Patterns of agreement between swab sampling and tissue sampling

E Andrea Nelson; Alexandra Wright-Hughes; Sarah Brown; Benjamin A Lipsky; Michael Backhouse; Moninder Bhogal; Mwidimi Ndosi; Catherine Reynolds; Gill Sykes; Christopher Dowson; Michael Edmonds; Peter Vowden; Edward B Jude; Tom Dickie; Jane Nixon

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Nelson EA, Wright-Hughes A, Brown S, et al. Concordance in diabetic foot ulceration: a cross-sectional study of agreement between wound swabbing and tissue sampling in infected ulcers. Southampton (UK): NIHR Journals Library; 2016 Nov. (Health Technology Assessment, No. 20.82.)

Cover of Concordance in diabetic foot ulceration: a cross-sectional study of agreement between wound swabbing and tissue sampling in infected ulcers

Concordance in diabetic foot ulceration: a cross-sectional study of agreement between wound swabbing and tissue sampling in infected ulcers.

Show details

Contents

< Prev Next >

Chapter 2Patterns of agreement between swab sampling and tissue sampling

Introduction

For infected DFUs, the accurate identification of pathogens, rather than colonising bacteria, is a prerequisite for selecting targeted antibiotic therapy to ensure optimal patient outcomes and avoid the acquisition of antibiotic resistance. Currently available evidence from the main diabetic foot infection guidelines (NICE,³⁷ IDSA¹^,²^,³⁶ and the IWGDF⁴²^,⁵⁶) and other studies⁶²^,⁶³ is not sufficiently robust to advise clinicians on the best technique to identify pathogens in DFUs.

Objectives

The primary objective of the COncordance in DIabetic Foot Infection (CODIFI) main agreement study was to assess the level of agreement and patterns of disagreement between culture results from specimens taken by both surface swabs and tissue sampling from DFUs with suspected infection. We were interested in comparing three major microbiological parameters:

reported presence of isolates likely to be pathogens
the number of bacterial pathogens reported
the presence of antimicrobial resistance among likely pathogens.

Secondary objectives of the main agreement study were to compare rates of sampling-related adverse events (AEs) and the costs of sampling using each of the two techniques.

Methods

Study design

A multicentre, cross-sectional study involving 400 patients with a DFU with suspected infection requiring antibiotic therapy was conducted (Figure 1). Consenting patients had both a swab and tissue sample taken from their suspected infected DFU for conventional plating and culture.

FIGURE 1

Study flow diagram. Adapted from Nelson EA, Backhouse MR, Bhogal MS, Wright-Hughes A, Lipsky BA, Nixon J, et al. Concordance in diabetic foot ulcer infection. BMJ Open 2013;3:e002370. Used under Creative Commons Attribution-Non Commercial 2.0 licence. (more...)

Eligibility

All patients at least 18 years of age with a DFU that the attending clinician suspected was either a new case of infection or a chronic infection were screened for enrolment against the eligibility criteria below.

A DFU was considered to be any open wound on the foot (below the malleoli/ankle) in a patient with a diagnosis of diabetes mellitus. Each patient underwent an eligibility screen by a member of the research team, prior to entry, and an anonymised log was used to capture patient demographics along with reasons for not entering the study.

Inclusion criteria

Patient had a diagnosis of diabetes mellitus (type 1 or type 2).
Patient had a suspected foot ulcer infection, with or without bone involvement, based on clinical signs and symptoms using IDSA/IWGDF³⁶^,⁴² criteria and the judgement of the investigator.
The clinical plan was to treat the patient with antibiotics for their infected ulcer.
Patient was at least 18 years of age at the time of signing the consent form.

Exclusion criteria

The clinician deemed it inappropriate to take a tissue sample or a swab sample for any reason.
The patient had already been recruited to the study.

Recruitment and registration

The study was approved by the Sheffield Research Ethics Committee (reference 11/YH/0078) and had central and local NHS permissions at each participating centre prior to data collection.

Patients were recruited from multidisciplinary primary and secondary care-based foot ulcer/diabetic clinics and hospital wards, by a member of the research team (usually a clinical research nurse). Potential patients were provided with a patient information leaflet outlining all aspects of the study and given the chance to read it and to ask any questions they may have about the study. Written informed consent was documented by the patient and member of the local team. Informed written consent was obtained from all patients prior to entering the study.

Patients were registered via a 24-hour automated telephone registration system that automatically sent confirmation of successful registration through to the site.

Assessments

Sample acquisition

Clinicians in the participating sites participated in a study information session to instruct them on techniques for swab and tissue sample acquisition. An e-learning package was also developed and issued to all sites, detailing study procedures, including video footage of the correct use of both sampling techniques.

After wound cleansing (using sterile saline and gauze) and debridement (removal of necrotic tissue, foreign material, callus, undermining, usually with sharp instruments), a physician or podiatrist obtained specimens from the wound for cultures in one of the following ways:

Rubbing a sterile, cotton-tipped swab over the wound surface to sample superficial wound fluid and tissue debris. The swab was pressed with sufficient pressure on the wound bed to capture expressed wound fluid and was positioned deep in the ulcer to collect from likely infected areas. This is the wound swabbing technique described by Levine et al.⁴⁸
Immediately after the cotton swab had been collected, a tissue sample was removed from the same area of the ulcer bed. This procedure was performed using sterile equipment and aseptic technique, involved removal of a small piece of wound tissue at the base of the wound by scraping or scooping using a dermal curette or sterile scalpel blade.

Sample transport and processing

Each sample was placed individually in the standard transport medium used at the site and delivered to the local medical microbiology laboratory in accordance with routine clinical practice. A UK national standard method was used for collecting and processing samples.⁶⁵^,⁶⁶ Both samples from each patient were processed in the same local laboratory as routine clinical samples. Neither sample was labelled as having been taken as part of a clinical study. Our goal was to ensure that, as far as possible, the reports reflected current sample processing methods in each laboratory, rather than these samples having received special attention or processing.

Clinical assessments

A member of the research team used a case report form (see Appendix 5) to record patient demographics, diabetes status and foot health history, including current or proposed antibiotic treatment and wound dressings. Details of the index ulcer were recorded using each of the Perfusion, Extent/Size, Depth/Tissue loss, Infection, Sensation (PEDIS),⁶⁷ Wagner grade,⁶⁸ and Clinical Signs and Symptoms Classification for Infection⁶⁹ schemes. The research team also filled out other study-related documentation, which was forwarded to the study co-ordinating centre at the University of Leeds.

Centre differences questionnaire

A ‘centre differences questionnaire’ aimed to capture, from each centre and laboratory processing samples, details relating to the clinical acquisition of samples, specimen transport, sample analysis, methods of reporting results of samples by the laboratory and local antibiotic protocols for infected DFUs.

End points

Coprimary end points

In order to assess agreement and patterns of disagreement between results from the swab and tissue samples, three coprimary end points were defined.

Reported presence of likely pathogens

The first coprimary end point was originally defined as the reported presence or not of the following likely pathogens, identified by the UK Health Protection Agency (HPA) as likely pathogens from limb-threatening DFUs:⁶⁵^,⁶⁶

Staphylococcus aureus Rosenbach 1884 (categorised by the presence or absence of meticillin resistance)
Streptococcus species Rosenbach 1884
Enterobacter aerogenes Hormaeche and Edwards 1960
Escherichia coli (Migula 1895) Castellani and Chalmers 1919
Pseudomonas species Migula 1984
Corynebacterium species Lehmann and Neumann 1896
anaerobic cocci (i.e. mixed anaerobes)
Fusobacterium species Knorr 1922
Bacteroides fragilis (Veillon and Zuber 1898) Castellani and Chalmers 1919
Prevotella bivia (Holdeman and Johnson 1977) Shah and Collins 1990.

A revised definition was implemented to include the most prevalent pathogens, defined as those reported in ≥ 10% of patients (in either swab or tissue samples). This overall prevalence rate was determined based on statistical justification of the sample size calculation; we also used clinical discretion to determine whether or not the end point would include pathogens with an overall prevalence below 10%.

An overall summary of pathogens reported⁵⁹ allowed for the comparison of all pathogens reported within each sample and an assessment of whether or not agreement was influenced by any of a number of covariates.

Antimicrobial resistance

Presence or absence of resistance to antibiotics to which the specific species is ordinarily susceptible among likely pathogens, as reported by standard techniques for:

meticillin-resistant S. aureus (MRSA)
meticillin-resistant coagulase-negative staphylococci (CNS)
vancomycin-resistant Enterococcus species.

Number of pathogens

Number of pathogens reported per specimen.

Secondary end points

Adverse events

The secondary end point relating to AEs was the number of patients with a study-related event categorised as an expected AE, defined as bleeding of concern attributable to the sampling method or patient-reported pain before and after each sampling technique, or as a related unexpected serious adverse event (RUSAE).

Costs

A full economic evaluation was beyond the scope of this study. The cost data collected were the laboratory costs, including all components used in processing and reporting of swab and tissue samples. Costs of these procedures were requested from the microbiologists at study centres.

Derivation involving reported pathogens

Microbiology laboratories reported pathogens at a range of taxonomic levels (species, genus, family and group); therefore, the end points for the prevalence, overall summary and number of pathogens required derivation in order to allow for comparison of pathogens reported within each sample at a meaningful level.

Staphylococcus aureus is used in reference to non-MRSA, whereas MRSA is used to describe S. aureus that is meticillin resistant.

The majority of pathogens were included at the genus level, with the exception of S. aureus (identified at the species level) and vancomycin-resistant and non-resistant Enterococcus spp. (included separately by vancomycin resistance). The following groups of pathogens were also included as part of the first coprimary end point for reported presence or absence: Gram-positive cocci, Gram-negative cocci, Gram-positive bacilli, Gram-negative bacilli, anaerobes (where possible as anaerobic cocci or anaerobic rods), CNS and Enterobactereaceae (including coliforms).

Furthermore, the following isolate designations were considered unlikely to represent pathogenic organisms in a sample from a DFU and were not included in the end points: yeasts, skin flora, normal flora, mixed flora, skin organisms, bacterial flora, enteric flora and faecal flora.

Statistical methods

Sample size

The sample size calculation was based on the primary outcome of reported ‘presence or absence of a pathogen’ for the whole sample overall.

To be confident that swabs adequately sampled wound flora, it was assumed that the chance corrected agreement between swab and tissue samples needed to be at least ‘good’ (usually defined as a κ-statistic > 0.6).⁷⁰ Of course, the κ-statistic alone does not convey the distribution of disagreement between swabs and tissue samples. Good overall agreement, with balanced disagreement, would be clinically important if tests were to be regarded as interchangeable. Therefore, the total sample size was based on there being good agreement and reasonably balanced disagreement for clinically important and less prevalent pathogens.

Using a two-sided McNemar’s test at the 5% level of significance, a sample size of 399 patients would provide 80% power to detect a difference of ≥ 3% in the reported presence of a pathogen, assuming an overall prevalence of the pathogen of 10% and 5% disagreement between the swab and tissue samples. This amount of agreement would also result in a κ-statistic of ≈ 0.7, and the calculation was based on the expected prevalence of less common pathogens, such as Pseudomonas (present in 10% of samples in Pellizer et al.⁵⁸). It was, therefore, planned that a total of 400 patients would be recruited. Further details of the sample size calculation are provided in Appendix 1.

Analysis methods

All data analyses and summaries were performed using SAS^® version 9.2 (SAS Institute Inc., Cary, NC, USA),⁷¹ with the exception of exact confidence intervals (CIs) only, which were calculated within R version 3.0 (The R Foundation for Statistical Computing, Vienna, Austria).⁷² All significance tests were two-sided and conducted at the 5% level of significance, with p-values and 95% CI provided where appropriate.

Patient populations

The full analysis set consisted of all patients registered and consented to take part in the study, regardless of their adherence to the study protocol or eligibility violation.

The evaluable population consisted of all registered and consented patients with evaluable swab and tissue samples. Patients for whom the swab or tissue samples were not successfully collected or were lost, or for whom the sample results were lost, were excluded from this evaluable population.

The per-protocol (PP) population consisted of all registered and consented patients for whom there were no protocol violations. Patients who did not satisfy the eligibility criteria, or those for whom a protocol deviation in the collection or processing of either sample had occurred, were excluded from the PP population.

Coprimary end point analysis

Reported presence of likely pathogens

The first coprimary end point was defined for each patient as the reported presence or absence of each likely pathogen reported from the result of culture of the swab and tissue sample. Patients for whom either the swab or tissue sample result was not available were excluded from the primary end point, with analysis conducted on the evaluable population.

For each likely pathogen, cross-tabulations of reported presence were generated to investigate agreement and the pattern of disagreement. For each pathogen, the following statistics are presented:

overall percentage prevalence, calculated as the proportion of patients with the pathogen reported from either the swab or tissue sample results
the percentage of patients for whom the swab and tissue sample results agreed or disagreed in the reported presence of the pathogen
the unadjusted κ-statistic (with 95% CIs) and the prevalence- and bias-adjusted kappa (PABAK), which evaluates chance corrected agreement between swab and tissue samples. Strength of agreement according to the κ-statistic was categorised as shown in Table 1
the difference in percentage prevalence between the swab and tissue samples (tissue percentage minus swab percentage) with 95% CIs (accounting for paired samples)
McNemar’s test of the difference between the swab and tissue samples in the percentage of samples reporting the pathogen accounting for paired samples.

TABLE 1

The κ-statistic

When the proportion of disagreement was low, leading to cell counts in the cross-tabulations of < 5, a small-sample binomial version of McNemar’s test was used to provide an exact p-value⁷³ and the exact 95% CI for the difference in percentage prevalence was calculated using an inductive method.⁷⁴

Cross-tabulations on the semiquantitative extent of bacterial growth (none, + to +++) and weighted κ-statistics by type of DFU (neuropathic, ischaemic, without inferential statistics) were also generated.

An overall summary of the pathogens reported was further generated,⁵⁹ with each patient’s pair of results (swab and tissue sample) coded as follows: swab and tissue sample report the same pathogens; swab reports same pathogens as tissue sample plus extra pathogens; tissue sample reports same pathogens as swab plus extra pathogens; both tissue sample and swab report different pathogens (with or without overlap in pathogen). Multinomial logistic regression analysis was used to model the proportion of patients in each category, compared with the reference category ‘swab and tissue sample report the same pathogens’ on pre-specified baseline factors to investigate their relevance in determining agreement between sample results. These factors included type of ulcer (ischaemic or neuroischaemic vs. neuropathic); Wagner ulcer grade (1 to 5); recent (on the day of sampling) systemic or topical antimicrobial therapy; or dressing and wound duration (< 56 days vs. ≥ 56 days, and continuous on the log-scale). The centre from which the patient was enrolled was included in each model as a random effect in order to allow for additional variability in outcome by centre and estimates of the effect of baseline factors without directly requiring the estimation of individual centre effects. Estimates of odds ratios for each covariate are presented along with 95% CIs and p-values (based on the change in –2 log-likelihood).

Reported presence of antimicrobial resistance among likely pathogens

Meticillin-resistant S. aureus, meticillin-resistant CNS and vancomycin-resistant Enterococcus were the three antimicrobial-resistant pathogens identified for exploration. For each of these resistant pathogens, cross-tabulations were created (reported presence or absence of resistant pathogen) and the following statistics presented: PABAK, unadjusted κ-statistic and overall percentage agreement. McNemar’s test was used to test for a difference between swab and tissue sampling techniques in the proportion of samples in which the specified resistant pathogen was reported.

For each resistant pathogen the following codes were also created: resistant pathogen reported by swab not tissue sample; resistant pathogen reported by tissue sample not swab; swab and tissue sample results agree. To determine if agreement is influenced by specified covariates, multinomial regression modelling was planned to model these categories on type of ulcer (predominantly neuropathic or ischaemic), ulcer grade, pre-sampling antibiotic therapy, pre-sampling antimicrobial dressing, wound duration and centre.

Number of pathogens reported

Summaries (including cross-tabulations) on the number of pathogens reported per specimen were generated for swab versus tissue samples. Samples were further coded as follows: tissue sample had two or more extra pathogens reported; tissue sample had one extra pathogen reported; tissue sample and swab had the same number of pathogens reported; swab had one extra pathogen reported; swab had two or more extra pathogens reported.

Ordinal logistic regression analysis, based on the proportional odds model,⁷⁵ was used to model the number of pathogens reported per specimen on pre-specified baseline factors to investigate their relevance in determining agreement between sample results. These factors included type of ulcer (ischaemic or neuroischaemic vs. neuropathic); Wagner ulcer grade (1–5); recent systemic or topical antimicrobial therapy or dressing; and wound duration (< 56 days vs. ≥ 56 days, and continuous on the log-scale). Centre was included in each model as a random effect in order to allow for additional variability in outcome by centre and estimates of the effect of baseline factors without directly requiring the estimation of individual centre effects. Estimates of odds ratios for each covariate are presented along with 95% CIs and p-values (based on the change in –2 log-likelihood).

Analysis population

Patients for whom both swab and tissue sample results were available were included in the coprimary end points, with analysis conducted on the evaluable population.

Missing data

As part of the study design, efforts were made to collect complete data; however, where data remained missing, this was assumed to be missing at random, and multiple imputation (MI)⁷⁶ was used to impute missing baseline covariates, thereby allowing inclusion of the 28 (7.1%) patients with missing data for at least one candidate baseline factor. The pattern and prevalence of missing data among covariates considered within the regression analysis of the coprimary end points are presented in Appendix 1, Table 81.

The outcome and all baseline covariates (including type of ulcer, Wagner ulcer grade, recent systemic or topical antimicrobial therapy or dressing, wound duration) to be considered in each regression analyses were included in the MI models alongside centre. A total of 10 imputations were conducted using the Markov chain Monte Carlo (MCMC) method⁷⁷ with multiple chains, initial values from the expectation–maximisation (EM) algorithm, 200 burn-in iterations, and the assumption of normality for factors with missing data (thus, imputations were made on a continuous scale).⁷¹ For dichotomous factors, imputations were not restricted for ‘implausible values’ and thus continuous imputations were rounded to plausible values for the dichotomous factor (with a small proportion of missing data the bias introduced as a result of this method is minimal).⁷⁸ This method was used as the pattern of missing data was arbitrary and non-monotone.

For the 10 imputed data sets, the odds ratios generated through the regression analyses were combined using Rubin’s rules;⁷⁹ therefore, reported estimates reflect the average of estimates across the imputed data sets, and estimated standard errors include variability across the imputed sets as well as the usual uncertainty in parameter estimates. The mean change in –2 log-likelihood was used to calculate the overall p-value.

Derivation

Semi-quantitative extent of bacterial growth

A number of common scales were used to quantify the extent of growth of a pathogen, specific to each recruiting site. In order of severity of growth within a scale, these were: +/++/+++; +/++/+++/++++; scanty/light/moderate/heavy; scanty/+/++/+++; light, moderate, heavy. The reported growth for each pathogen was derived onto one 3-point scale reported as +/++/+++ (Table 2).

TABLE 2

Derivation of extent of growth

Weighted κ-statistic for cross-tabulations on the extent of growth

κ-statistic weights were selected to reflect the ordinal nature of extent of growth, in which the difference between a sample with an extent of growth of + and ++ is far smaller than the difference between ++ and +++, owing to the increase in dilution factors used to determine the extent of growth (10-fold increase). To account for this relationship, while allowing greater differentiation between the highest level of growth (+++), the following exponential values were assigned to each level of growth, from which linear Cicchetti–Allison agreement weights were derived:⁸⁰ pathogen not reported (= 1), + (= 2.7), ++ (= 7.4), +++ (= 20.1).

As the choice of values for each level of growth was somewhat arbitrary, a sensitivity analysis assessed the impact of these weights, in which levels of growth were assigned the following linear values: pathogen not reported – 0, + – 1, ++ – 2, +++ – 3.

Summary of pathogens

To account for pathogens reported at various taxonomic ranks and to determine whether or not swab and tissue results reported the same pathogens, pathogens were compared according to pre-defined groups set out in Appendix 1 [i.e. largely at the genus level, and at the higher group level where further detail was not reported from the laboratory result (e.g. Gram-positive cocci rather than S. aureus)]. For example, where a pathogen was reported at the species level it was compared with the corresponding alternative sample at the genus level (e.g. E. coli belongs to the Escherichia genus). If, however, one sample reported the pathogen at a taxonomic rank higher than the genus level, such as ‘Gram-negative bacilli’ with the corresponding alternative sample reporting the pathogen in more detail (in this scenario ‘Escherichia’), then we did not class the patient’s results as reporting the same pathogens. This was based on clinical relevance of pathogens and overcame discrepancies in the level of reporting.

Summary and number of pathogens

The summary and number of pathogens reported per specimen was calculated independently for both the swab and tissue samples.

Samples were identified where more than one strain or species of pathogen (in which we were interested in the genus level or higher) was reported. In these samples, a single pathogen at the level of interest was retained for comparison with the corresponding swab or tissue sample in the summary of pathogens and with the count of the number of pathogens within the sample.

Samples from which results of a Gram-stained smear had been reported in addition to those from the culture were identified by the reporting of the following groups of pathogens: Gram-positive bacilli, Gram-negative bacilli, Gram-positive cocci, Gram-positive cocco-bacillus and Gram-negative cocci. Gram stain results were then compared with pathogens reported within the corresponding culture result, and pathogens belonging to the group of pathogens reported by the Gram stain were further identified. Where a pathogen belonged to the same group as that reported by the Gram stain, it was considered likely that both referred to the same pathogen and the corresponding Gram stain result was excluded from the summary and number of pathogens reported from the swab or tissue sample. For example, where both Gram-positive cocci and S. aureus were reported, because S. aureus is a type of Gram-positive cocci, only S. aureus was retained in the summary and number of pathogens. Conversely, where the results of a Gram stain were provided and no pathogens identified by the culture belonged to the group identified by the Gram stain, all pathogens were included. Details of all samples for which this derivation was applied are in Appendix 1.

Secondary end point analysis

Adverse events

Safety analyses presents summaries of all expected AEs (bleeding of concern that is attributable to either of the sampling methods, pain as reported by the patient before and immediately after acquisition of each sample) and RUSAEs. The number of events and number of patients with events are also summarised.

Results

Sample size

In total, 680 patients were screened for recruitment into CODIFI and 401 patients were enrolled between November 2011 and May 2013. One patient was excluded as the written informed consent was lost. Of 27 centres, 25 recruited patients into the study; Figure 2 shows the number of patients recruited and screened per centre, and Figure 3 shows the overall, monthly and cumulative recruitment of patients to the study together with our original target. See Appendix 1 for full centre names.

FIGURE 2

Screening and recruitment by centre. SJUH, St James’s University Hospital.

FIGURE 3

Monthly and cumulative recruitment.

Analysis populations

The numbers of patients recruited to CODIFI and included in the full analysis set, the evaluable population and the PP population are each summarised in Table 3.

TABLE 3

Number of patients in each analysis population

Full analysis set

No patient withdrew consent for the samples to be used for research purposes and hence only the one patient without informed consent was excluded from the full analysis set.

Evaluable population

The evaluable population consisted of the 395 (98.8%) patients who had both swab and tissue sample results available. Patients with a protocol deviation involving the loss of one or both samples or results were excluded. The number of patients excluded from the evaluable population and the reasons are summarised in Table 4.

TABLE 4

Reasons for exclusion from the evaluable and PP populations

Per-protocol population

The PP population consisted of the 386 (96.5%) patients without an eligibility violation or protocol deviation. The number of patients excluded from the PP population and the reasons are summarised in Table 4. Given that only an additional nine patients were excluded from the PP population compared with the evaluable population, no analyses were repeated for the PP population.

Study conduct

Figure 4 presents a flow diagram depicting the study conduct and analysis population.

FIGURE 4

Patient flow diagram.

Baseline characteristics

Tables 5–13 summarise the baseline characteristics, including patient demographics, information about diabetes, clinical details, ulcer characteristics, PEDIS classification, clinical signs and symptoms, and antibiotic regimens immediately pre and post sampling. Because the evaluable population was very similar to the full analysis set with respect to baseline characteristics, characteristics of the full sample only are detailed below.

TABLE 5

Baseline patient demographics and type of recruiting site

TABLE 13

Summary of patients’ baseline antibiotic regimen (pre sampling) and proposed new antibiotic regimen (immediately post sampling)

TABLE 10

Clinical signs and symptoms classification

Tables 5 and 6 summarise patient demographics and diabetes details, respectively. The median age of patients was 63 years (range 26–99 years); 79% of patients were male; and the majority of patients (94.3%) were of white ethnic origin. Recruitment of patients was from outpatient clinics for 79.8% of patients, hospital wards for 13.3% and community clinics for 7%. The median duration of diabetes in enrolled patients was 15 years (range 2 weeks–57 years); 14.5% and 85.5% of patients had type 1 or type 2 diabetes, respectively; and the vast majority of patients (96.3%) were receiving treatment for their diabetes.

TABLE 6

Baseline diabetes details

Tables 7 and 8 summarise patients’ ulcer characteristics. The total number of DFUs ranged from one to seven per patient, with one ulcer observed for 67.0% of patients, two ulcers for 19.5%, and three or more ulcers for 13.6% of patients. The anatomic site of the index ulcer, from which both the swab and tissue samples were obtained, was most commonly the plantar surface (43.0%), the digital surface (22.5%), the dorsum (14.0%) or the apex (i.e. tip of toe, 11.8%). The duration of the index ulcer varied to a large degree, with a median of 1.84 months (range 3 days–12 years). A total of 72.0% of patients’ index ulcers were incident as opposed to recurrent. Only 3.5% of ulcers were solely ischaemic, 50.5% of ulcers were neuropathic only, and 45.5% of ulcers were both ischaemic and neuropathic.

TABLE 7

Ulcer characteristics

TABLE 8

Index ulcer characteristics

Tables 9–11 summarise ulcer characterisation according to the PEDIS criteria, clinical signs and symptoms, and Wagner scale. Almost all patients (98%) had a grade 1 or 2 perfusion rating (no critical limb ischaemia); approximately equal proportions of patients had a grade 1–3 depth/tissue loss rating; the majority of patients (93.3%) had grade 2 sensation (loss of protective sensation); and the majority of patients had an infection of either grade 2 (inflammation of skin/subcutaneous tissue only, 37.3%) or grade 3 (extensive erythema deeper than skin/subcutaneous tissue, 59.3%). The majority of patients had an ulcer debridement undertaken at the baseline visit (87.8%), with the median area measuring 1.77 cm² (range 0.01–138.2 cm²). The clinical signs and symptoms classification of patients’ index ulcers revealed that 31.8% of patients had a foul wound odour; 42.5% had pocketing in the wound; 56.3% had discoloured granulation tissue; 51.0% had friable granulation tissue; 31.3% had a recent increase in pain, as opposed to the 2.3% who had a recent decrease in pain; 61.5% had a recent increase in wound size; and 31.5% had a breakdown of epithelium. Furthermore, of all index ulcers, 34.0% were classified as grade 1 (superficial diabetic ulcer); 33.5% were classified as grade 2 (ulcer extension to ligament, tendon, joint capsule or deep fascia without abscess or osteomyelitis); 30.5% were classified as grade 3 (deep ulcer with abscess, osteomyelitis or joint sepsis); 1.8% were classified as grade 4 (gangrene localised to a portion of forefoot or heel); and 0.3% were classified as grade 5 (extensive gangrenous involvement of the entire foot).

TABLE 9

Perfusion, Extent/Size, Depth/Tissue loss, Infection, Sensation classification and ulcer debridement

TABLE 11

Wagner grade

Tables 12 and 13 and Figure 5 summarise the antibiotic regimens patients were prescribed immediately pre and post sampling. Prior to sampling, 60.3% of patients had been treated with an antimicrobial dressing or agent on the infected ulcer. Furthermore, 46.8% of patients were on a systemic antibiotic regimen, with the most frequently prescribed antibiotics being flucloxacillin (31.1%), clindamycin (18.3%), co-amoxiclav (13.1%), ciprofloxacin (13.1%) and metronidazole (7.2%). The patient’s antibiotic regimen was changed following clinical assessment and specimen sampling, but before microbiology results were available, in 62.0% of patients. Among the 42.0% of patients who were not on an antibiotic regimen prior to sampling, treatment was initiated immediately post sampling. Finally, 6.5% of patients were not on an antibiotic regimen prior to sampling and did not have an antibiotic regimen initiated immediately post sampling.

TABLE 12

Antibiotic therapy: pre and post sampling

FIGURE 5

Prescribed antibiotics pre and post sampling (antibiotics are not mutually exclusive).

Microbiology results

Microbiology reports of culture results for swab and tissue samples produced a total of 79 different microbial isolates from the 395 evaluable patients.

Table 14 presents the number of patients with at least one pathogen reported. At least one pathogenic isolate was reported from swab results in 277 (70.1%) patients and from tissue results in 340 (86.1%) patients. On swab sample results, only isolates not likely to be pathogenic (defined as mixed skin flora, normal flora, enteric flora, yeast, faecal flora) were reported for 39 (9.9%) patients, and no isolates were reported at all for 79 (20.0%) patients. Based on tissue results, only isolates not likely to be pathogenic were reported for 15 (3.8%) patients and no isolates were reported at all for 40 (10.1%) patients.

TABLE 14

Summary of the reporting of pathogens from results of patients’ swab and tissue samples

Table 15 presents the pathogens most frequently reported, following their grouping at a range of taxonomic levels. The most frequently reported groups of pathogens from at least one of the patient’s swab or tissue sample were Gram-positive cocci (70.6%), Gram-negative bacilli (36.7%), Enterobacteriaceae including coliforms (26.6%), anaerobes (23.8%) and Gram-positive bacilli (11.1%). The most frequently reported genus- and species-level pathogens were S. aureus (35.7%), Streptococcus (16.7%), Enterococcus (14.9%), CNS (12.2%), Corynebacterium (9.4%), Pseudomonas (8.6%) and MRSA (8.1%). The prevalence of additional genus- and species-level pathogens were all < 6%.

TABLE 15

Overall prevalence of grouped pathogens