So here is the deal. D-dimer is a good way to rule out PE, but it is also a good way to over order CTPE’s. It needs to be used wisely.

There are a number of risk stratification rules that can be used to eval for PE. The best known is PERC. But you only use the PERC pathway if the provider has a low pre-test probability that the patient has a PE. “Pre-test probability” is a very fuzzy concept. If a patient comes in with chest pain, Mike may have a pretest probability that they have a PE of 4%, I may have it at 6% and Heidi may have it at 2%. Where do we get those numbers? We pull them out of our experience and to some extent out of thin air. In general the emergency medicine overlords have said that we need to catch 98% of PE’s, which means that getting our (post-test) probability below 2% is the magic number where we do not need to order more test before we say there is a low chance of a PE.

Lucky for us we have the Wells Criteria. The Wells Criteria are a number of factors that when you put them together give you a pretest probability that is based on some science. (One key point is that the physicians gestalt is a major part of the Well's criteria scoring, so your gut is not taken out of the picture--and once you have seen enough patients to have a well developed gestalt, your gut may be better at placing patients into low and high risk categories than the Wells criteria). Overall the goal is to slot your patient into a risk category; if they fall into the low risk category, you then check your work by using the PERC decision algorithm.

        So where does d-dimer come into play. This depends on who you ask. I know some physicians who think it is a garbage test and they never use it. Their argument is that if you think they have a PE, you should scan them. If you don’t think they have a PE, you should use your brain and then not-scan them. Their argument is backed up by studies that show that using d-dimers leads to huge rates of false negative CT scans. One singe-center study showed that of 118 patients with an elevated d-dimer and concern for a clot, 5 patients had a PE. The costs of the >95% of the scans that were negative was over $200,000 and there is about a 1/10 chance that one patient scanned in that cohort will develop cancer from their scan.
        
        While I think that the above decision making pathway (Scan or Don’t Scan) has some merit, it is not the generally accepted way to do it.  The recommendations are risk stratify, generally with the Well's criteria—>PERC if low risk—>if positive PERC then get a d-dimer—>if dimer positive get a CTPE.
    
        Let's go back to the question of pre-test probability. Every person with influenza or pneumonia who comes through the door will be PERC positive and may have an elevated d-dimer. So we need to be smart about how we use the test. If the story is clearly infectious or clearly traumatic, your pre-test should be below 2% for PE and you are done at that point. No dimer. No CTPE. 

If you have a patient with CP and SOB and leg swelling, then your pre-test probability is not low based on your gestalt (which is much better than any quantified algorithm at putting patients into a high risk category). Then you get the CTPE because the situation is yelling at you, “Hey, over here, I’m a PE.”

Penaloza, A, Roy, PM, at. al. Comparison of the Unstructured Clinician Gestalt, the Wells Score, and the Revised Geneva Score to Estimate PretestProbability for Suspected Pulmonary Embolism. Annals of Emergency Medicine. Volume 62, No 2. Page 117-124: August, 2013

Chopra N, Doddamreddy P, Grewal H, Kumar PC. An elevated D-dimer value: a burden on our patients and hospitals. Int J Gen Med. 2012;5:87–92. doi:10.2147/IJGM.S25027