"Should I get a D-Dimer test or CT chest angiogram on my patient with atypical chest pain to rule-out a pulmonary embolism?" This is a common question asked by emergency physicians on a routine basis.
Here are 3 clinical prediction rules: PERC, Wells, and Simplified Geneva Score. Personally, I've never used the Geneva Score, but it's worth looking at.
NOTE: These rules should be used with caution, because none of these scoring protocols are perfect. For instance, in a very recent publication in the Journal of Thrombosis and Haemostasis, the authors found that the PERC rule does not actually safely exclude PEs. Big bummer for us clinicians.
Thanks to Dr. Kit Tainter (Mount Sinai PGY-4 EM resident) for coming up with the idea for this card!
Or, if you want the contrarian point of view - Dr. Newman and Dr. Schriger in the most recent Annals make the case that our testing and treatment strategies for pulmonary embolism are more harmful than helpful. So, if you believe it, you could throw away this card, never test for PE, and sleep well at night - although it would probably be more reasonable to diagnose and treat the obvious, physiologically significant - and far more infrequent - PEs.
ReplyDeletePubmed:
http://www.ncbi.nlm.nih.gov/pubmed/21621091
With my blog entry for same:
http://emlitofnote.blogspot.com/2011/06/testing-for-pulmonary-embolism-is-more.html
Excellent point. It's a critical and very controversial issue. Thanks for the link.
ReplyDeleteI think the Newman/Schriger review is excellent, and very provocative. Perhaps our treatment and testing thresholds should be adjusted on a societal level.
ReplyDeleteRegarding the Hugli study, they looked at a very different prevalence of PE than the Kline PERC validation study from 2008. This is a reminder of an important concept of pre-test probability that I think is often overlooked.
Perhaps serial application of clinical prediction aids could stratify patients into a group with a low prevalence of PE in whom we can then safely apply the PERC rule; or even better, maybe someone will give us some data on whom we really need to diagnose or treat in the first place!
I'm not very smart when it comes to statistics (or other things) but it seems weird that the recent study had such a high prevalence of the disease. When the prevalence is so high, is a prediction rule even applicable?
ReplyDeleteWould love to hear Jeff Kline's thoughts!
ReplyDeleteInteresting comments from everyone. Yes, prevalence plays a big role when you are talking about positive and negative predictive values (PPV, NPV). I'm not even sure what the prevalence rate is at my site -- presumably on the low end, from what I can guess.
ReplyDeleteIn the end, the right thing to do is to change our testing thresholds on a societal level, but our medicolegal systems makes this change impossible.