Antimicrobial resistance in diabetic foot osteomyelitis: diagnosis and treatment

INTRODUCTION

Diabetic foot osteomyelitis is a limb-threatening complication of foot ulceration in people with diabetes and one of the most frequent reasons for non-traumatic lower-extremity amputation worldwide.(1) Bone infection complicates between 20% and 60% of diabetic foot infections, with prevalence rising from around 20% in moderate infections to more than 50% in severe presentations, and it accounts for 20 to 70% of all foot infections in this population.(2-4) Once infection reaches bone, remission usually requires prolonged antibiotic exposure and, in many patients, surgical resection, both of which extend hospitalization and increase cost, disability and the risk of limb loss.(5) The burden falls unevenly across health systems: ulceration that precedes bone infection recurs frequently, and each episode mobilizes wound care, imaging, antimicrobial therapy and, in advanced cases, revascularization and amputation within a coordinated multidisciplinary foot service.(5) Because the infection occupies cortical and medullary bone, it behaves differently from soft tissue infection in both its natural history and its response to treatment.(2)

Antimicrobial resistance has altered the calculus of care. Multidrug-resistant organisms were identified in 71.5% of patients with diabetic foot osteomyelitis in one retrospective cohort, and resistance among foot isolates has risen across the past decade, so empirical regimens that once covered methicillin-resistant staphylococci and Pseudomonas aeruginosa now fail against an expanding set of gram-negative and biofilm-forming pathogens.(6,7) The evidence that connects resistance patterns to diagnostic strategy and treatment selection is dispersed across diagnostic accuracy studies, microbiology series, randomized trials of antibiotic duration and route, and surgical case series, without a synthesis that links the microbiology of the infected bone to the choice of sampling, agent, route and duration.(1,2) The objective of this review is to synthesize current evidence on the diagnosis and antimicrobial treatment of diabetic foot osteomyelitis, with emphasis on antimicrobial resistance and its implications for treatment selection.

DEVELOPMENT

Bone infection should be suspected in most severe soft tissue infections of the diabetic foot, because the prevalence of osteomyelitis reaches 60% in that subgroup and drives the decision between medical and surgical management.(2) A systematic review that underpins the 2023 guideline confirms that clinical examination, the probe-to-bone test and imaging form the basis of suspicion, with confirmation reserved for a bone sample obtained aseptically.(12) The probe-to-bone test and plain radiography constitute the first-line combination: probing to bone in an ulcer with high pretest probability supports the diagnosis, and radiography documents cortical destruction when it is present.(2,13,14) The combination has been formalized in meta-analysis, which pooled the plain radiograph and the probe-to-bone test as a single bedside strategy for early recognition and found the pairing useful in practice.(13) The erythrocyte sedimentation rate combined with the probe-to-bone test has likewise been proposed for fast early diagnosis in settings without advanced imaging.(14)

In a diagnostic accuracy study of 128 patients, the probe-to-bone test combined with inflammatory biomarkers reached a sensitivity of 75% and a specificity of 24%, whereas its combination with plain radiography reached 76% sensitivity and 23% specificity, so the combined tests detect most cases but rule in poorly.(3) Plain radiography is inexpensive and specific for established cortical destruction, but its sensitivity for early bone infection is low, a normal film does not exclude osteomyelitis, and the modality is better suited to monitoring than to first diagnosis.(15) The low specificity of the bedside combinations implies that a positive probe-to-bone test in a low-prevalence setting should be followed by a confirmatory sample rather than by empirical escalation alone.(3,15) In a high-prevalence context, by contrast, the same combination is sufficient to begin therapy while awaiting culture results.(2)

Serum biomarkers add limited discrimination between bone and soft tissue infection when used alone. A meta-analysis of 14 studies with 1698 patients reported a pooled sensitivity of 0.78 and a specificity of 0.72 for C-reactive protein, a sensitivity of 0.77 and a specificity of 0.74 for erythrocyte sedimentation rate, and a sensitivity of 0.88 and a specificity of 0.81 for procalcitonin, which showed the highest accuracy but still remained insufficient to replace sampling.(16) Composite cut-off values for C-reactive protein improve standardization across laboratories but do not resolve the underlying overlap between infected and uninfected states.(17) These biomarkers are therefore best interpreted as supporting evidence that raises or lowers pretest probability, not as stand-alone diagnostic tests.(16) Their main clinical value is in monitoring response once the diagnosis has been established by other means.(17) A normal procalcitonin or C-reactive protein likewise does not exclude osteomyelitis when bone is exposed, so biomarkers should not be used to withhold sampling or imaging in a high-risk ulcer.(16,17)

Integrated in sequence, the tests perform differently than in isolation. The probe-to-bone test and plain radiography identify the patients who need a bone sample, biomarkers adjust the urgency of that sample, and advanced imaging localizes the infection and its extent before surgery.(2,3,18) This stepwise logic matches the guideline recommendation to reserve the reference-standard bone culture for patients in whom the pretest probability is high, while cheaper tests exclude bone infection in the remainder.(12) Only a culture, however, converts the diagnosis into a treatment plan, because the agent and the duration both depend on the organism recovered.(2,19)

Among imaging techniques, white blood cell SPECT/CT and conventional MRI were compared prospectively in 47 patients using bone biopsy as the reference standard; SPECT/CT achieved a sensitivity of 85% and a specificity of 79%, whereas MRI achieved 73% and 43%, and after antibiotic treatment both methods performed equivalently, with sensitivity and specificity of 75% each.(18) The lower specificity of MRI in that cohort reflects the marrow changes that accompany treated and ischemic bone, which limits its use as a stand-alone rule-in test despite its high spatial resolution.(18) SPECT/CT therefore offers a viable non-invasive alternative when available, although its higher radiation burden and cost restrict routine use.(18) Serial radiographs document progression or healing and remain the cheapest longitudinal tool once the diagnosis is secure.(15) Neither modality replaces tissue sampling, because imaging cannot identify the organism or its susceptibility profile.(18) The reported performance of the available methods is summarized in Table 1 (Table 1).

Microbiological confirmation rests on a bone sample obtained percutaneously or during surgery, because culture of a bone specimen is the reference standard that distinguishes osteomyelitis from soft tissue infection and guides targeted therapy.(2,20) A systematic review and meta-analysis of percutaneous bone biopsy documented its value for identifying the causative pathogen, although yield depends on technique and on the timing of antibiotics.(20) The choice between percutaneous and intraoperative sampling is practical rather than biological, because percutaneous biopsy avoids anesthesia and can be repeated whereas intraoperative biopsy allows simultaneous resection and dead-space management.(19,20) Histopathological and microbiological examination of the same bone biopsy are only partially concordant, and prior antibiotic exposure reduces the recovery of bacterial pathogens without abolishing it, which argues for sampling before prolonged therapy whenever feasible.(21,22) Deep tissue cultures approximate bone biopsy results but recover the colonizing flora of the ulcer more than the organisms inside bone.(23) Superficial wound swabs are not reliable for bone infection for the same reason, because they reflect the colonizing flora of the ulcer surface rather than the organisms invading bone.(2,23) A randomized trial of 84 participants compared antibiotic regimens guided by percutaneous bone biopsy or by ulcer bed biopsy and found 12-month remission of 31.4% and 39.4%, respectively, with a relative risk of not achieving remission of 1.13, so bone biopsy was not superior to the simpler ulcer bed sample.(19) The trial was double-blind and analyzed both intention-to-treat and per-protocol populations, which strengthens the inference that the two sampling strategies are interchangeable for remission.(19) Molecular tools recover additional organisms that culture misses: amplicon sequencing and molecular culture identified taxa not detected by conventional culture in paired bone and ulcer bed specimens, and metagenomic next-generation sequencing increased pathogen detection in proof-of-concept cohorts.(24,25) Metagenomic and ribosomal transcript profiling in Hispanic patients further showed biofilm-associated bacteria that conventional culture underestimates.(26) The clinical weight of these additional taxa remains uncertain, because detection does not always distinguish colonizer from pathogen.(24) The value of molecular methods will depend on demonstrating that organisms detected only by sequencing change treatment or outcome, a question that the current proof-of-concept studies leave open.(24,25)

The microbiology of diabetic foot osteomyelitis is dominated by gram-positive cocci, with Staphylococcus aureus as the most common pathogen, followed by coagulase-negative staphylococci and enterococci.(6,27) In a cohort of 85 patients undergoing amputation, Staphylococcus species represented 30.6% of isolates, with a seasonal shift toward gram-negative organisms in warmer months that alters empirical choices within a single region.(28) Regional series confirm that the microbial diversity underlying the infection is broad and that susceptibility profiles vary by centre, which complicates any universal empirical recommendation.(27) The dominance of staphylococci has two practical consequences: empirical coverage must include methicillin-resistant strains where local prevalence warrants it, and targeted therapy can narrow to a single anti-staphylococcal agent once susceptibility is known.(6,27) The host features that predict resistant organisms, diabetes duration, wound duration, anemia and ischemia, overlap with the features that delay wound healing, so the patient at highest risk of a resistant infection is also the patient least able to clear it.(6)

Gram-negative bacilli are increasingly reported. A multicentre European cohort spanning two decades documented a rising prevalence of Enterobacteriaceae as pathogens of diabetic foot osteomyelitis, which changes empirical therapy because these organisms require agents directed against Enterobacterales rather than against staphylococci.(11) Among patients with multidrug-resistant infection, Staphylococcus aureus and Enterococcus faecalis predominate among gram positives, whereas Proteus mirabilis, Escherichia coli, Klebsiella pneumoniae and Enterobacter cloacae predominate among gram negatives.(6) Retrospective cohorts link these organisms to specific host features and to prognosis after treatment, which supports risk-based rather than uniform empirical regimens.(29) The rise of Enterobacteriaceae also tracks the selective pressure of prolonged broad-spectrum therapy and prior hospitalization, both of which are common in this population, and the shift is clinically silent until culture results return.(11) Seasonal and geographic variation in pathogen distribution argues against fixed empirical protocols and in favor of centre-specific antibiograms updated at regular intervals.(27,28) The interpretation of a polymicrobial culture is itself a source of divergence, because the same bone specimen may yield several organisms whose individual contribution to the infection is uncertain.(21,24) In concordance studies, not every isolate recovered from bone required targeted coverage, which argues for treating the dominant pathogen while covering the remainder only when clinically indicated.(21)

Resistance is common and spans multiple drug classes, and a systematic review of 28 studies published between 2014 and 2024 documented rising resistance among diabetic foot isolates.(7) Multidrug-resistant organisms were found in 71.5% of 102 patients with osteomyelitis, and diabetes duration, wound duration, anemia and lower-limb ischemia were independent predictors, integrated in a nomogram with an area under the curve of 0.906.(6) Comparable risk factors for multidrug-resistant organisms have been reported in ulcer cohorts, which reinforces the value of local surveillance data when empirical therapy must begin before cultures return.(30) Biofilm and intracellular persistence add a biological layer to resistance: staphylococci invade the submicron osteocyte lacuna-canaliculi network of bone, and biofilm-forming bacteria persist despite susceptible phenotypes in vitro, so susceptibility reports may overstate the effectiveness of agents that do not reach sequestered bacteria.(26,31) This biology explains part of the gap between in vitro susceptibility and clinical recurrence.(31)

The clinical consequence of this resistance burden is that empirical therapy is now probabilistic rather than prescriptive. A regimen chosen before culture must cover the most likely organisms in the local population, accept the risk of missing an unexpected resistant isolate, and be narrowed as soon as susceptibility results return.(6,32) The nomogram that predicts multidrug-resistant organisms from diabetes duration, wound duration, anemia and ischemia is one attempt to make that probability explicit at the bedside.(6) Risk-based empirical therapy, rather than uniform broad-spectrum therapy, therefore emerges as the least harmful default in settings with heterogeneous resistance.(30,32)

Antimicrobial treatment has moved toward shorter courses. Meta-analytic evidence indicates that shorter antibiotic courses do not worsen outcomes in diabetic foot osteomyelitis, consolidating the six-week reference standard established by earlier randomized work.(33,34) A protocol for a multicentre, single-blind randomized trial now compares four against six weeks in nonsurgically treated disease, which signals continued interest in shortening exposure.(34) The four-versus-six-week question is being tested in a multicentre design precisely because the six-week standard rests on a single earlier trial rather than on a series of confirmatory studies.(34) After amputation, three weeks of postoperative antibiotics were comparable to six weeks when residual bone remained culture-positive, so prolonged postoperative courses may be unnecessary.(35) These data align with a broader body of evidence that favors the shortest effective course once the infected bone has been adequately managed, because longer exposure increases toxicity, cost and selection pressure without improving remission.(33) The duration decision nevertheless depends on whether infected bone was fully resected, because medically treated or partially resected disease behaves differently from completely excised disease.(34,35) For patients treated without surgery, the six-week course remains the reference because residual infected bone requires sustained exposure, whereas fully resected disease allows shorter postoperative courses.(33,35)

Two unresolved questions dominate the treatment literature. The first is whether the six-week standard can be compressed to four weeks without losing remission, a question under test in a multicentre randomized protocol.(33,34) The second is whether patients with residual positive bone cultures after amputation need any postoperative antibiotics at all, because three weeks and six weeks already appear equivalent.(35) Both questions matter for resistance, because every week of unnecessary exposure selects resistant flora in the patient and in the ward.(33) The answers will refine, rather than overturn, the current short-course direction.(34)

The route of administration is likewise changing. In a propensity-matched cohort of 49 pairs with unresected osteomyelitis, treatment failure did not differ between oral and intravenous therapy.(36) A systematic review of 26 studies with 972 patients reported complete resolution with oral monotherapy in 75.2% and, after intravenous step-down, in 56.5%, with no resolution or amputation in 21.6% and 23.1%, respectively.(37) A scoping review of 25 studies found comparable clinical outcomes between oral and intravenous routes, and most of the included studies were retrospective, which limits causal claims but reflects real-world practice.(38) Oral use in hospitalized patients increased after the OVIVA trial, although uptake remains inconsistent across centres.(39) Highly bioavailable oral agents therefore appear sufficient for the majority of patients, with intravenous therapy reserved for those who cannot absorb or tolerate oral drugs.(36,37) The advantage of the oral route depends on the bioavailability of the chosen agent, which the reviewed studies did not standardize.(36,37) The shift to oral therapy also reduces catheter-related complications and enables earlier discharge, which matters for a population with long treatment courses.(38) Oral step-down likewise matches the economic reality of long outpatient courses, where intravenous access is the main driver of cost and complication.(38,39) Empirical therapy must cover methicillin-resistant staphylococci in centres where their prevalence exceeds a locally defined threshold, while avoiding routine antipseudomonal coverage when Pseudomonas is uncommon, and the microbiology-driven order set illustrates how such thresholds can be operationalized.(6,40)

Anti-biofilm regimens with rifampicin against staphylococci and fluoroquinolones against gram-negative bacilli are biologically appealing but lack randomized support.(41) A randomized placebo-controlled trial of rifampin to reduce pedal amputations in veterans with diabetes did not demonstrate benefit, which cautions against adding rifampin solely on the basis of biofilm activity.(42) The discrepancy between in vitro biofilm efficacy and clinical outcome indicates that trials must measure remission and limb salvage rather than surrogate endpoints.(41,42) Until such trials are available, rifampin should be reserved for specific indications defined in guidelines rather than used routinely.(41)

Long-acting agents such as dalbavancin provide outpatient alternatives in real-world cohorts, where they reduce the need for daily intravenous access.(43) Narrow empiric antibiotic spectra were not associated with worse outcomes in bone and joint infections, which supports de-escalation once susceptibility data are available.(32) In turn, the choice of agent should follow the local antibiogram and the patient's prior antibiotic exposure, because prior therapy is among the strongest predictors of resistant organisms.(6,32) Susceptibility-guided prescribing therefore becomes both a treatment decision and a resistance-containment measure.(32) The same principle governs empirical selection, because a regimen that overestimates the local resistance burden exposes the patient and the unit to avoidable selection pressure.(32,40)

Surgery and antibiotics are complementary rather than competing strategies. Expert consensus ranked ischemia first among reasons to operate, with 91% agreement, followed by gangrene, sepsis, frailty and major soft tissue loss, and more than 90% of experts favored surgery for exposed, destroyed or hindfoot bone.(44) The consensus was built from a review of 118 articles and questionnaires completed by experts who had published on the choice between first-line antibiotics and direct surgery.(44) It reflects a stratification in which antibiotics alone are reserved for patients without those features, whereas direct resection is preferred when the bone is exposed or destroyed.(44) The decision between medical and surgical treatment is therefore a trade-off between the morbidity of resection and the longer antibiotic exposure of conservative management, and the consensus criteria in ischemia and gangrene resolve that trade-off in favor of surgery.(44)

Conservative surgery guided by histopathological analysis of the infected bone and margins produced resolution in a series of 25 patients, with targeted antibiotics adjusted to the resected specimen.(4) Complete resection of infected bone does not consistently improve outcomes: residual osteomyelitis at the resection margin was not associated with recurrence in one cohort, and positive bone margins remain controversial because histologic residue does not reliably predict failure.(45-47) Residual osteomyelitis at the margin can even be managed without postoperative antibiotics in selected cases, which questions the value of chasing a clean histologic margin at the expense of foot architecture.(46) The margin debate has practical weight, because a conservative resection preserves foot architecture and offloading capacity at the cost of possibly leaving histologically infected bone.(46,47)

Local antibiotic delivery fills the surgical dead space with high local concentrations. In the BIG D-FOOT randomized trial, bioabsorbable calcium sulphate granules delivering tobramycin or vancomycin were superior to placebo for post-surgical infective complications, and the trial was terminated early after 20 cases because of the advantage of the active arm.(48) The trial was double-blind and placebo-controlled, with a composite primary endpoint of dehiscence, infection, recurrence and new adjacent osteomyelitis at 12 weeks.(48) Bioabsorbable antibiotic-loaded bone substitutes combine local delivery with osteoconductive dead-space filling, although the evidence remains largely observational and heterogeneous.(49) A meta-analysis of antibiotic-impregnated cement pooled six studies with 82 patients and 85 procedures, and separate syntheses of local delivery systems support infection control and limb preservation without strong comparative data.(50,51) A narrative review of 19 studies concluded that local antibiotics can be added to standard systemic therapy in selected cases rather than replacing it.(52) Bone graft substitutes also improved outcomes of calcaneal osteomyelitis in a retrospective cohort.(53) The common thread is that local carriers achieve concentrations above the minimum inhibitory concentration of common pathogens while limiting systemic toxicity, and they convert a two-stage dead-space problem into a single-stage procedure when bone loss is limited.(49) Local carriers are most useful when the bone defect is contained, because diffusion into adjacent soft tissue is limited and the carrier acts mainly within the cavity.(49,50)

Despite these gains, re-infection occurred in 27.5% of 171 patients within 12 months of treatment, driven by ischemia, delayed healing and prior myocardial infarction, and re-infection multiplied the odds of subsequent hospitalization.(54) The recurrence burden implies that surgical and antimicrobial success must be measured over at least a year and that revascularization and wound closure are inseparable from infection control.(54) Only with adequate perfusion and durable soft tissue cover can the gains of targeted antimicrobial therapy be sustained, and this triad of infection control, revascularization and wound closure defines the treatment endpoint.(54)

Stewardship closes the loop between diagnosis and treatment. A microbiology-driven electronic health record algorithm with an order set reduced unnecessary antipseudomonal prescribing in diabetic foot infections, and MRSA nares screening allowed de-escalation of anti-methicillin-resistant Staphylococcus aureus agents when negative.(40,55) Narrow empiric antibiotic spectra were not associated with worse outcomes in bone and joint infections, which supports de-escalation after sampling rather than broad empirical escalation.(32) These interventions reduce selection pressure without compromising clinical outcomes, and they depend on reliable, timely, actionable microbiology results available at the point of prescribing.(32,40)

Practice remains variable: a survey of North American infectious disease specialists documented wide variation in oral and suppressive antibiotic use for osteoarticular infections, and the same variability appears in the diabetic foot literature.(56) Variation of this kind indicates that the evidence gap is organizational as well as biological, because clinicians apply different thresholds for oral step-down and for suppressive therapy.(56) Standardized order sets and audit feedback are the operational tools that close that gap.(40) Audit of prescribing against the local antibiogram closes the feedback loop, because it identifies the empirical choices that require later correction.(40,56)

The synthesis also exposes an organizational gap. Microbiology-driven prescribing requires a laboratory that reports susceptibility in time to influence the first follow-up decision, and multidisciplinary clinics that can act on that report.(40,57) Where such infrastructure is absent, the reviewed algorithms remain aspirational, which is why the Latin American literature calls for local antibiograms before local protocols.( 57)

Multidisciplinary teams that integrate podiatry, infectious diseases, vascular surgery and clinical pharmacy are the setting in which these recommendations operate, and recent syntheses describe algorithms that link wound classification, revascularization and antimicrobial stewardship to improve limb preservation.(57) Diagnostic stewardship is the counterpart of therapeutic stewardship, because biomarkers and imaging that refine pretest probability reduce unnecessary antibiotic exposure in patients without bone infection.(3,16) The evidence across the five axes is summarized in Table 2 (Table 2).

DISCUSSION

The principal finding of this synthesis is that antimicrobial resistance has become the organizing principle of diabetic foot osteomyelitis care: diagnosis increasingly seeks a reliable microbiological sample, and treatment has shifted from prolonged empirical intravenous courses toward short, oral, targeted regimens supported by surgery and stewardship.(1) Each diagnostic and therapeutic choice now hinges on local resistance epidemiology rather than on a fixed empirical protocol, which explains the convergence of the reviewed studies despite their clinical heterogeneity.(6,7) In effect, the question has changed from which antibiotic to start to which sample, which duration and which patient needs surgery, and the answer depends on the local antibiogram.(1) This represents a maturation of the field from a one-size-fits-all empirical approach toward a stratified, microbiology-anchored one in which the same pathogen may warrant different therapy in different centres.(1,6) The 71.5% prevalence of multidrug-resistant organisms in one recent cohort illustrates why that stratification is no longer optional.(6) A prevalence of that magnitude means that empirical regimens selected without local data will fail in a substantial minority of patients before cultures return.(6,7)

This position is concordant with current guidelines and with several recent comparative studies. The IWGDF/IDSA 2023 update and the earlier 2019 guideline both prioritize appropriate specimen collection and culture-based therapy over empirical escalation, and the 2019 document already recommended limiting antibiotic duration and route according to infection extent.(1,9) The oral-first evidence summarized here extends the OVIVA principle to the diabetic foot, where propensity-matched and systematic review data showed equivalence between oral and intravenous therapy.(36-38) The short-duration position is likewise supported by the six-week landmark trial, whose non-inferiority has been confirmed by subsequent meta-analyses.(33,58) Local antibiotic delivery aligns with these trends by concentrating drug at the infected site while sparing systemic exposure, and its strongest randomized signal came from a trial stopped early for superiority.(48,49) Earlier narrative syntheses reached the same broad conclusion that antibiotic therapy and surgery must be matched to the extent and microbiology of the infection.(59) Current reviews of diabetic foot infection likewise frame resistance as the dominant constraint on empirical therapy and place sampling at the centre of the algorithm.(60) The convergence across guideline, trial and observational sources persists despite the absence of large confirmatory trials for several of these recommendations.(1,59) The practical consequence is that a clinician can justify a short oral course in most patients once susceptibility is known, without fearing inferiority to prolonged intravenous therapy.(36,37)

Some findings conflict with the prevailing narrative and require explanation. The randomized trial that found bone biopsy not superior to ulcer bed biopsy challenges the reference-standard status of bone sampling, but its remission rates were low in both arms and its population had advanced, conservatively treated disease, which limits transfer to surgical candidates; the ulcer bed sample may also capture the same organisms when the ulcer communicates directly with bone.(19) The modest 43% specificity of MRI in one prospective comparison departs from the classic view of MRI as a highly sensitive test, a difference explained by the use of bone biopsy rather than clinical outcome as the reference and by treatment-related marrow changes.(18) The positive BIG D-FOOT trial was terminated early after 20 cases, an advantage that early stopping can exaggerate, whereas the wider local-delivery literature remains observational and heterogeneous.(48,52) The absence of benefit from adjunctive rifampin in a randomized trial tempers enthusiasm for anti-biofilm regimens that are otherwise supported only by observational data, and it highlights the distance between in vitro biofilm activity and clinical outcome.(41,42) The equivalence of antibiotics and conservative surgery observed in earlier randomized work similarly depends on strict patient selection, because expert opinion now reserves surgery for ischemia, gangrene and destroyed bone.(44,61) These divergences are consistent with the heterogeneity of the disease rather than with fundamental contradictions in the evidence.(18,19) The divergence is most informative where it concerns the reference standard and the imaging gold standard, because both affect every subsequent decision.(18,19) The early stopping of the local-delivery trial and the null rifampin trial illustrate the same methodological point: single favorable or null results should not, on their own, reset practice.(42,48)

The review has concrete limitations, each with a remedy. Selection was purposive and language-restricted to English and Spanish, so relevant reports in other languages may have been missed; a future systematic review with a preregistered protocol would quantify that gap.(8) Diagnostic accuracy estimates came largely from single-centre studies with variable reference standards, which inflates heterogeneity; standardized definitions and core outcome sets would improve comparability.(18) Resistance data are dominated by retrospective cohorts with inconsistent multidrug-resistance definitions, so pooled prevalence should be interpreted cautiously and complemented by prospective surveillance.(6,7) Publication bias is plausible for local delivery, where a favorable early-stopped trial coexists with small positive series; preregistration and non-inferiority designs would reduce that risk.(48) The narrative design itself precludes quantitative pooling, which limits the precision of any single estimate reported here, and the publication window may miss older but still relevant controlled trials.(8) These limitations do not invalidate the direction of the synthesis but they do mean that individual estimates should be read as indicative rather than definitive.(8) Finally, most evidence originates in high-income settings, so extrapolation to Latin American practice requires local validation rather than direct transfer, and the scarcity of regional antibiograms compounds that uncertainty.(5)

The practical implication is a microbiology-driven pathway: obtain a bone or ulcer bed sample before prolonged therapy, favor short oral courses once susceptibility is known, reserve surgery for ischemia, gangrene, sepsis or destroyed bone, and embed antibiotic stewardship within a multidisciplinary foot team.(1,40) Clinics in Latin America, where the burden is high and resistance surveillance remains thin, stand to gain most from local antibiograms and integrated limb-preservation programs, because empirical decisions there are least supported by local data.(5,57) Research should prioritize randomized comparisons of duration and route in surgical candidates, validation of molecular diagnostics against clinical outcomes, and longitudinal resistance surveillance in Latin American centers, which the current literature addresses only partially.(60,62) Such work would turn the current qualitative convergence into quantitative guidance for empirical therapy.(62) In the interim, teams that measure and report their own resistance rates and outcomes will be best placed to adapt these recommendations to their setting, and that local measurement is itself a form of stewardship.(57)

CONCLUSIONS

Current evidence supports a microbiology-driven strategy in which bone or ulcer bed sampling, short oral antibiotic courses and selective surgery are combined to limit antimicrobial resistance in diabetic foot osteomyelitis. Resistance profiles, dominated by staphylococci with rising gram-negative and multidrug-resistant organisms, must guide both empirical and targeted therapy. Stewardship and multidisciplinary care are the mechanisms that translate this evidence into limb salvage.

References

1. 1 Senneville É, Albalawi Z, van Asten SA, Abbas ZG, Allison G, Aragón-Sánchez J, et al. IWGDF/IDSA Guidelines on the Diagnosis and Treatment of Diabetes-related Foot Infections (IWGDF/IDSA 2023). Clin Infect Dis. 2023:ciad527. doi: 10.1093/cid/ciad527
2. 2 Senneville EM, Lipsky BA, van Asten SAV, Peters EJ. Diagnosing diabetic foot osteomyelitis. Diabetes Metabolism Res. 2020;36(S1):e3250. doi: 10.1002/dmrr.3250
3. 3 Herrera-Casamayor M, Sanz-Corbalán I, Tardáguila-García A, López-Moral M, Lázaro-Martínez JL, García-Álvarez Y. Comparing the Diagnostic Accuracy of the Probe-to-Bone Test, Plain Radiography, and Serum Biomarkers in Detecting Diabetic Foot Osteomyelitis. JCM. 2026;15(2):500. doi: 10.3390/jcm15020500
4. 4 Da Ros R, Assaloni R, Michelli A, Brunato B, Miranda C. Antibiotic and Surgical Treatment of Diabetic Foot Osteomyelitis: The Histopathological Evidence. Antibiotics. 2024;13(12):1142. doi: 10.3390/antibiotics13121142
5. 5 Chávez Moya MO, Cabrera LH, Ramírez LM, González González MA, Pazos AA, Fernández Feitó LB, et al. From ulcer to remission: podiatric and technological strategies to prevent ulceration and amputation in the diabetic foot. SAP Podiatry. 2026;3:160. doi: 10.62486/pod2026160
6. 6 Hu L, Xu S, Yang S, Meng G. Characteristics of multidrug-resistant bacteria infection and analysis of risk factors in diabetic foot complicated with osteomyelitis. Journal of Tissue Viability. 2026;35(4):101068. doi: 10.1016/j.jtv.2026.101068
7. 7 Li W, Sadeh O, Chakraborty J, Yang E, Basu P, Kumar P. Multifaceted Antibiotic Resistance in Diabetic Foot Infections: A Systematic Review. Microorganisms. 2025;13(10):2311. doi: 10.3390/microorganisms13102311
8. 8 Bus SA, Monteiro‐Soares M, Game F, van Netten JJ, Apelqvist J, Fitridge R, et al. Standards for the development and methodology of the 2023 IWGDF guidelines. Diabetes Metabolism Res. 2024;40(3):e3656. doi: 10.1002/dmrr.3656
9. 9 Lipsky BA, Senneville É, Abbas ZG, Aragón‐Sánchez J, Diggle M, Embil JM, et al. Guidelines on the diagnosis and treatment of foot infection in persons with diabetes (IWGDF 2019 update). Diabetes Metabolism Res. 2020;36(S1):e3280. doi: 10.1002/dmrr.3280
10. 10 Schaper NC, van Netten JJ, Apelqvist J, Bus SA, Hinchliffe RJ, Lipsky BA, et al. Practical Guidelines on the prevention and management of diabetic foot disease (IWGDF 2019 update). Diabetes Metabolism Res. 2020;36(S1):e3266. doi: 10.1002/dmrr.3266
11. 11 Soldevila-Boixader L, Murillo O, Waibel FWA, Schöni M, Aragón-Sánchez J, Gariani K, et al. The increasing prevalence of Enterobacteriaceae as pathogens of diabetic foot osteomyelitis: A multicentre European cohort over two decades. International Journal of Infectious Diseases. 2025;154:107843. doi: 10.1016/j.ijid.2025.107843
12. 12 Senneville É, Albalawi Z, van Asten SA, Abbas ZG, Allison G, Aragón‐Sánchez J, et al. Diagnosis of infection in the foot of patients with diabetes: A systematic review. Diabetes Metabolism Res. 2024;40(3):e3723. doi: 10.1002/dmrr.3723
13. 13 Calvo-Wright MDM, Álvaro-Afonso FJ, López-Moral M, García-Álvarez Y, García-Morales E, Lázaro-Martínez JL. Is the Combination of Plain X-ray and Probe-to-Bone Test Useful for Diagnosing Diabetic Foot Osteomyelitis? A Systematic Review and Meta-Analysis. JCM. 2023;12(16):5369. doi: 10.3390/jcm12165369
14. 14 Xu J, Cheng F, Li Y, Zhang J, Feng S, Wang P. Erythrocyte Sedimentation Rate Combined With the Probe-to-Bone Test for Fast and Early Diagnosis of Diabetic Foot Osteomyelitis. The International Journal of Lower Extremity Wounds. 2021;20(3):227-231. doi: 10.1177/1534734620923278
15. 15 Harrison HSG, Peters EJG, Maas M, Lavery LA, Beekman KM. Plain Radiography for Diagnosing and Monitoring Foot Osteomyelitis in Persons With Diabetes: Accuracy, Limitations and Clinical Utility. Wound Repair Regeneration. 2026;34(1):e70128. doi: 10.1111/wrr.70128
16. 16 Mei S, Chen H, Dai J. Is procalcitonin superior to CRP and ESR in the diagnosis of diabetic foot osteomyelitis? A systematic review and meta-analysis. BMC Infect Dis. 2026;26(1):796. doi: 10.1186/s12879-026-13071-6
17. 17 Liu L, Hou J, Tao Y, Zhou G, Xiao L, Li J, et al. Diagnostic accuracy of C-reactive protein for diabetic foot osteomyelitis: a retrospective study and meta-analysis of composite cut-off values. J Wound Care. 2026;35(Sup4a):S28-S36. doi: 10.12968/jowc.2025.0340
18. 18 Öz OK, Sherwood AM, Crisologo PA, Killeen AL, Bhavan K, La Fontaine J, et al. Prospective comparison of WBC SPECT/CT and conventional MRI for diagnosing osteomyelitis and monitoring treatment response in diabetic foot infections. Diabetologia. 2026;69(4):1049-1059. doi: 10.1007/s00125-025-06652-8
19. 19 Lagrand RS, Sabelis LWE, Gramberg MCTT, Ahmad M, van den Bosch AJF, Brekelmans W, et al. Bone Biopsy not Superior to Ulcer Bed Biopsy–guided Antibacterial Therapy on Remission of Diabetic Foot Osteomyelitis: A Randomized Controlled Trial. Clin Infect Dis. 2026;82(3):436-442. doi: 10.1093/cid/ciaf684
20. 20 Schechter MC, Ali MK, Risk BB, Singer AD, Santamarina G, Rogers HK, et al. Percutaneous Bone Biopsy for Diabetic Foot Osteomyelitis: A Systematic Review and Meta-Analysis. Open Forum Infectious Diseases. 2020;7(10):ofaa393. doi: 10.1093/ofid/ofaa393
21. 21 Kadiri L, Maher A, Bridgen A. Concordance of Microbiological and Histopathological Examination of Bone Biopsies in Diagnosing Diabetic Foot Osteomyelitis, a Systematic Review. Journal of Foot and Ankle Research. 2026;19(4):e70223. doi: 10.1002/jfa2.70223
22. 22 Lavery LA, Tarricone AN, Madsen AWE, Thirugnanasambandam N, Gee A, Wukich DK, et al. Does Antibiotic Treatment Before Bone Biopsy Affects the Identification of Bacterial Pathogens From Bone Culture in Diabetic Foot Osteomyelitis? A Systematic Review and Meta Analysis. Wound Repair Regeneration. 2026;34(3):e70171. doi: 10.1111/wrr.70171
23. 23 Ulusoy S, Kılınç İ, Coşkun B, Ayhan M. Are Deep Tissue Cultures a Reliable Alternative to Bone Biopsy for Diagnosing Diabetic Foot Osteomyelitis? A Comparative Diagnostic Study. Diagnostics. 2025;15(7):880. doi: 10.3390/diagnostics15070880
24. 24 Gramberg MCTT, Knippers C, Lagrand RS, van Hattem JM, de Goffau MC, Budding Budding AE, et al. Concordance between culture, Molecular Culture and Illumina 16S rRNA gene amplicon sequencing of bone and ulcer bed biopsies in people with diabetic foot osteomyelitis. BMC Infect Dis. 2023;23(1):505. doi: 10.1186/s12879-023-08472-w
25. 25 Morsli M, Magnan C, Salipante F, Dubois A, Schuldiner S, Cellier N, et al. Enhancing diabetic foot osteomyelitis diagnosis with metagenomics next‐generation sequencing, proof of concept. Diabet Med. 2026;43(4):e70235. doi: 10.1111/dme.70235
26. 26 Díaz-Velis L, Salvador-Sagüez F, Roach F, Mancilla E, Campos MA, Ruiz-Gil T, et al. Metagenomic and ribosomal transcript profiles of diabetic foot osteomyelitis in Hispanic patients: underestimated bacteria in biofilm persistence. Front Cell Infect Microbiol. 2026;15:1729196. doi: 10.3389/fcimb.2025.1729196
27. 27 Kulasegaran N, Vangaveti V, Norton R, Malabu U. The Microbial Diversity and Antimicrobial Susceptibility Profile Underlying Diabetic Foot Osteomyelitis: A Retrospective Study Conducted in North Queensland, Australia. Foot & Ankle Orthopaedics. 2024;9(3):24730114241281503. doi: 10.1177/24730114241281503
28. 28 Shin CS, Kim DW, Kim JK, Kim TH. Seasonality of Bacterial Strains in Diabetic Foot Osteomyelitis: Implications for Empiric Antibiotic Therapy in a Temperate Region with Distinct Seasons. JCM. 2026;15(5):2064. doi: 10.3390/jcm15052064
29. 29 Ren J, Gao H, Luo W, Lu S, Fu C, Wang H, et al. Risk Factors, Microbiology, and Prognosis of Diabetic Foot Osteomyelitis: A Retrospective Cohort Study. Endocrine Practice. 2025;31(5):578-584. doi: 10.1016/j.eprac.2025.01.007
30. 30 Bao Y, Shi Y, Zhu C, Liang Z, Xu Q, Zhang Z. Microbiological characteristics and risk factors for multidrug-resistant organism infections in diabetic foot ulcer. J Wound Care. 2026;35(1):59-67. doi: 10.12968/jowc.2023.0041
31. 31 Jensen LK, Birch JM, Jensen HE, Kirketerp‐Møller K, Gottlieb H. Bacterial invasion of the submicron osteocyte lacuna–canaliculi network ( OLCN ): a part of osteomyelitis disease biology. APMIS. 2023;131(7):325-332. doi: 10.1111/apm.13312
32. 32 Krag LB, Peterlin AAN, Gleipner-Andersen E, Gottlieb H. Impact of Narrow Empiric Antibiotic Spectrum and Patient Characteristics on Clinical Outcomes in Bone and Joint Infections: A Retrospective Cohort Study. Antibiotics. 2026;15(6):620. doi: 10.3390/antibiotics15060620
33. 33 Motaganahalli S, Reynolds G, Vogrin S, Haikerwal S, Richards T, Peel T, et al. Impact of antibiotic duration in management of diabetic foot osteomyelitis: A systematic review and meta-analysis. CMI Communications. 2025;2(3):105095. doi: 10.1016/j.cmicom.2025.105095
34. 34 Ghosh A, Rastogi A, Sahana PK, Agrawal AC, Mishra AK, Mohanty D, et al. Four-Week vs Six-Week Antibiotic Therapy in the Management of Nonsurgically Treated Diabetic Foot Osteomyelitis: Protocol for a Multicentric, Single-Blind Randomized Clinical Trial. JMIR Res Protoc. 2026;15:e93492. doi: 10.2196/93492
35. 35 Petithomme-Nanrocki M, Slitine I, Diallo S, Crouzet M, Mostaert M, Moysset P, et al. Three versus six weeks of post-amputation antibiotic therapy in diabetic forefoot osteomyelitis with positive culture for residual infected bone. Infectious Diseases Now. 2024;54(7):104975. doi: 10.1016/j.idnow.2024.104975
36. 36 Tkatch C, Bair S, Hale AJ. Comparison of Oral Versus Intravenous Antibiotics for the Treatment of Diabetic Foot Osteomyelitis: A Propensity-Matched, Retrospective Cohort Study. JAPMA. 2026;116(4):47. doi: 10.3390/japma116040047
37. 37 Sood S, Geng R, Waked J, Mufti A, Sriranganathan A, Sibbald RG. Use of Oral Antibiotics for Diabetic Foot Osteomyelitis: A Systematic Review. Advances in Skin & Wound Care. 2025;38(10):519-526. doi: 10.1097/asw.0000000000000343
38. 38 Meddas N, Gachet B, Piraux A, Senneville E, Drudi LM, Brousseau-Foley M, et al. Perspectives on antibiotic management of diabetic foot osteomyelitis: A scoping review on routes of administration. Diabetes Research and Clinical Practice. 2026;231:113035. doi: 10.1016/j.diabres.2025.113035
39. 39 Rich D, Daneman, MD, MSc N, Lam MD, MSc PW. Trends in oral antibiotic treatment in patients hospitalized with diabetic foot osteomyelitis: A single-centre experience. Journal of the Association of Medical Microbiology and Infectious Disease Canada. 2025;10(2):179-185. doi: 10.3138/jammi-2024-0042
40. 40 Acbo AM, Koura-Mola N, McSweeney T, Bao H, Chang M, Cowman K, et al. Healing soles: a microbiology-driven electronic health record-algorithm and order set to decrease antipseudomonal use in diabetic foot infections, a retrospective, observational, quasi-experimental study. ASHE. 2025;5(1):e89. doi: 10.1017/ash.2025.59
41. 41 Senneville E, Gachet B, Blondiaux N, Robineau O. Do Anti-Biofilm Antibiotics Have a Place in the Treatment of Diabetic Foot Osteomyelitis?. Antibiotics. 2023;12(2):317. doi: 10.3390/antibiotics12020317
42. 42 Bessesen MT, Doros G, Henrie AM, Harrington KM, Hermos JA, Bonomo RA, et al. A multicenter randomized placebo controlled trial of rifampin to reduce pedal amputations for osteomyelitis in veterans with diabetes (VA INTREPID). BMC Infect Dis. 2020;20(1):23. doi: 10.1186/s12879-019-4751-3
43. 43 Núñez LM, Benito AR, Bereciartua E, Morata L, Escudero-Sánchez R, Nadal JS, et al. Real-World Use of Dalbavancin in Diabetic Foot Osteomyelitis: PEDDAL Study. JCM. 2025;14(19):6705. doi: 10.3390/jcm14196705
44. 44 Gloor I, Faulhaber S, Schöni M, Waibel FWA, Aragón-Sánchez J, Lipsky BA, et al. Rationales for Antibiotic Treatment Versus Direct Surgical Resection in Various Strata of Diabetic Foot Osteomyelitis—Literature Review and Expert Opinions. Antibiotics. 2026;15(8):797. doi: 10.3390/antibiotics15080797
45. 45 Lavery LA, Tarricone AN, Reyes MC, Suludere MA, Sideman MJ, Siah MC, et al. Does complete resection of infected bone improve clinical outcomes in patients with diabetic foot osteomyelitis?. International Wound Journal. 2024;21(10):e70072. doi: 10.1111/iwj.70072
46. 46 Aragón‐Sánchez J, Víquez‐Molina G, López‐Valverde ME, Rojas‐Bonilla JM, Segura‐Retana E. Residual osteomyelitis at the resection margin after conservative surgery is not associated with the recurrence of diabetic foot infection and may successfully be treated without postoperative antibiotic therapy. Diabet Med. 2023;40(10):e15162. doi: 10.1111/dme.15162
47. 47 Aragón-Sánchez J, Víquez-Molina G, López-Valverde ME. Controversial Issues Regarding Positive Bone Margins in Surgery for Diabetic Foot Osteomyelitis: A Pilot Study. The International Journal of Lower Extremity Wounds. 2024;23(1):109-115. doi: 10.1177/15347346211041267
48. 48 Monami M, Bordoni L, Ragghianti B, Silverii GA, Mannucci E. Efficacy and safety of a bio ‐absorbable antibiotic delivery in calcium sulphate granules for the treatment of osteomyelitis in patients with diabetic foot: A randomized, double blinded, controlled clinical study The BIG D‐ FOOT study. Diabetes Obesity Metabolism. 2025;27(5):2552-2560. doi: 10.1111/dom.16254
49. 49 Monami M, Ragghianti B, Masulli M, Lipsky BA. Antibiotic-loaded bone substitutes in the surgical management of diabetic foot osteomyelitis: Current evidence and clinical perspectives. Diabetes Research and Clinical Practice. 2026;237:113294. doi: 10.1016/j.diabres.2026.113294
50. 50 Yammine K, Mansour J, Otayek J, Alqaysi B, Daher J, Assi C. The Use of Antibiotic-Impregnated Cement for Infection or Mechanical Support in Diabetic Foot Osteomyelitis: A Systematic Review and Meta-Analysis. Foot & Ankle Specialist. 2026;19(3):309-316. doi: 10.1177/19386400241274624
51. 51 Shen D, Huang K, Guo Q, Ma G, Ding L. The Efficacy of Local Antibiotic Delivery Systems Therapy in the Management of Diabetic Foot Osteomyelitis: A Systematic Review and Meta-Analysis. The International Journal of Lower Extremity Wounds. 2026;25(2):379-388. doi: 10.1177/15347346241266062
52. 52 Soldevila-Boixader L, Fernández AP, Laguna JM, Uçkay I. Local Antibiotics in the Treatment of Diabetic Foot Infections: A Narrative Review. Antibiotics. 2023;12(1):124. doi: 10.3390/antibiotics12010124
53. 53 Da Ros R, Barro E, Michelli A, Assaloni R, Brunato B, Nardi M, et al. Diabetic Calcaneal Osteomyelitis: Outcomes Before and After the Introduction of Bone Graft Substitutes—A Retrospective Cohort Study. Diabetology. 2026;7(9):177. doi: 10.3390/diabetology7090177
54. 54 Lavery LA, Reyes MC, Najafi B, Coye TL, Sideman M, Siah MC, et al. The infected diabetic foot: Risk factors for re‐infection after treatment for diabetic foot osteomyelitis. Wound Repair Regeneration. 2025;33(1):e13246. doi: 10.1111/wrr.13246
55. 55 Saito AK, Goetz MB, Wu S. When Can a MRSA Nares Swab Guide Antibiotic Stewardship?. JAMA Intern Med. 2025;185(2):230. doi: 10.1001/jamainternmed.2024.6436
56. 56 Cortés-Penfield N, Beekmann SE, Polgreen PM, Ryan K, Marschall J, Sekar P. Variation in North American Infectious Disease Specialists' Practice Regarding Oral and Suppressive Antibiotics for Adult Osteoarticular Infections: Results of an Emerging Infections Network (EIN) Survey. Open Forum Infectious Diseases. 2024;11(6):ofae280. doi: 10.1093/ofid/ofae280
57. 57 Rashki S, Haddadi MH, Marzhoseyni Z, Khaledi M, Yekani M, Shooraj M, et al. Challenges and Solutions in Advanced Management of Diabetic Foot Infections: A Review of Recent Studies. Journal of Diabetes Research. 2025;2025(1):6715157. doi: 10.1155/jdr/6715157
58. 58 Tone A, Nguyen S, Devemy F, Topolinski H, Valette M, Cazaubiel M, et al. Six-Week Versus Twelve-Week Antibiotic Therapy for Nonsurgically Treated Diabetic Foot Osteomyelitis: A Multicenter Open-Label Controlled Randomized Study. Diabetes Care. 2015;38(2):302-307. doi: 10.2337/dc14-1514
59. 59 Aicale R, Cipollaro L, Esposito S, Maffulli N. An evidence based narrative review on treatment of diabetic foot osteomyelitis. The Surgeon. 2020;18(5):311-320. doi: 10.1016/j.surge.2020.01.007
60. 60 Castellino LM, Crisologo PA, Chhabra A, Öz OK. Diabetic Foot Infections. Infectious Disease Clinics of North America. 2025;39(3):465-482. doi: 10.1016/j.idc.2025.02.014
61. 61 Lázaro-Martínez JL, Aragón-Sánchez J, García-Morales E. Antibiotics Versus Conservative Surgery for Treating Diabetic Foot Osteomyelitis: A Randomized Comparative Trial. Diabetes Care. 2014;37(3):789-795. doi: 10.2337/dc13-1526
62. 62 Aragón-Sánchez J, Víquez-Molina G, Uçkay I, Rojas-Bonilla JM, Lipsky BA. A research-based, current approach to diabetes-related acute foot infections and chronic osteomyelitis. Expert Review of Endocrinology & Metabolism. 2025;20(6):481-495. doi: 10.1080/17446651.2025.2474110