Current issues of ACP Journal Club are published in Annals of Internal Medicine


D-dimer levels detected DVT in patients hospitalized for stroke rehabilitation

ACP J Club. 1997 Mar-Apr;126:43. doi:10.7326/ACPJC-1997-126-2-043

Related Content in the Archives
D-dimer levels detected DVT in patients hospitalized for stroke rehabilitation

Source Citation

Harvey RL, Roth EJ, Yarnold PR, Durham JR, Green D. Deep vein thrombosis in stroke. The use of plasma D-dimer level as a screening test in the rehabilitation setting. Stroke. 1996 Sep;27:1516-20.



To determine whether plasma D-dimer level was a useful screening test for deep venous thrombosis (DVT) in patients who were hospitalized for stroke rehabilitation.


Blinded comparison of plasma D-dimer levels measured by an enzyme-linked immunosorbent assay and patient characteristics with venous duplex ultrasonography (VDU) to detect DVT.


Rehabilitation unit of a Chicago hospital in the United States.


105 patients (mean age 63 y, 50% women) who were hospitalized in a stroke rehabilitation unit. Inclusion criteria were acute ischemic or hemorrhagic stroke within the previous 3 months and lack of mobility (patients could walk < 30 m at admission). Exclusion criteria were renal failure, cancer, presence of a filter through the vena cava, or receipt of warfarin.

Description of tests and diagnostic standard

Plasma D-dimer and VDU were done within 24 hours of each other. The clinical features that were considered were age, sex, date of stroke, and date of admission. The D-dimer level analyses were done in duplicate in a blinded manner with the Asserachrom D-Di enzyme immunoassay kit (Diagnostica Stago). VDU was done by an experienced technologist and interpreted by a vascular surgeon. DVT was determined to be present if a thrombus was visualized on B-mode VDU or if noncompressibility was noted in proximal deep veins (femoral, superficial femoral, and popliteal).

Main outcome measures

Actual sensitivity and specificity at different plasma D-dimer levels and estimated sensitivity and specificity if the study was repeated in additional samples of patients (jackknife validity analysis).

Main results

14 patients (13%) had DVT. No clinical features predicted DVT. Using results directly from the study, the sensitivity, specificity, and {likelihood ratios for a positive and negative test}* for DVT for D-dimer levels of > 1591 ng/mL were 93% (95% CI 66% to 99.8%), 79% (CI 69% to 87%), {4.4 and 0.9}*, respectively, and for D-dimer levels of > 1092 ng/mL were 100% (CI 99% to 100%), 66% (55% to 76%), {3.1, and 0.0}*, respectively. The jackknife-adjusted sensitivity and specificity for DVT were 79% and 78% for plasma D-dimer levels of > 1591 ng/mL and 100% and 66% for D-dimer levels of > 1092 ng/mL.


Plasma D-dimer levels ≤ 1092 ng/mL excluded DVT in patients who were hospitalized for stroke rehabilitation. D-dimer levels > 1092 ng/mL require confirmation of DVT by other tests.

Sources of funding: American Heart Association; Department of Health and Human Services; the Women's Board of the Rehabilitation Institute of Chicago.

For article reprint: Dr. R.L. Harvey, Rehabilitation Institute of Chicago, 345 East Superior Street, Chicago, IL 60611, USA. FAX 312-908-1833. E-mail

*Numbers calculated from data in article.


The potential use of plasma D-dimer levels for screening or diagnosing DVT or pulmonary embolism has been assessed during the past decade. The study by Harvey and colleagues supports and extends previous work and is important because it raises the possibility that screening for DVT may be done using an inexpensive, noninvasive test. However, D-dimer levels were tested against compression ultrasonography rather than contrast venography. Compression VDU is not sensitive or specific in asymptomatic persons (1).

D-dimer levels almost always increase with thromboembolic events; however, levels also increase in inflammatory states. Therefore, although D-dimer tests are very sensitive (> 90%), they are not specific (< 80%). Hence, D-dimer levels may be more useful for screening for DVT but less useful for its diagnosis. Other screening tests for DVT (such as liquid crystal thermography [2]) have been assessed in patients with stroke, but these tests also have high sensitivity and poor specificity.

Once a screening test has been identified, its introduction into routine clinical practice depends on availability, standardization, and cost. D-dimer measurements are not universally available, and measurement methods have yet to be standardized. Further, cutoff points will probably vary between hospitals, and it may be necessary for individual laboratories to develop their own cutoff values.

The current uncertainties relating to D-dimer measurement methods suggest that it should remain a research tool. Physicians should continue to screen for DVT using clinical examination, although it is insensitive (2), and to diagnose DVT anatomically using VDU or venography. It is also important that measures to prevent DVT be prescribed for all patients with stroke; immobilized patients should have compression stockings, patients with ischemic stroke should receive aspirin, and heparin should be limited to high-risk patients.

Philip M. Bath, MD
King's College School of Medicine & DentistryLondon, England, UK


1. Jongbloets LM, Lensing AW, Koopman MM, Buller HR, ten Cate JW. Lancet. 1994;343:1142-4.

2. Cameron EW, Sachdev D, Gishen P, Martin JF. Eur J Clin Invest. 1991;21:548-50.