Swabs versus tissue samples for infected diabetic foot ulcers: the CODIFI2 RCT

Background
Foot ulcers affecting people with diabetes (diabetic foot ulcers [DFUs]) often become infected, potentially leading to amputation. Suspected DFU infection is treated with immediate empiric antimicrobials, with wound samples for culture and sensitivity (C&S) collected to optimise antibiotic therapy. Collecting samples with swabs is easier than obtaining tissue , but report fewer pathogens and more contaminants. Compared with standard C&S laboratory methods, molecular microbiology identifies more organisms. How these differences affect clinical decisions or outcomes is currently unknown.

Objectives
Main study: To determine if taking tissue samples, versus swabs from suspected infected DFU affects ulcer healing, antibiotic prescribing, costs of care, and patient safety.
Sub-study 1 (SS1). To determine the agreement between microbiology results from C&S versus molecular techniques, and assess whether intention of prescribers to change antimicrobials differs based on sampling methods.
Sub-study 2 (SS2). A health-economic perspective of the expected application of empiric and/or targeted treatment regimens and the cost-consequences of treatment decisions based on SS1.
Sub-study 3 (SS3): To compare questionnaire response rates for theoretically-informed versus standard participant letters.
Sub-study 4 (SS4). To explore clinician perspectives on DFU sampling and processing techniques.

Design
Main study: Randomised controlled trial of results of performing tissue sampling versus swabbing of wounds in people with suspected mild or moderate infected DFU. Individually randomised (allocation concealed), 1:1, tissue or swab sampling for suspected DFU infection. Follow-up 12 - 24-months. A priori sample size estimate 730.
SS1: Cross-sectional agreement study of microbiology results from molecular versus C&S techniques and virtual clinic. DFUs sampled for standard microbiology and central laboratory analysis (molecular).
SS2: Exploratory cost-consequence analysis of molecular processing and the likelihood of empiric and targeted treatment based on treatment decisions from SS1.
SS3: Randomised trial of theoretically-informed versus standard participant letters.
SS4: Qualitative study explored clinicians’ perspectives regarding sampling and processing techniques.

Setting and participants
21 UK DFU clinics. Participants with suspected mild or moderate infected DFUs.

Main outcome measures
Main study: Time to ulcer healing (primary outcome blinded assessment), proportion of ulcers healed, antibiotic usage, ulcer area reduction at 4-weeks, hospitalisation duration, time to death, quality of life, and cost-effectiveness.
SS1: Extent of agreement regarding presence or absence of pathogens from standard versus molecular microbiology. Decision to revise antimicrobials based on sampling method.
SS2: Modelled costs (from virtual clinic) of antimicrobial change for standard or molecular analysis.
SS3: Response rate for questionnaires.

Results
The trial was stopped early, after enrolling only 149 participants, due to poor recruitment.
Main study: The hazard ratio for wound healing for patients undergoing tissue versus swab sampling was 1.01 (95% CI 0.65 to 1.55). The swab group had both higher QALYs and lower costs across most time points.
SS1: Agreement between C&S and molecular microbiology was low. A higher proportion of molecular versus C&S, vignettes led to recommendations to change antimicrobials (difference 20.5% (95% CI 8.9 to 31.1%)).
SS2: The modelled costs of molecular processing versus C&S were £120 higher per wound.
SS3: Response rates were 14.8% higher at 52 weeks (95% CI -3.2% to 31.2%) using a theoretically informed versus standard cover letter.
SS4: Clinicians were not confident about replacing C&S with molecular microbiology, as the result reporting was unfamiliar to them.

Limitations
The trial was underpowered..

Conclusions
Trial recruitment was challenging during the COVID pandemic and its aftermath. Whilst the results leave substantial uncertainty regarding differences in healing between the sampling methods, tissue sampling appeared costlier and was associated with lower QALYs than swabbing.