Re‐infection after treatment for moderate and severe diabetic foot infections Authors: Lawrence A Lavery, Arthur N Tarricone, Easton C Ryan, Peter A Crisologo, Matthew Malone, Mehmet A Suludere, Lee C Rogers, Dane K Wukich Published: 2024 ABSTRACT AbstractTo investigate risk factors for re‐infection and compare the outcomes in people with diabetic foot infections. A retrospective chart review was conducted, and 294 hospitalised patients with moderate to severe diabetic foot infections (DFIs) were analysed for this study. The diagnosis and classification of the severity of infection was based on the International Working Group on the Diabetic Foot (IWGDF) infection guidelines. Skin and soft tissue infections were diagnosed based on clinical observations as per IWGDF classification in addition to ruling out any suspected osteomyelitis (OM) through negative bone culture, MRI or WBC SPECT CT. OM was confirmed by bone culture or histopathology. Clinical outcomes were based on a 12‐month follow‐up period. All dichotomous outcomes were compared using χ 2 with an alpha of 0.05. The result of this study shows a 48% rate of re‐infection in people admitted to our hospital with moderate and severe diabetic foot infections (DFI). Patients with osteomyelitis present during the index admission were 2.1 times more likely to experience a re‐infection than patients with soft tissue infection (56.7% vs. 38.0% respectively). In the univariate analysis, risk factors for re‐infection included osteomyelitis, non‐healing wounds, prolonged wound healing, antidepressants and leukocytosis. In the regression analysis, the only risk factor for re‐infection was wounds that were not healed >90 days (HR =2.0, CI: 1.5, 2.7, p = 0.001). Re‐infection is very common in patients with moderate and severe diabetic foot infections. Risk factors include osteomyelitis, non‐healing wound, prolonged wound healing, antidepressants and leukocytosis.Keywords: diabetic foot, infection, osteomyelitis, re‐infection, risk factors INTRODUCTION The global prevalence of diabetes is increasing each year. Diabetic foot ulcers (DFUs) have an annual incidence of 2%–7%, 1 and infected DFUs are a major cause of hospitalisation and non‐traumatic lower limb amputation. 2 , 3 , 4 , 5 Among patients with DFUs treated in the outpatient setting, 60% develop infections and 20% of those involve infection of the bone. When infection severity requires hospitalisation, the ratio of soft tissue to bone infection is reversed. About 68% of patients admitted for a foot infection have osteomyelitis (OM) and 32% have soft tissue infections (STI). 6 , 7 Re‐infection varies widely and ranges from 11% to 44%. 8 , 9 , 10 , 11 The purpose of this study was to evaluate the incidence and risk factors for re‐infection in patients admitted to hospital with moderate and severe diabetic foot infections. MATERIALS AND METHODS This study was approved by the Human Subjects Protection Board at the University of Texas Southwestern Medical Center and Parkland Hospital. A retrospective chart review was conducted on the medical records of 353 hospital admitted patients with moderate to severe diabetic foot infections over an 8‐year period. The classification for severity of infection was based on the International Working Group on the Diabetic Foot (IWGDF) infection guidelines. Inclusion criteria for patients consisted of a diagnosis of diabetes mellitus based on American Diabetes Association criteria, 12 age between 18 and 89 years, and hospital admission due to a moderate to severe diabetic foot infection. 13 Baseline measurements of erythrocyte sedimentation rate (ESR) and C‐reactive protein (CRP) collected within 72 h of admission. We used either positive bone culture or bone histopathology to confirm the diagnosis of osteomyelitis. A negative MRI, SPECT CT or bone biopsy were used to rule out bone infection, as false negatives are low. 5 , 14 , 15 Re‐infection was defined as soft tissue or bone infection at the study foot, after the initial infection was resolved during the 12‐month follow‐up period. The patient chart extraction process provided all demographic data including pertinent medical history, social history, wound characteristics, lab results, clinical outcomes and surgical and/or vascular interventions. Medical history included peripheral neuropathy, foot ulceration, peripheral vascular disease and amputation. Sensory neuropathy was defined as abnormal vibration sensation with a 128 Hz tuning fork or abnormal sensation with 10‐gram Semmes‐Weinstein monofilament. Peripheral arterial disease (PAD) was defined as either an ABI <0.9 or >1.3. The presence and severity of infection was based on criteria from the International Working Group on the Diabetic Foot—Diabetic Foot Infection Guidelines. 16 All statistics were computed using SAS 9.4. Categorical variables are presented in this study as frequency (percentage), and continuous variables are reported as means ± standard deviation. Proportional comparisons were made using the Chi‐squared (χ 2) test, and continuous variable comparisons were made using the Mann–Whitney U test. The measure of association between outcomes and covariates was determined with odds ratios (OR) with their 95% confidence intervals (CI) and are presented as OR [confidence interval]. A p‐value ≤0.05 was considered significant for all tests used in this study. RESULTS A total of 353 patients were initially identified, and 294 patients ultimately met inclusion criteria and had at least a 12‐month follow‐up documented in electronic medical records; 137 (46.6%) with STI and 157 (53.4%) with OM patients. When we compared patients that were excluded (EX) because they did not have 12 months of follow‐up and patients that completed the study (COM), patients were less likely to complete a 12‐month follow‐up if they had a STI versus OM (EX 66.0% vs. COM 46.6%, p = 0.02), retinopathy (EX. 8.2% vs. COM 17.4%, p = 0.03), did not require surgery (EX 75.6% vs. COM 24.4%, p < 0.01), did not require an amputation (EX 81.9% vs. COM 18.1%, p < 0.01) and did not require revascularization (EX 86.2% vs. COM 13.8%, p < 0.01). Demographic and patient characteristics are summarised in Table 1, and 12‐month outcomes are summarised in Table 2 and Figure 1. There were no significant differences in age, body mass index (BMI), glycated haemoglobin, PAD, sensory neuropathy or chronic kidney disease (CKD) between patients with and without re‐infection. The overall re‐infection was 48.0%. Complete wound healing was associated with significantly lower re‐infection rates as 116 of 153 (75.8%) who healed their wound did not experience a re‐infection compared 89 of 141 (63.1%) who experienced a re‐infection (OR 0.5, 95% CI 0.3–0.9, p = 0.02). TABLE 1Patients demographics and characteristics in patients with and without re‐infection.No re‐infection (n = 153, 52.0%)Re‐infection (n = 141, 48.0%)OR (95% CI) p‐valueAge53.1 ± 10.152.4 ± 11.633.0 to 69.00.56Age > 706 (3.9)8 (5.7)1.2 (0.4 to 3.7)0.70Body mass Index (BMI) (kg/m2)32.1 ± 9.232.5 ± 10.221.4 to 48.90.73Sex (% male)114 (74.5)107 (75.9)1.1 (0.6 to 1.8)0.78Diabetes type II148 (96.7)132 (93.6)0.50 (0.2 to 1.5)0.22Ulcer history92 (60.1)96 (68.1)1.4 (0.9 to 2.3)0.16Amputation history50 (32.7)54 (38.3)1.2 (0.7 to 1.9)0.54Retinopathy47 (30.7)43 (30.5)1.0 (0.6 to 1.6)1.00PAD109 (71.2)97 (68.8)0.9 (0.5 to 1.5)0.65Sensory neuropathy136 (88.9)131 (92.9)1.6 (0.7 to 3.7)0.23CKD62 (40.5)67 (47.5)1.3 (0.8 to 2.1)0.23Dialysis/ End Stage Renal Disease (ESRD)15 (9.8)13 (9.2)0.9 (0.4 to 2.0)0.86Systemic Inflammatory Response Syndrom(SIRS) criteria57 (37.3)54 (38.3)1.0 (0.7 to 1.7)0.85Osteomyelitis68 (44.4)89 (63.1)2.1 (1.3 to 3.4)<0.01Healed116 (75.8)89 (63.1)0.5 (0.3 to 0.9)0.02Time to heal, days121.2 ± 94.5191.2 ± 113.725.0 to 349.0<0.01Time to heal >90 days55 (35.9)67 (47.5)3.4 (1.8 to 6.2)<0.01Methicillin Resistant Staphlyococcus Aureus (MRSA) in Index6 (3.9)4 (2.8)0.7 (0.2 to 2.6)0.61MedicationsACE inhibitors72 (47.1)68 (48.2)1.1 (0.7 to 1.7)0.84Antidepressant5 (3.3)16 (11.3)3.8 (1.4 to 10.6)<0.01Beta blocker65 (42.4)48 (31.4)0.7 (0.4 to 1.1)0.14Calcium channel blocker50 (32.7)42 (37.6)0.9 (0.5 to 1.4)0.59Insulin123 (80.4)119 (82.3)1.3 (0.7 to 2.4)0.37Statin75 (49.0)70 (49.6)1.0 (0.6 to 1.6)0.92Steroid5 (3.3)5 (3.5)1.1 (0.3 to 3.8)0.90Non Steriod Anti‐Inflammatory Drug (NSAID)106 (69.3)96 (68.1)0.9 (0.6 to 1.6)0.83Laboratory valuesGlycated haemoglobin9.2 ± 2.59.1 ± 2.55.4 to 13.30.61Glycated haemoglobin >1057 (37.3)41 (29.1)0.7 (0.4 to 1.1)0.14Albumin3.3 ± 0.63.4 ± 0.62.4 to 4.30.37Albumin <3.042 (29.8)39 (30.2)1.0 (0.6 to 1.7)0.97White Blood Cell Count (WBC)10.0 ± 3.611.3 ± 4.85.4 to 19.50.01ESR70.6 ± 35.874.6 ± 36.816.0 to 130.00.34CRP6.6 ± 8.58.2 ± 8.40.2 to 28.00.10 Note: mean and standard deviation (SD) presented for continuous variables. Categorical variables presented as N (%). Length of stay, duration of antibiotics and time to heal are expressed in days.Abbreviations: AKA, above knee amputation; BKA, below knee amputation; CKD, chronic kidney disease; CRP, C‐reactive protein; ESR, erythrocyte sedimentation rate; PAD, peripheral arterial disease. Patients demographics and characteristics in patients with and without re‐infection. Note: mean and standard deviation (SD) presented for continuous variables. Categorical variables presented as N (%). Length of stay, duration of antibiotics and time to heal are expressed in days. Abbreviations: AKA, above knee amputation; BKA, below knee amputation; CKD, chronic kidney disease; CRP, C‐reactive protein; ESR, erythrocyte sedimentation rate; PAD, peripheral arterial disease. TABLE 2Outcomes during 12‐month follow‐up in patients with and without re‐infections diabetic foot infections.No Re‐infection (n = 153)Re‐infection (n = 141)OR (95% CI) p‐valueIndex outcomesIndex length of stay11.3 (9.7)10.6 (8.5)(2.0 to 29.0)0.52Duration of antibiotics9.9 (5.3)11.2 (4.3)(1.0 to 17.0)0.03Surgical intervention109 (71.2)123 (87.2)2.8 (1.5 to 5.1)<0.01Number of surgeries1.4 (1.3)3.3 (2.6)(0.0 to 7.0)<0.01AmputationMinor amputation65 (42.5)74 (52.5)1.5 (0.9 to 2.4)0.09Major amputation6 (3.92)22 (15.6)4.5 (1.8 to 11.5)<0.01Revascularization13 (8.5)18 (12.8)1.6 (0.7 to 3.4)0.231‐Year outcomesHealed within 1 year116 (75.8)89 (63.1)0.5 (0.3 to 0.9)0.02Time to heal121.2 (94.5)191.2 (113.7)(25.4 to 349.0)<0.01Foot‐related re‐admission18 (11.8)121 (85.8)45.4 (22.9 to 89.8)<0.01Length of stay, foot related17.5 (19.1)30.7 (24.9)(4.0 to 74.4)<0.01Mortality6 (3.9)3 (2.1)0.5 (0.1 to 2.2)0.37 Note: Mean and standard deviation (SD) presented for continuous variables. Categorical variables presented as N (%). Length of stay, duration of antibiotics and time to heal are expressed in days.Abbreviations: AKA, above knee amputation; BKA, below knee amputation. Outcomes during 12‐month follow‐up in patients with and without re‐infections diabetic foot infections. Note: Mean and standard deviation (SD) presented for continuous variables. Categorical variables presented as N (%). Length of stay, duration of antibiotics and time to heal are expressed in days. Abbreviations: AKA, above knee amputation; BKA, below knee amputation. This bar chart compares the clinical outcomes in patients with and without and re‐infection within the 12‐month study period. Patients with a re‐infection had significantly more surgical intervention, more below knee amputation (BKA) and above knee amputation (AKA), lower wound healing rate, a more foot‐related re‐admission to the hospital. *p < 0.05. The 12‐month outcomes in patients that experienced re‐infection were significantly worse than patients who did not experience re‐infections. Patients with a re‐infection had significantly fewer wounds that healed (63.1% vs. 75.8%, p = 0.02), longer time to heal (191.2 ± 113.7 days vs. 121.2 ± 94.5 days, p = <0.01), more foot‐related re‐admissions (85.8% vs. 11.8%, p = <0.01) and a longer foot‐related hospital length of stay (30.7. ± 24.9 vs. 17.5 ± 19.1 days, p = <0.01). OM, wounds that did not heal, prolonged time to heal, wound duration >90 days, antidepressants and leukocytosis were associated with re‐infection during the follow‐up period in the univariate analysis. Wound healing time of more than 90 days while receiving treatment was associated with more than threefold increased risk of re‐infection in a multiple logistic regression model (OR 3.1, 95% CI 1.7–6.1, p = <0.001) (Table 3) The Kaplan–Meier survival analysis for time to wound healing demonstrated that patients with re‐infection were twice as likely not to heal (Figure 2, HR = 2.0, CI: 1.5, 2.7, p = 0.001) compared to patients without re‐infection. TABLE 3Risk factors associated with re‐infection.Odds ratioStandard deviation95% confidence interval p‐valueTime to heal >90 days3.11.01.76.0<0.01Chronic kidney disease1.50.50.82.80.26Osteomyelitis1.70.60.93.20.08Previous ulcer1.40.40.72.60.35Healed0.30.30.03.00.27ACE inhibitor1.10.40.62.10.73Antidepressant2.81.70.99.40.09Beta blockers0.70.20.31.30.21Calcium channel blocker1.00.40.52.00.98NSAID0.70.20.41.40.31Statin0.70.20.41.30.28Steroid (oral or inhaled)1.41.30.28.60.75 Note: Risk factors in patients treated for moderate and severe diabetic foot infections. Risk factors associated with re‐infection. Note: Risk factors in patients treated for moderate and severe diabetic foot infections. Kaplan–Meier survival for time until wounds healed in patients without re‐infection compared to patients with re‐infection. Y‐axis shows the proportion with open wounds. HR =2.0 [CI:1.5,2.7], (p = 0.001), indicating patients with re‐infection are 2.0 times more likely not to heal compared to patients without re‐infection. DISCUSSION The results of this study demonstrate a 48% rate of re‐infection in people admitted to our hospital with moderate and severe diabetic foot infections (DFI). Patients with OM present during the index admission were 2.1 times more likely to experience a re‐infection than patients with soft tissue infections. In contrast to our results, there is a wide range of re‐infection rates reported by other authors. After treatment for moderate and severe infections, re‐infection ranges from 11% to 44%. 8 , 9 , 10 , 11 For instance, Aragon‐Sanchez and colleagues reported re‐infection of 10.5%, with no difference between patients with soft tissue and bone infections (STI 6.3% vs. OM 14.0%, p = 0.07), although the length of follow‐up was not clear. The low rate of re‐infection in complex wounds is interesting, especially considering that RCTs of Wagner 1 diabetic foot ulcers with highly selective study populations have infection incidence of 14%–36% in 12–16 weeks. 17 , 18 , 19 , 20 We have been unable to identify other studies that evaluate risk factors for re‐infections in diabetic foot infections with soft tissue and bone infections, even though re‐infection is one of the most important underlying factors in re‐admission and amputation. 17 , 18 , 19 , 20 The incidence of re‐infection in the year following the index hospitalisation was high with 38.0% in patients with soft tissue infections and 56.7% % in osteomyelitis patients. Re‐infections were probably the result of a second infection in patients with non‐healing wounds rather than recurrence because of failed treatment of the index infection. Osteomyelitis, wounds that did not heal, longer times to heal, antidepressants and leukocytosis were associated with re‐infection in the univariate analysis. Wounds of >90 days duration were the only significant factor in the regression. We have not been able to identify other studies that identify factors associated with re‐infection. Lavery and colleagues have previously reported risk factors for diabetic foot infections from a cohort of patients treated in an outpatient clinic with DFUs. Risk factors included ulcers with exposed bone (OR 6.7), wound duration >30 days (4.7), recurrent wounds (2.4), wounds with a traumatic aetiology (2.4) and PAD (1.9). 21 Patients with osteomyelitis, non‐healing wounds, prolonged healing times and those on antidepressants are at higher risk for re‐infection and should be closely monitored with long‐term follow‐up and tailored treatment protocols, including more aggressive therapies for bone infections. Multidisciplinary care, incorporating mental health support and addressing health disparities in underserved populations, is critical. Rigorous diagnostic protocols, particularly for soft tissue infections, should be standardised and advanced wound care technologies can help prevent re‐infection in high‐risk patients. Healthcare systems should focus on coordinated care to reduce re‐admissions, with special attention to underserved populations who face barriers to care. Like all studies, this study had weaknesses that merit discussion. Selection bias is likely for several reasons. First, the study population may not be generalizable to the general population because our institution is a safety net hospital that serves a large population of underrepresented minorities who often are unfunded or underfunded. This patient population experiences health disparity has challenges with literacy, transportation and obtaining recommended treatments such as medications, physical therapy and durable medical equipment compared with more affluent populations. Secondly, patients that had satisfactory results and no complications after discharge from hospital may not have needed to continue clinical follow‐up for 12 months. Based on inclusion criteria, we may have systematically removed patients with the best results from the study. We excluded lower risk patients because patients with soft tissue infections that did not require surgery were more likely not to stay in the system for a one‐year evaluation. Our electronic medical records are linked to other hospital systems in the area, so we could review hospital admissions, labs, imaging and clinic visits in the area. Thirdly, soft tissue infections were diagnosed based on clinical observations in addition to having either a negative MRI, SPECT CT or bone culture and histology. This is a more stringent definition of soft tissue infection because we wanted to have confidence that osteomyelitis was not missed or misclassified. Most studies use clinical judgement to define soft tissue infection. 7 , 22 , 23 The rate of false negatives in MRI and WBC SPECT CT is low, so misdiagnosis was unlikely. 24 , 25 , 26 Because we used a strict, objective definition of moderate and severe bone and soft tissue infection, the selection bias in this study favoured including more severe soft tissue infections. Infections with cellulitis that responded quickly to antibiotic therapy were not included because there was no concern for a deep infection or osteomyelitis. In those cases, advanced imaging (MRI and SPECT CT) was not performed and reserved for situations when the treating physician had a high index of suspicion for deep infection and osteomyelitis. We also acknowledge the ultimate decision for performing a surgery or amputation was made by a surgeon, potentially introducing bias into our outcomes. The finding of increased infection rate among subjects taking antidepressants was an interesting observation; however, this could be an artefact due to the relatively small number of patients taking antidepressants. This merits further study. Finally, measurement bias was likely because various operational definitions were used to define disease processes such as the interpretation of histology of osteomyelitis. Not every patient had documentation of the presence or absence of all the risk factors, co‐morbidities and outcomes that we evaluated; and 14%–24% of the patients that were initially identified were excluded because they did not have at least a 12‐month follow‐up. Despite the limitations, there are several strengths to this study. The operational definitions that were used to identify patients with osteomyelitis used the ‘gold standard’ of bone culture or histology. We recognise that other centres may use a positive ‘probe to bone test’ and abnormal radiographs to diagnose osteomyelitis; however, using these criteria to define osteomyelitis may include patients without bone infection with Charcot arthropathy or periosteal elevation from a healing fracture or patients that ‘probe to bone’ without infection. Patients are likely to be misdiagnosed as soft tissue infections when they do not show changes on radiographs or do not ‘probe to bone’. Another strength of this study is the substantial number of patients included in the analysis, and the fact that they were consistently evaluated by an experienced diabetic foot service. As a result, all the subjects included in the study had a uniform evaluation of neuropathy, wound characteristics and peripheral arterial disease. In conclusion, the results of this study showed that re‐infection was very common during the 12‐month follow‐up. Osteomyelitis, non‐healing wounds, longer times to heal, antidepressants and leukocytosis were associated with re‐infection. These findings can assist physicians and administrators plan for the services and re‐admissions expected in bone and soft tissue infections in patients with diabetes. CONFLICT OF INTEREST STATEMENT The authors do not report any potential conflicts of interest. Lavery LA, Tarricone AN, Ryan EC, et al. Re‐infection after treatment for moderate and severe diabetic foot infections. Int Wound J. 2024;21(11):e70123. doi: 10.1111/iwj.70123 DATA AVAILABILITY STATEMENT Data are available at reasonable request. REFERENCES Associated Data Data Availability Statement Data are available at reasonable request.