Bacterial profile and antimicrobial resistance patterns of infected diabetic foot ulcers in low- and middle-income countries: a systematic review and meta‑analysis

Abstract

Background:

Diabetic foot ulcers (DFUs) are often polymicrobial infections, necessitating antibiotic therapy. However, the high prevalence of antibiotic misuse and the growing issue of antibiotic resistance in low- and middle-income countries (LMICs) exacerbate the challenges in managing these ulcers, contributing to the risk of lower-limb amputations.

Objectives:

This systematic review reported the bacterial profile and antibiotic resistance patterns associated with DFUs in LMICs.

Design:

A systematic review and meta-analysis.

Data sources and methods:

A comprehensive search was conducted in PubMed, SCOPUS, and Web of Science to retrieve relevant articles, which were screened according to predefined selection criteria. Publication bias was assessed with a funnel plot and Egger’s regression test. A meta-analysis was conducted in RStudio version 4.5.2, using the random-effects model. Meta-regression and sensitivity analysis were conducted to assess the sources of heterogeneity and robustness of the included studies, respectively.

Results:

The most frequently isolated bacteria were Staphylococcus aureus (22.64%), Escherichia coli (12.37%), Pseudomonas aeruginosa (8.49%), Klebsiella pneumoniae (5.19%), P. mirabilis (4.61%), and Coagulase-negative Staphylococci (4.86%). The prevalence of key antibiotic-resistant bacteria was as follows: Methicillin-resistant S. aureus (38.65%), Extended-spectrum beta-lactamase (ESBL)-producing E. coli (50.74%), and ESBL-producing K. pneumoniae (41.54%). S. aureus demonstrated the highest resistance to penicillin (96.87% (93.30; 99.32)), azithromycin (76.42% (64.54; 86.65)), and cefoxitin (68.19% (46.83; 86.67)). E. coli showed the highest resistance to ampicillin (86.60% (77.14; 94.18)) and cotrimoxazole (69.27% (57.35; 80.18)).

Conclusion:

Pathogenic bacteria isolated from DFUs in LMICs exhibit high rates of antibiotic and multidrug resistance. Alternative therapies, such as phage therapy, should be explored as potential treatments for DFUs to mitigate the impact of antibiotic resistance and decrease the rate of diabetes-related amputations.

Trial registration:

Open Science Framework (OSF) registry, available at https://doi.org/10.17605/OSF.IO/DPX9F.

Introduction

Diabetes mellitus ranks among the top four chronic diseases globally and is deemed one of the fastest-growing public health challenges.^1,2 Since 1980, the number of individuals living with diabetes has quadrupled, with approximately 80% residing in low- and middle-income countries (LMICs). ³ The International Diabetes Federation (IDF) reports that four out of every five adults with diabetes live in LMICs, underscoring the disproportionate burden on these regions. ⁴ Diagnosis of diabetes in LMICs is usually made late, sometimes with complications already manifesting, making management more challenging. ⁵ Furthermore, diabetes management in LMICs is compounded by under-resourced health facilities, limited availability of medications, and high costs involved in the proper management of the condition. ⁶

Notable diabetic complications, including diabetic neuropathy and peripheral vascular disorders, have been identified as the major risks to the development of diabetic foot ulcers (DFUs), another complication associated with uncontrolled or poorly managed diabetes.^7,8 About 50% of patients with diabetes have diabetic peripheral neuropathy (DPN), a complicated condition with several etiologies. Foot infections, such as ulcers that usually result in amputations, are common in long-term peripheral neuropathy. ⁹ The progression of DPN, for which there are currently no effective treatment options, can be slowed in type 1 diabetes through good glycemic management. ¹⁰ However, this approach is less effective in type 2 diabetes, ¹¹ suggesting that the development of diabetic neuropathy is ultimately unavoidable.

Amputation can often be avoided with proper management of DFUs, which includes the use of antibiotics, as bacterial infections contribute to the progression of these ulcers. ¹² However, antimicrobial resistance poses a significant challenge to timely healing, even in well-resourced health systems like that of the United States, where non-healing diabetic foot wounds result in more than 100,000 amputations annually, with DFUs being the primary cause in 60% of these cases. ¹³ Notably, individuals with poor glycemic management face a 26% higher risk of amputation compared to those who maintain good glycemic control. ³ The situation is even more difficult in LMICs, which bear the majority of diabetes-related lower-limb amputations, a significantly higher prevalence of poor glycemic control, ¹⁴ and a higher burden of antibiotic resistance. ¹⁵

The existing challenges to diabetes management in LMICs underscore the urgent need to assess bacterial profiles and antibiotic resistance patterns to inform effective policy planning and implementation for diabetes care. Previous systematic reviews have focused on regions such as Africa and Sub-Saharan Africa (SSA). For instance, a review of the bacterial profile and antibiotic resistance of DFUs in Africa reported on only Staphylococcus aureus and Pseudomonas aeruginosa isolates, with prevalence of 19.9% and 11.8%, respectively. ¹⁶ A similar review on SSA identified S. aureus (34.34%), Escherichia coli (21.16%), and P. aeruginosa (20.98%) as predominant isolates, with these strains exhibiting resistance rates exceeding 50% for commonly used antibiotics. ¹⁷ Coverage of other regions, such as South Asia, with economic challenges similar to Africa, is lacking, limiting a comprehensive assessment of DFU management and outcomes in LMICs. To the best of our knowledge, no systematic review and meta-analysis has comprehensively assessed the bacterial profile and antibiotic resistance patterns of DFUs in LMICs. This systematic review addresses this gap by synthesizing data on bacterial isolates from DFUs and antibiotic resistance from studies conducted in LMICs. The findings of this study will inform rational antibiotic use in the care of DFUs and ultimately improve care outcomes.

Methodology

Search strategy

This study adhered to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. ¹⁸ The review was registered in the Open Science Framework (OSF) registry, available at https://doi.org/10.17605/OSF.IO/DPX9F. Thorough searches were conducted in PubMed, SCOPUS, and the Web of Science from 2nd to 8th December 2024 and updated to include studies published from 9th December 2024 to 15th December 2025. The study included countries in low-income, lower-middle-income, and upper-middle-income economies, as defined by the World Bank Data. ¹⁹

The following search string was used in the databases:

(“Antimicrobial resistance” OR “AMR” OR “Antibiotic resistance” OR “Bacterial resistance”) AND (“Diabetes” OR “Diabetes mellitus” OR “Type 1 diabetes mellitus” OR “T1DM” OR “Type 2 diabetes mellitus” OR “T2DM”) AND (“Diabetic foot ulcer” OR “Diabetic foot”) AND (“Afghanistan” OR “North Korea” OR “South Sudan” OR “Burkina Faso” OR “Liberia” OR “Sudan” OR “Burundi” OR “Madagascar” OR “Syria” OR “Central African Republic” OR “Malawi” OR “Togo” OR “Chad” OR “Mali” OR “Uganda” OR “Democratic Republic of Congo” OR “DRC” OR “Mozambique” OR “Yemen” OR “Eritrea” OR “Niger” OR “Ethiopia” OR “Rwanda” OR “Gambia” OR “Sierra Leone” OR “Guinea-Bissau” OR “Somalia” OR “Angola” OR “Bangladesh” OR “Benin” OR “Bhutan” OR “Bolivia” OR “Cabo Verde” OR “Cambodia” OR “Cameroon” OR “Comoros” OR “Congo” OR “Côte d'Ivoire” OR “Ivory Coast” OR “Djibouti” OR “Egypt” OR “Eswatini” OR “Ghana” OR “Guinea” OR “Haiti” OR “Honduras” OR “India” OR “Jordan” OR “Kenya” OR “Kiribati” OR “Kyrgyz Republic” OR “Lao” OR “Lebanon” OR “Lesotho” OR “Mauritania” OR “Micronesia” OR “Morocco” OR “Myanmar” OR “Nepal” OR “Nicaragua” OR “Nigeria” OR “Pakistan” OR “Papua New Guinea” OR “Philippines” OR “Samoa” OR “São Tomé and Principe” OR “Senegal” OR “Solomon Islands” OR “Sri Lanka” OR “Tajikistan” OR “Tanzania” OR “Timor-Leste” OR “Tunisia” OR “Uzbekistan” OR “Vanuatu” OR “Vietnam” OR “West Bank and Gaza” OR “Zambia” OR “Zimbabwe” OR “Albania” OR “Equatorial Guinea” OR “Moldova” OR “Algeria” OR “Fiji” OR “Mongolia” OR “Argentina” OR “Gabon” OR “Montenegro” OR “Armenia” OR “Georgia” OR “Namibia” OR “Azerbaijan” OR “Grenada” OR “North Macedonia” OR “Belarus” OR “Guatemala” OR “Paraguay” OR “Belize” OR “Indonesia” OR “Peru” OR “Bosnia and Herzegovina” OR “Iran” OR “Serbia” OR “Botswana” OR “Iraq” OR “South Africa” OR “Brazil” OR “Jamaica” OR “St. Lucia” OR “China” OR “Kazakhstan” OR “St. Vincent and the Grenadines” OR “Colombia” OR “Kosovo” OR “Suriname” OR “Costa Rica” OR “Libya” OR “Thailand” OR “Cuba” OR “Malaysia” OR “Tonga” OR “Dominica” OR “Maldives” OR “Türkiye” OR “Dominican Republic” OR “Marshall Islands” OR “Turkmenistan” OR “Ecuador” OR “Mauritius” OR “Tuvalu” OR “El Salvador” OR “Mexico” OR “Ukraine”)

Study selection

Two independent researchers sequentially screened search results by performing title/abstract and full-text screening. A third researcher acted as the arbiter in discrepancies, ensuring a consensus was reached. Studies published in the English language, containing bacterial profiles and antibiotic resistance data of DFUs and conducted in LMICs, were included in our study. Exclusion criteria included systematic reviews, narrative reviews, meta-analyses, case series, and case reports; studies conducted in developed countries; and studies presenting incomplete antibiotic resistance data of DFUs.

Data extraction

Two researchers extracted the data, and a third researcher validated the extracted data. The following relevant characteristics were extracted from the included studies for the systematic review: first author, study location, study period, study design, type of diabetes, age of study participants, number of study participants, Wagner’s classification, type of samples collected, and bacterial identification method used. For the meta-analysis, the frequency of bacteria isolated from DFUs, their antibiotic resistance rates, and prevalence data for multidrug-resistant pathogens were extracted. Multidrug resistance (MDR) was defined as drug resistance to at least three classes of antibiotics, while extensive drug resistance (XDR) was defined as drug resistance to all but two classes of antibiotics.^20,21 MDR and XDR data were extracted as reported despite differences in antimicrobial susceptibility testing methodologies used across the studies.

Quality assessment

A modified Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist was used to evaluate the quality of the included observational studies. ²² Each of the selected 12 items on the checklist was assessed, with a “Yes” assigned if a study provided adequate information in accordance with the item’s recommendations, and a “No” assigned if the information was inadequate or unclear. The twelve items were weighted equally in the assessment. Studies were categorized as High quality if they received “Yes” for 9–12 items, Moderate quality for 5–8 items, and Low quality for 1–4 items (Supplemental Table 5).

Statistical analysis

The meta-analysis was conducted using RStudio version 4.5.2 (Massachusetts, USA) and the meta package, with the data subjected to the random-effects model. The Freeman-Tukey double arcsine transformation was applied to standardize variances across the 42 studies. The pooled resistances were calculated using the restricted maximum-likelihood estimator method (REML). The heterogeneity of the study results was assessed with the I² statistic, where values of 0%, 25%, 50%, and ⩾75% indicated no, low, moderate, and high heterogeneity, respectively. Confidence intervals (95% CI) for the I² statistics were determined using the Jackson method. A funnel plot and Egger’s regression test were utilized to visualize and assess the statistical significance of publication bias. Furthermore, a sensitivity analysis and meta-regression analysis were conducted to identify potential sources of heterogeneity in the meta-analysis. A p-value of less than 0.05 was considered statistically significant.

Results

Search results

A total of 173 search results were obtained from the search. Forty (40) duplicate studies were deleted, and the remaining 133 were screened. Fifty-eight (58) studies were excluded on the grounds of being “false positives” (records not relevant to the review question) or reviews during title and abstract screening, and 75 studies were assessed for eligibility in the full-text screening stage. Forty-four (44) studies were finally included in this study^{23 –66} (Figure 1), and the remaining 31 studies were excluded for various reasons (Supplemental Table 6).

Figure 1.

PRISMA flowchart detailing the screening process.

Description of study characteristics

The 44 included studies comprised data from 5903 diabetes patients with DFUs. They were conducted between 2007 and 2024 across SSA (five countries), North Africa and the Middle East (six countries), South Asia (three countries), East Asia and the Pacific (two countries), and Latin America and the Caribbean (four countries). A total of 22 studies were prospective cross-sectional studies, 11 were retrospective cross-sectional studies, and one was a cohort study. A total of 18 studies included both type 1 and type 2 diabetes patients, while 22 did not specify diabetes type. Participants’ ages ranged from 14 to 92 years. The types of samples included wound swabs (18.2%), tissue biopsies (13.6%), wound swabs/tissue biopsies (18.2%), pus (4.5%), and wound aspirates, either alone or in combination with other sample types. Half (22) of the studies used Wagner’s classification to assess ulcer severity, while 17 (38.6%) studies did not assess wound severity. Of the remaining five studies, two used the University of Texas classification system, two used the University of Texas/Wagner classification system, and one used the University of Texas/Wagner/PEDIS classification system. Bacterial identification methods included biochemical tests (28 studies), VITEK 2 Compact system (seven studies), VITEK-MS mass spectrometer and VITEK2-Compact system (one study), MALDI-TOF (two studies), morphological examination (one study), and biochemical tests/polymerase chain reaction (one study). Thirty-six (36) of our included studies were of high quality, and the remaining eight (8) were of moderate quality (Supplemental Table 1).

For the meta-analysis, 904 antibiotic resistance data points, comprising 623 for Gram-negative organisms and 281 for Gram-positive organisms, were extracted from the included studies. The total number of isolates tested per study ranged from 7 to 590. All isolates were included in the meta-analysis regardless of sample size. Resistance patterns for individual bacterial strains with over one percent prevalence in the bacterial profile were determined: S. aureus (259 resistance data), S. agalactiae (11 resistance data), E. faecalis (11 resistance data), Acinetobacter baumannii (32 resistance data), E. coli (278 resistance data), Klebsiella pneumoniae (132 resistance data), P. aeruginosa (70 resistance data), P. mirabilis (59 resistance data), P. vulgaris (27 resistance data), and Morganella morganii (25 resistance data). However, due to statistical considerations, resistance rates of antibiotics tested on fewer than 30 bacterial isolates were not reported.

Profile of bacterial isolates

A total of 7244 bacterial isolates were reported across the included studies. Gram-negative bacteria (58.43%) were the predominant isolates in DFUs, while Gram-positive bacteria accounted for 41.57%. The most frequently isolated bacterial strains were S. aureus (22.64%), E. coli (12.37%), P. aeruginosa (8.49%), K. pneumoniae (5.19%), P. mirabilis (4.61%), and Coagulase-negative Staphylococci (CoNS; 4.86%) (Figure 2). The prevalence of the remaining bacterial species can be found in Figure 2 and Supplemental Table 2.

Figure 2.

Profile of bacterial isolates.

Profile of resistant bacterial isolates

Among S. aureus isolates (N = 1472), 38.65% were identified as Methicillin-resistant S. aureus (MRSA). Out of 71 Enterococci spp., 8.45% were VRE. The prevalence of carbapenem resistance among E. coli and P. aeruginosa was 17.86% (n/N = 5/28) and 9.09% (n/N = 4/44), respectively. Extended-spectrum beta-lactamases (ESBLs) were identified in 62.16% (n/N = 23/37) Acinetobacter spp., 50.74% (n/N = 138/272) E. coli, 41.45% (n/N = 27/65) K. pneumoniae, and 35.09% (n/N = 20/57 P. aeruginosa (Figure 3).

Figure 3.

Profile of resistant bacterial isolates.

CRE, Carbapenem-resistant Enterobacterales, ESBL, Extended-spectrum beta-lactamase, MRSA, Methicillin-resistant Staphylococcus aureus, VRE, Vancomycin-resistant Enterococci.

Antibiotic resistance of Gram-positive bacterial isolates

Antibiotic resistance estimates in S. aureus (Supplemental Figure 2) ranged from 5.47% (2.25; 9.62) for rifamycin to 96.87% (93.30; 99.32) for penicillin. For S. agalactiae (Supplemental Figure 3), high resistance rates were observed for clindamycin (31.35% (0.78; 75.15)) and tetracycline (66.54% (42.39; 87.34)). E. faecalis (Supplemental Figure 4) had pooled resistance of 1.47% (0.00; 6.21) for linezolid, 22.53% (2.98; 47.97) for vancomycin, and 35.42% (0.00; 99.31) for ampicillin (Table 1).

Table 1.

Pooled antibiotic resistance of predominant Gram-positive bacterial isolates.

Class of antibiotics	S. aureus		S. agalactiae		E. faecalis
	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)
Ampicillin	—	—	—	—	3	35.42 (0.00; 99.31)
Azithromycin	3	76.42 (64.54; 86.65)	—	—	—	—
Cefoxitin	15	68.19 (46.83; 86.67)	—	—	—	—
Ceftaroline	1	44.44 (28.46; 61.01)	—	—	—	—
Ciprofloxacin	26	42.94 (33.35; 52.76)	—	—	—	—
Clindamycin	31	39.37 (30.63; 48.43)	4	31.35 (0.78; 75.15)	—	—
Cotrimoxazole	22	32.36 (23.26; 42.08)	—	—	—	—
Erythromycin	34	52.63 (45.05; 60.16)	3	28.09 (0.00; 78.99)	—	—
Gentamicin	35	27.28 (20.30; 34.79)	—	—	—	—
Levofloxacin	14	39.65 (25.92; 54.11)	1	36.36 (10.00; 67.31)	—	—
Linezolid	5	8.50 (0.16; 23.44)	—	—	1	1.47 (0.00; 6.21)
Penicillin	23	96.87 (93.30; 99.32)	—	—	4	16.70 (0.00; 51.36)
Rifampicin	14	5.47 (2.25; 9.62)	—	—	—	—
Teicoplanin	2	15.35 (7.94; 24.46)	—	—	—	—
Tetracycline	24	41.87 (31.15; 52.96)	3	66.54 (42.39; 87.34)	—	—
Vancomycin	10	28.48 (13.71; 45.84)	—	—	3	23.53 (0.00; 75.95)

CI, confidence interval; NS, number of studies; R, resistance.

Antibiotic resistance of Gram-negative bacterial isolates

A. baumannii had the highest pooled resistance to both cefepime (91.30% (63.96; 100.00)) and ceftazidime (91.30% (63.96; 100.00)), while maintaining lower resistance to polymyxin B (19.44% (7.90; 34.19)). For E. coli, the highest resistance rates were recorded for cefuroxime (68.98% (49.76; 85.62)), cotrimoxazole (69.27% (57.35; 80.18)), and ampicillin (86.60% (77.14; 94.18)). Pooled resistance in K. pneumoniae ranged from 25.62% (7.77; 47.82) for ertapenem to 85.37% (72.66; 94.80) for colistin. P. aeruginosa exhibited a high resistance to aztreonam (52.63% (30.66; 74.12)), cefepime (55.08% (30.14; 78.87)), and meropenem (61.34% (27.66; 90.51)). P. mirabilis had pooled resistance ranging from 3.39% (0.00; 13.02) for meropenem to 41.85% (1.92; 89.84) for cefuroxime. Similarly, resistance rates in P. vulgaris ranged from 8.67% (1.25; 20.08) for tobramycin to 45.53% (29.94; 61.47) for ciprofloxacin. M. morganii had resistance of 61.57% (38.23; 82.93) to ciprofloxacin (Table 2; Supplemental Figures 5–11).

Table 2.

Pooled antibiotic resistance of predominant Gram-negative bacterial isolates.

Class of Antibiotics	A. baumannii		E. coli		K. pneumoniae		P. aeruginosa		P. mirabilis		P. vulgaris		M. morganii
	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)	NS	R (%) (95% CI)
Amikacin	2	87.91 (75.96; 96.60)	17	24.32 (14.08; 36.04)	11	44.32 (22.84; 66.69)	—	—	6	23.93 (5.55; 47.92)	2	25.58 (0.00; 93.67)	—	—
Amoxicillin-clavulanic acid	—	—	13	61.46 (47.47; 74.66)	8	61.83 (27.92; 91.43)	—	—	—	—	—	—	—	—
Ampicillin	—	—	20	86.60 (77.14; 94.18)	—	—	—	—	1	17.65 (6.37; 32.52)	—	—	—	—
Ampicillin-sulbactam	—	—	9	65.23 (53.13; 76.55)	5	47.22 (17.15; 78.22)	—	—	2	23.34 (11.64; 37.24)	—	—	—	—
Aztreonam	—	—	7	60.84 (48.56; 72.54)	4	54.34 (10.76; 94.68)	3	52.63 (30.66; 74.12)	—	—	—	—	—	—
Cefepime	2	91.30 (63.96; 100.00)	18	52.13 (39.06; 65.08)	8	60.99 (49.39; 72.10)	11	55.08 (30.14; 78.87)	4	28.92 (5.46; 59.34)	—	—	—	—
Cefotaxime	—	—	14	60.84 (48.56; 72.54)	5	56.99 (36.11; 76.87)	—	—	3	27.31 (1.85; 64.68)	2	42.79 (27.71; 58.52)	—	—
Cefoxitin	—	—	6	45.40 (22.62; 69.12)	—	—	—	—	—	—	—	—	—	—
Ceftazidime	2	91.30 (63.96; 100.00)	16	50.99 (34.62; 67.26)	8	54.55 (28.80; 79.27)	11	43.96 (20.97; 68.23)	4	30.67 (6.10; 61.90)	—	—	—	—
Ceftriaxone	—	—	20	54.64 (42.48; 66.56)	9	55.27 (34.53; 75.28)	—	—	5	30.80 (10.30; 55.42)	—	—		—
Cefuroxime	—	—	11	68.98 (49.76; 85.62)	8	53.34 (30.08; 75.98)	—	—	3	41.85 (1.92; 89.84)	—	—	—	—
Chloramphenicol	—	—	7	47.23 (23.81; 71.26)	—	—	—	—	—	—	—	—	—	—
Ciprofloxacin	—	—	22	65.54 (55.61; 74.92)	12	58.05 (38.88; 76.31)	12	42.40 (28.36; 56.98)	7	35.92 (20.61; 52.51)	4	45.53 (29.94; 61.47)	4	61.57 (38.23; 82.93)
Colistin	—	—	2	36.78 (0.00; 99.87)	1	85.37 (72.66; 94.80)	5	38.83 (4.99; 80.13)	—	—	—	—	—	—
Cotrimoxazole	—	—	18	69.27 (57.35; 80.18)	8	44.37 (29.47; 59.70)	—	—	5	41.06 (8.10; 77.97)	—	—	—	—
Ertapenem	—	—	4	18.10 (0.01; 51.52)	2	25.62 (7.77; 47.82)	—	—	—	—	—	—	—	—
Gentamicin	4	71.74 (59.71; 82.56)	19	42.23 (33.87; 50.79)	8	50.31 (27.15; 73.42)	—	—	6	35.89 (27.95; 44.19)	—	—	—	—
Imipenem	4	75.13 (60.50; 87.80)	12	22.97 (11.29; 36.85)	9	28.61 (9.60; 51.39)	13	39.13 (19.42; 60.55)	3	5.78 (0.00; 21.47)	—	—	—	—
Levofloxacin	3	81.30 (34.38; 100.00)	9	67.46 (56.25; 77.87)	5	52.98 (17.94; 86.68)	6	50.60 (40.97; 60.20)	—	—	—	—	—	—
Meropenem	—	—	9	20.94 (9.62; 34.71)	6	26.39 (16.41; 37.41)	5	61.34 (27.66; 90.51)	2	3.39 (0.00; 13.02)	—	—	—	—
Piperacillin-tazobactam	3	83.79 (56.72; 99.72)	15	25.19 (13.91; 38.21)	9	28.52 (11.72; 48.23)	—	—	4	15.63 (0.06; 42.92)	—	—	—	—
Polymyxin B	1	19.44 (7.90; 34.19)	—	—	—	—	2	28.44 (7.80; 55.11)	—	—	—	—	—	—
Tetracycline	—	—	6	65.75 (57.48; 73.61)	—	—	—	—	—	—	—	—	—	—
Tobramycin	2	45.30 (22.67; 68.84)	4	50.41 (34.03; 66.76)	—	—	2	40.46 (30.56; 50.74)	1	39.77 (29.75; 50.24)	2	8.67 (1.25; 20.08)	—	—

CI, confidence interval; NS, number of studies; R, resistance.

Prevalence of multidrug resistance among bacterial isolates

For Gram-positive bacteria, the prevalence of MDR for S. aureus and E. faecalis was 89.73% (32.10; 100.00) and 69.01% (46.24; 87.92), respectively. Among Gram-negatives, the highest prevalence of MDR was observed in E. coli (76.52% (55.12; 93.03)), followed by P. aeruginosa (55.84% (41.44; 69.86)) and K. pneumoniae (36.93% (27.85; 46.44); Table 3; Supplemental Figure 12). Few. E. coli and K. pneumoniae isolates exhibited XDR (Supplemental Figure 13).

Table 3.

Prevalence of multidrug resistance among bacterial isolates.

Bacterial isolates	Number of studies	MDR (%) (95% CI)
E. faecalis	2	89.73 (32.10; 100.00)
S. aureus	8	69.01 (46.24; 87.92)
E. coli	7	76.52 (55.12; 93.03)
K. pneumoniae	3	36.93 (27.85; 46.44)
P. aeruginosa	3	55.84 (41.44; 69.86)
P. mirabilis	3	68.44 (0.12; 100.00)
P. vulgaris	1	36.46 (13.47; 66.73)

MDR, Multidrug resistance.

Meta-regression

Three covariates, namely, bacterial isolates, Gram stain, and antibiotics, were included in a model to identify sources of heterogeneity in our study. All bacterial isolates and several antibiotics showed significant positive associations (p < 0.05), contributing to the observed heterogeneity in our meta-analysis. However, due to redundancy, the meta-regression model dropped the Gram stain covariate (Supplemental Table 3).

Sensitivity analysis

To evaluate the robustness of our study selection process, all 8 moderate-quality studies^{28,33,43,49,52,56,58,62} were excluded, leaving data from 36 high-quality studies for E. coli resistance data. A comparison of resistance rates for all included studies and the 36 high-quality studies showed that most antibiotics exhibited slight (<5%) variations in resistance rates (Supplemental Table 4). The consistent or slight variations in resistance rates for most antibiotics suggest that excluding moderate-quality studies did not significantly affect the findings. Therefore, we conclude that our selection process was relatively robust.

Heterogeneity and publication bias

A visual inspection of the funnel plot (Supplemental Figure 1) indicated wide variations in sample sizes of the included studies. The Egger’s regression test with a p-value of <0.0001 confirmed a publication bias among our included studies. The estimated antibiotic resistance of Gram-positive and Gram-negative bacterial isolates had significant heterogeneity (H > 1 and I² > 75%).

Discussion

Uncontrolled diabetes predisposes diabetics to several complications, including diabetic neuropathy, which is closely linked to the development of DFUs. ³ While these complications may be prevented with early diagnosis and proper management of diabetes, these practices are often more common in high-income countries compared to LMICs. ³ Diabetic patients in LMICs usually develop complications due to late diagnosis of their chronic condition, coupled with poor medication adherence, among other challenges. ⁶⁷ These challenges are exacerbated by the under-resourced health systems and the escalating global antimicrobial resistance crisis, making the management and treatment of complications such as DFUs particularly difficult. The high rates of antibiotic misuse in LMICs further exacerbate the situation, worsening outcomes for diabetics with foot ulcers, often leading to amputations. ⁶⁸

The poor blood flow to the lower limbs in diabetic patients with neuropathy implies a reduction in the migration of immune cells during localized invasions or infections. ⁶⁹ This makes the affected limbs suitable for microbial colonization and infection. Our study recorded more than 30 species of bacteria isolated from DFUs, with some belonging to the WHO’s critical and high-priority categories of bacteria. These included opportunistic pathogens such as CoNS, as well as ESKAPE pathogens, including E. coli, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and E. faecalis, which were among the top bacterial isolates from DFUs, reflecting the suitability of the diabetic foot environment for bacterial pathogens. The most frequently isolated bacterial strains were Gram-negative (58.43%), which is consistent with the findings of Zubair, ⁷ who reported that Gram-negative bacteria were the most common cause of DFUs. The three most common bacterial species were S. aureus (22.64%), E. coli (12.37%), and P. aeruginosa (8.49%). These findings were mirrored by a systematic review by Wada and colleagues, which reported S. aureus (34.34%), E. coli (21.16%), and P. aeruginosa (20.98%) as the most prevalent bacterial isolates from DFUs in SSA. ¹⁷ The frequent isolation of ESKAPE pathogens found in this study indicates a significant burden of difficult-to-treat infections, further compromising the prognosis of DFUs.

CoNS had a notable prevalence of 4.86% among the diabetic foot ulcer isolates. They are usually normal flora of the skin and hence regarded as concomitants in clinical samples, such as swab samples. ²⁶ With over half (24) of the included studies taking swab samples of DFUs, some of the CoNS isolated would likely be non-pathogenic. A higher prevalence of S. aureus compared to CoNS was also observed in this study. Similarly, a systematic review from Africa reported S. aureus as the predominant DFU pathogen. ¹⁶ An inverse relationship between some CoNS and S. aureus has been reported, with certain CoNS capable of inhibiting S. aureus virulence through and regulatory mechanisms during infections.^26,70

The availability of antibiotics in LMICs due to poor control policies enables the abuse of antibiotics in patients with DFUs, as they tend to abuse antibiotics through self-medication, with the hope of their foot ulcers healing. ⁷¹ The resistance patterns exhibited variations across several antibiotic classes and to varying degrees. Among the critical-priority strains, A. baumannii showed the highest resistance burden, with resistance exceeding 90% to third- and fourth-generation cephalosporins and high resistance to carbapenems. The relatively lower resistance to polymyxin B reflects its more restricted use as a last-line agent in many LMICs.

Regarding high-priority strains, S. aureus exhibited high resistance to commonly prescribed antibiotics in LMICs, including penicillin (96.7%), azithromycin (80.4%), erythromycin (52.4%), and cefoxitin (64.5%). The high cefoxitin resistance observed in this study suggests a substantial burden of MRSA, which is supported by the high prevalence of MRSA among resistant isolates. This aligns with global and regional estimates indicating persistent MRSA circulation in DFUs in LMICs. ¹⁶ A recent global meta-analysis reported an overall MRSA prevalence of approximately 13% in African DFU isolates, with higher rates in hospital-based studies and tertiary care settings. ⁷² These findings underscore a significant challenge to the fight against antimicrobial resistance and diabetic foot care outcomes, with factors such as ease of access to antibiotics and high antibiotic misuse notable in LMICs fueling the observed resistance rates.^71,73,74 Lower resistance rates to rifampicin, linezolid, and teicoplanin observed in this review are consistent with LMIC studies, where these agents are less frequently used due to cost, limited availability, and stewardship restrictions.^75,76 While the pooled estimate for vancomycin resistance in S. aureus reflects the reported data from the included studies, we acknowledge that this rate exceeds typical global surveillance reports. This may be attributed to heterogeneity and variations in susceptibility testing methods across the included studies.

The high pooled resistance to ampicillin observed in E. faecalis in this study should be interpreted with caution. Only three studies reported ampicillin resistance for E. faecalis, with one study ⁶⁶ reporting complete resistance (14/14 isolates), which substantially influenced the pooled estimate. Sensitivity analysis (Supplemental Figure 14) excluding this study reduced the pooled estimate to 4.13%, suggesting that a limited number of heterogeneous studies influenced the initial pooled estimate. This low estimate observed in the sensitivity analysis is more consistent with previously reported rates.^46,47

The high resistance of E. coli and K. pneumoniae to cotrimoxazole, cephalosporins, and fluoroquinolones is consistent with widespread ESBL production reported in DFU isolates across Africa. ¹⁷ In SSA, pooled analyses have shown that Enterobacterales isolated from DFUs frequently exhibit resistance to three or more antibiotic classes, limiting empiric treatment options. ¹⁶ Although carbapenem resistance was lower than cephalosporin resistance, its presence is clinically significant and reflects increasing carbapenem use in tertiary hospitals. ⁷⁷ The high colistin resistance (85.37%) observed in K. pneumoniae aligns with recent reports of plasmid-mediated colistin resistance (mcr genes) emerging in LMICs, raising concerns about the erosion of last-resort therapies. ⁷⁸ However, this notably high colistin resistance was derived from a single included study. ³¹ This limits its reliability and underscores the need for expanded surveillance by future studies to determine if it reflects a broader trend.

These findings indicate a significant challenge in the management of DFUs, particularly in LMICs, due to disparities in antibiotic access ⁷⁹ and diabetes treatment access. ⁸⁰ Other factors, such as peripheral vascular disorders and cigarette smoking, have been reported to be associated with an increased risk of ESKAPE pathogen infections in DFUs. ⁸¹ The high rates of infection and antibiotic resistance observed among ESKAPE pathogens isolated from DFUs support the high rate of diabetic foot amputations, as these pathogens are difficult to treat. ⁸² In addition, the generally high resistance of bacterial pathogens isolated from DFUs suggests an expected increase in diabetes-related amputations in the foreseeable future.

Multidrug resistance was high among isolated species, including E. faecalis (89.73%), E. coli (76.52%), and S. aureus (69.01%). However, these rates were largely influenced by studies with high heterogeneity. Despite this, the high MDR rates are supported by our findings of high prevalence of resistant strains, including ESBL-producing organisms, MRSA, and carbapenem-resistant Enterobacterales, which are known for their significant impact on global healthcare, including extended hospital stay duration, high rates of diabetic foot amputations and mortality, and healthcare costs. ⁸³ However, these high MDR rates could have been influenced by differences in the methodologies used to report antibiotic susceptibility across the included studies. Nonetheless, a mixture of these resistant pathogens in the diabetic foot ulcer environment, coupled with the reduced immunity in diabetics because of the effects of hyperglycemia,^84,85 could drive resistome expansion through the transfer of resistance determinants such as plasmids, transposons, and bacteriophages,^86,87 ultimately leading to the development of multidrug resistance among the ulcer flora.

Cefoxitin, commonly used as a surrogate marker for methicillin resistance, showed a 68.19% resistance rate, whereas the reported MRSA prevalence was 38.65%. This discrepancy may be explained by differences in both the number of studies and the number of isolates contributing to each estimate. Specifically, MRSA prevalence was derived from 33 studies involving 1472 isolates, whereas cefoxitin resistance was reported in 15 studies comprising 365 isolates. Notably, 10 of the 15 studies assessing cefoxitin resistance used it as a surrogate for methicillin resistance. Furthermore, the estimates were generated using different approaches: MRSA prevalence was calculated as a crude proportion, while cefoxitin resistance was pooled using a random-effects meta-analysis. This methodological difference, together with significant heterogeneity (I² = 93.5% (90.8%; 95.4%); p < 0.0001) among the cefoxitin resistance studies, may have further contributed to the observed variation.

Our study had some limitations, the first of which was heterogeneity. This could be attributed to pooling studies from different countries with differences in sampling techniques and laboratory methods, and wide variations in antimicrobial resistance patterns. In assessing the sources of heterogeneity in our meta-analysis, several bacterial strains and antibiotic classes were identified as contributors. In addition, the high variability in the analyzed resistance data could account for the heterogeneity. Second, publication bias was present among the included studies, which may in part reflect our inclusion of only English-language studies. Third, 38.6% of included studies did not report wound severity scores, potentially introducing bias in the severity scores reported. Furthermore, the wide prediction intervals observed in our meta-analysis could be attributed to variation in the number of isolates tested and differences in the microbiological examinations conducted.

Conclusion

DFUs in LMICs are caused by several strains of pathogenic bacteria with high antibiotic resistance. To address this, properly implementing antibiotic stewardship programs in these countries is essential to curb the rise in antibiotic resistance. Foot care education should be encouraged in patients to ensure that they seek help early. While mass screening programs for early diagnosis of diabetes mellitus have been initiated by several LMICs, our general findings could be attributed to the differences in healthcare infrastructure, treatment accessibility, and health financing capabilities between developed and developing countries within these regions. Addressing these challenges at the national level would improve the management of diseases such as diabetes mellitus in LMICs. Furthermore, alternative therapies, such as phage therapy, should be explored as potential treatments for DFUs. In recent times, several studies^{88 –90} have reported the potential utilization of phage therapy to mitigate the impact of antibiotic resistance and decrease the rate of diabetes-related amputations.