The ALT-70 predictive model outperforms thermal imaging for the diagnosis of lower extremity cellulitis: A prospective evaluation.

Abstract

We previously demonstrated that dermatology consultation substantially reduces the rates of misdiagnosis of cellulitis; however, broad implementation of dermatology consultation is impractical on account of existing practice patterns and reimbursement systems. Meanwhile, efforts to improve diagnostic accuracy have culminated in point-of-care tools, including the ALT-70 predictive model for lower extremity cellulitis and thermal imaging. To prospectively evaluate the performance of the ALT-70 predictive model and thermal imaging in diagnosing lower extremity cellulitis in a head-to-head comparison. We collected ALT-70 and thermal imaging data from patients with presumed lower extremity cellulitis and compared classification measures and accuracy for the ALT-70 predictive model, thermal imaging, and combination testing (ALT-70 predictive model plus thermal imaging). We enrolled 67 patients with ALT-70 and thermal imaging data. The ALT-70 predictive model conferred the highest sensitivity (97.8%) and negative predictive value (90.9%), whereas combination testing had the highest specificity (71.4%) and positive predictive value (86.6%). The ALT-70 predictive model had improved classification measures compared with thermal imaging. Combination testing conferred a marginal benefit compared with the ALT-70 predictive model alone. Single-center design may limit generalizability. The ALT-70 predictive model outperformed thermal imaging in diagnosing lower extremity cellulitis. The accuracy of the ALT-70 predictive model was high and consistent with its performance in previously published literature. Broad implementation of the ALT-70 predictive model in clinical practice may decrease the rates of misdiagnosis of lower extremity cellulitis.