These studies pretty much all end up saying the same thing - academic faculty can't agree on the presence or absence of differentiating characteristics between abscess and cellulitis. This particular study is in a pediatric population, and, there's a lot of kappa and absolute agreement to comb through in their tables, but, basically, about 20% of the time two attendings substantially disagreed. The authors then follow this up by observing that an I&D was performed 75% time, and purulent material was found 92% of the time.
The best conclusion from this might be - if there's some ambiguity, put a scalpel in it. I'd say this is reasonable - because we've seen a hundred times the child who bounces into the ED on day 3 of cephalexin for cellulitis because what he really had was a MRSA abscess to begin with.
Or, if you have an ultrasound with a high-frequency probe, you might be able to differentiate homogenous hyperemia from fluid collection.
"Interexaminer Agreement in Physical Examination for Children With Suspected Soft Tissue Abscesses"
www.ncbi.nlm.nih.gov/pubmed/21629150
