Interventions in the management of diabetes‐related foot infections: A systematic review

The optimal approaches to managing diabetic foot infections remain a challenge for clinicians. Despite an exponential rise in publications investigating different treatment strategies, the various agents studied generally produce comparable results, and high‐quality data are scarce. In this systematic review, we searched the medical literature using the PubMed and Embase databases for published studies on the treatment of diabetic foot infections from 30 June 2018 to 30 June 2022. We combined this search with our previous literature search of a systematic review performed in 2020, in which the infection committee of the International Working Group on the Diabetic Foot searched the literature until June 2018. We defined the context of the literature by formulating clinical questions of interest, then developing structured clinical questions (Patients‐Intervention‐Control‐Outcomes) to address these. We only included data from controlled studies of an intervention to prevent or cure a diabetic foot infection. Two independent reviewers selected articles for inclusion and then assessed their relevant outcomes and methodological quality. Our literature search identified a total of 5,418 articles, of which we selected 32 for full‐text review. Overall, the newly available studies we identified since 2018 do not significantly modify the body of the 2020 statements for the interventions in the management of diabetes‐related foot infections. The recent data confirm that outcomes in patients treated with the different antibiotic regimens for both skin and soft tissue infection and osteomyelitis of the diabetes‐related foot are broadly equivalent across studies, with a few exceptions (tigecycline not non‐inferior to ertapenem [±vancomycin]). The newly available data suggest that antibiotic therapy following surgical debridement for moderate or severe infections could be reduced to 10 days and to 3 weeks for osteomyelitis following surgical debridement of bone. Similar outcomes were reported in studies comparing primarily surgical and predominantly antibiotic treatment strategies in selected patients with diabetic foot osteomyelitis. There is insufficient high‐quality evidence to assess the effect of various recent adjunctive therapies, such as cold plasma for infected foot ulcers and bioactive glass for osteomyelitis. Our updated systematic review confirms a trend to a better quality of the most recent trials and the need for further well‐designed trials to produce higher quality evidence to underpin our recommendations.


| INTRODUCTION
Diabetes-related foot infections (DFIs) are associated with considerable morbidity, a worsened quality of life, and a marked increase in the risk of lower extremity amputation. 1,2Because appropriate treatment will very likely improve the outcome of these infections, we have reviewed the available evidence to help establish evidencebased criteria for selecting treatment.The present report updates and, by consolidating the results of previous and current literature searches, replaces the International Working Group on the Diabetic Foot (IWGDF) systematic review of the treatment of DFI conducted in 2019 and published in 2020. 3The review focuses on studies of all types of therapeutic interventions that could help inform the working group on developing recommendations for the IWGDF guideline on diagnosis and treatment of DFI.This review does not focus on definitions of infection or on methods for diagnosis; for our review on the accuracy of diagnostic procedures in DFIs, we refer to our parallel publication on this topic. 4

| METHODS
We performed the literature search for this systematic review for the period from 30 June 2018 to 30 June 2022 on the basis of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 5On 10 May 2022, we prospectively registered the systematic review in the PROSPERO 2022 database for systematic reviews, which assigned it the number CRD42022324812 (https://www.crd.york.ac.uk/PROSPEROFILES/324795_STRATEGY_ 20220426.pdf).
We began by defining the population (patients) of interest (P), interventions (I) performedand outcomes (O) assessed that we would attempt to address.The IWGDF editorial board and 10 external experts (not members of the infection working group) from various geographical regions worldwide then reviewed these questions and population intervention control outcomes (PICOs) for their clinical relevance.Using their input, we revised the PICOs to their final form for this review.Some PICOs from the 2019 systematic review only underwent small textural changes to improve clarity.One PICO was split into two different PICOs for readability and clarity.These changes did not have consequences for the identified publications before the 2019 search.With our oversight, a medical librarian performed electronic database searches using the databases of MEDLINE (PubMed), EMBASE, and Scopus, using a combination of MeSH and keyword terms.
We included studies on persons aged 18 years or older, with diabetes mellitus of any type who have an infection of the foot (diagnosed by any clinical, laboratory, or imaging methods) that involves skin, soft tissue, bone, or other structures, caused by any microorganism.We reviewed any study of an intervention (e.g.antibiotic, antiseptic, surgery and adjunctive therapy) to prevent or cure the infection in a person with a diabetes-related complication of the foot after our previous literature search.In addition to the subjects who received a specific intervention, all included studies had to have a contemporaneously studied set of subjects who received a control intervention.The control intervention could be a placebo, a sham device or sham procedure, a type of intervention or medicine different from the index intervention, no therapy, or usual clinical care.We only included outcomes that were relevant to an infectious aspect of the foot.These could include clinical cure of infection, requirement for lower extremity amputation, occurrence of a new infection, death, hospitalisation, resolution of a foot ulcer, eradication of microbial pathogens, quality of life, adverse effects, or cost of treatment.The infection working group agreed that acceptable study designs could include meta-analyses, systematic reviews, randomized controlled trials (RCTs), non-randomised comparative studies, casecontrol studies, and prospective and retrospective cohort studies.
We excluded papers that were conducted on non-human subjects, review articles, case series without a contemporaneous control group, studies in which the reported data on the evaluation of the diabetic population was not individualised, and studies that included fewer than 15 patients with diabetes.We used the same search string as the one we employed in 2018 with the addition of new terms in relation to new antibiotics and other therapeutic interventions. 3 The search criteria were only augmented with specific antimicrobials and techniques that became available after 2018.The search string was designed to identify all prospective and retrospective studies, in any language, that evaluated interventions for the treatment of DFIs in the given population and that were published after our previous search, that is, between 30 June 2018 and 30 June 2022.We also searched the ClinicalTrials.gov(https://clinicaltrials. gov) and the World Health Organization-International Clinical Trials Registry Platform trial registries (http://apps.who.int/trialsearch/default.aspx)for studies that appeared to meet our criteria.For studies of potential interest, we made an attempt to contact the designated investigator for outcome results, but we included no study identified by this process.To test the search terms we intended to employ, we first created a set of 20 key publications that we knew should be in the scope of the systematic review that had to be identified in the literature search.Our search terms identified all 20 publications.
After conducting the literature search, we divided the papers retrieved and assigned one-sixth of the papers to one of six infection working group teams of two members each.These working group members, working independently, reviewed their assigned publications by title and abstract to determine eligibility on the basis of the presence of the criteria listed above (appropriate population, study design, outcome(s) measurement, and intervention(s)), using the COVIDENCE online software (https://app.covidence.org).After the two members of each team reached a consensus on which papers met the criteria, they obtained and independently reviewed the full paper of all potentially eligible publications using the same key criteria to determine final eligibility for inclusion. 6Any disagreements between assessors were discussed until a consensus was reached, with a third assessor being involved if needed.All included full-text publications were assessed for risk of bias with forms of the Dutch Cochrane Centre by two independent assessors.The SIGN level of evidence was determined for each publication (https://www.sign.ac.uk/assets/ study_design.pdf)and combined with the risk of bias score. 7pending on the number of questions answered with 'yes' on the 10 items of the Cochrane scoring sheet, risk of bias for each study was very low when scoring ≥8/10, low when scoring 6-7/10 or high when scoring ≤5/10.After appropriate data were extracted from each included paper, they were summarised in a standardised evidence table that included study design, risk of bias, setting, follow-up, study population and characteristics, the variable or condition assessed, the study intervention and the control intervention, results of analyses, and an open field for comments.Through both electronic communications and an in-person meeting, each member of the working group reviewed and discussed the content of the evidence tables.Working group member(s) did not participate in the selection or the discussion of a paper if they were (co)-authors of that paper.
In the Results Section, risk of bias assessment and evidence tables are shown of studies found in the updated search.For those details on earlier studies, we refer to our previous systematic review. 6In the description of the results and in the evidence statements, we used the information from studies identified in both the previous and this updated search.

| Evidence statements
Based on the strength of the available evidence, we formulated evidence statements with the accompanying assessment of the quality of the evidence, according to the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) methodology. 8e authors rated the certainty of the evidence for each formulated evidence statement as 'high', 'moderate', 'low', or 'very low' in regard to the strength of confidence in estimates of the effect an intervention on patient-important outcomes.GRADE defines 'high' as 'We are very confident that the true effect lies close to that of the estimate of the effect'; 'moderate' as 'We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different'; 'low' as 'Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect', and 'very low' as 'We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect'.The rating was determined based on the level of evidence, risk of bias, (in)consistency of results, (im)precision, (in) directness, publication bias, effect size and evidence of doseresponse relation.Each evidence statement was phrased in accordance with the methods described by GRADE.All authors discussed these evidence statements until consensus was reached.After each evidence statement, the related literature is discussed.

| RESULTS
The PRISMA flowchart with the study selection process is shown in Figure 1.The risk of bias assessment of each paper can be found in Table 1.The full evidence table of the papers published after the 2019 search can be found in Appendix S1.
We report on the following topics:

| PICO 1
In a person with diabetes and a soft tissue infection of the foot, is any particular antibiotic regimen (specific agent[s], route of administration, duration) better than any other regarding the resolution of infection, recurrence of infection, and the acquisition of antimicrobial resistance?

Summary of the literature
A recent systematic review 30 reached the same conclusions as our 2019 systematic review 3 regarding the absence of any strong evidence to recommend a specific antibiotic with the highest efficacy.
With some exceptions, 41-43 many of the previously described studies on soft-tissue infection (STI) in the foot of persons with diabetes were characterised by suboptimal trial design and reporting.Only two studies did report a difference in outcome between the two tested antibiotics.One was a study that suggested the inferiority of tigecycline to ertapenem with or without vancomycin soft-tissue diabetes-related foot infections (ST-DFIs) 42 and the other a study that suggested the inferiority of ertapenem to piperacillintazobactam in a subset of persons with a severe ST-DFIs. 43e high-quality study compared the results of therapy with tigecycline (alone) and ertapenem (with or without the addition of vancomycin) in hospitalised subjects with an acute DFI of any severity. 42The study assessed individuals with diabetic foot infection without osteomyelitis (primary study) and with osteomyelitis (sub study).The tigecycline regimen did not meet the primary study endpoint of non-inferiority to the ertapenem � vancomycin regimen.
The percentage of adverse events (primarily nausea and insomnia) was significantly higher in the tigecycline-treated group.The other high-quality study was a non-inferiority, multi-centre trial of ertapenem versus piperacillin/tazobactam (with or without the addition of vancomycin in either group) in subjects with moderate or severe soft tissue DFIs without osteomyelitis. 43The outcomes suggest that ertapenem was clinically non-inferior to piperacillin/tazobactam in patients with moderate or severe DFIs.In a subset analysis, subjects with a severe DFI treated with ertapenem had a significantly lower clinical resolution rate, compared with subjects treated with piperacillin/tazobactam (91.5% vs. 97.2%[119/130 vs. 139/143], p = 0.04).
Because the study was not powered to detect statistical differences between study treatments in the severe DFI stratum, it is hard to draw solid conclusions from this observation.There were no significant differences in adverse events in the ertapenem group compared with the piperacillin/tazobactam group.In another non-inferiority trial comparing ertapenem to piperacillin/tazobactam in subjects with DFIs (SIDESTEP) published in 2005, the authors found no statistically significant differences in outcomes between the treatment arms. 43,44llowing our 2019 review, we identified 3 additional, recent papers that compared the efficacy of different systemic antibiotics for the treatment of ST-DFIs. 14,21,33In a series of 794 initial episodes of DFIs in 419 patients, including 339 episodes of DFO, all patients were treated with surgical debridement. 14Beta-lactam antibiotics were used in 631 episodes (79%) including oral amoxicillinclavulanate at a daily dose ranging from 2 to 3 g for a median of 20 days (interquartile range, 12-30 days) in 301 episodes.After a median follow-up of 3.3 years, amoxicillin-clavulanate and non-betalactam antibiotics (mostly fluoroquinolones, vancomycin, and clindamycin) resulted in comparable remission (respectively 74% and 79%: p = 0.15).A total of 61 methicillin-resistant Staphylococcus spp.
were identified, which presumably were treated with non-betalactam antibiotics.The routine use of amoxicillin-clavulanate in patients treated for DFIs is, however, limited by the increasing prevalence of methicillin-resistant staphylococci and resistant gramnegative rods in many countries in the world.Importantly, these data do not apply to patients treated medically (i.e., without any surgical intervention) for DFI or DFO.A small randomised-controlled trial with high risks of bias compared linezolid (600 mg bid) and ampicillin-sulbactam (1.5-3 g q6h iv), and oral amoxicillin-clavulanate (500-875 mg every 8-12 h) as empirical treatment of infected diabetic foot ulcers. 21The duration of treatment ranged from 7 to prospective study from Turkey, which included 284 patients hospitalised for a moderate or severe DFI, a multivariate analysis was performed to identify the risk factors of reinfection, major amputation, and death during a 6-month follow-up. 33A major amputation was necessary in 12.7% of the population.Four independent risk factors for major amputation were identified, including vascular insufficiency (p = 0.004), hospital readmission (p = 0.009), Creactive protein (CRP) >130 mg/L (p = 0.007), and carbapenems use (p = 0.005).However, there are doubts about the statistical analyses performed in this study.In the multivariate models, there was inadequate adjustment for the severity of the infection and carbapenems might have been used in more severe or nonresponding cases.This limitation makes it difficult to draw reliable conclusions. 33We added 'acquisition of antimicrobial resistance' in the 2023 PICO, a term that was not included in the 2019 PICO.We did not find outcome data for antimicrobial resistance in the extracted data of the papers identified in the 2019 systematic review.
T A B L E 1 Risk of bias assessment of 32 included studies in the 2023 systematic review.

Evidence statement
There were no differences in all potential clinical outcomes among antibiotics compared in studies of soft-tissue-DFIs (except in a study advocating the tigecycline inferiority to ertapenem with or without vancomycin).

Certainty of evidence
Moderate, based on numerous studies including recent RCTs and older randomised and non-randomised studies.

Summary of the literature
4][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63][64][65] Two more recent single-centre retrospective studies 13,17 did not show any link between antibiotic duration and microbiological failure or clinical recurrences of DFI.The first one collated 1018 moderate or severe DFI episodes (392 episodes of DFO including those with residual bone infection after amputation), of which 313 cases involved revascularisation. 13Surgical debridement and amputation were performed in respectively 824 episodes (81%), and 596 (59%) cases.The median total duration of antibiotic therapy was 20 days and the median follow-up was 3 years.On multivariate analysis, the duration of antibiotic therapy did not affect the risk of recurrence (HR 1.0, 95% CI 0.99-1.01).Stratifying the analysis according to the type of DFI (ST vs. DFO) did not modify the results.The second single-centre retrospective study included for a 20-year period (2000-2020) 721 episodes of DFI treated with systemic antibiotics and surgery. 17In a prospective randomised-controlled trial, involving patients with moderate or severe ST-DFIs treated with surgical debridement, 35 received 10 days of antibiotic therapy, and 31 other patients, 20 days.This pilot study reported preliminary data from a larger RCT that is underway.Similar rates of cure and antibiotic-related adverse events were recorded in both groups. 29Limitations of this study were that it was underpowered, that it had a large non-inferiority margin of 25%, and that the proportion of participants with a moderate versus those with a severe ST-DFIs was not reported.In all studies investigating the duration of total antibiotic therapy, 'time zero' always started with the (intraoperative) debridement.So far, the influence of presurgical antibiotic duration on the ultimate outcomes after surgery has not been evaluated.The limited data from the pilot study suggested that antibiotic therapy for DFIs following surgical debridement for 20 and 10 days leads to comparable outcomes. 29duction of the duration of antibiotic treatment from 14, as recommended in 2019, to 10 days in surgically treated patients could therefore provide an opportunity to reduce the exposure of patients to potential antibiotic-related adverse effects.

Evidence statement
No conclusive data could be identified to determine the optimal duration of systemic antibiotic therapy in relation with the outcome of soft-tissue DFIs, while limited data advocate that a 10-day duration may be enough for moderate or severe DFIs treated with surgical debridement.

Certainty of evidence
Low, based on retrospective studies and one small recent RCT with a risk of bias.

| PICO 2
In a person with diabetes and a bone and/or joint infection of the foot, is any particular antibiotic regimen (specific agent[s], route of administration, total and parenteral duration) better than any other regarding the resolution and recurrence of infection.

Summary of evidence
[49][50][51][63][64][65][66][67] Seven of these RCTs compared the use of a beta-lactam/beta-lactamase inhibitor combination antibiotic in DFO against one of the following agents: imipenem/cilastatin 46,47 ; cefoxitin 48 ; ofloxacin 49 ; linezolid 50 ; ertapenem 44 ; or moxifloxacin. 51Results of each of these studies reported no significant differences in outcomes among the different antibiotic regimens with the exception of two studies. 42,48The first of these was a sub study of 118 participants with osteomyelitis in the large RCT reported a higher cure rate patients treated with ertapenem � vancomycin compared with those treated with of tigecycline (discussed above in the skin and soft tissue infection section). 42In the other single centre, double-blind study, 36 subjects were treated with either cefoxitin or ampicillin/sulbactam. 48 Subjects in the cefoxitin group had a significantly higher rate of 'cure' than subjects in the ampicillin/sulbactam group, but outcome 'cure or improvement' was not statistically different.
The quality of most, but not all (see Appendix S1), of these studies was generally high.In one non-inferiority RCT, randomised subjects with a variety of severe osseous and joint infections, including approximately 20% subjects with diabetic foot osteomyelitis, to treatment with an oral versus a parenteral antibiotic regimen. 68,69ere were no significant differences in treatment outcomes between the two routes of therapy for the various types of infections combined.
Unfortunately, the authors did not provide separate outcomes for subjects with DFO, which makes the results of the study hard to apply to the general population of persons with DFO.
We identified four new studies 15,16,18,33 since the systematic review of 2020, 3 and another systematic review mentioned earlier. 30ether oral antibiotic therapy is superior to intravenous antibiotic therapy for residual osteomyelitis following amputation for DFIs was addressed in two recent studies. 14,16A retrospective study included a total of 65 evaluable patients without providing data on a power calculation of the necessary population size. 16Failure was defined as the need for a revision surgery (including debridement and irrigation or proximal amputation), and/or a persistent nonhealing wound with draining sinus tract at the surgical site within 12 months after the initial amputation or remission.An RCT compared the efficacy of 6-week versus 12-week antibiotic therapy in the nonsurgical treatment of diabetic foot osteomyelitis. 18 The study by Saltoglu and co-workers mentioned earlier, also included patients with DFO.The smaller number of this subgroup limits the possibility to draw reliable conclusions in patients with DFO. 33idence statement There were no differences in all potential clinical outcomes among antibiotics compared in studies of DFO (except in a study advocating the tigecycline inferiority to ertapenem with or without vancomycin).

Certainty of evidence
Moderate, based on numerous studies including RCTs and older randomised and non-randomised studies.

Summary of the literature
In the 2020 systematic review, the most important study covering this topic was an RCT that compared 6 weeks versus 12 weeks of antibiotics for patients with a DFO treated with antibiotics, but not with surgery. 3,65In this study, the outcomes were comparable between the two treatment arms. 65In the literature search that encompassed studies published between 2018 and 2022, we identified two additional studies.One was a retrospective multicentre study from Turkey in which the cyclic lipopeptide antibiotic daptomycin and the glycopeptide antibiotic teicoplanin were compared and that included 16 patients (8 in each group) diagnosed with DFO on the basis of imaging and/or bone biopsy results. 35 An RCT mentioned earlier compared the efficacy of 6-week versus 12-week antibiotic therapy in the nonsurgical treatment of diabetic foot osteomyelitis with different antibiotics. 18No significant difference in clinical outcomes between the two groups was recorded.The conclusions of the study are limited by the small number of patients (n = 30), the use of different antibiotic regimens in the two groups of patients, a short follow-up period, and the lack of information about bone microbiology.This study had the same conclusions as a previous RCT on the same topic. 65idence statement A 6-week duration of systemic antibiotic therapy seems appropriate for DFOs treated medically.

Certainty of evidence
Moderate, based on one recent RCT and one recent retrospective, combined with older (non-randomised) studies, all with a high risk of bias.

Summary of the literature
Compared with our 2020 review, there were more recent studies that specifically evaluated the duration of antibiotic treatment in combination with surgical debridement of infected bone tissue in patients with a DFO.We identified two studies that addressed the duration of systemic antibiotic therapy for DFO treated surgically. 15,32 prevents from assessing whether failures were related to residual DFO or not.CRP monitoring does not seem to be of any help in determining the outcomes of the infection and thus the optimal duration of the antibiotic treatment of DFI including DFO. 28 Evidence statement A 3-week period of systemic antibiotic therapy seems appropriate for DFO treated with surgical debridement of infected bone tissues; duration and mode of administration (oral vs. intravenous) of the antibiotic therapy following amputation for a DFI with or without residual DFO does not seem to influence the outcome.

| PICO 3
In a person with diabetes and osteomyelitis of the foot, are there circumstances in which non-surgical (antibiotic only) treatment is as safe and effective (in achieving remission) as surgical treatment (combined with antibiotic therapy)?

Summary of the literature
Our previous systematic review 3 identified two low-quality studies that suggested that early surgical debridement reduced the likelihood of a major amputation. 70,71Our search identified two additional single-centre, retrospective studies that investigated the effect of treatment with 'early' surgery, including toe, foot and leg amputations (variously defined, but usually within 72 h of presentation) versus delayed surgery (i.e., 3-6 days after admission) in hospitalised patients with a severe, deep DFI, with or without osteomyelitis. 24,40 the first study, an above knee amputation was required in 2 out of 26 patients who initially underwent conservative treatment and deferred a below the knee amputation, with a median delay of 75.5 days (23-87) after the initial recommendation to undergo a below knee amputation recommendation, and in 2 out of 18 who proceeded directly to below the knee amputation. 40The median duration of the antibiotic therapy was significantly higher in the group of patients with deferred amputation (55 days [42-78] vs. 17 days [10-37]; p = 0.0017).During the follow-up of about 4 years, the hazard ratio for death was 12.2 (95% CI, 1.58-94.8) in patients in the deferred group compared with the other patients.The second study evaluated outcomes in 668 patients with moderate or severe DFI after dividing their time to specialised treatment into quartiles. 24Patients with DFO were not clearly described or evaluated separately.The delay in referral was quite long between groups (Q1 <9 days, Q2 9-21 days, Q3 21-59 days, Q4 >59 days).Patients with the longest delays were significantly more likely to require a major amputation.
Both studies are limited by a high risk of bias, especially including a lack of randomisation of the subjects and a lack of standardised protocols for diagnosis and for surgical (or medical and antimicrobial) treatment.Although the effect of early surgical intervention seems large, the low quality of the studies considerably reduces the quality of evidence of the evidence statement.

Evidence statement
Early surgery in patients hospitalised for severe DFIs requiring surgical intervention (e.g. to drain an abscess) appears to reduce the likelihood of a major lower extremity amputation.

Certainty of evidence
Low, based on two older and two more recent, single-centre, retrospective studies.

Summary of the literature
The conclusion of the 2020 systematic review was that treatment with a primarily surgical or primarily non-surgical (antibiotic) approach in selected patients with forefoot DFO without peripheral artery diseases and without exposed bone or abscesses, yields similar outcomes.This was primarily based on one RCT that did not show a difference in outcome between patients treated with surgical intervention and antibiotics, compared with those treated without surgery but with antibiotics. 63In our current systematic review, we identified 3 additional studies that compared surgical versus non-surgical treatment of DFO. 12,22,38In addition, we found a systematic review that also identified these studies. 37A prospective cohort study from Spain evaluated a cohort of 116 people admitted to hospital for DFO with a 1 year follow-up. 38The majority of subjects required surgery and antibiotics (82.8%), the others were treated with antibiotics, but without surgery.Patients with extensive soft tissue infection and critical limb ischaemia were excluded from the study.The authors did not report the proportion of wounds that healed in each group.There were no differences in the time to heal or complications in the two groups.
A retrospective cohort study from the US reported the results of 90 people with moderate and severe DFIs. 22The study compared subjects who required surgery with bone resection and antibiotics and local debridement without bone resection and antibiotics.There was a significantly higher proportion of subjects with DFO in the surgical treatment group (79%) compared with the antibiotic treatment group (55%; p = 0.01).Despite this, the subjects in the surgery group were 1.8 times more likely to heal (69% vs. 38%) than those in the antibiotic treatment group.A limitation of the study was that the authors included mixed group of DFO and STI patients.In another retrospective cohort, 60 people from Israel with DFO of the toe were evaluated for at least 2 years. 12The outcomes of subjects undergoing toe amputation for DFO were compared with those not undergoing amputation.Subjects in the non-amputation group had a significantly higher rate of re-hospitalisation for re-infection.This resulted in 33% of subjects ultimately requiring toe amputation.There was no difference in the length of hospitalisation for the index infection or leg amputations during the follow-up period between study groups.
Limitations of this study were dominated by the small number of patients, and the lack of standardisation for confusing factors such as perfusion status, type and duration of the antibiotic therapy.
Overall, compared with a previous RCT 63 that showed similar outcomes between surgical and non-surgical approaches of DFO, non-randomised studies favour better outcomes and fewer complications when surgery is performed.Given the inherent lower chance of bias of RCTs, we value the evidence of the RCT higher than the non-randomised studies and therefore we rated the certainity of evidence as moderate.

Evidence statement
Surgical and conservative (non-surgical) approaches for the treatment of DFO lead to comparable outcomes.

Certainty of evidence
Moderate, on the basis of one recent prospective cohort and two recent retrospective studies, all with risk of bias, one systematic review, and older non-randomised and one older randomised study.

| PICO 4
In a person with diabetes and a foot infection, does the addition of any specific adjunctive/topical treatment to systemic antibiotic therapy and surgery improve the resolution rate of infection and time to reach this resolution, recurrence of infection, need for surgical intervention including amputation and duration of the antibiotic therapy?

Summary of the literature
The 2020 review identified five RCTs [72][73][74][75][76][77] and one systematic review 78 on the treatment of DFI with topical antimicrobial agents.A 2017 Cochrane systematic review concluded that there was little difference in the rate of treatment-related adverse events with topical versus systemic antibiotic therapy. 78One RCT compared the results of treatment with a topical application of the antimicrobial peptide pexiganan versus treatment with an oral antibiotic (ofloxacin). 75The data demonstrated equivalent results in rates of clinical improvement, microbiological eradication and wound healing, while the incidence of adverse events was higher in the ofloxacin group.
Despite these promising results, a recent large unpublished (except for a summary in ClinicalTrials.gov)study of topical therapy for a mild DFI with pexiganan found that it was not superior to placebo (standard of care treatment alone) 76,77 and no further, recent, studies were identified.[74] Pathogen eradication was high in one study in mild infections. 73fortunately, this single-blinded, single-centre study was underpowered. 73In another study, authors found non-significant differences in clinical cure, and pathogen eradication. 72In a study of moderate DFIs, the clinical cure rate for subjects in the gentamicincollagen sponge group was worse than subjects in the control group at treatment day 7, but significantly better 2 weeks after discontinuing treatment. 74The study was marred by a modification of the inclusion criteria (to enhance enrolment) during the study, failure to reach the recruitment target, and a high withdrawal rate, making it is difficult to interpret the reported findings.All subjects in the studies tolerated the gentamicin-collagen sponge well.
0][81] Drawing conclusions from these studies was severely limited by their weak trial designs, incomplete reporting and possible sources of bias and we did not identify additional studies.27]31 One RCT of low quality compared the efficacy of systemic antibiotics (intravenous ciprofloxacin 200 mg bid) either with or without gentamicin cream for the treatment of DFIs in 140 patients (70 each group) who underwent surgical intervention for infected or necrotic tissue removal. 26Patients with DFO, ischaemic foot, severe immune suppressions, patients already using gentamicin, alcohol or substance abusers were excluded from the study.Clinical cure was defined as the absence of clinical signs of infection, and negative culture swab 7 days after the intervention.Significantly better outcomes (reduction in inflammation, culture negativity, clinical cure rate and microbiological eradication) were recorded in the gentamicin group versus the control group (respectively, p = 0.03; p = 0.001; p = 0.02 and p = 0.03).The other RCT included a small smallsize population of patients (n = 36) treated with or without antibiotic (vancomycin, cefoperazone, or gentamicin)-loaded bone cement for the primary treatment of diabetic neuropathic foot ulcers complicated by osteomyelitis. 27The results of this low-quality study suggest that antibiotic-loaded bone cement versus no bone cement is associated with lower postoperative pain, shorter hospital length of stay, reduced hospital cost, and fewer dressing renewals.A small case-control study compared the outcomes of bone resection alone or in combination with adjuvant antibiotic-impregnated calcium sulphate for the treatment of diabetic foot osteomyelitis in 46 patients who underwent surgical bone resection. 31 was used in some cases and not in others.A small controlled beforeafter study explored if the intraoperative addition of a local vancomycin powder in a series of 38 patients with ulcerated calcaneal osteomyelitis reduced the incidence of surgical revisions and if it was associated with future colonisation by multi-resistant pathogens. 9ter an average follow-up of 26.1 � 15.0 months, the intervention failed to demonstrate any clinical benefit.Based on the data from 13 studies and a total of 798 patients, a systematic review assessed the role of local antibiotic delivery systems (gentamicin-impregnated sponges, poly methyl methacrylate, calcium sulphate) as an adjunct to surgery in DFIs. 25 Re-operation rates, major amputation, and mortality were similar in patients treated with either antibiotic-eluting devices or standard treatments.
Our new search identified one RCT that examined the use of sodium hypochlorite (also known as Dakin's) solution for irrigation of diabetic foot ulcer infections versus standard rising with normal saline. 19A total of 90 evaluable hospitalised patients for DFI (45 in each group) were randomly treated with diluted hypochlorite solution (30 min of irrigation each other day) or normal saline for irrigation of diabetic foot ulcer infections in addition to the antibiotic therapy with imipenem/cilastatin or piperacilline/tazobactam.This single-centre trial conducted in Jordan suggests that replacing normal saline irrigation with 0.1% sodium hypochlorite followed by soaking the ulcer with 0.08% sodium hypochlorite (and pursued after hospital discharge) significantly improved ulcer healing and decreased the number of amputations (2.2% vs. 28.9%),hospitalisations and mortality (2.2% vs. 13.3%).Infection resolution rate was higher in the intervention group (35.6% vs. 22.2%) but not statistically significant (p = 0.17).The infection severity is not described in either group of participants, neither are the compliance with the saline irrigation or the reasons of the amputations.The single centre design in one geographical location might further diminish the applicability in other centres.Although the effect size is large, these limitations make it very difficult to rely on the conclusions about the superiority of the hypochlorite versus saline irrigations in persons with infected DFIs.Overall, the effects of local antibiotic or antiseptic treatments on the outcomes of DFIs/DFOs are based on low certainty of evidence and the effect size seems small in the majority of studies with the exception of one study to hypochlorite. 19

Evidence statement
There is insufficient evidence that local antibiotic or antiseptic treatment improves outcomes in diabetic foot infections, except possibly sodium hypochlorite.There is very low-quality evidence that hypochlorite irrigation benefits ulcer healing, amputations and rehospitalisation.

Certainty of evidence
Low, based on retrospective studies and RCTs, all of low quality.

Evidence statement
In a patient with diabetes and foot infection due to Pseudomonas species, there is currently no evidence of improved wound healing or other clinical benefits from acidifying agents.

Certainty of the evidence
Very low, based on one retrospective cohort study with high risk of bias.

Summary of the literature
In the previous systematic review, we identified one cohort study that addressed the question of whether or not using a percutaneous bone biopsy and an antibiotic regimen containing rifampicin for gram-positive organisms would help improve outcomes in primarily non-surgical management of DFO. 66It is very possible that the better outcomes in the group that was treated based on bone biopsy and with rifampicin were the result of confounding variables, especially the fact that patients in one of the highest enrolling centres only received a rifampicin-containing regimen if they underwent a bone culture.In the recent literature search, one large size multicentre retrospective cohort study from the US compared the clinical outcomes of patients treated for DFO with antibiotic regimens that included or did not include rifampicin. 39The study included 6174 rifampicin.The logistic regression confirmed the significant association between rifampicin use and the studied events (0.65; 95% CI, 0.43-0.96;p = 0.04).These outcomes need to be confirmed in an RCT currently ongoing in the US before we can advise on the use of rifampicin in patients with DFO. 86The evidence statement on the benefit of rifampicin treatment in persons with DFO is only based on studies with very high risk of (confounding) bias.

Evidence statement
In a patient with diabetic foot osteomyelitis, available data suggest that rifampicin combination antimicrobial therapy leads to fewer major amputations and lower mortality.

Certainty of evidence
Very low, based on one recent and one older large size multicentre retrospective study with a high risk of bias.

Summary of the literature
In the 2020 systematic review, 3 we found insufficient high-quality evidence to assess the effect of negative pressure wound therapy (NPWT) on infection-related outcomes in patients with DFI. 87,88We identified one study in our recent literature search, that examined the potential impact of using 0.1% polyhexanide-betaine irrigation in NPWT. 23In a large clinical trial, 150 patients with diabetes and moderate or severe DFIs that required surgical drainage were randomised to receive NPWT either with or without 0.1% polyhexanide-betaine irrigation.The study included hospitalised adults in one medical centre in the USA with wound size 5-100 cm 2 .
There were no significant differences between the two groups in wound size after index surgery, duration of antibiotics, number of surgeries during the index hospitalisation, duration of NPWT, surgical wound closure, wound dehiscence of surgically closed wounds, proportion of healed wounds, time to wound healing, length of hospitalisation, re-infection, all cause re-hospitalisation, and foot specific re-hospitalisation.Outcomes from this randomised clinical trial suggest no added benefit from using 0.1% polyhexanide-betaine irrigation in NPWT.A recent retrospective study of low quality that included 106 patients with ST-DFIs (some with foot ischaemia) compared the efficacy of surgical debridement with or without use of ultrasound debridement and NPWT in combination with sequential cycles of antiseptic injection into the wound cavity. 34Control patients were treated with topical tamed iodine solutions and polyethylene glycol ointments with or without foot plastic surgery depending on the foot ulcer outcome.Those treated with ultrasound debridement and NPWT had lower number of major amputations (p = 0.016) and reduced length of hospital stay (p = 0.001).However, the quality of this study was such that no firm conclusion can be drawn on the beneficial effect of NWPT in DFIs.

Evidence statement
Available data are inconsistent regarding the suggest a potential benefit of in using NPWT with or without irrigation.

Certainty of evidence
Low, based on one RCT with minimal risk of bias and other nonrandomised studies.

Summary of the literature
Two small observational retrospective cohort single-centre studies examined the safety and efficacy of surgical bone debridement and systemic antibiotic therapy with or without filling bioactive glass S53P4 into the infected debrided bone in subjects with DFO. 10,20e two studies included a total of 66 hospitalised patients in Italy 10 and Slovenia. 20Both studies differed on the location of DFO (first metatarsophalangeal joint with adjacent bone osteomyelitis vs.
different locations of foot osteomyelitis) and the extent of bone resection (limited vs. resection of all the infected bone).While one study 10 found a significantly higher rate of resolution of DFO than in subjects treated with standard treatments (18 [90%] vs. 13

Evidence statement
Limited available data is contradictory about the benefit of bioactive glass S53P4 for the treatment of DFO.

Certainty of evidence
Low, based on two recent small-sized observational retrospective cohort studies with serious risk of bias.

Summary of the literature
One small RCT examined the effect of application of cold atmospheric plasma or placebo in addition to standard care therapy on wound healing in patients with diabetes and chronic foot ulcer infection. 36The study included 43 participants from the hospital or ambulatory clinics in Germany.Cold atmospheric plasma therapy yielded a significant increase in wound healing, both in total area reduction and time to relevant wound area reduction.However, there was no significant difference in clinical resolution of infection or microbial burden between cold atmospheric plasma and placebo by the end of treatment.No cold atmospheric plasma therapy-related adverse events occurred during treatment.

Evidence statement
Available data do not suggest improved diabetic foot infectionrelated outcomes from cold atmospheric plasma.

Certainty of evidence
Low, based on one RCT with high risk of bias.

| DISCUSSION
In this updated systematic review on interventions for the management of persons with DFIs, the largest number of studies were related to antibiotic treatments for soft-tissue DFIs and DFO.As reported in our previous systematic review, 3 the separation of these two groups is debatable, as the various studies used different definitions for both entities, the percentage of subjects with DFO was sometimes small, and infected bone was removed prior to inclusion in most trials. 3This may explain the apparent resolution of a substantial number of included cases labelled as having osteomyelitis with only a relatively short course of antibiotic therapy.In addition to short-term measures of microbiological response and apparent clinical cure, studies of the treatment of osteomyelitis should optimally include some measures (clinical, laboratory, and imaging) of long-term clinical remission.Finally, this separation appears a bit artificial given that both soft-tissue DFIs and DFO may present concomitantly in a nonnegligible proportion of the patients.
We identified a total of 5418 articles published between 2018 and 2022, of which 32 met our inclusion criteria.Overall, the quality of the studies tended to increase in comparison with previous systematic reviews, noting however that there is still a need for more and its duration, especially following surgical intervention for both soft-tissue DFIs and DFO.These data confirm the current international trend to reduce the exposition of patients to the potential negative effects of antibiotics.
We did not find any new studies that addressed the possibility to reduce the 6-week duration of antibiotic therapy for DFO treated medically (without any resection of the infected bones).The new studies we identified confirm that there is a place for conservative treatment of DFO in selected patients and that early surgery is associated with improved outcomes in patients with moderate and severe DFIs.We could not identify clinical trials of sufficient quality that support the use of any topical intervention for the management of DFIs.
In order to improve the quality of future research we suggest that researchers in this area should strictly adhere to the agreed definitions for the diagnosis of soft-tissue DFIs and DFOs, for describing their treatment as well as its outcomes to enable comparison of studies and to translate these findings into recommendations for daily practise.
28 days.Ulcer healing was recorded in 17/25 (68%) patients in the linezolid group and in 14/25 (56%) other patients (p = 0.76).In this study, no microbiology results were provided, which makes the interpretation of the results difficult.The use of carbapenem antibiotics is usually necessary for the treatment of multiresistant DFIs. 3,45In a multicenter observational F I G U R E 1 Flow diagram for 2023 systematic review on the intervention for diabetes-related foot infections.PETERS ET AL.

[Correction added on 9 -
November-2023, after first online publication, 'from the same group' has been removed from the sentence.]As in the other study, clinical failure was not associated with the total duration of antibiotic therapy.Both studies could not determine a threshold for the optimal duration of antibiotic therapy to prevent recurrences of DFI and clinical failures.It remained unclear how participants were recruited and how many of them were included in both studies.
DFO was diagnosed on a combination of a positive probe-to-bone probe test with abnormalities on radiography compatible with bone involvement, and elevated laboratory tests including leucocytosis and erythrocyte sedimentation rate.Patients were randomly assigned to a 6-week clindamycin or 12-week ciprofloxacin regimen.No significant difference in clinical outcomes between the two groups was recorded (11 complete improvements out of 15 in both groups) at the end of a 3-month post-end of treatment follow-up.The conclusions of the study are limited by the small number of patients(30), the use of different antibiotic regimens in the two groups of patients, a short follow-up period, and the lack of information about bone microbiology.
Remission based on the resolution of clinical signs of infection assessed at the end of the antibiotic therapy (primary outcome) and 1 month later (secondary outcome) was recorded in both groups in 7 (87.5%) in both groups.The daily dose of daptomycin was 500 mg and ranged from 400 to 800 mg for teicoplanin.Daptomycin and teicoplanin were administered for a mean duration of respectively, 35.1 � 22.7 and 45.2 � 27.4 days.The main limitations of this study concern (i) the definition of remission of DFO limited to clinical signs of infection, (ii) the very short follow-up duration, (iii) the small number of patients and (iv) the absence of data regarding the use of a concomitant antibiotic during treatment.

[ 61 .
9%], respectively p = 0.03) and a lower probability of requiring additional antibiotic therapy, the other study20 did not find any benefit in terms of infection resolution and ulcer healing in comparison to the standard treatment (10/10 vs. 9/12; p = 0.22).The available data on the potential benefit of bioactive glass in persons with DFOs is only based on two low-quality non-randomised studies.
high-quality studies to underpin clinical practice in the management of DFIs.Unfortunately, data about the assessment of new antimicrobials in the settings of DFIs are still very limited, although new antibiotics with activity against multi-resistant strains of staphylococci, enterococci and gram-negative rods have appeared in the last years.New data have emerged regarding the mode of administration of the antibiotic therapy (i.e.intravenous vs. oral route of application) A higher postoperative healing rate, longer mean healing duration and lower recurrence rate were recorded in the intervention group, but the authors did not state why calcium sulphate PETERS ET AL.