PracticeUpdate: Dermatology - Vol 1 - No.1 - 2017

EDITOR’S PICKS 6

Journal of the American Academy of Dermatology A predictive model for diagnosis of lower extremity cellulitis COMMENT By Boris D Lushniak MD, MPH T he word diagnosis is derived through Latin from Greek from a word meaning “to dis- cern, distinguish.” Since the time of our

Take-home message • In this cross-sectional study of 259 patients admitted through the emergency department between 2010 and 2012 with a diagnosis of lower extremity cel- lulitis, 30.5%were ultimately found to have beenmisdiagnosed with cellulitis. Variables associated with a true diagnosis of cellulitis included asymmetric involvement, leukocytosis, tachycardia, and an age >70. These variables were converted into a predictive model named the ALT-70 cellulitis score (asymmetry, 3 points; leukocytosis, 1 point; tachycardia, 1 point; age >70 years, 2 points). A score of 0–2 points indicates >83.3% likelihood of pseudocellulitis, and a score of >5 points indicates >82.2% likelihood of cellulitis. • A novel model incorporating the four variables of asymmetry, leukocytosis, tachycardia, and age >70 years is capable of predicting pseudocellulitis and cellulitis with a likelihood of >80%. This model may be useful in the emergency department for early identification and may reduce healthcare costs by precluding hospital admissions for pseudocellulitis. Abstract BACKGROUND Cellulitis has many clinical mimickers (pseudocellulitis), which leads to fre- quent misdiagnosis. OBJECTIVE To create a model for predicting the likelihood of lower extremity cellulitis. METHODS A cross-sectional review was performed of all patients admitted with a diag- nosis of lower extremity cellulitis through the emergency department at a large hospital between 2010 and 2012. Patients discharged with diagnosis of cellulitis were categorized as having cellulitis, while those given an alternative diagnosis were considered to have pseudocellulitis. Bivariate associations between predictor variables and final diagnosis were assessed to develop a 4-variable model. RESULTS In total, 79 (30.5%) of 259 patients were misdiagnosed with lower extremity cellulitis. Of the variables associated with true cellulitis, the 4 in the final model were asymmetry (unilateral involvement), leukocytosis (white blood cell count ≥10,000/uL), tachycardia (heart rate ≥90 bpm), and age ≥70 years. We converted these variables into a points system to create the ALT-70 cellulitis score as follows: Asymmetry (3 points), Leukocytosis (1 point), Tachycardia (1 point), and age ≥70 (2 points). With this score, 0–2 points indicate ≥83.3% likelihood of pseudocellulitis, and ≥5 points indicate ≥82.2% like- lihood of true cellulitis. LIMITATIONS Prospective validation of this model is needed before widespread clinical use. CONCLUSION Asymmetry, leukocytosis, tachycardia, and age ≥70 are predictive of lower extremity cellulitis. This model might facilitate more accurate diagnosis and improve patient care. A predictive model for diagnosis of lower extremity cellulitis: a cross-sectional study. J Am Acad Dermatol 2017 Feb 16;[EPub Ahead of Print], AB Raff, QY Weng, JM Cohen, et al.

initial training in medical school, we have been primed to follow an odyssey to become mas- ter diagnosticians. The goal to “getting it right” is critical because of the myriad repercussions of being wrong. For example, incorrect diagno- ses can lead to unnecessary treatment (one of the concerns in the issue of antimicrobial resist- ance) or a pathway of delayed treatment and potential increased morbidity or mortality. Using the skills of history taking, the physical exam, diagnostic procedures, and laboratory tests, we are at times awed by our acumen and success in getting it right and perhaps humbled by the opposite. In the practice of medicine, we are committed to the ongoing pursuit to improve our diagnostic skills using any variety of new tools. In this endeavour to do our jobs better, we have been blessed with high tech (and oftentimes very costly) advancements and cutting-edge labora- tory techniques. Yet, there is plenty to learn from population-based pattern recognition, which then allows us to establish diagnostic criteria. In this article, the authors delineate the impact of this common infection as well as the impact of misdiagnosis, which is attributed to a lack of “accurate or reliable diagnostic studies.” Using a retrospective, cross-sectional chart review of emergency department patients, comparisons were made between admitting and discharge diagnoses. The end product of the analyses was a point system taking into account asym- metry, leukocytosis, tachycardia, and age ≥70 (ALT-70), with proposed cutoff points and clini- cal interpretation. So, can we add this to our diagnostic tool kit? Well, consider it a work in progress. Two points to consider: 1) the threshold is set at 80% positive or negative predictive value – that is, nothing is per- fect; and 2) further validation is needed before widespread clinical use. Yet this shows that pop- ulation-based pattern recognition can play a part in expanding our diagnostic tools.

Dr Lushniak is Professor and Chair of the Department of Preventive Medicine and Biostatistics and Professor of Dermatology, F. Edward Hébert School of Medicine at the Uniformed Services University of the Health Sciences in Bethesda, Maryland.

PRACTICEUPDATE DERMATOLOGY