This study aimed to identify features associated with length of hospital stay (LOHS), length of intravenous antibiotic therapy (LIVAT) and six-week outcomes for patients with lower limb cellulitis, and to test the Eron/Passos classification of cellulitis in the New Zealand system.

Eighty-five variables were collected prospectively from a cohort of 51 inpatients admitted to Christchurch hospital. The primary end-point for analysis was LOHS. LIVAT and six-week outcomes were secondary end-points.

On univariate analysis use of diuretics, living alone, cellulitis acuity, a creatinine concentration of >0.1mmol/l, poor mobility, pulse >90bpm, age >70 years, oedema extent, chronic oedema, ulceration, neutrophil count >10×10⁹/l, erythema area >1000cm² and haemoglobin concentration less than normal were significantly (P=<0.05) associated with LOHS. A stay of ≤3 days was associated with less oedema, absence of diuretic use and less acute cellulitis. A stay of >7 days was associated with use of diuretics, living alone, age >70 years, more oedema, erythema area >1000cm², haemoglobin less than normal, ulceration, creatinine >0.1mmol/l and poor mobility. The presence of a discharge was associated with LIVAT.

Multivariate analysis accounted for 48% of the variance in LOHS and 16% for LIVAT. Use of diuretics, neutrophil count >10×10⁹/l and oedema score were independently associated with LOHS, with oedema score associated with short stay and diuretic use with long stay.

The Eron/Passos system was not helpful so a new scoring system was devised which successfully classified patients into length of stay groups.

The clinical features analysed accounted for half of the variance in LOHS. An important reason may be physician discretion. If so, our scoring system based on these results could be used in a clinical pathway to improve patient care. This tool would need to be evaluated prospectively.