Chronic Suppression of Periprosthetic Joint Infections with Oral Antibiotics Increases Infection-Free Survivorship

Transcription

1 1220 COPYRIGHT Ó 2015 BY THE JOURNAL OF BONE AND JOINT SURGERY, INCORPORATED Chronic Suppression of Periprosthetic Joint Infections with Oral Antibiotics Increases Infection-Free Survivorship Marcelo B.P. Siqueira, MD, Anas Saleh, MD, Alison K. Klika, MS, Colin O Rourke, MS, Steven Schmitt, MD, Carlos A. Higuera, MD, an Wael K. Barsoum, MD Investigation performe at the Orthopaeic an Rheumatologic Institute, Clevelan Clinic Founation, Clevelan, Ohio Backgroun: The clinical benefit of chronic suppression with oral antibiotics as a salvage treatment for periprosthetic joint infection is unclear. The purpose of this stuy was to compare infection-free prosthetic survival rates between patients who receive chronic oral antibiotics an those who i not following irrigation anebriement with polyethylene exchange or two-stage revision for periprosthetic joint infection. Methos: We reviewe the recors on all irrigation anebriement proceures with polyethylene exchange an twostage revisions performe at our institution from 1996 to 2010 for hip or knee periprosthetic joint infection. Of 625 patients treate with a total of 655 eligible revisions, ninety-two receive chronic oral antibiotics for a minimum of six months an were eligible for inclusion in our stuy. These patients were compare with a matche cohort (ratio of 1:3) who i not receive chronic oral antibiotics. Results: The five-year infection-free prosthetic survival rate was 68.5% (95% confience interval [CI] = 59.2% to 79.3%) for the antibiotic-suppression group an 41.1% (95% CI = 34.9% to 48.5%) for the non-suppression group (hazar ratio [HR] = 0.63, p = 0.008). Stratification by the type of surgery an the infecting organism showe a higher five-year survival rate for the patients in the suppression group who unerwent irrigation anebriement with polyethylene exchange (64.7%) compare with those in the non-suppression group who unerwent irrigation anebriement with polyethylene exchange (30.4%, p < ) an a higher five-year survival rate for the patients in the suppression group who ha a Staphylococcus aureus infection (57.4%) compare with those in the non-suppression group who ha a Staphylococcus aureus infection (40.1%, p = 0.047). Conclusions: Chronic suppression with oral antibiotics increase the infection-free prosthetic survival rate following surgical treatment for periprosthetic joint infection. Patients who unerwent irrigation anebriement with polyethylene exchange an those who ha a Staphylococcus aureus infection ha the greatest benefit. Level of Evience: Therapeutic Level III. See the Instructions for Authors for a complete escription of levels of evience. Peer Review: This article was reviewe by the Eitor-in-Chief an one Deputy Eitor, an it unerwent bline review by two or more outsie experts. It was also reviewe by an expert in methoology an statistics. The Deputy Eitor reviewe each revision of the article, an it unerwent a final review by the Eitor-in-Chief prior to publication. Final corrections an clarifications occurreuring one or more exchanges between the author(s) an copyeitors. Periprosthetic joint infection is a evastating complication of total knee an hip arthroplasty that is associate with substantial morbiity 1,2, mortality 3, an economic buren 4,5. The risk of infection following total joint arthroplasty is approximately 1% 4,6,7. Treatment guielines for periprosthetic joint infection recommen irrigation anebriement with polyethylene exchange for acute infections cause by low-virulence pathogens aroun well-fixe implants an two-stage revision Disclosure: None of the authors receive payments or services, either irectly or inirectly (i.e., via his or her institution), from a thir party in support of any aspect of this work. One or more of the authors, or his or her institution, has ha a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomeical arena that coul be perceive to influence or have the potential to influence what is written in this work. No author has ha any other relationships, or has engage in any other activities, that coul be perceive to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitte by authors are always provie with the online version of the article. J Bone Joint Surg Am. 2015;97:

2 1221 TABLE I Results of Univariate Analyses Comparing Baseline Characteristics Between Suppression an Non-Suppression Groups Variable Suppression Group (N = 92) Non-Suppression Group (N = 276) P Value Charlson comorbiity inex* 4 [3, 5] 4 [2, 5] 0.34 Age (yr) 63.7 ± ± BMI (kg/m 2 ) 33.6 ± ± Sex 0.90 Female 36 (39.1) 112 (40.6) Male 56 (60.9) 164 (59.4) Inex surgery 0.63 Irrigation anebriement with polyethylene 54 (58.7) 152 (55.1) exchange 2-stage revision 38 (41.3) 124 (44.9) No. of previous revisions* 1 [0, 3] 1 [0, 2] 0.37 Pathogen 0.33 S. aureus 44 (47.8) 114 (41.3) Non-S. aureus 48 (52.2) 162 (58.7) Joint 0.94 Knee 71 (77.2) 210 (76.1) Hip 21 (22.8) 66 (23.9) Duration of symptoms* (ays) 30 [7, 90] 14 [5, 45] Duration of intravenous antibiotic therapy* (wk) 6 [6, 6] 6 [6, 6] 0.17 Previous joint infection anywhere 41 (44.6) 130 (47.1) 0.76 Infecting organism class 0.21 Virulent 54 (58.7) 147 (53.2) Inolent# 31 (33.7) 55 (20.0) Fungal an aci-fast bacilli 0 1 (0.3) Miscellaneous an contaminants 5 (5.4) 22 (7.2) Multiple organisms 18 (19.6) 35 (12.7) 0.13 *The values are given as the meian with the 25th an 75th percentiles in brackets. The values are given as the mean an stanareviation. The values are given as the number of patients with the percentage in parentheses. Inclues S. aureus, Enterococcus, an gram-negative organisms. #Inclues coagulase-negative Staphylococcus an Propionibacterium species. for infections that are chronic, cause by a high-virulence pathogen, or aroun a loose component 8.Atwotosix-weekpostoperative course of pathogen-specific intravenous antibiotic therapy is also recommene in both cases 9. Despite optimal surgical an meical treatments, failure rates are still high, with five-year prosthetic survival rates ranging from 38.4% 10 to 64.7% 11. Chronic antibiotic suppression is an unproven metho that has been use in an attempt to increase the chance of retaining a functional prosthesis in certain patients, typically those who have unergone surgical treatment for periprosthetic joint infection but have a high risk of relapse an/or for whom the next surgical step woul be limb-threatening. There are no clearly efine criteria for increase risk of relapse, although common factors inclue a history of multiple joint infections, immunosuppression, comorbiities that preispose to periprosthetic joint infection, an a virulent pathogen 9,12. Caniates for chronic antibiotic suppression must also have the ability to tolerate sieeffects. The purpose of this stuy was to evaluate patients who unerwent irrigation anebriement with polyethylene exchange or two-stage revision in orer to (1) compare infectionfree prosthetic survivorship between patients who i an those who i not unergo subsequent chronic oral antibiotic suppression; (2) analyze infection-free survivorship stratifie by the type of surgery, infecting organism, an involve joint; an (3) etermine factors associate with failure of chronic suppression with oral antibiotics an buil a nomogram to preoperatively preict the probability of failure. Materials an Methos After institutional review boar approval was obtaine, we use a combination of Current Proceural Terminology (CPT)-4 an International Classification of Diseases, Ninth Revision (ICD-9) coes to ientify all revision total knee an hip arthroplasties performe from 1996 to 2010 at a single institution. Of 10,411 proceures that were ientifie, 7111 that were one for inications other than periprosthetic joint infection, 2510 that were not either an irrigation anebriement with polyethylene exchange or a complete two-stage revision,

3 1222 Fig. 1 Flowchart showing eligible cases of chronic antibiotic suppression an matche cases without suppression. I&D = irrigation anebriement, PJI = periprosthetic joint infection, SA = S. aureus, an C.I. = comorbiity inex. an 135 that were one for a conition that i not fulfill the criteria for periprosthetic joint infection 12 were exclue. This left 655 eligible proceures in 625 patients for the stuy (Fig. 1). Of the 655 proceures, 379 were two-stage revisions (240 knees an 139 hips) an 276 were irrigation anebriement proceures with polyethylene exchange (207 knees an sixty-nine hips). Chart review an obituary atabase searches were performe. All cases were comanage with the infectious iseases service at our hospital. Patients receive broa-spectrum intravenous antibiotics following the first stage of the two-stage revision or following irrigation an ebriement with polyethylene exchange. An organism-specific antibiotic regimen was starte after culture results were obtaine. One hunre an twenty patients receive chronic antibiotic suppression, efine as treatment with oral antibiotics for a minimum of six months following the initial course of intravenous antibiotics. As we are aware of no current specific inication for chronic antibiotic suppression, the ecision of whether to offer this treatment was iniviualize. Patients with intraoperative cultures that were positive for virulent pathogens receive antibiotic suppression if they ha a risk factor for reinfection (a history of multiple joint infections, previous faile surgery for periprosthetic joint infection, retaine implants an/or immunosuppression). Patients with less virulent pathogens or negative cultures receive antibiotic suppression if they ha multiple risk factors for reinfection. The patients were contacte by telephone an questione regaring compliance with the antibiotic therapy, aitional joint surgery, joint pain, anrainage. Outcomes aneath rates were recore. Exclusion criteria inclue non-compliance (n = 9), less than one year of followup (n = 12), an initiation of oral antibiotic therapy after the onset of new rainage (n = 7). This resulte in ninety-two patients eligible for the stuy. The non-suppression group was selecte from the remaining pool of 505 patients in whom a total of 535 periprosthetic joint infections ha been manage without chronic antibiotic suppression. In an effort to manage the inherent selection bias, the suppression-group patients were matche with similar, non-suppressiongroup patients. Propensity score matching was use for age, sex, boy mass inex (BMI), Charlson comorbiity inex, number of previous operations on the affecte joint, affecte joint (hip or knee), type of surgery (irrigation anebriement with polyethylene exchange versus two-stage revision) preceing the antibiotic suppression, an infecting organism (Staphylococcus aureus versus non-s. aureus). Although the Charlson comorbiity inex was evelope to preict mortality within one year after hospital amission 13,14, it has been shown to be an inepenent preictor of periprosthetic joint infection in both the hip 15 an the knee 16. Propensity scores were estimate on the scale of the log-os base on the above variables with use of a logistic regression moel. Patients in the suppression group were matche to those in the non-suppression group with the closest propensity scores without replacement in a ratio of 1:3 (suppression:non-suppression). The covariate balance resulting from propensity score matching was checke with use of stanarizeifferences, variance ratios of propensity scores, an variance ratios of resiuals orthogonal to propensity scores for each covariate. Balance was foun to be aequate without restricting which patients were matche by enforcing a matching caliper. The primary outcome variable was infection-free prosthetic survival, with aitional surgery ue to infection or eath as the en points. All patients were inclue in the analysis even if the time to the outcome was less than one year. Further stratification on the basis of the affecte joint, type of inex surgery, an infecting organism was performe. Treatment failure was the outcome variable use to compare the patients in the suppression group for whom the

4 1223 TABLE II Intravenous an Oral Antibiotic Regimens for the Suppression Group Accoring to Pathogen* Pathogen (No. of Patients) IV Antibiotic Regimen [Ajunctive PO Antibiotic Regimen] (No. of Patients) Chronic Suppressive Antibiotic Regimen (No. of Patients) Methicillin-sensitive Staphylococcus aureus (31) Oxacillin 2 g qi or q4h (14) Vancomycin 1 or 1.5 or 2 g bi or q or q48h (8) Cefazolin 1 or 2 g ti (4) Imipenem 500 mg qi (1) Daptomycin 500 mg q (1) Ceftriaxone 2 g q (1) Ampicillin/sulbactam 2/1 g qi (1) Dicloxacillin 500 mg qi or ti or bi (11) Doxycycline 100 mg qi or bi or q (7) Cephalexin 500 mg qi or ti or bi (6) Trimethoprim/sulfamethoxazole 160/800 mg bi (5) Minocycline 100 mg q (1) Amoxicillin 500 mg ti (1) Clinamycin 300 mg bi (1) Clinamycin 900 mg ti (1) Piperacillin/tazobactam 3/0.375 g ti (1) Methicillin-resistant coagulase-negative staphylococci (21) Vancomycin 1 or 1.25 or 1.5 g bi or q (14) Daptomycin 500 or 650 mg q (3) Oxacillin 2 g qi (1) Clinamycin 600 mg ti (1) Linezoli 600 mg bi (1) Tigecycline 50 mg bi (1) Doxycycline 100 mg bi or q (12) Rifampin 300 mg bi (2) Trimethoprim/sulfamethoxazole 160/800 mg bi or q (4) Clinamycin 300 mg bi (2) Dicloxacillin 500 mg qi (2) Cephalexin 250 mg bi (1) Methicillin-resistant Staphylococcus aureus (13) Vancomycin 1 or 1.25 or 1.5 g bi or q (8) Daptomycin 350 or 500 mg q (2) Oxacillin 2 g qi (1) Linezoli 600 mg bi (1) Doxycycline 100 mg q (8) Trimethoprim/sulfamethoxazole 160/800 mg bi (3) Clinamycin 300 mg bi (1) Erythromycin 400 mg bi (1) Doxycycline 100 mg bi (1) Methicillin-sensitive coagulase-negative staphylococci (12) Vancomycin 1 or 1.25 or 1.5 g bi or q (8) Oxacillin 1 or 2 g q4h or qi (2) Ceftriaxone 2 g q (1) Daptomycin 600 mg q (1) Doxycycline 100 mg bi or q (8) Dicloxacillin 250 or 500 mg ti (3) Cephalexin 500 mg ti (1) Cefaroxil 500 mg bi (1) Linezoli 600 mg bi (1) Enterococci (5) Ampicillin 2 g q4h (2) Penicillin 1 million U qi (1) Tigecycline 50 mg bi (1) Amoxicillin 500 mg ti (3) Doxycycline 100 mg bi (1) Ciprofloxacin 500 mg bi (1) Vancomycin 1 g q (1) Virians streptococci (5) Ceftriaxone 2 g q (2) Piperacillin/tazobactam 3/0.375 g qi (1) Clinamycin 900 mg ti (1) Cefazolin 1 g ti (1) Amoxicillin 500 mg ti (2) Trimethoprim/sulfamethoxazole 160/800 mg bi (2) Moxifloxacin 400 mg q (1) Doxycycline 100 mg bi (1) Group-B streptococci (3) Penicillin G 2 million U q4h (1) Ampicillin 2 g qi (1) Amoxicillin 500 mg ti (2) Dicloxacillin 250 mg ti (1) Vancomycin 1 g bi (1) Diphtheroi-like bacilli (2) Vancomycin 1 g bi [ciprofloxacin 500 mg bi (1) Doxycycline 100 mg bi or q (2) or rifampin 300 mg PO bi (1)] (2) Enterobacteriaceae (2) Imipenem 500 mg qi [ciprofloxacin 750 mg bi] (1) Ciprofloxacin 750 mg bi (2) Ciprofloxacin 750 mg bi (1) Propionibacterium acnes (2) Vancomycin 1 or 1.5 g bi or q (2) Doxycycline 100 mg bi or q (2) Pseuomonas aeruginosa (1) Imipenem 500 mg qi [ciprofloxacin 750 mg bi] (1) Ciprofloxacin 500 mg bi (1) Staphylococcus lugunensis (1) Oxacillin 2 g q4h [rifampin 300 mg bi] (1) Cephalexin 500 mg ti (1) Ureaplasma species (1) Ceftaziime 2 g q (1) Azithromycin 500 mg q (1) Negative culture (1) Vancomycin 1 g bi (1) Doxycycline 100 mg bi (1) *IV = intravenously, PO = orally, qi = four times a ay, q4h = every four hours, ti = three times a ay, bi = twice a ay, q = every ay, an q48h = every fortyeight hours. The number of pathogens as up to more than the total number of ninety-two patients because some patients ha polymicrobial infections. Also, the number of antibiotic regimens may a up to more than the total number of pathogens because some patients ha multiple regimens at once.

5 1224 TABLE III Intravenous Antibiotic Regimens for the Non-Suppression Group Accoring to Pathogen* Pathogen (No. of Patients) IV Antibiotic Regimen [Ajunctive PO Antibiotic Regimen] (No. of Patients) Methicillin-sensitive Staphylococcus aureus (61) Oxacillin 2 g q4h or qi [rifampin 300 mg bi (9)] (25) Vancomycin 500 mg or 1 or 1.25 g bi or q [rifampin 300 or 600 mg bi (2)] (16) Cefazolin 1 or 2 g bi or ti (9) Penicillin G 3 million U q4h [rifampin 300 mg bi (1)] (3) Daptomycin 600 mg q (2) Piperacillin/tazobactam 3/0.375 g ti (1) Ceftriaxone 1 g q (1) Amoxicillin 1 g ti (1) Linezoli 600 mg bi (1) Clinamycin 900 mg ti (1) Ampicillin/sulbactam 2/1 g qi (1) Doxycycline 100 mg bi (1) Methicillin-resistant Staphylococcus Vancomycin 500 mg or 1 or 1.25 or 1.5 g bi or q or q48h or q72h aureus (53) [rifampin 300 mg bi (14)] (45) Piperacillin/tazobactam 3/0.375 g ti (3) Piperacillin 4 g ti (1) Ceftriaxone 1 g q (1) Oxacillin 2 g q4h (1) Ampicillin 2 g qi (1) Linezoli 600 mg bi (1) Cefazolin 2 g ti (1) Methicillin-sensitive coagulase-negative Vancomycin 1 or 1.25 or 1.5 g bi or q or q48h [rifampin 300 mg bi (4)] (29) staphylococci (34) Cefazolin 1 or 2 g ti (2) Oxacillin 2 g q4h (1) Daptomycin 600 mg q (1) Piperacillin/tazobactam 3/0.375 g qi (1) Methicillin-resistant coagulase-negative Vancomycin 750 mg or 1 or 1.25 or 1.5 g bi or q [rifampin 300 or 600 mg bi (2)] (24) staphylococci (29) Clinamycin 600 mg ti (2) Daptomycin 500 mg q (2) Aztreonam 1 g q8h (1) Diphtheroi-like bacilli (9) Vancomycin 750 mg or 1 or 1.5 g bi or q or q48h [ciprofloxacin 750 mg bi (6)] (8) Gentamycin 50 mg q (1) [Ciprofloxacin 500 mg bi (1)] [Azithromycin 500 mg q (1)] Enterococci (8) Vancomycin 1.5 g bi (4) Ampicillin 2 g qi (2) Piperacillin/tazobactam 3/0.375 g qi [ciprofloxacin 750 mg bi (1)] (2) Pseuomonas aeruginosa (8) Piperacillin/tazobactam 3/0.375 g qi or ti or bi [ciprofloxacin 500 mg bi (1)] (7) Daptomycin 570 mg q [ciprofloxacin 500 mg bi] (1) Virians streptococci (6) Penicillin G 3 million U q4h (2) Vancomycin 1 g bi or q (2) Ampicillin 2 g qi (1) Ceftriaxone 1 g q (1) Group-B streptococci (6) Vancomycin 1 g bi [rifampin 300 mg bi (1)] (2) Oxacillin 2 g q4h (1) Ceftriaxone 2 g q (1) Penicillin G 3 million U ti (1) continue

6 1225 TABLE III (continue) Pathogen (No. of Patients) IV Antibiotic Regimen [Ajunctive PO Antibiotic Regimen] (No. of Patients) Escherichia coli (5) Vancomycin 1 g bi [ciprofloxacin 500 mg bi (2)] (2) Ertapenem 1 g q [metroniazole 250 mg ti] (1) [Levofloxacin 750 mg q (1)] [Ciprofloxacin 500 mg bi (1)] Unspecifie b-hemolytic streptococci (5) Vancomycin 1 or 1.5 g q (3) Clinamycin 900 mg ti (1) Penicillin G 4 million U ti (1) Klebsiella pneumoniae (4) Tigecycline 50 mg bi (2) Piperacillin/tazobactam 3/0.375 g ti (1) Daptomycin 570 mg q [ciprofloxacin 500 mg bi] (1) Propionibacterium acnes (3) Vancomycin 1 g q (2) Ampicillin 2 g ti (1) Serratia marcescens (3) Vancomycin 1 g bi [ciprofloxacin 750 mg bi (1)] (3) Gentamycin 50 mg q (2) Peptostreptococcus species (3) Oxacillin 2 g qi [rifampin 300 mg bi] (1) Vancomycin 1.25 g bi (1) Cefazolin 2 g bi (1) Streptococcus pneumoniae (2) Vancomycin 1.25 g bi (1) Linezoli 300 mg bi (1) Group-C streptococci (2) Vancomycin 1 g bi (2) Group-A streptococci (1) Vancomycin 1 g bi (1) Group-G streptococci (1) Ampicillin 2 g qi (1) Enterobacteriaceae (1) Ceftriaxone 1 g q (1) Bacteroies fragilis (1) Metroniazole 250 mg ti (1) Cania parapsilosis (1) [Fluconazole 400 mg q (1)] Proteus mirabilis (1) Ceftaziime 1 g q (1) Morganella morganii (1) Ciprofloxacin 400 mg q (1) Staphylococcus lugunensis (1) Oxacillin 2 g q4h [rifampin 600 mg q] (1) Negative culture (32) Vancomycin 1 or 1.5 g bi or q or q48h [rifampin 300 mg bi (1) or ciprofloxacin 500 mg bi (1)] (12) Cefazolin 1 g ti (7) Oxacillin 2 g q4h [rifampin 300 mg bi (1)] (2) Daptomycin 450 mg q (2) Vancomycin 1 g bi [ciprofloxacin 500 mg bi] (1) Piperacillin/tazobactam 3/0.375 g ti (1) Ticarcillin/clavulanate 3/0.1 g q (1) Ciprofloxacin 400 mg q (1) Doxycycline 100 mg bi (1) Penicillin G 4 million U q4h (1) Clinamycin 900 mg ti (1) Ceftriaxone 2 g q (1) [Levofloxacin 500 mg q (1)] *IV = intravenously, PO = orally, q4h = every four hours, qi = four times a ay, bi = twice a ay, q = every ay, ti = three times a ay, q48h = every forty-eight hours, q72h = every seventy-two hours, an q8h = every eight hours. The number of pathogens as up to more than the total number of 276 patients because some patients ha polymicrobial infections. Also, the number of antibiotic regimens may a up to more than the total number of pathogens because some patients ha multiple regimens at once.

7 1226 TABLE IV Cox Proportional Hazars Moel Estimates of Survival, with Ajustment for Matching Covariates Variable HR 95% CI P Value Chronic suppressive antibiotics <0.001 No. of previous revisions Non-S. aureus infection Age (per year) Hip joint Charlson comorbiity inex (per inex point) Male sex BMI (per inex point) suppression faile with those for whom it i not fail. Failure was efine, as escribe by Diaz-Leezma et al. 17, as (1) subsequent surgical intervention for infection after the inex proceure; (2) persistent fistula, rainage, or joint pain at the last follow-up visit; or (3) eath relate to the periprosthetic joint infection. Any unresolverainage at the last follow-up visit was consiere a failure. Pain was consiere to inicate a failure only when it was severely ebilitating an prevente any kin of walking at the time of the last follow-up. Because we coul not etermine the causes of most eaths, we consiereeath within one year after the surgery as the cutoff point for failure; this was base on the finings of Zmistowski et al. 3, who showe that the highest ifferential mortality between septic an aseptic revisions occurre within one year. The Fisher exact test or Pearson chi-square test was use for categorical ata, whereas the Wilcoxon rank sum test or Welch two-sample t test was use for continuous ata. Infection-free survival was estimate by using the Kaplan-Meier metho. Hazar ratios (HRs) for comparison between the suppression an nonsuppression groups were calculate with use of Cox proportional hazars regression moels, with ajustment for variables that were previously use for matching. Subset analyses were performe by selecting the matche groups in which all subjects belongetothesubsetofinterest.thiswasonetopreservethematchingcharacteristics of the original ata, an the propensity score balance was rechecke within these subsets to verify that this ha been accomplishe. A log-rank test was use to compare prosthetic survival between the suppression an non-suppression groups in the subset cohorts. Further moeling was performe with Cox proportional hazars moelstoverifytheeffectofchronicantibiotictherapyacrosstheifferentsubsets. This moel compare the benefits of suppression between the subsets of infecting organism (methicillin-resistant S. aureus [MRSA] versus methicillin-sensitive S. aureus [MSSA] versus non-s. aureus), affecte joint (hip versus knee), an type of surgery (irrigation anebriement with polyethylene exchange versus two-stage revision). Failures within the suppression group were moele as a function of clinically relevant variables, with use of a penalize logistic regression moel 18.Asa result of a low ratio of events to parameters, this moel inclue only three variables an use a rige-type penalty to correct for possible overfitting. A significance level of 5% was use, an analyses were one with use of R software (version 3.0.2; R Founation for Statistical Computing, Vienna, Austria). The penalize moel was fit with use of rms (regression moeling strategies) in the R package 19. of the patients in the suppression group an 186 (67.4%) of those in the non-suppression group ha ha previous revisions (mean, 2.7 ± 1.4 an 2.4 ± 1.7 previous revisions, respectively) for any cause. With the exception of two knees an one hip (all in the non-suppression group), which ha a resection arthroplasty, all knees were treate with static spacers an all hips were treate with articulate spacers at the time of explantation. The time interval between removal an reimplantation was variable, averaging 21.1 ± 9.7 weeks (range, 6.6 to 46.3 weeks) in the suppression group an 15 ± 8 weeks (range, 3.1 to 45.6 weeks) in the non-suppression group. The meian uration of postoperative intravenous therapy with antibiotics was six weeks in both groups. The mean uration of oral antibiotic suppression was 63.5 ± 38.3 months (range, six to months). The specific antibiotic regimens are epicte in Tables II an III. The five-year infection-free prosthetic survival rate was 68.5% (95% confience interval [CI] = 59.2% to 79.3%) for the antibiotic-suppression group compare with 41.1 % (95% CI = 34.9% to 48.5%) for the non-suppression group (HR = 0.63, p = 0.008) (Fig. 2). The Cox proportional hazars moel showe Source of Funing This stuy was internally fune by our institution. Results The baseline characteristics of the suppression an nonsuppression groups are epicte in Table I. The mean uration of follow-up (an stanareviation) was 69.1 ± 38.2 months (range, 2.2 to months) in the suppression group an 41.6 ± 40.2 months (range, less than one to 183 months in the non-suppression group). Sixty-two (67.4%) Fig. 2 Kaplan-Meier infection-free prosthetic survival curves for suppression an non-suppression patient groups. The blue line represents the suppression group, an the re line represents the non-suppression group; the shae areas surrouning the lines represent the 95% CI.

8 1227 TABLE V Comparisons of the Effects of Suppression Across Groups Stratifie by Infecting Organism, Affecte Joint, an Type of Surgery Subset Interaction HR 95% CI Methicillin-sensitive S. aureus vs. non-s. aureus Methicillin-resistant S. aureus vs. non-s. aureus Methicillin-resistant vs. methicillin-sensitive S. aureus Hip vs. knee Irrigation anebriement with polyethylene exchange vs. 2-stage revision that a higher number of previous revisions preicte a ecrease survival rate (HR = 1.12 [95% CI = 1.04 to 1.21]; p = 0.005) an a non-s. aureus infection preicte an increase survival rate (HR = 0.69 [95% CI = 0.51 to 0.94]; p = 0.018) (Table IV). Stratification by the type of inex surgery showe that, in the group that unerwent irrigation anebriement with polyethylene exchange, the patients treate with chronic antibiotic suppression ha an increase five-year infection-free survival rate(64.7% [95% CI = 49.7% to 77.3%]) compare with the non-suppression group(30.4%[95%ci= 22.4% to 39.6%]; p < ). There was no ifference in survival between the suppression an nonsuppression groups following two-stage revision (p = 0.14) (Fig. 3). Fig. 3 Kaplan-Meier infection-free prosthetic survival curves for subset cohorts. The blue line represents the suppression group, an the re line represents the non-suppression group; the shae areas surrouning the lines represent the 95% CI. Fig. 3-A Iniviuals who unerwent irrigation anebriement with polyethylene exchange (p < for the ifference between the suppression an non-suppression groups). Fig. 3-B Iniviuals who unerwent a two-stage revision (p = 0.14). Fig. 3-C Iniviuals with an S. aureus infection (p = 0.047). Fig. 3-D Iniviuals with a non-s. aureus infection (p = 0.62). Fig. 3-E Iniviuals with an infection in the hip (p = 0.001). Fig. 3-F Iniviuals with an infection in the knee (p = 0.01).

9 1228 TABLE VI Results of Univariate Analyses Comparing Baseline Characteristics Between Patients for Whom Chronic Oral Antibiotic Suppression Di Not Fail an Those for Whom It Faile Variable No Failure (N = 60) Failure (N = 32) P Value Age at revision* (yr) ± ± Sex Male 34 (56.7) 22 (68.8) Female 26 (43.3) 10 (31.3) BMI* (kg/m 2 ) ± ± Inex surgery >0.99 Irrigation anebriement with polyethylene exchange 35 (58.3) 19 (59.4) 2-stage revision 25 (41.7) 13 (40.6) Duration of intravenous antibiotic therapy (wk) 6 [6, 6] 6 [6, 6] 0.26 Duration of symptoms (ays) 30 [6, 77.5] 28 [13.25, 83.75] 0.23 Onset of infection 0.70 Early 24 (40.0) 12 (37.5) Late 36 (60.0) 20 (62.5) Joint Knee 41 (68.3) 30 (93.8) Hip 19 (31.7) 2 (6.3) No. of previous revisions* 1.47 ± ± Infecting organism class 0.76 Virulent 35 (58.3) 19 (59.4) Inolent# 19 (31.7) 12 (37.5) Fungal an aci-fast bacilli 0 0 Miscellaneous an contaminants 4 (6.7) 1 (3.1) Multiple organisms 10 (16.7) 9 (28.8) 0.27 Pathogen 0.93 S. aureus 28 (46.7) 16 (50.0) Non-S. aureus 32 (53.3) 16 (50.0) Duration of antibiotic suppression* (mo) ± ± Charlson comorbiity inex* 4.00 ± ± American Society of Anesthesiologists score 3 [3, 3] 3 [2.75, 3] 0.89 Smoker 4 (6.7) 3 (9.4) 0.69 Diabetes 15 (25.0) 11 (34.4) 0.48 Inflammatory arthropathy 10 (16.7) 1 (3.1) Malignancy 7 (11.7) 2 (6.3) 0.49 Steroi use 9 (15.0) 2 (6.3) 0.32 Heart isease 47 (78.3) 24 (75.0) 0.92 *The values are given as the mean an stanareviation. The values are given as the number of patients with the percentage in parentheses. The values are given as the meian with the 25th an 75th percentiles in brackets. Inclues S. aureus, Enterococcus, an gram-negative organisms. #Inclues coagulase-negative Staphylococcus an Propionibacterium species. Stratification by the affecte joint showe a significant increase in the infection-free survival rate in the suppression group, compare with the non-suppression group, both after revision for a hip infection (87.2% [95% CI = 60.2% to 96.8%] compare with 49.0% [95% CI = 32.9% to 65.4%]; p = 0.001) an after revision for a knee infection (65.8% [95% CI = 53.0% to 76.7%] compare with 43.2% [95% CI = 35.4% to 51.4%]; p = 0.01). Stratification by the infecting organism showe that, when the pathogen was S. aureus, the five-year infection-free survival rate in the suppression group (57.4% [95% CI = 39.2% to 73.8%]) was increase compare with that in the non-suppression group

10 1229 between the suppression group with MSSA infection (63.6% [95% CI = 45.1% to 78.8%]) an the non-suppression group with MSSA infection (48.1% [95% CI = 34.4% to 62.1%]; p = 0.09). The suppression-group patients with gram-negative infection ha an increase infection-free survival rate (74.7% [95% CI = 36.7% to 94.1%]) compare with the non-suppression group with gram-negative infection (20.3% [95% CI = 10.6% to 39.2%]; p = 0.04). (40.1% [95% CI = 29.6% to 51.6%]; p = 0.047). There was no ifference in survival between the suppression an nonsuppression groups when the patients ha a non-s. aureus infection (p = 0.62). The results of aitional stratification, by whether the patient ha MRSA, MSSA, or gram-negative infection, are shown in Figure 4, although this analysis i not preserve the original matching quality between the suppression an non-suppression groups. The benefit of chronic antibiotic suppression compare across ifferent subsets was non-significant for the entire moel (p = 0.22) an at every interaction (Table V). Thirty-two patients (34.8%) in the suppression group an 115 (41.7%) in the non-suppression group met the criteria for failure of treatment. Comparisons between the patients for whom chronic antibiotic suppression faile an those for whom it i not fail are epicte in Table VI. Failures ha a lower association with hip infection than with knee infection (HR = 0.10, p = 0.013) an were associate with a higher number of prior revisions (HR = 2.83, p = 0.026) (Table VII). Fig. 4 Kaplan-Meier infection-free prosthetic survival curves for subsets with MRSA (Fig. 4-A), MSSA (Fig. 4-B), an gram-negative (Fig. 4-C) infection. The blue line represents the suppression group, an the re line represents the non-suppression group; the shae areas surrouning the lines represent the 95% CI. Suppressive antibiotics increase the five-year infection-free prosthetic survival rate for patients with MRSA infection, with a survival rate of 50.8% (95% CI = 22.2% to 78.9%) in the suppression group an 28.2% (95% CI = 16.5% to 43.8%) in the non-suppression group (p = 0.05). There was no ifference in the infection-free survival rate Discussion With the rates of total knee an hip arthroplasties performe in the Unite States expecte to increase by 673% an 174%, respectively, from 2005 to , strategies to avoi limbthreatening outcomes an high mortality rates associate with periprosthetic joint infection are neee. In the current stuy, we examine how chronic suppressive antibiotics affect infectionfree prosthetic survival rates after surgical intervention for periprosthetic joint infection. The five-year infection-free survival rate in the suppression group was 68.5% compare with 41.1% in the non-suppression group (p = 0.008). The increase survival rate in the suppression group is consistent with the finings in another stuy, which showe a three-year success rate of 78% after treatment of 112 periprosthetic joint infections with ebriement, antibiotics, an implant retention 21. Other stuies ocumenting the management of periprosthetic joint infection with antibiotics an prosthesis retention have emonstrate both positive 22,23 an negative 24 outcomes; however, subsequent chronic suppression with oral antibiotics was not consistently use. Stratification by the type of surgery an infecting organism showe that the patients who benefite the most from chronic suppressive antibiotic therapy were those who unerwent irrigation anebriement with polyethylene exchange an those with S. aureus infection. The benefit of suppression following irrigation anebriement with polyethylene exchange has alreay been ocumente an is in accorance with finings in the current literature Persistence of a latent infection is common

11 1230 TABLE VII Penalize Regression Moel Estimates of Factors Associate with Treatment Failure Variable HR 95% CI P Value No. of previous revisions* BMI Hip infection *Effect for a three-unit change. Effect for a 12.7-unit change. in patients with retaine implants an thus antibiotic suppression seems to be a reasonable alternative that avois the nee for a more invasive two-stage revision. Antibiotic suppression following two-stage proceures i not seem to affect prosthetic survival. The current stuy also showe that, among the patients who ha S. aureus infection, those who unerwent postoperative suppression ha a higher infection-free survival rate than those who i not receive postoperative suppression. This conflicts with the finings in the stuy by Brant et al. 28,in which suppression i not influence the outcome of surgical treatment of periprosthetic joint infections cause by S. aureus. Possible explanations for this iscrepancy are the newer, more efficacious antibiotic regimens use in the present stuy, the longer course of intravenous antibiotics postoperatively (six weeks versus four weeks in the stuy by Brant et al.), an ifferences in the en points that were use. Furthermore, it has been suggeste that S. aureus survives in the intracellular environment 29, which may account for its higher recurrence rate an thus the nee for a long-term oral antibiotic regimen. Finally, chronic antibiotic therapy i not seem to influence infection-free survival after revisions for non-s. aureus infections. This was probably ue to the greater success attaine with the surgery an by the increase susceptibility of these pathogens to intravenous agents use in the first six weeks after surgery. Numerous stuies have shown the efficacy of chronic suppression even of non-s. aureus infections 25-27, but the absence of a control group in those stuies preclues any efinite conclusions from being reache. TABLE VIII Finings in Previous Stuies of Chronic Antibiotic Suppression Stuy No. Mean Duration of Follow-up (mo) Mean Duration of Antibiotic Suppression (Range) (mo) Previous Surgical Treatment Failure Criteria No. (%) of Successful Cases Johnson an Bannister 30 (1986) Goulet et al. 27 (1988) Tsukayama et al. 31 (1991) Segreti et al. 25 (1998) Rao et al. 26 (2003) Current stuy 25 knees (1.2-59) Excision of sinus tract, ebriement, exchange arthroplasty 19 hips (2-120) Incision anrainage (11), no prior surgery (8) 8 knees, 5 hips 12 knees, 6 hips 19 knees, 15 hips, 2 elbows 71 knees, 21 hips Persistent ischarge, joint pain Removal or revision of prosthesis, increasing symptoms of infection (6-48) Surgical ebriement Prosthesis removal for recurrent infection (4-103) Surgical ebriement Persistence of symptoms of infection (pain, rainage, etc.) (6-128) Surgical ebriement Development of progressive pain, loosening of implant, or rainage (6-165) Irrigation anebriement with polyethylene exchange (54), 2-stage revision (38) Aitional surgery ue to infection, persistent rainage or joint pain, eath within 1yr 2 (8%) 9 (47%) 3 (23%) 15 (83%) 31 (86%) 60 (65%)

12 1231 The multivariate moels constructe for our stuy confirme the overall benefit of chronic antibiotic therapy. However, this analysis was not able to etermine the subset from which this benefit was mostly erive, probably because of the limite number of events in each subset an overfitting of the ata within the moel. The 34.8% failure rate for the ninety-two patients who receive chronic antibiotic therapy for a minimum of six months was higher than the rates in the most recently publishe stuies 25,26. Segreti et al. 25 reviewe the cases of eighteen patients who ha unergone surgical ebriement an six to eight weeks of intravenous antibiotics followe by prolonge antibiotic suppression. At five years, only three ha persistent symptoms of infection. Rao et al. 26 prospectively examine thirty-six patients who ha unergone the same protocol of surgical ebriement an four to six weeks of intravenous antibiotics prior to antibiotic suppression. The failure rate was 13.9% (five of the thirty-six patients) after 4.4 years of follow-up. Possible explanations for the higher failure rates observe in the present stuy inclue the higher prevalence of invasive pathogens (47.8% of the infections were cause by S. aureus compare with 38.9% in the other stuies 25,26 ) an the inclusion of eaths within the first year after treatment as failures. Oler stuies on chronic antibiotic suppression 27,30,31 showe higher failure rates, probably as a result of less efficient antibiotic regimens (Table VIII). In our stuy, patients for whom suppressive antibiotic treatment faile ha ha more prior joint revisions. The association of a higher number of previous revisions with increase failure rates is in accorance with the finings in other stuies 32,33. We also foun that patients with knee periprosthetic joint infection ha a higher rate of failure than those with hip infection. In a review of 9245 cases, Pulio et al. 6 reporte an increase prevalence of periprosthetic joint infection after total knee arthroplasty than after total hip arthroplasty. This oes not explain why knees have an inferior response to treatment of periprosthetic joint infection, but it provies some evience of an increase susceptibility to periprosthetic joint infection in knees. Our stuy was limite by its sample size. This is a common limitation of stuies of chronic antibiotic suppression for periprosthetic joint infection because it is a last-resort treatment for an infrequent complication of joint arthroplasty. The small sample size compromise the ability of the multivariate analyses stratifie by the infecting organism, affecte joint, an type of surgery to provie results similar to those in the survivorship analysis. In aition, sampling bias was an inherent limitation of our stuy when the suppression group was compare with the non-suppression group. In an effort to overcome this bias, propensity score matching with a 3:1 matching ratio was one for eight variables. Also, the retrospective nature of the stuy preclue the use of stanarize surgical techniques an antibiotic regimens. Finally, the six subset cohorts ha lower-quality propensity score matching than the cohort as a whole. This was the case because the matching was esigne to prouce goo matches for all subjects rather than within subsets. In summary, chronic suppression with oral antibiotics resulte in superior infection-free survival rates after surgical treatment for periprosthetic joint infection compare with those observe without suppression. Patients who unerwent irrigation anebriement with polyethylene exchange an those with S. aureus infection showe the greatest benefit. Knee infection an a greater number of prior revisions were ientifie as variables associate with treatment failure. n Marcelo B.P. Siqueira, MD Anas Saleh, MD Alison K. Klika, MS Colin O Rourke, MS Steven Schmitt, MD Carlos A. Higuera, MD Wael K. Barsoum, MD Departments of Orthopaeic Surgery (M.B.P.S., A.S., A.K.K., C.A.H., an W.K.B.), Quantitative Health Sciences (C.O R.), an Infectious Diseases (S.S.), Clevelan Clinic Founation, Clevelan, OH aress for M.B.P. Siqueira: bogliom@ccf.org. aress for A. Saleh: anas.ar.saleh@gmail.com. aress for A.K. Klika: klikaa@ccf.org. aress for C. O Rourke: orourkc@ccf.org. aress for S. Schmitt: schmits@ccf.org. aress for C.A. Higuera: higuerc@ccf.org. aress for W.K. Barsoum: barsouw@ccf.org References 1. Darouiche RO. Treatment of infections associate with surgical implants. N Engl J Me Apr 1;350(14): Bozic KJ, Ries MD. The impact of infection after total hip arthroplasty on hospital an surgeon resource utilization. J Bone Joint Surg Am Aug;87 (8): Zmistowski B, Karam JA, Durinka JB, Casper DS, Parvizi J. Periprosthetic joint infection increases the risk of one-year mortality. J Bone Joint Surg Am Dec 18;95(24): Kurtz SM, Lau E, Schmier J, Ong KL, Zhao K, Parvizi J. Infection buren for hip an knee arthroplasty in the Unite States. J Arthroplasty Oct;23(7): Epub 2008 Apr Klouche S, Sariali E, Mamouy P. Total hip arthroplasty revision ue to infection: a cost analysis approach. Orthop Traumatol Surg Res Apr;96 (2): Pulio L, Ghanem E, Joshi A, Purtill JJ, Parvizi J. Periprosthetic joint infection: the incience, timing, an preisposing factors. Clin Orthop Relat Res Jul;466 (7): Epub 2008 Apr Phillips JE, Crane TP, Noy M, Elliott TS, Grimer RJ. The incience of eep prosthetic infections in a specialist orthopaeic hospital: a 15-year prospective survey. J Bone Joint Surg Br Jul;88(7): Parvizi J, Aeli B, Zmistowski B, Restrepo C, Greenwal AS. Management of periprosthetic joint infection: the current knowlege: AAOS exhibit selection. J Bone Joint Surg Am Jul 18;94(14):e Osmon DR, Berbari EF, Berent AR, Lew D, Zimmerli W, Steckelberg JM, Rao N, Hanssen A, Wilson WR; Infectious Diseases Society of America. Diagnosis an management of prosthetic joint infection: clinical practice guielines by the Infectious Diseases Society of America. Clin Infect Dis Jan;56(1):e1-25. Epub 2012 Dec Buller LT, Sabry FY, Easton RW, Klika AK, Barsoum WK. The preoperative preiction of success following irrigation anebriement with polyethylene exchange for hip an knee prosthetic joint infections. J Arthroplasty Jun;27 (6): e1-4. Epub 2012 Mar Sabry FY, Buller L, Ahme S, Klika AK, Barsoum WK. Preoperative preiction of failure following two-stage revision for knee prosthetic joint infections. J Arthroplasty Jan;29(1): Epub 2013 May 23.

13 Parvizi J, Gehrke T, Chen AF. Proceeings of the International Consensus on Periprosthetic Joint Infection. Bone Joint J Nov;95-B(11): Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new metho of classifying prognostic comorbiity in longituinal stuies: evelopment an valiation. J Chronic Dis. 1987;40(5): Charlson M, Szatrowski TP, Peterson J, Gol J. Valiation of a combine comorbiity inex. J Clin Epiemiol Nov;47(11): Soohoo NF, Farng E, Lieberman JR, Chambers L, Zingmon DS. Factors that preict short-term complication rates after total hip arthroplasty. Clin Orthop Relat Res Sep;468(9): SooHoo NF, Lieberman JR, Ko CY, Zingmon DS. Factors preicting complication rates following total knee replacement. J Bone Joint Surg Am Mar;88(3): Diaz-Leezma C, Higuera CA, Parvizi J. Success after treatment of periprosthetic joint infection: a Delphi-base international multiisciplinary consensus. Clin Orthop Relat Res Jul;471(7): Epub 2013 Feb Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic moels: issues in eveloping moels, evaluating assumptions an aequacy, an measuring an reucing errors. Stat Me Feb 28;15(4): Harrel FE Jr. rms: regression moeling strategies. R package version Accesse 2014 June Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary an revision hip an knee arthroplasty in the Unite States from 2005 to J Bone Joint Surg Am Apr;89(4): Byren I, Bejon P, Atkins BL, Angus B, Masters S, McLary-Smith P, Gunle R, Berent A. One hunre an twelve infecte arthroplasties treate with DAIR (ebriement, antibiotics an implant retention): antibiotic uration an outcome. J Antimicrob Chemother Jun;63(6): Epub 2009 Mar Aboltins CA, Page MA, Buising KL, Jenney AW, Daffy JR, Choong PF, Stanley PA. Treatment of staphylococcal prosthetic joint infections with ebriement, prosthesis retention an oral rifampicin an fusiic aci. Clin Microbiol Infect Jun;13 (6): Epub 2007 Feb Mont MA, Walman B, Banerjee C, Pacheco IH, Hungerfor DS. Multiple irrigation, ebriement, an retention of components in infecte total knee arthroplasty. J Arthroplasty Jun;12(4): Deirmengian C, Greenbaum J, Lotke PA, Booth RE Jr, Lonner JH. Limite success with open ebriement an retention of components in the treatment of acute Staphylococcus aureus infections after total knee arthroplasty. J Arthroplasty Oct;18(7)(Suppl 1): Segreti J, Nelson JA, Trenholme GM. Prolonge suppressive antibiotic therapy for infecte orthopeic prostheses. Clin Infect Dis Oct;27(4): Rao N, Crossett LS, Sinha RK, Le Frock JL. Long-term suppression of infection in total joint arthroplasty. Clin Orthop Relat Res Sep;(414): Goulet JA, Pellicci PM, Brause BD, Salvati EM. Prolonge suppression of infection in total hip arthroplasty. J Arthroplasty. 1988;3(2): Brant CM, Sistrunk WW, Duffy MC, Hanssen AD, Steckelberg JM, Ilstrup DM, Osmon DR. Staphylococcus aureus prosthetic joint infection treate with ebriement an prosthesis retention. Clin Infect Dis May;24(5): Drancourt M, Stein A, Argenson JN, Zannier A, Curvale G, Raoult D. Oral rifampin plus ofloxacin for treatment of Staphylococcus-infecte orthopeic implants. Antimicrob Agents Chemother Jun;37(6): Johnson DP, Bannister GC. The outcome of infecte arthroplasty of the knee. J Bone Joint Surg Br Mar;68(2): Tsukayama DT, Wicklun B, Gustilo RB. Suppressive antibiotic therapy in chronic prosthetic joint infections. Orthopeics Aug;14(8): Sherrell JC, Fehring TK, Oum S, Hansen E, Zmistowski B, Dennos A, Kalore N; Periprosthetic Infection Consortium. The Chitranjan Ranawat Awar: fate of twostage reimplantation after faile irrigation anébriement for periprosthetic knee infection. Clin Orthop Relat Res Jan;469(1): Bejon P, Berent A, Atkins BL, Green N, Parry H, Masters S, McLary-Smith P, Gunle R, Byren I. Two-stage revision for prosthetic joint infection: preictors of outcome an the role of reimplantation microbiology. J Antimicrob Chemother Mar;65(3): Epub 2010 Jan 6.