An algorithm combining clinical probability assessment, D-dimer testing, and CT was evaluated for managing suspected pulmonary embolism in 3306 patients. Using this approach, pulmonary embolism was correctly excluded in 1057 low probability patients with normal D-dimers and 1505 patients with negative CT scans, with a very low 3-month VTE risk of 0.5-1.3%. CT confirmed pulmonary embolism in 674 high probability patients. This diagnostic strategy allowed safe management decisions in 97.9% of patients with clinically suspected pulmonary embolism.

1. Effectiveness of Managing Suspected Pulmonary
Embolism Using an Algorithm Combining Clinical
Probability, D-Dimer Testing, and Computed
Online article and related content Tomography
current as of July 10, 2008.
Writing Group for the Christopher Study Investigators
JAMA. 2006;295(2):172-179 (doi:10.1001/jama.295.2.172)
http://jama.ama-assn.org/cgi/content/full/295/2/172
Correction Contact me if this article is corrected.
Citations This article has been cited 30 times.
Contact me when this article is cited.
Topic collections Radiologic Imaging; Diagnosis; Computed Tomography; Cardiovascular System;
Venous Thromboembolism
Contact me when new articles are published in these topic areas.
Related Articles published in Does This Patient Have Deep Vein Thrombosis?
the same issue Philip S. Wells et al. JAMA. 2006;295(2):199.
Diagnosing Pulmonary Embolism With Improved Certainty and Simplicity
Russell D. Hull. JAMA. 2006;295(2):213.
Pulmonary Embolism
Sharon Parmet et al. JAMA. 2006;295(2):240.
Related Letters An Algorithm for Managing Suspected Pulmonary Embolism
Hisato Takagi et al. JAMA. 2006;295(22):2603.
In Reply:
Mathilde Nijkeuter et al. JAMA. 2006;295(22):2604.
Subscribe Email Alerts
http://jama.com/subscribe http://jamaarchives.com/alerts
Permissions Reprints/E-prints
permissions@ama-assn.org reprints@ama-assn.org
http://pubs.ama-assn.org/misc/permissions.dtl
Downloaded from www.jama.com by guest on July 10, 2008

2. ORIGINAL CONTRIBUTION
Effectiveness of Managing Suspected
Pulmonary Embolism Using an Algorithm
Combining Clinical Probability, D-Dimer
Testing, and Computed Tomography
Writing Group for the
Christopher Study Investigators* Context Previous studies have evaluated the safety of relatively complex combina-
tions of clinical decision rules and diagnostic tests in patients with suspected pulmo-
T
HE MAIN CHALLENGE IN THE DI- nary embolism.
agnostic workup of patients Objective To assess the clinical effectiveness of a simplified algorithm using a di-
with clinically suspected pul- chotomized clinical decision rule, D-dimer testing, and computed tomography (CT) in
monary embolism is to accu- patients with suspected pulmonary embolism.
rately and rapidly distinguish the ap- Design, Setting, and Patients Prospective cohort study of consecutive patients
proximately 25% of patients who have with clinically suspected acute pulmonary embolism, conducted in 12 centers in the
the disease and require anticoagulant Netherlands from November 2002 through December 2004. The study population of
treatment from the 75% who do not.1,2 3306 patients included 82% outpatients and 57% women.
A number of new approaches have im- Interventions Patients were categorized as “pulmonary embolism unlikely” or “pul-
proved the diagnostic process for pul- monary embolism likely” using a dichotomized version of the Wells clinical decision
monary embolism. The first is the com- rule. Patients classified as unlikely had D-dimer testing, and pulmonary embolism was
bination of a clinical decision rule such considered excluded if the D-dimer test result was normal. All other patients under-
as the Wells score,3 which categorizes went CT, and pulmonary embolism was considered present or excluded based on the
patients as low, intermediate, or high results. Anticoagulants were withheld from patients classified as excluded, and all pa-
clinical probability of pulmonary em- tients were followed up for 3 months.
bolism, with a D-dimer test. Several Main Outcome Measure Symptomatic or fatal venous thromboembolism (VTE)
management studies have shown that during 3-month follow-up.
pulmonary embolism can be safely Results Pulmonary embolism was classified as unlikely in 2206 patients (66.7%). The
ruled out without the need for addi- combination of pulmonary embolism unlikely and a normal D-dimer test result oc-
tional imaging in patients with low curred in 1057 patients (32.0%), of whom 1028 were not treated with anticoagu-
clinical probability and a normal D- lants; subsequent nonfatal VTE occurred in 5 patients (0.5% [95% confidence inter-
dimer test result, occurring in 20% to val {CI}, 0.2%-1.1%]). Computed tomography showed pulmonary embolism in 674
patients (20.4%). Computed tomography excluded pulmonary embolism in 1505 pa-
40% of patients.3-5 In these studies, 3
tients, of whom 1436 patients were not treated with anticoagulants; in these patients
categories of likelihood were used. the 3-month incidence of VTE was 1.3% (95% CI, 0.7%-2.0%). Pulmonary embo-
However, a retrospective analysis sug- lism was considered a possible cause of death in 7 patients after a negative CT scan
gested that the clinical utility of the (0.5% [95% CI, 0.2%-1.0%]). The algorithm was completed and allowed a manage-
Wells score could be further increased ment decision in 97.9% of patients.
by using 2 instead of 3 categories of Conclusions A diagnostic management strategy using a simple clinical decision rule,
clinical probability, dichotomizing pa- D-dimer testing, and CT is effective in the evaluation and management of patients
tients as either likely or unlikely to have with clinically suspected pulmonary embolism. Its use is associated with low risk for
had a pulmonary embolism,3 but no subsequent fatal and nonfatal VTE.
large prospective studies evaluating this JAMA. 2006;295:172-179 www.jama.com
dichotomization have been carried out.
*The Authors/Writing Group and Investigators of Leiden University Medical Center, Section of Vascu-
See also pp 199 and 213 and the Christopher Study are listed at the end of this lar Medicine, Department of General Internal Medi-
Patient Page. article. cine, C4-68, Albinusdreef 2, 2333 ZA, Leiden, the
Corresponding Author: Menno V. Huisman, MD, Netherlands (m.v.huisman@lumc.nl).
172 JAMA, January 11, 2006—Vol 295, No. 2 (Reprinted) ©2006 American Medical Association. All rights reserved.
Downloaded from www.jama.com by guest on July 10, 2008

3. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
Another advancement is computed trasound (CUS) of the leg veins, and a consecutive patients with clinically sus-
tomography (CT), which has emerged normal multidetector-row CT scan. In pected pulmonary embolism to evalu-
as a prominent imaging technique for the that study, all patients with high prob- ate the effectiveness of this novel man-
exclusion or confirmation of pulmo- ability for pulmonary embolism had to agement strategy.
nary embolism, as well as the detection undergo pulmonary angiography after
of alternative diagnoses.6-10 However, a normal CT and normal CUS. A more ef- METHODS
critical missing piece of information has ficient strategy would consist of an al- Study Design
been whether it is safe to withhold an- gorithm with a dichotomized decision The Christopher Study was a prospec-
ticoagulation treatment after a CT that rule, D-dimer testing, and CT, in which tive cohort study evaluating a diagnos-
is negative for pulmonary embo- pulmonary embolism is considered ex- tic algorithm consisting of sequential
lism.11,12 In a recent study,13 recurrent ve- cluded in patients with an unlikely clini- application of a clinical decision rule,
nous thromboembolism (VTE) oc- cal probability score and a normal D-dimer testing, and CT within 24
curred in 1.7% of patients who initially D-dimer test result, while CT is used in hours of presentation (FIGURE). All pa-
had a low or intermediate probability for all other patients as the sole imaging tients were followed up for a period of
pulmonary embolism using the Geneva method to make management deci- 3 months after presentation to docu-
score,14 an abnormal D-dimer test re- sions. Therefore, we performed a pro- ment the occurrence of subsequent
sult, normal bilateral compression ul- spective study in a large cohort of symptomatic VTE.
Figure. Diagnostic Flowchart
3503 Patients With Clinically
Suspected Pulmonary
Embolism
197 Excluded
184 Met Exclusion Criteria
13 Did Not Provide Consent
3306 Study Patients
Clinical Decision Rule
2206 Pulmonary Embolism 1100 Pulmonary Embolism
Unlikely (Decision Rule Likely (Decision Rule
Score ≤4) Score >4)
D-Dimer Test
1057 Normal D-Dimer Test Result 1149 Abnormal D-Dimer Test 2249 Computed Tomography
(D-Dimer ≤500 ng/mL) Result (D-Dimer >500 ng/mL) Indicated
1057 Pulmonary Embolism 1505 Pulmonary Embolism 674 Pulmonary Embolism 20 Inconclusive 50 Computed Tomography
Excluded Excluded Confirmed Not Performed
1028 Did Not Receive Treatment∗ 1436 Did Not Receive Treatment† 674 Received Treatment 18 Did Not Receive Treatment 45 Did Not Receive Treatment
Follow-up at 3 mo Follow-up at 3 mo Follow-up at 3 mo Follow-up at 3 mo Follow-up at 3 mo
5 Nonfatal Event 11 Nonfatal Event 9 Nonfatal Event 1 Nonfatal Event 1 Nonfatal Event
4 Pulmonary Embolism 3 Pulmonary Embolism 3 Pulmonary Embolism 1 Pulmonary Embolism 0 Pulmonary Embolism
1 Deep Vein Thrombosis 8 Deep Vein Thrombosis 6 Deep Vein Thrombosis 0 Deep Vein Thrombosis 1 Deep Vein Thrombosis
0 Fatal Pulmonary Embolism 7 Fatal Pulmonary Embolism 11 Fatal Pulmonary Embolism 0 Fatal Pulmonary Embolism 1 Fatal Pulmonary Embolism
2 Lost to Follow-up 1 Lost to Follow-up 1 Lost to Follow-up 0 Lost to Follow-up 0 Lost to Follow-up
*Excludes 29 patients treated with anticoagulant therapy for reasons other than venous thromboembolism.
†Excludes 69 patients treated with anticoagulant therapy for reasons other than venous thromboembolism.
©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, January 11, 2006—Vol 295, No. 2 173
Downloaded from www.jama.com by guest on July 10, 2008

4. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
attending physician using a validated 4.0 mL/s and bolus tracking in the com-
Table 1. Clinical Decision Rule*
clinical decision rule (TABLE 1).3 Pul- mon pulmonary artery to get optimal
Variable Points
monary embolism was classified as “un- contrast opacification of the pulmo-
Clinical signs and symptoms 3.0
of deep vein thrombosis (minimum likely” with a clinical decision rule score nary arteries.
of leg swelling and pain with of 4 or less points, and “likely” with a The pulmonary arteries were evalu-
palpation of the deep veins)
Alternative diagnosis less likely 3.0 score of more than 4 points. This cut- ated up to and including the subseg-
than pulmonary embolism off was chosen because it has been mental vessels from the level of the
Heart rate 100/min 1.5 shown to give an acceptable VTE diag- aortic arch to the lowest hemidia-
Immobilization ( 3 d) or surgery 1.5
in the previous 4 wk nostic failure rate of 1.7% to 2.2% in phragm. Pulmonary embolism was
Previous pulmonary embolism 1.5 combination with a normal D-dimer test diagnosed if contrast material outlined
or deep vein thrombosis
Hemoptysis 1.0 result.3 An estimated 300 attending phy- an intraluminal defect or if a vessel
Malignancy (receiving treatment, 1.0 sicians in the participating hospitals was totally occluded by low-
treated in the last 6 mo or palliative)
used the clinical decision rule with the attenuation material on at least 2 adja-
*Clinical probability of pulmonary embolism unlikely: 4 or
less points; clinical probability of pulmonary embolism study participants. cent slices. These patients received
likely: more than 4 points. Source: Wells et al.3
In patients with a clinical decision low-molecular-weight heparin or
rule indicating pulmonary embolism unfractionated heparin, followed by
Patients unlikely, a D-dimer concentration was vitamin K antagonists, according to
Consecutive patients with clinically sus- measured, using either the VIDAS D- local practice. In patients without pul-
pected pulmonary embolism, defined dimer assay (Biomerieux, Marcy monary embolism, the presence or
as a sudden onset of dyspnea, sudden L’Etoile, France) or the Tinaquant as- absence of an alternative diagnosis was
deterioration of existing dyspnea, or say (Roche Diagnostica, Mannheim, recorded and anticoagulant treatment
sudden onset of pleuritic chest pain Germany). A D-dimer concentration of was withheld. The CT was considered
without another apparent cause, were 500 ng/mL or less was defined as nor- inconclusive if the images could not
potentially eligible for the study. Pa- mal. In patients with pulmonary em- be interpreted because of motion arti-
tients presenting to the emergency ward bolism unlikely and a normal D- facts due to movements of the patient
(outpatients) and inpatients were eli- dimer test result, the diagnosis of or the heart or if there was insufficient
gible. Patients presenting to an outpa- pulmonary embolism was considered contrast enhancement of the pulmo-
tient office were sent directly to the excluded and anticoagulant treatment nary arteries. The management of
emergency department for evaluation. was withheld. Those patients who had patients in whom the CT could not be
Patients were recruited between No- a combination of clinical decision rule performed or who had an inconclusive
vember 2002 and September 2004. indicating pulmonary embolism un- CT scan was left to the discretion of
Exclusion criteria were treatment with likely with an abnormal D-dimer test the attending physician.
therapeutic doses of unfractionated or result, or who had a clinical decision The decision of the presence or ab-
low-molecular-weight heparin for more rule indicating pulmonary embolism sence of pulmonary embolism was
than 24 hours, life expectancy less than likely, underwent CT. made by trained attending radiolo-
3 months, pregnancy, geographic inac- gists who were blinded to any specific
cessibility precluding follow-up, age Radiological Evaluation patient clinical information. By proto-
younger than 18 years, allergy to intra- Computed tomography was per- col design they knew that a patient re-
venous contrast agents, renal insuffi- formed using either single-detector row ferred for CT either had a D-dimer level
ciency (creatinine clearance 30 mL/ or multidetector-row systems. Pa- that was above 500 ng/mL or a clinical
min [ 0.5 mL/s]), logistic reasons (eg, tients were examined during sus- decision rule score that was higher than
unavailability of CT, patient too ill to un- pended inspiration. The single- 4 points, but did not know which of
dergo CT scanning), or hemodynamic detector row CT parameters were 3-mm these items was the reason for perform-
instability. Five academic and 7 gen- slice thickness with a 2-mm reconstruc- ing a CT scan.
eral urban hospitals in the Netherlands tion interval at 120 kV/140 mAs, 120
participated. The institutional review to 140 mL of nonionic contrast mate- Outcome Measures
boards of all participating hospitals ap- rial containing 350 mg of iodine per mL The primary outcome of the study was
proved the study protocol, and written with an injection speed of 3.0 mL/s and the incidence of symptomatic VTE
or oral informed consent was obtained an injection delay of 16 seconds. Mul- events during 3 months of follow-up,
from all participants. tidetector-row CT parameters were defined as fatal pulmonary embolism,
1.25-mm slice thickness with a 1.2-mm nonfatal pulmonary embolism, or
Clinical Decision Rule reconstruction interval at 120 kV/120 deep vein thrombosis (DVT). An inde-
and D-Dimer Assay mAs, 80 to 100 mL of nonionic con- pendent adjudication committee,
Patients with clinically suspected pul- trast material containing 350 mg of io- whose members were unaware of the
monary embolism were evaluated by an dine per mL with an injection speed of patient’s allocation within the diagnos-
174 JAMA, January 11, 2006—Vol 295, No. 2 (Reprinted) ©2006 American Medical Association. All rights reserved.
Downloaded from www.jama.com by guest on July 10, 2008

5. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
tic algorithm, evaluated all suspected Sample size was based on an assump-
Table 2. Baseline Demographic and Clinical
VTE and deaths. A diagnosis of pul- tion of a 1% incidence of VTE in both Characteristics of Study Population
monary embolism or DVT was based patient groups5,9 and a goal to keep the (N = 3306)*
on a priori defined and generally upper limit of the 95% confidence in- Characteristic Value
accepted criteria.15 Deaths were classi- terval (CI) below 2.7%, which has been Age, mean (SD), y 53.0 (18.4)
fied as caused by pulmonary embolism reported as the upper limit of the range Female 1897 (57.4)
Outpatients 2701 (81.7)
in case of confirmation by autopsy, in of recurrent VTE after a normal angio- Duration of complaints, 2 (1-5)
case of an objective test positive for gram.16 We calculated that approxi- median (IQR), d
Paralysis 91 (2.8)
pulmonary embolism prior to death, mately 1000 patients would have to be Immobilization or recent surgery 610 (18.5)
or if pulmonary embolism could not included in each group, using a 2-sided Previous venous thromboembolism 480 (14.5)
COPD with treatment 341 (10.3)
be confidently excluded as the cause type I error of .05 and a type II error of Heart failure with treatment 243 (7.4)
of death. .20. Since we expected that approxi- Malignancy 476 (14.4)
Follow-up consisted of a scheduled mately 30% of patients would have a Estrogen use, women 438 (23.1)
Clinical symptoms of deep vein 190 (5.7)
outpatient visit or telephone inter- classification of pulmonary embolism thrombosis
view at 3 months. Patients were addi- unlikely by clinical decision rule and Heart rate 100/min 867 (26.2)
Hemoptysis 176 (5.3)
tionally instructed to contact the study a normal D-dimer test result,5 a total
Abbreviations: COPD, chronic obstructive pulmonary dis-
center or their general practitioner im- study population of 3300 patients was ease; IQR, interquartile range.
*Data are presented as number and percentage unless
mediately in the event of symptoms sug- needed. otherwise indicated.
gestive of DVT or pulmonary embo- Exact 95% CIs were calculated
lism. At each visit, information was around the observed incidences using
obtained on complaints suggestive of StatXact software, version 5 (Cytel Soft- dicating pulmonary embolism un-
VTE, including acute onset of dys- ware Corp, Cambridge, Mass). Descrip- likely and were tested for D-dimer con-
pnea, acute worsening of existing dys- tive parameters were calculated using centrations (Figure). The prevalence of
pnea, acute onset of chest pain, unilat- SPSS software, version 11.5 (SPSS Inc, pulmonary embolism in these pa-
eral leg swelling and leg pain, as well Chicago, Ill). For statistical differ- tients was 12.1% (266/2206; 95% CI,
as interval initiation of anticoagu- ences, the Fisher exact test was used; 10.7%-13.5%) vs 37.1% (408/1100;
lants. In case of clinically suspected statistical significance was set at P .05. 95% CI, 34.2%-40.0%) in those with a
DVT or pulmonary embolism, objec- clinical decision rule indicating pul-
tive diagnostic tests were required, in- RESULTS monary embolism likely (P .001).
cluding CUS for suspected DVT, and Study Patients Among the 1149 patients classified as
ventilation-perfusion scintigraphy or A total of 3503 consecutive patients unlikely but with an abnormal D-
CT for suspected pulmonary embo- with clinically suspected pulmonary dimer test result, the prevalence of pul-
lism. In case of death, information was embolism were screened, of whom 184 monary embolism was 23.2% (266/
obtained from the general practi- (5.3%) were excluded because of pre- 1149). D-dimer test results were normal
tioner, from the hospital records, or defined exclusion criteria: more than 24 in 1057 (32.0%) patients, and in these
from autopsy. hours of low-molecular-weight hepa- patients, pulmonary embolism was con-
rin (n = 50), life expectancy less than 3 sidered excluded. Of the 2206 pa-
Statistical Analysis months (n = 47), pregnancy (n = 26), tients undergoing D-dimer testing, 968
The 2 primary analyses were inci- geographic inaccessibility precluding (44%) had a VIDAS D-dimer test per-
dence of symptomatic VTE during fol- follow-up (n=20), renal insufficiency formed; 1238 patients (56%) had a Ti-
low-up, confirmed by objective test- (n = 26), logistic reasons (n = 10), age naquant D-dimer test.
ing, in (1) the group of patients in younger than 18 years (n=4), and al- Of the 2249 patients with either ab-
whom anticoagulant treatment was lergy to intravenous contrast agent normal D-dimer concentrations
withheld based on a classification of (n = 1). In addition, 13 patients re- (n=1149) or a clinical decision rule in-
pulmonary embolism unlikely by clini- fused consent (Figure). The final study dicating pulmonary embolism likely
cal decision rule and a normal D- population of 3306 participants in- (n=1100), 2199 underwent CT. In the
dimer test result, and (2) the group of cluded 2701 (81.7%) outpatients and other 50 patients a CT was indicated but
patients in whom anticoagulant treat- 605 (18.3%) inpatients; the baseline not performed because of lack of ve-
ment was withheld based on a CT scan demographic and clinical characteris- nous access, extreme obesity, DVT con-
that excluded pulmonary embolism. tics of the 3306 study patients are firmed by CUS prior to CT, or a dete-
Additional analyses were performed for shown in TABLE 2. riorating clinical condition prior to CT.
fatal pulmonary embolism in these Multidetector-row CT was performed
groups, as well as among the patients Results of Diagnostic Algorithm in 1939 patients and single-detector row
with a normal CT scan and an alterna- Of the 3306 included patients, 2206 CT in 260 patients. Computed tomog-
tive diagnosis on CT separately. (66.7%) had a clinical decision rule in- raphy excluded pulmonary embolism
©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, January 11, 2006—Vol 295, No. 2 175
Downloaded from www.jama.com by guest on July 10, 2008

6. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
0.2%-1.0%]); it was proven by au-
Table 3. Venous Thromboembolic Events (VTEs) During 3-Month Follow-up (n = 3138)*
topsy in 2 and attributed as the cause
Total VTEs, Fatal Pulmonary Embolism,
Variable No. No. (%) [95% CI] No. (%) [95% CI] of death in 5 (TABLE 4). Follow-up was
Pulmonary embolism unlikely 1028 5 (0.5) [0.2-1.1] 0 (0) [0.0-0.3] incomplete in one of the 1436 pa-
and normal D-dimer test result tients (0.1%). In a “worst case” sce-
Pulmonary embolism excluded by CT 1436 18 (1.3) [0.7-2.0] 7 (0.5) [0.2-1.0] nario in which this patient would have
CT normal 764 9 (1.2) [0.5-2.2] 3 (0.4) [0.1-1.1] developed VTE, the incidence of VTE
CT alternative diagnosis 672 9 (1.3) [0.6-2.5] 4 (0.6) [0.1-1.5] would have been 19 of 1436 (1.3%
Pulmonary embolism diagnosed by CT 674 20 (3) [1.8-4.6] 11 (1.6) [0.8-2.9] [95% CI, 0.8%-2.1%]).
Abbreviations: CI, confidence interval; CT, computed tomography. Rates of VTE during follow-up were
*A total of 168 patients were excluded due to treatment with anticoagulation outside of protocol, inconclusive CT, or
CT not performed. comparable for inpatients and outpa-
tients (1.4% ([95% CI, 0.4%-3.1%]) vs
1.2% [95% CI, 0.7%-2.1%], respec-
in 1505 patients (45.5% of the study [95% CI, 0.3%-1.4%]). There were no fa- tively). Among the patients who did not
population). In these patients, 702 tal pulmonary embolisms. Eight (0.8%) receive anticoagulants, similar inci-
(21.2% of the study population) had ad- of the 1057 patients died of other causes. dences of VTE were observed in those
ditional diagnostic information visual- Of the study population, 605 were in- with a normal CT scan (9/764 [1.2%]
ized on CT: pneumonia (n=212), pleu- patients, and 56 of these had a deci- {95% CI, 0.5%-2.2%}) and those with
ral effusion (n = 163), malignancy sion rule indicating pulmonary embo- additional diagnostic information on
(n=50), and other diagnoses (n=277). lism unlikely and a normal D-dimer test CT (9/672 [1.3%] {95% CI, 0.6%-
Pulmonary embolism was confirmed in result (9.3%). No VTE was observed at 2.5%}) (Table 3). Similar incidences of
674 patients (20.4% of the study popu- follow-up in these patients (VTE rate, VTE were observed in untreated pa-
lation). Computed tomography was in- 0% [95% CI, 0%-6.4%]). The results for tients who underwent multidetector-
conclusive in 20 patients (0.9%). Hence, inpatients and outpatients were com- row CT (14/1266 [1.1%] {95% CI,
the diagnostic algorithm could be com- parable (VTE rate, 0% [95% CI, 0%- 0.6%-1.9%}) vs single-detector row CT
pleted according to the protocol in 3256 6.4%] vs 0.5% [95% CI, 0.2%-1.2%]). (4/170 [2.4%] {95% CI, 0.6%-5.9%}).
patients (98.5%) and allowed a man- There were no significant differences Twenty patients returned with symp-
agement decision in 3236 patients between patients at academic and gen- toms of pulmonary embolism during
(97.9%). eral hospitals. follow-up. Computed tomography was
The VIDAS D-dimer assay had a true- again used as the diagnostic method in
Patients With Pulmonary negative rate of 44.2% (428/968 pa- 13 of these 20 patients and was nor-
Embolism Unlikely and tients) and the Tinaquant D-dimer as- mal in all. No VTE was demonstrated
Normal D-Dimer Test Result say had a true-negative rate of 50.8% at later follow-up.
Of the 1057 patients with the combina- (629/1238 patients) (P .002). The The overall mortality rate in pa-
tion of a clinical decision rule indicat- negative predictive values for the VIDAS tients in whom CT excluded pulmo-
ing pulmonary embolism unlikely and and Tinaquant assays were 100% (95% nary embolism was 8.6% (129 pa-
a normal D-dimer test result, 29 pa- CI, 99.1%-100%) and 99.2% (95% CI, tients).
tients (2.7%) were treated with oral an- 98.1%-99.7%), respectively.
ticoagulants during follow-up for vari- Patients With CT That Was
ous reasons other than VTE. Three of the Patients With CT Excluding Inconclusive or Not Performed
1028 remaining patients returned with Pulmonary Embolism Of the 20 patients with an inconclu-
symptomatic VTE events (2 nonfatal pul- Of the 1505 patients in whom CT ex- sive CT scan, pulmonary embolism was
monary embolism, 1 DVT) during the cluded pulmonary embolism, 69 (4.6%) demonstrated by ventilation-perfu-
3-month follow-up. In 25 patients, the received anticoagulants during fol- sion lung scan in 2 patients, and they
protocol was violated and a CT or a ven- low-up for various reasons other than received anticoagulant treatment. Dur-
tilation-perfusion scan was performed VTE. Of the 1436 patients who did not ing follow-up, 1 of the 18 remaining pa-
while not indicated. Pulmonary embo- receive anticoagulant treatment, 18 ex- tients had a nonfatal VTE event. Of the
lism was diagnosed in 2 of these 25 pa- perienced VTE events during the 50 patients in whom CT was indi-
tients. Therefore, the incidence of VTE 3-month follow-up (1.3% [95% CI, cated but not performed, 3 had pulmo-
was 5 of 1028 (0.5% [95% CI, 0.2%- 0.7%-2.0%]). Eleven of these patients nary embolism demonstrated by ven-
1.1%]) (TABLE 3). Two patients were lost had nonfatal symptomatic thrombo- tilation-perfusion lung scan, and 2
to follow-up (0.2%). In a “worst case” embolic events (3 pulmonary embo- patients had DVT demonstrated by
scenario, in which these 2 patients would lism and 8 DVT). Fatal pulmonary em- CUS; during follow-up, 1 of the re-
have developed VTE, the incidence of bolism was presumed to have occurred maining 45 patients had a fatal pulmo-
VTE would have been 7 of 1028 (0.7% in the other 7 patients (0.5% [95% CI, nary embolism, while DVT occurred in
176 JAMA, January 11, 2006—Vol 295, No. 2 (Reprinted) ©2006 American Medical Association. All rights reserved.
Downloaded from www.jama.com by guest on July 10, 2008

7. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
1 patient. The mortality rate for incon- pulmonary embolism. No further di- In contrast to our simple algorithm,
clusive CT was 5% (1/20) and for CT agnostic testing was necessary in the a recent study13 used a more complex
not performed, 14% (7/50). third of our patients who had an un- flowchart with sequential testing that
likely clinical probability score in com- included clinical probability assess-
Patients With Pulmonary bination with a normal D-dimer test re- ment, D-dimer assay, CUS, CT, as well
Embolism Confirmed by CT sult, with a 3-month incidence of VTE as pulmonary angiography to exclude
Of the 674 patients in whom CT dem- of 0.5%. Computed tomography effec- pulmonary embolism in patients with
onstrated pulmonary embolism, 20 pa- tively ruled out pulmonary embolism high likelihood and negative workup.
tients (3.0%) had a recurrent VTE de- in all other patients without using other As the authors pointed out, their study
spite anticoagulant treatment. This imaging tests (3-month incidence of was not a true outcome study, since
included 11 fatal pulmonary embo- VTE in those with a negative CT of CUS was performed in all patients
lism, 3 nonfatal pulmonary embolism, 1.3%). The algorithm was pragmatic in with abnormal D-dimer levels, and
and 6 DVT. One patient was lost to fol- that it could be completed in 98.5% of patients with abnormal CUS and a
low-up. The overall mortality in this the eligible patients and allowed a man- normal CT scan were treated with
group was 7.2% (55 patients). agement decision in 97.9%. anticoagulation. That study had a
Other management studies have smaller sample size (674 patients) and
COMMENT documented the safety of a low clini- a higher rate of exclusion (25% vs
This large cohort study of 3306 con- cal probability in combination with a 5.6% in our study).
secutive patients with clinically sus- normal D-dimer concentration for the To improve the simplicity and util-
pected pulmonary embolism demon- exclusion of pulmonary embo- ity of their decision rule, Wells et al
strates that the use of a diagnostic lism.3-5,17 In these studies, the rate of proposed changing their model from
algorithm consisting of a dichoto- VTE during follow-up ranged from 0% the original 3 categories (low, moder-
mous decision rule, D-dimer testing, to 1.5%. However, because the sample ate, high) to 2 categories (pulmonary
and CT scan can guide treatment de- size was limited, upper confidence lim- embolism unlikely and pulmonary
cisions with a low risk for subsequent its were as high as 6.0%.3-5,15 embolism likely).3 Our study is the
Table 4. Deaths Attributed to Pulmonary Embolism
Results of
Patient Computed Anticoagulant Past Time of Death
Sex Age, y Tomography Therapy Medical History After Enrollment , d Circumstances of Death
Male 60 Normal No COPD, alcohol abuse 3 Sudden death at home
Female 65 Alternative diagnosis: No Colon cancer; multiple 18 Dehydration due to
pulmonary metastases in liver, chemotherapy-induced
metastases spleen, adrenal glands diarrhea; morphine for pain
complaints; sudden death
Male 46 Normal No Multiple myeloma 40 Bedridden due to complaints
of pain associated
with myeloma; sudden death
at home; autopsy result:
pulmonary embolism
Female 69 Alternative diagnosis: No Progressive dyspnea 41 Computed tomography at day 34
interstitial in past half year showed pulmonary embolism;
pneumonia due to interstitial progressive respiratory
pneumonia insufficiency; ventilator
dependency; palliative care;
autopsy result: pulmonary
embolism and bilateral pneumonia
Female 60 Alternative diagnosis: No COPD, breast cancer 75 Immobilization in electric wheelchair
pericarditis in nursing home; gradual
carcinomatosa worsening, cardiac failure
due to pericarditis
Female 77 Alternative diagnosis: No Hypertension 86 Collapse on street
pneumonia; at review, with swollen face
a subsegmental
pulmonary embolism
had been missed
at inclusion
Female 31 Normal No Pulmonary embolism 94 Antibiotics for CAPD peritonitis;
in 2002, diabetes, sudden death
renal insufficiency,
estrogen use
Abbreviations: CAPD, continuous ambulatory peritoneal dialysis; COPD, chronic obstructive pulmonary disease.
©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, January 11, 2006—Vol 295, No. 2 177
Downloaded from www.jama.com by guest on July 10, 2008

8. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
first to prospectively validate the safety treated and further follow-up was un- Authors/Christopher Study Writing Group: Arne van
Belle, MD, Department of Pulmonary Medicine, Aca-
of the dichotomized score in combina- eventful, making this unlikely. demic Hospital, Maastricht; Harry R. Buller, MD, PhD,
¨
tion with the D-dimer assay. Com- Fourth, the use of multidetector- Department of Vascular Medicine, Academic Medi-
cal Center, Amsterdam; Menno V. Huisman, MD, De-
pared with a combination using the row CT has the potential for overdiag- partment of General Internal Medicine/Endocrinol-
3-category classification, this approach nosis by imaging very small periph- ogy, Leiden University Medical Center, Leiden; Peter
has the potential to increase the num- eral subsegmental emboli. Because M. Huisman, MD, Department of Radiology, Hilver-
sum Hospital, Hilversum; Karin Kaasjager, MD, De-
ber of patients in whom pulmonary patients did not undergo confirma- partment of Internal Medicine, Rijnstate Hospital, Arn-
embolism can be excluded by approxi- tory pulmonary angiography, our study hem; Pieter W. Kamphuisen, MD, Department of
General Internal Medicine, Radboud University Medi-
mately 50%.3,17 design did not permit assessing the cal Center, Nijmegen; Mark H. H. Kramer, MD, De-
Despite concerns that the sensitivity false-positive rate of CT scans. Only partment of Internal Medicine, Meander Medical
Center, Amersfoort; Marieke J. H. A. Kruip, MD, De-
of CT for pulmonary embolism is lower 10% of our patients underwent single- partment of Hematology, Erasmus University Medi-
than that of pulmonary angiogra- detector row CT, so we could not make cal Center, Rotterdam; Johanna M. Kwakkel-van Erp,
MD, Department of Pulmonary Medicine, Rijnstate
phy,18,19 the observed risk of subse- a meaningful comparison of the im- Hospital, Arnhem; Frank W. G. Leebeek, MD, De-
quent symptomatic VTE in those pa- pact of each test. However, the overall partment of Hematology, Erasmus University Medi-
tients in whom pulmonary embolism prevalence of pulmonary embolism in cal Center, Rotterdam; Mathilde Nijkeuter, MD, De-
partment of General Internal Medicine/Endocrinology,
was excluded by CT was comparable to our study (20%) is comparable to the Leiden University Medical Center, Leiden; Martin H.
the risk reported after a normal pulmo- prevalence in a previous multicenter Prins, MD, Department of Clinical Epidemiology, Aca-
demic Hospital, Maastricht; Maaike Sohne, MD, De-
nary angiogram (3-month incidence, study performed with single-detector partment of Vascular Medicine, Academic Medical
1.3% [95% CI, 0.7%-2.0%] vs 1.7% [95% row CT (24%).9 This does not support Center, Amsterdam; and Lidwine W. Tick, MD, De-
partment of Hematology, Leiden University Medical
CI, 1.0%-2.7%],16 respectively). In ad- a concern that multidetector-row CT Center, Leiden; all in the Netherlands.
dition, in our study fatal pulmonary em- technology will lead to a high number Author Contributions: Drs Huisman and Prins had
bolism occurred in 0.5% (95% CI, 0.2%- of false-positive results. full access to all of the data in the study and take
responsibility for the integrity of the data and the
1.0%) of patients in whom CT had Finally, a definitive cause of death accuracy of the data analysis.
excluded pulmonary embolism, com- could not be established for all pa- Study concept and design: Belle, Buller, M. V. Huisman,
¨
P. Huisman, Kaasjager, Kamphuisen, Kramer, Kruip,
pared with 0.3% (95% CI, 0.02%- tients with normal test results who died Kwakkel-van Erp, Leebeek, Nijkeuter, Prins, Sohne,
0.7%) after normal pulmonary angiog- during follow-up. However, pulmo- Tick.
raphy.16 Computed tomography has the nary embolism was assigned as the Acquisition of data: Belle, Buller, M. V. Huisman,
¨
P. Huisman, Kaasjager, Kamphuisen, Kramer, Kruip,
potential advantage of providing addi- cause of death if it could not be confi- Kwakkel-van Erp, Leebeek, Nijkeuter, Prins, Sohne,
tional diagnostic information for the pre- dently excluded, a conservative assump- Tick.
Analysis and interpretation of data: Belle, Buller, ¨
senting symptoms in patients without tion that strengthens our conclusions M. V. Huisman, P. Huisman, Kaasjager, Kamphuisen,
pulmonary embolism. about low risk for this strategy. Kramer, Kruip, Kwakkel-van Erp, Leebeek, Nijkeuter,
Prins, Sohne, Tick.
Several potential limitations in our The generalizability of our findings Drafting of the manuscript: Belle, Bu ller, M. V.
¨
study require comment. First, the should be considered. The baseline clini- Huisman, P. Huisman, Kaasjager, Kamphuisen, Kramer,
absence of pulmonary embolism was cal characteristics and the incidence of Kruip, Kwakkel-van Erp, Leebeek, Nijkeuter, Prins,
Sohne, Tick.
not verified by pulmonary angiogra- pulmonary embolism for our study Critical revision of the manuscript for important in-
phy. However, the clinical outcome population are comparable with those tellectual content: Belle, Bu ller, M. V. Huisman,
¨
P. Huisman, Kaasjager, Kamphuisen, Kramer, Kruip,
after a 3-month follow-up is widely observed in other population-based stud- Kwakkel-van Erp, Leebeek, Nijkeuter, Prins, Sohne,
accepted as an appropriate alternative ies, except for a somewhat younger mean Tick.
to establish the safety of a diagnostic age.5,10,12 The low proportion of pa- Statistical analysis: Prins.
Obtained funding: Belle, Buller, M. V. Huisman,
¨
strategy, given a near-complete follow- tients excluded and the enrollment of P. Huisman, Kaasjager, Kamphuisen, Kramer, Kruip,
up.20 consecutive patients who were referred Kwakkel-van Erp, Leebeek, Nijkeuter, Prins, Sohne,
Tick.
Second, while our cohort study has to both academic and nonacademic hos- Administrative, technical, or material support: Belle,
the strength of minimal loss to fol- pitals further supports broad applicabil- Bu ller, M. V. Huisman, P. Huisman, Kaasjager,
¨
Kamphuisen, Kramer, Kruip, Kwakkel-van Erp,
low-up (3 patients, 0.1%) and indepen- ity of these results, as does the similar Leebeek, Nijkeuter, Prins, Sohne, Tick.
dent blinded adjudication of all out- rates of VTE during follow-up between Study supervision: Belle, Bu ller, M. V. Huisman,
¨
comes, a randomized controlled study inpatients and outpatients. P. Huisman, Kaasjager, Kamphuisen, Kramer, Kruip,
Kwakkel-van Erp, Leebeek, Nijkeuter, Prins, Sohne,
design would have allowed a direct com- In conclusion, a diagnostic manage- Tick.
parison to other validated strategies. ment strategy using a simple clinical de- Financial Disclosures: None reported.
Funding/Support: This study was supported in
Third, CT was again used to exclude cision rule, D-dimer testing, and CT is part by unrestricted grants from the participating
pulmonary embolism in 13 of 20 pa- as effective as other more complex di- hospitals.
tients who returned during follow-up agnostic strategies in the evaluation and Role of the Sponsors: The boards of the respective hos-
pitals did not have any specific role in the design and
with symptoms after CT had excluded management of patients with clini- conduct of the study; in the collection, management,
pulmonary embolism at baseline. Al- cally suspected pulmonary embolism. analysis, and interpretation of the data; or in the prepa-
ration, review, or approval of the manuscript.
though these could represent false- Its use is associated with low risk for Adjudication Committee: K. Hamulyak, H. ten Cate,
negative results, these patients were not subsequent fatal and nonfatal VTE. Academic Hospital Maastricht, the Netherlands.
178 JAMA, January 11, 2006—Vol 295, No. 2 (Reprinted) ©2006 American Medical Association. All rights reserved.
Downloaded from www.jama.com by guest on July 10, 2008

9. ALGORITHM FOR MANAGING PULMONARY EMBOLISM
Christopher Study Investigators: In addition to the Amphia Hospital, Breda: C. van Guldener, J. Y. Hospital, Meppel: H. de Korte, C. B. Kos, L. Later-
Writing Committee, the following institutions and in- Mijnsbergen, M. F. A. M. Sturm; Spaarne Hospital, veer, W. C. J. van Veldhuizen; Radboud University
vestigators participated in the study: Meander Medi- Haarlem: C. de Swart, P. M. Kuijer, J. G. Schrama, Medical Center, Nijmegen: S. J. H. Bredie, C. E.
cal Center, Amersfoort: C. J. M. Halkes, B. Heggel- A.v.d. Velde; Hilversum Hospital, Hilversum: M. M. van Die, Y. F. Heijdra, J. W. M. Lenders; Erasmus Medi-
man, M. Nix; Academic Medical Center, Amster- van der Eerden, P. J. H. Janssen, R. Jansen, S. Lo- cal Center, Rotterdam: K-S. G. Jie, A. H. Kars, A. H.
dam: P. Bresser, D. R. Kool, S. S. K. S. Phoa, B. Rekke; batto; Leiden University Medical Center, Leiden: van den Meiracker, P. M. T. Pattynama; Medical
Rijnstate Hospital, Arnhem: H. M. H. Grandjean, E. A. Compier, H. C. J. Eikenboom, A. de Roos; Aca- Center Rijnmond Zuid, Rotterdam: J. M. de Borst,
F. O. H. W. Kesselring, J. J. Mol, E. F. Ullmann; demic Hospital, Maastricht: G. Snoep; Diakonessen A. van Houten, H. T. Teng.
REFERENCES
1. Lee AY, Hirsh J. Diagnosis and treatment of 8. Mullins MD, Becker DM, Hagspiel KD, et al. The Perrier A. Assessing clinical probability of pulmonary
venous thromboembolism. Annu Rev Med. 2002;53: role of spiral volumetric computed tomography in the embolism in the emergency ward: a simple score. Arch
15-33. diagnosis of pulmonary embolism. Arch Intern Med. Intern Med. 2001;161:92-97.
2. The PIOPED Investigators. Value of the ventilation/ 2000;160:293-298. 15. Buller HR, Davidson BL, Decousus H, et al. Sub-
¨
perfusion scan in acute pulmonary embolism: results 9. van Strijen MJ, de Monye W, Schiereck J, et al. cutaneous fondaparinux versus intravenous unfrac-
of the Prospective Investigation of Pulmonary Embo- Single-detector helical computed tomography as the tionated heparin in the initial treatment of pulmo-
lism Diagnosis (PIOPED). JAMA. 1990;263:2753-2759. primary diagnostic test in suspected pulmonary nary embolism. N Engl J Med. 2003;349:1695-1702.
3. Wells PS, Anderson DR, Rodger M, et al. Deriva- embolism: a multicenter clinical management 16. van Beek EJ, Brouwerst EM, Song B, et al. Clini-
tion of a simple clinical model to categorize patients’ study of 510 patients. Ann Intern Med. 2003;138: cal validity of a normal pulmonary angiogram in pa-
probability of pulmonary embolism: increasing the 307-314. tients with suspected pulmonary embolism—a criti-
model’s utility with the SimpliRED D-dimer. Thromb 10. Musset D, Parent F, Meyer G, et al. Diagnostic cal review. Clin Radiol. 2001;56:838-842.
Haemost. 2000;83:416-420. strategy for patients with suspected pulmonary em- 17. Wolf SJ, McCubbin TR, Feldhaus KM, et al. Pro-
4. Kruip MJ, Slob MJ, Schijen JH, et al. Use of a clini- bolism: a prospective multi-center outcome study. spective validation of Wells criteria in the evaluation
cal decision rule in combination with D-dimer con- Lancet. 2002;360:1914-1920. of patients with suspected pulmonary embolism. Ann
centration in diagnostic workup of patients with 11. Moores LK, Jackson WL Jr, Shorr AF, et al. Meta- Emerg Med. 2004;44:503-510.
suspected pulmonary embolism: a prospective man- analysis: outcomes in patients with suspected pulmo- 18. Van Strijen MJ, De Monye W, Kieft GJ, Patty-
agement study. Arch Intern Med. 2002;162:1631- nary embolism managed with computed tomo- nama PM, Prins MH, Huisman MV. Accuracy of
1635. graphic pulmonary angiography. Ann Intern Med. single-detector spiral CT in the diagnosis of pulmo-
5. Ten Wolde M, Hagen PJ, MacGillavry MR, et al. 2004;141:866-874. nary embolism: a prospective multicenter cohort
Non-invasive diagnostic work-up of patients with clini- 12. Perrier A, Roy PM, Aujesky D, et al. Diagnosing study of consecutive patients with abnormal perfu-
cally suspected pulmonary embolism: results of a man- pulmonary embolism in outpatients with clinical as- sion scintigraphy. J Thromb Haemost. 2005;3:
agement study. J Thromb Haemost. 2004;2:1110- sessment, D-dimer measurement, venous ultra- 17-25.
1117. sound, and helical computed tomography: a multi- 19. Perrier A, Bounameaux H. Validation of helical
6. Schoepf UJ, Goldhaber SZ, Costello P. Spiral com- center management study. Am J Med. 2004;116:291- computed tomography for suspected pulmonary
puted tomography for acute pulmonary embolism. 299. embolism: a near miss? J Thromb Haemost. 2005;3:
Circulation. 2004;109:2160-2167. 13. Perrier A, Roy P-M, Sanchez O, et al. Multide- 14-16.
7. Rathbun SW, Raskob GE, Whitsett TL. Sensitivity tector-row computed tomography in suspected pul- 20. Kruip MJ, Leclercq MG, van der Heul C, et al. Di-
and specificity of helical computed tomography in the monary embolism. N Engl J Med. 2005;352:1760- agnostic strategies for excluding pulmonary embo-
diagnosis of pulmonary embolism: a systematic review. 1768. lism in clinical outcome studies: a systematic review.
Ann Intern Med. 2000;132:227-232. 14. Wicki J, Perneger TV, Junod AF, Bounameaux H, Ann Intern Med. 2003;138:941-951.
©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, January 11, 2006—Vol 295, No. 2 179
Downloaded from www.jama.com by guest on July 10, 2008