Risk Factors Predicting Cellulitis Diagnosis in a Prospective Cohort Undergoing Dermatology Consultation in the Emergency Department

Results

Of 1359 patients screened for eligibility, 104 patients with presumed lower extremity cellulitis undergoing dermatology consultation were included in this study (Figure). The mean patient age (SD) was 60.4 (19.2) years, and 63.5% of patients were male. In the study population, 63 (60.6%) patients received a final diagnosis of cellulitis. The most common pseudocellulitis diagnosis identified was venous stasis dermatitis, which occurred in 12 (11.5%) patients with concomitant cellulitis and in 12 (11.5%) patients mimicking cellulitis (Table).

Patient selection flowchart. Patient screening and selection methodology for final study cohort (n=104).

Univariate comparisons revealed a diverse set of historical, examination, and laboratory factors associated with cellulitis diagnosis. Diagnosis of cellulitis was associated with unilateral presentation, recent trauma to the affected site, and history of cellulitis or onychomycosis. Diagnosis of cellulitis also was associated with elevated white blood cell count, absolute neutrophil count, C-reactive protein, body mass index, hematocrit, and platelet count; age less than 75 years; and lower serum sodium and serum chloride levels. These were the independent factors included in the multivariate analysis, which consisted of a logistic regression model for prediction of cellulitis (eTable).

Multivariate logistic regression on all preliminary factors significantly associated with cellulitis diagnosis in univariate comparisons demonstrated leukocytosis, which was defined as having a white blood cell count exceeding 11,000/μL, unilateral presentation, history of onychomycosis, and trauma to the affected site as significant independent predictors of cellulitis diagnosis; history of cellulitis approached significance (eTable). Unilateral presentation and leukocytosis were the strongest predictors; having either of these factors had a sensitivity of 93.7% and a negative predictive value of 76.5%.

Comment

Importance of Identifying Pseudocellulitis—Successful diagnosis of cellulitis can be confounded by pseudocellulitis that can present concomitantly with or in lieu of cellulitis itself. Although cellulitis mostly affects the lower extremities in adults, pseudocellulitis also was common in this study population of patients with suspected lower extremity cellulitis, occurring both as a mimicker and concomitantly with cellulitis with substantial frequency. Notably, among patients with both venous stasis dermatitis and cellulitis diagnosed, most patients (n=10/12; 83.3%) had unilateral presentations of cellulitis as evidenced by signs and symptoms more notably affecting one lower extremity than the other. These findings suggest that certain pseudocellulitis diagnoses may predispose patients to cellulitis by disrupting the skin barrier, leading to bacterial infiltration; however, these pseudocellulitis diagnoses typically affect both lower extremities equally,¹ and asymmetric involvement suggests the presence of overlying cellulitis. Furthermore, the most common pseudocellulitis entities found, such as venous stasis dermatitis, hematoma, and eczema, do not benefit from antibiotic treatment and require alternative therapy.¹ Successful discrimination of these pseudocellulitis entities is critical to bolster proper antibiotic stewardship and discourage unnecessary hospitalization.

Independent Predictors of Cellulitis—Unilateral presentation and leukocytosis each emerged as strong independent predictors of cellulitis diagnosis in this study. Having either of these factors furthermore demonstrated high sensitivity and negative predictive value for cellulitis diagnosis. Other notable risk factors were history of onychomycosis, cellulitis, and trauma to the affected site. Prior studies have identified similar historical factors as predisposing patients to cellulitis.^7-9 Interestingly, warmth of the affected area on physical examination emerged as strongly associated with cellulitis but was not included in the final predictive model because of its subjective determination. These factors may be especially important in diagnosing cellulitis in patients without concerning vital signs and with concomitant or prior pseudocellulitis.

Study Limitations—This study was limited to patients with uncomplicated presentations to emphasize discrimination of factors associated with cellulitis in the absence of suggestive signs of infection, such as vital sign abnormalities. Signs such as fever and tachypnea have been previously correlated to outpatient treatment failure and necessity for hospitalization.^10-12 This study instead focused on patients without concerning vital signs to reduce confounding by such factors in more severe presentations that heighten suspicion for infection and increase likelihood of additional treatment measures. For such patients, suggestive historical factors, such as those discovered in this study, should be considered instead. Interestingly, increased age did not emerge as a significant predictor in this population in contrast to other predictive models that included patients with vital sign abnormalities. Notably, older patients tend to have more variable vital signs, especially in response to physiologic stressors such as infection.¹³ As such, age may serve as a proxy for vital sign abnormalities to some degree in such predictive models, leading to heightened suspicion for infection in older patients. This study demonstrated that in the absence of concerning vital signs, historical rather than demographic factors are more predictive of cellulitis.

Conclusion

Unilateral presentation and leukocytosis emerged as strong independent predictors of lower extremity cellulitis in patients with uncomplicated presentations. Having either of these factors had a sensitivity of 93.7% and a negative predictive value of 76.5%. Other factors such as history of cellulitis, onychomycosis, and recent trauma to the affected site emerged as additional predictors. These historical, examination, and laboratory characteristics may be especially useful for successful diagnosis of cellulitis in varied practice settings, including outpatient clinics and EDs.