The Different Microbial Etiology of Prosthetic Joint Infections according to Route of Acquisition and Time after Prosthesis Implantation, Including the Role of Multidrug-Resistant Organisms

Journal of Clinical Medicine Article The Different Microbial Etiology of Prosthetic Joint Infections according to Route of Acquisition and Time after Prosthesis Implantation, Including the Role of Multidrug-Resistant Organisms Natividad Benito 1,2,* , Isabel Mur 1,2, Alba Ribera 3, Alex Soriano 4, Dolors Rodríguez-Pardo 5, Luisa Sorlí6, Javier Cobo 7, Marta Fernández-Sampedro 8, María Dolores del Toro 9, Laura Guío10, Julia Praena 11, Alberto Bahamonde 12, Melchor Riera 13, Jaime Esteban 14 , Josu Mirena Baraia-Etxaburu 15, Jesús Martínez-Alvarez 16, Alfredo Jover-Sáenz 17, Carlos Dueñas 18, Antonio Ramos 19, Beatriz Sobrino 20, Gorane Euba 3, Laura Morata 4, Carles Pigrau 5, Juan P. Horcajada 6, Pere Coll 2,21, Xavier Crusi 22, Javier Ariza 3and on behalf of the REIPI (Spanish Network for Research in Infectious Disease) Group for the Study of Prosthetic Joint Infections/GEIO (Group for the Study of Osteoarticular Infections), SEIMC (Spanish Society of Infectious Diseases and Clinical Microbiology) † 1Infectious Diseases Unit, Hospital de la Santa Creu i Sant Pau - Institut d’InvestigacióBiomèdica Sant Pau, 08025 Barcelona, Spain; [email protected] 2Department of Medicine, Universitat Autònoma de Barcelona, 08193 Barcelona, Spain; [email protected] 3Department of Infectious Diseases, Hospital Universitari Bellvitge, 08097 Barcelona, Spain; [email protected] (A.R.); [email protected] (G.E.); [email protected] (J.A.) 4Department of Infectious Diseases, Hospital Clínic Universitari, 08036 Barcelona, Spain; [email protected] (A.S.); [email protected] (L.M.) 5Department of Infectious Diseases, Hospital Universitari Vall d’Hebron, 08035 Barcelona, Spain; [email protected] (D.R.-P.); [email protected] (C.P.) 6Department of Infectious Diseases, Parc de Salut Mar, 08003 Barcelona, Spain; [email protected] (L.S.); [email protected] (J.P.H.) 7Department of Infectious Diseases, Hospital Universitario Ramón y Cajal – IRYCIS, 28034 Madrid, Spain; javier[email protected] 8Department of Infectious Diseases, Hospital Universitari Valdecilla, 39008 Santander, Spain; [email protected] 9Department of Infectious Diseases, Clinical Microbiology and Preventive Medicine, Hospital Universitario Virgen Macarena, 41009 Sevilla, Spain; mdeltor[email protected] 10 Department of Infectious Diseases, Hospital Universitario Cruces, 48903 Bilbao, Spain; [email protected] 11 Department of Infectious Diseases, Clinical Microbiology and Preventive Medicine, Hospital Universitario Virgen del Rocío, 41013 Sevilla, Spain; [email protected] 12 Department of Internal Medicine-Infectious Diseases, Hospital Universitario del Bierzo, 24404 Le ó n, Spain; [email protected] 13 Infectious Diseases Unit, Department of Internal Medicine, Hospital Universitario Son Espases, 07120 Palma de Mallorca, Spain; melchor[email protected] 14 Department of Clinical Microbiology, IIS-Fundación Jiménez Díaz, 28040 Madrid, Spain; [email protected] 15 Department of Infectious Diseases, Hospital Universitario de Basurto, 48013 Bilbao, Spain; josumirena.baraia-etxabur[email protected] 16 Department of Orthopedic and Traumatology, Hospital Universitario Central de Asturias, 33011 Oviedo, Spain; [email protected] 17 Unit of Nosocomial Infection, Hospital Universitari Arnau de Vilanova, 25198 Lleida, Spain; ajover[email protected] 18 Department of Internal Medicine, Hospital Clínico Universitario de Valladolid, 47003 Valladolid, Spain; [email protected] J. Clin. Med. 2019,8, 673; doi:10.3390/jcm8050673 www.mdpi.com/journal/jcm J. Clin. Med. 2019,8, 673 2 of 15 19 Infectious Diseases Unit, Department of Internal Medicine, Hospital Universitario Puerta de Hierro, 28222 Madrid, Spain; [email protected] 20 Department of Infectious Diseases, Hospital Regional Universitario Málaga, 29010 Málaga, Spain; [email protected] 21 Department of Clinical Microbiology, Hospital de la Santa Creu i Sant Pau - Institut d’Investigació Biomèdica Sant Pau, 08025 Barcelona, Spain 22 Department of Orthopedic and Traumatology, Hospital de la Santa Creu i Sant Pau - Institut d’Investigaci ó Biomèdica Sant Pau, 08025 Barcelona, Spain; xcr[email protected] *Correspondence: [email protected]; Tel.: +34-935-565-624; Fax: +34-935-565-938 †Members of the Group (in alphabetical order): Hospital Universitari Arnau de Vilanova, Lleida: Mercé García-González, Ferran Pérez-Villar, Laura Prats-Gispert, María Ramírez-Hidalgo. Hospital Universitario de Basurto, Bilbao: Mireia De-la-Peña-Trigueros, Iñigo L ó pez-Azkarreta, Miriam L ó pez-Mart í nez. Hospital Universitari de Bellvitge, Barcelona: Jaime Lora-Tamayo (currently, Hospital 12 de Octubre), Oscar Murillo, Salvador Pedrero. Hospital el Bierzo, León: Juan Amador, Luis García-Paíno, Susana Parrondo, Carmen Raya. Hospital Universitario Central de Asturias, Oviedo: Alfonso Moreno. Hospital Universitario de Burgos, Burgos: María Aránzazu Blanco-Martínez-de-Morentin, Laura Rodríguez-Fernández Genoveva Zapico-Aldea. Hospital Clínic Universitari, Barcelona: Luis Lozano, Ernesto Muñoz-Mahamud. Hospital Universitario Cruces, Bilbao: Josu Merino-Pérez. Parc de Salut Mar, Barcelona: Albert Alier, Núria Prim, Luis Puig. IIS-Fundación Jiménez Díaz: Álvaro Auñon, Antonio Blanco, Joaquín García-Cañete, Raul Parrón-Cambero. Hospital Universitario Marqués de Valdecilla, Santander: M Carmen Fariñas, Ricardo Manuel-Bloque, Carlos Salas-Venero. Hospital Universitario Puerta de Hierro: Jesús Campo-Loarte, Elena Múñez-Rubio, Isabel Sánchez-Romero. Hospital Universitario Ramón y Cajal, Madrid: Rosa Escudero-Sánchez, Patricia Ruiz-Carbajosa. Hospital de la Santa Creu i Sant Pau, Barcelona: Aránzazu Gonz á lez, Jos é Carlos Gonz á lez, Marcos Jord á n, Alba Rivera. Hospital Universitario Son Espases, Palma de Mallorca: Carmen Marinescu, Francisco Montaner, Antonio Ramírez. Hospital Universitari Vall d’Hebron, Barcelona: Pablo S. Corona, Maily Lung, Mireia Puig-Asensio. Hospital Universitario Virgen Macarena, Sevilla: Miguel Ángel Muniain-Ezcurra, Cecilia Peñas-Espinar, Ana Isabel Suárez. Hospital Universitario Virgen del Rocío, Sevilla: María JoséGómez, Macarena López-Pliego. Received: 17 April 2019; Accepted: 9 May 2019; Published: 13 May 2019   Abstract: The aim of our study was to characterize the etiology of prosthetic joint infections (PJIs)—including multidrug-resistant organisms (MDRO)—by category of infection. A multicenter study of 2544 patients with PJIs was performed. We analyzed the causative microorganisms according to the Tsukayama’s scheme (early postoperative, late chronic, and acute hematogenous infections (EPI, LCI, AHI) and “positive intraoperative cultures” (PIC)). Non-hematogenous PJIs were also evaluated according to time since surgery: <1 month, 2–3 months, 4–12 months, >12 months. AHIs were mostly caused by Staphylococcus aureus (39.2%) and streptococci (30.2%). EPIs were characterized by a preponderance of virulent microorganisms (S. aureus, Gram-negative bacilli (GNB), enterococci), MDROs (24%) and polymicrobial infections (27.4%). Conversely, coagulase-negative staphylococci (CoNS) and Cutibacterium species were predominant in LCIs (54.5% and 6.1%, respectively) and PICs (57.1% and 15.1%). The percentage of MDROs isolated in EPIs was more than three times the percentage isolated in LCIs (7.8%) and more than twice the proportion found in AHI (10.9%). There was a significant decreasing linear trend over the four time intervals post-surgery for virulent microorganisms, MDROs, and polymicrobial infections, and a rising trend for CoNS, streptococci and Cutibacterium spp. The observed differences have important implications for the empirical antimicrobial treatment of PJIs. Keywords: prosthetic joint infections; microbial etiology; classification schemes for prosthetic joint infections; antimicrobial empirical treatment; multidrug-resistant organisms J. Clin. Med. 2019,8, 673 3 of 15 1. Introduction While the risk of prosthetic infection in patients undergoing joint replacement could be considered low (hips: 0.2–1.5%, knees: 0.4–1.5%, shoulders: 0.8–2%), the high frequency of these procedures converts the combination of low risks into a substantial burden of infection [ 1 ]. Prosthetic joint infection (PJI) is a devastating complication associated with major patient morbidity and high healthcare and societal costs (recent estimated costs of 20,000–40,000 dollars per infection) [1]. Biofilm formation when microorganisms attach to the surface of prosthetic devices plays a crucial role in the pathogenesis of PJI [ 2 ]. This poses a challenge for the diagnosis of biofilm-embedded microorganisms, and antimicrobial therapy in biofilm-associated pathogens is of limited efficacy [3]. Antimicrobial therapy combined with surgery is required to cure PJIs [ 4 – 7 ]. A common management approach is to start broad-spectrum antimicrobial agents after obtaining intraoperative samples for culture [ 8 , 9 ]. The importance of adequate initial empirical antimicrobial therapy in the outcome of infections is well-known [ 10 ] and seems to be critical in patients with PJIs treated with debridement, antibiotics, and implant retention [ 11 ]. Vancomycin combined with a broad-spectrum beta-lactam such as piperacillin-tazobactam has been recently studied as an initial treatment [ 12 , 13 ] but has been associated with a high rate of adverse effects [ 13 ]. After the pathogen has been identified and antimicrobial susceptibility results are available, the most effective narrow-spectrum antibiotic regimen is selected for continuation of therapy [ 4 – 8 , 10 ]. Nevertheless, a significant number of patients (5–35%) have negative cultures [ 8 ]. In this situation, empirical antimicrobial therapy is even more important, but more difficult to decide on, taking into consideration that the patient will have to receive it for the several weeks to months that it takes to cure a PJI [4,8]. Knowledge of the microbiological spectrum of PJIs is essential for guiding empirical antibiotic therapy. There are, however, no specific recommendations for the most appropriate empirical treatment for PJIs. We previously characterized the microbial etiology of PJI in a large cohort of patients [ 14 ], but the causative microorganisms can vary significantly, depending on the infection route and the time interval between index surgery and onset of symptoms, which can help guide empirical treatment [ 4 , 8 , 12 , 15 ]. These differences may include the involvement of multidrug-resistant microorganisms (MDRO), although this aspect has not been previously studied. There are several useful classifications of PJI in different categories, namely based on the mode of acquisition and/or time from prosthesis implantation, but none of them are universally accepted [ 8 , 16 ]. A good deal of what is currently known about the microbial etiology of different categories of PJI is based on studies that are limited by small sample sizes [ 11 , 15 , 17 – 31 ] and describe single-center experiences [ 12 , 15 , 17 – 24 , 26 – 28 , 30 – 32 ]. Most focus on specific types of infection, [ 11 , 21 , 23 , 25 , 29 ] surgical strategies used during treatment [ 22 , 31 ], or include only infections occurring within a limited period of time after prosthesis implantation [ 30 , 33 ]. Consequently, the results do not adequately represent the percentages of the different microorganisms involved in different types of infection across the full range of PJIs Our aim was to characterize the etiology of PJIs—including MDROs—according to the category of infection, in a large cohort of consecutive patients with PJIs. The results would enable tailoring empirical antimicrobial therapy to the clinical situation, offering coverage of the most likely microorganisms but narrowing the antimicrobial spectrum, which is crucial in antimicrobial stewardship program, and potentially reducing adverse effects. 2. Methods 2.1. Setting, Study Design, and Patients This was an ambi-directional observational study carried out at 19 hospitals in different areas of Spain. The study was performed within the framework of the Spanish Network for Research in Infectious Diseases (REIPI) (www.reipi.org) and included the participation of the Group for the Study of Osteoarticular Infections (GEIO) (https://seimc.org/grupos-de-estudio/geio) belonging to the Spanish Society of Clinical Microbiology and Infectious Diseases (SEIMC). The REIPI Group for the J. Clin. Med. 2019,8, 673 4 of 15 Study of PJIs and the GEIO form a multicenter collaborative group of infectious disease specialists, microbiologists, and orthopedic specialists across Spain, with extensive experience at orthopedic infection management and had previous joint publications [14,29,34–36]. All consecutive adult patients diagnosed with PJIs between 2003 and 2012 were included. Excluded were relapse episodes of infections that were first diagnosed before the study period. 2.2. Data Collection This study was ambi-directional, with both prospective and retrospective data collection. Data were first acquired from the REIPI cohort of consecutive patients with PJI who were prospectively enrolled from 2003 through 2006. The cohort characteristics have been described elsewhere [ 29 , 34 ]. Apart from clarifications concerning key variables, no further data on this prospective cohort was requested. For the retrospective phase, data of patients who developed PJIs from 2007 through 2012 were retrospectively collected from REIPI and other hospitals that met the criteria for participation. The three criteria included hospitals with orthopedic surgery, the use of proper identification procedures to ensure the inclusion of all consecutive cases diagnosed at the hospital, and that ascertainment bias was minimized, and finally, most of the data needed to resolve queries were either available or easily accessible. A standard case report form designed specifically for this study was used at all sites to collect data. Data of patients with PJI were obtained from electronic databases used at most of the participating hospitals with prospectively collected information on patients with PJI, and from the patient’s medical records held at each hospital as required. Completed case report forms were sent to the coordinating center for data entry, or the site investigators entered the variables directly into the common electronic database. The coordinating center for this study was the Hospital de la Santa Creu i Sant Pau (Barcelona, Spain). The study was approved by the Institutional Review Board at the Hospital de la Santa Creu i Sant Pau before data collection started. All case report forms were reviewed at the coordinating center. 2.3. Clinical Data and Definitions The following information was collected: Patient demographics and underlying conditions, characteristics of the arthroplasty, risk factors for MDROs, classification of the PJI, and microbiological diagnosis. All variables were predefined to ensure standardized data collection in participating hospitals. Individual patient data recorded included age and gender; comorbidities and immunosuppressive therapy; the patient’s American Society of Anesthesiologists (ASA) score before the surgical procedure (typically the arthroplasty implant) closest to the diagnosis of infection; previous exposure to antibiotics ( ≥ 7 days) or hospitalization in the previous 90 days ( ≥ 2 days); receipt of hemodialysis, intravenous therapy, wound care or specialized nursing care at home in the 30 days preceding the last surgical procedure or onset of hematogenous PJI; residence in a nursing home or long-term care facility. Information of the arthroplasty collected included: The reasons for implantation and the date performed, site, time from admission to implantation, primary or revision arthroplasty, cemented vs. uncemented, and the use of antibiotic-loaded bone cement. Date of diagnosis, classification of the PJI type, and the number of cultured samples and their results were also recorded. Cefazolin —sometimes cefuroxime, depending on the center, — was used as an antimicrobial prophylaxis in surgery. Vancomycin or teicoplanin was used for patients who were allergic to penicillin. Baseline comorbidities were quantified using the Charlson comorbidity score [ 37 ]. A diagnosis of PJI was established using the 2011 Musculoskeletal Infection Society definition [ 38 ]. The microbial etiology of PJI was established when the same organism (indistinguishable by common laboratory tests including genus and species identification or common antibiogram) was isolated in two or more periprosthetic cultures yielded [ 4 ]. When the diagnostic criteria for PJI were met, virulent microorganisms (such as Staphylococcus aureus) isolated in a single periprosthetic tissue/biopsy sample were also considered causative organisms [ 7 ]. MDRO was defined following Magiorakos et al. (acquired non-susceptibility to at least one agent in three or more specified antimicrobial categories) [39]. J. Clin. Med. 2019,8, 673 5 of 15 The Tsukayama scheme was used to classify PJIs. This scheme divides PJIs into four categories, based partly on presumed mode of infection and time since surgery: [ 15 , 40 ] a) Early postoperative infection (EPI): PJI diagnosed within one month of the index surgery (usually implantation of joint prosthesis, but also later procedures performed at the arthroplasty site); b) late chronic infection (LCI): PJIs with an insidious clinical course diagnosed >1 month after the index operation; both EPI and LCI are considered perioperatively acquired; c) acute hematogenous infection (AHI): PJI associated with documented or suspected antecedent bacteremia and characterized by acute onset of symptoms in the affected joint with the prosthesis; d) positive intraoperative culture (PIC): PJI diagnosed when at least two specimens, from a minimum of five obtained at the time of revision surgery, are positive after culture; infection was not clinically obvious or suspected at the time of the revision [ 15 , 40 ]. AHI and PIC can occur any time after surgery. EPI and AHI are acute PJIs that can be treated and potentially cured with debridement and antibiotics, without removal of the prosthesis. Another commonly used classification scheme based only on time since index surgery classifies PJIs as early (develops <3 months after surgery), delayed (3 to 12 or 24 months after surgery), or late (>12 or 24 months after surgery) [ 8 , 41 ]. This chronological framework was used to further analyse all cases of non-hematogenous PJI. 2.4. Statistical Analysis Medians and interquartile ranges were used to summarize continuous variables, and absolute numbers and percentages of total samples for categorical variables. Statistical analyses were based on differences in the percentages of causative microorganisms/groups of organisms of PJI in the four categories of the Tsukayama scheme (EPI, LCI, AHI, PIC). The causative microorganisms of non-haematogenous PJIs diagnosed according to time since surgery (within the first month, in months 2/3, months 4–12, and more than 12 months after the index surgery), were also compared. These percentages were compared using the χ 2 test or Fisher’s exact test as appropriate. To determine statistically significant linear trends in the proportions of infection caused by specific microorganisms/groups of organisms over time the Mantel-Haenszel χ 2 test for trend was used. The data were analyzed using SPSS, version 22.0 (IBM SPSS, Chicago, IL, USA) 3. Results Overall, 2524 episodes of PJI were diagnosed during the study period in 19 participating hospitals located in eight of the 17 administrative regions of Spain. All were university hospitals except for one. In 17 hospitals, they had more than 500 beds, and two had between 400 and 500 beds. The characteristics of the patients are outlined in Table 1. Most infections occurred in hip or knee arthroplasties; 77.2% affected primary arthroplasties. The most common reason for joint replacement was degenerative joint disease. LCI was the most frequent type of infection, accounting for 47.4% (1178) of cases, followed by EPI (35.7%, 888), AHI (11.6%, 288) and PIC (5.3%). A microbiological diagnosis was obtained in 2288 cases (90.6%) and significantly more frequently in EPI (94.5%, 839) and AHI (92%, 256) than in LCI (89.3%, 1052) (p<0.001). The causative microorganisms of PJI using Tsukayama’s classification are shown in Tables 2and 3. Overall, coagulase-negative staphylococci (CoNS) were the most common group of microorganisms involved in PJI (with Staphylococcus epidermidis as the most frequent species). They were more often isolated in chronic (>50% of cases) than in acute infections. Whereas CoNS represented almost 30% of EPIs, they were involved in less than 10% of AHI. S. aureus on the other hand, was the microorganism most often involved in acute infection and the leading causative species of EPI and AHI. Streptococci were significantly more common in AHI, while enterococci were more frequent in EPI than in other categories of PJI. J. Clin. Med. 2019,8, 673 6 of 15 Table 1. Characteristics of patients with prosthetic joint infection diagnosed between 2003 and 2012. Characteristic No. of Cases (n=2524) Median age (IQR), years 74 (13) Female gender 1508 (59.7) Underlying conditions •Any comorbid condition 1594 (63.3) •Diabetes mellitus 592 (23.5) •Heart disease 506 (20.1) •Chronic obstructive pulmonary disease 299 (11.9) •Cancer 231 (9.2) •Neurological disease 221 (8.8) •Chronic kidney disease 195 (7.7) • Systemic rheumatic (connective tissue) disease 175 (6.9) •Immunosuppressive treatment 168 (6.7) •Liver disease 164 (6.5) •Rheumatoid arthritis 129 (5.1) Charlson score, median (IQR) 1 (2) Index arthroplasty site •Hip 1244 (49.3) #Hemiarthroplasty 249 (9.9) #Total arthroplasty 995 (39.5) •Knee 1219 (48.3) •Shoulder 46 (1.8) •Other 15 (0.6) ASA score, median (IQR) 2 (1) Indication for index arthroplasty* •Primary joint replacement 1938 (77.2) #Osteoarthritis 1264 (52.4) #Fracture 417 (17.3) #Avascular necrosis 51 (2.1) #Rheumatoid arthritis 32 (1.3) #Tumor 31 (1.3) #Septic arthritis sequelae 12 (0.5) #Other 43 (1.8) • Revision arthroplasty (prior joint arthroplasty) 573 (22.8) #Aseptic loosening 292 (12.1) #Infection 158 (6.6) #Dislocation 32 (1.3) #Periprosthetic fracture 25 (1) #Implant failure or fracture 13 (0.5) #Other 29 (1.2) Unless stated otherwise, data refer to numbers (%) of patients with the indicated characteristic. ASA =American Society of Anesthesiologists; IQR =interquartile range. * Information on indication for index arthroplasty was not available for 112 (4.4%) procedures. Aerobic Gram-negative bacilli (GNB), both Enterobacteriaceae and non-fermenting Gram-negative bacilli, were much more frequently involved in EPI than in the other types of infection. Escherichia coli, however, was isolated almost as frequently in AHI as in EPI. Cutibacterium (formerly Propionibacterium) spp. were more common in chronic than in acute PJI, but with a significantly higher proportion in the PIC group (15.1%) than in the LCI (6.1%). Polymicrobial infections were much more frequent in the EPI category (27.4%) than in other types, and MDROs (both methicillin-resistant S. aureus (MRSA) and multidrug-resistant GNB, including extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae) were also much more commonly isolated in EPIs than in other categories of infection. Ciprofloxacin-resistant GNB were also more common in EPIs. J. Clin. Med. 2019,8, 673 7 of 15 Table 2. Aerobic Gram-positive cocci involved in prosthetic joint infections using the four categories of the Tsukayama classification. Microorganism or Microorganism Group Early Postoperative Infections n=839 Acute Hematogenous Infections n=265 Late Chronic Infections n=1052 Positive Intraoperative Cultures n=126 p-Value Total no. (%)* Staphylococcus species 505 (60.2) 122 (46) 776 (73.8) 84 (66.7)<0.001 •Coagulase-negative Staphylococci (CoNS) 236 (28.1) 23 (8.7) 573 (54.5) 72 (57.1)<0.001 #Staphylococcus epidermidis 130 (15.5) 11 (4.2) 355 (33.7) 36 (28.6)<0.001 #Staphylococcus lugdunensis 2 (0.2) 6 (2.3) 31 (2.9) 4 (3.2)<0.001 #Staphylococcus capitis 8 (1) 0 (0) 25 (2) 2 (1.6) 0.014 #Staphylococcus hominis 8 (1) 0 (0) 22 (2.1) 0 (0) 0.014 #Staphylococcus warneri 5 (0.6) 0 (0) 11 (1) 3 (2.4) 0.065 #Staphylococcus auricularis 2 (0.2) 0 (0) 12 (1.1) 1 (0.8) 0.048 #CoNS not identified to species level 89 (10.6) 6 (2.3) 168 (16) 23 (23)<0.001 •Staphylococcus aureus 299 (35.6) 104 (39.2) 224 (21.3) 12 (9.5) <0.001 Streptococcus species 36 (4.3) 80 (30.2) 85 (8.1) 5 (4) <0.001 •Streptococcus agalactiae 8 (1) 28 (10.9) 28 (2.7) 0 (0) <0.001 •Viridans group streptococci not identified to species level 6 (0.7) 12 (4.5) 25 (2.4) 2 (1.6) 0.002 •Streptococcus mitis group 8 (1) 5 (1.9) 16 (1.5) 2 (1.6) 0.632 •Streptococcus anginosus group 3 (0.4) 8 (3) 12 (1.1) 1 (0.8) 0.005 •Streptococcus pyogenes 10 (1.2) 5 (1.9) 1 (0.1) 0 (0) 0.003 •Streptococcus pneumoniae 0 (0) 10 (3.8) 2 (0.2) 0 (0) <0.001 •Streptococcus dysgalactiae 2 (0.2) 7 (2.6) 1 (0.1) 0 (0) <0.001 Enterococcus species 106 (12.6) 6 (2.3) 66 (6.3) 4 (3.2) <0.001 •Enterococcus faecalis 95 (11.3) 5 (1.9) 55 (5.2) 3 (2.4) <0.001 •Enterococcus faecium 7 (0.8) 1 (0.4) 5 (0.5) 0 (0.0) 0.584 * Percentages marked in bolded blue are statistically significant highest percentages in that row. Two percentages marked in bolded blue in the same row refer to the highest percentages (both are significantly higher than the other two percentages), but with no statistically significant differences between them. Table 3. Microorganisms and group of microorganisms (other than aerobic Gram-positive cocci) involved in prosthetic joint infections according to the four categories of the Tsukayama classification. Microorganism or Microorganism Group Early Postoperative Infections n=839 Acute Hematogenous Infections n=265 Late Chronic Infections n=1052 Positive Intraoperative Cultures n=126 p-Value Total no. (%)* Aerobic Gram-negative bacilli 395 (47.1) 60 (22.6) 161 (15.3) 14 (11.1) <0.001 •Enterobacteriaceae 303 (36.1) 49 (18.5) 106 (10.1) 6 (4.8) <0.001 #Escherichia coli 129 (15.4) 33 (12.5) 41 (3.9) 3 (2.4) <0.001 #Proteus spp. 75 (8.9) 4 (1.5) 27 (2.6) 2 (1.6) <0.001 #Enterobacter spp. 73 (8.7) 5 (1.9) 19 (1.8) 0 (0) <0.001 #Klebsiella spp. 48 (5.7) 1 (0.4) 1 (0.4) 9 (0.9) <0.001 #Morganella morganii 26 (3.1) 4 (1.5) 11 (1) 1 (0.8) =0.009 #Serratia marcescens 13 (1.5) 0 (0) 6 (0.6) 0 (0) =0.028 •Non-fermenting Gram-negative bacilli 137 (16.3) 9 (3.4) 62 (5.9) 9 (7.1) <0.001 #Pseudomonas spp. 128 (15.3) 8 (3) 59 (5.6) 6 (4.8) <0.001 #Acinetobacter spp. 10 (1.2) 0 (0) 2 (0.2) 1 (0.8) 0.021 Aerobic Gram-positive bacilli 16 (1.9) 5 (1.9) 29 (2.8) 4 (3.2) 0.555 •Corynebacterium species 16 (1.9) 1 (0.4) 29 (2.8) 4 (3.2) 0.087 #Corynebacterium striatum 9 (1.1) 0 (0) 7 (0.7) 1 (0.8) 0.321 # Corynebacterium spp. without identification to species level 3 (0.4) 1 (0.4) 14 (1.3) 2 (1.6) 0.081 Anaerobic Gram-positive bacilli 19 (2.3) 3 (1.1) 73 (6.9)** 22 (17.3)** <0.001 •Cutibacterium spp. 17 (2) 3 (1.1) 64 (6.1)** 19 (15.1)** <0.001 Anaerobic Gram-positive cocci†8 (1) 2 (0.8) 23 (2.2) 0 (0) 0.042 Anaerobic Gram-negative bacilli 12 (1.4) 1 (0.4) 8 (0.8) 0 (0) 0.182 •Bacteroides group 10 (1.1) 1 (0.3) 5 (0.4) 0 (0) 0.145 Mycobacterium species 2 (0.2) 0 (0) 8 (0.8) 0 (0) 0.183 Fungi 8 (1.5) 4 (1.5) 16 (1.5) 2 (1.6) 0.723 •Candida spp. 8 (1) 3 (1.1) 14 (1.3) 2 (1.6) 0.854 Multidrug-resistant organisms 201 (24) 29 (10.9) 82 (7.8) 6 (4.8) <0.001 •Methicillin-resistant S. aureus 92 (11) 22 (8.3) 58 (5.5) 5 (4) <0.001 •Multidrug-resistant Gram-negative bacilli 112 (13.3) 7 (2.6) 25 (2.4) 1 (0.8) <0.001 • Extended-spectrum beta-lactamase producing Enterobacteriaceae 36 (4.3) 2 (0.8) 4 (0.4) 0 (0.0) <0.001 Ciprofloxacin-resistant Gram-negative bacilli 63 (7.5) 7 (2.6) 20 (1.9) 3 (2.4) <0.001 Polymicrobial infections 230 (27.4)** 17 (6.3) 143 (13.1)** 7 (5.6) <0.001 * Percentages marked in bolded blue are the highest statistically significant percentages in that row. Two percentages marked in bolded blue in the same row with no other marks, refer to the highest percentages (both are significantly higher than the other two percentages), but with no statistically significant differences between them. ** The two highest percentages in that row but with statistically significant differences between them. † Finegoldia magna 5, Parvimonas micra 5, Peptostreptococcus anaerobius 3, Peptococcus niger 4, Peptostreptococcus not identified to species level 15. J. Clin. Med. 2019,8, 673 8 of 15 With respect to non-haematogenous infections, a microbiological diagnosis was obtained for 94.5%, 92.7%, 92%, and 88.2% of cases in the first month, months 2/3, 4–12 and more than 12 months after index surgery, respectively, with a statistically significant decreasing linear trend (p<0.001). With respect to PJIs diagnosed in the first month after surgery (EPI according to the Tsukayama classification) versus those diagnosed in the second or third month after surgery (also early infections according to other common classifications) [ 5 , 36 ], the former were significantly more often caused by S. aureus, enterococci, aerobic GNB (with more than twice the percentage of both Enterobacteriaceae and non-fermenting GNB than in months 2 and 3), MDROs (both MRSA and multidrug-resistant GNB, including ESBL-producing Enterobacteriaceae) and polymicrobial infections, and less often caused by CoNS, streptococci, and Cutibacterium spp. (Table 4). Furthermore, infections diagnosed more than three months after surgery were more frequently caused by CoNS, and less commonly by Enterobacteriaceae than those diagnosed in the first two to three months after surgery. No other differences between these four time intervals were observed. When the four periods of time after the index surgery were considered overall, a statistically significant decreasing linear trend was observed for infections caused by S. aureus, enterococci, Enterobacteriaceae, non-fermenting GNB (mainly Pseudomonas spp.), MDROs (both MRSA and multidrug-resistant GNB), and polymicrobial infections (p<0.001 in each case) and a rising trend for infections caused by CoNS (p<0.001), streptococci (p=0.015) and anaerobic bacteria (mainly Cutibacterium spp.) (p<0.001). Table 4. Microorganisms and groups of organisms involved in non-hematogenous prosthetic joint infections according to time of infection after surgery ( ≤ 1 month, 2–3 months, 4–12 months, >12 months). Microorganism or Microorganism Group PJI within 1 Month after Surgery n=844 PJI 2-3 Months after Surgery n=243 PJI 4-12 Months after Surgery n=277 PJI >12 Months after Surgery n=619 p-Value Total no. (%)* Staphylococcus species •Coagulase-negative staphylococci 236 (28.2) ** 107 (44) ** †167 (60.3)†348 (56.2)<0.001 #Staphylococcus epidermidis 132 (15.6) ** 68 (28) ** †106 (38.3)†203 (32.8)<0.001 #Staphylococcus lugdunensis 2 (0.2) ** 3 (1.3) ** 10 (3.6) 22 (3.6)<0.001 •Staphylococcus aureus 301 (35.7) 60 (24.7) 60 (21.7) 108 (17.4) <0.001 Streptococcus species 36 (4.3) ** 25 (10.3) ** 14 (5.1) 49 (7.9) <0.001 #Streptococcus agalactiae 8 (0.9) ** 11 (4.5) ** 6 (2.2) 10 (1.6) 0.003 # Viridans group streptococci not identified to species level 6 (0.7) ** 7 (2.9) ** 2 (0.7) †18 (2.9)†0.003 Enterococcus species 106 (12.6) 23 (9.5) 15 (5.4) 32 (5.4) <0.001 Aerobic Gram-negative bacilli 396 (46.9) ** 50 (20.6) ** †37 (13.4) †37 (13.4) <0.001 •Enterobacteriaceae 303 (35.9) ** 37 (15.2) ** †25 (9) †48 (7.8) <0.001 #Escherichia coli 129 (15.3) 12 (4.9) 10 (3.6) 21 (3.4) <0.001 #Proteus spp. 75 (8.9) 7 (2.9) 7 (2.5) 14 (2.3) <0.001 #Enterobacter spp. 73 (8.6) ** 11 (4.5) ** †2 (0.7) †6 (1) <0.001 #Klebsiella spp. 48 (5.7) 2 (0.8) 4 (1.4) 3 (0.5 <0.001 •Non-fermenting Gram-negative bacilli 138 (16.4) ** 16 (6.6) ** 11 (4) 41 (6.6) <0.001 #Pseudomonas spp. 128 (15.2) ** 17 (7) ** 11 (4) 35 (5.7) <0.001 Aerobic Gram-positive bacilli 16 (1.9) 3 (1.2) 7 (2.5) 23 (3.7) 0.083 Anaerobic Gram-positive bacilli 19 (2.3) ** 14 (5.7) ** 16 (5.8) 61 (9.7) <0.001 •Cutibacterium spp. 17 (2) ** 12 (4.9) ** 16 (5.8) 51 (8.2) <0.001 Anaerobic Gram-positive cocci 8 (0.9) 4 (1.6) 5 (1.8) 13 (2.1) 0.330 Anaerobic Gram-negative bacilli 12 (1.4) 2 (0.8) 1 (0.4) 5 (0.8) 0.409 Mycobacterium species 2 (0.2) 1 (0.9) 4 (1.4) 2 (0.3) 0.068 Fungi 8 (0.9) 3 (1.2) 3 (1.1) 12 (1.9) 0.418 Multidrug-resistant organisms 202 (23.9) 20 (8.2) 20 (7.2) 43 (6.9) <0.001 •Methicillin-resistant S. aureus 93 (11) 14 (5.8) 14 (5.1) 30 (4.8) <0.001 •Multidrug-resistant Gram-negative bacilli 112 (13.3) 7 (2.5) 5 (1.8) 14 (2.3) <0.001 • Extended-spectrum beta-lactamases producing Enterobacteriaceae 36 (4.3) 1 (0.4) 2 (0.7) 1 (0.3) <0.001 Ciprofloxacin-resistant Gram-negative bacilli 84 (10) 5 (2.1) 4 (1.4) 11 (1.8) <0.001 Polymicrobial infections 230 (27.2) 36 (14.7) 32 (11.1) 77 (11.1) <0.001 * Percentages marked in bolded blue are the highest statistically significant percentages in that row. Two percentages marked in bolded blue in the same row with no other indication refer to the highest percentages (both are significantly higher than the other two percentages) with no statistically significant differences between them. ** and † indicate adjacent percentages with statistically significant differences between them in the same row. J. Clin. Med. 2019,8, 673 9 of 15 Figure 1shows the main microorganisms/groups of organisms involved in each of the Tsukayama categories of PJI (A) and in non-haematogenous PJIs according to time since index surgery (B). J. Clin. Med. 2019, 8, x FOR PEER 12 of 18  Figure 1. Main microorganisms or group of microorganisms involved in prosthetic joint infections according to Tsukayama’s classification (A), and in non-hematogenous prosthetic joint infection according to time since index surgery (B). 4. Discussion Empirical antimicrobial treatment is based on a diagnosis of infection without knowing the causative microorganism, while covering those most likely to be involved in particular clinical situations [42]. This multicenter study, which is, to our knowledge the largest series to have analyzed this question, found significant differences in the microbial etiology of different types of Α Β Figure 1. Main microorganisms or group of microorganisms involved in prosthetic joint infections according to Tsukayama’s classification ( A ), and in non-hematogenous prosthetic joint infection according to time since index surgery (B). 4. Discussion Empirical antimicrobial treatment is based on a diagnosis of infection without knowing the causative microorganism, while covering those most likely to be involved in particular clinical situations [ 42 ]. This multicenter study, which is, to our knowledge the largest series to have analyzed this question, found significant differences in the microbial etiology of different types of prosthetic joint infection. Our results show that the choice of empirical antibiotics for early postoperative infections (EPI) (<1 month after surgery) is the most challenging, since these infections are characterized by a preponderance of diverse and virulent pathogens, mainly S. aureus, Enterobacteriaceae, Pseudomonas aeruginosa and enterococci, MDROs, and polymicrobial infections. These microorganisms and groups of organisms were isolated progressively less frequently in non-hematogenous PJIs after the first month following index surgery, whereas a steady linear increase over time was observed for less