Cellulitis is an infection of the skin and skin-associated structures characterized by redness, warmth, swelling, and pain of the affected area. Cellulitis most commonly occurs in middle-aged and older adults and frequently affects the lower extremities.1 Serious complications of cellulitis such as bacteremia, metastatic infection, and sepsis are rare, and most cases of cellulitis in patients with normal vital signs and mental status can be managed with outpatient treatment.2
Diagnosis of cellulitis can be confounded by a number of similarly presenting conditions collectively known as pseudocellulitis, such as venous stasis dermatitis and deep vein thrombosis.1 Misdiagnosis of cellulitis is common, with rates exceeding 30% among hospitalized patients initially diagnosed with cellulitis.3,4 Dermatology or infectious disease assessment is considered the diagnostic gold standard for cellulitis4,5 but is not always readily available, especially in resource-constrained settings.
Most cases of uncomplicated cellulitis can be managed with outpatient treatment, especially because serious complications are rare. Frequent misdiagnosis leads to repeat or unnecessary hospitalization and antibiosis. Exceptions necessitating hospitalization usually are predicated on signs of systemic infection, severe immunocompromised states, or failure of prior outpatient therapy.6 Such presentations can be distinguished by corresponding notable historical or examination factors, such as vital sign abnormalities suggesting systemic infection or history of malignancy leading to an immunocompromised state.
We sought to evaluate factors leading to the diagnosis of cellulitis in a cohort of patients with uncomplicated presentations receiving dermatology consultation to emphasize findings indicative of cellulitis in the absence of clinical or historical factors suggestive of other conditions necessitating hospitalization, such as systemic infection.
Methods
Study Participants—A prospective cohort study of patients presenting to an emergency department (ED) between October 2012 and January 2017 at an urban academic medical center in Boston, Massachusetts, was conducted with approval of study design and procedures by the relevant institutional review board. Patients older than 18 years were eligible for inclusion if given an initial diagnosis of cellulitis by an ED physician. Patients were excluded if incarcerated, pregnant, or unable to provide informed consent. Other exclusion criteria includedinfections overlying temporary or permanent indwelling hardware, animal or human bites, or sites of recent surgery (within the prior 4 weeks); preceding antibiotic treatment for more than 24 hours; or clinical or radiographic evidence of complications requiring alternative management such as osteomyelitis or abscess. Patients presenting with an elevated heart rate (>100 beats per minute) or body temperature (>100.5 °F [38.1 °C]) also were excluded. Eligible patients were enrolled upon providing written informed consent, and no remuneration was offered for participation.
Dermatology Consultation Intervention—A random subset of enrolled patients received dermatology consultation within 24 hours of presentation. Consultation consisted of a patient interview and physical examination with care recommendations to relevant ED and inpatient teams. Consultations confirmed the presence or absence of cellulitis as the primary outcome and also noted the presence of any pseudocellulitis diagnoses either occurring concomitantly with or mimicking cellulitis as a secondary outcome.
Statistical Analysis—Patient characteristics were analyzed to identify factors independently associated with the diagnosis of cellulitis in cases affecting the lower extremities. Factors were recorded with categorical variables reported as counts and percentages and continuous variables as means and standard deviations. Univariate analyses between categorical variables or discretized continuous variables and cellulitis diagnosis were conducted via Fisher exact test to identify a preliminary set of potential risk factors. Continuous variables were discretized at multiple incremental values with the discretization most significantly associated with cellulitis diagnosis selected as a preliminary risk factor. Multivariate analyses involved using any objective preliminary factor meeting a significance threshold of P<.1 in univariate comparisons in a multivariate logistic regression model for prediction of cellulitis diagnosis with corresponding calculation of odds ratios with confidence intervals and receiver operating characteristic. Factors with confidence intervals that excluded 1 were considered significant independent predictors of cellulitis. Analyses were performed using Python version 3.8 (Python Software Foundation).