|

Received: 5 January 2017    Accepted: 24 February 2017

DOI: 10.1111/ijlh.12665

REVIEW ARTICLE

Review of D-­dimer testing: Good, Bad, and Ugly

L.-A. Linkins1,2 | S. Takach Lapner1,2

1
Department of Medicine, McMaster
University, Hamilton, ON, Canada
2
Department of Medicine, University of
Alberta, Edmonton, AB, Canada
Correspondence
Dr. Lori-Ann Linkins, McMaster University,
Hamilton, ON, Canada.
Email: linkinla@mcmaster.ca
Abstract
D-­dimer assays are commonly used in clinical practice to exclude a diagnosis of deep
vein thrombosis or pulmonary embolism. More recently, they have been also been
used to guide patients with unprovoked venous thromboembolism (VTE) when faced
with the decision to continue or stop anticoagulation after initial treatment is com-
plete. D-­dimer assays vary widely with respect to the antibody used, method of cap-
ture, instrumentation required, and calibration standard. These differences have an
important influence on the operating characteristics of the assays. Consequently, the
evidence available in the literature for one assay cannot simply be extrapolated to
another. In this review, we will outline the general properties of D-­dimer assays, dis-
cuss the concept of raising the D-­dimer threshold used in diagnosis of VTE according
to pretest probability and age, and provide clinical perspective on the role of D-­dimer
testing in the diagnosis and prognosis of VTE.

KEYWORDS
D-dimer, venous thromboembolism, DVT, PE, diagnosis, prognosis

With the generations that followed, there was dramatic improve-

1 | INTRODUCTION
1.1 | What is D-­dimer?
During hemostasis, the formation of fibrin clots by the coagulation
system in response to vascular injury is balanced by the breakdown
of clot by the fibrinolytic system. D-­dimers are one of several frag-
ments that are produced when plasmin, an enzyme activated through
the fibrinolytic pathway, cleaves fibrin to break down clots. It con-
sists of two covalently bound fibrin D domains that were cross-­linked
by factor XIII when the clot was formed. This fragment forms unique
epitopes that can be targeted by monoclonal antibodies in D-­dimer
assays to confirm that the coagulation cascade is generating thrombin.
The role of D-­dimer assays in clinical medicine has evolved over
time. When D-­dimer assays were first introduced in the 1970s,
they were primarily used to check for evidence of disseminated in-
travascular coagulation, a condition that promotes rapid turnover of
clot formation and breakdown within the microvasculature. These
first-­generation assays detected both fibrinogen and fibrinogen deg-
radation products and were therefore more of a blunt tool than a fine-­
tuned diagnostic instrument.
ment in the performance characteristics of D-­dimer assays and conse-
quently a shift of their clinical utility into the field of acute thrombosis
medicine. Now, clinicians routinely use them as part of a diagnostic al-
gorithm to exclude the diagnosis of thrombosis within the lower limbs
(deep vein thrombosis; DVT) or the lungs (pulmonary embolism; PE).
More recently, D-­dimer assays have also been used to predict which
patients are more likely to experience recurrent thrombosis when an-
ticoagulants are stopped.
In the discussion below, we will provide a brief overview of the
general properties of D-­dimer assays; compare and contrast assay
types; discuss the importance of the threshold chosen to define a pos-
itive and negative result; and provide clinical context for the role of
D-­dimer testing in the diagnosis and prognosis of venous thrombo-
embolism (VTE).
1.2 | Determinants of D-­dimer concentration
Low levels of D-­dimers are detectable in healthy individuals, as small
amounts of fibrinogen are converted to fibrin physiologically. D-­dimer
levels in healthy individuals increase with age; in a population-­based

98  | © 2017 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/ijlh Int J Lab Hem. 2017;39(Suppl. 1):98–103.
LINKINS and LAPNE |
      99

cohort, the upper 95th percentile for D-­dimer was 2.5 times greater in
people over the age of 70 compared to people less than 50 years old.1
D-­dimer levels are increased in almost all cases of acute VTE.
However, any process that increases fibrin production or breakdown
also increases D-­dimer levels. Examples of these conditions include
pregnancy, inflammation, cancer, and surgery. Therefore, an elevated
D-­dimer is not a specific test for VTE.
Among patients with VTE, D-­dimer levels vary with clot burden,
timing of measurement, and initiation of treatment. Among patients
with confirmed PE, higher D-­dimer levels are observed among patients
with larger emboli, such as those involving more than 50% of the lung
volume, compared to patients with smaller thromboses.2 Similarly,
among patients with DVT, higher D-­dimer levels are observed among
patients who have thrombosis extending above the level of the knee
compared to patients who have thrombosis confined to their calf
3
(proximal vs distal DVT). D-­dimer levels fall with increasing dura-
tion of VTE symptoms. Patients with confirmed DVT who have had
symptoms for more than 7 days have lower D-­dimer concentrations
3
compared to patients with DVT and a shorter duration of symptoms.
D-­dimer levels also fall rapidly after initiation of treatment (eg hep-
arin, low-­molecular-­weight heparin, vitamin K antagonists). Within
24 hours of heparin treatment, D-­levels fall by approximately 25%.4
Therefore, D-­dimer testing is most accurate in patients with larger clot
burden who have symptoms for less than a week and who have D-­
dimer testing performed before initiation of therapy.
1.3 | Principles of D-­dimer measurement
The process of measuring D-­dimer can be conceptually divided into
two steps: (i) D-­dimer fragments must be captured by monoclonal
antibodies, and (ii) the D-­dimer plus monoclonal antibodies must be
detected and quantified. Captured antibodies are immobilized on a
larger structure, such as a microplate well or membrane, or linked
to a latex bead or red cell. Detection antibodies can be labeled and
produce a colorimetric or fluorescent reaction that is quantified after
binding captured D-­dimer or can be linked to a large particle such as
a bead or red blood cell to induce agglutination. In sandwich-­based
assays, the capture and detection antibodies may have specificity
Learning Objectives
At the conclusion of this presentation, participants should
be able to do the following:
1. Summarize the general properties of D-dimer assays
2. Explain the concept of varying D-dimer thresholds ac-
cording to pretest probability and age
3. Describe the role of D-dimer testing in the diagnosis and
prognosis of venous thromboembolism
for different epitopes on the D-­dimer molecule. It follows that a D-­
dimer molecule must have the two different epitopes present to be
detected by such assays.5 In agglutination based assays, the same
monoclonal antibody serves as both capture and detection antibody.
Only ­D-­dimer fragments with multiple identical epitopes are captured
with these assays.5
1.4 | Assay types
The range of commercially available D-­dimer assays is daunting. A
recent publication reported 30 different assays using 20 different
monoclonal antibodies.6 The assays differ in the D-­dimer epitope tar-
geted by the antibody, method of capture and detection (as discussed
above), instrumentation required, and calibration standard. A brief
overview of the major categories of D-­dimer assays is given below.7,8
(Table 1) For a more comprehensive listing of assay types, see the re-
view by Riley et al.8
1.4.1 | Enzyme-­linked immunosorbent assays (ELISA)
Microplate ELISA
Plasma is added to microtiter wells that are coated with D-­dimer an-
tibody. After incubation, a second labeled antibody also specific for
D-­dimer is added to the well that binds to immobilized D-­dimer mol-
ecules and then produces a colorimetric reaction. Due to high sensitiv-
ity of these assays to low levels of D-­dimer, they are considered the

T A B L E   1   Comparison of categories of D-­dimer assays

ELISA ELFA Latex-­enhanced immunoturbidimetric Whole-­blood point of care

Description Quantitative Quantitative Quantitative Qualitative

Turnaround time 2-­4 h 35 min 15 min 2-­5 min
Sensitivitya (95% CI) 94% (86-­97) 96% (89-­98) 93% (89-­95) 83% (67-­93)
Specificitya (95% CI) 53% (38-­68) 46% (31-­61) 53% (46-­61) 71% (57-­82)
Advantages High sensitivity High sensitivity Comparable sensitivity to ELISA Can be performed at
Fully automated Fully automated bedside
Higher specificity
Disadvantages Labor-­intensive Moderate specificity Moderate specificity Observer dependent
Moderate specificity Lower sensitivity

ELISA, enzyme-­linked immunosorbent assay; ELFA, enzyme-­linked immunofluorescence assay.

a
Summary estimates for diagnosis of DVT reported in systematic review by Di Nisio et al.11
 |
100       LINKINS
reference standard assay for D-­dimer. However, they are time con-
suming and require specialized personnel and therefore are impracti-
cal for routine D-­dimer determination.
1.4.2 | Enzyme-­linked immunofluorescence assays
(ELFA)
Designed along the same principles as the microplate ELISA, ELFAs
are fully automated and binding of the detection antibody produces a
fluorescent product. These assays have a rapid turnaround time and
can be performed on single samples.
1.4.3 | Latex assays
These assays use latex particles coated with D-­dimer antibod-
ies. They are available in qualitative, semiquantitative, and
quantitative formats. Second-­generation latex-­enhanced immu-
noturbidimetric assays are fully automated, quantitative assays
that measure transmission of light with a photometric analyzer.
If D-­dimer antigen is present in the plasma, the antibody-­coated
latex beads agglutinate around molecules of D-­dimer producing
a large aggregate. The degree of light absorption by these larger
particles can then be measured. These assays have similar operat-
ing characteristics to ELFAs.
1.4.4 | Whole-­blood assays
A drop of whole blood is added to a slide followed by a reagent con-
taining a hybrid, bispecific antibody that binds to D-­dimer and a red
blood cell membrane antigen. If D-­dimer antigen is present, erythro-
cyte agglutination occurs. The agglutination is detected visually by the
operator. Other rapid point-­of-­care assays apply a similar principle,
using slides with a chromatography membrane containing immobi-
lized monoclonal antibodies conjugated to colloidal gold particles or
fluorescent-­labeled monoclonal antibodies that produce reactions
that are read by portable fluorometers.
From a laboratory perspective, D-­dimer testing is further compli-
cated by lack of a standardized unit of measure. D-­dimer levels are re-
ported as a unit of mass per volume. However, there are two different
units used to describe D-­dimer mass: (i) purified D-­dimer units (DDUs)
or (ii) fibrinogen equivalent units (FEUs). FEUs express the mass of D-­
dimer as the equivalent mass of fibrinogen that would be needed to
produce the D-­dimer in the sample. One FEU has approximately two
times the mass of one DDU; therefore, 1 μg/L of D-­dimer as measured
in DDU is about equal to 2 μg/L in FEU.5,9 There is no standardization
with respect to the mass unit used to report D-­dimer levels, with some
laboratories reporting D-­dimer concentration using DDU, while oth-
ers report FEU. Many laboratories also do not clearly identify which
mass unit they use. Furthermore, the magnitude of the unit of measure
varies widely across laboratories using the same assays (eg ng/mL,
μg/mL, μg/L). An additional problem is the lack of a universal calibrator
that can be used to standardize D-­dimer concentration across assay
types.6,10
and LAPNE
From a clinician’s perspective, variation in assay types means
variation in operating characteristics. Systematic reviews and meta-­
analyses of studies that have evaluated D-­dimer assays for diagnosis
of DVT and PE have reported sensitivities ranging from 69% to 97%
and specificity from 43% to 99%.11 Consequently, the results of clini-
cal studies performed using one assay cannot simply be extrapolated
to another.
Key point: All D-­dimer assays are not made equal. Each assay must
be independently and prospectively validated within the target popu-
lation of interest.
2 | D-­D IMER TESTING FOR DIAGNO SIS OF
DVT AND PE
A D-­dimer assay should not be used as stand-­alone test to diagnose
VTE. As outlined above, elevated D-­dimer levels are nonspecific; con-
sequently, a positive D-­dimer result in a patient suspected to have
VTE should be followed by an imaging test to confirm the diagnosis
(eg ultrasound of leg, computed tomography pulmonary angiogram of
lungs).
A negative D-­dimer result can exclude the diagnosis of VTE without
further testing, but only if the sensitivity of the test is high (>98%).12
However, high sensitivity comes at the cost of specificity. A D-­dimer
with high sensitivity (eg ELISA) will miss few patients who have VTE,
but due to low specificity, it will also produce a large number of false-­
positive results. False-­positive D-­dimer results lead to unnecessary
ultrasounds and CT scans. Ultrasounds cost money and time, but are
relatively innocuous. On the other hand, CT scans require exposure to
radiation and the risk of contrast dye-­induced nephropathy.
To improve the utility of D-­dimer testing in patients with suspected
VTE, D-­dimer measurement is usually combined with other tests as
part of a diagnostic algorithm. (Figure 1) A key component of validated
Suspected VTE
Clinical Pretest Probability
Low Moderate* High
D-dimer Diagnosc Imaging
Negave Posive Negave Posive
Diagnosc Imaging
Negave Posive
VTE Excluded Treat VTE Excluded Treat
F I G U R E   1   Diagnostic algorithm for venous thromboembolism.
*D-­dimer testing may also be performed to exclude VTE in this
category depending on the assay used (Linkins et al.21)
LINKINS and LAPNE
diagnostic algorithms is the application of a clinical decision rule be-
fore performing any diagnostic tests, including D-­dimer measure-
ment. Clinical decision rules are comprised of independent predictors
for disease (ie signs, symptoms, risk factors) that have been derived
from clinical studies and incorporated into a score. The score divides
patients into different groups according to their likelihood of having
DVT/PE (eg low/moderate/high, or likely/unlikely). Several clinical de-
cision rules for diagnosis of DVT and PE have been developed, but
the most validated include the Wells’ score and the Geneva score.13-18
In a typical VTE diagnostic algorithm, patients who are scored as
“low” or “unlikely” to have VTE have a D-­dimer level drawn. (Figure 1)
If the D-­dimer result is negative, VTE is considered excluded and no
further testing is performed. This is the primary role of a D-­dimer
assay—to safely exclude the diagnosis of VTE without requiring any
further investigations. On the other hand, if the D-­dimer result is pos-
itive, imaging tests are ordered. If the patient is scored as “moderate/
high” or “likely” to have VTE, D-­dimer testing is omitted and imaging
tests are performed. A D-­dimer result is unhelpful when the pretest
probability of VTE is “likely” for three reasons: (i) the result is unlikely
to be negative even if VTE is not present (ie many of the factors that
increase the risk of VTE will also increase D-­dimer levels, eg cancer,
recent surgery); (ii) a positive result still requires confirmatory imaging;
and (iii) there is uncertainty that D-­dimer testing can safely exclude
VTE in this subgroup. Algorithmic approaches similar to this one have
been well validated in clinical studies.19,20
Key point: D-­dimer assays play an important role in diagnostic al-
gorithms for VTE. They should always be used within the context of
pretest likelihood for VTE.
2.1 | D-­dimer thresholds
The D-­dimer level which defines the boundary between a “positive”
and “negative” result is referred to as the threshold or cutoff level. The
threshold is determined by the manufacturer, and it is important to
note that it is not equivalent to the upper limit of the reference inter-
val for the assay. Traditionally, the threshold used to define a positive
test was intentionally set low to maximize the sensitivity of D-­dimer
testing and limit the likelihood of missing patients with DVT or PE.
However, as discussed previously, maximizing sensitivity comes at the
cost of lowering specificity. If specificity is too low, the end result is
reduced clinical utility of the assay because few patients will have a
negative result (ie the majority will have a positive test and need to
proceed to diagnostic imaging).
Investigators have attempted to improve the clinical utility of D-­
dimer testing by raising the threshold that defines a positive test, pri-
marily among patients with “low” or “unlikely” clinical pretest probability
(C-­PTP), as determined by a clinical prediction rule. The principal behind
this approach is that the negative predictive value (NPV) of a diagnostic
test (ie the probability that a patient with a negative result truly does not
have the disease) is dependent on the prevalence of disease within the
population being tested. If the prevalence is low, the negative predictive
value of a test will increase because a negative result is more likely to be
a true negative. Reciprocally, if prevalence is low, the positive predictive
|
      101
value (PPV) of a test will decrease because a positive result is more likely
to be a false positive. In other words, if the prevalence of DVT in a pop-
ulation is low, elevated D-­dimer levels are more likely to be due to con-
ditions other than DVT; therefore, PPV is low and NPV is high.
2.1.1 | Varying D-­dimer threshold according to
C-­PTP
Varying the threshold according to a patient’s baseline C-­PTP for VTE
has been shown to improve the clinical utility of D-­dimer testing. As
previously discussed, the Wells’ score is a validated clinical decision
rule that separates patients into pretest probability categories—”low”
(5% prevalence of VTE), “moderate” (25%), or “high” (60%)—based
on the presence or absence of certain risk factors, physical signs,
and symptoms.13 VTE prevalence is low in patients with a low Wells’
score; therefore, a high D-­dimer threshold (less sensitive) can be used
to exclude VTE within this subgroup. Conversely, VTE prevalence is
high in patients with a high Wells’ score, which lowers the NPV of D-­
dimer testing and limits the safety and usefulness of D-­dimer testing
in this subgroup. Furthermore, hospitalized patients commonly have
other reasons to have elevated D-­dimer levels (eg malignancy, infec-
tion, surgery); therefore, a negative D-­dimer result in this subgroup
is both unlikely and unhelpful. Consequently, varying the D-­dimer
threshold according to C-­PTP and patient admission status may be
more efficient than using a one-­threshold-­fits-­all approach.
To test this strategy, Linkins et al. randomized patients with sus-
pected first acute DVT either to a uniform strategy where the same
D-­dimer threshold was used for all patients (<500 μg/L; STA®-­Liatest®
D-­Di; Stago, Asnières sur Seine Cedex, France), or a selective strategy
where the D-­dimer threshold varied according to C-­PTP.21 In the se-
lective strategy, a higher D-­dimer threshold (<1000 μg/L) was used for
patients with low C-­PTP; the manufacturer’s recommended threshold
(<500 μg/L) was used for patients with moderate C-­PTP; and D-­dimer
testing was not performed in outpatients with a high C-­PTP or inpa-
tients (irrespective of C-­PTP). The incidence VTE at 3 months was the
same in both arms (0.5%; difference 0.0%, 95% CI: −0.8% to 0.8%)
which confirmed the hypothesis that a higher D-­dimer threshold can
safely be used in patients with low C-­PTP. Furthermore, the selective
strategy resulted in 22% fewer patients requiring D-­dimer testing and
8% fewer requiring ultrasonography compared with uniform testing.
The primary limitation of this strategy is concern about the ability to
extrapolate the results to other D-­dimer assays.
2.1.2 | Varying D-­dimer threshold according to age
Another strategy that improves the clinical utility of D-­dimer test-
ing is to vary the threshold according to patient age. It is known that
­D-­dimer levels naturally increase with age; therefore, older patients
are less likely to have a negative result even if they do not have VTE.1
Righini et al. proposed that the D-­dimer threshold for patients above
50 years of age could be safely increased by multiplying their age in
years by 10 (eg D-­dimer threshold for a 60-­year-­old patient would
be 600 μg/L instead of the manufacturer’s threshold of 500 μg/L).22
 |
102       LINKINS
In their prospective trial, consecutive patients presenting to emer-
gency departments with suspected PE had C-­PTP determined by the
Wells’ score (likely or unlikely) or the revised Geneva score (likely or un-
likely). Patients who were categorized as “PE likely” proceeded directly
to diagnostic imaging. Patients who were categorized as “PE unlikely”
underwent D-­dimer testing. Six different quantitative D-­dimer assays
were used (VIDAS D-­Dimer Exclusion [bioMérieux, Marcy l’Etoile,
France]; Tina-­quant [Roche, Basel, Switzerland]; Cobas h 232 [Roche,
Basel, Switzerland]; STA-­Liatest D-­Dimer [Stago, Asnières sur Seine
Cedex, France]; D-­Dimer HS 500 [IL Diagnostics, Orangebury, NY];
and Innovance D-­Dimer [Siemens AG, Munich, Germany]). For patients
<50 years of age, the threshold for a positive test was 500 μg/L. For
patients ≥50 years of age, the threshold for a positive test was calcu-
lated by multiplying the patient’s age in years by 10. Patients with a
positive D-­dimer underwent diagnostic imaging, patients with a nega-
tive D-­dimer had no further testing and were not anticoagulated, and
all patients were followed for 3 months.
The investigators reported that the 3-­month rate of VTE in pa-
tients who had PE excluded based on a D-­dimer level higher than
500 μg/L, but lower than their age-­adjusted threshold, was 0.3% (95%
CI: 0.1%-­1.7%). Furthermore, in patients over age 75 with pretest
probability of “PE unlikely,” using the age-­adjusted threshold increased
the proportion of patients who had PE excluded without the need for
imaging by 23%.
Others have questioned the use of the age-­adjusted D-­dimer. In
an analysis of individual patient data from two clinical studies, Takach
Lapner et al. compared the proportion of patients who had VTE ex-
cluded using (i) age-­adjusted D-­dimer interpretation, (ii) a uniformly
higher D-­dimer threshold in all patients, and (iii) a higher threshold
only in younger patients (ie an opposite age-­adjusted strategy, but the
same average D-­dimer threshold as age-­adjusted interpretation).23
The results from this analysis showed that the gain in specificity with
the age-­adjusted strategy merely reflected use of a higher average
­D-­dimer threshold overall (ie 620 μg/L) and not just the selective use
of a higher threshold in the elderly. Furthermore, variation in D-­dimer
results with different commercial assays can make interpretation of
the age-­adjusted threshold difficult. For example, in one retrospective
study, an 86-­year-­old patient with PE had D-­dimer levels that ranged
24
from 590 to 1170 ng/mL across five commercial assays.
Key point: The clinical utility of D-­dimer testing is limited by a one-­
threshold-­fits-­all approach. Varying the threshold according to C-­PTP
or age improves utility, but the safe threshold values differ from one
assay to another. Laboratories and clinicians must be familiar with the
literature on the D-­dimer assay used at their institution to interpret
results.25
3 |  D-­D IMER TESTING FOR PROGNOSIS
In recent years, the role of D-­dimer testing has expanded to include
prediction of risk of recurrent VTE. VTE in patients who have no obvi-
ous provoking risk factors (eg recent surgery, immobilization of a limb,
cancer) is referred to as “unprovoked” or “idiopathic.” After 3 months
and LAPNE
of anticoagulant therapy, patients with unprovoked VTE are informed
that the risk of recurrence while off of anticoagulant therapy is 10%
in the first year and 30% over 5 years.26 Anticoagulant therapy carries
a risk of bleeding that increases with age; therefore, it is not a benign
treatment. Some patients will be uncomfortable with this level of risk
of recurrent VTE and choose to remain on anticoagulation indefinitely,
whereas others will choose to stop it. However, a substantial propor-
tion of patients are undecided about what to do. D-­dimer testing can
be helpful in these cases to aid decision-­making.
From our discussion about diagnostic testing for VTE, we know
that D-­dimer levels increase with acute thrombosis. For most patients,
the level drops over time; however, for some patients, the level re-
mains elevated which is thought to indicate a persistent thrombotic
state. Meta-­analyses of clinical trials have shown that the risk of re-
current VTE in patients with a positive D-­dimer after 3 months of an-
ticoagulant therapy is higher (8.9% per year; 95% CI: 5.8% to 11.9%)
than in patients with a negative D-­dimer (3.5% per year; 95% CI: 2.7%
to 4.3%).27 Overall, the evidence suggests that a positive D-­dimer
after completion of 3 months of anticoagulant therapy in patients with
a first unprovoked VTE predicts a risk of recurrence that is approxi-
mately double the risk in patients with a negative D-­dimer.
While D-­dimer testing to predict recurrence is used at many centers,
individual practice varies. As with diagnosis of VTE, there are concerns
about extrapolating the results of studies using one D-­dimer assay to
another assay. There is also controversy over the timing of when the
prognostic D-­dimer level should be drawn after stopping anticoagulant
therapy (3 weeks? 2 months?) and the threshold that should be used to
indicate a positive result (manufacturer’s threshold? a different thresh-
old?). One meta-­analysis using patient-­level study data suggested that
while timing and threshold level may not influence the predictive ability
of D-­dimer testing, gender does appear to be a factor.28 For reasons
that are not entirely clear, a negative D-­dimer in patients who have
completed 3 months of treatment does not predict a lower risk of re-
currence in men with the same accuracy as it does in women.
As with diagnosis of VTE, D-­dimer testing should not be used as
a stand-­alone test for predicting risk of recurrence. Clinical prediction
rules which include the result of D-­dimer testing as one factor in the
score have been developed.29-31 Furthermore, it is important to con-
sider the risk of bleeding on anticoagulant therapy as well as the pa-
tient’s personal preferences when making this important decision.
Key point: D-­dimer testing after 3 months of anticoagulant therapy
for a first unprovoked VTE can guide decisions about duration of ther-
apy, but should be used as part of a clinical decision model.
4 | CONCLUSIONS
D-­dimer testing plays an important role in the exclusion of VTE and,
to a lesser extent, in the prediction of recurrence in patients with a
first acute unprovoked VTE. However, wide variation in the types and
operating characteristics of D-­dimer assays means study results with
one assay cannot be extrapolated to another. Much work remains
to be done with respect to standardization of the performance and
LINKINS and LAPNE |
      103

reporting of D-­dimer assays as well as translation of results of D-­ 17. Klok FA, Kruisman E, Spaan J, et al. Comparison of the revised Geneva
dimer studies into clinical practice. score with the Wells rule for assessing clinical probability of pulmo-
nary embolism. J Thromb Haemost. 2008;6:40‐44.
18. Carrier M, Wells PS, Rodger MA. Excluding pulmonary embolism at
the bedside with low pre-­test probability and D-­dimer: safety and
CO NFLI CTS OF I NT E RE ST
clinical utility of 4 methods to assign pre-­test probability. Thromb Res.
Linkins received D-­dimer assays in-­kind from BioMérieux and Stago 2006;117:469‐474.
19. Geersing GJ, Zuithoff NP, Kearon C, et  al. Exclusion of deep vein
for a peer-­funded research study. Takach Lapner reports no conflict
thrombosis using the Wells rule in clinically important subgroups: in-
of interests. dividual patient data meta-­analysis. BMJ. 2014;348:g1340.
20. Carrier M, Righini M, Djurabi RK, et al. VIDAS D-­dimer in combination
with clinical pre-­test probability to rule out pulmonary embolism. A
REFERENCES systematic review of management outcome studies. Thromb Haemost.
2009;101:886‐892.
1. Haase C, Joergensen M, Ellervik C, Joergensen MK, Bathum L. Age-­
21. Linkins LA, Bates SM, Lang E, et al. Selective D-­dimer testing for di-
and sex-­dependent reference intervals for D-­dimer: evidence for a
agnosis of a first suspected episode of deep venous thrombosis: a
marked increase by age. Thromb Res. 2013;132:676‐680.
randomized trial. Ann Intern Med. 2013;158:93‐100.
2. De Monye W, Sanson BJ, Mac Gillavry MR, et  al. Embolus loca-
22. Righini M, Van Es J, Den Exter PL, et al. Age-­adjusted D-­dimer cutoff
tion affects the sensitivity of a rapid quantitative D-­dimer assay in
levels to rule out pulmonary embolism: the ADJUST-­PE study. JAMA.
the diagnosis of pulmonary embolism. Am J Respir Crit Care Med.
2014;311:1117‐1124.
2002;165:345‐348.
23. Takach Lapner S, Julian JA, Linkins LA, Bates SM, Kearon C.
3. Chapman CS, Akhtar N, Campbell S, Miles K, O’Connor J, Mitchell VE.
Questioning the use of an age-­adjusted D-­dimer threshold to exclude
The use of D-­Dimer assay by enzyme immunoassay and latex agglu-
venous thromboembolism: analysis of individual patient data from
tination techniques in the diagnosis of deep vein thrombosis. Clin Lab
two diagnostic studies. J Thromb Haemost. 2016;14:1953‐1959.
Haematol. 1990;12:37‐42.
24. Mullier F, Vanpee D, Jamart J, et al. Comparison of five D-­dimer re-
4. Couturaud F, Kearon C, Bates SM, Ginsberg JS. Decrease in sensitivity
agents and application of an age-­adjusted cut-­off for the diagnosis
of D-­dimer for acute venous thromboembolism after starting antico-
of venous thromboembolism in emergency department. Blood Coagul
agulant therapy. Blood Coagul Fibrinolysis. 2002;13:241‐246.
Fibrinolysis. 2014;25:309‐315.
5. Reber G, de Moerloose P. D-­dimer assays for the exclusion of venous
25. Goodwin AJ, Higgins RA, Moser KA, et  al. Issues surrounding age-­
thromboembolism. Semin Thromb Hemost. 2000;26:619‐624.
adjusted D-­dimer cutoffs that practicing physicians need to know
6. Longstaff C, Adcock D, Olson JD, et al. Harmonisation of D-­dimer -­A
when evaluating patients with suspected pulmonary embolism. Ann
call for action. Thromb Res. 2016;137:219‐220.
Intern Med 2016;166:361‐363.
7. Adam SS, Key NS, Greenberg CS. D-­dimer antigen: current concepts
26. Kearon C. Long-­term management of patients after venous thrombo-
and future prospects. Blood. 2009;113:2878‐2887.
embolism. Circulation. 2004;110:I10‐I18.
8. Riley RS, Gilbert AR, Dalton JB, Pai S, McPherson RA. Widely
27. Verhovsek M, Douketis JD, Yi Q, et al. Systematic review: D-­dimer to
Used Types and Clinical Applications of D-­Dimer Assay. Lab Med.
predict recurrent disease after stopping anticoagulant therapy for unpro-
2016;47:90‐102.
voked venous thromboembolism. Ann Intern Med. 2008;149:481–490.
9. Jennings I, Woods TA, Kitchen DP, Kitchen S, Walker ID. Laboratory
28. Douketis J, Tosetto A, Marcucci M, et al. Patient-­level meta-­analysis:
D-­dimer measurement: improved agreement between methods
effect of measurement timing, threshold, and patient age on ability of
through calibration. Thromb Haemost. 2007;98:1127‐1135.
D-­dimer testing to assess recurrence risk after unprovoked venous
10. Olson JD, Cunningham MT, Higgins RA, Eby CS, Brandt JT. D-­dimer: sim-
thromboembolism. Ann Intern Med. 2010;153:523‐531.
ple test, tough problems. Arch Pathol Lab Med. 2013;137:1030‐1038.
29. Rodger MA, Kahn SR, Wells PS, et al. Identifying unprovoked throm-
11. Di Nisio M, Squizzato A, Rutjes AW, Buller HR, Zwinderman AH, Bossuyt
boembolism patients at low risk for recurrence who can discontinue
PM. Diagnostic accuracy of D-­dimer test for exclusion of venous throm-
anticoagulant therapy. CMAJ. 2008;179:417‐426.
boembolism: a systematic review. J Thromb Haemost. 2007;5:296‐304.
30. Eichinger S, Heinze G, Jandeck LM, Kyrle PA. Risk assessment of
12. Bates SM, Takach Lapner S, Douketis JD, et al. Rapid quantitative D-­
recurrence in patients with unprovoked deep vein thrombosis or
dimer to exclude pulmonary embolism: a prospective cohort manage-
pulmonary embolism: the Vienna prediction model. Circulation.
ment study. J Thromb Haemost. 2016;14:504‐509.
2010;121:1630‐1636.
13. Wells PS, Hirsh J, Anderson DR, et al. A simple clinical model for the
31. Tosetto A, Iorio A, Marcucci M, et  al. Predicting disease recur-
diagnosis of deep-­vein thrombosis combined with impedance pleth-
rence in patients with previous unprovoked venous thromboem-
ysmography: potential for an improvement in the diagnostic process.
bolism: a proposed prediction score (DASH). J Thromb Haemost.
J Intern Med. 1998;243:15‐23.
2012;10:1019‐1025.
14. Wells PS, Anderson DR, Rodger M, et al. Derivation of a simple clini-
cal model to categorize patients probability of pulmonary embolism:
increasing the models utility with the SimpliRED D-­dimer. Thromb
Haemost. 2000;83:416‐420. How to cite this article: Linkins L-A, Takach Lapner S. Review
15. Le Gal G, Righini M, Roy PM, et al. Prediction of pulmonary embolism of D-­dimer testing: Good, Bad, and Ugly. Int J Lab Hem.
in the emergency department: the revised Geneva score. Ann Intern
2017;39(Suppl. 1):98–103. https://doi.org/10.1111/ijlh.12665
Med. 2006;144:165‐171.
16. Chagnon I, Bounameaux H, Aujesky D, et al. Comparison of two clini-
cal prediction rules and implicit assessment among patients with sus-
pected pulmonary embolism. Am J Med. 2002;113:269‐275.