Characteristics Associated With Diagnostic Error in Emergency Department Patients With Potential Cellulitis

Prior studies have reported 19-34% of cellulitis cases in the ED are overdiagnosed. These diagnostic errors are a threat to patient safety and unnecessary hospital admissions. Our objective was to identify factors associated with diagnostic errors in ED patients with potential cellulitis.

We conducted a secondary analysis of prospective data from patients presenting to the ED with dermatologic lower extremity complaints with visible erythema. Diagnostic consensus regarding cellulitis was determined by six independent, multidisciplinary physicians. The treating ED physician’s diagnosis in the patient’s chart was compared to the consensus panel diagnosis to identify errors. We used four multivariable logistic regression models to evaluate factors associated with diagnostic error: 1) patient characteristics alone (age, sex, race, area deprivation index); 2) patient characteristics plus provider-documented signs and symptoms (area of concern, laterality, redness, swelling, tenderness, pain, warmth, and drainage); 3) patient characteristics plus clinical features (fever, maximum heart rate, maximum temperature of complaint via thermal imaging camera and disposition); and 4) patient characteristics plus past medical history (diabetes, end stage renal disease, active cancer, neutropenia, peripheral arterial disease, prior cellulitis diagnosis, lymphedema and immunosuppressive medications).

This analysis included 192 patients of which 50(26%) had diagnostic errors and of those 48(96%) were overdiagnosed with cellulitis by the treating physician. Models evaluating patient characteristics alone, clinical features, and past medical history found no significant associations with diagnostic error. In the patient characteristics plus provider-documented signs and symptoms model, an acute dermatologic manifestation overlying a joint (foot, ankle, or knee) (OR: 3.88, 95%CI 1.82-8.46, p<0.001); visible redness (OR: 4.60, 95%CI 1.67-15.26, p=0.006); or pain (OR: 3.70 95%CI 1.32-10.58, p=0.013) were all significantly associated with higher odds of diagnostic error.

We found skin conditions overlying joints and those involving visible redness and pain had increased odds of overdiagnosis of cellulitis. This work has the potential to support the targeted development of tools to improve diagnostic accuracy in patients with potential cellulitis.