). Strategies are needed to safely decrease the utilization of CTs to control rising health care costs and minimize the associated risks of anaphylaxis, contrast-induced nephropathy, and radiation-induced carcinogenesis. The use of compression ultrasonography (US) to identify deep vein thromboses (DVTs) in hemodynamically stable patients with signs and symptoms suggestive of PE is highly specific for the diagnosis of PE and may represent a cost-effective alternative to CT imaging.

Objectives:  The objective was to analyze the cost-effectiveness of a selective CT strategy incorporating the use of compression US to diagnose and treat DVT in patients with a high pretest probability of PE.

Methods:  The authors constructed a decision analytic model to evaluate the scenario of an otherwise healthy 59-year-old female in whom PE was being considered as a diagnosis. Two strategies were used. The selective CT strategy began with a screening compression US. Negative studies were followed up with a CTA, while patients with positive studies identifying a DVT were treated as though they had a PE and were anticoagulated. The universal CT strategy used CTA as the initial test, and anticoagulation was based on the CT result. Costs were estimated from the 2009 Medicare data for hospital reimbursement, and professional fees were obtained from the 2009 National Physician Fee Schedule. Clinical probabilities were obtained from existing published data, and sensitivity analyses were performed across plausible ranges for all clinical variables.

Results:  In the base case, the selective CT strategy cost $1,457.70 less than the universal CT strategy and resulted in a gain of 0.0213 quality-adjusted life-years (QALYs). Sensitivity analyses confirm that the selective CT strategy is dominant above both a pretest probability for PE of 8.3% and a compression US specificity of 87.4%.

Conclusions:  A selective CT strategy using compression US is cost-effective for patients provided they have a high pretest probability of PE. This may reduce the need for, and decrease the adverse events associated with, CTAs.

Pulmonary embolism (PE), while common and potentially lethal, is a treatable disease if recognized.1 The Wells Criteria, Geneva Score, and Revised Geneva Score established the concept of risk-stratifying patients with suspected PE to allow physicians to determine a patient’s pretest likelihood of having the disease based on a small number of risk factors.2–4 While the use of the D-dimer assay helped to successfully rule out PE in low-risk patients,4–7 the evaluation of higher risk patients still involves the use of either computed tomography angiography (CTA) or ventilation–perfusion (V/Q) imaging. CTA has been shown to have a higher sensitivity8,9 and has become the standard method for diagnosis of PE.6,10

The use of CT for vascular indications has increased 235% between 1991 and 2002.11 In 2006 alone, 67 million CT scans were performed in the United States, 25% to 40% of which were ordered in the emergency department (ED).12 However, recent literature has shown that there are significant risks associated with CTA, including the risks of radiation-induced carcinogenesis, contrast-induced nephropathy, and anaphylaxis.12–17

The radiation dose from CTA varies with the technology used, the imaging parameters selected, and the size of the patient, but a typical CTA results in an effective dose of approximately 8 mSv. The resultant lifetime-attributable risk of radiation-induced carcinogenesis above baseline varies depending upon sex and age of exposure. These values represent lifetime-attributable risks from all cancer types combined, most importantly breast, lung, and thyroid. Additionally, repeat CTAs are a common source of additional radiation in the ED population being evaluated for PE.18

In addition to its risks due to ionizing radiation, CTA is increasingly recognized as a source of iatrogenic nephropathy. Radiographic contrast media is the third most common cause of hospital-induced renal failure and 11% of inpatient renal failure is due to contrast-induced nephropathy.19 One recent study found that the incidence of contrast-induced nephropathy in ED patients receiving CTA was between 4 and 12%.17 Another study by the same group found that the incidence of contrast-induced nephropathy was 11% and that the likelihood of severe renal failure and death in the patients were increased by 7 and 8%, respectively.16 Anaphylaxis, while uncommon, is another complication. Two studies looking at adverse events associated with nonionic contrast found that the incidence of anaphylactic reactions ranged from 0.042% to 0.126%, and overall mortality rates ranged from 0 to 0.0051%.14,15

The use of compression ultrasound (US) to detect deep vein thrombosis (DVT) in patients with suspected PE is not a new concept and has been endorsed in recent guidelines as a method of reducing the use of CTA in appropriate patients.20,21 An early Dutch study in patients with suspected PE found that the use of compression US reduced the use of other imaging by 22%, at the expense of 2% to 4% of patients being unnecessarily treated for venous thromboembolism (VTE).22 However, this study was not performed in ED patients and was conducted prior to the routine use of CTA. A more recent study found that, among patients with either a positive D-dimer or a high pretest probability for PE by the Geneva Score, 39% were found to have a PE when a DVT was detected by compression US, giving US a 99% specificity for detecting PE in an unselected population.2

As more risks of CTA are discovered, the development of evaluation strategies designed to decrease its excessive utilization have become imperative. While the management of DVTs and PEs is similar, they are not exactly the same. This study was undertaken because the diagnosis of a DVT with concomitant signs and symptoms of PE almost certainly suggests a PE. Our objective was to determine the cost-effectiveness of the routine use of compression US to diagnose DVT in patients with suspected PE to forgo CTA and presumptively treat with heparin. Our hypothesis was that it would be cost-effective to use this selective CT strategy in otherwise healthy patients with a high pretest probability of PE.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Study Design

This was a study using decision analysis to estimate the cost-effectiveness and quality of life improvements for patients who would receive compression US before possible CTA for diagnosis of PE.

Major Model Assumptions

We constructed a decision analytic model using standard software (TreeAge 2009, Williamstown, MA) to evaluate a decision tree and perform sensitivity analyses. The base-case patient is a 59-year-old hemodynamically stable female who presents with new-onset symptoms of PE, and our model assumes a societal perspective. She does not have any history of prior VTE or known thrombophilia, nor does she have other thoracic findings necessitating inevitable CT imaging. Her age was based on the mean age of patients evaluated for PE in numerous studies.6,22–24 We assumed a pretest probability of PE in this patient of 37.5%, based on established risk stratification techniques using the Wells criteria, and ran the analysis using a range of pretest probabilities to determine for which patients the decision strategies are cost-effective. The base-case estimates, as well as the ranges of possibilities for each variable, are included in Table 1.2,3,7,8,10,13–17,22,25–43 Additionally utilities and costs are displayed in Tables 230,33,44–48 and 3,46,49,50 respectively.

Table 1.    Base-case Estimates and Range of Values Used in Sensitivity Analyses
VariableBase-case EstimateRange     Reference(s)
  1. Data are obtained from published literature to establish the base-case values. Clinically plausible ranges are used in the sensitivity analyses to evaluate their influence on the decision strategy.

  2. CIN = contrast-induced nephropathy; CTA = computed tomography angiogram; CUS = compression ultrasound; DVT = deep vein thrombosis; ICH = intracerebral hemorrhage; PE = pulmonary embolism; V/Q = ventilation–perfusion.

Pretest probability of PE (high)0.3750–0.57
Probability of anaphylaxis0.000730.00001–0.00914,15
Probability of cancer from CT0.00050.00001–0.00113
Probability of cancer from V/Q0.00010.0001–0.00125
Probability of CIN0.060.04–0.1216,17,26
Probability of death from anaphylaxis (among patients with anaphylaxis)0.00510.0001–0.0114,15
Probability of death from cancer10.1–126,27
Probability of death from CIN00–116,17,26
Probability of death from major bleeding0.0930.01–0.928
Probability of death from ICH0.450.1–126,28,29
Probability of death from PE despite anticoagulation0.010.005–0.230–34
Probability of death from untreated PE0.050–0.226,35
Probability of DVT in the setting of PE0.290.09–0.52,22
Probability of ICH if major hemorrhage0.0370–0.426,38
Probability of major hemorrhage0.020.01–0.226,28
Probability of severe renal failure from CIN0.050.04–0.1216,17,26
Probability of untreated PE0.0250–0.226,35
Sensitivity of CTA0.830.5–18
Specificity of CTA0.960.5–18
Sensitivity of CUS for DVT0.620.5–13,38,48,51,52
Specificity of CUS for DVT0.970.5–12,10,48,52–55

We assumed that the decision to obtain CTA imaging for the evaluation of PE had already been made by the clinician through the use of risk stratification, either with or without the use of a D-dimer test. D-dimer testing was not incorporated into the decision model. Two diagnostic strategies were evaluated (Figure 1): 1) a selective CT strategy—CTA only performed after a negative compression US and 2) a universal CT strategy—CTA performed on all patients, with no US. We assumed a societal perspective when determining direct and indirect medical costs. Additionally, we used a willingness-to-pay threshold of US$50,000 per quality-adjusted life-year (QALY).51 While several authors have suggested raising this level, the $50,000 per QALY threshold is the most conservative and most commonly used measure of the cost-effectiveness of an intervention.52 While the base case involves a 59-year-old female, we ran this model using 35-, 45-, 55-, 65-, and 75-year-old females in whom PE was the primary diagnostic concern, to determine the effect of age on the decision strategy.


Figure 1.  Simplified decision tree demonstrating both decision strategies (universal CT vs. selective CT) of a 59-year-old female with a high pretest probability of PE. (A) The tree is read left to right and the initial decision node is denoted with a square. Chance nodes are represented with a circle and payoff nodes represent the final outcomes (triangle) that assign QALYs and costs. (B) Anaphylaxis node. (C) Contrast induced nephropathy node. (D) Anticoagulation node. (E) Cancer Markov node. CIN = contrast-induced nephropathy; CUS = compression ultrasound; CT = computed tomography; PE = pulmonary embolism; QALY quality-adjusted life-years.

Download figure to PowerPoint

We varied the time frame of the analysis based upon the duration of morbidity. For patients with uncomplicated PEs who were anticoagulated, we used a 6-month time horizon. For patients with contrast-induced nephropathy or anaphylaxis, the time horizon was 48 hours unless the patient developed permanent renal failure requiring dialysis. Cases of intracerebral hemorrhage (ICH) occurred within the anticoagulation period, and patients’ quality of life was reduced throughout their projected life expectancy. However, cases of major bleeding assumed only a 6-month impact on quality of life. Last, radiation-induced carcinogenesis was modeled as a Markov node, which allows for the evaluation of risk when there is an ongoing probability of a disease. Because radiation permanently increases the risk of cancer, which could occur during any of the remaining years of the patients’ lives, a Markov node incorporates this ongoing risk.

Two strategies were evaluated for these patients. The selective CT strategy used compression US to evaluate for DVT in all patients. Any patient with a negative compression US had a CT performed to evaluate for PE. If compression US was positive for DVT, no further imaging was performed, and the patient was admitted to the hospital with the diagnosis of PE, anticoagulated with heparin for 6 days, and followed for 6 months of outpatient anticoagulation on 5 mg of warfarin daily. We assumed that any bleeding complications associated with anticoagulation occurred after the initial hospitalization and that every patient diagnosed with a PE would be treated with anticoagulation. Low-molecular-weight heparin is an alternative to unfractionated heparin, and sensitivity analyses were performed to account for a plausible range of associated complications between the two.

The second strategy, universal CT, involved CTA in all patients with suspected PE, with no US. Patients with positive CTAs were anticoagulated in the same fashion as the selective CT patients detailed above. Test characteristics of CTA were incorporated into the model with a sensitivity of 86% and specificity of 96%, with sensitivity analyses conducted to evaluate the impact of variation of the test characteristics on the decision strategy.8 Studies that were nondiagnostic had a V/Q scan performed. We assumed that, if both studies were performed, the subsequent V/Q would be diagnostic.

While many different malignancies can develop as a result of CTA imaging, we primarily used the costs and disutilities of non–small cell lung carcinoma, the most common type of pulmonary cancer (84.9%).27 Additionally, since non–small cell lung carcinoma is often diagnosed late in its course,53 we assumed that patients who developed non–small cell lung carcinoma were diagnosed at an advanced stage for the purposes of calculating costs, utilities, and mortality. Since these patients could have either resectable or unresectable lesions, we ran a range of sensitivity analyses around these probabilities using clinically plausible ranges.

Clinical Probabilities

We assumed that each CTA was performed with an effective radiation dose of 8 mSv. A 59-year-old woman exposed to 8 mSv has a lifetime associated risk of radiation-induced cancer of 0.05%, according to the age and sex-specific BEIR VII data set.27 Patients who had a V/Q scan were exposed to 2 mSv, with a lifetime associated risk of radiation-induced carcinoma of 0.01%.25 Based on similar studies, we assumed that all patients diagnosed at an advanced stage of malignancy would die, but ran wide sensitivity analyses to evaluate this assumption’s effect on the decision strategy. Furthermore, we used a Markov model to vary the age at which the diagnosis was made.26

The major complication associated with anticoagulating patients with PEs is hemorrhage. We separated the outcomes into three scenarios: no bleeding, ICH, and gastrointestinal (GI) bleeding. Either ICH or GI bleeding could result in survival or death.

For patients treated with anticoagulation for PE, the base-case likelihood of death from the PE was 1%. This was derived from several studies reporting the range from 0.7% to 5.3%.30–33 Studies with unselected patients had probabilities on the higher end of this range, while studies in hemodynamically stable patients treated at home for PE noted lower mortality rates.34 The base-case probability of major hemorrhage was 2%, which was obtained from studies evaluating use of anticoagulants in the treatment of PE.28,36 Subsequent rates of ICH and death were derived from these same studies.

The base-case probability of contrast-induced nephropathy was 6% based on two studies that found that the risk of nephropathy ranged from 4% to 12%.16,17 Although the first study did not find any evidence of severe renal failure or death associated with contrast-induced nephropathy, a later study by the same group found that the probabilities of renal failure and death (among the patients with renal failure) were 7 and 8%, respectively.16 However, these studies were based at single institutions and had limited sample sizes. Therefore, in our base case, we were more conservative and used a probability of contrast-induced nephropathy of 6%, a probability of severe renal failure among those who developed nephropathy of 5%, and a probability of death from severe renal failure of 0%. Sensitivity analyses were conducted to evaluate the effect of contrast-induced nephropathy and subsequent morbidity on the decision model. Lastly, we assumed the rate of anaphylaxis from CT contrast dye to be 0.07%, with an overall mortality rate of 0.00037%.14,15

To determine the rate of positive compression US in the setting of suspected PE, we first needed to identify the sensitivity and specificity of compression US for DVT, since we could find no clinical studies evaluating compression US as a “rule-in” strategy that have been conducted in ED patients. Initial sensitivity and specificity were set at 0.62 and 0.97, respectively, based on studies of asymptomatic patients screened for DVT, since many patients with PE no longer have signs of DVT.3,37 These values were taken from data evaluating compression US for the suspicion of DVT performed and read by radiologists. The criterion standard used in these studies was either serial compression US or venography. The probability of DVT, along with published sensitivity and specificity rates of compression US, were used to calculate the rate of positive findings in the setting of suspected PE.2,22 Using test characteristics for compression US, the positive predictive value is 96%, and the positive likelihood ratio is 30.3. Sensitivity analyses were conducted using clinically plausible ranges. False-positive rates were calculated as (1 – specificity). The false-negative rate of compression US was not factored into our model since all patients with negative US in the selective CT strategy underwent CTA.

Quality-of-life Measures (Disutilities)

We used QALYs as the major measure of disease burden, the standard method for cost-effectiveness analyses.54 QALYs take into account the quality of life adjusted for age and remaining life expectancy and allow for comparison between health care interventions. For this analysis, the life expectancy for our 59-year-old base-case patient was 23.8 years.45 To determine the effect of a disease state on quality of life, we first evaluated the time she would spend in that disease state. If temporary, as in the case of anaphylaxis, contrast-induced nephropathy, PE, and non-ICH hemorrhage, a disutility would be subtracted from her quality-adjusted life expectancy. However, if the patient developed a disease that could permanently affect her long-term health (untreated PE, cancer, ICH), the utility (or quality of life) drawn from prior studies evaluating specific medical conditions (Table 2) was multiplied by her life expectancy to determine her remaining QALYs.23,33,46–48 Since these QALYs occurred in the future, they were discounted using a standard 3% rate to determine their current value.49

Table 2.    Assumptions for QALY Calculations
VariableBase-case EstimateReference(s)
  1. CIN = contrast-induced nephropathy; ICH = intracerebral hemorrhage; PE = pulmonary embolism; QALY = quality-adjusted life-year.

Discount rate3%44
Life expectancy of 59-year-old female23.8 years45
Disutilitity of anaphylaxis (48-hour time horizon)0.000846
Disutilitity of CIN (48-hour time horizon)0.000847
Disutilitity of untreated PE (6-month time horizon)0.072532
Disutility of PE and treated (6-month time horizon)0.85532
Utility of ICH0.647
Utility of major bleeding episode0.8747
Utility of cancer0.6926,48,49


We used 2009 Medicare data for diagnosis-related groups (DRGs) and relative value units as a surrogate for costs (Table 3). Hospital costs were calculated using the following DRGs: 165 (ICH), 175 and 176 (PE with and without critical care), 181 and 182 (respiratory neoplasm with and without critical care), 294 and 295 (DVT with and without critical care), 377 and 379 (renal failure with and without critical care), 684 (GI hemorrhage), and 915 (allergy with critical care). Critical care DRGs were used if the patient went into cardiac arrest, but otherwise the non–critical care DRGs were used. Professional costs were based on the 2009 National Physician Fee Schedule for outpatient treatment as well as the mean length of stay for inpatient treatment.49 Current procedural terminology (CPT) code 99222 was used for hospital day 1 and CPT code 99231 was used for each subsequent complete or partial hospital day. Costs of imaging studies were derived from relative value units for bilateral lower extremity US, CT pulmonary angiography, and V/Q scanning. The values of these three were $261, $505, and $217, respectively.55

Table 3.    Costs
VariableBase-case Estimate ($)Range ($)Reference(s)
  1. CIN = contrast-induced nephropathy; CUS = compression ultrasound; ICH = Intracerebral hemorrhage; PE = pulmonary embolism; V/Q = ventilation–perfusion.

CT angiogram505.29250–75049
Death from anaphylaxis80,489.5250,000–100,00046
Death from cancer69,076.4320,000–100,00049,50
Death from CIN88,354.2050,000–118,00049
Death from major hemorrhage74,768.0654,000–94,00049,50
Death from PE77,155.1554,000–100,00049,50
ICH and survive34,193.8214,000–54,00049,50
Major hemorrhage and survive13,830.182,500–21,00049,50
PE and survive33,283.4419,000–66,00049,50
Severe renal failure24,936.2214,000–34,00049,50
V/Q scan217.48100–75049,50

Costs of anticoagulation (Table 4)23,55–57 reflected the patient being admitted for 6 days of inpatient treatment with unfractionated heparin and daily 5 mg of warfarin, ancillary costs and supplies while hospitalized (converted to 2009 dollars using the medical consumer price index), and a 6-month outpatient course of anticoagulation with 5 mg of warfarin daily. Outpatient follow-up assumed one physician office visit (CPT code 99212) at 3 months, monthly anticoagulation clinic visits for fingerstick international normalized ratio (INR) checks (CPT code 99211), and 6 months of medication costs for warfarin.57

Table 4.    Anticoagulation Expenses
 Cost ($)Reference(s)
  1. CPT = current procedural terminology; INR = international normalized ratio.

Total anticoagulation costs715.9023,56
Daily pharmacy costs7.8623
Ancillary supplies31.4323
Warfarin 5 mg (monthly supply)13.9957
Anticoagulation clinic (CPT 99211)18.1555
INR (CPT 85610)5.7455
Physician visit (CPT 99212)36.7155

Since a single inpatient visit resulted in reimbursement for only one diagnosis, we chose the most expensive diagnosis using Medicare data. However, if conditions required multiple separate hospitalizations, their costs could sum. For example, if a patient diagnosed with a PE was discharged and developed a hemorrhage as an outpatient, the cost of his or her care increased. However, a patient who developed anaphylaxis during a CT that diagnosed a PE would incur only the cost of PE admission. Patients who died were assumed to accrue Medicare charges two standard deviations beyond the mean under the assumption that they required additional care before they died.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Base-case Results

The base-case revealed that the selective CT strategy resulted in a cost savings of $1,457.70 and a life expectancy increase of 0.0213 QALYs (Table 5).

Table 5.    Base-case Analysis
 Cost ($)Effectiveness (QALY)Cost/Eff ($/QALY)Marginal Cost ($)Marginal Effectiveness (QALY)Result
  1. CT = computed tomography; QALY = quality-adjusted life-year.

Universal CT9,051.9416.80911,133.43   
Selective CT7,594.2416.83041,045.38(1,457.70)0.0213Dominant strategy

Sensitivity Analyses

All variables were analyzed to evaluate their effects on the decision strategy. A “cost-effective” strategy was one whose cost/effectiveness ratio was below the societally established willingness-to-pay threshold of $50,000 per QALY. A “dominant” strategy was one in which the costs are lower and the effectiveness was higher than the alternative strategy.

The probabilities that affected the decision strategy included: PE, death from PE following anticoagulation, major bleeding following anticoagulation, cancer, and death from cancer. Additionally, the sensitivity of CTA and specificity of compression US affected the decision strategy. Finally, costs of both PE and death from cancer, as well as the disutilities associated with the short-term morbidity of PE and untreated PE, also affected the model.

The selective CT strategy was the dominant strategy above a pretest probability of PE of 8.3% and was cost-effective at all testing thresholds (Figure 2). Similarly, the selective CT strategy was dominant above a specificity of compression US of 0.874 and was cost-effective at all testing thresholds (Figure 3). The selective CT strategy was also dominant below a probability of death from PE despite anticoagulation of 2.45% and was cost-effective from 2.45% to 3.1%. Above this threshold, the universal CT strategy was cost-effective and became dominant at 20%. For the probability of major bleeding, the selective CT strategy was dominant from 0% to 4.2% and then cost-effective until 7.3%, above which the universal CT strategy was cost-effective.


Figure 2.  Marginal cost-effectiveness ratio of selective CT strategy as a function of the probability of PE. CT = computed tomography; PE = pulmonary embolism; QALY = quality-adjusted life-year.

Download figure to PowerPoint


Figure 3.  Marginal cost-effectiveness ratio of selective CT strategy as a function of the specificity of CUS. CT = computed tomography; CUS = compression ultrasound; QALY = quality-adjusted life-year.

Download figure to PowerPoint

The selective CT strategy was the preferred strategy across all CTA sensitivities; it was dominant below a sensitivity of 0.92 and cost-effective above. Sensitivity analyses were also performed around the risks of CTA, including the probabilities of anaphylaxis, death from anaphylaxis, contrast-induced nephropathy, death from nephropathy, renal failure, and radiation-induced carcinoma. Contrast-induced nephropathy and renal failure did not affect the decision, and the selective CT strategy remained cost-effective. Likewise, anaphylaxis and death from anaphylaxis also did not affect the decision strategy. The sensitivity analysis around radiation-induced carcinoma found that the selective CT strategy was cost-effective from 0.027% to 0.037%, above which it became dominant. For mortality from cancer, the selective CT strategy was cost-effective until a rate of death of 19% and then dominant above this.

All other variables that affected the decision strategy are reflected in a tornado diagram that demonstrates thresholds for cost-effectiveness and dominance (Figure 4). Several notable sensitivity analyses did not alter the decision strategy, including age and the specificity of CTA.


Figure 4.  Tornado diagram demonstrating marginal cost-effectiveness ratio of universal CT strategy in dollars per QALY. The dotted line at $50,000 represents the willingness-to-pay threshold. Numbers to the right of this line (in white) indicate the probability of each variable at the willingness to pay threshold. Bars to the left of the dotted line represent SCTS as a cost-effective strategy. If they continue beneath zero and are indicated by an arrow, this indicates a dominant strategy. CIN = contrast-induced nephropathy; CTA = computed tomography angiography; CUS = compression ultrasound; DVT = detect deep vein thrombosis; PE = pulmonary embolism; QALY = quality-adjusted life-year.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Our decision model demonstrates that a selective CT strategy using compression US prior to CTA in the ED evaluation of patients with a high pretest probability for PE is cost-effective. In the base-case scenario, the selective CT strategy was less expensive ($1,457.70) and resulted in a higher number of QALYs (0.0213).

To test the robustness of our model, we conducted sensitivity analyses on every variable to determine their effect on the decision strategy. The probability of PE is the primary clinical driver of this model, since emergency physicians can use established risk stratification tools to determine a patient’s pretest probability of having a PE. This pretest probability can then guide the utilization of laboratory tests and imaging studies.7,58 In our model, the use of the selective CT strategy was the preferred strategy at all levels of pretest probability for PE and was dominant above 8.3%. This suggests that compression US may be a cost-effective method of reducing radiation exposure in patients with a suspected PE. Additionally, since the selective CT strategy is dominant at higher pretest probabilities for PE, adverse events (such as unnecessary anticoagulation) are less likely to occur as the pretest probability for PE increases.

Adverse events from CTA, including anaphylaxis, contrast-induced nephropathy, and possible death from either, did not affect the decision model. However, the probability of major bleeding did. While the base value used was 1%, the selective CT strategy is dominant to 4.8% and cost-effective until 7.3%. This is due primarily to false-positive compression US results, the resulting unnecessary anticoagulation, and the increased morbidity and cost of complications, particularly ICH. Calculating the rate of false-positive compression US as 1 – specificity, the selective strategy is cost-effective at all testing thresholds and dominant above a compression US specificity of 0.874. Nevertheless, a probability of major bleeding of 7.3% is fairly high, and alternative forms of heparin have been found with lower rates of hemorrhage.23

Radiation-induced carcinoma is also a major driver of this decision model. As the probability of cancer increases to 0.037%, the selective strategy becomes cost-effective and, above this, it is the dominant strategy. While cancer may take years to develop, factoring this into our decision model demonstrates that even the level of radiation received from a single CTA is enough to warrant alternative imaging strategies. In light of the fact that the selective strategy is preferred across the wide range of ages and adverse events demonstrated above, our model suggests that it may be a cost-effective alternative.

It should be noted that the selective CT strategy also lends itself to situations in which patients cannot receive a CTA, including contrast allergy, pregnancy, renal failure, or hemodynamic instability. In fact, a review of VTE diagnosis in pregnant patients recommended that a positive compression US should be managed as a PE.59

Finally, since the selective CT strategy demonstrates a willingness-to-pay threshold below $50,000 per QALY across a wide range of test characteristics (both sensitivity and specificity) for compression US in a patient population with a high pretest probability of PE, our study suggests that it may be feasible for non-US fellowship-trained emergency physicians to perform the studies themselves in this select population, especially since most EDs do not have compression US available on nights and weekends.38


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

This analysis was conducted with existing published data that represent a number of different clinical scenarios and may not reflect the clinical scenario depicted in this model, nor its rates of risks and complications, which is an inherent limitation of all cost-effectiveness analyses. For example, we did not model low pretest probability patients who may have received a D-dimer instead of imaging as the first step in their work-up. We performed sensitivity analyses to attempt to account for this. In addition, we assumed that anticoagulation with heparin was the sole initial treatment for PE. While other options such as low-molecular-weight heparin exist, we accounted for them by performing sensitivity analyses around the complication rates. For the sake of simplicity, we did not account for the possibility that a CT result could lead to alternative or incidental diagnoses, as there was no limit to the number of these diagnoses or their subsequent clinical workups.

Furthermore, this study does not look at younger patients who would develop types of cancer different from the lung malignancies in this model. However, younger patients are more likely to have a poorer quality of life given their longer life span, and our model would presumably be cost-effective in these patients as well.

It should be noted that compression US would add to the length of stay in the ED for patients with suspected PEs because most EDs do not have continual availability of compression US and require technicians to be called in after hours. This increased ED time prevents other patients from being seen, and this opportunity cost was not factored into the model due to the difficulty in quantifying it. Additionally, while emergency physicians could potentially perform the compression US, not all EDs have US readily available, and many emergency physicians are not credentialed in this diagnostic skill, limiting its availability and utility.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Using a decision analytical model, we found that a pulmonary embolism diagnostic strategy using lower extremity compression ultrasound followed by selective computed tomography angiography is cost-effective. This strategy may be a reasonable approach to reduce the incidence of adverse events associated with computed tomography angiography.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References
  • 1
    Courtney DM, Kline JA. Prospective use of a clinical decision rule to identify pulmonary embolism as likely cause of outpatient cardiac arrest. Resuscitation. 2005; 65:5764.
  • 2
    Le Gal G, Righini M, Sanchez O, et al. A positive compression ultrasonography of the lower limb veins is highly predictive of pulmonary embolism on computed tomography in suspected patients. Thromb Haemost. 2006; 95:9636.
  • 3
    Wells PS, Lensing AW, Davidson BL, Prins MH, Hirsh J. Accuracy of ultrasound for the diagnosis of deep venous thrombosis in asymptomatic patients after orthopedic surgery. A meta-analysis. Ann Intern Med. 1995; 122:4753.
  • 4
    Wicki J, Perneger TV, Junod AF, Bounameaux H, Perrier A. Assessing clinical probability of pulmonary embolism in the emergency ward: a simple score. Arch Intern Med. 2001; 161:927.
  • 5
    Kearon C, Ginsberg JS, Douketis J, et al. An evaluation of D-dimer in the diagnosis of pulmonary embolism: a randomized trial. Ann Intern Med. 2006; 144:81221.
  • 6
    van Belle A, Buller HR, Huisman MV, et al. Effectiveness of managing suspected pulmonary embolism using an algorithm combining clinical probability, D-dimer testing, and computed tomography. JAMA. 2006; 295:1729.
  • 7
    Wells PS, Anderson DR, Rodger M, et al. Excluding pulmonary embolism at the bedside without diagnostic imaging: management of patients with suspected pulmonary embolism presenting to the emergency department by using a simple clinical model and D-dimer. Ann Intern Med. 2001; 135:98107.
  • 8
    Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med. 2006; 354:231727.
  • 9
    Anderson DR, Kahn SR, Rodger MA, et al. Computed tomographic pulmonary angiography vs ventilation-perfusion lung scanning in patients with suspected pulmonary embolism: a randomized controlled trial. JAMA. 2007; 298:274353.
  • 10
    Righini M, Le Gal G, Aujesky D, et al. Complete venous ultrasound in outpatients with suspected pulmonary embolism. J Thromb Haemost. 2009; 7:40612.
  • 11
    Kalra MK, Maher MM, D’Souza R, Saini S. Multidetector computed tomography technology: current status and emerging developments. J Comput Assist Tomogr. 2004; 28(Suppl 1):S26.
  • 12
    Mettler FA Jr, Bhargavan M, Faulkner K, et al. Radiologic and nuclear medicine studies in the United States and worldwide: frequency, radiation dose, and comparison with other radiation sources--1950–2007. Radiology. 2009; 253:52031.
  • 13
    Brenner DJ. Radiation risks potentially associated with low-dose CT screening of adult smokers for lung cancer. Radiology. 2004; 231:4405.
  • 14
    Caro JJ, Trindade E, McGregor M. The risks of death and of severe nonfatal reactions with high- vs low-osmolality contrast media: a meta-analysis. AJR Am J Roentgenol. 1991; 156:82532.
  • 15
    Katayama H, Yamaguchi K, Kozuka T, Takashima T, Seez P, Matsuura K. Adverse reactions to ionic and nonionic contrast media. A report from the Japanese Committee on the Safety of Contrast Media. Radiology. 1990; 175:6218.
  • 16
    Mitchell A, Jones A, Tumlin J, Kline J. Contrast induced nephropathy in the emergency department setting: incidence, risk-factors and outcomes [abstract]. Acad Emerg Med. 2009; 16(Suppl 1):S21.
  • 17
    Mitchell AM, Kline JA. Contrast nephropathy following computed tomography angiography of the chest for pulmonary embolism in the emergency department. J Thromb Haemost. 2007; 5:504.
  • 18
    Kline JA, Courtney DM, Beam DM, King MC, Steuerwald M. Incidence and predictors of repeated computed tomographic pulmonary angiography in emergency department patients. Ann Emerg Med. 2009; 54:418.
  • 19
    McCullough PA, Adam A, Becker CR, et al. Epidemiology and prognostic implications of contrast-induced nephropathy. Am J Cardiol. 2006; 98(6 Suppl 1):513.
  • 20
    Torbicki A, Perrier A, Konstantinides S, et al. Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J. 2008; 29:2276315.
  • 21
    Righini M, Le Gal G, Aujesky D, et al. Diagnosis of pulmonary embolism by multidetector CT alone or combined with venous ultrasonography of the leg: a randomised non-inferiority trial. Lancet. 2008; 371:134352.
  • 22
    Turkstra F, Kuijer PM, van Beek EJ, Brandjes DP, ten Cate JW, Buller HR. Diagnostic utility of ultrasonography of leg veins in patients suspected of having pulmonary embolism. Ann Intern Med. 1997; 126:77581.
  • 23
    Aujesky D, Smith KJ, Cornuz J, Roberts MS. Cost-effectiveness of low-molecular-weight heparin for treatment of pulmonary embolism. Chest. 2005; 128:160110.
  • 24
    Musset D, Parent F, Meyer G, et al. Diagnostic strategy for patients with suspected pulmonary embolism: a prospective multicentre outcome study. Lancet. 2002; 360:191420.
  • 25
    Mettler FA Jr, Huda W, Yoshizumi TT, Mahesh M. Effective doses in radiology and diagnostic nuclear medicine: a catalog. Radiology. 2008; 248:25463.
  • 26
    Lessler AL, Isserman JA, Agarwal R, Palevsky HI, Pines JM. Testing low-risk patients for suspected pulmonary embolism: a decision analysis. Ann Emerg Med. 2010; 55:31626.
  • 27
    Board on Radiation Effects Research. Health Risks from Exposure to Low Levels of Ionizing Radiation: BEIR VII, Phase 2. Washington, DC: National Academies Press, 2006.
  • 28
    Linkins LA, Choi PT, Douketis JD. Clinical impact of bleeding in patients taking oral anticoagulant therapy for venous thromboembolism: a meta-analysis. Ann Intern Med. 2003; 139:893900.
  • 29
    Punthakee X, Doobay J, Anand SS. Oral-anticoagulant-related intracerebral hemorrhage. Thromb Res. 2002; 108:316.
  • 30
    Buller HR, Davidson BL, Decousus H, et al. Subcutaneous fondaparinux versus intravenous unfractionated heparin in the initial treatment of pulmonary embolism. N Engl J Med. 2003; 349:1695702.
  • 31
    Doyle NM, Monga M. Thromboembolic disease in pregnancy. Obstet Gynecol Clin North Am. 2004; 31:31944.
  • 32
    Duriseti RS, Shachter RD, Brandeau ML. Value of quantitative D-dimer assays in identifying pulmonary embolism: implications from a sequential decision model. Acad Emerg Med. 2006; 13:75566.
  • 33
    Douketis JD, Kearon C, Bates S, Duku EK, Ginsberg JS. Risk of fatal pulmonary embolism in patients with treated venous thromboembolism. JAMA. 1998; 279:45862.
  • 34
    Almahameed A, Carman TL. Outpatient management of stable acute pulmonary embolism: proposed accelerated pathway for risk stratification. Am J Med. 2007; 120(10 Suppl 2):S1825.
  • 35
    Calder KK, Herbert M, Henderson SO. The mortality of untreated pulmonary embolism in emergency department patients. Ann Emerg Med. 2005; 45:30210.
  • 36
    Quinlan DJ, McQuillan A, Eikelboom JW. Low-molecular-weight heparin compared with intravenous unfractionated heparin for treatment of pulmonary embolism: a meta-analysis of randomized, controlled trials. Ann Intern Med. 2004; 140:17583.
  • 37
    Bressollette L, Nonent M, Oger E, et al. Diagnostic accuracy of compression ultrasonography for the detection of asymptomatic deep venous thrombosis in medical patients--the TADEUS project. Thromb Haemost. 2001; 86:52933.
  • 38
    Burnside PR, Brown MD, Kline JA. Systematic review of emergency physician-performed ultrasonography for lower-extremity deep vein thrombosis. Acad Emerg Med. 2008; 15:4938.
  • 39
    Kearon C, Ginsberg JS, Hirsh J. The role of venous ultrasonography in the diagnosis of suspected deep venous thrombosis and pulmonary embolism. Ann Intern Med. 1998; 129:10449.
  • 40
    Kline JA, O’Malley PM, Tayal VS, Snead GR, Mitchell AM. Emergency clinician-performed compression ultrasonography for deep venous thrombosis of the lower extremity. Ann Emerg Med. 2008; 52:43745.
  • 41
    Frazee BW, Snoey ER, Levitt A. Emergency department compression ultrasound to diagnose proximal deep vein thrombosis. J Emerg Med. 2001; 20:10712.
  • 42
    Jacoby J, Cesta M, Axelband J, Melanson S, Heller M, Reed J. Can emergency medicine residents detect acute deep venous thrombosis with a limited, two-site ultrasound examination? J Emerg Med. 2007; 32:197200.
  • 43
    Jang T, Docherty M, Aubin C, Polites G. Resident-performed compression ultrasonography for the detection of proximal deep vein thrombosis: fast and accurate. Acad Emerg Med. 2004; 11:31922.
  • 44
    Shepard DS. Cost-effectiveness in Health and Medicine. By M.R.Gold, J.ESiegel, L.B.Russell, M.C.Weinstein (eds). New York: Oxford University Press, 1996 [book review]. J Ment Health Policy Econ. 1999; 2:9192.
  • 45
    Heron M, Hoyert DL, Murphy SL, Xu J, Kochanek KD, Tejada-Vera B. Deaths: final data for 2006. Natl Vital Stat Rep. 2009; 57:1134.
  • 46
    Pepper PV, Owens DK. Cost-effectiveness of the pneumococcal vaccine in healthy younger adults. Med Decis Making. 2002; 22(5 Suppl):S4557.
  • 47
    Tengs TO, Wallace A. One thousand health-related quality-of-life estimates. Med Care. 2000; 38:583637.
  • 48
    Das P, Ng AK, Earle CC, Mauch PM, Kuntz KM. Computed tomography screening for lung cancer in Hodgkin’s lymphoma survivors: decision analysis and cost-effectiveness analysis. Ann Oncol. 2006; 17:78593.
  • 49
    Centers of Medicare and Medicaid Services. 2009 National Physician Fee Schedule Relative Value File. Hyattville, MD: U.S. Department of Health and Human Services, 2009.
  • 50
    Centers of Medicare and Medicaid Services. Acute Inpatient PPS. Short Stay Inpatient by Diagnosis Related Group. AOR FY09 Final Notice Data. Hyattville, MD: U.S. Department of Health and Human Services, 2009.
  • 51
    Kaplan RM, Bush JW. Health-related quality of life measurement for evaluation research and policy analysis. Health Psychol. 1982; 1:6180.
  • 52
    Shiroiwa T, Sung YK, Fukuda T, Lang HC, Bae SC, Tsutani K. International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ. 2010; 19:42237.
  • 53
    Fischer B, Lassen U, Mortensen J, et al. Preoperative staging of lung cancer with combined PET-CT. N Engl J Med. 2009; 361:329.
  • 54
    Ubel PA, Hirth RA, Chernew ME, Fendrick AM. What is the price of life and why doesn’t it increase at the rate of inflation? Arch Intern Med. 2003; 163:163741.
  • 55
    Centers of Medicare and Medicaid Services. 2009 Clinical Diagnostic Laboratory Fee Schedule. Hyattville, MD: U.S. Department of Health and Human Services, 2009.
  • 56
    Eckman MH, Rosand J, Greenberg SM, Gage BF. Cost-effectiveness of using pharmacogenetic information in warfarin dosing for patients with nonvalvular atrial fibrillation. Ann Intern Med. 2009; 150:7383.
  • 57 Warfarin sodium-coumadin, 5mg tables #30. Available at: Accessed Oct 10, 2010.
  • 58
    Kline JA, Mitchell AM, Kabrhel C, Richman PB, Courtney DM. Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost. 2004; 2:124755.
  • 59
    Nijkeuter M, Ginsberg JS, Huisman MV. Diagnosis of deep vein thrombosis and pulmonary embolism in pregnancy: a systematic review. J Thromb Haemost. 2006; 4:496500.