Which test(s) can best identify the causative pathogen(s) and result in tailored use of antibiotics?

Objective To determine the extent of agreement and patterns of disagreement between wound swab and tissue samples in patients with an infected diabetic foot ulcer (DFU). Design Multicentre, prospective, cross-sectional study. Setting Primary and secondary care foot ulcer/diabetic outpatient clinics and hospital wards across England. Participants Inclusion criteria: consenting patients aged ≥18 years; diabetes mellitus; suspected infected DFU. Exclusion criteria: clinically inappropriate to take either sample. Interventions Wound swab obtained using Levine’s technique; tissue samples collected using a sterile dermal curette or scalpel. Outcome measures Coprimary: reported presence, and number, of pathogens per sample; prevalence of resistance to antimicrobials among likely pathogens. Secondary: recommended change in antibiotic therapy based on blinded clinical review; adverse events; sampling costs. Results 400 consenting patients (79% male) from 25 centres. Most prevalent reported pathogens were Staphylococcus aureus (43.8%), Streptococcus (16.7%) and other aerobic Gram-positive cocci (70.6%). At least one potential pathogen was reported from 70.1% of wound swab and 86.1% of tissue samples. Pathogen results differed between sampling methods in 58% of patients, with more pathogens and fewer contaminants reported from tissue specimens. The majority of pathogens were reported significantly more frequently in tissue than wound swab samples (P<0.01), with equal disagreement for S. aureus and Pseudomonas aeruginosa. Blinded clinicians more often recommended a change in antibiotic regimen based on tissue compared with wound swab results (increase of 8.9%, 95% CI 2.65% to 15.3%). Ulcer pain and bleeding occurred more often after tissue collection versus wound swabs (pain: 9.3%, 1.3%; bleeding: 6.8%, 1.5%, respectively). Conclusion Reports of tissue samples more frequently identified pathogens, and less frequently identified non-pathogens compared with wound swab samples. Blinded clinicians more often recommended changes in antibiotic therapy based on tissue compared with wound swab specimens. Further research is needed to determine the effect of the additional information provided by tissue samples. Trial registration number ISRCTN52608451.

To undertake a systematic review of the diagnostic performance of clinical examination, sample acquisition and sample analysis in infected foot ulcers in diabetes. Nineteen electronic databases plus other sources were searched. To be included, studies had to fulfil the following criteria: (i) compare a method of clinical assessment, sample collection or sample analysis with a reference standard; (ii) recruit diabetic individuals with foot ulcers; (ii) present 2 x 2 diagnostic data. Studies were critically appraised using a 12-item checklist. Three eligible studies were identified, one each on clinical examination, sample collection and sample analysis. For all three, study groups were heterogeneous with respect to wound type and a small proportion of participants had foot ulcers due to diabetes. No studies identified an optimum reference standard. Other methodological problems included non-blind interpretation of tests and the time lag between index and reference tests. Individual signs or symptoms of infection did not prove to be useful tests when assessed against punch biopsy as the reference standard. The wound swab did not perform well when assessed against tissue biopsy. Semiquantitative analysis of wound swab might be a useful alternative to quantitative analysis. The limitations of these findings and their impact on recommendations from relevant clinical guidelines are discussed. Given the importance of this topic, it is surprising that only three eligible studies were identified. It was not possible to describe the optimal methods of diagnosing infection in diabetic patients with foot ulceration from the evidence identified in this systematic review.