The Natural History of Antibiotic-Treated Limb Cellulitis: Data Derived from a Randomised Trial of Cellulitis of over 400 Patients Allows the Determination of Time to Recovery and Clarifies the Utility of Systemic Observations, Blood Tests, Pain Scores and Local Limb Measurements.

Background:Despite being a common condition the natural history of cellulitis is poorly described. This makes it difficult to determine recovery at different time points and to give the patient a likely duration of the condition. We have used data collected as part of a clinical trial on the treatment of limb cellulitis to describe the day-by-day changes in physiology, pain, blood parameters and limb measurements. We have used these variables to estimate severity, and relate this to time to recovery.

Methods: At entry into the trial patients had routine observations, limb measurements and blood tests. Pain (VAS) and well-being scores were also recorded. The patients were followed up with 2 subsequent face-face visits and a telephone questionnaire up to 40 days after entering the study. Patients kept a record of pain and other features.

Results: 410 patients from 20 centers, entered the study. Because the actual follow-up days varied from patient to patient, data was available from recruitment for every day up until 15 days for 20 variables and up until 40 days for well-being and physical activity. Plots were constructed for each variable both from the onset of local features and from the start of antibiotic therapy. In general, limb swelling increased over the first few days before starting to decrease at 6 days, erythema was still evident at 10 days and pain decreased uniformly. Baseline severity and age were correlated with the likelihood of prolonged disability.

Baseline values and back to normal activities (recovery) at Day 5 follow up.

1. Limb: Arm: 42/84 (51%); Leg: 108/251 (43%); p=0.20
2. Affected skin area: ≤4%: 101/183 (55%); >4%: 49/150 (33%); p<0.001
3. CRP: <37mg/L: 82/159 (52%); >37mg/L: 59/163 (36%); p=0.007
4. Temperature: Afebrile: 94/107 (53%), Febrile: 56/156 (36%); p=0.002

Low pain scores, normal neutrophil counts and normal renal function were significantly associated with recovery. Early antibiotic therapy was not significantly associated with recovery at Day 5.

Conclusion: Local features of cellulitis may progress or remain static for the first few days despite other parameters improving and should not be used to influence antibiotic therapy. The extent of cellulitis at diagnosis is the feature most strongly predictive of time to recovery.