The infected diabetes‐related foot: Comparison of erythrocyte sedementation rate/albumin and C‐reactive protein/albumin ratios with erythrocyte sedimentation rate and C‐reactive protein to differentiate bone and soft tissue infections

The objective of this study was to evaluate the effectiveness of C‐reactive protein (CRP)/albumin, erythrocyte sedimentation rate (ESR)/albumin ratio, ESR, CRP and albumin to differentiate bone and soft tissue infection in persons with diabetes. We retrospectively evaluated 242 individuals admitted to hospital with diabetes‐related foot infections (DFI). We categorised DFI cases as either bone (OM) or soft tissue infection based on bone culture and/or histology. We evaluated the diagnostic accuracy of CRP, ESR, albumin, CRP/albumin and ESR/albumin as biomarkers to diagnose OM in persons with diabetes. The median age was 53 years (74% male). There were 224 diabetes‐related patients of which 125 had been diagnosed with osteomyelitis. The ESR/albumin and CRP/albumin ratios cut‐points were >17.84 and >1.83, respectively. ESR/albumin and CRP/albumin ratios had similar diagnostic parameters: AUC (0.71, 0.71), sensitivity (70.0%, 57.0%), specificity (62.0%, 75.0%), positive predictive value (67.0%, 71.0%) and negative predictive value (66.0% and 71.0%). In contrast diagnostic efficiency of CRP and ESR were AUC 0.71 and 0.71, sensitivity (45.6%, 71.2%), specificity (85.5%, 60.7%), positive predictive value (70.0%, 65.9%) and negative predictive value (59.5%, 66.4%), respectively. When comparing area under the curves, the results showed that ESR/albumin was not significantly different to ESR alone (Delong test pvs ESR >0.1). Similarly, CRP/albumin was not significantly different to CRP alone (Delong test pvs CRP >0.1). In conclusion, ESR/albumin and CRP/albumin ratios provided comparable results as using ESR and CRP alone.

Diabetes-related foot osteomyelitis (OM) can be challenging to diagnose.Clinical examination, traditional radiographs and standard laboratory tests often have a unique presentation in comparison to patients without diabetes.Diabetes-related peripheral sensory neuropathy can mask painful sensations associated with infection, removing a person's innate alarm mechanism to injury.For this reason, diabetic foot infections can be overlooked, and early diagnosis can be missed.
Evaluation of the ulcer to determine if it extends to bone is helpful to diagnose OM. 1 However, the 'probe to bone test' (PTB) is difficult to perform and interpret, especially for inexperienced clinicians. 2 In addition, PTB is not routinely performed or even recognised by many emergency room physicians, internists, nurse practitioners or physician assistants.Plain radiographs often lag by weeks before changes are evident, so acute and subacute cases are missed at a time when treatment may be most effective.
Advanced imaging such as magnetic resonance imaging (MRI) and single-photon emission computed tomography (SPECT/CT) should not be used unless there is a high index of suspicion for OM.These tests are expensive and not often readily available.If we can identify diagnostically accurate inflammatory markers able to consistently identify bone infections, we may also be able to identify patient who require further advanced imaging, avoiding unnecessary tests and cost in low-risk patients.
Inflammatory markers are inexpensive with easy access.In previous studies C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) have been shown to perform better than newer biomarkers such as procalcitonin and interleukins to diagnose. 3,4However, a recent meta-analysis has shown that procalcitonin may have superior performance to both ESR and CRP in differentiating grade 2 and grade 3 diabetes-related foot ulcerations. 5Previous studies have also described the utility of ESR and CRP in monitoring OM. 6,7 Recently, several authors have evaluated the ratio of CRP and albumin to identify total joint infections, 8 infection versus inflammation in patients with systemic lupus erythematosus 9 and OM in diabetes-related foot infections 10 as well as to predict death. 11,12The goal of this study was to evaluate CRP/albumin and ESR/albumin ratios to diagnose OM to determine if they improved on the diagnostic characteristics of ESR and CRP alone.

| METHODS
This retrospective study was approved by our institution's research ethics board.A review of electronic medical records were used to evaluate patients matching our inclusion criteria.We included patients with type 1 or type 2 diabetes based on American Diabetes Association criteria, 18 to 89 years of age, admitted to hospital between 12 June 2009, and 21 February 2017, with soft tissue or bone infection.The definition of STI according to the IWGDF foot infection criteria was used. 13We defined OM as either a positive pathology or bone culture.Patients who had chronic kidney disease with a GFR < 30 mL/min were excluded from this analysis.
In addition, we collected data on diagnosis, treatment, comorbidities, recent medication use, results of laboratory tests and clinical signs and symptoms of the presenting infections.Blood samples for ESR, CRP, albumin and WBC are collected as part of standard practise for each admission as part of routine work-up.Patients who had missing biomarker data for ESR, CRP and albumin were also excluded.
To summarise baseline characteristics, parametric data were assessed using t-tests, whereas categorical variables were analysed using chi-square tests and a p-value of <0.05 was considered significant.Receiver operating characteristic (ROC) curves were generated to determine the diagnostic value of each test for the assessment of diabetes-related foot bone and soft tissue infection.
The area under the curve (AUC) and 95% confidence interval (CI) were calculated.The discriminatory value of curves was interpreted as excellent (0.9 to 1), good (0.8 to 0.89), fair (0.7 to 0.79), poor (0.6 to 0.69), or as failing or having no discriminatory capacity (0.5 to 0.59). 14e optimal threshold for each marker as a diagnostic tool for differentiating diabetes-related foot soft tissue and bone infections was determined using the Youden index. 15Delong tests were used to compare the AUCs of different biomarkers.The Delong test is a nonparametric statistical test used to compare the AUC of two or more correlated ROC curves.The Delong test considers the correlation between the ROC curves and provides a p-value indicating the statistical significance of the difference between the AUCs. 16The sensitivity and specificity of the diagnostic tests, predictive values and likelihood ratios were also calculated.Statistical analyses were performed using R (The R Foundation).

| RESULTS
We identified a total of 242 patients representing 117 STI cases and 125 OM cases.The median age of patients was 53 years (interquartile range [IQR], 45 to 60).There were 180 (74%) male patients and 62 (26%) female patients.There was no significant difference between age, gender, BMI or glycated haemoglobin between the two groups (Table 1).
The ROC curves showed that ESR/albumin ratio, CRP/albumin ratio, ESR and CRP all had similar AUC (0.71, 0.71, 0.709, 0.708).The AUCs of these biomarkers were between 0.7 and 0.8, indicating that they had fair performance in differentiating between STI and OM (Figures 1 and 2).

| DISCUSSION
The results of this study suggest that ESR/albumin and CRP/albumin ratios were similar to CRP and ESR alone in diagnosing OM in people with diabetes-related foot infections.Our goal was to determine if we could find better screening laboratory tests that could be used across disciplines to diagnose OM.Our ESR, CRP, ESR/albumin and CRP/albumin results are similar to previous published studies that evaluate diagnostic parameters. 17,18rtually all foot infections are associated with a break in the protective barrier of the epidermis and may involve deeper structures of the foot architecture.Diabetes foot ulcers are commonly colonised with microorganisms from the surrounding skin flora or environment, but this does not necessarily equate to the fact that all foot ulcers will develop an infection; however, the incidence of infection is exceedingly high.For instance, in randomised clinical trials to treat diabetes-related foot ulcers, the incidence of  soft tissue infection alone can be diagnostic dilemma. 4,183][24][25][26][27] Non-diabetic OM is often defined just above reference values of 20-30 mm/h for ESR.In contrast, OM cut-points for diabetics are higher at 60-67 mm/h for ESR and 7.9-14.0mg/dL for CRP. 4,28Consequently, applying non-diabetes-related cut-points could lead to misdiagnosis of OM in diabetic patients.For instance, non-diabetic OM of the foot had cut-points of 45.5 mm/h for ESR and 3.45 mg/dL for CRP. 29 were only able to identify two studies that used CRP/albumin ratio in diabetes-related foot infections. 10,30In a retrospective cohort study, Aragon-Sanchez and colleagues evaluated 245 patients with moderate and severe diabetes-related foot infections.They used SIRS criteria to determine severe infections 13 and CRP, ESR, CRP/Albumin, albumin and neutrophil to lymphocyte ratio as biomarkers to predict severe infection.ESR, albumin, neutrophil to lymphocyte ratio, CRP and CRP/albumin were associated with severe infection. 30Similar results were found in Eran and colleagues' work evaluating OM.They reported results of a retrospective cohort study of 97 patients with diabetes-related foot infection and used MRI to define the presence of bone infection.They evaluated ESR, CRP, WBC, CRP/albumin ratio to diagnoses OM.CRP/albumin had the highest AUC; however, CRP and ESR and albumin were also effective to identify OM. 10 Albumin is an inflammatory marker that decreases when there is systemic inflammation.2][33][34] At that time, low serum albumin was not associated with diabetes-related foot infection or inflammation, but it is reasonable to understand how this association could have been mislabelled as a nutritional marker.Recent studies have reported an association with poor nutritional status and markers of inflammation in patients with acute illness. 35The severity of the inflammation driving lower albumin could be mistakenly interpreted as poor nutrition among systemically ill subjects.The diagnostic parameters for albumin by itself are poor (Table 2).
There are several advantages and limitations for this study.One of the advantages of this study was that we used bone culture or bone histology as the reference standard to define OM.Most studies use clinical parameters or MRI as the reference standard.
IWGDF and IDSA have recommended using bone biopsy to define OM. 13 Patients with no bone infection had either a negative bone biopsy or a negative MRI or SPECT CT.We did not differentiate acute, subacute or chronic OM for the purpose of evaluation.As a practical matter, we have been unable to find a cogent operational definition of OM chronicity.We did not differentiate cases with a concomitant soft tissue abscess or extensive soft tissue infection versus OM with no soft tissue infection.We are not sure if acute OM provides a different inflammatory signature compared to chronic OM, or if OM with abscess provides different biomarker characteristics than a severe soft tissue infection.We did not have adequate information from the retrospective data to make good delineations of these variables.

2. 19 (
IQR: 0.75 to 4.98) and 0.63 (IQR: 0.18 to 1.74) (p < 0.001).The median ESR/Albumin Ratio was 25.61 (IQR: 21.34 to 30.21) and 14.85 (IQR: 12.47 to 17.30) ( p < 0.001).ESR values were 82 mm/h (IQR: 55.0 to 120.0 mm/h) in the OM group and 54.0 mm/h (IQR: The results from the Delong test for AUC comparisons are shown in Table3.The results show that ESR/albumin was not significantly different to ESR alone (Delong test p vs. ESR >0.1).Similarly, CRP/albumin was not significantly different to CRP alone (Delong test pvs CRP >0.1).However, the result of the Delong test showed that ESR, CRP, CRP/albumin and CRP/albumin significantly outperformed albumin alone as a diagnostic marker for OM.

F I G U R E 1
Combined receiver operator characteristics curve for CRP, ESR, albumin, CRP/albumin, ESR/albumin to diagnose diabetesrelated foot osteomyelitis.The results show no difference in the degree of separability of soft tissue and bone infection cases between CRP, ESR, CRP/albumin and ESR/albumin.Albumin had the lowest degree of separability between cases.CRP, C-reactive protein; ESR, erythrocyte sedimentation rate. in 12 weeks is usually 17%-36%. 19-21The initiating mechanisms of microbial infection in diabetes-related foot ulcers remain unknown.Scant reports propose that persons with diabetes are usually compromised hosts, so they do not have normal, clinical inflammatory responses to infections agents, and they have an exaggerated inflammatory response.Differentiating bone and soft tissue infection, or more often soft tissue abscess and OM versus

F I G U R E 2
Faceted receiver operator characteristics curve for CRP, ESR, albumin, CRP/albumin and ESR/albumin to diagnose diabetesrelated foot osteomyelitis.Shaded areas coincide with 95% confidence intervals.The results show no difference in the degree of separability of soft tissue and bone infection cases between CRP, ESR, CRP/albumin and ESR/albumin.Albumin had the lowest degree of separability between cases.CRP, C-reactive protein; ESR, erythrocyte sedimentation rate.

1
Study demographics and group differences.
Results of the receiver operator characteristics curve analysis for 242 cases.CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; NPV, negative predictive value; PPV, positive predictive value.T A B L E 3 Delong test for comparison of AUCs.Abbreviations: AUC, area under the curve; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate.
vious studies have shown promising results with CRP/albumin in diabetes-related foot infections, our findings did not improve diagnostic characteristics for diabetes-related foot OM compared to CRP and ESR alone.Further research is needed to determine the potential of using CRP/albumin and CRP/albumin as a diagnostic tool for OM.T A B L E 2