Risk Stratification for Cellulitis Versus Noncellulitic Conditions of the Lower Extremity: A Retrospective Review of the NEW HAvUN Criteria

Materials and Methods

This retrospective chart review was approved by the Yale University (New Haven, Connecticut) institutional review board (HIC#1409014533). Final diagnosis, demographic data, clinical manifestations, and relevant diagnostic laboratory values of 57 patients were obtained from a database in the dermatology department’s consultation log and electronic medical record database (December 2011 to December 2014). The presence of each clinical symptom—acute onset, erythema, pyrexia, history of associated trauma, tenderness, unilaterality, and leukocytosis—was assigned a score equal to 1; values were tallied to achieve a final score for each patient (Table 1). Patients who were seen initially as a consultation for possible cellulitis but given a final diagnosis of stasis dermatitis or lipodermatosclerosis were included (Table 2).

Clinical Criteria
The clinical criteria were developed based largely on clinical experience and relevant secondary literature.^15-17 At the patient encounter, presence of each of the variables (Table 1) was assessed according to the following definitions:

• acute onset: within the prior 72 hours and more indicative of an acute infective process than a gradual and chronic consequence of venous stasis
• erythema: a subjective clinical marker for inflammation that can be associated with cellulitis, though darker, erythematous-appearing discolorations also can be seen in patients with chronic venous hypertension or valvular incompetence^4,15
• pyrexia: body temperature greater than 100.4°F
• history of associated trauma: encompassing mechanical wounds, surgical incisions, burns, and insect bites that correlate closely to the time course of symptomatic development
• tenderness: tenderness to light touch, which may be more common in patients afflicted with cellulitis than in those with venous insufficiency
• unilaterality: a helpful distinguishing feature that points the diagnosis away from a dermatitislike clinical picture, especially because bilateral cellulitis is rare and regarded as a diagnostic pitfall¹⁸
• leukocytosis: white blood cell count greater than 10.0×10⁹/L and is reasonably considered a cardinal metric of inflammatory processes, though it can be confounded by immunocompromise (low count) or steroid use (high count)

Statistical Analysis
Odds ratios (ORs) were calculated and χ² analysis was performed for each presenting symptom using JMP 10.0 analytical software (SAS Institute Inc). Each patient was rated separately by means of the clinical feature–based scoring system for the calculation of a total score. After application of the score to the patient population, receiver operating characteristic curves were constructed to identify the optimal score threshold for discriminating cellulitis from dermatitis in this group. For each clinical feature, P<.05 was considered significant.

Results

Our cohort included 32 male and 25 female patients with a mean age of 63 and 61 years, respectively. The final clinical diagnosis of cellulitis was made in 20 patients (35%). An established diagnosis of cellulitis was assigned based on a dermatology evaluation located within our electronic medical record database (Table 2).

Each clinical parameter was evaluated separately for each patient; combined results are summarized in Table 3. Acute onset (≤3 days) was a clinical characteristic seen in 80% (16/20) of cellulitis cases and 22% (8/37) of noncellulitis cases (OR, 14.5; P<.001). Erythema had similar significance (OR, 10.3; prevalence, 95% [19/20] vs 65% [24/37]; P=.012). Pyrexia possessed an OR of 99.2 for cellulitis and was seen in 85% (17/20) of cellulitis cases and only 5% (2/37) of noncellulitis cases (P<.001).

A history of associated trauma had an OR of 36.0 for cellulitis, with 50% (10/20) and 3% (1/37) prevalence in cellulitis cases and noncellulitis cases, respectively (P<.001). Tenderness, documented in 90% (18/20) of cellulitis cases and 43% (16/37) of noncellulitis cases, had an OR of 11.8 (P<.001).

Unilaterality had 100% (20/20) prevalence in our cellulitis cohort and was the only characteristic within the algorithm that yielded an incalculable OR. Noncellulitis or stasis dermatitis of the lower extremity exhibited a unilateral lesion in 11 cases (30%), of which 1 case resulted from a unilateral tibial fracture. Leukocytosis was seen in 65% (13/20) of cellulitis cases and 8% (3/37) of noncellulitis cases, with an OR for cellulitis of 21.0 (P<.001).

All parameters were significant by χ² analysis (Table 3).