Hello,
I'm a long-time lurker, first-time poster. As a microbiology lab. tech with experience using qPCR and immunoassays for Salmonella, I think I can offer some insight here.
Like Charles C said, it's a question of limit of detection. If you look at methods at the FDA BAM (http://www.fda.gov/F...s/ucm070149.htm) you ideally are detecting one viable cell per 25 g of material. That limit is not going to be achievable by qPCR, immunoassay, or similar methods without enrichment - so there's no way to dispense with the enrichment step and maintain that L.O.D. As the original post alluded to, there is no way to establish a correlation between the number of cells post-enrichment and pre-enrichment, so by doing the enrichment the method is no longer quantitative.
If you had something that is highly contaminated (say, hundreds of cells per gram), yes you can do a direct DNA extraction and run qPCR. I suppose this would be useful as an in-process control but it wouldn't meet criteria for finished-products.