Infect Dis Ther (2022) 11:2177–2203 https://doi.org/10.1007/s40121-022-00701-0 ORIGINAL RESEARCH Rates and Predictors of Treatment Failure in Staphylococcus aureus Prosthetic Joint Infections According to Different Management Strategies: A Multinational Cohort Study—The ARTHR-IS Study Group Reinaldo Espı́ndola . Venanzio Vella . Natividad Benito . Isabel Mur . Sara Tedeschi . Eleonora Zamparini . Johannes G. E. Hendriks . Luisa Sorlı́ . Oscar Murillo . Laura Soldevila . Mathew Scarborough . Claire Scarborough . Jan Kluytmans . Mateo Carlo Ferrari . Mathias W. Pletz . Iain Mcnamara . Rosa Escudero-Sanchez . Cedric Arvieux . Cecile Batailler . Frédéric-Antoine Dauchy . Wai-Yan Liu . Jaime Lora-Tamayo . Julia Praena . Andrew Ustianowski . Elisa Cinconze . Michele Pellegrini . Fabio Bagnoli . Jesús Rodrı́guez-Baño . Maria Dolores del Toro . The ARTHR-IS Group Received: June 23, 2022 / Accepted: September 20, 2022 / Published online: October 15, 2022 Ó The Author(s) 2022 ABSTRACT However, it is necessary to reassess the inci- dence and risk factors for treatment failure (TF) of Staphylococcus aureus PJI (SA-PJI) including Introduction: Guidelines have improved the functional loss, which has so far been neglected management of prosthetic joint infections (PJI). as an outcome. Methods: A retrospective cohort study of SA-PJI Jesús Rodrı́guez-Baño and Maria Dolores del Toro was performed in 19 European hospitals contributed equally as senior authors of the study. between 2014 and 2016. The outcome variable was TF, including related mortality, clinical Members of The ARTHR-IS Group are listed in failure and functional loss both after the initial Acknowledgements section. surgical procedure and after all procedures at Supplementary Information The online version 18 months. Predictors of TF were identified by contains supplementary material available at https:// logistic regression. Landmark analysis was used doi.org/10.1007/s40121-022-00701-0. R. Espı́ndola N. Benito  I. Mur Infectious Diseases and Microbiology Clinical Unit, Department of Medicine, Universitat Autònoma de Hospital Universitario Virgen Macarena, Seville, Barcelona, Barcelona, Spain Spain S. Tedeschi V. Vella  E. Cinconze  M. Pellegrini  F. Bagnoli Department of Medical and Surgical Sciences, GlaxoSmithKline (GSK), Siena, Italy University of Bologna, Bologna, Italy N. Benito  I. Mur S. Tedeschi  E. Zamparini Infectious Diseases Unit, Hospital de la Santa Creu i Infectious Diseases Unit, IRCCS Azienda Sant Pau, Barcelona, Spain Ospedaliero Universitaria di Bologna, Bologna, Italy N. Benito  I. Mur J. G. E. Hendriks Sant Pau Institute for Biomedical Research, Department of Orthopaedic Surgery and Trauma, Barcelona, Spain Máxima MC, Eindhoven, The Netherlands 2178 Infect Dis Ther (2022) 11:2177–2203 to avoid immortal time bias with rifampicin dL, hip fracture and additional joint surgery not when debridement, antibiotics and implant related to persistent infection. retention (DAIR) was performed. Conclusions: TF remains common in SA-PJI. Results: One hundred twenty cases of SA-PJI Functional loss accounted for a substantial were included. TF rates after the first and all proportion of treatment failures, particularly surgical procedures performed were 32.8% and after prosthesis removal. Use of rifampicin after 24.2%, respectively. After all procedures, func- DAIR was associated with a protective effect. tional loss was 6.0% for DAIR and 17.2% for Among the risk factors identified, anaemia and prosthesis removal. Variables independently obesity have not frequently been reported in associated with TF for the first procedure were previous studies. Charlson C 2, haemoglobin \ 10 g/dL, bacter- Trial registration: This study is registered at aemia, polymicrobial infection and additional clinicaltrials.gov, registration no. debridement(s). For DAIR, TF was also associ- NCT03826108. ated with a body mass index (BMI) [ 30 kg/m2 and delay of DAIR, while rifampicin use was protective. For all procedures, the variables associated with TF were haemoglobin \ 10 g/ L. Sorlı́ I. Mcnamara Department of Infectious Diseases, Hospital del Mar, Department of Orthopaedics, Norfolk and Norwich Institut Hospital del Mar d’Investigacions Mèdiques University Hospital, Norwich, UK (IMIM), Barcelona, Spain R. Escudero-Sanchez L. Sorlı́ Department of Infectious Diseases, Hospital Universitat Pompeu Fabra, Barcelona, Spain Universitario Ramón y Cajal, Madrid, Spain O. Murillo  L. Soldevila R. Escudero-Sanchez  J. Rodrı́guez-Baño  Department of Infectious Diseases, Hospital M. D. del Toro (&) Universitari Bellvitge, IDIBELL, Barcelona, Spain Centro de Investigación en Red de Enfermedades M. Scarborough  C. Scarborough Infecciosas (CIBERINFEC), Instituto de Salud Carlos Bone Infection Unit, Nuffield Orthopaedic Centre, III, Madrid, Spain Oxford University Hospitals NHS Foundation Trust, e-mail: mdeltoro@us.es Oxford, UK C. Arvieux J. Kluytmans Department of Infectious Diseases, Centre Department of Infection Control, Amphia Hospital, Hospitalier Universitaire de Rennes, Rennes, France Breda, The Netherlands C. Batailler J. Kluytmans Orthopedic Surgery Department, Croix Rousse Department of Medical Microbiology, University Hospital, Lyon, France Medical Center Utrecht, Utrecht University, F.-A. Dauchy Utrecht, The Netherlands Department of Infectious and Tropical Diseases, M. C. Ferrari Centre Hospitalier Universitaire de Bordeaux, Prosthetic-Joint Replacement Unit, Humanitas Bordeaux, France Research Hospital, Milan, Italy W.-Y. Liu M. W. Pletz Department of Orthopaedic Surgery and Trauma, Center for Infectious Diseases and Infection Catharina Hospital, Eindhoven, The Netherlands Control, Jena University Hospital, Jena, Germany Infect Dis Ther (2022) 11:2177–2203 2179 Graphical abstract: Microbiology, Hospital Universitario Virgen del W.-Y. Liu Rocı́o, Seville, Spain Department of Orthopaedic Surgery and Trauma, Máxima MC, Eindhoven, The Netherlands A. Ustianowski J. Lora-Tamayo Infectious Diseases Unit, North Manchester General Department of Internal Medicine, Hospital Hospital, Manchester, UK Universitario, 12 de Octubre, Madrid, Spain J. Rodrı́guez-Baño  M. D. del Toro J. Praena Infectious Diseases and Microbiology Clinical Unit, Clinical Unit of Infectious Diseases and Hospital Universitario Virgen Macarena, Avda Dr Fedriani 3, 41009 Seville, Spain 2180 Infect Dis Ther (2022) 11:2177–2203 PLAIN LANGUAGE SUMMARY Key Summary Points Staphylococcus aureus is one of the most virulent bacteria and frequently causes prosthetic joint Why carry out this study? infections. Staphylococcus aureus has been associated Knowledge of the treatment of this type of with a higher rate of prosthetic joint infection has advanced in recent years, and infection (PJI) and treatment failure treatment guidelines have led to improved compared with other pathogens. management. Typically, the successful treat- ment of these infections has been determined by Changes in the management of S. aureus clinical cure, that is, the symptoms of infection PJI (SA-PJI) in recent decades make it have disappeared, but has not taken into account necessary to reassess the incidence and loss of function (such as significant difficulties risk factors for treatment failure, including walking), which is critical for the patient’s qual- functional loss which has previously been ity of life. Our aim in this study was to evaluate neglected as an outcome. the success of current management strategies for S. aureus prosthetic joint infection, including What was learned from the study? recovery of functionality, and the factors that A considerable proportion of SA-PJIs failed predict why some of these infections are not after initial surgical treatments, although cured, to identify areas for improvement. a substantial part of them can be rescued In a multinational cohort of 128 patients with by additional procedures. S. aureus prosthetic joint infection, rates of treatment failure were found to be high, with Significant functional loss must be significant rates of loss of function, especially considered in addition to clinical failure, when the prosthesis needed to be removed. Loss mainly in patients in whom the prosthesis of function was less frequent when the infection was removed, with no possibility of was initially treated with surgical cleaning with- recovery. out removal of the prosthesis, even when this In patients managed with debridement, procedure failed at first. We found that anaemia antibiotic and prosthesis retention and obesity were associated with lower treatment (DAIR), the loss of function was lower success, and that the probability of treatment even if this procedure fails. success increased when surgical cleaning with- out prosthesis removal was performed early, and Anaemia and obesity were risk factors for when the antibiotic rifampicin was used in treatment failure that are rarely reported. combination with another antibiotic. The importance of some risk factors for treatment failure was reinforced, including the protective role of Keywords: Prosthetic joint infection; rifampicin-based treatment in DAIR. Staphylococcus aureus; Outcome; Clinical failure; Functional failure J. Rodrı́guez-Baño  M. D. del Toro Department of Medicine, School of Medicine, University of Sevilla, Seville, Spain DIGITAL FEATURES J. Rodrı́guez-Baño  M. D. del Toro This article is published with digital features, Biomedicine Institute of Sevilla (IBiS), Seville, Spain including [list digital features available e.g. a Present Address: video abstract and slide deck], to facilitate R. Espı́ndola understanding of the article. To view digital Hospital Universitario de Valme, Seville, Spain Infect Dis Ther (2022) 11:2177–2203 2181 features for this article go to https://doi.org/10. account clinical and functional aspects, and to 6084/m9.figshare.21163246. investigate the predictors of TF according to different management strategies. The assess- ment of both objectives will be carried out for INTRODUCTION both the first surgical procedure and all addi- tional procedures performed on patients in Staphylococcus aureus is widely recognized as a whom the first one failed. key microorganism causing prosthetic joint infection (PJI), and has been associated with higher rates of treatment failure (TF) compared METHODS with other aetiologies [1]. However, studies evaluating TF have traditionally focused on Study Design, Sites and Period outcomes related primarily with control of the infection [2–8], while functional outcomes have This study is part of the ARTHR-IS project (reg- been given little consideration. A definition of istered at clinicaltrials.gov: NCT03826108) and TF that integrates clinical and functional was designed to evaluate the incidence, risk aspects could provide a more realistic measure- factors and predictors of SA-PJI TF after primary ment of the consequences of S. aureus-PJI (SA- hip and knee arthroplasty. A retrospective PJI). cohort study was conducted in 19 European TF rates may vary according to patient hospitals to identify patients older than 18 years characteristics, infection type (early, delayed or who received a primary arthroplasty between 1 late) and surgical management. Implementa- January 2014 and 31 December 2016 and tion of the recommendations included in the developed post-surgical knee or hip PJI due to guidelines for the management of PJI [9, 10] S. aureus within the first year after the proce- may have positively influenced changes in the dure. The participating sites, located in Spain, rates and predictors of TF in general, and par- Italy, France, Germany, UK and the Nether- ticularly in patients managed with debride- lands, were selected using the CLIN-NET net- ment, antibiotics and implant retention (DAIR). work (https://www.combacte.com/about/clin- For SA-PJI, however, there is little published net/), based on their research experience and data on the outcomes associated with different data collection capability. therapeutic strategies and their predictors, and The STROBE recommendations were fol- the timing and role of rifampicin in combina- lowed for the reporting of the study (Supple- tion with other antibiotics, particularly in mentary Table S1). patients undergoing DAIR, remain controversial. Participants Assessments of TF tend to be analysed for the first surgical procedure performed to treat the Patients with post-surgical hip or knee PJI due infection, which is important to evaluate the to S. aureus diagnosed within the first year after outcomes associated for the decision about primary arthroplasty were included. The criteria which procedure must be performed. However, used to define SA-PJI were as follows: presence additional procedures are usually performed if of at least one sign or symptom of PJI, including that first procedure fails to cure the infection, joint pain and/or swelling, or a sinus tract which may rescue some patients but could also communicating with the prosthesis; and the affect the functional outcome. Evaluations of TF isolation of S. aureus from (a) C one joint aspi- considering all the procedures performed are rate culture, (b) C two periprosthetic tissue not usually made. Such an evaluation would samples and (c) blood cultures with no other provide a more global view of the final conse- obvious source of infection. The patients were quences of SA-PJI. identified by reviewing medical records from The objectives of this study were to provide updated rates of TF in SA-PJI, taking into 2182 Infect Dis Ther (2022) 11:2177–2203 microbiological laboratory databases, local PJI inconsistencies. The study was approved by the surveillance databases and discharge reports. Ethics Committees at each site (Supplementary Table S5). The need to obtain written informed Variables and Definitions consent was waived owing to the retrospective nature of the study and anonymized data, The primary outcome variable was TF until except in the case of the French hospitals where month 18 after the first surgical procedure per- a letter of non-opposition was sent to eligible formed, and was analysed separately both for patients. All patients included in these centres the first surgical procedure performed (mim- therefore gave their authorization to icking an intention-to-treat analysis for that participate. procedure) and for all procedures performed (including those performed after failure of the Statistical Analysis first one). TF was defined as a composite vari- able including SA-PJI-related mortality, clinical For bivariate analysis of the association of failure and functional failure. Clinical failure exposure variables with TF, relative risks with was defined as persistence or recurrence of signs 95% confidence interval (CI) were calculated; or symptoms of infection. For the analysis of p values were calculated by Chi-square or Fish- the first procedure, this also included the need er’s exact test, as appropriate. Continuous vari- for an additional course of antibiotics beyond ables were categorized after analysing for the initial one, the need to use long-term sup- stratified associations with TF. Multivariable pressive antibiotic therapy and need of pros- analyses were performed by logistic regression: thesis removal if not performed as the initial the effect of study site was controlled for using a surgical procedure. Functional failure was generalized linear mixed model in which study defined as impeded or significantly impaired sites were considered random effects. Variables walking due to prosthetic loosening or the need with p value \ 0.15 in bivariate analysis, and to perform a Girdlestone procedure or those considered as potentially relevant from arthrodesis. Finally, TF was also analysed in the clinical judgement, were entered into the subgroup of patients who underwent DAIR as models and selected using a manual stepwise the first procedure. backward procedure. Variables with Potential predictors of TF were selected p value \ 0.1 were kept in the models. according to previous studies [2–8] and addi- Collinearity and modification effects between tional hypotheses developed by the project variables were studied when clinically sound. team, and are included in Tables 1 and 2. The predictive ability of each model was DAIR as primary treatment procedure was examined by calculating their areas under the considered appropriate if it was per- receiver operating characteristic (AUROC) formed \ 21 days from onset of PJI symptoms, curves with 95% CIs. For the effect of rifampicin there was no sinus tract communicating with in the subgroup of patients who underwent the joint prosthesis and replacement of poly- DAIR as first surgical treatment, a propensity ethylene or mobile components was performed score (PS) was calculated using a non-parsimo- according to IDSA guidelines [9]. The defini- nious multivariate logistic regression model, in tions for other variables are included in the which the outcome variable was combination Supplementary Table S2. therapy with rifampicin. As further sensitivity analyses for the impact of rifampicin therapy, Data Collection and Ethical Aspects parsimonious multivariate logistic regression models were performed, in which one or two Data collection was supervised locally by staff other variables were removed. In addition, to with relevant expertise in the field. Data were avoid immortal time bias, landmark analysis entered into an anonymized electronic case was used, excluding patients who died or failed report form and checked for missing values and treatment in the first 21 days after debridement. Infect Dis Ther (2022) 11:2177–2203 2183 Table 1 Characteristics of 128 patients with Staphylococ- Table 1 continued cus aureus prosthetic joint infections (SA-PJI) Variables No. of patients Variables No. of patients (percentage), except (percentage), except where specified where specified Methicillin-resistant 28 (21.9) Spain 52 (40.6) Staphylococcus aureus France 15 (11.7) Polymicrobial infection 36 (28.1) Germany 7 (5.5) Bacteraemia 25 (19.5) United Kingdom 16 (12.5) Days from arthroplasty to onset of 24 (15–36) Italy 17 (13.3) SA-PJI symptoms; median The Netherlands 21 (16.4) (IQR) Male sex 65 (50.8) Symptoms and signs of SA-PJI Born abroad 2 (1.5) Fever 32 (25.0) Age in years; median (IQR) 73 (59.3–80.8) Joint pain 70 (54.7) Body mass index units; median 31.2 (25.6–35.1) Suppuration 89 (69.5) (IQR) Cellulitis 38 (29.7) ASA 3-4 assessment for primary 64 (50%) Wound dehiscence 56 (43.8) arthroplasty Articular swelling 18 (14.1) Charlson comorbidity index; 1 (1–2) Sinus tract 8 (6.3) median (IQR) Laboratory data at diagnosis of Comorbidities SA-PJI; median (IQR) Chronic heart failure 17 (13.3) Haemoglobin (g/dL) 10.5 (9.5–11.4) Chronic pulmonary disease 30 (23.4) Blood leucocytes (cells/lL) 9600 (7550–12,950) Diabetes mellitus 29 (22.7) C-reactive protein (mg/L) 84.5 (23.2–224.2) Chronic renal insufficiency 5 (3.9) Erythrocyte sedimentation rate 68.5 (39.8–92.3) Reason for arthroplasty (mm/h) Osteoarthritis 83 (64.8) Type of initial surgical procedure Hip fracture 34 (26.6) performed to treat SA-PJI Osteonecrosis 5 (3.9) Debridement and prosthesis 99 (77.3) Others 6 (4.6) retention Type of arthroplasty Partial removal and 6 (4.7) reimplantation Total hip arthroplasty 50 (39.1) One-stage replacement and 4 (3.1) Partial hip arthroplasty 27 (21.1) reimplantation Total knee arthroplasty 47 (36.7) Two-stage replacement and 13 (10.2) Partial knee arthroplasty 4 (3.1) reimplantation 2184 Infect Dis Ther (2022) 11:2177–2203 Table 1 continued characteristics are presented in Table 1. Infec- tion-related symptoms started a median of 24 Variables No. of patients (IQR 15–36) days after the primary arthroplasty (percentage), except (Supplementary Fig. S1), while the first surgical where specified procedure for treatment of infection was per- formed a median of 4 days (IQR 1–11) after Girdlestone procedure (hip 6 (4.7) symptom onset. Bacteraemia occurred in 25 resection arthroplasty) cases (19.5%). Overall, out of 128 PJI cases, 28 Days from onset of SA-PJI 4 (1–11) (21.9%) were due to methicillin-resistant S. au- symptoms to first surgical reus (MRSA) strains. procedure performed; median Figure 1 shows patient outcomes according (IQR) to the first and additional surgical procedures performed. The rate of TF after the first proce- Days of antibiotic treatment to dure was 32.8% (42 patients; 95% CI treat PJI; median (IQR) 25.2–41.3%). TF was due to clinical failure in 27 Empirical intravenous 2 (1–4) cases (21.1%), related deaths in 9 (7%) and loss of function in 6 (4.7%). Median days until fail- Target intravenous 16 (11–34) ure was 126 (IQR 34–335). Deaths occurred at a Target oral 50 (33–80) median of 21 (IQR 13–48) days after the first surgical procedure performed. Total 73 (56–96) After further surgical interventions, 11 Target oral antibiotics patients who failed the first procedure (9 with persistent infection, 2 with functional failure Rifampicin 103 (80.5) due to prosthesis loosening after DAIR) were Fluoroquinolones 73 (63.5) rescued. After 18 months of follow-up, TF was 24.2% (95% CI 17.5–32.3%). The reasons for Clindamycin 17 (14.8) failure were 9 related deaths (7.0%), 11 cases of Trimethoprim–sulfamethoxazole 15 (13.0) clinical failure and 11 cases of functional loss Tetracyclines 3 (2.6) (8.5%, respectively). Excluding functional loss, the failure rate was 15.5%. Linezolid 4 (3.5) Of the 99 patients who received DAIR as a Others 3 (2.6) first intervention to treat the SA-PJI (Fig. 1), 31 (31.3%) failed treatment due to death (n = 6), clinical failure (n = 23) or loss of function (n = 2). Despite further interventions, 15 were still failing at the end of the 18-month follow- RESULTS up. Of the 29 patients who received other types Patient Characteristics and Treatment of first interventions to treat SA-PJI, 11 (37%) Failure Rates failed treatment, broken down as follows: rela- ted death (n = 3), clinical failure (n = 4) and loss A total of 130 cases of SA-PJI were detected, and of function (n = 4). Notwithstanding further 128 were included (relevant follow-up data were interventions, 7 were still failing at the end of missing for the other two). The median number the 18-months follow-up. of cases per hospital was 7 (interquartile range A summary of the rates and reasons for [IQR] 5–9). The median age of patients was treatment failure is provided in Supplementary 73 years (IQR 59–81 years); 65 (50.8%) were Table S3, including the rate of TF for prosthesis males; 77 (60.2%) had hip arthroplasty (50 total removal as first procedure. and 27 partial) and 51 (39.8%) knee arthro- plasty (47 total and 4 partial). Patient Table 2 Bivariate analysis of potential predictors of treatment failure among patients with SA-PJI: after the first surgical procedure, after DAIR, and after all surgical procedures performed Failure after the initial surgical Failure after DAIR as first surgical Failure at 18 months after all procedure (n = 128; failures = 42) procedure (n = 99; failures = 31) procedures performed (n = 128; failures = 31) Variables No. Relative risk p- No. Relative risk p- No. Relative risk p- failure (%) (95% CI) Value failure (%) (95% CI) Value failure (%) (95% CI) Value Age Infect Dis Ther (2022) 11:2177–2203 C 80 years 13 (40.6) 1.6 (0.7–3.6) 0.277 8 (38.1) 1.5 (0.5–4.1) 0.450 12 (37.5) 2.4 (1.1–5.8) 0.043 \ 80 years 29 (30.2) 23 (29.5) 19 (19.8) Sex Male 22 (33.8) 1.1 (0.5–2.3) 0.800 15 (30.6) 1.07 (0.4–2.5) 0.880 17 (26.2) 1.2 (0.5–2.8) 0.378 Female 20 (31.7) 16 (32.0) 14 (22.2) Charlson index C 2 Yes 21 (47.7) 2.7 (1.3–5.9) 0.009 16 (44.4) 2.6 (1.1–6.1) 0.033 15 (34.1) 2.2 (0.9–5.0) 0.059 No 21 (25.0) 15 (23.8) 16 (19.0) Haemoglobin \ 10 mg/dl Yes 23 (48.9) 3.1 (1.4–6.7) 0.003 18 (50.0) 3.8 (1.6–9.4) 0.002 18 (38.3) 3.2 (1.4–7.5) 0.005 No 19 (23.5) 13 (20.6) 13 (16.0) Leukocytes C 7500/lL Yes 29 (31.9) 0.9 (0.4–2.2) 0.882 19 (27.9) 1.9 (0.5–7.6) 0.383 24 (26.4) 2.3 (0.7–7.4) 0.211 No 10 (33.3) 9 (37.5) 4 (13.3) C-reactive protein C 150 mg/La Yes 18 (36.7) 1.3 (0.6–2.8) 0.478 10 (31.3) 1.0 (0.4–2.5) 0.992 15 (30.6) 1.7 (0.7–3.8) 0.221 No 22 (30.6) 19 (31.1) 15 (20.8) 2185 Table 2 continued 2186 Failure after the initial surgical Failure after DAIR as first surgical Failure at 18 months after all procedure (n = 128; failures = 42) procedure (n = 99; failures = 31) procedures performed (n = 128; failures = 31) Variables No. Relative risk p- No. Relative risk p- No. Relative risk p- failure (%) (95% CI) Value failure (%) (95% CI) Value failure (%) (95% CI) Value Erythrocyte sedimentation rate C 60 mm/hb Yes 9 (42.9) 2.1 (0.5–8.6) 0.261 3 (37.5) 1.3 (0.2–8.6) 0.557 8 (38.1) 8.6 (0.9–78.7) 0.051 No 4 (26.7) 4 (30.8) 1 (6.7) 2 Body mass index [ 30 kg/m Yes 24 (34.8) 1.2 (0.6–2.5) 0.608 21 (36.8) 1.9 (0.8–4.5) 0.122 15 (21.8) 0.7 (0.3–1.7) 0.479 No 18 (30.5) 10 (23.8) 16 (27.1) Hip fracture as the reason for arthroplasty Yes 17 (50.0) 2.7 (1.2–6.2) 0.010 12 (48.0) 2.7 (1.04–6.8) 0.035 17 (50.0) 5.7 (2.3–13.7) 0.001 No 25 (26.6) 19 (25.7) 14 (14.9) Type of arthroplasty Total knee arthroplasty 12 (25.5) Ref 10 (27.0) Ref 7 (14.9) Ref Total hip arthroplasty 14 (28.0) 1.1 (0.5–2.8) 0.784 11 (26.8) 0.9 (0.4–2.7) 0.984 8 (16.0) 1.1 (0.4–3.3) 0.880 Partial hip arthroplasty 14 (51.9) 3.1 (1.1–8.5) 0.025 9 (50.0) 2.7 (0.8–8.7) 0.097 14 (51.9) 6.1 (2.0–18.5) 0.001 Partial knee arthroplasty 2 (50.0) 2.9 (0.4–23.0) 0.310 1 (33) 1.3 (0.1–16.6) 0.815 2 (50.0) 5.7 (0.7–47.5) 0.107 c Radiological signs of infection Yes 6 (42.8) 4.6 (0.8–26.0) 0.080 5 (44.0) 1.76 (0.2–7.7) 0.866 2 (20.0) 0.56 (0.1–2.8) 0.479 No 20 (35.1) 14 (32.6) 19 (31.1) Fever [ 38 °C: Infect Dis Ther (2022) 11:2177–2203 Table 2 continued Failure after the initial surgical Failure after DAIR as first surgical Failure at 18 months after all procedure (n = 128; failures = 42) procedure (n = 99; failures = 31) procedures performed (n = 128; failures = 31) Variables No. Relative risk p- No. Relative risk p- No. Relative risk p- failure (%) (95% CI) Value failure (%) (95% CI) Value failure (%) (95% CI) Value Yes 13 (40.6) 1.6 (0.7–3.6) 0.190 9 (40.9) 1.7 (0.6–4.6) 0.199 10 (31.3) 1.6 (0.7–3.9) 0.200 No 29 (30.2) 22 (28.6) 21 (21.9) Infect Dis Ther (2022) 11:2177–2203 Sinus tract at diagnosis: Yes 6 (75.0) 7 (1.3–36.0) 0.015 2 (100) – 0.096 6 (75.0) 11.4 (2.2–59.9) 0.003 No 36 (30.0) 29 (29.9) 25 (20.8) Methicillin-resistant S. aureus Yes 13 (46.4) 2.1 (0.9–5.0) 0.068 8 (40.0) 1.6 (0.6–4.5) 0.249 11 (39.3) 2.6 (1.0–6.4) 0.035 No 29 (29.0) 23 (29.1) 20 (20.0) Polymicrobial infection Yes 15 (41.7) 1.7 (0.8–3.8) 0.130 13 (40.6) 1.8 (0.7–4.5) 0.160 11 (30.6) 1.6 (0.6–3.7) 0.205 No 27 (29.3) 18 (26.9) 20 (21.7) Presence of bacteraemia Yes 13 (52.0) 2.7 (1.1–6.7) 0.023 9 (52.9) 3.1 (1.05–8.9) 0.037 10 (40.0) 2.6 (1.0–6.6) 0.040 No 29 (28.2) 22 (26.8) 21 (20.4) Type of first surgical therapy Debridement or partial replacement 32 (30.5) 0.5 (0.2–1.4) 0.229 – – – 22 (21.0) 0.4 (0.2–1.1) 0.061 Total prosthesis replacement 10 (43.5) 9 (39.1) Polyethylene/mobile component replacementd Yes – – – 14 (24.1) 0.4 (0.2–1.1) 0.067 – – – No 17 (41.5) 2187 Table 2 continued 2188 Failure after the initial surgical Failure after DAIR as first surgical Failure at 18 months after all procedure (n = 128; failures = 42) procedure (n = 99; failures = 31) procedures performed (n = 128; failures = 31) Variables No. Relative risk p- No. Relative risk p- No. Relative risk p- failure (%) (95% CI) Value failure (%) (95% CI) Value failure (%) (95% CI) Value Days from symptom onset to surgery [ 21 Yes – – – 6 (54.5) 3.3 (0.8–11.1) 0.081 – – No 5 (45.5) e Appropriate indication for DAIR Yes – – – 0 (20.8) 0.38 0.029 – – – (0.15–0.92) No 21 (51.0) Need for additional surgery not due to persistent infectionf Yes 16 (55.2) 3.45 (1.5–8.1) 0.004 11 (55.0) 3.6 (1.3–9.9) 0.013 13 (44.8) 3.6 (1.5–8.9) 0.005 No 26 (26.3) 9 (45.0) 18 (18.2) Additional procedures due to failure of first procedure 0 – – – – – 12 (14.0) Ref 1 – 3 (21.4) 1.68 (0.4–6.9) 0.471 2 8 (50.0) 6.17 (1.9–19.5) 0.002 3 3 (60.0) 9.25 (1.4–61.2) 0.021 C4 5 (71.4) 12.3 0.047 (1.0–146.8) Inadequate empirical antimicrobial therapyg Infect Dis Ther (2022) 11:2177–2203 Table 2 continued Failure after the initial surgical Failure after DAIR as first surgical Failure at 18 months after all procedure (n = 128; failures = 42) procedure (n = 99; failures = 31) procedures performed (n = 128; failures = 31) Variables No. Relative risk p- No. Relative risk p- No. Relative risk p- failure (%) (95% CI) Value failure (%) (95% CI) Value failure (%) (95% CI) Value Yes 5 (55.6) 3.5 (0.9–14.3) 0.077 5 (55.6) 4.8 (1.1–20.5) 0.038 4 (44.4) 3.4 (0.8–14.1) 0.096 No 22 (26.2) 13 (20.6) 16 (19.0) Infect Dis Ther (2022) 11:2177–2203 Rifampicin used in antibiotic regimen Yes 31 (30.1) 0.5 (0.2–1.3) 0.138 22 (26.2) 0.2 (0.1–0.7) 0.009 22 (21.4) 0.5 (0.2–1.2) 0.104 No 11 (44.0) 9 (60.0) 9 (36.0) Fluoroquinolone used in antibiotic regimen Yes 20 (27.4) 0.6 (0.3–1.2) 0.133 16 (26.7) 0.4 (0.2–1.4) 0.216 12 (16.4) 0.4 (0.2–0.8) 0.018 No 22 (40.0) 15 (38.5) 19 (34.5) Rifampicin plus quinolones used in regimen Yes 18 (27.7) 0.6 (0.3–1.4) 0.210 14 (25.5) 0.6 (0.4–1.2) 0.160 11 (16.9) 0.4 (0.2–0.9) 0.040 No 24 (38.1) 17 (38.6) 20 (31.7) DAIR debridement and implant retention a Data available for 121 patients b Data available for 36 patients c Data available for 71 patients. Radiological signs of infection is defined as the presence of periprosthetic lucency or signs of loosening of prosthetic components d Applies only in case of DAIR e Performed \ 21 days from symptom onset to surgery, absence of sinus tract and replacement of polyethylene or mobile components f Additional interventions caused by bleeding, haematoma or devitalized tissues g Data for 93 patients, defined as inactive antibiotics in vitro used before the susceptibility results were available 2189 2190 Infect Dis Ther (2022) 11:2177–2203 Fig. 1 Outcome of patients with SA-PJI according to type loss of function). Similar explanations apply to other of procedure performed to treat the infection. This surgical procedures. aFailure: a composite variable that figure shows the outcome of patients with SA-PJI included related mortality, clinical failure and functional according to the first type of procedure performed to loss (see Methods section). bIncludes: persistence or relapse treat the infection. For boxes on the left, 99 patients of infection, need for extra courses of antibiotics after receiving initial debridement, antibiotic and implant initial scheduled treatment and need for suppressive retention (DAIR) were included: treatment was successful antibiotic therapy. cIncludes severe impairment of limb in 68 patients and 31 failed, including 6 deaths, 23 clinical function that impedes or makes walking very difficult, failures and 2 loss of function. The 25 surviving patients including Girdlestone or arthrodesis. *Two aseptic loosen- with failure after the first procedure underwent subsequent ing. **Two arthrodesis (one hip, one knee), four prosthesis procedures. After 18 months and considering additional removal waiting for a new prosthesis. ***One arthrodesis procedures, 15 were still failing (9 clinical failures and 6 Analysis of Variables Associated Charlson index C 2 [adjusted OR 2.5 (95% CI with Treatment Failure 1.1–5.9)], haemoglobin \ 10 g/dL [aOR 3.6 (95% CI 1.6–8.4)], bacteraemia [aOR 2.7 (95% Table 2 presents the bivariate analysis of risk CI 1.0–7.6)], polymicrobial infection [aOR 3.1 factors for TF after the initial surgery. Variables (95% CI 1.2–8.2)] and need for additional pro- with p-value \ 0.15 were Charlson index C 2, cedures not due to persistent infection [aOR 2.7 haemoglobin \ 10 g/dL, hip fracture as the (95% CI 1.1–6.6)] as variables independently reason for arthroplasty, type of arthroplasty, associated with TF following the first surgical radiological signs of infection, sinus tract, procedure (Table 3). The AUROC of the model MRSA, polymicrobial infection, bacteraemia, for the observed data was 0.82 (95% CI need for additional procedures not due to per- 0.74–0.89). sistent infection, inadequate empirical antimi- In the specific analysis performed to inves- crobial treatment, and treatment with tigate the factors associated with TF of SA-PJI rifampicin and fluoroquinolones. After con- after DAIR as the first surgical procedure, the trolling for hospital and type of initial surgery variables with p-values \ 0.15 identified in performed, multivariable analysis retained bivariate analysis were the same as for the whole Infect Dis Ther (2022) 11:2177–2203 2191 Table 3 Multivariate analysis of associations of potential predictors of treatment failure among patients with SA-PJI after the first surgical procedure, after DAIR, and after all surgical procedures performed Failure after the initial Failure after Failure at 18 months surgical procedure debridement and after the initial and performed (n = 128; prosthesis retention as additional surgical failures = 42)a first surgery procedure procedures performed performed (n = 99; (n = 128; failures = 31)c failures = 31)b Variable Adjusted OR p- Adjusted OR p- Adjusted OR P- (95%IC) Value (95%IC) Value (95%IC) Value Charlson index C 2 2.5 (1.1–5.9) 0.030 2 Body mass index [ 30 kg/m – – 3.9 (1.2–12.6) 0.021 – Hip fracture – – – 4.6 (1.6–12.9) 0.005 Haemoglobin \ 10 g/dL 3.6 (1.6–8.4) 0.003 5.2 (1.9–14.5) 0.002 2.5 (1.0–6.6) 0.059 Bacteraemia 2.7 (1.0–7.6) 0.058 6.4 (1.7–24.2) 0.007 – – Days from symptom onset to surgery – – 1.03 (1.01–1.06) 0.003 – – Polymicrobial infection 3.1 (1.2–8.2) 0.021 – – Additional surgery not due to 2.7 (1.1–6.6) 0.033 4.4 (1.4–13.9) 0.044 3.2 (1.1–8.9) 0.028 persistent infection Rifampicin therapy – – 0.2 (0.1–0.7) 0.015 – – BMI body mass index a The AUROC curve of the model was 0.82 (95% CI 0.74–0.89) b The AUROC curve of the model was 0.84 (95%CI 0.76–0.93). The model prediction was similar when rifampicin was introduced at more than 21 or 28 days c The AUROC curve of the model was 0.80 (95% CI 0.71–0.90) cohort, except for radiological signs of infection haemoglobin \ 10 g/dL [aOR 5.2 (95% CI and MRSA, and with the addition of body mass 1.9–14.5)], bacteraemia [aOR 6.4 (95% CI index (BMI) [ 30 kg/m2, interval [ 21 days 1.7–24.2)], days from symptom onset to surgery from symptom onset to surgery, [aOR 1.03 (95% CI 1.01–1.06)], and need for polyethylene/mobile component replacement additional debridements after the first proce- and appropriate indication for DAIR (Table 2). dure not due to persistent infection [aOR 4.4 No statistically significant differences were (95% IC 1.4–13.9)]; use of rifampicin was a found between patients treated with and with- protective factor [aOR 0.2 (95% CI 0.1–0.7)]. out combination rifampicin therapy (Supple- The AUROC of the model for the observed data mentary Table S4). As we were unable to obtain was 0.84 (95% CI 0.76–0.93). The model pre- a PS sufficiently predictive for the probability of diction was similar when rifampicin was con- receiving rifampicin (AUROC curves \ 0.5), a PS sidered to be used for more than 21 or 28 days. was not useful to provide additional control for For the different models made in which one or confounding for the indication of rifampicin. two variables other than rifampicin were On multivariable analysis (Table 3), variables removed, the estimates of the OR for rifampicin independently associated with TF of DAIR were: ranked from 0.15 (95% CI 0.04–0.5) to 0.2 (95% BMI [ 30 kg/m2 [aOR 3.9 (95% CI 1.2–12.6)], CI 0.1–0.7). In the landmark analysis excluding 2192 Infect Dis Ther (2022) 11:2177–2203 patients who died or failed in the first 21 days critical to the quality of life of patients. Func- after debridement, use of rifampicin remained tional loss is significantly influenced by the protective for TF [adjusted ORs (95%CI), 0.22 surgical procedures performed and this infor- (0.06–0.80); AUROC curve of the model, 0.82 mation is therefore relevant to the decision- (95% CI 0.72–0.91)]. making process. Table 2 also presents the bivariate analysis of When DAIR was analysed as the initial pro- risk factors for patients who failed after all pro- cedure, the TF rate was 31.3% (29.2% without cedures performed. In addition to the variables considering functional loss), which is slightly identified for the first procedure, TF was also higher than reported in more recent observa- associated with age [ 80 years, prosthesis tions [7, 12] but lower than in older publica- removal as first procedure, need to perform tions [2–4, 6, 13] (Table 4). However, the TF rate additional joint surgery not due to persistent decreased to 21.2% after additional procedures, infection and non-use of rifampicin and fluo- and to 15% if functional loss was not taken into roquinolones in combination. On multivariate account (which is more consistent with defini- analysis, hip fracture [aOR 4.6 (95% CI tions in previous reports). The lower TF rates for 1.6–12.9)], haemoglobin level \ 10 g/dL [aOR DAIR reported by the latest studies (and by this 2.5 (95% CI 1.0–6.6)] and need to perform one) could be attributed to better patient additional joint surgery not due to persistent selection for this treatment strategy and to the infection [aOR 3.2 (95% IC 1.1–8.9)] were involvement of multidisciplinary teams in the independently associated with TF at 18 months management of PJI. Indeed, Bouaziz et al. [13] (Table 3). The AUROC of the model for the found an overall TF rate of 42%, which observed data was 0.80 (95% CI 0.71–0.90). decreased to 30% when DAIR was performed according to the latest guidelines [9]. Our data further suggest that appropriate patient selec- DISCUSSION tion favours more positive outcomes, and that an initial TF can be rescued without significant In this study, we found that nearly a third of functional loss in a considerable number of initial surgical procedures resulted in TF. The TF patients. rate decreased when further surgical procedures The overall TF rates, including clinical and were performed. Importantly, we estimated the functional updates, for prosthesis removal in impact of significant functional loss. When our cohort (34.4%) may seem relatively high DAIR was used as the first procedure, even when compared with other studies, but when though it is a less aggressive strategy, additional only clinical cure was considered: the rate of TF procedures rescued a significant proportion of for SA-PJI was 17.2%, which is similar to that initial failures without increasing loss of func- found in other series [4, 5, 11]. The high pro- tion. In addition, the predictors of TF for SA-PJIs portion of TF after prosthesis removal related to were identified, and the role of rifampicin in functional loss is noteworthy and reinforces the patients undergoing DAIR was confirmed. importance of early diagnosis of SA-PJI to A review of the literature on SA-PJI studies increase the likelihood of being treated with focusing on treatment outcomes highlighted DAIR. the difficulties of comparing different study Since the decision to perform DAIR or results owing to heterogeneity in study design, remove the prosthesis as first procedure is case definitions adopted, length of follow-up strongly influenced by patient and infection and types of analyses used (Table 4). Overall, characteristics, we did not try to compare the previously reported TF rates ranged from 0% to outcomes of the two procedures as they are not 16.6% for prosthesis removal [4, 5, 8, 11] and comparable. Instead, we focused our analysis on 13.6% to 63% for DAIR [2, 4–8, 12–15]. It should identifying potential predictors for TF. Regard- also be noted that, to the best of the authors’ ing the variables identified, the Charlson index knowledge, functional outcome was not con- is a predictor of survival and also of prognosis of sidered at all in previous studies, despite it being many infections; similarly, other studies have Table 4 Summary of published studies on the outcome and management of PJI including C 20 cases focusing on S. aureus References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design Brandt et al. 1980–1991 Post-surgical knee and hip DAIR (n = 33) Failure: persistence or Failure: Multivariable (HR): DAIR (1997) [2] Retrospective SA-PJI Median 28 days intravenous recurrence of clinical signs 59% at 1 year performed [ 2 days after cohort, N = 33 (3 MRSA) (IV) antibiotics (beta- of PJI symptom onset 63% at 2 years unicentric lactams, 91%; vancomycin, Infect Dis Ther (2022) 11:2177–2203 9%) Salgado et al. 1998–2004 Hip and knee SA-PJI DAIR (n = 20) Failure: relapse, reinfection, Failure: 38%, Multivariable: (2007) [3] Retrospective N = 45 (12 MRSA) Partial/total 1 T (n = 4) death-related, other signs of 32% in MSSA, MRSA, TKA, retention of cohort, clinical failure prosthesis 2 T (n = 15) 50% in MRSA unicentric Resection arthroplasty Follow-up: median (n = 6) 190 days (range Median 42 days IV 4–2279 daus) antibiotics (beta-lactams or vancomycin, 51% with rifampicin). In four DAIR, oral rifampicin-quinolone Vilchez et al. 2000–2007 Early DAIR (n = 53) Cure: Absence of infection Failure: 24.5% Multivariable (HR): (2010) Retrospective (\ 2 months, \ 15 days Mean 10.6 days iv antibiotics symptoms, aseptic loosening Follow-up: 2 years CRP [ 22 mg/dl [14] cohort, of symptoms) SA-PJI, (cloxacillin if MSSA, that required exchange Need for 2nd DAIR unicentric total and partial hip and vancomycin if MRSA) and prosthesis knee prosthesis Prosthesis age [ 25 days 88 days oral (levofloxacin N = 53 (4 MRSA) plus rifampicin) 2193 Table 4 continued 2194 References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design Joulie et al. 2001–2006 Any total and partial hip DAIR (n = 30), 1 T Cure: ESR and/or CRP Cure: Multivariable for cure: (2011) [4] Prospective and knee SA-PJI (n = 15), 2 T (n = 25), normal, non-inflammatory DAIR, 57%; Monomicrobial cohort, N = 95 evaluable (25% resection (n = 10) scar with no fistula, no 1 T, 94%; Prosthesis removal unicentric MRSA) Mean 7 days IV antibiotics antibiotics 2 T, 86.2%; (reference) (not specified, adapted to since discharge and no Resection, 34% susceptibility) and reintervention needed 122.3 days oral (not Follow-up: specified, adapted to minimum of susceptibility, 64% 12 months combined with rifampicin) Senneville 2000–2006 Total hip and knee SA-PJI DAIR (n = 41) (performed if Cure: no local or systemic Cure: 78.6%; Multivariable for cure: et al. Retrospective N = 98 (17.3% MRSA) symptoms \ 4 weeks) signs of infection, no need DAIR, 78.0%; ASA score B 2, use of (2011) [5] cohort, 1 T (n = 14) for reintervention or new rifampicin- 1 T, 100%; unicentric antibiotic therapy, no fluoroquinolone 2 T (n = 26) 2 T, 84.6%; infection-related deaths Resection arthroplasty Resection, 44.4%; (n = 9) Arthrodesis, Arthrodesis (n = 8) 62.5% Mean 7 days IV antibiotics Follow-up: (not specified, adapted to minimum of susceptibility) and 2 years 3–6 months oral (rifampicin, n = 68; rifampicin-fluorquinolone, n = 39). Six DAIR with suppressive treatment Infect Dis Ther (2022) 11:2177–2203 Table 4 continued References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design Lora-Tamayo 2003–2010 Early and haematogenous DAIR (n = 345) Failure: infection-related Failure: 45% Multivariable (HR) for early et al. Retrospective hip (total and partial) Median days of antibiotics: death, prosthesis (infection-related failure: male, rheumatoid (2013) [6] cohort, and knee SA-PJI 94 in MRSA and 91 in replacement, need for death, 7%) arthritis, bacteraemia, multicentre N = 345 (23% MRSA) MSSA. Rifampicin further debridements polymicrobial, – Early failure Infect Dis Ther (2022) 11:2177–2203 combinations in 88%: beta- CRP [ 100 mg/L (30 days), 29% lactams (13%) or Failure during therapy: ; Failure during quinolones (75%, mainly higher age, antibiotic therapy, levofloxacin) in MSSA, and immunosuppressive drugs, 32%; glycopeptides (18%, namely MRSA, sinus tract, vancomycin), cotrimoxazole – Failure after abnormal radiography, antibiotic therapy, need C 2 debridements, (46%), linezolid (24%) or 39%; no use of rifampicin clindamycin (10%) in MRSA Follow-up: 2 years Failure after therapy: haematogenous infection, debridement delay, need for C 2 debridements Gómez- 2003–2010 Hip and knee SA-PJI Implant removal Failure: local failure and/or Failure: 15.6% Multivariable for failure: Junyent Retrospective N = 249 (early, n = 141; (161 initial therapy, 88 all-cause mortality within Local failure: 9.3% C 2 comorbidities et al. cohort, haematogenous, n = 26; salvage) 60 days Mortality: 12.8% (2021) multicentre chronic, n = 82) 1 T, n = 17 (6.8%) Follow-up: median [11] 2 T, n = 188 (75.5%) 781 days, Hip resection, n = 44 interquartile (17.7%) range [IQR] 355–1375 days 2195 Table 4 continued 2196 References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design Betz et al. 1996–2012 Hip (total and partial) PJI DAIR (n = 38) Persistence or recurrence of Failure: 18.4% Univariate analysis: no (2015) Retrospective N = 29 (12 MSSA, 17 All exchange of mobile parts. PJI MSSA, 9.5% factors associated with [15] cohort, MRSA, 9 streptococci) Antibiotic adapted to failure MRSA, 29.4% unicentric susceptibility, with median Streptococci, 0% duration of Follow-up: 12 weeks (IV for a median of minimum of 14 days). Rifampicin use: 23 2 years (60.5%) Bouaziz et al. 2000–2010 Hip and knee MSSA-PJI DAIR (n = 62) Failure: need for further Failure: 42% Multivariable (HR) for (2018) Retrospective N = 85 (early, 33; delayed, Antibiotic not specified, surgery to control PJI (or to Early, 10 (30%) failure: non-compliance [13] cohort, two 11; late, 45) except for rifampicin treat superinfection), with surgical IDSA Delayed, 4 (36%) centres amputation or infection- guidelines (namely DAIR Rifampicin use: early 20 Late, 23 (51%) related death performed if duration of (62%), delayed 7 (63%), late Follow-up: symptoms \ 3 weeks or 26 (58%) 2.8 ± 2.2 years joint age \ 30 days, and stable implant without sinus tract) Lesens et al. 2010–2014 Hip, knee, shoulder, elbow DAIR (n = 137) Failure: prosthesis removal, Failure: 25% Multivariable (HR) for (2018) Retrospective SA-PJI Mean duration antibiotic: death, additional Mortality: 8.8% failure: longer AB [12] cohort, N = 137 (early acute, 12.6 weeks debridement or course of duration (protective), Follow-up: 2 years multicentre n = 63; early chronic, antibiotics, clinical and rifampicin regimen Rifampicin use: 85 (65%) n = 26; late acute, microbiological signs of (protective), smoking, Rifampicin plus infection early infection (protective) n = 35; late chronic, fluorquinolone: 63 (47.4%) n = 13) Suppressive treatment: 14 MRSA 27 (19.7%) (10.2%) Infect Dis Ther (2022) 11:2177–2203 Table 4 continued References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design Wouthuyzen- 2005–2015 Early hip and knee MSSA- DAIR (n = 58) Failure: need for revision Cure: 86.4% Not performed Bakker Retrospective PJI treated with Duration antibiotic therapy surgery (87.5% in et al. cohort, two rifampicin plus about 90 days. IV cloxacillin and/or suppressive levofloxacin (2018) [7] centres levofloxacin (n = 40) or or flucloxacillin 7–14 days antimicrobial therapy group, 84.2% in moxifloxacin (n = 19) moxifloxacin Infect Dis Ther (2022) 11:2177–2203 followed rifampicin plus because of persistent N = 58 fluorquinolone infection, death-related group) infection, reinfection or Follow-up: relapse with S. aureus minimum 2 years Muñoz- 2016–2017 Hip (total and partial) and DAIR (n = 55) Failure: all-cause death within Overall failure: Univariate analysis: Gallego Prospective knee SA-PJI Prosthesis removal (n = 30) 90 days, persistent or 36.4% (DAIR, Global failure associated et al. cohort, N = 85 (19 MRSA) relapsing signs 47.2%; prosthesis with DAIR Duration and type of (2020) [8] multicentre of staphylococcal infection, removal, 16.6%) Early (\ 90 days): 45 antibiotic not specified DAIR: failure associated need for salvage therapy Follow-up: at least with delay of debridement Chronic ([ 90 days): 21 except for extra debridements 1 year and no use of rifampicin Haematogenous: 19 in the first 30 days, suppressive antimicrobial treatment 2197 Table 4 continued 2198 References Years of Types of infection and Number of patients and Cure or failure definition Outcome Factors associated with diagnosis and number of patients management outcome design ARTHR-IS 2014–2016 Hip (total and partial) and DAIR (n = 99) Clinical and functional failure Overall clinical and Multivariable analysis: cohort Retrospective knee postsurgical SA-PJI 1 T (n = 10) after the first procedure and functional failure Failure after first surgical cohort, N = 128 (28 MRSA) at 18 months after first surgical procedure: Charlson C 2, 2 T (n = 13) multicentre Clinical failure: all-cause procedure: 32.8% haemoglobin Resection (n = 6) (DAIR, 33.3%; deaths in first 2 months, SA- level \ 10 mg/dL, Median days of antibiotics: PJI-related death, persistence prosthesis bacteraemia, polymicrobial 16 IV and 50 oral. or recurrence of signs or removal, 37.1%) infection and need to Rifampicin combinations in symptoms of infection; need Considering only perform additional 103 (80.5%); with for additional course of clinical failure: debridement quinolones in 42 (65.8%) antibiotics including 28.1% In DAIR: those above and suppressive therapy, removal Failure at BMI [ 30, and time of the prosthesis if DAIR. 18 months: between symptom onset Functional failure: 24.2% (DAIR: and DAIR; rifampicin use significantly impaired 21.2%; prosthesis was protective walking due to prosthesis removal, 34.4%) At 18 months: haemoglobin loosening, Girdlestone or Considering only level \ 10 mg/dL, hip arthrodesis clinical failure: fracture and need for 15.5% (DAIR: additional joint, surgical 15.5%; prosthesis procedures not related to removal: 17.2%) infection control Follow-up: 18 months 1 T one-stage replacement, 2 T two-stage replacement, CRP C-reactive protein, DAIR debridement, antibiotics, and implant retention, ESR erythrocyte sedimentation rate, HR hazard ratio, IV intravenous, MRSA methicillin-resistant S. aureus., MSSA methicillin-susceptible S. aureus, PJI prosthetic joint infection, SA-PJI prosthetic joint infection produced by S. aureus, TKA total knee arthroplasty Infect Dis Ther (2022) 11:2177–2203 Infect Dis Ther (2022) 11:2177–2203 2199 used the ASA index [5] or C 2 comorbidities for selection and immortal time bias. In our study, similar reasons [11]. Anaemia was previously on the other hand, we found that rifampicin identified as a predictor in one study [16], but combinations were associated with a protective was not assessed at all in most of the others. effect, even after performing sensitivity and Anaemia is a potential marker of nutritional landmark analyses. status that can increase tissue hypoxia or even Although previous studies have found higher indicate a systemic inflammatory process. As in TF rates in PJIs caused by MRSA compared with our study, bacteraemia, polymicrobial infection susceptible strains [3, 15], our data did not and the need for additional debridement were demonstrate this association, which is in line also predictors of TF in other studies [6, 14]. with more recent observations [5, 6]. Whether Since DAIR is the most frequent initial pro- this is due to the use of anti-MRSA drugs with cedure, we also analysed predictors of TF in this good bioavailability and anti-biofilm activity, subgroup. Apart from the above variables, obe- such as linezolid, would require further studies. sity and delay in performing DAIR were also Finally, we also analysed the predictors of TF identified as risk factors, while the use of after all procedures had been performed. Apart rifampicin had a protective effect. Obesity is a from other factors, hip fracture increased the risk factor for PJI and was also found to be risk of TF, probably reflecting the frailty of the associated with TF in hip PJI undergoing two- patients affected. In these patients, the first stage replacement [17], but was not even con- surgical approach was crucial, since initial TF sidered in most studies. Obesity could be asso- was followed by functional failure in all cases ciated with wound complications, additional (data not shown). In a previous multicentre debridement, impaired innate immune cohort study of patients with hip PJIs, fracture response and changes in the pharmacokinetics was also associated with clinical failure and of some antimicrobial drugs [18]. Delayed DAIR worse functional prognosis [25]. The risk of TF following onset of PJI symptoms is a known also increased when additional joint surgery not factor for TF regardless of the microorganism due to persistent infection was performed; this involved, but is particularly pertinent in the may have been due to tissue damage, delayed case of SA-PJI [2, 6, 12–14]. Establishing a healing or facilitation of bacterial threshold is complex. The 21-day threshold for superinfections. performing DAIR recommended by the guide- This study has some limitations that should lines [9] was based on one small-scale study be considered when interpreting the results. experience [19], and a delay of [ 2 days in The sample size was too small to investigate patients with SA-PJI treated with beta-lactams predictors of other initial surgical procedures was associated with increased TF in another and may have been insufficient to detect addi- study [2] but not when a rifampicin-fluoro- tional predictors of TF; its retrospective design quinolone combination was used [6–14, 16]. In limited the available variables; residual con- our cohort, we found an increased risk for each founding is also possible; we did not collect data day of delay, supporting the recommendation about ethnicity of the patients; finally, some that debridement should be performed as early changes in management may have occurred as possible. during the study period. Some strengths include Rifampicin in combination with other that it is a multinational study, the definitions antibiotics (mainly fluoroquinolones) was of TF including clinical and functional aspects reported to increase cure rates in a small ran- and the good predictive ability of the models domized trial [19] and in observational studies developed. [5, 6, 12, 20–22]. However, two recent meta- analyses found controversial results on the role of rifampicin: one found no benefit in staphy- CONCLUSIONS lococcal infections [23] and the other only a limited impact [24]. The studies included in the In conclusion, we observed that a considerable meta-analysis had a considerable risk of proportion of SA-PJIs failed after initial surgical 2200 Infect Dis Ther (2022) 11:2177–2203 treatments, although a substantial part of them Universitaire de Bordeaux—Bordeaux, France), were recovered with further procedures. Signif- Remco J.A. van Wensen (Department of icant functional loss should be considered Orthopaedic Surgery & Trauma, Catharina alongside clinical failure, and the importance of Hospital, Eindhoven, the Netherland), Gabriella certain risk factors for TF was confirmed, Lindergard (North Manchester General Hospi- including the protective role of rifampicin- tal—Manchester, UK) collaborated in the data based treatment in DAIR. collection. Funding. This work was supported by the ACKNOWLEDGEMENTS Innovative Medicines Initiative Joint Under- taking (grant agreement No. 115523), COM- We thank all study participants for their col- BACTE-NET consortium (European Union FP7/ laboration, without which this manuscript 2007–2013 and GlaxoSmithKline Biologicals SA, would have been impossible. as EFPIA partner). R.E, L.S, O. M, R. E-S, J. L–T, J. Other members of the ARTHR-IS group are: P, J. R-B and MD. del T are members of the Nienke Cuperus and Giuseppe Manfré (Julius Spanish Network for Research in Infectious Center for Health Sciences and Primary Care, Diseases (REIPI), supported by Plan Nacional de University Medical Center Utrecht, Utrecht, the I ? D ? i 2013-2016 and Instituto de Salud Netherlands) collaborated in the selection pro- Carlos III, Subdirección General de Redes y cess of the participating centers. Ana Isabel Centros de Investigación Cooperativa, Ministe- Suárez-Barrenechea and Alvaro Pascual-Her- rio de Ciencia, Innovación y Universidades, nandez (Department of microbiology, Hospital Spanish Network for Research in Infectious Universitario Virgen Macarena), Alba Rivera Diseases (REIPI RD16/0016/0001; 0002; 0005; (Deparment of Microbiology, Hospital de la 0009; 0011; 0015), co- financed by European Santa Creu i Sant Pau), Xavier Crusi and Marcos Development Regional Fund ‘‘A way to achieve Jordán (Department of traumatology, Hospital Europe’’, Operative Program Intelligence de la Santa Creu i Sant Pau), Nicolò Rossi Growth 2014-2020. The study sponsor is also (Department of Medical and Surgical Sciences, funding the journal’s Rapid Service Fee. Role of University of Bologna, Bologna, Italy), Tessa funding source: The funders had no influence van de Kerkhof (Department of Orthopaedic on the analysis and decision to published; Surgery & Trauma, Máxima MC, Eindhoven, GlaxoSmithKline Biologicals SA was provided the Netherlands; Department of Orthopaedic the opportunity to review a version of this Surgery & Trauma, Catharina Hospital, Eind- manuscript for factual accuracy; authors are hoven, the Netherlands), Juan P. Horcajada, solely responsible for final content and Joan Gómez-Junyent and Albert Alier (Hospital interpretation. del Mar, Institut Hospital del Mar d’Investiga- cions Mèdiques, Universitat Pompeu Fabra, Authorship. All named authors meet the Barcelona, Spain), Miranda van Rijen and Jan- International Committee of Medical Journal nie Romme (Amphia Hospital, Breda, Nether- Editors (ICMJE) criteria for authorship for this lands), Juliane Ankert (Jena University article, take responsibility for the integrity of Hospital—Jena, Germany), Celia Whitehouse the work as a whole, and have given their and Adrian Jones (Norfolk and Norwich approval for this version to be published. University Hospital—Norwich, UK), Javier Cobo and Javier Moreno (Hospital Universitario Author Contributions. Espindola R, Vella V, Ramón y Cajal—Madrid, Spain), Anne Meheut Rodrı́guez-Baño J, Del Toro MD contributed in (Centre Hospitalier Universitaire de Rennes— the conception and design of the study, analysis Rennes, France), Claire Gledel (Orthopedic sur- and interpretation of data and drafting the gery department, Croix Rousse Hospital—Lyon, article. Benito N, Mur I, Tedeschi S, Zamparini France), Pauline Perreau (Centre Hospitalier E, Hendriks J, Sorlı́ L, Murillo O, Soldevila L, Scarborough M, Scarborough C, Kluytmans J, Infect Dis Ther (2022) 11:2177–2203 2201 Ferrari MC, Pletz M, Mcnamara I, Escudero- participating centers, with the names and ref- Sanchez R, Arvieux C, Batailler C, Dauchy F-A, erences, are reflected in the table S5 of the Liu W-Y, Lora-Tamayo J, Praena J, Ustianowski J supplementary material. contributed in the acquisition of data and revising the article. Cinconze E, Pellegrini M Data Availability. The datasets generated and Bagnoli F participated in the analysis and and analyzed during the current study are interpretation of data and revision the article. available from the corresponding author on All authors gave their final approval of the ver- reasonable request. sion to be submitted. Open Access. This article is licensed under a Prior Presentation. This study was previ- Creative Commons Attribution-NonCommer- ously presented as a poster at the 32nd Euro- cial 4.0 International License, which permits pean Congress of Clinical Microbiology & any non-commercial use, sharing, adaptation, Infectious Diseases (ECCMIC), held in Lisbon, distribution and reproduction in any medium Portugal, April 23–26. or format, as long as you give appropriate credit to the original author(s) and the source, provide Disclosures. Venanzio Vella, Elisa Cinconze, a link to the Creative Commons licence, and Michele Pellegrini and Fabio Bagnoli are indicate if changes were made. The images or employees of the GSK group of companies. other third party material in this article are Venanzio Vella, Michele Pellegrini and Fabio included in the article’s Creative Commons Bagnoli hold shares in the GSK group of com- licence, unless indicated otherwise in a credit panies. Fabio Bagnoli holds patents pending line to the material. If material is not included and issued patents on Staphylococcus aureus in the article’s Creative Commons licence and vaccine formulations. Venanzio Vella, Elisa your intended use is not permitted by statutory Cinconze, Michele Pellegrini and Fabio Bagnoli regulation or exceeds the permitted use, you declare no other financial or non-financial will need to obtain permission directly from the relationships and activities. Reinaldo Espindola, copyright holder. To view a copy of this licence, Natividad Benito, Isabel Mur, Sara Tedeschi, visit http://creativecommons.org/licenses/by- Eleonora Zamparini, Johannes G.E. Hendriks, nc/4.0/. Luisa Sorlı́, Oscar Murillo, Laura Soldevila, Mathew Scarborough, Claire Scarborough, Jan Kluytmans, Mateo Carlo Ferrari, Mathias W. Pletz, Iain Mcnamara, Rosa Escudero-Sanchez, REFERENCES Cedric Arvieux, Cecile Batailler, Frédéric- Antoine Dauchy, Wai-Yan Liu, Jaime Lora- 1. Tande AJ, Patel R. Prosthetic joint infection. Clin Tamayo, Julia Praena, Andrew Ustianowski, Microbiol Rev. 2014;27:302–45. https://doi.org/10. 1128/CMR.00111-13. Jesús Rodrı́guez-Baño and Maria Dolores Del Toro have no conflicts to declare. At present, 2. Brandt CM, Sistrunk WW, Duffy MC, et al. Sta- the affiliation of Reinaldo Espindola is Infec- phylococcus aureus prosthetic joint infection trea- tious Diseases and Microbiology Clinical Unit, ted with debridement and prosthesis retention. Clin Infect Dis 1997;24:914–919. https://pubmed. Hospital Universitario Virgen de Valme, Seville, ncbi.nlm.nih.gov/9142792/. Accessed 4 June 2021. Spain. https://doi.org/10.1093/CLINIDS/24.5.914. Compliance with Ethics Guidelines. The 3. Salgado CD, Dash S, Cantey JR, Marculescu CE. study was approved by the Ethics Committees Higher risk of failure of methicillin-resistant Sta- phylococcus aureus prosthetic joint infections. Clin at each site. The need to obtain written Orthop Related Res. 2007:48–53. https://pubmed. informed consent was waived due to the retro- ncbi.nlm.nih.gov/17534195/. Accessed 3 June spective nature of the study and anonymized 2021. https://doi.org/10.1097/BLO. data, except in the case of the French hospitals. 0b013e3181123d4e. The approval of the ethics committee of all the 2202 Infect Dis Ther (2022) 11:2177–2203 4. Joulie D, Girard J, Mares O, et al. Factors governing retrospective multicenter study. Antibiotics the healing of Staphylococcus aureus infections 2021;10:1–13. https://pubmed.ncbi.nlm.nih.gov/ following hip and knee prosthesis implantation: a 33530523/. Accessed 3 June 2021. https://doi.org/ retrospective study of 95 patients. Orthop Trauma- 10.3390/antibiotics10020118. tol Surg Res 2011;97:685–692. https://pubmed.ncbi. nlm.nih.gov/22000281/. Accessed 20 Mar 2022. 12. Lesens O, Ferry T, Forestier E, et al. Should we https://doi.org/10.1016/J.OTSR.2011.05.013. expand the indications for the DAIR (debridement, antibiotic therapy, and implant retention) proce- 5. Senneville E, Joulie D, Legout L, et al. Outcome and dure for Staphylococcus aureus prosthetic joint predictors of treatment failure in total hip/knee infections? A multicenter retrospective study. Eur J prosthetic joint infections due to staphylococcus Clin Microbiol Infect Dis 2018;37:1949–1956. aureus. Clin Infect Dis 2011;53:334–340. https:// https://pubmed.ncbi.nlm.nih.gov/30083889/. pubmed.ncbi.nlm.nih.gov/21810745/. Accessed 3 Accessed 20 Mar 2022. https://doi.org/10.1007/ June 2021. https://doi.org/10.1093/CID/CIR402. S10096-018-3330-7. 6. Lora-Tamayo J, Murillo O, Iribarren JA, et al. A large 13. Bouaziz A, Uçkay I, Lustig S, et al. Non-compliance multicenter study of methicillin-susceptible and with IDSA guidelines for patients presenting with methicillin-resistant staphylococcus aureus pros- methicillin-susceptible Staphylococcus aureus thetic joint infections managed with implant prosthetic joint infection is a risk factor for treat- retention. Clin Infect Dis 2013;56:182–194. https:// ment failure. Med Mal Infect 2018;48:207–211. pubmed.ncbi.nlm.nih.gov/22942204/. Accessed 3 https://pubmed.ncbi.nlm.nih.gov/29122410/. June 2021. https://doi.org/10.1093/cid/cis746. Accessed 3 June 2021. https://doi.org/10.1016/J. MEDMAL.2017.09.016. 7. Wouthuyzen-Bakker M, Tornero E, Morata L, et al. Moxifloxacin plus rifampin as an alternative for 14. Vilchez F, Martı́nez-Pastor JC, Garcı́a-Ramiro S, levofloxacin plus rifampin in the treatment of a et al. Outcome and predictors of treatment failure prosthetic joint infection with Staphylococcus in early post-surgical prosthetic joint infections due aureus. Int J Antimicrob Agents 2018;51:38–42. to Staphylococcus aureus treated with debridement. https://pubmed.ncbi.nlm.nih.gov/28668678/. Clin Microbiol Infect 2011;17:439–444. https:// Accessed 20 Mar 2022. https://doi.org/10.1016/J. pubmed.ncbi.nlm.nih.gov/20412187/. Accessed 3 IJANTIMICAG.2017.04.011. June 2021. https://doi.org/10.1111/J.1469-0691. 2010.03244.X. 8. Muñoz-Gallego I, Viedma E, Esteban J, et al. Genotypic and phenotypic characteristics of 15. Betz M, Abrassart S, Vaudaux P, et al. Increased risk staphylococcus aureus prosthetic joint infections: of joint failure in hip prostheses infected with Sta- insight on the pathogenesis and prognosis of a phylococcus aureus treated with debridement, multicenter prospective cohort. Open forum Infect antibiotics and implant retention compared to Dis 2020;7. https://pubmed.ncbi.nlm.nih.gov/ Streptococcus. Int Orthop 2015;39:397–401. 33005695/. Accessed 20 Mar 2022. https://doi.org/ https://pubmed.ncbi.nlm.nih.gov/25183296/. 10.1093/OFID/OFAA344. Accessed 20 Mar 2022. https://doi.org/10.1007/ S00264-014-2510-Z. 9. Osmon DR, Berbari EF, Berendt AR, et al. Diagnosis and management of prosthetic joint infection: 16. Swenson RD, Butterfield JA, Irwin TJ, Zurlo JJ, Davis Clinical practice guidelines by the infectious dis- CM. Preoperative anemia is associated with failure eases Society of America. Clin Infect Dis 2013;56. of open debridement polyethylene exchange in https://pubmed.ncbi.nlm.nih.gov/23223583/. acute and acute hematogenous prosthetic joint Accessed 13 June 2021. https://doi.org/10.1093/ infection. J Arthroplasty 2018;33:1855–1860. CID/CIS803. https://pubmed.ncbi.nlm.nih.gov/29555498/. Accessed 4 June 2021. https://doi.org/10.1016/J. 10. Ariza J, Cobo J, Baraia-Etxaburu J, et al. Executive ARTH.2018.01.042. summary of management of prosthetic joint infec- tions. Clinical practice guidelines by the Spanish 17. Ahmad SS, Orlik L, Ahmad SJS, Albers CE, Sieben- Society of Infectious Diseases and Clinical Micro- rock KA, Klenke FM. Obesity and smoking predict biology (SEIMC). Enferm Infecc Microbiol Clin the results of two-stage exchange in septic revision 2017;35:189–195. https://linkinghub.elsevier.com/ hip arthroplasty: a cohort study. Orthop Traumatol retrieve/pii/S0213005X16302816. Accessed 1 Apr Surg Res 2019;105:467–471. https://pubmed.ncbi. 2019. https://doi.org/10.1016/j.eimc.2016.08.012. nlm.nih.gov/30922806/. Accessed 4 June 2021. https://doi.org/10.1016/j.otsr.2019.01.006. 11. Gómez-Junyent J, Lora-Tamayo J, Baraia-Etxaburu J, et al. Implant removal in the management of 18. Falagas ME, Kompoti M. Obesity and infection. prosthetic joint infection by staphylococcus aureus: Lancet Infect Dis 2006;6:438–446. https://pubmed. outcome and predictors of failure in a large ncbi.nlm.nih.gov/16790384/. Accessed 20 Mar Infect Dis Ther (2022) 11:2177–2203 2203 2022. https://doi.org/10.1016/S1473- 23. Aydın O, Ergen P, Ozturan B, Ozkan K, Arslan F, 3099(06)70523-0. Vahaboglu H. Rifampin-accompanied antibiotic regimens in the treatment of prosthetic joint 19. Zimmerli W, Widmer AF, Blatter M, Frei R, Ochsner infections: a frequentist and Bayesian meta-analysis PE. Role of rifampin for treatment of orthopedic of current evidence. Eur J Clin Microbiol Infect Dis implant-related staphylococcal infections: a ran- 2021;40:665–671. https://pubmed.ncbi.nlm.nih. domized controlled trial. J Am Med Assoc 1998;279: gov/33125602/. Accessed 20 Mar 2022. https://doi. 1537–1541. https://pubmed.ncbi.nlm.nih.gov/ org/10.1007/S10096-020-04083-4. 9605897/. Accessed 4 June 2021. https://doi.org/ 10.1001/jama.279.19.1537. 24. Scheper H, Gerritsen LM, Pijls BG, Van Asten SA, Visser LG, De Boer MGJ. Outcome of debridement, 20. El Helou OC, Berbari EF, Lahr BD, et al. Efficacy and antibiotics, and implant retention for staphylococ- safety of rifampin containing regimen for staphy- cal hip and knee prosthetic joint infections, focused lococcal prosthetic joint infections treated with on rifampicin use: a systematic review and meta- debridement and retention. Eur J Clin Microbiol analysis. Open forum Infect Dis 2021;8. https:// Infect Dis 2010;29:961–967. https://pubmed.ncbi. pubmed.ncbi.nlm.nih.gov/34258321/. Accessed 20 nlm.nih.gov/20505968/. Accessed 4 June 2021. Mar 2022. https://doi.org/10.1093/OFID/OFAB298. https://doi.org/10.1007/s10096-010-0952-9. 25. Del Toro MD, Nieto I, Guerrero F, et al. Are hip 21. Widmer AF, Gaechter A, Ochsner PE, Zimmerli W. hemiarthroplasty and total hip arthroplasty infec- Antimicrobial treatment of orthopedic implant-re- tions different entities? The importance of hip lated infections with rifampin combinations. Clin fractures. Eur J Clin Microbiol Infect Dis 2014;33: Infect Dis 1992;14:1251–1253. https://pubmed. 1439–1448. https://pubmed.ncbi.nlm.nih.gov/ ncbi.nlm.nih.gov/1623081/. Accessed 4 June 2021. 24671411/. Accessed 4 June 2021. https://doi.org/ https://doi.org/10.1093/clinids/14.6.1251. 10.1007/s10096-014-2091-1. 22. Beldman M, Löwik C, Soriano A, et al. If, when, and how to use rifampin in acute staphylococcal Publisher’s Note periprosthetic joint infections, a multicentre observational study. Clin Infect Dis 2021. https:// Springer Nature remains neutral with regard to jurisdic- pubmed.ncbi.nlm.nih.gov/33970214/. Accessed 4 tional claims in published maps and institutional June 2021. https://doi.org/10.1093/cid/ciab426. affiliations.