Predicting cellulitis recurrence using baseline patient factors and early clinical response

Abstract

Cellulitis is a common bacterial skin infection that most often affects the lower legs and frequently recurs. Clinicians often prescribe antibiotics for longer than recommended to try to improve outcomes, although it remains unclear whether prolonged therapy reduces recurrence. Instead, individual patient factors, such as comorbidities, illness severity, and early treatment response, may be more important in influencing outcomes. A one-size-fits-all approach to treatment is unlikely to be optimal, and the potential for personalised treatment based on individual patient factors warrants further investigation.

I first conducted an electronic health records (EHR) study of 4,938 UK adults with cellulitis to identify factors predicting recurrence (before death) within 90 days. Multivariable logistic regression with backwards elimination in complete cases was used to develop a points-based risk score. The Baseline Recurrence Risk in Cellulitis (BRRISC) score comprises eight variables (age, heart rate, urea, platelets, albumin, previous cellulitis, venous insufficiency, and liver disease) and ranges from 0-15. Performance, assessed by the C-index, was modest at 0.65 (95% CI: 0.63-0.68). Categorising patients as low (score 0-1), medium (2-5) and high (6-15) risk, recurrence increased fourfold; 3.2% (95% CI: 2.3-4.4%), 9.7% (8.7-10.8%), and 16.6% (13.3-20.4%), respectively.

In practice, antibiotic decisions are often guided by treatment response. Therefore, I conducted a prospective cohort study of 202 adults with lower limb cellulitis to externally validate the BRRISC score and assess whether markers of early clinical response (not available from EHRs), could enhance its predictive value. Score performance was maintained (C-index=0.75 (0.64-0.86)), although there was only weak evidence that an extended score, incorporating day 2/3 severity of skin blistering, improved prediction of ‘hospital-attended recurrence’ to 0.83 (0.75-0.92).

Assessment of skin warmth is one aspect of clinical response that remains subjective and unreliable. To explore more objective approaches, I conducted a methods comparison study within the prospective cohort, focusing on two key technologies increasingly studied for measuring skin temperature in cellulitis. All devices demonstrated significant daily reductions in affected limb temperature, with the largest decrease observed using the thermal imaging camera. However, there was considerable variability in measurement capability among the non-contact infrared thermometers.

Recurrence risk in cellulitis varies markedly according to patient factors captured at baseline. Further research should explore how baseline risk stratification can personalise treatment by tailoring antibiotic duration or targeting non-antibiotic strategies to prevent recurrence. Evaluating whether more advanced thermal or conventional imaging analysis can more accurately capture acute treatment response is also a key area for future investigation.

Date of Award	Oct 2025
Original language	English
Awarding Institution	University of Brighton
Supervisor	Martin Llewelyn (Supervisor), Professor Gail Hayward (Supervisor) & Prof Sarah Walker (Supervisor)