Detecting PEs: Role of VQ scan =============================== Yes, I'm very glad you asked me that question (as Richard Nixon used to say). The PIOPED study (1) was designed to answer precisely that question. I had alluded to it in a previous post but had been unable (until recently) to find the article. In this study 931 patients underwent V/Q scanning and the results compared to angiography. The following table shows the scan sensitivity and specificity with increasingly liberal criteria for PE: Scan Category Sensitivity Specificity High prob 41% 97% High and Intermed 82% 52% High, Inter and Low 98% 10% What these results indicate is that if one insists on the presence of a high probabilty scan to diagnose PE then (100-41) = 59% of them will be missed. If instead one accepts _either_ a high or intermediate probabilty scan then the false negative rate is still (100-82) = 18%. Of course if one accepts _any_ abnormal scan then 98% would be detected but with only 10% specificity. Ninety percent of these patients would be needlessly anticoagulated. The authors then compared scan results with a priori clinical suspicion (The a priori probabilities are: high >80%; intermed = 20-79%; low <20%) Suspicion--> High Intermediate Low All Scan Category High 96% 88% 56% 87% Intermediate 66% 28% 16% 30% Low 40% 16% 4% 14% Normal 0% 6% 2% 4% The last (normal) category actually included a 'near normal' subcategory. The consensus being that a totally normal V/Q scan reliably excludes clinically significant PE. Of note is that a person with high clinical suspicion, even the presence of a low probabilty scan, still has a 40% chance of PE! I believe that this result confirms your suspicions about the unreliability of V/Q scanning _alone_ to exclude PE. A small study (3) done previously also noted a relatively low sensitivity and specificity of scans. The high probabilty scans had only a 65% (as opposed to 87% in PIOPED) specificity indicating the hazards of anticoagulating patients based exclusively on V/Q. Another study (7) confirmed the dangers of excluding PE based upon intermediate probabilty interpretations. Single V/Q mismatches although sometimes considered low probability actually belong in the intermediate probaility group (8,9) and carry about a 30% risk of PE. Considering the large number of patients who are deemed to be at risk with non-diagnostic V/Q scans, it is perhaps surprising at the small number of angiograms that are actually done (6). Several stradegies utilizing ultrasonography to initiate (5) but not exclude (4) anticoagulation have been developed. In the former case it may be useful as a method to avoid angiography in patients with intermediate probability scans who have DVT by ultrasonagraphy. Finally a recent study appears to question the sensitivity of helical CT scans particularly for subsegmental defects (2). In summary V/Q scanning _used_alone_ has an unaceptably high rate of false positive and false negative results and needs to be supplemented with a priori probability estimates, venous ultrasonography and, in some cases, pulmonary angiography. H. Louzon MD references: Re: dismissing low-prob V/Q scans, you might want to scare up a copy of Hull's 1995 paper from Arch Intern Med (155):1845-1851 showing that Lo-prob v/q patients with poor underlying cardiopulmonary reserves were nearly 60 times more likely to die from PE than patients with 'worse' v/q scans who had better cardiopulmonarty reserves (quite an interesting paper). --Craig Feied, MD ---------------------- >one of the pulmonlogists suggested that in this case, bilateral leg >dopplers could be done in lieu of angio. if negative, send home for >serial dopplers. His point was that even if she'd had a PE, if the >dopplers are negative she won't have another one. Yes, I've heard this proposed. I've also reviewed a number of legal cases related to deaths caused this way. The concept is fatally flawed because the negative predictive value of duplex ultrasound in a patient with PE is extremely low. In patients with proven PE, the source cannot be identified in nearly one-third, yet they go on to re-embolize and die just like the ones whose DVT source is visible. >Questions: >do you agree with the pulmonologist? The pt had decent reserve >according to those studies as her p02 was 61 (the studies report p02 >less than 50 is necessary for dx of poor cardiopulm reserve). However, >she felt quite SOB and felt better with 02. pco2 was 38. No. These guys are crazy. They read one little slice of a huge body of literature and then they want to play games with the single most common cause of unexpected death in the world. The likelihood of immediate mortality goes up as the immediate PO2 goes down, but the overall mortality is the same regardless of the PO2 at the time the diagnosis is made. Immediate-death-risk-stratification based on PO2 is silly when we can do cardiac ultrasound to assess RV dysfuntion, which is the direct measure of decompensation. And all of this risk-stratification has value ONLY if it is being used to decide who gets some otherwise-unavailable intervention, like pulmonary embolectomy. I can remove one lung and not produce much symptomatic dysfunction nor hypoxemia. By the time patients are profoundly SOB, hypoxemic, or have an iota of transient hypotension they are well into their 'reserves' and are headed for death or permanent cor pulmonale. I thrombolyse everybody with exhausted cardiopulmonary reserves. Most pulmonologists just want to avoid thinking about PE, because they can't see it on bronchoscopy. Cardiologists do a much better job, for obvious reasons. > >if we sent her home, at what time interval would you perform serial >dopplers? I'd never send home a patient with probable PE without an angiogram. I feel so strongly about this that I testify for the plaintiffs in such cases. As to when to re-duplex -- in a patient with suspicion of DVT or PE with a NORMAL perfusion scan (or a negative angiogram) and normal duplex, I bring them back and scan them every 24 hours until I can be sure what is going on. It is not uncommon for the second duplex to show so much thrombus that we think the first was incompetently performed. Sometimes this is true, but thrombus can grow incredibly quickly. I've seen videos of thrombus forming around an intravascular ultrasound transducer and filling a vessel for 10-15 cm within just a few minutes. > >do you buy the claim that if the pt has occult pelvic thrombus, it would >not cause a problem until it extended into the thigh and was therefore >visible? This is the kind of canard that just won't die. No reasonable study has ever suggested such a thing, and there is absolute proof that the opposite is true. Death from thrombus at any site has been documented to be common and correlates with the quantity of clot. People die from arm DVT, from pelvic DVT, and from calf DVT at rates that are very close to the death rates from thigh DVT. > >lastly, what is the yield on dopplering asymptomatic legs? > Unfortunately, the yield is very poor. Several prospective 'gold-standard' studies showed that up to 60% of cases of DVT are not detected when duplex is used to examine asymptomatic legs. This doesn't even include all the cases where thrombus was in the pelvis or abdomen -- only includes leg DVT that could be seen on venogram. --Craig Feied, MD