Critically Appraised Topic:
Is compression ultrasound sufficient for diagnosing deep vein thromboses in the outpatient arena?
Date: Jan 14, 1998
Appraised by Eric Terman, MD
Clinical Bottom Lines:
1. Ultrasound with a follow-up study between 5 and 7 days, after the initial test, makes a bad outcome in patients with potential DVTs very unlikely. Much of this risk relates to the prevalence of the condition.
2. If there is high suspicion of a clot as a result of known predisposing factors then the diagnosis needs to be pursued with more vigor.
Consecutive outpatients were determined to have a normal or abnormal scan.
If abnormal the patients had a venogram, and if this was not done, they were started on heparin.
If the scan was normal, it was repeated 5 to 7 days later.
End points were death, or thromboembolic event.
Exclusions: if compression ultrasonography could not be done or there were technical limitations, follow-up was not assured, the patient had received therapeutic doses of heparin for more than 24 hours, or if the patient was pregnant.
Patients were followed up to 3 months after initial contact.
A two x two table of the data shows:
|number of patients||bad outcome (%)|
|nl scan||335||2 (0.6%)|
|1 abnl scan||70||4 (5.7%)|
Calculations: The positive likelihood ratio of having one abnormal scan and then having a bad outcome is: 4/6 divided by 64/397= 4.135.
The negative likelihood ratio of no bad outcome after two negative tests is 2/6 divided by 333/397=0.397
1. Although this article did not address sensitivities and specificities, since a gold standard was not used for all patients, it did address a more important issue. What is the chance of having a bad event if your scans are negative? The answer looks like its very low. The reason its low is that the prevalence is low. This is why although the numbers look impressive the calculated likelihood ratios are not that impressive.
2. The researchers intentionally did not include a gold standard for negative tests because they wanted to see what the long term results would be if the patients were not treated. If they had obtained a venogram which showed clot it probably would have been unethical to withhold treatment. Unfortunately it is difficult to know if patients did well because the chances of a bad outcome were so low to start or for another reason.
3. The study probably needed more patients so that there were more events.
4. Although I would feel comfortable withholding heparin for a normal person, if your suspicion for clot is high secondary to known predisposing factors, eg cancer, lupus..., then you probably need to pursue the diagnosis more.
5. The study is actually good, but it depends on the prevalence of the disease.
Birdwell BG, Raskob GE, Whitsett TL, Durica SS, Comp PC, George JN, Tytle TL and McKee PA. The Clinical Validity of Normal Compression Ultrasonography in Outpatients Suspected of Having Deep Venous Thrombosis. Annals of Internal Medicine 1998: 128:1 1-7.