Role of bone biopsy and deep tissue culture for antibiotic stewardship in diabetic foot osteomyelitis

Abstract

Objectives

To describe organisms most frequently identified on bone biopsy or deep tissue culture and determine how culture data impacted antibiotic management in patients with diabetic foot osteomyelitis (DFO).

Methods

We retrospectively reviewed patients admitted with a diabetic foot ulcer (DFU) between 3 March 2018 and 31 December 2019 and selected for patients diagnosed with infectious osteomyelitis (OM) of the lower extremity. We stratified patients by whether a bone biopsy or deep tissue culture was obtained and compared rates of antibiotic utilization with chi-squared and Fisher’s exact tests.

Results

Of 305 patients with a DFU, 152 (50%) were clinically diagnosed with DFO. Forty-seven patients received 41 deep tissue cultures and 29 bone biopsy cultures for a total of 70 cultures. Of 45 (64%) positive cultures, 36 (80%) had Gram-positive organisms and 19 (42%) had Gram-negative organisms. MDR organisms were isolated in 7 (15%) patients. Culture data resulted in antibiotic changes in 41 (87%) patients. Therapy was narrowed in 29 (62%) patients and broadened due to inadequate empirical coverage in 4 (9%) patients. Culture data from 18 (40%) patients showed susceptibility to an oral treatment regimen with high bioavailability. There was no significant difference in rates of antibiotic utilization at discharge between patients who underwent bone biopsy or deep tissue culture relative to those who did not (77% versus 75%, P = 0.86), although less MRSA coverage was used (34% versus 50%, P = 0.047).

Conclusions

In patients with DFO, deep tissue and bone biopsy cultures were infrequently obtained but resulted in targeted therapy changes in most patients. Culture data usually allowed for narrowing of antibiotics but revealed inadequate empirical coverage in a subset of patients.

Introduction

Nearly one-third of people with diabetes will develop a diabetic foot ulcer (DFU) over the course of their lifetime, and up to 50% of individuals with a DFU will develop a diabetic foot infection (DFI).^1,2 While DFIs almost always begin as an open wound in the skin and soft tissue, they can spread contiguously to deeper structures such as bone, leading to diabetic foot osteomyelitis (DFO). Osteomyelitis (OM) is present in about half of all patients requiring hospitalization for a DFI and accounts for 20% of moderate DFIs and up to 60% of severe DFIs.^3,4 Up to 80% of patients with DFO may ultimately require amputation, leading to significant morbidity, reduced quality of life, prolonged hospitalizations and a 5-year mortality rate near 50%.^1,5,6

Appropriate antimicrobial therapy may mitigate the harms associated with DFO. Although bone culture and histology are the gold standard for identifying the causative organism in DFO, bone biopsy is not always performed, even in settings where resistant organisms are common.^5,7 Absence of microbiological diagnosis coupled with the high rate of poor clinical outcomes in patients with DFO often leads to overtreatment with empirical antibiotics, which can include excessively broad coverage, parenteral instead of enteral therapy, or unnecessarily long or repeated treatment courses.⁸ Unnecessary antibiotic courses may increase the risk of adverse effects from antibiotic therapy and drive antimicrobial resistance.⁹ Despite the prevalence of DFI, limited research exists that examines antimicrobial stewardship in this field.

The primary aim of this study was to: (1) describe organisms most frequently identified on bone biopsy or deep tissue culture; and (2) describe the effect of culture data on antibiotic management in patients hospitalized with DFO as part of usual care. The secondary aim of this study was to compare antibiotic utilization between those who received bone biopsy or deep tissue culture compared with those who did not.

Methods

Data collection

We conducted a retrospective cohort study of adults (age ≥ 18 years) with DFO hospitalized at Northwestern Memorial Hospital (NMH). We queried our Enterprise Data Warehouse (EDW) for the index encounter of adult patients with an active ICD-10 code of ‘diabetes’ and ‘ulcer’ who were discharged between 3 March 2018 and 31 December 2019. The start date for the cohort correlated with adoption of our current electronic medical record system. We conducted manual chart review and excluded patients with a non-foot ulcer or those without clear documentation of an ulcer to establish a DFU cohort. We then selected for patients with a discharge diagnosis of OM of the lower extremity to create a suspected DFO cohort. The diagnosis of OM was clinical and was adjudicated by the treating physician. We extracted demographic, clinical and laboratory data including age, sex, discharge diagnoses, antibiotic use and microbiological data for patients who underwent bone or deep tissue biopsy. Bone biopsies were collected for diagnostic and therapeutic purposes. Deep tissue biopsy was defined as a non-bone biopsy sample collected as part of a debridement and/or amputation. In patients with DFO who underwent bone or deep tissue biopsy, superficial wound culture data were also collected when available to determine concordance. Concordance was defined as the superficial wound culture identifying all pathogens detected on bone or deep tissue biopsy without additional pathogens.

Data analysis

We compared patients with DFO who received bone or deep tissue biopsy with patients with DFO who did not receive bone or deep tissue biopsy during their index hospitalization. Outcome measures included rates of antibiotic utilization and the spectrum of antibiotic coverage. Broadening of antibiotic therapy was defined as changing to a drug that covered more resistant or more species of bacteria. Narrowing of antibiotic therapy was defined as changing to a drug that covered fewer resistant organisms or fewer species of bacteria.

We calculated frequencies and counts of demographics, microbiological diagnoses and antibiotic utilization using descriptive statistics. All proportions were compared using chi-squared test or Fisher’s exact test, based on counts. R (4.1.1, Vienna, Austria) was used for analysis.

Ethics

The study was approved by the Northwestern University Institutional Review Board.

Results

Our initial study cohort included 305 patients with a diagnosis of DFU. Of these patients, 242 (79%) had a superimposed DFI, which included the diagnoses of cellulitis, wound infection, OM and/or gangrene. OM was the most prevalent, affecting 152 (63%) of individuals with a DFI during their hospitalization. Individuals diagnosed with DFO were predominately male (73%) with an average age of 63.4 years (range 32–100 years) on hospital admission.

Of patients diagnosed with DFO, 47 (31%) received bone biopsy culture and/or deep tissue culture. Most of these patients (81%) had their bone biopsy or deep tissue culture obtained during surgery for amputation. In total, 29 bone biopsy cultures were obtained with a 31% positivity rate, and 41 deep tissue cultures were obtained with an 88% positivity rate. There were 45 positive cultures altogether, of which 36 cultures (80%) grew Gram-positive organisms and 19 cultures (42%) grew Gram-negative organisms. There was no significant difference in the prevalence of Gram-positive organisms (89% versus 78%, P = 0.456) or Gram-negative organisms (33% versus 44%, P = 0.546) found on positive bone biopsy cultures compared with positive deep tissue cultures. Superficial wound cultures were obtained in 21 (45%) patients and were concordant with deep tissue or bone biopsy culture in 2 (10%) of patients. In both cases of concordance, MRSA was the identified organism.

Of the 47 patients with DFO for whom deep tissue or bone biopsy culture data were available, 40 (85%) had at least one positive culture. Of patients with positive cultures, 24 (60%) patients had polymicrobial infections. The Gram-positive bacteria most commonly isolated from patients were Streptococcus spp. (30%), Staphylococcus aureus (28%) and CoNS (17%) (Table 1). The most prevalent Gram-negative bacteria isolated were Proteus spp. (13%), Klebsiella spp. (11%), Pseudomonas aeruginosa (6%) and Escherichia coli (6%). MDR organisms (MDROs) were isolated in seven (15%) patients, including five cases of MRSA (10%) and two cases of ESBL-producing organisms (4%).

Table 1.

Bacteria isolated from deep tissue and bone biopsy culture of DFO and effect on antibiotic therapy

	Patients (n = 47)	Percent
Positive culture	40	85.1
Gram-positive organisms	32	68.1
S. aureus	13	27.7
CoNS	8	17.0
Streptococcus spp.	14	29.8
Enterococcus spp.	12	25.5
Gram-negative organisms	17	36.2
P. aeruginosa	3	6.4
Proteus spp.	6	12.8
Klebsiella spp.	5	10.6
E. coli	3	6.4
MDROs	7	14.9
MRSA	5	10.6
ESBL-producers	2	4.3
Antibiotics changed	41	87.2
Narrowed	29	61.7
Broadened	4	8.5

Most patients (87%) were prescribed antibiotics in the 48 h prior to obtaining bone or deep tissue cultures. The most common empirically prescribed antibiotics were vancomycin (70%) and piperacillin/tazobactam (57%) (Figure 1). After obtaining cultures, antibiotics were changed in 41 (87%) of patients. Therapy was narrowed in 29 (62%) patients and broadened due to inadequate empirical coverage in 4 (9%) patients. Culture-directed therapy was started in four (9%) patients in which it had initially been withheld. The antibiotics used most frequently after obtaining culture included vancomycin (19%), ciprofloxacin (19%) and metronidazole (15%). Thirteen patients (28%) were discharged from the hospital without any antibiotics at the discretion of the treating physician. Culture data from 18 (40%) patients showed susceptibility to an oral treatment regimen with high bioavailability, including clindamycin, doxycycline, trimethoprim-sulfamethoxazole or ciprofloxacin.

Figure 1.

Rate of antibiotic utilization: empirical versus culture directed. In patients who underwent deep tissue or bone biopsy, the most common empirical antibiotics were vancomycin and piperacillin/tazobactam. Vancomycin was discontinued in 51% of patients after culture data were obtained, although it was one of the most common culture-directed antibiotics, along with ciprofloxacin and metronidazole (a). Overall, culture data allowed for the discontinuation of MRSA coverage (vancomycin, daptomycin, linezolid, doxycycline, trimethoprim-sulfamethoxazole) in 47% of patients and anti-pseudomonal coverage (aztreonam, cefepime, piperacillin/tazobactam, meropenem) in 49% of patients (b).

There was no significant difference in rates of antibiotic utilization at discharge between patients who underwent bone biopsy or deep tissue culture relative to those who did not [36 (77%) versus 79 (75%), P = 0.86]. This trend held true for both oral [22 (47%) versus 62 (59%), P = 0.16] and IV [22 (47%) versus 37 (35%), P = 0.18] antibiotics. However, MRSA coverage was prescribed less frequently in patients who underwent bone biopsy or deep tissue culture [16 (34%) versus 53 (50%), P = 0.047]. There was also a trend towards less P. aeruginosa coverage [15 (32%) versus 44 (42%), P = 0.36], but this did not reach statistical significance.

Discussion

In this study, OM was the most common infection diagnosed in patients admitted with DFIs, consistent with rates previously described.⁷ Most patients with DFO were diagnosed clinically and treated with empirical antibiotics, as rates of deep tissue culture and bone biopsy were low and obtained in only 31% of patients. This was due to samples being obtained predominantly at the time of amputation, as percutaneous bone biopsies are not performed at our institution. Studies have suggested that percutaneous image-guided biopsy does not result in a microbiological diagnosis that routinely impacts antibiotic choice or outcomes.^10–12 However, in our study, deep tissue and bone biopsy culture data resulted in targeted therapy changes in the majority (87%) of patients. This suggests that operative samples may have greater clinical utility than percutaneous samples, although further studies are needed.

Based on deep tissue and bone biopsy culture results, antimicrobial therapy was most frequently (62%) narrowed. This has important implications for antimicrobial stewardship, as more targeted therapies can potentially limit adverse effects from empirical antibiotic therapy and avoid the development of bacterial resistance that occurs with empirical use of broad-spectrum antibiotics.^8,13 Future studies could evaluate the use of empirical antimicrobials with a narrower spectrum in cases of suspected DFO and reserve broader therapy for clinically deteriorating patients or those with MDRO growth on deep tissue or bone biopsy culture.

DFO in our study was frequently polymicrobial, with Streptococcus spp., S. aureus and CoNS being the most commonly isolated pathogens, consistent with previous literature.^7,14 A subset (15%) of patients with DFO who underwent bone or deep tissue biopsy in our cohort grew MDROs, including MRSA and ESBL-producing organisms, which have been found to be increasingly prevalent in cases of DFI.^15,16 Empirical treatment for OM may not always cover MDROs, particularly resistant Gram-negative bacteria.⁷ Our study reflected this phenomenon, as empirical coverage was inadequate in 9% of patients with DFO who underwent bone or deep tissue biopsy. This reiterates the importance of bone biopsy culture to identify patients with MDROs not covered by empirical therapy in order to target treatment, since widespread and prolonged use of empirical broad-spectrum antibiotics have been associated with an increased prevalence of MDRO in this population.¹⁷

A limitation of our study was that the majority (81%) of patients receiving deep tissue or bone biopsy at our institution did so during amputation. Severity of DFI is correlated with rate of amputation, so patients in whom deep tissue or bone biopsy culture data was obtained likely had more severe disease.¹⁸ Although guidelines advocate for cessation of antibiotics if clean amputation margins are obtained, in practice, antibiotics are frequently continued following amputation due to concern for poor wound healing or inadequate margins.^7,13 The lack of antibiotic cessation may account for our failure to detect a difference in the rate of antibiotic utilization between patients receiving deep tissue or bone biopsy culture relative to those who did not. Additional prospective studies are needed to determine if deep tissue or bone biopsy culture significantly impacts the rates of antibiotic use in patients with DFO.

Overall, in patients admitted with DFO, deep tissue and bone biopsy cultures were infrequently obtained but resulted in targeted therapy changes in most patients. While culture data usually allowed for narrowing of antibiotics, it revealed inadequate coverage in a subset of patients. Improved efforts to obtain deep tissue or bone biopsy cultures in patients presenting with DFI with concern for OM would allow for improved antibiotic stewardship in this field.

Funding

This study was carried out as part of our routine work.

Transparency declarations

To the best of our knowledge, no conflict of interest, financial or other, exists with regards to the information presented in this manuscript.