Can point-of-care tests assist antibiotic prescribing decisions for diabetic foot ulcer infections? (INDUCE study)
In 2011-12, the financial cost of diabetic foot ulcers (DFUs) and amputation in NHS England was £650 million, approaching 1% of its budget. Diagnosis of DFU infection remains entirely clinical and antibiotics are often prescribed empirically, risking increased antibiotic resistance. Conversely, missing early bacterial infection increases the risk of serious infection and amputation.Inflammatory biomarkers may aid DFU infection diagnosis by clinicians working in the community, if results are available at point of care.
Our aim was to develop a composite DFU infection assay, investigating three inflammatory biomarkers: venous C-reactive protein (CRP), venous procalcitonin and calprotectin levels in wound exudate, the latter being resistant to wound protease degradation. We tested both venous and point-of-care CRP and developed a novel calprotectin assay for wound exudate, based on the faecal test already available. We also developed and sought feedback on an online educational tool to assist in clinical assessment of DFUs.Our inclusion criterion was the presence of a DFU with either mild or no infection. Exclusion criteria were immunosuppression or receipt of antibiotics within the previous two weeks. Antibiotics were prescribed according to the judgement of the clinician at initial assessment and participants were reviewed one week later. Our gold standard for the definition of DFU infection was the clinician's overall impression based on their reviews at week 0 and week 1, incorporating the response to antibiotics, if given, while being blinded to all test results.
From September 2014 to September 2015, 67 of 363 potential participants were recruited from community podiatry clinics in two UK regions. Following early withdrawal of one participant from each group, 28 (43%) had mildly infected DFUs and 37 (57%) were non-infected, based on the week 1 gold standard assessment.Median baseline CRP was slightly higher in the infected group, 7.50 mg/ml compared to 6.00 mg/ml, however the area under the receiver operating characteristic curve (AUROC) was 0.52, demonstrating that predictive efficacy was no better than by chance. Most of the venous procalcitonin levels were below the assay’s lower limit and the median level was slightly lower for infected ulcers. Median calprotectin levels were higher in infected ulcers, 1437ng/ml compared with 879ng/ml, but with an insufficient AUROC of 0.56.
The sensitivity and specificity of wound exudate calprotectin levels in isolation are insufficient for calprotectin to be used as a biomarker of DFU infection, however the assay could be combined with other biomarkers to produce a future composite assay. Based on our results, venous CRP and procalcitonin levels should not be pursued as point-of-care tests for DFU infection, either alone or combined with other inflammatory markers.The INDUCE study has developed and received detailed feedback on a DFU educational tool, which will be made freely available online.