a b s t r a c t

Objectives: The diagnosis of acute pulmonary embolism (PE) in trauma patients is challenging. This study evalu- ated the diagnostic value of simplified Wells and simplified revised Geneva scores to predict PE in femur fracture patients in emergency department (ED).

Methods: All consecutive adult patients with femur fractures and elevated D-dimer levels (N 0.5 ug/mL) who underwent CTPA within 72 h of injury from January 2010 to December 2014 were included. The simplified Wells and simplified revised Geneva scores were applied to evaluate the clinical probability of PE.

Results: Among 519 femur fracture patients, 446 patients were finally included, and 23 patients (5.2%) were di- agnosed with acute PE. The median values of simplified Wells and simplified revised Geneva scores [0 (IQR: 0-1) vs. 0 (IQR: 0-0), P = 0.23; 3 (IQR: 2-4) vs. 3 (IQR: 2-3), P = 0.48] showed no differences between the PE (n = 23) and non-PE (n = 423) groups. Using the simplified Wells score, 98% of the patients were categorized into the “PE unlikely” group. The sensitivity, specificity, positive predictive value, and negative predictive value of the sim- plified revised Geneva score (>=3 points) for the diagnosis of PE were 74%, 35%, 6%, and 96%, respectively.

Conclusion: In femur fracture patients with elevated D-dimer levels, the simplified Wells and simplified revised Geneva scores have limited predictive value. However, the simplified revised Geneva score of b 3 points may be possibly used as a diagnostic tool.

(C) 2017

Introduction

Acute pulmonary embolism (PE) is a well-known fatal complication of major trauma [1-4], and recent literature has suggested that a signif- icant percentage of post-trauma PEs occur within the early period (within 72 h of injury) [5-10]. Importantly, lower limb fractures have been identified as independent risk factors for early PEs after trauma [7]. Strategies to diagnose PE using clinical prediction scores and serum D-dimer levels are widely utilized, but their use in orthopedic trauma patients remains challenging, as traumatic injury can lead to el- evated D-dimer levels independent of PE in the acute period after injury [11].

The Wells score and the revised Geneva score are the most widely used clinical Prediction tools for the detection of PE [12,13]. Both scores assign different weights to variables and stratify patients into 3 clinical probabilities (low, moderate, and high) and 2 groups (PE likely or

* Corresponding author at: Department of Emergency Medicine, University of Ulsan College of Medicine, Asan Medical Center, 88 Olympic-ro 43 gil, Songpa-gu, Seoul 05505, Republic of Korea.

E-mail address: [email protected] (W.Y. Kim).

¹ These authors contributed equally to this work.

unlikely). Both scores have been extensively validated and have become important strategies for diagnosing PE. Recently, these scoring systems were simplified to increase their clinical usefulness [14,15]. However, the simplified Wells score and the simplified revised Geneva score have not been validated in orthopedic trauma patients. Furthermore, no study has specifically examined the value of the diagnostic perfor- mance of the two scores for early PE in femur fracture patients who have elevated D-dimer levels at the emergency department (ED). Thus, the objective of this study was to evaluate the diagnostic value of simplified Wells and simplified revised Geneva scores to predict the need for computed tomography pulmonary angiography to con- firm clinically suspected PE (elevation of D-dimer) in femur fracture pa- tients at ED admission.

Methods

Study design and population

This retrospective, single-center study was conducted at the ED of a university-affiliated, tertiary referral center in Seoul, Korea, with an an- nual census of approximately 100 000 visits. Before commencing this

http://dx.doi.org/10.1016/j.ajem.2017.03.023

0735-6757/(C) 2017

Table 1

Clinical prediction rules for pulmonary embolism.

Simplified Wells score	Points	Simplified revised Geneva score	Points
Previous pulmonary embolism or deep vein thrombosis	+1	Previous pulmonary embolism or deep vein thrombosis	+1
Heart rate N 100 beats per minute	+1	Heart rate
Surgery or immobilization within 4 weeks	+1	75-94 beats per minute	+1
Hemoptysis	+1	>=95 beats per minute	+2
Active cancer	+1	Surgery or fracture within 1 month	+1
Clinical signs of deep vein thrombosis	+1	Hemoptysis	+1
alternative diagnosis less likely than pulmonary embolism	+1	Active cancer	+1
		Unilateral lower limb pain	+1
		Pain on lower limb deep venous palpation and unilateral edema	+1
		Age N 65 years	+1
Clinical probability		Clinical probability
Unlikely	<= 1	Unlikely	<= 2
Likely	N 1	Likely	N 2

study, our institutional review board approved the review of patient data and waived the requirement for informed consent (S2015-0627- 0001). A total of 1301 consecutive adult patients (>= 18 years old) with femur fractures admitted to our ED from January 2010 to December 2014 were eligible to participate in this study. From this popula- tion, we included 446 patients who had elevated D-dimer levels (N 0.5 ug/mL) and underwent subsequent CTPA within 72 h of injury. Patients were excluded if they had normal D-dimer levels or did not undergo a D-dimer test or CTPA within 72 h of their injury. The clinical probability of PE was assessed using simplified Wells and simplified revised Geneva scores at ED admission (Table 1) [14,15].

CTPA scans were considered positive if the following criteria were satisfied: Contrast material failed to fill the entire lumen because of a central filling defect (the artery may be enlarged compared with similar arteries); contrast material surrounded a partial filling defect on a cross- sectional image; there was contrast material between the central filling defect and the artery wall on an in-plane, longitudinal image; there was a peripheral intraluminal filling defect that formed an acute angle with the artery wall [16]. Board-certified radiologists made formal interpre- tive reports of CTPA scans. PE is usually classified as proximal or distal depending on the location of the emboli identified on the CT scan. Usually, when the emboli are located within the main or lobar arteries, they are reported as proximal PE, and emboli that are segmental or subsegmental are reported as distal PE [17]. In this present study, we di- agnosed PE if emboli were located within the main or lobar arteries in CTPA within 72 h after injury. CTPA is considered in femur fracture pa- tients with dyspnea, chest discomfort, or Oxygen desaturation (b 90%) in our ED unless patients refuse a CT scan.

Data collection

Demographic and clinical data, such as age, sex, symptoms, signs, co- morbidity, injury mechanism, vector, and fracture site, the presence of pulmonary embolism, and the variables of simplified Wells and revised Geneva scores for the affected patients, were collected from electronic medical records. Patient vital signs and relevant laboratory findings after admission to the ED were also collected. The duration between CT scan and injury was also calculated. A reviewer (D.H.C) calculated both simplified scores for each patient based on data from electronic medical records and in a blind manner.

Statistical analysis

Patients were divided into PE likely and PE unlikely groups accord- ing to the simplified Wells and revised Geneva scores, and their PE prev- alence was calculated. To assess the ability of the simplified Wells and revised Geneva scores to predict PE, area under the receiver-Operator characteristics curves (AUCs) were calculated. Sensitivity, specificity, positive predictive values, and negative predictive values were also cal- culated. Continuous variables were expressed as means with standard deviations or medians with interquartile ranges (IQRs) if the assump- tion of a normal distribution was violated. Categorical variables were re- ported in terms of numbers and percentages. Femur fracture patients with PE were compared to those without PE. All variables were tested for normal distribution using the Kolmogorov-Smirnov test. The Student’s t-test was used to compare the mean values of normally dis- tributed continuous variables, whereas the Mann-Whitney U test was

Fig. 1. Patient flow diagram. CTPA = computed tomography pulmonary angiography; ED = emergency department; PE = pulmonary embolism.

used to compare non-normally distributed continuous variables. The chi-squared or Fisher’s exact test was used for categorical variables. A 2-sided P value of b 0.05 was considered to be statistically significant. All statistical analyses were performed using SPSS version 21.0 (SPSS Inc., Chicago, IL).

Results

From January 2010 to December 2014, 519 patients were examined for the presence of PE on a CT scan and 73 patients were excluded from

the study because they either underwent CTPA over 72 h after injury (n = 66), did not undergo serum D-dimer testing (n = 5), or had a nor- mal D-dimer level (<= 0.5 ug/mL) (n = 2). Thus, 446 patients were finally included in the study population (Fig. 1), and 23 patients (5.2%) in this cohort were diagnosed with acute PE.

The baseline characteristics of the study patients are summarized in Table 2. The two groups were similar in terms of demographic and clin- ical variables, including the components of the simplified Wells and re- vised Geneva scores. The median age of the cohort was 78 years, and 135 patients (30.3%) were male. Most of the patients (98.9%) had

Table 2

Demographic and clinical findings for the study patients.

	Total	PE group	Non-PE group	P value
	(N = 446)	(N = 23)	(N = 423)
Age	78.0 (71.0-82.0)	79.0 (72.0-84.0)	78.0 (71.0-82.0)	0.634
Male	135 (30.3%)	10 (43.5%)	125 (29.6%)	0.157
Smoking	32 (7.2%)	2 (7.4%)	30 (7.1%)	0.676
Initial vital signs
Systolic BP	143 (127-158)	145 (125-158)	143 (127-159)	0.899
Diastolic BP	83 (74-93)	80 (71-95)	83 (74-93)	0.813
Heart rate	83 (72-94)	84 (65-91)	83 (72-95)	0.863
Respiratory rate	20 (20 -20)	20 (20-20)	20 (18-20)	0.889
BT (?C)	36.6 (36.3-37.0)	36.5 (36.1-37.0)	36.6 (36.3-37.0)	0.342
Saturation (%)	98.0 (96.0-99.0)	96.0 (95.0-99.0)	98.0 (96.0-99.0)	0.264
injury type				0.247
Slip	397 (89.0%)	19 (82.6%)	378 (89.4%)
Traffic accident	27 (6.1%)	3 (13.0%)	24 (5.7%)
Fall	12 (2.7%)	0 (0%)	12 (2.8%)
Others	10 (2.2%)	1 (4.3%)	9 (2.1%)
Injury mechanism				1.000
Blunt	441 (98.9%)	23 (100%)	418 (98.8%)
Penetrating	1 (0.2%)	0 (0%)	1 (0.2%)
Others	4 (0.9%)	0 (0%)	4 (0.9%)
Fracture site of femur				0.904
Neck	196 (43.9%)	11 (47.8%)	185 (43.7%)
Intertrochanteric	224 (50.2%)	11 (47.8%)	213 (50.4%)
Other site	26 (5.8%)	1 (4.3%)	25 (5.9%)
Concomitant fracture of other site	53 (11.9%)	2 (8.7%)	51 (12.1%)	1.000
Injury severity
AIS score	3 (3-3)	3 (3-3)	3 (3-3)	0.721
ISS	9 (9-9)	9 (9-9)	9 (9-9)	0.662
PE probability score Simplified Wells score	0 (0-0)	0 (0-1)	0 (0-0)	0.234
Simplified Wells score >= 2	7 (1.6%)	1 (4.3%)	6 (1.4%)	0.311
Simplified revised Geneva score	3 (2-3)	3 (2-3)	3 (2-3)	0.477
Simplified revised Geneva score >= 3	293 (65.7%)	17 (73.9%)	276 (65.2%)	0.394
elapsed time between injury time and ED presentation time (hours)	5.0 (2.0-11.0)	9.0 (4.0-23.0)	5.0 (2.0-11.0)	0.038
Elapsed time between injury time and CT scan time (hours)	13.0 (9.0-23.3)	21.0 (12.0-39.0)	12.0 (9.0-23.0)	0.002
Comorbidity
history of AMI	21 (4.7%)	0 (0%)	21 (5.0%)	0.616
Angina	54 (12.1%)	2 (8.7%)	52 (12.3%)	1.000
Atrial fibrillation	36 (8.1%)	0 (0%)	36 (8.5%)	0.241
Heart failure	16 (3.6%)	1 (4.3%)	15 (3.5%)	0.578
History of valvular heart disease	24 (5.4%)	0 (0%)	24 (5.7%)	0.626
Peripheral vessel disease	3 (0.7%)	0 (0%)	3 (0.7%)	1.000
Coagulopathy	7 (1.6%)	1 (4.3%)	6 (1.4%)	0.311
Hypertension	288 (64.6%)	16 (69.6%)	272 (64.3%)	0.607
Diabetes mellitus	153 (34.3%)	7 (30.4%)	146 (34.5%)	0.688
Previous venous thromboembolism	7 (1.6%)	2 (8.7%)	5 (1.2%)	0.046
Cancer	18 (4.0%)	1 (4.3%)	17 (4.0%)	1.000
Immobilization	2 (0.4%)	1 (4.3%)	1 (0.2%)	0.101
Medication use Aspirin	118 (26.5%)	5 (21.7%)	113 (26.7%)	0.598
Warfarin	21 (4.7%)	2 (8.7%)	19 (4.5%)	0.296
Antiplatelets	40 (9.0%)	0 (0%)	40 (9.5%)	0.250
Laboratory test Platelets (x103/uL)	187 (151-218)	204 (165-216)	186 (150-219)	0.322
PT (INR)	1.02 (0.97-1.08)	1.02 (0.95-1.07)	1.02 (0.97-1.08)	0.584
aPTT (seconds)	27.5 (25.5-29.4)	28.7 (26.8-29.7)	27.4 (25.5-29.3)	0.072
D-dimer (ug/mL)	26.1 (13.5-35.2)	14.8 (11.8-35.2)	26.3 (13.5-35.2)	0.321
ICU care during hospitalization	121 (27.1%)	9 (39.1%)	112 (26.5%)	0.184
In-hospital mortality	7 (1.6%)	0 (0%)	7 (1.7%)	1.000

Values are expressed as median with interquartile range and n (%).

PE = pulmonary embolism; BP = blood pressure; BT = body temperature; AIS = Abbreviated Injury Scale; ISS = Injury Severity Score; CT = computed tomography; ED = emergency department; AMI = acute myocardial infarction; PT = prothrombin time; INR = international normalized ratio; aPTT = activated partial thromboplastin time, ICU = intensive care unit.

blunt trauma, and the intertrochanteric femur was the most frequent fracture site (50.2%), followed by the femoral neck (43.9%). There was no statistical difference in PE probability scores in those with and with- out PE [simplified Wells score, 0 (IQR: 0-1) vs. 0 (IQR: 0-0), P = 023;

simplified revised Geneva score, 3 (IQR: 2-3) vs. 3 (IQR: 2-3), P = 0.48]. Also, there was no statistical difference in PE probability categori- zation of likely and unlikely groups in those with and without PE [sim- plified Wells score >= 2, 1 (4.3%) vs. 6 (1.4%), P = 0.31; simplified revised Geneva score >= 3, 17 (73.9%) vs. 276 (65.2%), P = 0.39]. Median time be- tween injury and CT scan was statistically different between the pa- tients with and without PE [21.0 h (IQR: 12.0-39.0) vs. 12.0 h (IQR: 9.0-23.0), P = 0.002]. Seven patients (1.6%) died during hospitalization, all from the non-PE group.

Using the simplified Wells score, 98.2% (439/447) of the study pa- tients were categorized into the PE unlikely group, and 5.0% (22/439) were diagnosed with PE (Table 3). Using the simplified revised Gene- va score, 34.2% (153/447) of the study patients were categorized into the PE unlikely group, and 3.9% (6/153) were diagnosed with PE. The abilities of the simplified Wells score and simplified revised Geneva score to predict PE, expressed as AUCs, were 0.548 (95% CI: 0.421-0.676), and 0.541 (95% CI: 0.429-0.653), respectively. The

simplified Wells score predicted PE with a sensitivity of 4.4% (95% CI: 0.1-22.0), a specificity of 98.6% (95% CI: 96.9-99.5), a positive predictive value of 14.3% (95% CI: 0.4-57.9), and a negative predictive value of 95.0% (95% CI: 92.5-96.8). The simplified revised Geneva score predicted PE with a sensitivity of 73.9% (95% CI: 51.6-89.8), a specificity of 34.8% (95% CI: 30.2-39.5), a positive predictive value of 5.8% (95% CI: 3.4-9.1), and a negative predictive value of 96.1% (95%

CI: 91.7-98.6).

Discussion

This study examined the usefulness of simplified Wells and simpli- fied revised Geneva scores for predicting early PE in 446 femur fracture patients who had elevated levels of D-dimer upon emergency depart- ment admission. The simplified Wells score was limited in its ability to predict PE risk. However, a simplified revised Geneva score of b 3 points was able to rule out PE with a 97% negative predictive value. Only 6 pa- tients with a simplified Revised Geneva score of b 3 points had a PE, and there was no mortality in these patients. We believe that the observa- tions reported here have potentially significant implications for the decision to perform CTPA for PE in femur neck fracture patients. Additionally, if we only performed CTPA only in the 153 patients with a simplified revised Geneva score of b 3 points, we could reduce 34.2% of the costs associated with the use of CTPA.

The simplified Wells and simplified revised Geneva scores have been widely integrated and validated as the first step in Diagnostic strategies

Table 3

Simplified Wells and simplified revised Geneva score classification and accuracy with which these scores predict pulmonary embolism in femur fracture patients with elevated D-dimer levels

Clinical probability Pulmonary embolism

Simplified Wells score? Simplified revised Geneva score⁺ Sensitivity, % (95% CI) 4.4 (0.1-22.0) 73.9 (51.6-89.8)

Specificity, % (95% CI) 98.6 (96.9-99.5) 34.8 (30.2-39.5)

PPV, % (95% CI) 14.3 (0.4-57.9) 5.8 (3.4-9.1)

NPV, % (95% CI) 95.0 (92.5-96.8) 96.1 (91.7-98.6)

PLR, (95% CI) 3.1 (0.4-24.4) 1.1 (0.9-1.5)

NLR, (95% CI) 1.0 (0.9-1.1) 0.8 (0.4-1.5)

CI = confidence interval; PPV = positive predictive value; NPV = negative predictive value; PLR = positive likelihood ratio; NLR = negative likelihood ratio.

?² for clinical probability (unlikely vs. likely) and pulmonary embolism.

* P = 0.311

⁺ P= 0.394.

in patients with clinically suspected PE [12,13,15]. These are the most widely used and extensively validated predictive models avail- able for the assessment of PE probability in medical and surgical patients presenting with clinically suspected PE, but not in trauma patients [12-15,18]. In addition, the revised Geneva score was derived and validated in outpatients, and the Wells rule has been derived and validated both in outpatients and inpatients, but not in patients visiting ED [12-15,18]. Their aim is to separate patients into two or three catego- ries of clinical pre-test probability, corresponding to categories of PE prevalence [12,13,15]. A recent study demonstrated that the Wells score is not correlated with the development of PE in orthopedic trauma patients [4]. The applicability of two simplified scores in the context of orthopedic trauma, which is one of the major risk factors of PE, has not yet been validated. In addition, recent studies have reported that early stage post-trauma PE comprised 25% to 43% of all post-trauma PE [7-9]. To the best of our knowledge, there have been no previous studies evaluating the diagnostic value of simplified Wells and simpli- fied revised Geneva scores in predicting early PE in femur fracture patients.

The superiority of the simplified revised Geneva score in our study may be associated with the fully standardized criteria itself and Geneva score-specific variables such as “age N 65 years” and “surgery or fracture within 1 month,” that are absent in the simplified Wells score. The Wells model has been criticized due to subjective criteria such as “alternative diagnosis less likely than pulmonary embolism.” The use of clinical judgment in assigning this score has been shown to be dependent on the presence of other variables in the Wells score [19]. Although such clinical judgment remains one of the most informative criteria in this model [20], other Alternative diagnoses would be possible in trauma pa- tients, which make this criterion inapplicable. In our present study, 369 (82.7%) patients had a 0 score on the simplified Wells score, suggesting a lack of efficacy of risk assessment in our cohort. In contrast to the sim- plified Wells score, 1 point was routinely added for the variable ‘surgery or fracture within 1 month’ in the simplified revised Geneva score, since all patients had femur fractures. Besides, 392 (87.9%) patients were older than 65 years and 316 (70.9%) patients had a heart rate of 75 beats per minute or faster, which each added another point. As a result, the simplified revised Geneva score showed diverse scores from 1 to 5 in our cohort, and this may have contributed to the increased discrimi- nant power of the simplified revised Geneva score, especially in our femur fracture patients.

To improve the efficacy of predicting models, we have attempted to combine these two predictive scores. However, both scores demonstrat- ed poor agreement with respect to clinical probability, and 286 (64.1%) patients were classified differently. Only 7 (1.6%) patients were classi- fied as “likely to develop PE” with the simplified Wells score, and all of them were classified as “likely to develop PE” with the simplified re- vised Geneva score. However, only one of those patients was diagnosed with early PE, suggesting a lack of predictive value in combining the two scores.

This study had several limitations. First, there was an inevitable selection bias. Although the clinical data for the 446 femur fracture patients presenting with clinically suspected PE (dyspnea, chest dis- comfort, or oxygen desaturation) and elevated D-dimer levels undergo- ing CTPA were evaluated, the remaining patients in the cohort were not evaluated. Additionally, a small number of patients (23/446, 5.2%) were diagnosed with acute PE in our study, which would result selection bias. Thus further prospective validation study will be needed. Second, the retrospective design of the study imposed intrinsic limitations. The sim- plified Wells and revised Geneva scores were assessed retrospectively, and some variable data were missing. These missing values were assumed to be normal but this strategy could have affected the results. Third, the study was conducted in a single hospital and involved a relatively small sample size, which weakened our results. Finally, it re- mains unclear whether the early PE identified on CTPA was clinically significant.

Conclusion

Our current study provides insights into the predictive value of the simplified Wells and the simplified revised Geneva scores in femur fracture patients presenting with clinically suspected PE and elevated D-dimer levels. A simplified revised Geneva score of b 3 points was found to be acceptable to observe closely rather than performing emer- gent CTPA in these patients.

Disclosure

The authors have no conflicts of interest to disclose.

References

1. Malinoski D, Ewing T, Patel MS, Jafari F, Sloane B, Nguyen B, et al. Risk factors for ve- nous thromboembolism in critically ill trauma patients who cannot receive chemical prophylaxis. Injury 2013;44:80-5.
2. Schiff RL, Kahn SR, Shrier I, Strulovitch C, Hammouda W, Cohen E, et al. Identifying orthopedic patients at high risk for venous thromboembolism despite thromboprophylaxis. Chest 2005;128:3364-71.
3. Tornetta P, Bogdan Y. Pulmonary embolism in orthopaedic patients: diagnosis and management. J Am Acad Orthop Surg 2012;20:586-95.
4. Young MD, Daniels AH, Evangelista PT, Reinert SE, Ritterman S, Christino MA, et al. Predicting pulmonary embolus in orthopedic trauma patients using the Wells score. Orthopedics 2013;36:e642-7.
5. Spencer Netto F, Tien H, Ng J, Ortega S, Scarpelini S, Rizoli SB, et al. Pulmonary emboli after blunt trauma: timing, clinical characteristics and natural history. Injury 2012; 43:1502-6.
6. Benns M, Reilly P, Kim P. Early pulmonary embolism after injury: a different clinical entity? Injury 2014;45:241-4.
7. Coleman JJ, Zarzaur BL, Katona CW, Plummer ZJ, Johnson LS, Fecher A, et al. Factors associated with pulmonary embolism within 72 hours of admission after trauma: a multicenter study. J Am Coll Surg 2015;220:731-6.
8. Menaker J, Stein DM, Scalea TM. Incidence of early pulmonary embolism after injury.

   J Trauma 2007;63:620-4.

   Menaker J, Stein DM, Scalea TM. Pulmonary embolism after injury: more common than we think? J Trauma 2009;67:1244-9.

9. Kim YJ, Choi DH, Ahn S, Sohn CH, Seo DW, Kim WY. Timing of pulmonary embolisms in femur fracture patients: incidence and outcomes. J Trauma Acute Care Surg 2016; 80:952-6.
10. Meythaler JM, Fisher WS, Rue LW, Johnson A, Davis L, Brunner RC. Screening for ve- nous thromboembolism in traumatic brain injury: limitations of D-dimer assay. Arch Phys Med Rehabil 2003;84:285-90.
11. Wells PS, Anderson DR, Rodger M, Ginsberg JS, Kearon C, Gent M, et al. Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: increasing the models utility with the SimpliRED D-dimer. Thromb Haemost 2000; 83:416-20.
12. Le Gal G, Righini M, Roy PM, Sanchez O, Aujesky D, Bounameaux H, et al. Prediction of pulmonary embolism in the emergency department: the revised Geneva score. Ann Intern Med 2006;144:165-71.
13. Douma RA, Gibson NS, Gerdes VE, Buller HR, Wells PS, Perrier A, et al. Validity and clinical utility of the simplified Wells rule for assessing clinical probability for the ex- clusion of pulmonary embolism. Thromb Haemost 2009;101:197-200.
14. Klok FA, Mos IC, Nijkeuter M, Righini M, Perrier A, Le Gal G, et al. Simplification of the revised Geneva score for assessing clinical probability of pulmonary embolism. Arch Intern Med 2008;168:2131-6.
15. Stein PD, Fowler SE, Goodman LR, Gottschalk A, Hales CA, Hull RD, et al. Multidetec- tor computed tomography for acute pulmonary embolism. N Engl J Med 2006;354: 2317-27.
16. Pruszczyk P, Torbicki A, Pacho R, Chlebus M, Kuch-Wocial A, Pruszynski B, et al. Non- invasive diagnosis of suspected severe pulmonary embolism: transesophageal echo- cardiography vs spiral CT. Chest 1997;112:722-8.
17. Ceriani E, Combescure C, Le Gal G, Nendaz M, Perneger T, Bounameaux H, et al. Clin- ical prediction rules for pulmonary embolism: a systematic review and meta- analysis. J Thromb Haemost 2010;8:957-70.
18. Klok FA, Zidane M, Djurabi RK, Nijkeuter M, Huisman MV. The physician’s estimation ‘alternative diagnosis is less likely than pulmonary embolism’ in the Wells rule is de- pendent on the presence of other required items. Thromb Haemost 2008;99:244-5.
19. Kabrhel C, McAfee AT, Goldhaber SZ. The contribution of the subjective component of the Canadian Pulmonary Embolism Score to the overall score in emergency de- partment patients. Acad Emerg Med 2005;12:915-20.