J Bone Joint Infect 2017; 2(3):122-126. doi:10.7150/jbji.17129

Research Paper

Clinical and Microbiological Characteristics of Bacteroides Prosthetic Joint Infections

Neel Shah1, Douglas Osmon1, Aaron J. Tande1, James Steckelberg1, Rafael Sierra2, Randall Walker1, Elie F. Berbari1 Corresponding address

1. Division of Infectious Disease, Mayo Clinic College of Medicine, Rochester, Minnesota, USA;
2. Department of Orthopedic Surgery, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.

This is an open access article distributed under the terms of the Creative Commons Attribution (CC BY-NC) License. See http://ivyspring.com/terms for full terms and conditions.
How to cite this article:
Shah N, Osmon D, Tande AJ, Steckelberg J, Sierra R, Walker R, Berbari EF. Clinical and Microbiological Characteristics of Bacteroides Prosthetic Joint Infections. J Bone Joint Infect 2017; 2(3):122-126. doi:10.7150/jbji.17129. Available from http://www.jbji.net/v02p0122.htm


Clinical and microbiological characteristics of patients with Bacteroides prosthetic joint infection (PJI) have not been well described in the literature. The aim of this retrospective cohort study was to assess the outcome of patients with Bacteroides PJI and to review risk factors associated with failure of therapy. Between 1/1969 and 12/2012, 20 episodes of Bacteroides PJI in 17 patients were identified at our institution. The mean age of the patients in this cohort at the time of diagnosis was 55.6 years; 59% (n=10) had knee involvement. Twenty four percent (n=4) had diabetes mellitus, and 24% had a history of either gastrointestinal (GI) or genitourinary (GU) pathology prior to the diagnosis of PJI. Thirty five percent (n=6) were immunosuppressed. The initial medical/surgical strategy was resection arthroplasty (n=9, 50%) or debridement and implant retention (n=5, 28%). Thirty seven percent (n=7) were treated with metronidazole. Eighty percent (n=4) of patients that failed therapy had undergone debridement and retention of their prosthesis, as compared to none of those treated with resection arthroplasty. Seventy percent (n=14) of patient episodes were infection free at their last date of follow up. In conclusion, a significant proportion of patients with Bacteroides PJI are immunosuppressed and have an underlying GI or GU tract pathology. Retention and debridement of the prosthesis is associated with a higher risk of treatment failure.

Keywords: Bacteroides, prosthetic joint infection


Prosthetic joint infections (PJI) are the most common cause of failure of total joint arthroplasty, resulting in additional procedures, increased health care costs, and increased morbidity for the patient [1-3]. The number of cases of PJI is rising, commensurate with the increasing number of prosthetic joints being implanted in the United States [4-6]. The Infectious Disease Society of America published guidelines in 2012 to help guide clinicians in the diagnosis and management of PJI. These guidelines primarily focus on the management of common pathogens associated with PJI, namely staphylococci and streptococci, and do not comprehensively address less commonly encountered organisms associated with PJI.

Anaerobes account for 4% of PJI cases [6, 7]. Similar to other, more common pathogens, these organisms have the potential to create significant morbidity among patients, resulting in increased health care costs and prolonged hospital stay. Bacteroides species are obligately anaerobic, Gram-negative bacilli that are commensal members of the human gastrointestinal microbiota. [8-10]. Risk factors associated with clinical infections with Bacteroides sp. include immunosuppression, rheumatoid arthritis and intra-abdominal pathology [10, 11]. Despite the ability to produce significant patient morbidity, there are few data available on the medical and surgical management of Bacteroides PJI [11].

Given the paucity of data associated with Bacteroides PJI, we performed a retrospective cohort study to describe the demographics, clinical characteristics and outcomes of patients with Bacteroides PJI. Results of this study will help guide both clinicians and microbiologists to effectively assess and manage such infections

Materials and Methods

Study Population

The medical records of all episodes of Bacteroides sp. total hip or knee arthroplasty infection seen at the Mayo Clinic in Rochester, MN between January 1, 1969 and December 31, 2012 were reviewed. Bacteroides PJI was defined according to a strict case definition, outlined below. All episodes were followed from the date of PJI diagnosis until treatment failure, prosthesis removal, death or the end of the study period. This study was approved by our institutional review board, and all patients had previously consented to participating in future research studies.

Data Collection

Data were obtained from the electronic and written medical records and transferred onto a structured data sheet, using the RedCAP™ database. The data was then analyzed using the JMP® (version 10.0.0) statistical program software. Data on date of birth, gender, medical comorbidities, type of surgery, and antimicrobial therapy was abstracted.

Definitions of terms used

(i) Bacteroides PJI was said to be present if at least two separate cultures grew Bacteroides sp. isolated from intra-articular aspiration or surgical specimens, or a single positive culture of Bacteroides sp. was isolated, along with any of the following: (a) purulence surrounding the prosthesis observed at the time of periprosthetic aspiration or during surgery; (b) acute inflammation consistent with infection on histopathologic examination; or (c) the presence of a sinus tract, found on examination or during operative repair. This definition was adapted from the Infectious Disease Society of America's guidelines for the management of PJI [6].

(ii) Treatment failure was defined as the occurrence of PJI, as outlined above, caused by Bacteroides or any other organism at any time after the original therapy episode.

(iii) The original therapy episode was defined as treatment that was undertaken during the initial hospitalization after diagnosis of PJI was established.

Statistical Methods

The number and percentage of episodes in each of the medical and surgical groups were calculated. Differences in demographic factors, clinical variables, and microbiology between the groups were compared using the chi-square test for categorical variables, and Wilcoxon rank sum analysis for continuous variables. Comparisons of the overall efficacy between the medical and surgical treatment strategies were performed by the use of the Petot-Petot Wilcoxon test.


Study population

Between January 1, 1969 and December 31, 2012, 20 episodes of Bacteroides PJI in 17 patients were identified at the Mayo Clinic in Rochester, MN. Three of the patients had more than one episode, due to treatment failure. The overall mean age at the time of diagnosis was 55.6 years (range, 17 to 74 years). Ten (59%) patients had an infected knee arthroplasty infection, while seven (41%) had a hip arthroplasty infection. Twelve (71%) were females. All of the patients in this cohort were Caucasian. Nine patients (53%) had prosthesis placement at sites other than their primary site of infection, and 3 (18%) were associated with prior arthroplasty placement at the same site as the current PJI. Seven patients (41%) had prior PJI at the same site. One of these episodes was associated with a prior staphylococcal PJI, one with Pseudomonas, one with Serratia marcescens, and one was culture negative. There were 4 episodes associated with a prior Bacteroides PJI (one patient had 2 recurrences). Table 3 outlines the frequency of selected comorbidities in our study population. Of note, of the two patients (12%) that had an active malignancy at the time of diagnosis, one had osteosarcoma and one had multiple myeloma. Four patients (24%) had a history of gastrointestinal/genitourinary (GI/GU) pathology at the time of PJI diagnosis. This included a history of recurrent urinary tract infections, prior bowel resection, gastrointestinal perforation, cholecystectomy, and, hysterectomy. Six patients (35%) were immunosuppressed at the time of PJI diagnosis; of note, four of these patients were taking prednisone at doses of less than 20 mg daily. Two of these patients were receiving methotrexate, and one was on prednisone doses of greater than 20 mg daily. One patient was on more than one immunosuppressive agent at the time of their PJI diagnosis. All patients were taking these medications due to underlying connective tissue disease or malignancy; the most common primary connective tissue disease associated with immunosuppressive use was rheumatoid arthritis.

Diagnostic methods

The diagnosis of Bacteroides PJI was based on the presence of at least two positive Bacteroides cultures obtained either from periprosthetic tissue or joint fluid in 12 of the 20 total episodes. Of those that did not fulfill this criteria (8 episodes), the diagnosis was made based on the presence of a single positive culture and the presence of a sinus tract (2 of the 8 episodes) or the presence of purulence around the joint space (6 of the 8 episodes). Ten patient episodes (50%) had elevated sedimentation rate and C-reactive protein; in the remaining episodes, markers were either normal or were not performed. Table 1 outlines some of the key clinical features associated with the Bacteroides PJI encountered at our institution, including treatment and history of prior Bacteroides PJI. There were 4 episodes associated with prior Bacteroides infections, and eleven of the 20 total episodes of PJI (55%) were associated with coinfection with an organism other than Bacteroides. Companion organisms included coagulase negative staphylococci (most common), Staphylococcus aureus, Providencia stuartii, Escherichia coli, Group D and Group B streptococci, viridans group streptococci, and multiple anaerobes including Actinomyces species, Clostridium species, Peptococcus species and Peptostreptococcus species.

Microbiological Data

Table 2 summarizes the microbiological characteristics of the Bacteroides isolates isolated at the time of diagnosis. Of note, we did not find any significant difference in overall outcome when comparing monomicrobial vs. polymicrobial Bacteroides PJI (p=0.892),

 Table 1 

17 Patients with Bacteroides PJI, classified by species isolated, medical and surgical therapy and prior exposure

Patient NumberEpisodes Per PatientBacteroides sp.Primary Medical TherapySurgical TherapyPrior Infection with Bacteroides sp.Outcome per episode
11B. fragilisImipenemRetentionNoTreatment Success
21B. fragilisTetracyclineResectionYesTreatment Success
31B. melaninogenicusPenicillin G2- stageNoTreatment Success
41B. fragilisMetronidazoleChronic suppressionNoTreatment Success
52B. fragilis (both episodes)Clindamycin/MetronidazoleRetention/ResectionYes/YesTreatment Failure/Treatment Success
61B. melaninogenicusLincomycinRetentionNoTreatment Failure
72B. vulgatus (both episodes)Cefadroxil/Amoxicillin-ClavulanateRetention/Chronic suppressionNo/NoTreatment Failure/Treatment Failure
82B. fragilis (both episodes)Imipenem/ImipenemRetention/ResectionNoTreatment Failure/Treatment Success
91B. fragilisClindamycin2-stageNoTreatment Success
101B. ovatusNoneAmputationNoTreatment Success
111B. melaninogenicusPenicillin GResectionNoTreatment Success
121B. fragilisMetronidazoleResectionYesTreatment Success
131B. biviusMetronidazoleResectionNoTreatment Success
141B. ovatusMoxalactamResectionNoTreatment Success
151Bacteroides (species not identified)MetronidazoleResectionNoTreatment Success
161B. fragilisMetronidazoleResectionNoTreatment Success
171B. ovatusMetronidazole2-stageNoTreatment Failure
 Table 2 

Various Bacteroides species isolates and susceptibility data based on antibiotics tested

Antimicrobial AgentB fragilis MIC (mg/L)B. melaninogenicus MICB. vulgatus MICB. ovatus MICB. bivius MIC
Penicillin64, 50, >4, 6.25, 64, 64, 1000.5>6464
Clindamycin1, 1, 25,3.1, 12.5, 1, 1, <0.780.5, <0.5<0.5<0.5
Cefoxitin4, 12.5, 4, 4432<0.5
Metronidazole4, 1, 0.8, 1.5, 4, 42, 2<0.5
Imipenem0.5, 0.5, 0.5, 0.50.5, <0.5
Ampicillin12. 5
B-lactamase2 strains positive2 strains negative
 Table 3 

Clinical Factors observed with Bacteroides PJI

Clinical Factor# of Patients (% total)
Rheumatoid Arthritis6 (35%)
Immunosuppression6 (35%)
Diabetes mellitus4 (24%)
GI/GU tract pathology4 (24%)
Active smoking3 (18%)
Renal disease2 (12%)
Active malignancy2 (12%)
Aseptic necrosis of bone1 (6%)

Surgical Intervention

Of the 18 total episodes that underwent a surgical intervention, nine (50%) were managed with resection arthroplasty. Three (17%) episodes were treated with two-stage exchange procedure, with the mean duration between staged procedures of 409 days (range 202-572 days). Five episodes (28%) were managed with irrigation, debridement and retention of the prosthesis. One episode (6%) was managed with amputation and two episodes were managed with chronic antimicrobial suppressive therapy alone, without any surgical interventions.

Medical Therapy

Of the 19 total episodes that were treated with antimicrobial therapy, seven (37%) received a course of metronidazole, 3 (16%) received either clindamycin or lincomycin, 1 (5%) received tetracycline, and 8 (42%) received a β-lactam (imipenem, penicillin, amoxicillin-clavulanate, moxalactam, oxacillin). The mean duration of primary therapy was 57.1 days (range, 14-360 days). Three of the seven episodes that were managed with metronidazole developed peripheral neuropathy. One of these patients developed diarrhea not associated with Clostridium difficile colitis. Eight patient episodes received oral antimicrobial suppressive therapy following the initial therapy course, of which three were placed on metronidazole, two were on clindamycin, and one patient episode each was on penicillin VK, doxycycline and amoxicillin-clavulanate. Antibiotic poly (methyl methacrylamide) spacers were placed in only 2 episodes, of which one had a vancomycin/gentamicin spacer placed, and the other had a vancomycin/tobramycin spacer placed.

Outcome data

Table 1 outlines the various medical and surgical interventions provided for each patient and episode, including outcomes of each episode.


To our knowledge, this is the largest study of patients with Bacteroides prosthetic joint infections published in the literature. Based on our data, a significant proportion (35%) of patients with Bacteroides PJI had underlying immunosuppression, with a significant number of patients in this study presenting either with diabetes mellitus (24%), and rheumatoid arthritis or the use of immunosuppressive therapy (35% for each). Eighteen percent of the patients in this study had a history of active smoking. In addition, two of the patients (12%) had an active malignancy during their PJI diagnosis. Previous findings by other authors align with this finding [10]. Eight of the seventeen patients (47%) in this study had prior infection of their prosthesis at the same site as their PJI, suggesting that this too is a possible risk factor associated with Bacteroides PJI; of note, 4 of the patients previously had a Bacteroides PJI (at an outside institution), suggesting that prior infection with this organism increases the risk of reinfection. Our study found a relatively high proportion of patients (24%) with a history of underlying GI/GU pathology. Intra-abdominal pathology has been associated with Bacteroides infections, due to the fact that this organism is a commensal of the human gastrointestinal tract [8-10]. Our findings seem to reinforce this as well.

Based on our findings, a significant proportion of patients had a polymicrobial infection (55% of the total episodes). However, we did not find any significant difference in overall outcome when comparing monomicrobial vs. polymicrobial Bacteroides PJI (p=0.892). We noticed a high failure rate associated with debridement with prosthesis retention and chronic suppressive therapy alone, with 80% of the treatment failure episodes are accounted for by one of these two management strategies. In addition, among the patients who were treated successfully in this study, the majority underwent either resection arthroplasty or 2-stage exchange (9 of 14 episodes, or 64%). Together, these findings suggest that removal of all hardware may improve outcomes associated with these infections; however our cohort did not contain enough patients to make a definitive conclusion with regards to this. From an antimicrobial standpoint, the majority of our patients were treated with metronidazole or imipenem. In general, both of these agents are well tolerated for short periods of time, with minimal associated side effects. Both have good bone penetration as well, based on prior PK/PD studies [13]. The rate of Bacteroides resistance to metronidazole is still <1%. Bacteroides sp. are highly susceptible to carbapenems; however the rate of resistance to these agents is also increasing [12]. In this study, all of the tested isolates were susceptible to imipenem, and the majority were susceptible to metronidazole.

There are a number of limitations associated with our study inherent to its retrospective nature; there is the risk of recall bias, along with selection bias. Our study did not contain a sufficient number of patients to perform reliable outcome analyses, and therefore was restricted to being primarily descriptive in nature.

In conclusion, Bacteroides PJIs have distinctive clinical characteristics. The majority of isolates were susceptible to metronidazole, imipenem, tetracyclines and clindamycin. Episodes treated with prosthesis retention appeared to have worse outcomes than episodes treated with resection arthroplasty or 2-stage exchange. A high proportion of underlying immunosuppression and a history of GI/GU pathology were observed as significant risk factors in this cohort. Future cohort studies are warranted to confirm these findings.

Competing Interests

The authors have declared that no competing interest exists.


1. Galliera E, Drago L, Marazzi MG, Romano C, Vassena C, Corsi Romanelli MM. Soluble urokinase-type plasminogen activator receptor (suPAR) as new biomarker of prosthetic joint infection: Correlation with inflammatory cytokines. Clinica Chimica Acta. 2013;441:23-28

2. Marculescu CE, Cantey JR. Polymicrobial Prosthetic Joint Infections: Risk Factors and Outcome. Clin Orthop Relate Res. 2008;6:1397-1404

3. Trampuz A, Zimmerli W. Diagnosis and treatment of implant-associated septic arthritis and osteomyelitis. Curr Infect Dis Rep. 2008;5:394-403

4. Hsieh PH, Lee MS, Hsu KY, Chang YH, Shih HN, Ueng SW. Gram-negative Prosthetic Joint Infections: Risk Factors and Outcome of Treatment. CID. 2009;49(7):1036-1043

5. Sculco TP. The economic impact of infected joint arthroplasty. Orthopedics. 1995;18:871-3

6. Osmon DR, Berbari EF, Berendt AR, Lew D, Zimmerli W, Steckelberg JM, Rao N, Hanssen A, Wilson WR. Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Disease Society of America. CID. 2013;1:e1-e25

7. Tande AJ, Patel R. Prosthetic Joint Infection. Clin Microbiol Rev. 2014;2:302-45

8. Hickey MM, Davies UM, Dave J, Vogler M, Wall RA. Metronidazole Resistant Bacteroides fragilis Infection of a Prosthetic Joint. J Infect. 1990;20(2):129-33

9. Sanders KL, Kaplowitz LG. Late Infection with Bacteroides fragilis in a Prosthetic Joint. Scand J Infect Dis. 1980;12(3):235

10. Pattison RM, Cooke RPD, James SE. Bacteroides fragilis infection of a knee prosthesis after haemorrhoidectomy. Lancet. 1995;346(8982):1097

11. Roux AL, El Sayed F, Duffiet P, Bauer T, Heym B, Gaillars JL, Herrmann JL, Rottman M. Ruminococcus gnavus total hip arthroplasty infection in a 62-year-old man with ulcerative colitis. J Clin Microbiol. 2015;53(4):1428-30

12. Sadarangani SP, Cunningham SA, Jeraldo PR, Wilson JW, Khare R, Patel R. Metronidazole and carbapenem resistant Bacteroides thetaiotamicron isolated in Rochester, MN in 2014. Antimicrob Agents Chemotherap. 2015;59(7):4157-61

13. Summersgill JT, Schupp LG, Raff MJ. Comparative penetration of metronidazole, clindamycin, chloramphenicol, cefoxitin, ticaracillin and moxalactam into bone. Antimicrob Agents Chemother. 1982;4:601-603

Author contact

Corresponding address Corresponding author: Elie. F. Berbari, M.D. FIDSA, Division of Infectious Disease, Mayo Clinic College of Medicine, 200 1st Street SW, Rochester, MN 55905. Phone: (507) 255-6482. Fax: (507) 255-7767. E-mail: berbari.elieedu.

Published 2017-3-19