I don't remember the source, but it is because d-dimer should be order when the patient arrives in the ED, i.e. to help differentiate between CHF versus other possible factors, because this is when it was shown in studies to have predictive value, I think the studies looking at inpatient evaluation did not show a usefullness for ruling out PE (or maybe there haven't been enough of these studies done), I postulate because there are so many things that can happen inpatient to elevate d-dimer, i.e. surgery, infection, so it is less useful in this population, but maybe no one really knows why. I do believe that a very large percentage of ICU patients, and a lesser extent of general medicine patients have a positive d-dimer without PE . . . so it is less useful. Somebody please post these article references as I can't find them and haven't read them in a while.