Monte Minute‎ > ‎Monte Minute‎ > ‎

Pulmonary Embolism and the PIOPED Study - 8/28/2012

posted Aug 28, 2012, 1:55 PM by Rohit Das   [ updated Dec 27, 2012, 6:59 AM by Purnema Madahar ]

Today, noontime was all about feedback about our current state of affairs. I appreciate everyone coming and making your voices heard, and I assure you that I’ll do my best to echo them.

We did have our weekly resident journal club today, and talked pretty extensively about an old school, but seminal article, the PIOPED study (attached – hard to get a copy of this on the Einstein website since it’s from 1990..so make sure you save it!). We’ve talked a bit about PE in past dailys, but always worth adding a little more…

The PIOPED study was designed to assess the diagnostic accuracy and predictive value of ventilation/perfusion scanning in the diagnosis of PE. More specifically, the authors looked at the accuracy of variable “probability” V/Q scans. Just like any other test, V/Q scan results have different cutoffs, and each cutoff comes with its own diagnostic accuracy and predictive value. Table 1 of the PIOPED paper provides an overview of the various V/Q interpretation categories, ranging from “high probability” to “normal.” The investigators took what was ultimately 755 patients who had V/Q scans, and compared the results of those studies with the gold standard, pulmonary angiogram. So, let’s go over some of the key numbers:

  • Using “high probability” findings, V/Q scan had a sensitivity of 41%, and a specificity of 97%. When accounting for high, intermediate and low probabilities, the sensitivity increased to 96% and the specificity decreased to 10%.
  • This reinforces a central concept – when one lowers the cutoff for what defines a positive test, there will be a decreased amount of FALSE NEGATIVES, at the expense of MORE FALSE POSITIVES. Based on the definitions of sensitivity and specificity, this will INCREASE sensitivity and DECREASE specificity. Above is a perfect example.
  • More along the lines of V/Q scan’s predictive value, a high probability scan had a positive predictive value of 91%. A normal V/Q scan had a negative predictive value of around 90%l, and a low probability scan had a NPV of around 85%. Very importantly, patients with a high probability scan AND a history of previous VTE disease had a much lower PPV – around 75%.
  • Finally, when taking into consideration clinically assigned pre-test probabilities, the predictive capabilities of V/Q scan were enhanced:
    • High Pre-Test + High Prob V/Q Scan = PPV of 96%
    • Low Pre-Test + Low Prob V/Q scan = NPV of also around 96%
  • Again, central concept à the predictive value of a test is highly dependent on the prevalence of the disease you’re trying to diagnose in the population you’re testing. A positive test is much more valuable in a patient you suspect has the disease, and conversely, a negative test is much more valuable in patients you don’t suspect have the disease.
  • With other combinations, V/Q scan is not as predictive. For example, with an intermediate pre-test probability and a high probability V/Q scan, the post-test probability was 86%. That value is not terribly good, especially for a potentially harmful treatment (i.e., anticoagulation).

So, whenever you get a V/Q scan test back, try to interpret the results in the context of two things – the nature of the perfusion/ventilation defect (high vs. intermediate vs. low) AND your clinical suspicion for the patient having PE (nowadays based on Well’s criteria). PE has been extensively studied enough where, in combination, those two factors can give you a confident probability of whether or not a patient has, or does not have, pulmonary embolism.

We also talked about a very interesting commentary around the A-HeFT trial – a study which showed that an isosorbide-hydralazine combination led to decreased mortality in black patients with CHF. There has been a lot of criticism surrounding this article, particularly around how the study was focused around a particular ethnic group, in the absence of previous data in the general population. Several correspondences (one is attached) suggest that the drug should have been tested in the general population, and choosing self-identifed blacks as the target population was both a misleading suggestion of potential genetic predispositions to therapeutic response, and also a business tactic to get a patented drug on the market…so yea...some food for thought…

Remember the concepts we've talked about regarding PE, including assessing severity and overdiagnosis, and try to incorporate that into some fundamental knowledge around diagnostics. Hope this little bit helped...until tomorrow!

Value of the Ventilation/Perfusion Scan in Acute Pulmonary Embolism
PIOPED Investigators, JAMA 1990, Volume 263 (20): 2753-2759

Taylor et. al., NEJM 2004, Volume 351: 2049-2057

Bloche, NEJM 2004, Volume 351 (20): 2035-37

Comments