Download pdf - Diagnosis of DVT

7/29/2019 Diagnosis of DVT

1/68

CHEST Supplement

www.chestpubs.org CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e351S

ANTITHROMBOTIC THERAPY AND PREVENTION OF THROMBOSIS, 9TH ED: ACCP GUIDELINES

Summary of Recommendations

Note on Shaded Text: Throughout this guideline,shading is used within the summary of recommenda-tions sections to indicate recommendations that arenewly added or have been changed since the publica-tion of Antithrombotic and Thrombolytic Therapy:American College of Chest Physicians Evidence-BasedClinical Practice Guidelines (8th Edition). Recom-mendations that remain unchanged are not shaded.

3.1. In patients with a suspected first lowerextremity DVT, we suggest that the choice ofdiagnostic tests process should be guided by the

clinical assessment of pretest probability ratherthan by performing the same diagnostic tests in

all patients (Grade 2B).

Note: In considering this recommendation, five pan-elists voted for a strong recommendation and four votedfor a weak recommendation (one declined to vote andtwo did not participate). According to predeterminedcriteria, this resulted in weak recommendation.

3.2. In patients with a low pretest probability offirst lower extremity DVT, we recommend oneof the following initial tests: (i) a moderatelysensitive D-dimer, (ii) a highly sensitive D-dimer,

Background: Objective testing for DVT is crucial because clinical assessment alone is unreliableand the consequences of misdiagnosis are serious. This guideline focuses on the identification ofoptimal strategies for the diagnosis of DVT in ambulatory adults.

Methods: The methods of this guideline follow those described in Methodology for the Develop-ment of Antithrombotic Therapy and Prevention of Thrombosis Guidelines: AntithromboticTherapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines.

Results: We suggest that clinical assessment of pretest probability of DVT, rather than performingthe same tests in all patients, should guide the diagnostic process for a first lower extremity DVT(Grade 2B). In patients with a low pretest probability of first lower extremity DVT, we recom-mend initial testing with D-dimer or ultrasound (US) of the proximal veins over no diagnostictesting (Grade 1B), venography (Grade 1B), or whole-leg US (Grade 2B). In patients with mod-erate pretest probability, we recommend initial testing with a highly sensitive D-dimer, proximalcompression US, or whole-leg US rather than no testing (Grade 1B) or venography (Grade 1B).

In patients with a high pretest probability, we recommend proximal compression or whole-leg USover no testing (Grade 1B) or venography (Grade 1B).Conclusions: Favored strategies for diagnosis of first DVT combine use of pretest probability assess-ment, D-dimer, and US. There is lower-quality evidence available to guide diagnosis of recurrent DVT,upper extremity DVT, and DVT during pregnancy. CHEST 2012; 141(2)(Suppl):e351Se418S

Abbreviations: aOR5adjusted OR; CUS5compression ultrasonography; GRADE5Grades of Recommendation,Assessment, Development, and Evaluation; IPG5 impedance plethysmography; MR5magnetic resonance; PE5pul-monary embolism; US5ultrasonography

Diagnosis of DVT

Antithrombotic Therapy and Prevention of Thrombosis,9th ed: American College of Chest Physicians

Evidence-Based Clinical Practice Guidelines

Shannon M. Bates, MDCM; Roman Jaeschke, MD; Scott M. Stevens, MD;Steve Goodacre, MBChB, PhD; Philip S. Wells, MD; Matthew D. Stevenson, PhD;Clive Kearon, MD, PhD; Holger J. Schunemann, MD, PhD, FCCP; Mark Crowther, MD;Stephen G. Pauker, MD; Regina Makdissi, MD; and Gordon H. Guyatt, MD, FCCP


2/68

e352S Diagnosis of DVT

raphy(Grade 1B for all comparisons). If the prox-imal CUS is negative, we recommend no furthertesting compared with (i) repeat proximal CUSafter 1 week, (ii) whole-leg US, or (iii) venog-raphy(Grade 1B for all comparisons).

If the D-dimer is positive, we suggest furthertesting with CUS of the proximal veins rather

than (i) whole-leg US (Grade 2C) or (ii) venog-raphy(Grade 1B). If CUS of the proximal veins ispositive, we suggest treating for DVT and per-forming no further testing over performingconfirmatory venography(Grade 2C).

Remarks: In circumstances when high-quality venog-raphy is available, patients who are not averse to thediscomfort of venography, are less concerned aboutthe complications of venography, and place a high

value on avoiding treatment of false-positive resultsare likely to choose confirmatory venography if find-ings for DVT are less certain (eg, a short segment of

venous noncompressibility).

3.3. In patients with a moderate pretest proba-bility of first lower extremity DVT, we recom-mend one of the following initial tests: (i) ahighly sensitive D-dimer or (ii) proximal CUS,or (iii) whole-leg US rather than (i) no testing(Grade 1B for all comparisons) or (ii) venography(Grade 1B for all comparisons). We suggest initialuse of a highly sensitive D-dimer rather than US(Grade 2C).

Remarks: The choice between a highly sensitiveD-dimer test or US as the initial test will depend onlocal availability, access to testing, costs of testing,and the probability of obtaining a negative D-dimerresult if DVT is not present. Initial testing with USmay be preferred if the patient has a comorbid condi-tion associated with elevated D-dimer levels and islikely to have a positive D-dimer result even if DVTis absent. Whole-leg US may be preferred in patientsunable to return for serial testing and those withsevere symptoms consistent with calf DVT. In patients

with suspected first lower extremity DVT in whom

US is impractical (eg, when leg casting or excessivesubcutaneous tissue or fluid prevent adequate assess-ment of compressibility) or nondiagnostic, we suggestCT scan venography, MR venography, or MR directthrombus imaging could be used as an alternative to

venography.

If the highly sensitive D-dimer is negative, werecommend no further testing over further inves-tigation with (i) proximal CUS, (ii) whole-leg US,or (iii) venography(Grade 1B for all comparisons).If the highly sensitive D-dimer is positive, we

or (iii) compression ultrasound (CUS) of theproximal veins rather than (i) no diagnostic testing(Grade 1B for all comparisons), (ii) venography(Grade 1B for all comparisons), or (iii) whole-legultrasound (US) (Grade 2B for all comparisons).

We suggest initial use of a moderately sensitive(Grade 2C) or highly sensitive (Grade 2B) D-dimerrather than proximal CUS.

Remarks: The choice between a moderately sensitiveD-dimer test, a highly sensitive D-dimer test, or prox-imal CUS as the initial test will depend on local avail-ability, access to testing, costs of testing, and theprobability of obtaining a negative D-dimer result ifDVT is not present. Initial testing with US would bepreferred if the patient has a comorbid conditionassociated with elevated D-dimer levels and is likelyto have a positive D-dimer result, even if DVT is absent.In patients with suspected first lower extremity DVTin whom US is impractical (eg, when leg casting or

excessive subcutaneous tissue or fluid prevent ade-quate assessment of compressibility) or nondiagnos-tic, we suggest CT scan venography or magneticresonance (MR) venography, or MR direct thrombusimaging could be used as an alternative to venography.

If the D-dimer is negative, we recommend nofurther testing over further investigation with(i) proximal CUS, (ii) whole-leg US, or (iii) venog-

Revision accepted August 31, 2011.Affiliations: From the Department of Medicine (Drs Bates andCrowther), McMaster University and Thrombosis and Athero-

sclerosis Research Institute; the Departments of Medicine andClinical Epidemiology and Biostatistics (Drs Jaeschke, Schune-mann, and Guyatt), McMaster University, Hamilton, ON, Can-ada; the Department of Medicine (Dr Stevens), IntermountainMedical Center, Murray, UT; the School of Health and RelatedResearch (Drs Goodacre and Stevenson), University of Sheffield,Sheffield, England; the Department of Medicine (Dr Wells), Uni-

versity of Ottawa, Ottawa, ON, Canada; the Department of Med-icine (Dr Pauker), Tufts New England Medical Center, Boston,MA; and the Department of Medicine (Dr Makdissi), Universityof Buffalo, Buffalo, NY.Funding/Support: The Antithrombotic Therapy and Preventionof Thrombosis, 9th ed: American College of Chest PhysiciansEvidence-Based Clinical Practice Guidelines received support fromthe National Heart, Lung, and Blood Institute [R13 HL104758]and Bayer Schering Pharma AG. Support in the form of educa-

tional grants was also provided by Bristol-Myers Squibb; Pfizer,Inc; Canyon Pharmaceuticals; and sanofi-aventis US.Disclaimer: American College of Chest Physician guidelines areintended for general information only, are not medical advice,and do not replace professional medical care and physician advice,

which always should be sought for any medical condition. Thecomplete disclaimer for this guideline can be accessed at http://chestjournal.chestpubs.org/content/141/2_suppl/1S.Correspondence to: Shannon M. Bates, MDCM, HSC 3W11,Department of Medicine, 1280 Main St W, Hamilton, ON, L8S4K1, Canada; e-mail: [email protected] 2012 American College of Chest Physicians. Reproductionof this article is prohibited without written permission from theAmerican College of Chest Physicians (http://www.chestpubs.org/site/misc/reprints.xhtml).DOI: 10.1378/chest.11-2299
http://chestjournal.chestpubs.org/content/141/2_suppl/1Shttp://chestjournal.chestpubs.org/content/141/2_suppl/1Smailto:[email protected]://www.chestpubs.org/site/misc/reprints.xhtmlhttp://www.chestpubs.org/site/misc/reprints.xhtmlhttp://www.chestpubs.org/site/misc/reprints.xhtmlhttp://www.chestpubs.org/site/misc/reprints.xhtmlhttp://www.chestpubs.org/site/misc/reprints.xhtmlmailto:[email protected]://chestjournal.chestpubs.org/content/141/2_suppl/1Shttp://chestjournal.chestpubs.org/content/141/2_suppl/1S


3/68


imaged to exclude isolated iliac DVT. In patientswith suspected first lower extremity DVT in whomUS is impractical (eg, when leg casting or excessivesubcutaneous tissue or fluid prevent adequate assess-ment of compressibility) or nondiagnostic, we suggestCT scan venography, MR venography, or MR directthrombus imaging could be used as an alternative to

venography.

If proximal CUS or whole-leg US is positive forDVT, we recommend treatment rather than con-firmatory venography(Grade 1B).

In patients with a negative proximal CUS, werecommend additional testing with a highly sen-sitive D-dimer or whole-leg US or repeat prox-imal CUS in 1 week over no further testing(Grade 1B for all comparisons) or venography(Grade 2B for all comparisons). We recommendthat patients with a single negative proximalCUS and positive D-dimer undergo whole-leg

US or repeat proximal CUS in 1 week overno further testing (Grade 1B) or venography(Grade 2B). In patients with negative serial prox-imal CUS, a negative single proximal CUS andnegative highly sensitive D-dimer, or a negative

whole-leg US, we recommend no further testingover venography or additional US (Grade 1B fornegative serial proximal CUS and for negative singleproximal CUS and highly sensitive D-dimer; Grade 2Bfor negative whole-leg US).

We recommend that in patients with high pre-

test probability, moderately or highly sensitiveD-dimer assays should not be used as stand-alone tests to rule out DVT (Grade 1B).

3.5. If risk stratification is not performed inpatients with suspected first lower extremity DVT,

we recommend one of the following initial tests:(i) proximal CUS or (ii) whole-leg US rather than(i) no testing (Grade 1B), (ii) venography(Grade 1B),or D-dimer testing (Grade 2B).

Remarks: Whole-leg US may be preferred to prox-imal CUS in patients unable to return for serial testing

and those with severe symptoms consistent with calfDVT or risk factors for extension of distal DVT. Inpatients with suspected first lower extremity DVT in

whom US is impractical (eg, when leg casting orexcessive subcutaneous tissue or fluid prevent ade-quate assessment of compressibility) or nondiagnostic,

we suggest that CT scan venography, MR venog-raphy, or MR direct thrombus imaging could be usedas an alternative to venography.

We recommend that patients with a negative prox-imal CUS undergo testing with a moderate- or

recommend proximal CUS or whole-leg USrather than no testing (Grade 1B for all compari-sons) or venography(Grade 1B for all comparisons).

If proximal CUS is chosen as the initial test andis negative, we recommend (i) repeat proximalCUS in 1 week or (ii) testing with a moderate orhighly sensitive D-dimer assay over no further

testing (Grade 1C) or venography(Grade 2B). Inpatients with a negative proximal CUS but apositive D-dimer, we recommend repeat prox-imal CUS in 1 week over no further testing(Grade 1B) or venography(Grade 2B).

In patients with (i) negative serial proximal CUSor (ii) a negative single proximal CUS and nega-tive moderate or highly sensitive D-dimer, werecommend no further testing rather than fur-ther testing with (i) whole-leg US or (ii) venog-raphy(Grade 1B for all comparisons).

If whole-leg US is negative, we recommend nofurther testing over (i) repeat US in one week,(ii) D-dimer testing, or (iii) venography(Grade 1Bfor all comparisons). If proximal CUS is positive,

we recommend treating for DVT rather thanconfirmatory venography(Grade 1B). If isolateddistal DVT is detected on whole-leg US, we sug-gest serial testing to rule out proximal exten-sion over treatment (Grade 2C).

Remarks: Patients with abnormal isolated distal USfindings on whole-leg US who place a high value onavoiding the inconvenience of repeat testing and alow value on avoiding treatment of false-positiveresults are likely to choose treatment over repeat US.Patients with severe symptoms and risk factors forextension as outlined in Perioperative Management ofAntithrombotic Therapy. Antithrombotic Therapyand Prevention of Thrombosis, 9th ed: AmericanCollege of Chest Physicians Evidence-Based ClinicalPractice Guidelines are more likely to benefit fromtreatment over repeat US.

3.4. In patients with a high pretest probability

of first lower extremity DVT, we recommendeither (i) proximal CUS or (ii) whole-leg US overno testing (Grade 1B for all comparisons) or venog-raphy(Grade 1B for all comparisons).

Remarks: Whole-leg US may be preferred to prox-imal CUS in patients unable to return for serial testingand those with severe symptoms consistent with calfDVT. In patients with extensive unexplained legswelling, if there is no DVT on proximal CUS or

whole-leg US and D-dimer testing has not beenperformed or is positive, the iliac veins should be


4/68


of,2 mm), we suggest at least one furtherproximal CUS (day 71) or testing with a mod-erately or highly sensitive D-dimer (followed byrepeat CUS [day 71] if positive) rather thanno further testing or venography(Grade 2B).

Remarks: In patients with an abnormal proximal CUSat presentation that does not meet the criteria for the

diagnosis of recurrence, an additional proximal CUSon day 21 in addition to that on (day 71) may bepreferred. Patients who place a high value on anaccurate diagnosis and a low value on avoiding theinconvenience and potential side effects of a venog-raphy are likely to choose venography over misseddiagnosis (in the case of residual diameter increaseof,2 mm).

We recommend that patients with suspectedrecurrent lower extremity DVT and a negativehighly sensitive D-dimer or negative proximal

CUS and negative moderately or highly sensi-tive D-dimer or negative serial proximal CUSundergo no further testing for suspected recur-rent DVT rather than venography(Grade 1B).

If CUS of the proximal veins is positive, we rec-ommend treating for DVT and performing nofurther testing over performing confirmatoryvenography(Grade 1B for the finding of a new non-compressible segment in the common femoral or pop-liteal vein, Grade 2B for a 4-mm increase in venousdiameter during compression compared with that in

the same venous segment on a previous result).

Remarks: Patients with US abnormalities at pre-sentation that do not include a new noncompressiblesegment who place a high value on an accurate diag-nosis and a low value on avoiding the inconvenienceand potential side effects of a venography are likelyto choose venography over treatment (in the caseof4-mm increase in venous diameter).

4.2. In patients with suspected recurrent lowerextremity DVT and abnormal but nondiagnostic

US results (eg, an increase in residual venousdiameter of,4 but2 mm), we recommendfurther testing with venography, if available(Grade 1B); serial proximal CUS (Grade 2B) ortesting with a moderately or highly sensitiveD-dimer with serial proximal CUS as above if thetest is positive (Grade 2B), as opposed to othertesting strategies or treatment.

4.3. In patients with suspected recurrent ipsilat-eral DVT and an abnormal US without a priorresult for comparison, we recommend further

high-sensitivity D-dimer, whole-leg US, or repeatproximal CUS in 1 week over no further testing(Grade 1B) or venography(Grade 2B). In patients

with a negative proximal CUS, we suggestD-dimer rather than routine serial CUS (Grade 2B)or whole-leg US (Grade 2C). We recommendthat patients with a single negative proximalCUS and positive D-dimer undergo further

testing with repeat proximal CUS in 1 week orwhole-leg US rather than no further testing(Grade 1B for both comparisons).

We recommend that in patients with (i) nega-tive serial proximal CUS, (ii) a negative D-dimerfollowing a negative initial proximal CUS, or (iii)negative whole-leg US, no further testing be per-formed rather than venography(Grade 1B).

If proximal US is positive for DVT, we recom-mend treatment rather than confirmatory venog-raphy(Grade 1B). If isolated distal DVT is detectedon whole-leg US, we suggest serial testing torule out proximal extension over treatment(Grade 2C).

Remarks: Patients with abnormal isolated distal USfindings on whole-leg US who place a high value onavoiding the inconvenience of repeat testing and a low

value on avoiding treatment of false-positive results arelikely to choose treatment over repeat US. Patients withsevere symptoms and risk factors for extension as out-lined in Perioperative Management of AntithromboticTherapy. Antithrombotic Therapy and Prevention of

Thrombosis, 9th ed: American College of Chest Physi-cians Evidence-Based Clinical Practice Guidelines aremore likely to benefit from treatment over repeat US.

3.6. In patients with suspected first lowerextremity DVT, we recommend against the rou-tine use of CT venography or MRI (Grade 1C).

4.1. In patients suspected of having recurrentlower extremity DVT, we recommend initialevaluation with proximal CUS or a highly sensi-tive D-dimer over venography, CT venography,

or MRI (all Grade 1B).Remarks: Initial D-dimer testing with a high-sensitivityassay is preferable if prior US is not available forcomparison.

If the highly sensitive D-dimer is positive, we rec-ommend proximal CUS over venography, CTvenography, or MRI (Grade 1B for all comparisons).

In patients with suspected recurrent lowerextremity DVT in whom initial proximal CUS isnegative (normal or residual diameter increase


5/68


testing with venography, if available (Grade 1B) ora highly sensitive D-dimer(Grade 2B) over serialproximal CUS. In patients with suspected recur-rent ipsilateral DVT and an abnormal US withoutprior result for comparison and a negative highlysensitive D-dimer, we suggest no further testingover venography(Grade 2C). In patients with sus-pected recurrent ipsilateral DVT and an abnor-

mal US without prior result for comparison anda positive highly sensitive D-dimer, we suggestvenography if available over empirical treatmentof recurrence (Grade 2C).

Remarks: Patients who place a high value on avoidingthe inconvenience and potential side effects of a

venography are likely to choose treatment overvenography.

5.1. In pregnant patients suspected of havinglower extremity DVT, we recommend initial eval-

uation with proximal CUS over other initial tests,including a whole-leg US (Grade 2C), moderatelysensitive D-dimer (Grade 2C), highly sensitiveD-dimer(Grade 1B),or venography(Grade 1B).

5.2. In pregnant patients with suspected DVT inwhom initial proximal CUS is negative, we sug-gest further testing with either serial proximalCUS (day 3 and day 7) (Grade 1B) or a sensitiveD-dimer done at the time of presentation (Grade2B) over no further testing for DVT. We recom-mend that patients with an initial negative prox-

imal CUS and a subsequent negative sensitiveD-dimer or negative serial proximal CUS un-dergo no further testing for DVT (Grade 1B) andthat patients with positive D-dimer have anadditional follow-up proximal CUS (day 3 andday 7) rather than venography (Grade 1B) or

whole-leg US (Grade 2C).

5.3. In pregnant patients with symptoms sug-gestive of isolated iliac vein thrombosis (swellingof the entire leg, with or without flank, buttock,or back pain) and no evidence of DVT on stan-

dard proximal CUS, we suggest further testingwith either Doppler US of the iliac vein (Grade2C), venography (Grade 2C), or direct MRI(Grade 2C), rather than standard serial CUS ofthe proximal deep veins.

6.1. In patients suspected of having upperextremity DVT, we suggest initial evaluation

with combined modality US (compression witheither Doppler or color Doppler) over otherinitial tests, including highly sensitive D-dimeror venography(Grade 2C).

6.2. In patients with suspected upper extremityDVT in whom initial US is negative for thrombosisdespite a high clinical suspicion of DVT, we sug-gest further testing with a moderate or highly sen-sitive D-dimer, serial US, or venographic-basedimaging (traditional, CT scan, or MRI), ratherthan no further testing (Grade 2C).

In patients with suspected upper extremity DVTand an initial negative combined-modality US andsubsequent negative moderate or highly sensitiveD-dimer or CT or MRI, we recommend no fur-ther testing, rather than confirmatory venography(Grade 1C). We suggest that patients with an initialcombined negative modality US and positiveD-dimer or those with less than complete evalua-tion by US undergo venography rather than nofurther testing, unless there is an alternativeexplanation for their symptoms (Grade 2B), in

which case testing to evaluate for the presence

an alternative diagnosis should be performed. Wesuggest that patients with a positive D-dimer orthose with less than complete evaluation by USbut an alternative explanation for their symptomsundergo confirmatory testing and treatment ofthis alternative explanation rather than venog-raphy(Grade 2C).

Remarks: Further radiologic testing (serial US orvenographic-based imaging or CT/MR to seek analternative diagnosis) rather than D-dimer testing ispreferable in patients with comorbid conditions typi-cally associated with elevated D-dimer levels.

DVT is a common condition that affects approxi-mately one in 1,000 persons per year.1,2 Objective

testing for DVT is crucial because clinical assessmentalone is unreliable,3-6 and the consequences of misdiag-nosis are serious, including fatal pulmonary embolism(PE).7,8 Although anticoagulant therapy is effective,9 itsunnecessary use entails expense, inconvenience, andrisk of major hemorrhage.9 Only a minority of patientsevaluated for suspected DVT actually have the dis-ease.10 Therefore, diagnostic strategies must be able to

correctly rule in DVT when it is present and safelyrule out DVT when it is absent.

Three categories of tests are typically used to deter-mine the probability of DVT: (1) clinical probabilityassessment based on patient history and clinical find-ings, (2) D-dimer assays, and (c) imaging studies(most commonly venous ultrasonography [US] and lessfrequently venography, CT scan, or MRI). Diagnostictesting often requires that the results of more thanone assessment are combined. The goal of choosingone strategy over another is to improve patient out-comes in the most efficient manner.


6/68


of VTE during follow-up (a rate comparable to that seen whenDVT is excluded by venography) in management studies in whichtreatment is withheld on the basis of a negative result.13 Manage-ment studies that assess the follow-up frequency of VTE afternegative diagnostic testing provide no information regarding false-positive diagnoses for DVT. Patients who are misdiagnosed withDVT will be prescribed unnecessary anticoagulants and some willsuffer major bleeding as a result.

To overcome this limitation, we estimated the risk of majorbleeding associated with different diagnostic strategies. These esti-

mates were based on (1) the proportion of patients diagnosed withDVT (derived from sensitivity and specificity, with the assumptionthat all diagnosed DVT are treated), and (2) the frequency of majorbleeding with 3 months of therapeutic-dose anticoagulants incohort studies and randomized trials of patients with VTE. Becausethe evidence regarding major bleeding emerging from thesemodels is indirect, it is generally rated as no higher than moderatequality.

For those diagnostic tests that have been robustly evaluated inmanagement studies (ie, in patients with suspected first lowerextremity DVT), we have assessed the impact of various strategieson major bleeding (both fatal and nonfatal, in patients prescribedanticoagulants on the basis of a positive test result) and mortality, as

well as on the frequency of PE during follow-up (fatal and nonfatal)

after application of a given diagnostic strategy (see Table S1 for listof strategies) (tables that contain an S before the number denotesupplementary tables not contained in the body of the article andavailable instead in an online data supplement; see the Acknowl-edgments for more information). Management studies that fol-lowed cohorts of patients subjected to specific strategies for DVTdiagnosis were used to determine the proportion of patients ini-tially judged to be DVT-free who returned with symptomatic VTE.In order to identify the proportion and clinical course of patientsincorrectly classified as having DVT and to estimate the risk of PE(fatal and nonfatal) in patients incorrectly categorized, we used adecision analytic model based on methodology described in detailin previous publications.14,15 The model was originally developedto estimate the cost-effectiveness of diagnostic strategies. It was

updated to include estimates of the outcomes (see below) of patientswith DVT treated with anticoagulation for at least 3 monthsreported in a recent meta-analysis.9 Sensitivities and specificitiesfrom meta-analyses were used to determine the proportion ofpatients with proximal, distal, and no DVT subjected to each diag-nostic strategy who would be treated with anticoagulant therapy.

Based on the results of a previous meta-analysis of patientswith suspected symptomatic DVT of the leg, we estimated anoverall prevalence of proximal DVT of 19.0%,10 with prevalencesof 56.2%, 12.4%, and 3.4% in the high, moderate, and low pretestprobability groups, respectively. The overall prevalence of distalDVT was estimated to be 5%. Untreated distal DVT was assumednot to directly cause PE; we estimated the probability of propaga-tion to proximal veins of 21.4%. We estimated the probability thatpatients with treated proximal DVT would suffer a fatal PE to be

0.3% and a nonfatal PE to be 1.4% over 3 months.The model assumed that all bleeding events were attribut-

able to anticoagulation (ie, bleeding rates are not reported foruntreated patients). Patients receiving treatment had a 0.3% proba-bility of fatal bleeding, a 0.1% probability of nonfatal intracranialbleeding, and a 2.1% probability of major nonfatal non-intracranialbleeding over 3 months.9,14,15 All parameters were modeled with aprobability distribution to generate a credible range for the out-comes in question. The outputs from the model were the propor-tion of patients suffering the following events over the 3 monthsafter diagnostic assessment: (1) fatal PE, (2) nonfatal PE, (3) fatalbleeding, (4) nonfatal intracranial bleeding, and (5) major nonfa-tal, non-intracranial bleeding. Table S1 lists the 21 diagnosticalgorithms evaluated with this model.16-35

This article focuses on the identification of optimalstrategies for the diagnosis of clinically suspectedDVT in adults. Consecutive sections of this chapterconcentrate on first DVT, recurrent DVT, upperextremity DVT, and DVT during pregnancy. Most ofthe data come from evaluations of patients in theambulatory setting (ie, outpatient or ED), and ourrecommendations are most applicable to this patient

population. Recommendations for the treatment ofDVT once diagnosed can be found in Kearon et al.11

1.0 Methods

Article panelists identified questions related to the evaluationof adults with suspected DVT (Table 1). A broad overview search

was performed centrally and provided to all coauthors, who fol-lowed it with more specific searching as required. Recommenda-tions were developed from this evidence.

Eligible studies included both those addressing diagnosticaccuracy (cross-sectional accuracy studies) and studies that assessedclinical outcomes such as DVT or PE during follow-up (prospec-tive cohort management studies and randomized controlled trials[RCTs]). In typical management studies, investigators followuntreated patients with negative test results and record the pro-portion of patients who develop VTE. For each section, we devel-oped corresponding methodology tables that included informationon the study question (in terms of population, intervention, com-parator, and outcome), the type of evidence assessed (meta-analysisor original study; cross-sectional study or management cohortor randomized trial), and selected details of study execution(inclusion of consecutive patients and independence of test resultassessment). Findings of individual studies and meta-analyses arepresented in descriptive tables and, when feasible, overall find-ings relating to each question are summarized as Evidence Pro-files and Summary of Findings tables.

For accuracy studies, we extracted sensitivity and specificityand then estimated the effect on patient-important outcomes (eg,DVT, PE, death, bleeding in treated patients) that would be asso-ciated with this level of accuracy, assuming prevalences of DVTthat correspond to high, moderate, and low pretest probabilitycategories. For studies in which the diagnostic test was used tomanage patients (ie, management studies), the incidence of VTEduring follow-up was determined for patients in whom anticoagu-lation and additional diagnostic testing were withheld on the basisof negative test results.

Following the approach articulated by Grades of Recommen-dation, Assessment, Development, and Evaluation (GRADE) forformulation of recommendations related to diagnosis,12 we firstconsidered the quality of evidence (representing our confidence

that the testing strategy would result in patient outcomes that sup-port a particular recommendation). We initially considered studiesas providing high quality of evidence, unless rated down becauseof the following factors: risk of bias (eg, unrepresentative patients,lack of independent assessment of test and criterion standard),inconsistency (differences among study results), indirectness(with respect to the population studied, the tests performed, orthe outcome measured), lack of precision, and risk of publicationbias. Unless otherwise explicitly stated, the quality of evidenceobtained from cross-sectional accuracy studies was lowered byone level because of the indirectness with which sensitivity andspecificity corresponds to patient-important outcomes.

Typically, diagnostic strategies for DVT have been deemedacceptable if they have demonstrated no more than a 2% frequency


7/68


Table1StructuredClinicalQuestions

InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e

S

uspectedfirstDVT(Section3.0)

Whataretheconsequences

ofusingvenography

todiagnosefirst

DVT?

Patientswith

suspectedfirstDVT

Persistentintraluminal

fillingdefect

Morbiditycaused

by

teststrategy

RCTs

Observationalstudies

Prospectivecohortstudies(canbesingle

grouporsinglearmthatuseddiagnosticintervention

specifiedinquestion)

Cross-sectionalaccuracystudies

(ifinsufficientdatafromrandomizedcontrolled

managementtrialsorprospectivemanagement

studies)

Numberofnonevaluable

testresults


ofusingvenographyto

ruleoutfirstDVT?

Patientswith

suspectedfirstDVT

Negativevenography

V

TEduringadditional

testingor3-6mo

follow-upif

interventionnegative

forDVT

FN/1,000ofnegatives

(eg,post-TPofa

negativetest)

Morbiditycaused

by

teststrategy

RCTs



grouporsinglearmthatu

seddiagnosticintervention




managementtrialsorprospectivemanagementstudies)


ofusingCUSto

diagnoseproximal

DVT

Patientswith

suspectedfirstDVT

Noncompressiblevenous

segmentfromcommon

femoralveindownto

andincludingthe

trifurcationveins

Inallpatients

Iflowpre-TP

Ifmoderatepre-TP

Ifhighpre-TP

Ifpositivehighlysensitive

DD

Ifpositivemoderately

sensitive(SimpliRED)

DD

Ifnegativehighly

sensitiveDD

V

enography

FP/1,000ofpositive

(eg,post-TPofa

positivetest)if

managements

tudy

Specificityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









Ifnegativemoderately


DD

(Continued)


8/68


(Continued)

Table1Continued

InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingserialproximal

CUStoexcludeDVT

(regardlessofpre-TP)?

Patientswith

suspectedfirstDVT

ProximalCUSon

presentationandif

negativeafollow-up

testapproximately

1wklater

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upif

interventionnegative

forDVT

FN/1,000ofnegatives

(eg,post-TPofanegative

test)ifmanage

mentstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs








managementtrialsorprosp

ectivemanagementstudies)


ofusingwhole-legUS

todiagnosedistalDVT

Patientswith

suspectedfirstDVT


segmentisolatedto

thecalfveins(eg,

posteriortibial,

anteriortibial,

andperonealveins)

Inallpatients

Iflowpre-TP

Ifmoderatepre-TP

Ifhighpre-TP

Ifpositivehighly

sensitiveDD



DD

Ifnegativehighly

sensitiveDD


sensitiveDD

V

enographyorserial

proximalCUS

plusVTEduring

additional3-6mo

ifnegativefor

proximalDVT

FP/1,000ofpositive

(eg,post-TPof

apositivetest)

if

managements

tudy

Specificityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










ofusingasinglewhole-

legUStoexcludeDVT

(regardlessofpre-TP)?

Patientswith

suspectedfirstDVT

Negativesinglewhole-

legUSondayof

presentation

V

enographyorserial

proximalCUSplus

VTEduringadditional

3-6moifnegativefor

proximalDVT

FN/1,000ofnegatives

(eg,post-TPofa

negativetest)

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










9/68


10/68


11/68


12/68


13/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e

Newnoncompressible

segment

Increaseinresidual

venousdiameter

Newnoncompressible

segmentorincrease

inresidualvenous

diameter

Newnoncompressible

segmentcompared

withprevious

proximalCUSeither

atpresentationoron

follow-uptest(s)over

next7-10d

Newincreaseinresidual

venousdiameter

greaterthanspecified

comparedwithprevious

proximalCUSeither

atpresentation

oronfollow-uptest(s)

overnext7-10d

Newnoncompressible

segmentorchangein

residualdiameter

greaterthanspecified

comparedwithprevious

proximalCUSat

presentationoron


next7-10d



CUStoexclude

recurrentDVTin

theabsenceofthe

following?

Patientswith

suspectedrecurrent

DVT

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upif

interventionshows

noevidenceof

recurrence

FN/1,000(eg,po

st-TP

ofanegativetest)if

managements

tudy

RCTs









Sensitivityif

accuracystudy

Morbiditycaused

byteststrategy

Newnoncompressible

segment

ProximalCUSon

presentationandif

negativeornonew

noncompressible

segmentcompared

withprevious,follow-up

test(s)overnext

5-10d,examiningfor

newnoncompressible

segment

Table1Continued

(Continued)


14/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e

Changeinresidual

venousdiameter

ProximalCUSon

presentationandif

negativeorchangein

residualvenousdiameter

lessthanspecified

comparedwithprevious,


next5-10d,examining

forchangeinresidual

venousdiameter

ProximalCUSon

presentationandif

negativeornonew

noncompressiblesegment

orchangeinresidual

diametergreaterthan

specifiedcomparedwith

previous,follow-uptest(s)

overnext5-10d,

examiningfornew

noncompressiblesegment

orchangeinresidual

venousdiameter

Newnoncompressible

segmentorchangein

residualvenous

diameter


ofusingDDand

pre-TPtoexclude

suspectedrecurrent

DVT?

Patientswith

suspectedrecurrent

DVT

Negativemoderately


DDpluslow/moderate/

highpre-TPat

presentationOR

NegativehighlysensitiveDD

pluslow/moderate/

highpre-TPat

presentation

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upif

interventionshows

noevidenceof

recurrence

FN/1,000(eg,po

st-TP

ofanegativetest)if

managements

tudy

Sensitivityifaccu

racystudy

Morbiditycaused

bytest

strategy

RCTs









(Continued)

Table1Continued


15/68


16/68


17/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingvenographyto

ruleoutDVTduring

pregnancy?

Patientswith

suspectedrecurrent

DVT

Negativevenography

V

TEduringadditional

testingor3-6mo

follow-upif

interventionshows

noevidenceofDVT

FN/1,000(eg,po

st-TP

ofanegativetest)

Morbiditycaused

by

teststrategy

Radiationexposu

retothe

fetus(alsotera

togenicity,

increaseinthe

risk

ofchildhoodcancer,

fetalloss)

RCTs










ofusingCUSto

diagnoseproximal

DVTduring

pregnancy

Pregnantw

omen

withsuspected

DVT


segmentfromcommon

femoralveindownto

andincludingthe

trifurcationveinsand/

orabsenceofDoppler

flowintheiliacvein

Inallpatients

Iflowpre-TP

Ifmoderatepre-TP

Ifhighpre-TP


DD



DD

Ifnegativehighly

sensitiveDD



DD

V

enography

FP/1,000ofpositive(eg,

post-TPofapositivetest)

ifmanagementstudy

Specificityifaccu

racystudy

Morbiditycaused

bytest

strategy

RCTs











CUStoexcludeDVT

duringpregnancy,

regardlessofpre-TP?

Pregnantw

omen

withsuspected

DVT

ProximalCUSon

presentationandif

negativefollow-up

testswithin5-10d

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upifintervention

showsnoevidence

ofDVT

FN/1,000(eg,po

st-TP

ofanegativetest)if

managements

tudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









(Continued)

Table1Continued


18/68


19/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusinganegativeDD

toobviatetheneedfor

serialtestinginpatients

withanegativeproximal

CUSandmoderate

orhighpre-TPat

presentationduring

pregnancy?

Pregnantw

omen

withsuspected

DVT

Negativehighlysensitive

DDornegative

moderatelysensitive

(SimpliRED)DD

V

enographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









NegativeproximalCUS

plusmoderatepre-TP

NegativeproximalCUS

plushighpre-TP


ofusingpre-TPwitha

negativeproximal

CUStoexcludeDVT?

Pregnantw

omen

withsuspected

DVT

NegativeproximalCUS

pluslow/moderate/high

pre-TPatpresentation

V

enographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycausedby

teststrategy

RCTs











CUStoexcludeDVT

inpatientswithalow/

moderate/highpre-TP?

Pregnantw

omen

withsuspected

DVT

RepeatproximalCUS

(withinapproximately

1wk)

V

enographyorVTEduring

additionaltestingor

3-6mofollow-upif

interventionshows

noevidenceofDVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









(Continued)

Table1Continued


20/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofwhole-legUSto

diagnosedistalDVT

duringpregnancy

Pregnantw

omen

withsuspected

DVT


segmentisolatedto

thecalfveins(eg,

posteriortibial,

anteriortibial,and

peronealveins)

Inallpatients

Iflowpre-TP

Ifmoderatepre-TP

Ifhighpre-TP


DD



DD

Ifnegativehighlysensitive

DD



DD

V

enographyorserial

proximalCUS

plusVTEduring

additional3-6moif

negativefor

proximalDVT

FP/1,000ofpositive

(eg,post-TPofa

positivetest)if

managements

tudy

Specificityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










ofusingasinglewhole-

legUStoexcludeDVT

duringpregnancy?

Pregnantw

omen

withsuspected

DVT

Negativesinglewhole-leg

USondayof

presentation

V

enographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

FN/1,000(eg,po

st-TP

ofanegativetest)if

managements

tudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










ofusingahighlysensitive

DDasastand-alonetest

toexcludeDVTduring

pregnancy?

Pregnantw

omen

withsuspected

DVT


DDondayof

presentation

V

enographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

VenographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

RCTs









(Continued)

Table1Continued


21/68


22/68


23/68


24/68


25/68


26/68


27/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingserialCUSto

excludeupperextremity

DVTinpatientswitha

positiveDD?

NegativeCUSpluspositive

moderately(SimpliRED)

orhighlysensitiveDD

pluslowpre-TP

Patientswith

suspectedupper

extremityDVT

RepeatCUS(within

approximately1wk)

V

enographyorVTE

duringadditionaltesting

or3-6mofollow-upif

interventionshowsno

evidenceofDVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









NegativeCUSpluspositive

moderatelysensitive

(SimpliRED)orhighly

sensitiveDDplus

moderatepre-TP

NegativeCUSplus

positivehighlysensitive

DDplushighpre-TP


ofusingserialDoppler

UStoexcludeupper

extremityDVTin

patientswitha

positiveDD?

NegativeDopplerUS

pluspositivemoderately


orhighlysensitive

DDpluslowpre-TP

NegativeDopplerUS

pluspositivemoderately


orhighlysensitive

DDplusmoderate

pre-TP

Patientswith

suspectedupper

extremityDVT

RepeatDopplerUS


1wk)

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upif

interventionshows

noevidenceof

DVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









NegativeDopplerUS

pluspositivehighly

sensitiveDDplus

highpre-TP

(Continued)

Table1Continued


28/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingserialduplex

UStoexcludeupper

extremityDVTin

patientswithapositive

DD?

NegativeduplexUSplus

positivemoderately


orhighlysensitive

DDpluslowpre-TP

Patientswith

suspectedupper

extremityDVT

RepeatduplexUS


1wk)

V

enographyorVTE


or3-6mofollow-upif

interventionshowsno

evidenceofDVT

FN/1,000(eg,po

st-TPofa

negativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










positivemoderately


orhighlysensitive

DDplusmoderate

pre-TP


positivehighly

sensitiveDDplus

highpre-TP


ofusinganegativeDD

toobviatetheneedfor


withsuspectedupper

extremityDVTand

anegativeCUSand

moderateorhighpre-TP

atpresentation?

Patientswith

suspectedupper

extremityDVT


DDornegative

moderatelysensitive

(SimpliRED)DD

V

enographyorVTE


or3-6mofollow-upif

interventionshowsno

evidenceofDVT

FN/1,000(eg,po

st-TPofa

negativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









NegativeCUSpluslow

pre-TP

NegativeCUSpluslow

pre-TP

NegativeCUSplushigh

pre-TP

(Continued)

Table1Continued


29/68


(Continued)

InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusinganegativeDD

toobviatetheneedfor


withsuspectedupper

extremityDVTanda

negativeDopplerUSand

moderateorhighpre-TP

atpresentation?

Patientswith

suspectedupper

extremityDVT

Negativesensitive

DDornegative

moderatelysensitive

(SimpliRED)DD

V

enographyorVTE


or3-6mofollow-upif

interventionshowsno

evidenceofDVT

FN/1,000(eg,po

st-TPof

anegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









NegativeDopplerUS

pluslowpre-TP

NegativeDopplerUS

plusmoderatepre-TP

NegativeDopplerUS

plushighpre-TP


ofusinganegativeDD

toobviatetheneedfor


withsuspectedupper

extremityDVTanda

negativeduplexUS

andmoderateorhigh

pre-TPatpresentation?

Patientswith

suspectedupper

extremityDVT


DDornegative

moderatelysensitive

(SimpliRED)DD

V

enographyorVTE


or3-6mofollow-upif

interventionshowsno

evidenceofDVT

FN/1,000(eg,po

st-TPofa

negativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










lowpre-TP


lowpre-TP


highpre-TP

Table1Continued


30/68


31/68


InformalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingcontrastMR

venographytodiagnose

upperextremityDVT?

Patientswith

suspectedupper

extremityDVT

IntraluminalfillingdefectV

enography

FP/1,000ofpositive

(eg,post-TP

ofa

positivetest)if

managemen

tstudy

Specificityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










ofusingMRvenography

toexcludeupper

extremityDVT?

Patientswith

suspectedupper

extremityDVT

NegativeMRvenographyV

enographyorVTE

duringadditional

testingor3-6mo


showsnoevidence

ofDVT

FN/1,000(eg,po

st-TP

ofanegativetest)if

managemen

tstudy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs










ofusingMRdirect

thrombusimaging

todiagnoseupper

extremityDVT?

Patientswith

suspectedupper

extremityDVT

Highsignalintensity

V

enography

FP/1,000ofpositive

(eg,post-TP

ofa

positivetest)if

managemen

tstudy

Specificityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









Table1Continued

(Continued)


32/68


We have judged diagnostic strategies acceptable if we are con-fident that they meet all of the following conditions: (1) manage-ment studies have established that the probability of objectivediagnosis of symptomatic VTE during 3 to 6 months of fol-low-up after initial classification as DVT negative is 2%;(2) modeling suggests a, 0.1% (one in 1,000) risk of fatal PE(to illustrate, the calculated risk is 0.065% for routine venographyand 0.076% for routine serial proximal compression US [CUS])and nonfatal PE of,0.5% (five in 1,000) (to illustrate, the calcu-lated risk is 0.31% for routine venography and 0.37% for rou-

tine serial proximal CUS); (3) modeling suggests a ,0.1% (one in1,000) risk of fatal hemorrhage (to illustrate, the calculated riskis 0.071% for routine venography and 0.084% for routine serialCUS) and nonfatal intracranial hemorrhage of 0.035%(0.35/1,000) (to illustrate, the calculated risk is 0.025% for rou-tine venography and 0.03% for routine serial proximal CUS);and (4) modeling suggests a risk of death from PE or hemor-rhage of, 0.17% (1.7/1,000) (to illustrate, the calculated riskis 0.137% for routine venography and 0.159% for routine serialproximal CUS). These thresholds are admittedly arbitrary; those

who choose different threshold may reach different conclusions.In the accompanying evidence profiles, we have rated down thequality of evidence if the CI around the estimate of a false-negativeresult after initial diagnostic testing crosses our 2% threshold.

1.1 Implications of Values and Preferencesin the Diagnostic Process

When evaluating alternative diagnostic strategies in patientswith suspected DVT, harmful effects, cost, and patient preference(eg, test discomfort, inconvenience, and diagnostic uncertainty)need to be considered. Unless stated, the cost (eg, to the patient,a third-party payer, or society) associated with different diagnosticstrategies did not influence our recommendations.

Harmful effects of a given diagnostic strategy include not onlyacute (eg, renal toxicity) and long-term (eg, cancer secondary toradiation exposure) complications but also indirect complicationsassociated with the incorrect diagnosis (eg, bleeding) or exclusion

(subsequent DVT and/or PE) of DVT. A systematic review ofpatient preferences suggests that the disutility (unpleasantness)associated with an episode of nonfatal VTE and major nonfatalbleeding are similar.36 This assessment is also supported by theresults of a subsequent survey of all panelists for these guidelinesthat rated values and preferences associated with different stan-dardized clinical scenarios, including episodes of nonfatal VTEand nonfatal major bleeding, as well as the use of different anti-thrombotic therapies.36 Therefore, on average, we assume thatpatients attach equal value to nonfatal VTE and nonfatal majorbleeding events. However, we also took into account that valuesand preferences vary markedly between individual patients andthat there is often appreciable uncertainty about the averagepatient values we used.

We generally recommend against invasive diagnostic strategies

when a comparably accurate noninvasive alternative is available.This is because invasive tests are generally associated with greaterpatient discomfort, side effects (eg, reactions to contrast) and radia-tion exposure than noninvasive tests. However, we recommendinvasive testing over noninvasive testing if the benefits of a moreaccurate diagnosis outweigh these disadvantages. Individual patientpreferences relating to test discomfort and tolerance for diagnos-tic uncertainty influence this decision. We also acknowledge thataccess to types of diagnostic testing differs (eg, many centers donot perform ascending venography) and that it is appropriate forsuch factors to influence the choice of diagnostic testing.

In making recommendations, we have placed the burden ofproof with those who would claim a benefit with a more complex,invasive, or expensive diagnostic strategy. In the absence of suchIn

formalQuestion

PICOQuestion

Methodology

Popu

lation

Intervention

Comparator

Outcom

e


ofusingMRdirect

thrombusimaging

toexcludeupper

extremityDVT?

Patientswith

suspectedupper

extremityDVT

NegativeMRdirect

thrombusimaging

V

enographyorVTE

duringadditional

testingor3-6mo

follow-upif

intervention

showsnoevidence

ofDVT

FN/1,000(eg,po

st-TPof

anegativete

st)if

accuracystu

dy

Sensitivityifaccu

racystudy

Morbiditycaused

by

teststrategy

RCTs









ForPre-TPuseavailablemodelandspecify.Managementstudyisoneinwhichpatie

ntismanaged(ortreated)accordingtotest

results.CUS5compressionultrasonograph

y;DD5D-dimer;FN5false

negative;FP5falsepositive;MR5magneticresonance;RCT5randomizedcontrolle

dtrial;sensitiveDD5D-dimerwithsensitivity95%ingeneralpopulationofpatientswithsuspectedVTE;TP5test

probability;US5ultrasound.

Table1Continued


33/68


unable to definitely establish or exclude the diagnosisof DVT.

3.0 Diagnosis of Suspected FirstLower Extremity DVT

The limitations of contrast venography have led tothe development of other testing strategies for theevaluation of patients with suspected DVT.

3.1 Alternatives to Venography for the Evaluationof Suspected First Lower Extremity DVT

3.1.1 Pretest Probability Assessment: Althoughthe clinical diagnosis of DVT is nonspecific andindividual clinical features are of little value indiagnosing DVT,56 clinical prediction or pretestprobability estimates (structured and based on spe-cific criteria or unstructured and empirical) are ableto stratify patients into groups according to theirprobability of DVT.56 Several structured scoringsystems have been developed10,31,32,56-60; the most

well studied is the Wells score.6,10,31,56 This ruleincorporates signs, symptoms, and risk factors for

VTE to categorize patients as having a low, mod-erate, or high probability of DVT,31 with a preva-lence of DVT of 5.0% (95% CI, 4%-8%), 17% (95%CI, 13%-23%), and 53% (95% CI, 44%-61%),respectively.10 A modification of the Wells scorestratifies patients as being likely (prevalence of DVT,28%; 95% CI, 24%-32%) or unlikely (prevalence ofDVT, 6%; 95% CI, 4%-8%) to have DVT.32 The

Wells score has limitations. Interobserver reliabilityhas not been widely evaluated, although one studyconfirmed its reproducibility when used by residentphysicians.61 One study found that the model per-formed less well in a primary care setting.62

3.1.2 D-Dimer: D-dimer, a degradation product ofcross-linked fibrin, is typically elevated in patients

with acute DVT. However, because D-dimer levelsmay also be increased in a variety of nonthrombotic

disorders (eg, malignancy, disseminated intravascularcoagulation, increasing age, infection, pregnancy,following surgery or trauma, inflammatory condi-tions, atrial fibrillation, and stroke), D-dimer is a sen-sitive but nonspecific marker for VTE. Consequently,although a positive result is not useful in confirmingthe diagnosis of DVT, a negative result can aid in theexclusion of this diagnosis. In hospitalized and otheracutely ill patients commonly affected by the condi-tions listed above, D-dimer testing has less usefulnessbecause of the high frequency of false-positive results.A wide variety of D-dimer assays are available. In a

proof (eg, the strategy has not been assessed in managementstudies), we generally recommend against such strategies. Whenrecommendations were considered controversial by the panel(Recommendation 3.1), the results of panel votes are presented,along with the recommendation.

2.0Venography: ReferenceStandard for Diagnosis of DVT

Contrast venography is the criterion standard (eg,the benchmark or best-performing test) for the diag-nosis of DVT.13,37-39 In this technique, iodinated con-trast is injected into a dorsal foot vein to outline theentire deep venous system of the lower extremity.DVT is diagnosed by the presence of a constantintraluminal filling defect that is present in more thanone view; nonfilling of a venous segment despiterepeated injection is suspicious, but not diagnostic, ofDVT.40 Tables S2 to S4 present methods, descriptiveresults, and evidence profiles for diagnostic studies

assessing venography in patients with first suspectedlower extremity DVT. Withholding anticoagulants inpatients with suspected first DVT who have a techni-cally adequate normal venogram is associated with alow frequency of symptomatic DVT or PE during3 months of follow-up (1.2%; 95% CI, 0.2%-4.4%).13This frequency of subsequent disease is the standardagainst which all tests or diagnostic strategies used toexclude DVT are typically judged.

Venography is expensive, not uniformly available,uncomfortable for patients, and contraindicated inpatients with renal insufficiency and severe allergic

reactions to contrast medium. In 5% of patients, thedorsal foot vein cannot be cannulated.41 Even whenvenography is performed by experienced radiologists,inadequate imaging is common; in up to 20% of veno-grams there is inadequate visualization of a venoussegment.13,42-46 Further, venography can be difficultto interpret and the designation of DVT present orDVT absent is subject to a considerable degree ofboth intraobserver (k values ranging from 0.56-0.95)and interobserver (k values ranging from 0.47-0.92)

variation.43,47-52 Adverse reactions to contrast mediainclude dizziness and nausea (complicating between

1% and 4% of procedures53,54

), severe allergic reac-tions (in 0% [95% CI, 0%-2.4%]54 to 0.4% [95% CI,0.1%-0.4%]53 of patients) and post-venography DVT(confirmed by repeat venography in between 0%[95% CI, 0%-13.3%]55 to 2% [95% CI, 0%-12.6%]53of patients).

The above limitations make venography unsuitablefor routine use in patients presenting with suspectedDVT. Venography is now rarely used in clinical prac-tice and many hospitals are unable to perform theprocedure. However, venography can serve as a ref-erence standard and be used when other tests are


34/68


suspected DVT and suspected PE. Most studies ofCT scan venography have been done this way.

3.1.5 MRI: MRI can be applied using a variety oftechniques. Some techniques visualize blood flow

without the need for contrast agents because theyrely on the intrinsic properties of flowing blood (time-of-flight or phase-contrast venography). However,

the imaging of vascular structures is often improvedby the use of contrast agents, such as in IV gadolin-ium. MR contrast agents can be either injected into a

vein in the foot or into the arm with imaging timedfor optimal imaging of lower limb veins. Alternatively,MR can identify DVT by direct thrombus imaging.This technique involves visualizing thrombus (highsignal due to red cell methemoglobin in the clot)against a suppressed background. This technique hasthe advantages of being noninvasive and not requiringIV contrast agents. However, MRI is not routinelyaccessible for this purpose in most centers.

3.2 Evaluation of Diagnostic Strategies forSuspected First Lower Extremity DVT

Pretest probability assessment, D-dimer testing,and venous US have been extensively investigatedand are widely used either alone or in combination inpatients with a suspected first DVT. Evaluation ofCT scan venography and MRI in this patient popula-tion has been limited to accuracy studies.

Details of management studies in patients withsuspected first DVT that used pretest probability,

D-dimer, and proximal CUS are summarized inTable 2 and Tables S5 to S16.10,16-19,21-23,25,30-32,41,60,62,63,65,69-86Table 2 describes the consequences of using specificstrategies in terms of the probability of VTE beingdiagnosed during clinical follow-up when a givendiagnostic strategy suggests that DVT is not present.

As shown in Table 2 and Tables S5 to S16, the pre-test assessment (ie, prevalence of DVT) has a sig-nificant effect on the usefulness of D-dimer andproximal US. Categorizing patients as having a lowpretest probability for DVT eliminates the need for(1) radiologic imaging (eg, US) in those with a neg-

ative D-dimer, and (2) serial or repeat testing inthose with a normal proximal US. Although mostpatients with a positive CUS have a proximal DVT,this is progressively less true as pretest probabilitydeclines. In a study of 529 symptomatic patients,the posttest probability of DVT in those with a posi-tive CUS (as assessed by venography) was 100% inpatients with a high pretest probability, 96% in those

with a moderate pretest probability, and 63% inpatients with a low pretest probability.6

Tables 3 and 4 and Tables S17 to S23 summarize themethodology and results of studies assessing whole-leg

meta-analysis of 217 studies, enzyme-linked immu-nofluorescence assays (sensitivity 96%; 95% CI,89%-98%), microplate enzyme-linked immunosor-bent assays (ELISAs) (sensitivity 94%; 95% CI,86%-97%), and quantitative latex or immunoturbidi-metric assays (sensitivity 93%; 95% CI, 89%-95%)

were more sensitive for DVT than were the wholeblood D-dimer assay (sensitivity 83%; 95% CI,

67%-93%) and latex semiquantitative assays (sensi-tivity 85%; 95% CI, 68%-93%).63 Based on these data,ELISAs and enzyme-linked immunofluorescenceassays, along with the latex immunoturbidimetric assays,are generally termed highly sensitive, whereas the

whole blood D-dimer assay is considered moder-ately sensitive.10 Of these tests, the whole bloodD-dimer assay had the highest specificity (71%;95% CI, 57%-82% vs 46% [95% CI, 31%-61%] forenzyme-linked immunofluorescence assays; 53%[95% CI, 38%-68%] for microplate ELISAs, and 53%[95% CI, 46%-61%] for quantitative latex or immu-

noturbidimetric assays).63

3.1.3 Venous US: Venous US is the most widelyused imaging study for the diagnosis of DVT.64 Prox-imal CUS assesses compressibility of the femoral andpopliteal veins. The inability to fully collapse a venoussegment under gentle US probe pressure is consid-ered diagnostic of DVT. Although distal DVT may bepresent in patients with a normal proximal US, itis seldom if ever associated with important clinicalsequelae (PE or postthrombotic syndrome). However,as distal DVT may propagate proximally and lead to

PE, additional investigations, such as pretest proba-bility assessment, D-dimer testing, or a second prox-imal CUS performed 5 to 7 days later (serial or repeatUS), are needed to exclude distal DVT or, if distalDVT cannot be excluded, to detect early extensioninto the proximal veins.22,65,66 Whole-leg US assessesthe deep veins of both the proximal leg and calf. Thistechnique has been studied as a means of excludingDVT as a stand-alone test, eliminating the need for areturn visit for serial US. As whole-leg US results intreatment of distal DVT that will not extend,67,68 itcarries the risk of overtreatment.

3.1.4 CT Scan Venography: CT scan venographytypically involves injection of contrast media into anarm vein followed by helical CT imaging timed tocoincide with opacification of the deep veins of thelegs to allow assessment of these veins for thrombus.It therefore shares the disadvantage with conven-tional contrast venography of requiring administra-tion of IV contrast but does not require cannulationof a foot vein (although this technique can be used).CT scan venography can be combined with CT scanpulmonary angiography to provide imaging for both


35/68


Table2[Sections3.1-3.5]SummaryofOutcomesforDiagnosticStudiesAssessingDD,

Pre-TP,andProximal

USfortheDiagnosisofSuspectedFir

stLowerExtremityDVT

Pre-TP

DiagnosticStrateg

yUsedtoExcludeDVT

No.ofStudiesa

QualityofEvidence

NumberofPatientsSubjected

toGivenDiagnosticStrategy

NumberConsideredNegative

for

DVTatCompletion

ofTesting

OverallPrevalenceinPopulationofInterest

Incidence

ofVTEDuringFollow-upb

AmongTh

oseJudgedtoHaveDVT

ExcludedbySpecifiedDiagnosticStrategy

(ie,Post-T

PofDisease),%(95%CI)

Low

Moderatelysensitiv

eDDnegative

179

Moderate

206177

2.40.6(0.03

-2.7)

Meta-analysis,310,70,74

0.9

0.5(0.07-1.1

)

1.1(0.9-1.5)

HighlysensitiveDD

negative

516,18,25,81,82

High

1,270824

6.71.0(0.5-

1.7)


Moderate

0.5

0.4(0.04-1.1

)

Cardiac:0.4

(0.2-0.8)orTriage:0.9

(0.9-2.2)

SingleproximalUS

negative

417,30,31,7

8

High

944885

7.10.9(0.5-1.6)

DD(mixed)positiv

esingleproximal

USnegative

216,32

High

765198

50(0-1.5)

Low/moderateor

unlikely

Moderatelysensitiv

eDDnegative

160

Moderate

852500

9.61.4(0.7-

2.6)

DD(mixed)negative

132

Moderate

317218

5.00.92(0.2

-2.9)

HighlysensitiveDD

negative

121

Moderate

749481

N/A0.42(0.04-1.5)

Meta-analysis,174

Moderate

0.4(0.04-1.1)

SingleproximalUS

negative

132

Moderate

284272

5.61.5(0.5-

3.3)

DD(mixed)positiv

esingleproximal

USnegative

132

Moderate

31785

50(0-3.5)

Moderate

Moderatelysensitiv

eDDnegative


Moderate

4.4

3.5(1.4-6.9)

SimpliRED

4.9(3.6-6.8)orClearview

Simplify:5.2(4.1-6.5)

HighlysensitiveDD

negative

318,25,77

Moderate

655214

25.70.57(0.02-2.2)

Meta-analysis,210,70

Low

NPV:99(96-100);LR(negative)0.05(0.01-

0.21);estim

atedpost-testprevalence,1

Cardiac:1.7

(1.0-3.8)forpointofcareand

Triage:4.3

(2.0-9.7)

SingleproximalUS

negative

178

Moderate

144114

21.50.9(0.0

5-4.1)

SingleproximalUS

negativeDD

(mixed)negative

216,32

High

675325

22.40(0-0.9)

SerialproximalUS

negative

317,31,80

Moderate

N/A365

15.8(Basedon

2outof3studies)1.1

(0.4-2.5)0.6

(0.4-0.9)

Meta-analy

sis85

Moderate

HighlysensitiveDD

positiveSingle

proximalUSnegative

177

Moderate

13473

19.40(0-4.0)

SingleproximalUS

negativeDD

(mixed)positivesingle

proximalUSnegative

116

Moderate

42694

18.80(0-3.1)

(Continued)


36/68


Pre-TP

DiagnosticStrategyUsedtoExcludeDVT

No.ofStudiesa

QualityofEvidence

Numbe

rofPatientsSubjected

toGivenDiagnosticStrategy

NumberConsideredNegative

forDVTatCompletion

ofTesting

OverallPrevalenceinPopulationofInterest

Incidence

ofVTEDuringFollow-upb

AmongTh

oseJudgedtoHaveDVT

ExcludedbyS

pecifiedDiagnosticStrategy

(ie,Post-TPofDisease),%(95%CI)

Moderate/highor

likely

SingleproximalUS

negative

moderatelysensitiveDDnegative

130

Moderate

531148

58.80(0-2.0

)

SingleproximalUS

negativeDD

(mixed)negative

132

Moderate

24981

49.50(0-3.6

)

SerialproximalUSnegative

132

Moderate

246181

27.11.1(0.2

-3.4)

SingleproximalUS

negativeDD(mixed)

positivesingle

proximalUSnegative

132

Moderate

24997

28.50(0-3.0

)

SingleproximalUS

negativemoderately

sensitiveDDpos

itivesingleproximal

USnegative

130

Moderatenotto

use,lowtouse

53183

58.93.6(1.0

-9.1)

High

Moderatelysensitiv

eDDnegative

Meta-analysis,210,74

Moderate

19

21.4(8.5-37.9)

HighlysensitiveDD

negative


Moderate

NPV:92(81-97)

6.4(1.7-14.5)

Cardiac:6.5(3.8-13.7)orTriage:15.3

(7.4-30.1)

HighlysensitiveDD

negativesingle

proximalUSnegative

218,25

Low

35059

53.41.7(0-7

.8)

SerialproximalUSnegative

418,78,80,81

Moderate

291221

36.40.9(0.2

-2.8)

SingleproximalUS

negativehighly

sensitiveDDpos

itivesingleproximal

USnegative

125

Low

27936

59.52.8(0.1

-12.5)

SingleproximalUS

negativevenogram

negative

317,23,31

Low

16843

78.00(0-6.7

)

Table2Continued

(Continued)


37/68


38/68


DVT and either an abnormal sensitive D-dimer or ascore corresponding to DVT likely using the Wellstwo-level prediction rule were managed with serialproximal US. Whole-leg US was also performed, butresults were blinded and not used for management.Sixty-five patients (15.3%; 95% CI, 12.0%-18.8%)

were found to have DVT isolated to the calf. Ofthe 64 who completed follow-up, two patients with

isolated calf DVT experienced extension into theproximal system, which was detected on serial prox-imal US.

Tables S24 to S26 present the results of our mod-eling and decision analysis. In all tables, the denomi-nator is 1,000 patients managed according to eachdiagnostic strategy; Table S24 presents the numberof expected clinical events, and Table S25 presentsthe incremental number of events compared witha strategy of serial proximal US. Table S26 containsthe number of tests performed with each strategyand the incremental number of tests compared with

a serial proximal US strategy. The latter was used forcomparative purposes as it is one of the strategiesleast likely to produce false-negative results.

Although a whole-leg US strategy reduces the num-ber of US sessions compared with serial proximalUS, routine anticoagulation of patients with isolatedcalf DVT will result in a larger number receivingtreatment and an increase in bleeding complications.As only a relatively small portion of isolated calfDVT would propagate or embolize without treat-ment, some patients undergoing whole-leg US willreceive anticoagulation for a disease with a benign

prognosis if left untreated (Kearon et al).11 Theabsence of a clear safety advantage for the whole-legUS strategy is demonstrated in Table S24, whereinthe point estimate for the risk of fatal PE with a serialproximal US strategy is no higher than one involving

whole-leg US. Guidelines for determining whichpatients are likely to most benefit from anticoagulant

Table 3[Sections 3.2-3.5] Summary of Findings for Diagnostic Studies Evaluating Whole-Leg US in First SuspectedLower Extremity DVT: Prospective Cohort Management Studies

Diagnostic Strategy Used toExclude DVT

No. of Participants(Studies)a Outcome

Incidence of VTE During Follow-up inThose Judged to Have DVT Excluded

(ie, Post-TP of DVT), % (95% CI) Quality of Evidence

Single negative whole-legUS

4,731 (7)72,114-119 VTE during clinicalfollow-up (3 mo)

0.57 (0.25-0.89) Moderateb

Consequences in terms of presenting with VTE during clinical follow-up when a single whole-leg US is used to rule out suspected first lowerextremity DVT. GRADE Working Group grades of evidence: High quality: Further research is very unlikely to change our confidence in theestimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and maychange the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likelyto change the estimate. GRADE5Grades of Recommendation, Assessment, Development, and Evaluation; See Table 1 legend for expansion ofother abbreviations.aIncludes six management studies and one arm from an RCT.bPooled management studies are of high methodologic quality and results are highly concordant. Moderate quality due to only one RCT among theanalyzed studies.

US for the diagnosis of first DVT.14,19-21,24,26-29,64,87-123Although whole-leg US generates a larger number offalse-negative results than venography for the diag-nosis of isolated calf vein thrombosis, almost all posi-tive results will be true positives. Pooling data from34 studies comparing US with venography for calf-

vein DVT in symptomatic patients yielded a speci-ficity of 96.0% (95% CI, 95.2%-96.8%).87-122 Eight

management studies have assessed the safety ofwithholding anticoagulants based a negative whole-leg US.19-21,24,26-29 The results of seven of these studies

were pooled in a recent meta-analysis123 that foundthat the 3-month rate of VTE after a single negat