Mortality predictors and definition proposal for complicated coagulase-negative Staphylococcus bacteraemia: a multicentre prospective cohort study

For Peer Review Mortality predictors and definition proposal for complicated coagulase-negative Staphylococcus bacteremia. A multicenter prospective cohort study Journal: Clinical Microbiology and Infection Manuscript ID CLM-24-28675.R1 Article Type: Original Article Date Submitted by the Author: n/a Complete List of Authors: Varisco, Benedetta; ASST Santi Paolo e Carlo, Clinic of Infectious and Tropical Diseases, Department of Health Sciences Martínez Pérez-Crespo, Pedro María; Hospital Universitario de Valme, Unidad de Enfermedades Infecciosas y Microbiología Retamar, Pilar; Hospital Universitario Virgen Macarena, Unidad Clínica de Enfermedades Infecciosas y Microbiología López-Hernández, Inmaculada; Hospital Universitario Virgen Macarena, Unidad Clínica de Enfermedades Infecciosas y Microbiología Fariñas, Maria Carmen; University Hospital Marqués de Valdecilla, Infectious Diseases Fernández-Natal, Isabel; Complejo Asistencial Universitario de León, Clinical Microbiology María Teresa, Pérez-Rodríguez; Complexo Hospitalario Universitario de Vigo, Servicio de Medicina Interna Goikoetxea, Josune; Hospital Universitario de Cruces, Enfermedades Infecciosas Sánchez Calvo, Juan Manuel; Hospital Universitario de Jerez de la Frontera, UGC Enfermedades Infecciosas y Microbiología Clínica Buzon-Martin, Luis; Hospital Universitario de Burgos, Medicina Interna León , Eva; Hospital Universitario Nuestra Señora de Valme, Unidad de Enfermedades Infecciosas y Microbiología Vinuesa García, David; Hospital Universitario Clínico San Cecilio., Unidad de Gestión Clínica de Enfermedades Infecciosas. Reguera, José; Regional University Hospital of Malaga, Department of Infectious Diseases Bahamonde, Alberto; Hospital El Bierzo, Department of Internal Medicine. Fernandez, Jonathan; Hospital Universitario Central de Asturias, Microbiology Unit Rodríguez-Baño, Jesús; Hospital Universitario Virgen Macarena, Enfermedades Infecciosas, Microbiologa y Medicina Preventiva; Universidad de Sevilla, Medicina López-Cortés, Luis Eduardo ; Hospital Universitario Virgen Macarena, Unidad Clínica de Enfermedades Infecciosas y Microbiología; Universidad de Sevilla, Facultad de Medicina Key Words: Coagulase-negative staphylococci, Bloodstream infections, Complicated bacteremia, Mortality predictors, Catheter-related bloodstream infections Clinical Microbiology and Infection For Peer Review Abstract: Objective. To explore a definition for complicated coagulase-negative staphylococci bloodstream infections (CoNS BSI), and to identify predictors for mortality. Methods. Prospective cohort study conducted from October 2016 to March 2017 in 26 Spanish hospitals. Complicated CoNS BSI criteria included lack of early catheter removal in catheter-related cases, foreign indwelling implant, persistent bacteremia, fever ≥72 hours on active therapy, metastatic infection or deep-seated focus and infective endocarditis. Independent predictors for 30-day mortality were evaluated by Cox regression, and the impact of the definition of complicated bacteremia assessed. Results. Overall, 445 CoNS BSI cases were included; catheter-related infections were predominant (336/445, 75.5%). Complicated bacteremia was identified in 240/445 patients (53.9%); 30-day mortality in complicated and uncomplicated cases were 53/240 (22.1%) and 24/205 (11.7%), respectively (p=0.004). Predictors of 30-day mortality identified in the multivariate analysis included age (HR 1.03, 95%CI 1.01-1.05), cerebrovascular disease (HR 2.58, 95%CI 1.45-4.58), immunosuppressive therapy (HR 2.16, 95%CI 1.22-3.84), SOFA score (HR 1.09, 95%CI 1.03-1.16), and complicated bacteremia (HR 2.14, 95%CI 1.29-3.53). A catheter-related source of bacteremia was found to be protective (HR 0.49, 95%CI 0.30-0.80). When specific criteria to define complicated bacteremia were included, fever ≥72h was associated with increased risk of death (HR 2.52, 95%CI 1.52-4.17) and early catheter removal was protective (HR 0.47, 95%CI 0.26-0.83). Conclusions. A high proportion of patients presented complicated bacteremia according to the proposed criteria; these patients had higher hazards for mortality. Other mortality predictors were identified. Further studies would be needed to validate the proposed criteria. Page 1 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Seville, December 2024 Dear Prof. Paul, Editor, Clinical Microbiology and Infection Thank you for the constructive feedback and detailed suggestions regarding our manuscript, "Mortality predictors and definition proposal for complicated coagulase-negative Staphylococcus bacteremia. A multicenter prospective cohort study" (manuscript ID: CLM-24-28675). We appreciate the time and effort that both you and the reviewers have dedicated to this process. Based on your suggestions, we have made careful revisions to improve the clarity and strength of our work. The tracked and clean copies of the revised manuscript are submitted here for your consideration. Kind regards, Luis Eduardo López-Cortés On behalf of all authors Page 2 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review EDITOR COMMENTS TO THE AUTHOR Comment (1): Among inclusion criteria “there were systemic symptoms like fever or chills” – was this the actual criterion used for inclusion? As defined, this criterion is non-specific, cannot be replicated and probably leads to heterogeneity in patient inclusion. If a more precise criterion was used please define it in the manuscript. If this was the criterion, we have little confidence that the patients in the cohort has a clinically-significant CONS bacteremia. Response: Thank you for the important feedback. To be brief, we acknowledge that the criteria were not described with sufficient specificity. We have now clarified the specific inclusion criteria in the manuscript and Supplementary material, aligning them with those outlined in the study protocol. Comment (2): “Data were collected via an electronic case report” – please define data sources and collection methods. Were data manually collected from electronic patients files or paper charts? Was any automated data query tool used? Response: We have revised the manuscript to clarify the data collection methods, which now reads: "Data were collected prospectively from either electronic health records or paper charts at each center according to local availability. Data for each enrolled patient were entered into a standardized online case report form.” Comment (3): Under “Definitions and variables” please define “empiric therapy” – first 24 hours? 48 hours” what is the starting time point – blood culture collection time? In methods you defined Appropriate antimicrobial therapy and in the results you refer to “active” therapy – please use consistent terminology. Response: Thank you for your feedback. We have added the definition of "empiric therapy," which is now stated as "the antimicrobial regimen initiated prior to the identification of the infecting microorganism and its susceptibility profile; since Gram/MALDI reports were typically available within 24 hours, therapy was usually considered empirical during this time frame". Regarding the terminology inconsistency, we appreciate your observation. We find the term "active therapy" more accurate than "appropriate therapy." As such, we have revised the manuscript to ensure the consistent use of "active therapy" throughout the sections. Comment (4): Follow-up blood cultures were taken only to <30% of patients by physicians’ discretion. CONS are low virulence pathogens who do not necessary cause marked sepsis signs, Page 3 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review it is likely that some of the patients in whom follow-up blood cultures were not taken (presumably because they had no fever or other sepsis signs) also had prolonged bacteremia. I propose listing the indications for taking follow-up blood cultures (prospective criteria or retroactive reasons from the data) and re-naming the variable “Positive follow-up blood cultures” to “Positive follow-up blood cultures when indicated” or similar. Response: Thank you for highlighting this important point. Given the lack of guidelines for follow-up blood cultures (FUBC) in CoNS bacteremia, our observational study reflects clinical practice, with FUBC performed at the discretion of the treating physician. We recognize that positive FUBC results may represent only a subset of actual persistent bacteremia cases, as some patients not tested may have had undetected positive results. To address this, we have included this point in the Limitations section of the study. In response to your suggestion, we conducted a retrospective analysis of factors that may have influenced the decision to perform FUBC. As shown in Supplementary Table S1, patients with endocarditis or cardiac/endovascular devices were significantly more likely to undergo FUBC. We have summarized these findings in the Results section. Although we hypothesize that these listed factors likely influenced FUBC decisions, our dataset does not include information on the specific clinical indications for each FUBC, as noted in the Limitations. Regarding the variable name, please note that in Table 1 we include the variable “Blood cultures” (now called “Follow-up blood cultures performed”), and we state at Table foot that FUBC were performed at the discretion of attending physician, and refer to our analysis. The other variable is now called "Positive follow-up blood cultures among those performed." Comment (5): The reviewers were critical of the catheter extraction measure among the measures defining complicated bacteremia, since there is a consensus that catheter retaining is possible in the initial management of CONS CRABSI bacteremia. I agree that since there is no good evidence for this recommendation this remains a question. However, this variable is relevant only for the subgroup of patients with a CVC. The variable “early (<48h) catheter removal” in model 2 mixes two variables: Yes – removed in CRBSI, No – not removed in CRABSI or no CRABSI. This is not a useful predictor clinically, because for the individual patient the source of the bacteremia is known and if unknown, the variable is not relevant. To appropriately analyse the association between catheter extraction and mortality the analysis needs to be limited to patients with a CVC. I recommend revising model 2 to include a 3category variable of CRBSI extracted, CRBSI retained and no CVC, to allow a more appropriate evaluation of these factors in a model including all individual components of complicated bacteremia. Alternatively, or concomitantly, it would be of interest to evaluate catheter extraction in the subgroup of patients with a CVC, among whom risk factors for mortality need Page 4 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review to be re-evaluated to select the appropriate confounders for the regression analysis including catheter extraction. Response: We fully agree; in fact, Model 2 was performed using patients with CRBSI without early catheter removal as the reference category for early catheter removal (estimation provided in the model) and for non-CRBSI (we did not include the estimate for non-CRBSI due to the lack of association, but we agree it should be included). We now include this as Model 2a, and explained the stratification in the title of Table 4. We also addressed your second suggestion about adding an analysis of patients with a CVC and therefore we added a Model 2b including only patients with CRBSI. Comment (6): Please define the variables entered into the regression analyses, those removed in the stepwise selection process and all those retained with their significance. Describe the variables that were not entered due to correlations. Please document the number of patients included in each regression model and the statistics for the model – fitness and calibration. See our guidance for reporting of multivariable logistic regression models. https://www.clinicalmicrobiologyandinfection.com/article/S1198-743X(19)30592-0/fulltext. As a reviewer noted the model including all components of complicated bacteremia might have higher dependent to independent ratio than recommended resulting is overfitting. Response: Thank you for your valuable feedback. We added in the text that a detailed explanation of models’ development is explained in the Supplementary material (please see section "Multivariable regression model methodology"). Comment (7): Please include dispersion measures (SD, min-max, IQR, CI) for all continuous and outcomes and ratio maeasures reported in abstract, text and table and crude numbers (numerator/ denominator) for all percentages. Response: We have included the dispersion measures (IQR and CI) wherever they were missing in the abstract, text, and tables. Additionally, we specified all absolute numbers (numerator/denominator) for each percentage. Comment (8): Appropriate empiric therapy was not associated with mortality in your cohort. Your following statements in the discussion are unsubstantiated: “However, empiric therapy should be initiated if risk factors for adverse outcomes are identified” and lines 244-247 “Additionally, nearly half of the empiric therapies were ineffective against the isolated strain, underscoring the importance of starting treatment that covers methicillin-resistant strains when CoNS infections are suspected. This approach is supported by our finding that over 70% of isolates in our cohort were methicillin-resistant, consistent with other reports.” Please revise – I Page 5 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review suggest deleting these parts of the discussion. In addition, you note for the first time in the discussion that only 90% pf patients received targeted therapy once culture results were available – please include this variable in the tables. Response: We agree with the Editor. We have made the requested changes by deleting the specified parts of the discussion. Additionally, we have included the information on targeted therapy in the tables that was missing. Comment (9): Please check and reference lines 234-235 in the discussion on adherence to guidelines relating to catheter extraction. Most people follow the IDSA guidelines that as noted above allow retaining the catheter in CONS CRBSI. Response: Following the Editor’s comment, we have completely revised that part of the Discussion, recognizing that IDSA guidelines consider catheter retention in CoNS CRBSI. Page 6 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review REVIEWER #1 COMMENTS Comment (1): In this study, Spanish investigators were looking at patients with coagulasenegative staphylococcus (CoNS) bacteremia in a prospective study of bacteremia that enrolled at multiple hospitals in 2017. They sought to look at predictors of mortality and proposed an operational definition for complicated bacteremia. Table 1 shows the characteristics of the 445 patients included in the study. The authors defined complicated infections according to clinical criteria previously applied to S. aureus bacteremia plus they included lack of catheter removal within 48 hours if CRBSI was present. 30 day mortality was higher in the complicated bacteremia group. Table 4 shows multi-variate analysis where early catheter removal was associated with lower mortality. Overall: The authors are addressing an important topic, namely bloodstream infections due to coagulase-negative staphylococci. The main strengths of their study are the large sample size and multi-center design. The main weakness is that it is observational with all the drawbacks such a study design entails. Thus, although associations can be made, the clinical impact of this study is limited. Moreover, the authors have combined infection/patient characteristics, such as persistent fever/presence of in-dwelling device with practitioner decisions (i.e. early catheter removal) which ultimately obscures their definition of “complicated bacteremia”. Response: Thank you for your comment. We acknowledge the significant limitations inherent in observational studies. However, we believe that our study design, featuring prospective data collection, provides a valuable reflection of real-world practices in managing coagulase-negative staphylococcal bloodstream infections. Regarding the definition of "complicated bacteremia," we understand your concern about the integration of patient/infection characteristics and the physician's decision regarding early catheter removal. We would like to emphasize that our definition is not intended to be pathophysiologic but rather practical for clinical use. If lack of catheter removal impacts outcomes, we believe that considering it is useful from a clinical perspective. Comment (2): The authors could have actually derived risk factors for poor outcome rather than deciding on them beforehand and then seeing if they were associated with mortality. This would have strengthened their definition of “complicated infection”. Response: We appreciate your insightful comment, which allows us to clarify our study's purpose. Actually, our objective was hypothesis-based, specifically aiming to assess whether a pre-established definition for “complicated bacteremia” was associated with an increased risk of 30-day mortality instead of just exploring which variables are associated with outcome. While identifying broader risk factors for poor outcomes would indeed be a relevant approach, it aligns with a different objective than ours. By applying predefined criteria, we sought to evaluate their associations with mortality, both collectively and independently, within our operational definition. Page 7 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review To reflect this more clearly, we have revised the study objectives in the Introduction as follows: “This study aims to describe a large cohort of CoNS BSI cases, test whether predefined criteria for complicated bacteremia are associated with 30-day mortality and identify other outcome predictors.” Comment (3): There are statistical errors, such as in Table 1 when the authors compare uncomplicated and complicated infection in terms of whether patients have early catheter removal. Given that early catheter removal is part of the definition, it is not surprising that there is a statistically significant difference between the two groups. Given that this manuscript has complex statistics at its heart, this kind of comparison raises concern for the validity of the other findings. Response: The Reviewer is right; we deleted the statistical comparisons in Table 1 dealing with factors which are criteria for complicated bacteremia, as patients with uncomplicated bacteremia cannot meet these criteria. We also took this opportunity to recheck all the data before resubmission. We also clarify that in the Table foot. Comment (4): On line 233 the authors state that lack of early removal of a catheter in CoNS CRBSI indicates poor adherence to CRBSI management standards without giving a reference – however, the 2009 IDSA guidelines (which are most current) state that is appropriate to leave a catheter in place initially for CoNS CRBSI and see how patients do. The vast majority of patients are cured with this approach (see PMID 19780661 which is not referenced here) which in turn saves patients from undergoing unnecessary catheter change. Thus, I would suggest the authors temper this criticism unless they have other data to support it. Response: Thank you for your insightful comment; we fully agree with your observations. This concern was also raised by the Editor. We appreciate your suggestion regarding the reference, which has helped us clarify our perspective. Following your and the Editor’s comment, we have revised in deep the Discussion section dealing with this aspect. Comment (5): The authors should be careful about including catheter retention in any model without specifically understanding why catheters were retained. For example, if a patient is dying of another process, catheter retention may simply reflect the unwillingness of the treating physicians to put the patient through more procedures. Response: Thank you for your comment. We agree that understanding the specific reasons for catheter retention is important; however, our dataset does not include this information. Despite this limitation, the association of early catheter removal with mortality in our cohort remains Page 8 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Comment (6): Major points. Methods. Lines 142-144: How many variables went into the multivariate Cox regression analysis? Was the number in proportion to 77 events/deaths at day30? Please clarify and comment. Response: Thank you for raising this point, which was also highlighted in Comment (6) from the Editor. We hope that our modifications have clarified this aspect. Comment (7): Major point. Line 144: Interactions were evaluated – could the authors please specify this important step in the respective steps. Response: Interactions between complicated bacteremia and the other variables in the final model were assessed, as well as interactions between the variables and each individual criterion defining complicated bacteremia. No clinically or statistically significant interactions were observed. The explanation can now be read in the Section “Methods [3]. Multivariable regression model methodology” in Supplementary material. Comment (8): Major point. Results. If ‘metastatic complication’ occurred in <3% of cases, could this criterion be omitted? Response: Thank you for the comment. Since the definition for complicated bacteremia was pre-defined, we decided to retain all the criteria considered. To note, 5 of the 7 patients with metastatic complications met other criteria for "complicated bacteremia": 1 was a vascular prosthetic device carrier, 1 had an articular device, 2 had persistent fever, and 1 had a CRBSI without early catheter removal. Only two patients did not meet any other "complicated criteria," and the definition was solely attributed to the presence of metastatic foci. We have added this explanation in the Supplementary Material under the section "Methods [2]: Justification for the inclusion of “metastatic complications” as a criterion for “complicated bacteremia”." Comment (9): Minor point. General aspects. The manuscript could benefit from some language editing. Response: Thank you for your feedback. We have revised the manuscript for clarity and language quality to enhance its overall readability. Comment (10): Minor point. Abstract and throughout the manuscript. Suggest to use ‘infective’ endocarditis instead of ‘infectious’. Page 15 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Response: Thank you for the suggestion. We agree and have made the modification to use "infective" endocarditis instead of "infectious" throughout the manuscript and additional material. Comment (11): Minor point. Line 103: which (intravascular) catheters were included – please specify. Response: Cultures drawn from any type of intravascular catheter could be sent to microbiology for identification of CRBSI, if submitted alongside peripheral blood cultures. This information was added to the Supplementary Material under the section “Study microbiological and clinical inclusion criteria for CoNS bloodstream infection episodes,” specifically within the microbiological criteria. Comment (12): Minor point. Methods and Results. Please specify which implanted devices were included (so far outlined in table, but not in text). Response: Since this information is in the Table, we think is it sufficient, also given the words limit; however, we are open to following any specific guidance from the Editor on this matter. Comment (13): Minor point. Discussion. Line 214: please elaborate or give references to the ‘previous studies’ mentioned. Response: The reference was indeed provided at the end of the sentence, but we will move it within the sentence for improved clarity. We would finally like to declare the following: •In Table 2, the criterion of “primary endocarditis” was omitted in the previous version. While it was considered in the count, it was inadvertently excluded from the table. This has been corrected, and the criterion is now included in the revised table. •In response to the authors' and reviewers' comments, the current version of the manuscript is 2,593 words, which slightly exceeds the word limit. Page 16 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 1 1Mortality predictors and definition proposal for complicated coagulase-negative Staphylococcus 2bacteremia. A multicenter prospective cohort study 3 4Authors 5Benedetta Varisco1, Pedro María Martínez Pérez-Crespo2, Pilar Retamar-Gentil3,4, Inmaculada López 6Hernandez3,4, Mª Carmen Fariñas-Álvarez4,5, Isabel Fernández-Natal6, María Teresa Pérez7Rodríguez7, Ane Josune Goikoetxea Aguirre8, Juan Manuel Sánchez-Calvo9, Luis Buzón Martín10, 8Eva León-Jiménez2, David Vinuesa García11, José María Reguera-Iglesias12, Alberto Bahamonde9Carrasco13, Jonathan Fernández Suárez14, Jesús Rodríguez-Baño3,4*, Luis Eduardo López-Cortés3,4*, 10 on behalf of the PROBAC REIPI/GEIH-SEIMC/SAEI Group 11 12 Affiliations 13 1Department of Health Sciences, Clinic of Infectious Diseases, University of Milan, ASST Santi Paolo 14 e Carlo, Via A. Di Rudinì, 8, 20142, Milan, Italy. 15 2Unidad de Enfermedades Infecciosas y Microbiología, Hospital Universitario de Valme, 41014, 16 Sevilla, Spain. 17 3Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen 18 Macarena; Departamentos de Medicina y Microbiología, Facultad de Medicina, Universidad de 19 Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, 41009, Seville, Spain. 20 4CIBERINFEC, Instituto de Salud Carlos III, 28029, Madrid, Spain. 21 5Unidad de Enfermedades Infecciosas, Hospital Universitario Marqués de Valdecilla, Universidad de 22 Cantabria, IDIVAL, 39008, Santander, Spain. Page 17 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 2 23 6Department of Clinical Microbiology, Complejo Asistencial Universitario de León (CAULE), 24071, 24 León, Spain. 25 7 Infectious Diseases Unit, Department of Internal Medicine, Complexo Hospitalario Universitario de 26 Vigo, 36312, Vigo, Spain; Instituto de Investigación Biomédica Galicia Sur, Spain. 27 8 Unidad de Enfermedades Infecciosas, Hospital Universitario de Cruces, 48903, Bizkaia, Spain. 28 9 Unit of Infectious Diseases and Clinical Microbiology. Jerez De La Frontera University Hospital, 29 11407, Jerez De La Frontera, Cádiz, Spain. 30 10 Unidad de Gestión Clínica de Enfermedades Infecciosas, Hospital Universitario de Burgos, 09006, 31 Burgos, Spain. 32 11 Unidad Gestión Clínica Enfermedades Infecciosas, Hospital Universitario Clínico San Cecilio, 33 18016, Granada, Spain. 34 12Servicio de Enfermedades Infecciosas, Hospital Regional Universitario de Málaga, IBIMA Málaga, 35 29010, Málaga, Spain. 36 13Departamento de Medicina Interna, Hospital de El Bierzo, 24404, Ponferrada, Spain. 37 14Unidad de Microbiología, Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), 38 Hospital Universitario Central de Asturias, 33011, Oviedo, Spain. 39 *These authors contributed equally as senior authors. 40 41 Corresponding author 42 Luis Eduardo López-Cortés 43 Infectious Diseases and Microbiology Unit 44 Hospital Universitario Virgen Macarena 45 Department of Medicine, University of Sevilla Page 18 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 3 46 Biomedicines Institute of Sevilla 47 41009 Sevilla, Spain 48 Phone: +34 670 946 430 49 Email: [email protected] Page 19 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 4 50 Keywords. Coagulase-negative staphylococci; bloodstream infections; complicated bacteremia 51 Article type. Original article 52 Subject section. Bacterial Infections; Antimicrobial Stewardship 53 Abstract word count. 245 54 Text word count. 2593 55 Page 20 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 5 56 ABSTRACT 57 Objective. To explore a definition for complicated coagulase-negative staphylococci bloodstream 58 infections (CoNS BSI), and to identify predictors for mortality. 59 60 Methods. Prospective cohort study conducted from October 2016 to March 2017 in 26 Spanish 61 hospitals. Complicated CoNS BSI criteria included lack of early catheter removal in catheter-related 62 cases, foreign indwelling implant, persistent bacteremia, fever ≥72 hours on active therapy, metastatic 63 infection or deep-seated focus and infective endocarditis. Independent predictors for 30-day mortality 64 were evaluated by Cox regression, and the impact of the definition of complicated bacteremia assessed. 65 66 Results. Overall, 445 CoNS BSI cases were included; catheter-related infections were predominant 67 (336/445, 75.5%). Complicated bacteremia was identified in 240/445 patients (53.9%); 30-day 68 mortality in complicated and uncomplicated cases were 53/240 (22.1%) and 24/205 (11.7%), 69 respectively (p=0.004). Predictors of 30-day mortality identified in the multivariate analysis included 70 age (HR 1.03, 95%CI 1.01-1.05), cerebrovascular disease (HR 2.58, 95%CI 1.45-4.58), 71 immunosuppressive therapy (HR 2.16, 95%CI 1.22-3.84), SOFA score (HR 1.09, 95%CI 1.03-1.16), 72 and complicated bacteremia (HR 2.14, 95%CI 1.29-3.53). A catheter-related source of bacteremia was 73 found to be protective (HR 0.49, 95%CI 0.30-0.80). When specific criteria to define complicated 74 bacteremia were included, fever ≥72h was associated with increased risk of death (HR 2.52, 95%CI 75 1.52-4.17) and early catheter removal was protective (HR 0.47, 95%CI 0.26-0.83). 76 Page 21 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 6 77 Conclusions. A high proportion of patients presented complicated bacteremia according to the 78 proposed criteria; these patients had higher hazards for mortality. Other mortality predictors were 79 identified. Further studies would be needed to validate the proposed criteria. Page 22 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 7 80 INTRODUCTION 81 82 Coagulase-negative staphylococci (CoNS) have traditionally been regarded as low-virulence 83 organisms due to their role as skin commensals and frequent contaminants in microbiological 84 samples. They are now recognized as a leading cause of nosocomial bloodstream infections (BSI), 85 particularly catheter-related (CRBSI), due to their ability to form biofilms.1 Despite their high 86 prevalence, research into the morbidity and mortality of CoNS bacteremia remains limited, with 87 reported mortality rates ranging from 4% to 24%.2–7 Furthermore, treatment recommendations rely 88 on limited evidence, reflecting the challenge of distinguishing true infections from contamination.8 89 Nevertheless, a combination of accurate clinical evaluation and interpretation of microbiological 90 findings (particularly the number of positive blood cultures)9 can help reduce misdiagnoses. 91 In terms of treatment, the Infectious Diseases Society of America (IDSA) guidelines10 92 recommend treating uncomplicated CoNS CRBSI for 5-7 days if the catheter is removed, and for 1093 14 days with antibiotic lock therapy if the catheter is retained (level BIII evidence). However, to the 94 best of our knowledge, the impact of delayed catheter removal in high-risk patients has not been 95 systematically assessed. Additionally, criteria for complicated bacteremia used in the guidelines 96 include suppurative thrombophlebitis, endocarditis, osteomyelitis, and metastatic seeding, though 97 these criteria are not specific to CoNS infections.10 Whether additional criteria, such as prolonged 98 fever, persistent bacteremia, and lack of source control (considered for Staphylococcus aureus )11,12 99 are associated with worse outcomes in CoNS BSI remains unstudied. 100 This study aims to describe a large cohort of CoNS BSI cases, evaluate the association 101 between predefined criteria for complicated bacteremia and 30-day mortality, and identify other 102 outcome predictors. Page 23 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 8 103 104 METHODS 105 106 Design and setting 107 This study is part of the PROBAC project, a prospective cohort study involving patients >14 108 years admitted to 26 hospitals in Spain from October 1, 2016, to March 31, 2017, with clinically 109 significant BSI.13,14 The cohort included 18 tertiary care and 8 community hospitals. 110 111 Inclusion and exclusion criteria 112 To ensure the clinical significance of CoNS bloodstream infections (BSI), episodes were 113 included in the study if they met both microbiological and clinical prespecified criteria (Methods [1], 114 Supplementary Material). In summary, microbiological criteria required the isolation of the same 115 CoNS species with an identical susceptibility profile from at least two distinct blood culture sets, with 116 additional criteria for catheter-related infections.10 Clinical criteria included the presence of at least 117 two systemic infection signs (fever or hypothermia, tachycardia, tachypnoea, leukocytosis or 118 leukopenia). Additionally, other causes of infection were reasonably excluded; for patients in whom 119 the source of bacteremia was exceptional for CoNS (i.e., pneumonia, intra-abdominal infection, 120 meningitis, or unknown source in patients without catheters or prosthetic material), the opinion of a 121 second local investigator was sought. 122 The following exclusion criteria were applied: polymicrobial BSI; infections by 123 Staphylococcus lugdunensis ; 15 recurrent episodes of bacteremia, defined by the isolation of the same 124 species with clinical signs and symptoms consistent with infection, within three months after resolution 125 of the initial episode. Page 24 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 15 261 disease, and immunosuppressive therapy were linked to higher mortality, likely reflecting patient 262 frailty, while the SOFA score indicated acute severity. Additionally, catheter-related infections were 263 associated with lower mortality compared to infections from other sources. 264 In the absence of specific criteria for CoNS BSI, our study used modified criteria originally 265 designed for S. aureus complicated bacteremia. The finding that complicated bacteremia is 266 independently associated with higher mortality suggests that these criteria might be useful for CoNS 267 BSI as well. However, the separate analysis in the second model, where not all criteria were linked to 268 mortality, indicates the need for a tailored approach to CoNS BSI in future research. 269 FUBC were performed in fewer than 30% of cases in our cohort. Although specific guidelines 270 are lacking, obtaining them in high-risk patients with complicated bacteremia may be beneficial, 271 especially given the observed association between mortality and persistent fever, which could indicate 272 ongoing bacteremia. In contrast, FUBC may be unnecessary for low-risk cases, as suggested by Badia273 Cebada et al.23 Their trial comparing antibiotics versus no antibiotics in low-risk CoNS CRBSI found 274 only one case of persistent bacteremia in the no-antibiotics group. However, the small sample size due 275 to study interruption limits the strength of these findings. 276 Although our study found early catheter removal to be protective against mortality, it was 277 carried out in fewer than 60% of CRBSI cases. IDSA guidelines consider the possibility of catheter 278 retention in uncomplicated CoNS CRBSI cases.10 However, despite the fact that some studies suggest 279 that catheter retention in CoNS bloodstream infections may not significantly impact mortality, it 280 remains an independent risk factor for recurrence,24 while recurrence is rare after catheter removal.25 281 Our finding that early catheter removal is independently associated with a protective effect on 30-day 282 mortality suggests the need for further studies to better identify the subgroups of patients in whom 283 retaining the catheter is safe. Page 31 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 16 284 In our study, about one-third of patients did not receive empiric therapy at the onset of infection; 285 however, approximately 90% of these patients were initiated on targeted therapy once culture results 286 became available. Omitting empiric therapy may be justified if no risk factors are present and source 287 control (e.g. catheter removal) is achieved, potentially making subsequent targeted therapy 288 unnecessary. 23,26 289 Our study has limitations. First, the absence of a comparative group restricts our ability to 290 establish causal relationships and assess the proportion of mortality directly attributable to CoNS BSI 291 versus underlying conditions. Although a minority of patients did not complete the 30-day follow-up, 292 sensitivity analyses suggest the results are robust. We acknowledge that positive FUBC results likely 293 represent only a subset of persistent bacteremia cases, as some patients without FUBC may have had 294 undetected positive results; although we retrospectively analyzed factors influencing the decision to 295 perform FUBC, the data lacks clear indications for each case. Nonetheless, we believe including this 296 variable is important to highlight the need for further studies to clarify FUBC indications in CoNS 297 bloodstream infections. Management of patients was performed according to attending physicians and 298 therefore was not standardized; nonetheless, we believe the data reflect real-world clinical practice. 299 Despite efforts like investigator training and data monitoring, some data collection errors may still 300 exist. However, the study's strengths include a large sample size, and a multicenter, prospective design 301 conducted by infectious disease experts in hospitals with bacteremia programs, which enhances the 302 credibility of our findings. 303 In summary, our study identifies key mortality factors in CoNS BSI and emphasizes the need 304 for improved risk stratification. The proposed definition for complicated CoNS BSI needs validation 305 in further studies. 306 Page 32 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 17 307 AUTHOR CONTRIBUTIONS 308 309 Conceptualization: LELC, JRB. Methodology and statistical analysis: BV, LELC, JRB. 310 Original draft: BV. Review & editing: BV, LELC, JRB (all authors). Supervision: JRB. Project 311 administration: LELC. Funding acquisition: LELC, JRB. All authors contributed to the final 312 manuscript. 313 314 FUNDING 315 316 This research was supported by grants from Plan Nacional de I+D+i 2013-2016, Instituto de 317 Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio 318 de Ciencia, Innovación y Universidades (PI16/01432); the Spanish Network for Research in Infectious 319 Diseases (REIPI) (RD16/0016/0001; RD16/ 0016/0008); and co-financed by the European Regional 320 Development Fund under the 'A way to achieve Europe' Operational Program for Intelligent Growth 321 2014-2020. 322 323 TRANSPARENCY STATEMENT 324 325 PRG has served as a scientific advisor for Shionogi and Advanz, speaker for Angelini, and 326 Menarini. LELC has served as a scientific advisor for Novartis, speaker for MSD, Pfizer, Angelini, 327 and ViiV, and trainer for MSD and ViiV. The other authors declare no conflicts of interest. Page 33 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 18 328 REFERENCES 329 1. Heilmann C, Ziebuhr W, Becker K. Are coagulase-negative staphylococci virulent? Clinical 330 Microbiology and Infection. 2019;25(9):1071-1080. doi:10.1016/j.cmi.2018.11.012 331 332 2. Molina J, Peñuela I, Lepe JA, et al. Mortality and hospital stay related to coagulase-negative 333 Staphylococci bacteremia in non-critical patients. Journal of Infection. 2013;66(2):155-162. 334 doi:10.1016/j.jinf.2012.10.021 335 336 337 3. Park SY, Kwon KH, Chung JW, Huh HJ, Chae SL. Coagulase-negative staphylococcal 338 bacteremia: risk factors for mortality and impact of initial appropriate antimicrobial therapy 339 on outcome. European Journal of Clinical Microbiology and Infectious Diseases. 340 2015;34(7):1395-1401. doi:10.1007/s10096-015-2364-3 341 342 4. Hentzien M, Strady C, Vernet-Garnier V, et al. Prognostic factors associated with 30-day in343 hospital mortality in coagulase-negative Staphylococcus bacteraemia: no impact of 344 vancomycin minimum inhibitory concentration. Infect Dis. 2017;49(9):664-673. 345 doi:10.1080/23744235.2017.1323346 346 347 5. Cui J, Liang Z, Mo Z, Zhang J. The species distribution, antimicrobial resistance and risk 348 factors for poor outcome of coagulase-negative staphylococci bacteraemia in China. 349 Antimicrob Resist Infect Control. 2019;8(1):1-10. doi:10.1186/s13756-019-0523-5 350 351 6. Yamada K, Namikawa H, Fujimoto H, et al. Clinical characteristics of methicillin-resistant 352 coagulase-negative staphylococcal bacteremia in a tertiary hospital. Internal Medicine. 353 2017;56(7):781-785. doi:10.2169/internalmedicine.56.7715 354 355 7. Rosa RG, Dos Santos RP, Goldani LZ. Mortality related to coagulase-negative staphylococcal 356 bacteremia in febrile neutropenia: A cohort study. Canadian Journal of Infectious Diseases 357 and Medical Microbiology. 2014;25(1):14-18. doi:10.1155/2014/702621 358 359 8. Rahkonen M, Luttinen S, Koskela M, Hautala T. True bacteremias caused by coagulase 360 negative Staphylococcus are difficult to distinguish from blood culture contaminants. 361 European Journal of Clinical Microbiology and Infectious Diseases. 2012;31(10):2639-2644. 362 doi:10.1007/s10096-012-1607-9 363 364 9. Mirrett S, Weinstein MP, Reimer LG, Wilson ML, Reller LB. Relevance of the number of 365 positive bottles in determining clinical significance of coagulase-negative staphylococci in 366 blood cultures. J Clin Microbiol. 2001;39(9):3279-3281. doi:10.1128/JCM.39.9.3279367 3281.2001 368 369 10. Mermel LA, Allon M, Bouza E, et al. Clinical practice guidelines for the diagnosis and 370 management of intravascular catheter-related infection: 2009 update by the Infectious 371 Diseases Society of America. Clinical Infectious Diseases. 2009;49(1):1-45. 372 doi:10.1086/599376 373 374 11. Jung N, Rieg S. Essentials in the management of S. aureus bloodstream infection. Infection. 375 2018;46(4):441-442. doi:10.1007/s15010-018-1130-8 376 Page 34 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 19 377 12. López-Cortés LE, Del Toro MD, Gálvez-Acebal J, et al. Impact of an evidence-based bundle 378 intervention in the quality-of-care management and outcome of Staphylococcus aureus 379 bacteremia. Clinical Infectious Diseases. 2013;57(9):1225-1233. doi:10.1093/cid/cit499 380 381 13. Calo F, Retamar P, Perez-Crespo PMM, et al. Catheter-related bloodstream infections: 382 Predictive factors for Gram-negative bacteria aetiology and 30 day mortality in a multicentre 383 prospective cohort. Journal of Antimicrobial Chemotherapy. 2020;75(10):3056-3061. 384 doi:10.1093/jac/dkaa262 385 386 14. Mussa M, Martínez Pérez-Crespo PM, Lopez-Cortes LE, et al. Risk Factors and Predictive 387 Score for Bacteremic Biliary Tract Infections Due to Enterococcus faecalis and Enterococcus 388 faecium: a Multicenter Cohort Study from the PROBAC Project. Microbiol Spectr. 389 2022;10(4). doi:10.1128/spectrum.00051-22 390 391 15. Frank KL, Del Pozo JL, Patel R. From clinical microbiology to infection pathogenesis: How 392 daring to be different works for Staphylococcus lugdunensis. Clin Microbiol Rev. 393 2008;21(1):111-133. doi:10.1128/CMR.00036-07 394 395 16. Rodríguez Díaz JC, Guna Serrano M del R, Larrosa Escartín N, Marín Arriaza M. Diagnóstico 396 microbiológico de la bacteriemia y la fungemia: hemocultivos y métodos moleculares. 397 Sociedad Española de Enfermedades Infecciosas y Microbiología Clínica. 2017. Accessed 398 November 6, 2024. 399 https://seimc.org/contenidos/documentoscientificos/procedimientosmicrobiologia/seimc400 procedimientomicrobiologia62.pdf 401 402 17. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic 403 comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 404 1987;40(5):373-383. doi:https://doi.org/10.1016/0021-9681(87)90171-8. 405 406 18. Friedman ND, Kaye KS, Stout JE, et al. Health care-associated bloodstream infections in 407 adults: A reason to change the accepted definition of community-acquired infections. Ann 408 Intern Med. 2002;137(10):791-797. doi:10.7326/0003-4819-137-10-200211190-00007 409 410 19. Hilf M, Yu VL, Sharp J, Zuravleff JJ, Korvick JA, Muder RR. Antibiotic therapy for 411 Pseudomonas aeruginosa bacteremia: Outcome correlations in a prospective study of 200 412 patients. Am J Med. 1989;87(5):540-546. doi:10.1016/S0002-9343(89)80611-4 413 414 20. Fernando SM, Rochwerg B, Seely AJE. Clinical implications of the third international 415 consensus definitions for sepsis and septic shock (Sepsis-3). JAMA. 2018;190(36):E1058416 E1059. doi:10.1503/cmaj.170149 417 418 21. Chaves F, Garnacho-Montero J, del Pozo JL, et al. Diagnosis and treatment of catheter-related 419 bloodstream infection: Clinical guidelines of the Spanish Society of Infectious Diseases and 420 Clinical Microbiology and (SEIMC) and the Spanish Society of Spanish Society of Intensive 421 and Critical Care Medicine a. Med Intensiva. 2018;42(1):5-36. 422 doi:10.1016/j.medin.2017.09.012 423 424 22. Orihuela-Martín J, Rodríguez-Núñez O, Morata L, et al. Performance of differential time to 425 positivity as a routine diagnostic test for catheter-related bloodstream infections: a single426 centre experience. Clinical Microbiology and Infection. 2020;26(3):383.e1-383.e7. Page 35 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 20 427 doi:10.1016/j.cmi.2019.07.001 428 429 23. Badia-Cebada L, Carmezim J, Pérez-Rodríguez MT, et al. Randomized Clinical Trial of the 430 Need for Antibiotic Treatment for Low-Risk Catheter-Related Bloodstream Infection Caused 431 by Coagulase-Negative Staphylococci. Antibiotics. 2023;12(5). 432 doi:10.3390/antibiotics12050839 433 434 24. Raad I, Kassar R, Ghannam D, Chaftari AM, Hachem R, Jiang Y. Management of the catheter 435 in documented catheter-related coagulase-negative staphylococcal bacteremia: Remove or 436 retain? Clinical Infectious Diseases. 2009;49(8):1187-1194. doi:10.1086/605694 437 438 25. Hebeisen U, Babouee Flury B, Atkinson A, Marschall J, Buetti N. Catheter-related 439 bloodstream infections due to coagulase-negative staphylococci managed with catheter 440 removal: Recurrences are rare. Am J Infect Control. 2020;48(7):837-839. 441 doi:10.1016/j.ajic.2019.10.013 442 443 26. Hebeisen UP, Atkinson A, Marschall J, Buetti N. Catheter-related bloodstream infections with 444 coagulase-negative staphylococci: Are antibiotics necessary if the catheter is removed? 445 Antimicrob Resist Infect Control. 2019;8(1):1-8. doi:10.1186/s13756-019-0474-x 446 Page 36 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 1 1Mortality predictors and definition proposal for complicated coagulase-negative Staphylococcus 2bacteremia. A multicenter prospective cohort study 3 4Authors 5Benedetta Varisco1, Pedro María Martínez Pérez-Crespo2, Pilar Retamar-Gentil3,4, Inmaculada López 6Hernandez3,4, Mª Carmen Fariñas-Álvarez4,5, Isabel Fernández-Natal6, María Teresa Pérez7Rodríguez7, Ane Josune Goikoetxea Aguirre8, Juan Manuel Sánchez-Calvo9, Luis Buzón Martín10, 8Eva León-Jiménez2, David Vinuesa García11, José María Reguera-Iglesias12, Alberto Bahamonde9Carrasco13, Jonathan Fernández Suárez14, Jesús Rodríguez-Baño3,4*, Luis Eduardo López-Cortés3,4*, 10 on behalf of the PROBAC REIPI/GEIH-SEIMC/SAEI Group 11 12 Affiliations 13 1Department of Health Sciences, Clinic of Infectious Diseases, University of Milan, ASST Santi Paolo 14 e Carlo, Via A. Di Rudinì, 8, 20142, Milan, Italy. 15 2Unidad de Enfermedades Infecciosas y Microbiología, Hospital Universitario de Valme, 41014, 16 Sevilla, Spain. 17 3Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen 18 Macarena; Departamentos de Medicina y Microbiología, Facultad de Medicina, Universidad de 19 Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, 41009, Seville, Spain. 20 4CIBERINFEC, Instituto de Salud Carlos III, 28029, Madrid, Spain. 21 5Unidad de Enfermedades Infecciosas, Hospital Universitario Marqués de Valdecilla, Universidad de 22 Cantabria, IDIVAL, 39008, Santander, Spain. Page 37 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 2 23 6Department of Clinical Microbiology, Complejo Asistencial Universitario de León (CAULE), 24071, 24 León, Spain. 25 7 Infectious Diseases Unit, Department of Internal Medicine, Complexo Hospitalario Universitario de 26 Vigo, 36312, Vigo, Spain; Instituto de Investigación Biomédica Galicia Sur, Spain. 27 8 Unidad de Enfermedades Infecciosas, Hospital Universitario de Cruces, 48903, Bizkaia, Spain. 28 9 Unit of Infectious Diseases and Clinical Microbiology. Jerez De La Frontera University Hospital, 29 11407, Jerez De La Frontera, Cádiz, Spain. 30 10 Unidad de Gestión Clínica de Enfermedades Infecciosas, Hospital Universitario de Burgos, 09006, 31 Burgos, Spain. 32 11 Unidad Gestión Clínica Enfermedades Infecciosas, Hospital Universitario Clínico San Cecilio, 33 18016, Granada, Spain. 34 12Servicio de Enfermedades Infecciosas, Hospital Regional Universitario de Málaga, IBIMA Málaga, 35 29010, Málaga, Spain. 36 13Departamento de Medicina Interna, Hospital de El Bierzo, 24404, Ponferrada, Spain. 37 14Unidad de Microbiología, Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), 38 Hospital Universitario Central de Asturias, 33011, Oviedo, Spain. 39 *These authors contributed equally as senior authors. 40 41 Corresponding author 42 Luis Eduardo López-Cortés 43 Infectious Diseases and Microbiology Unit 44 Hospital Universitario Virgen Macarena 45 Department of Medicine, University of Sevilla Page 38 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 3 46 Biomedicines Institute of Sevilla 47 41009 Sevilla, Spain 48 Phone: +34 670 946 430 49 Email: [email protected] Page 39 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 4 50 Keywords. Coagulase-negative staphylococci; bloodstream infections; complicated bacteremia 51 Article type. Original article 52 Subject section. Bacterial Infections; Antimicrobial Stewardship 53 Abstract word count. 245 54 Text word count. 2593 55 Page 40 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 11 171 implant, persistent bacteremia, fever ≥72 hours on treatment, metastatic infection or deep-seated focus, 172 infectious infective endocarditis (these were previously applied to S. aureus bacteremia)11, or, in 173 CRBSI cases, absence of catheter removal within 48 hours. Data quality was monitored, with queries 174 sent to local investigators for inconsistencies or missing data. 175 176 Data analysis 177 Categorical variables were presented as counts/percentages, and continuous variables as 178 median with interquartile range (IQR). Categorical variables were compared using Chi-square or 179 Fisher exact tests, and continuous variables with Student’s t-tests or Mann-Whitney U tests. Stratified 180 analyses were performed to account for potential confounders and effect modifiers. 181 Survival curves for patients with complicated and uncomplicated bacteremia were plotted and 182 compared using the log-rank test. Patients without a 30-day assessment were censored at their last 183 follow-up. Independent predictors of 30-day mortality were identified using Cox regression, with 184 variables selected through a stepwise backward procedure, including those with a p-value <0.1 in 185 univariate analysis and those of clinical or literature-based relevance. (Methods [3], Supplementary 186 Material). Interactions were evaluated. 187 To assess the impact of complicated bacteremia on mortality, we developed two models: one 188 including the composite variable "complicated bacteremia" and another including instead each 189 criterion of complicated bacteremia individually. To assess the impact of complicated bacteremia on 190 mortality, two models were developed: one with a composite variable for complicated bacteremia and 191 another analyzing each criterion individually. 192 Sensitivity analyses were performed using logistic regression to test the robustness of findings, 193 accounting for 21 patients censored before day 30. Assumptions included all 21 surviving and random Page 47 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 12 194 imputation of two deaths in 10 iterations (Methods [4], Supplementary material). Additionally, we 195 repeated the multivariable models excluding patients classified as having a rapidly fatal underlying 196 disease according to the McCabe classification (i.e., considered to have a high probability of dying 197 within the next three months) to account for cases where a terminal condition might preclude certain 198 aspects of management and in whom death would likely be primarily attributable to the underlying 199 condition (Table S5). Statistical analyses were conducted using SPSS version 29.0.1.0 (SPSS Inc., 200 Chicago, IL, USA). 201 202 RESULTS 203 204 Characteristics of the study cohort 205 The PROBAC cohort included 6,313 episodes of bacteremia, with CoNS isolated from blood 206 cultures and not considered contaminants in 481 cases. After excluding 36 cases (28 polymicrobial 207 and 8 due to S. lugdunensis ), 445 patients (7.0% of the cohort) were included in the analysis (Table 208 1). 209 Among these patients, the majority (268/445, 60.2%) were male, with a median age of 67 years 210 (IQR 54-76). The most common source of infection was catheter-related (336/445, 75.5%). At onset, 211 109/445 patients (24.5%) had a SOFA score ≥2, and 327/445 isolates (73.5%) were methicillin212 resistant. FUBC Follow-up blood cultures were performed in 122/445 patients (27.4%), revealing 213 persistent bacteremia in 44 of themse cases (36.1% of patients with FUBCs), which represents 9.8% 214 of the total cohort. Comparative analysis between patients with and without FUBC (Table S1) showed 215 that patients with endocarditis (9.8% vs. 2.2%, respectively; p<0.001), cardiac prosthetic valves (7.4% 216 vs. 2.2%, p=0.008), and pacemakers (6.6% vs. 1.5%, p=0.005) more frequently had FUBC. Active Page 48 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 13 217 empiric therapy was administered to 155 patients (34.8%). Empiric therapy was administered in 67.2% 218 of cases (299/445), and it was active in vitro in 151 (33.9%). 219 In-hospital mortality was 14.8% (66/445), while 30-day mortality was 17.3% (77/445). Data 220 for 21/445 patients (4.7%) who were discharged alive and were not assessed at day 30 before day 30 221 were censored at their last follow-up date.. 222 223 Complicated and uncomplicated CoNS BSI 224 Among the 445 patients, 240 (53.9%) met the criteria for complicated BSI (Table 2). The most 225 common issue was delayed catheter removal in CRBSI (141/240 patients, 58.8%), followed by 226 persistent fever for ≥72 hours (77/240 patients, 32.1%) and the presence of a prosthetic device (64/240 227 patients, 26.7%). 228 Patients with complicated BSI had a higher prevalence of moderate-to-severe CKD (13.3% vs. 229 7.3%, p=0.045) and were more likely to be on immunosuppressive therapy (18.3% vs. 10.2%, 230 p=0.016). Septic shock was also more frequent (10% vs. 4.4%, p=0.024). Catheter-related infections 231 were more prevalent in complicated cases (80.8% vs. 69.3%, p=0.005), while an unknown bacteremia 232 source was more common in uncomplicated cases (18.1% vs. 5%, p<0.001). Additionally, oxacillin 233 resistance was higher among those with complicated BSI (80.4% vs. 65.4%, p<0.001). 234 Mortality was significantly higher in the complicated group (22.1% vs. 11.7%, p=0.004). The 235 proportion of censored patients was similar between groups (4.5% vs. 4.8%, p=0.88). Patients with 236 complicated BSI had a longer hospital stay (median 17 days [IQR 7-34] vs. 11 days [IQR 7-23], 237 p=0.07), but there were no significant differences in ICU admission or recurrence. Survival curves are 238 shown in Figure 1. 239 Page 49 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 14 240 Predictors of mortality in CoNS BSI cases 241 Bivariate analysis results are in Table 3, and multivariate results are in Table 4. Iin multivariate 242 model 1 (Table 4), which considered “complicated BSI” as a single entity, increased mortality was 243 associated with age (HR 1.03 per year, 95% CI 1.01-1.05), cerebrovascular disease (HR 2.58, 95% CI 244 1.45-4.58), immunosuppressive therapy (HR 2.16, 95% CI 1.22-3.84), SOFA score (HR 1.09 per score 245 point, 95% CI 1.03-1.16), and complicated bacteremia (HR 2.14, 95% CI 1.29-3.53). Conversely, a 246 catheter source of bacteremia was protective (HR 0.49, 95% CI 0.30-0.80) (Table 4). The AUROC of 247 the model was 0.75 (95%CI 0.70 – 0.81). 248 In models 2, the specific criteria for complicated bacteremia were included separately instead 249 of the composite variable. In model 2a, which included all patients evaluated each criterion for 250 “complicated BSI” separately, catheter removal was categorized as CRBSI without early catheter 251 removal (reference), CRBSI with early catheter removal, and non-CRBSI; , persistent fever (HR 2.52, 252 95% CI 1.52-4.17) was a risk factor for mortality, while early catheter removal in CRBSI was 253 protective (HR 0.47, 95% CI 0.26-0.83). The catheter-related origin was not linked to mortality in this 254 model.In Model 2b, which included only the CRBSI cases, the same components were found to be 255 associated with mortality, with similar hazard ratios. The AUROC of models 2a and 2b were 0.75 256 (95%CI 0.69–0.81) and 0.75 (95%CI 0.67–0.82), respectively. 257 Tables S21 and S32 show the bivariate analysis of in-hospital mortality and the multivariate 258 logistic regression models. Predictors of in-hospital mortality were consistent with those for 30-day 259 mortality. Sensitivity analysis, which imputed two deaths for patients discharged alive before day 30, 260 revealed adjusted odds ratios (aORs) for complicated bacteremia ranging from 1.90 to 2.06, with all 261 95% CIs above 1 (Table S43). Additionally, in the sensitivity analysis excluding the 28 patients with 262 a rapidly fatal underlying disease, complicated bacteremia remained associated with increased hazards Page 50 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 15 263 of mortality (HR 2.34, 95%CI 1.37–3.99), while catheter removal with lower mortality (HR 0.46, 264 95%CI 0.25–0.85) (Table S5). 265 266 DISCUSSION 267 268 This prospective cohort study analyzed patient demographics, clinical features, complications, 269 and outcomes of CoNS BSI. A definition for complicated bacteremia was explored, consistently linked 270 toassociated with higher 30-day mortality, with two of the defining criteria showing individual 271 association: persistent fever and delayed catheter removal. 272 To our knowledge, this is the largest cohort focused solely on CoNS BSI, underscoring the 273 relative lack of research compared to the extensive literature on S. aureus BSI. The patient 274 characteristics align with those typical of catheter-related BSI, such as advanced age and high 275 comorbidity rates. Notably, about 25% of patients had sepsis at onset. The 30-day all-cause mortality 276 rate was 17.3%, with previous studies reporting a wide range of mortality rates from 4% to 24% 2–7, 277 likely due to variations in patient populations and criteria for defining CoNS bacteremia. 278 Previous studies have identified risk factors for mortality including age, lack of active antibiotic 279 treatment, and comorbidities like renal or liver disease. 2–5 In our study, advanced age, cerebrovascular 280 disease, and immunosuppressive therapy were linked to higher mortality, likely reflecting patient 281 frailty, while the SOFA score indicated acute severity. Additionally, catheter-related infections were 282 associated with lower mortality compared to infections from other sources.2–5 283 In the absence of specific criteria for CoNS BSI, our study used modified criteria originally 284 designed for S. aureus complicated bacteremia. The finding that complicated bacteremia is Page 51 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 16 285 independently associated with higher mortality suggests that these criteria might be useful for CoNS 286 BSI as well. However, the separate analysis in the second model, where not all criteria were linked to 287 mortality, indicates the need for a tailored approach to CoNS BSI in future research. 288 Follow-up blood culturesFUBC were performed in fewer than 30% of cases in our cohort. 289 Although specific guidelines are lacking, obtaining them in considering follow-up blood cultures for 290 high-risk patients with complicated bacteremia may be beneficial, especially given the observed 291 association between mortality and persistent fever, which could indicate ongoing bacteremia. In 292 contrast, FUBC follow-up blood cultures may be unnecessary for low-risk cases, as suggested by 293 Badia-Cebada et al.23 Their trial comparing antibiotics versus no antibiotics in low-risk CoNS CRBSI 294 found only one case of persistent bacteremia in the no-antibiotics group. However, the small sample 295 size due to study interruption limits the strength of these findings. 296 Although our study found early catheter removal to be protective against mortality, it was 297 carried out in fewer than 60% of CRBSI cases. IDSA guidelines consider the possibility of catheter 298 retention in uncomplicated CoNS CRBSI cases.10 However, despite the fact that some studies suggest 299 that catheter retention in CoNS bloodstream infections may not significantly impact mortality, it 300 remains an independent risk factor for recurrence,24 while recurrence is rare after catheter removal.25 301 Our finding that early catheter removal is independently associated with a protective effect on 30-day 302 mortality suggests the need for further studies to better identify the subgroups of patients in whom Page 52 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 17 303 retaining the catheter is safe. This indicates poor adherence to CRBSI management standards, even in 304 hospitals with dedicated bacteremia programs, where early catheter removal is crucial. In cases without 305 risk factors for complications, removing the infected catheter alone might be sufficient and could 306 potentially reduce the need for antibiotics. However, supporting data for this approach remains limited 307 In our study, about one-third of patients did not receive empiric therapy at the onset of infection 308 onset; however, though approximately 90% of these patients were started initiated on on targeted 309 therapy once culture results were became available. Omitting empiric therapy may be justified if no 310 risk factors are present and source control (e.g. catheter removal) is achieved, potentially making 311 subsequent targeted therapy unnecessary , as previously mentioned. 23,2625–27 However, empiric therapy 312 should be initiated if risk factors for adverse outcomes are identified. These findings highlight the need 313 for careful initial assessment and risk stratification. 314 Additionally, nearly half of the empiric therapies were ineffective against the isolated strain, 315 underscoring the importance of starting treatment that covers methicillin-resistant strains when CoNS 316 infections are suspected. This approach is supported by our finding that over 70% of isolates in our 317 cohort were methicillin-resistant, consistent with other reports. 318 Our study has limitations. First, the absence of a comparative group restricts our ability to 319 establish causal relationships and assess the proportion of mortality directly attributable to CoNS BSI 320 versus underlying conditions. Although a minority of patients did not complete the 30-day follow-up, 321 sensitivity analyses suggest the results are robust. We acknowledge that positive FUBC results likely Page 53 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 18 322 represent only a subset of persistent bacteremia cases, as some patients without FUBC may have had 323 undetected positive results; although we retrospectively analyzed factors influencing the decision to 324 perform FUBC, the data lacks clear indications for each case. Nonetheless, we believe including this 325 variable is important to highlight the need for further studies to clarify FUBC indications in CoNS 326 bloodstream infections. Management of patients was performed according to attending physicians and 327 therefore wasa not standardized; nonetheless, we believe the data reflect real-world clinical practice. 328 Despite efforts like investigator training and data monitoring, some data collection errors may still 329 exist. However, the study's strengths include a large sample size, and a multicenter, prospective design 330 conducted by infectious disease experts in hospitals with bacteremia programs, which enhances the 331 credibility of our findings. 332 In summary, our study identifies key mortality factors in CoNS BSI and emphasizes the need 333 for improved risk stratification. The proposed definition for complicated CoNS BSI needs validation 334 in further studies. 335 336 AUTHOR CONTRIBUTIONS 337 338 Conceptualization: LELC, JRB. Methodology and statistical analysis: BV, LELC, JRB. Original draft: 339 BV. Review & editing: BV, LELC, JRB (all authors). Supervision: JRB. Project administration: 340 LELC. Funding acquisition: LELC, JRB. All authors contributed to the final manuscript. 341 342 FUNDING 343 Page 54 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 19 344 This research was supported by grants from Plan Nacional de I+D+i 2013-2016, Instituto de Salud 345 Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de 346 Ciencia, Innovación y Universidades (PI16/01432); the Spanish Network for Research in Infectious 347 Diseases (REIPI) (RD16/0016/0001; RD16/ 0016/0008); and co-financed by the European Regional 348 Development Fund under the 'A way to achieve Europe' Operational Program for Intelligent Growth 349 2014-2020. 350 351 TRANSPARENCY STATEMENT 352 353 PRG has served as a scientific advisor for Shionogi and Advanz, speaker for Angelini, and Menarini. 354 LELC has served as a scientific advisor for Novartis, speaker for MSD, Pfizer, Angelini, and ViiV, 355 and trainer for MSD and ViiV. The other authors declare no conflicts of interest. Page 55 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 20 356 REFERENCES 357 1. Heilmann C, Ziebuhr W, Becker K. Are coagulase-negative staphylococci virulent? Clinical 358 Microbiology and Infection. 2019;25(9):1071-1080. doi:10.1016/j.cmi.2018.11.012 359 360 2. Molina J, Peñuela I, Lepe JA, et al. Mortality and hospital stay related to coagulase-negative 361 Staphylococci bacteremia in non-critical patients. Journal of Infection. 2013;66(2):155-162. 362 doi:10.1016/j.jinf.2012.10.021 363 364 3. Park SY, Kwon KH, Chung JW, Huh HJ, Chae SL. Coagulase-negative staphylococcal 365 bacteremia: risk factors for mortality and impact of initial appropriate antimicrobial therapy 366 on outcome. European Journal of Clinical Microbiology and Infectious Diseases. 367 2015;34(7):1395-1401. doi:10.1007/s10096-015-2364-3 368 369 4. Hentzien M, Strady C, Vernet-Garnier V, et al. Prognostic factors associated with 30-day in370 hospital mortality in coagulase-negative Staphylococcus bacteraemia: no impact of 371 vancomycin minimum inhibitory concentration. Infect Dis. 2017;49(9):664-673. 372 doi:10.1080/23744235.2017.1323346 373 374 5. Cui J, Liang Z, Mo Z, Zhang J. The species distribution, antimicrobial resistance and risk 375 factors for poor outcome of coagulase-negative staphylococci bacteraemia in China. 376 Antimicrob Resist Infect Control. 2019;8(1):1-10. doi:10.1186/s13756-019-0523-5 377 378 6. Yamada K, Namikawa H, Fujimoto H, et al. Clinical characteristics of methicillin-resistant 379 coagulase-negative staphylococcal bacteremia in a tertiary hospital. Internal Medicine. 380 2017;56(7):781-785. doi:10.2169/internalmedicine.56.7715 381 382 7. Rosa RG, Dos Santos RP, Goldani LZ. Mortality related to coagulase-negative staphylococcal 383 bacteremia in febrile neutropenia: A cohort study. Canadian Journal of Infectious Diseases 384 and Medical Microbiology. 2014;25(1):14-18. doi:10.1155/2014/702621 385 386 8. Rahkonen M, Luttinen S, Koskela M, Hautala T. True bacteremias caused by coagulase 387 negative Staphylococcus are difficult to distinguish from blood culture contaminants. 388 European Journal of Clinical Microbiology and Infectious Diseases. 2012;31(10):2639-2644. 389 doi:10.1007/s10096-012-1607-9 390 391 9. Mirrett S, Weinstein MP, Reimer LG, Wilson ML, Reller LB. Relevance of the number of 392 positive bottles in determining clinical significance of coagulase-negative staphylococci in 393 blood cultures. J Clin Microbiol. 2001;39(9):3279-3281. doi:10.1128/JCM.39.9.3279394 3281.2001 395 396 10. Mermel LA, Allon M, Bouza E, et al. Clinical practice guidelines for the diagnosis and 397 management of intravascular catheter-related infection: 2009 update by the Infectious 398 Diseases Society of America. Clinical Infectious Diseases. 2009;49(1):1-45. 399 doi:10.1086/599376 400 401 11. Jung N, Rieg S. Essentials in the management of S. aureus bloodstream infection. Infection. 402 2018;46(4):441-442. doi:10.1007/s15010-018-1130-8 403 Page 56 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Table 1. Baseline patient and infection characteristics with bivariate analysis of complicated versus uncomplicated bacteremia. Variables N (%) except where specified All patients N = 445 Uncomplicated N = 205 (46.1%) Complicated N = 240 (53.9%) p -value1 Age, median (IQR) 67 (54 - 76) 67 (53 – 76) 67 (54 – 76) 0.950 Male gender 268 (60.2) 124 (60.5) 144 (60) 0.766 Age-adjusted Charlson, median (IQR) 4 (2 - 6) 4 (2 - 6) 4 (2 - 7) 0.281 Age-adjusted Charlson >3 261 (58.7) 118 (57.6) 143 (59.6) 0.666 Comorbidities and conditions Diabetes mellitus 65 (14.6) 31 (15.1) 34 (14.2) 0.776 Chronic liver disease 39 (8.8) 23 (11.2) 16 (6.7) 0.177 Chronic kidney disease, moderate-severe 47 (10.6) 15 (7.3) 32 (13.3) 0.045 Hemodialysis 9 (2.0) 1 (0.5) 8 (3.3) 0.077 Cerebrovascular disease 50 (11.2) 25 (12.2) 25 (10.4) 0.554 COPD 40 (9.0) 18 (8.8) 22 (9.2) 0.887 Chronic heart failure 37 (8.3) 16 (7.8) 21 (8.8) 0.719 Solid cancer 115 (25.8) 46 (22.4) 69 (28.8) 0.130 Immunosuppressive therapy 65 (14.6) 21 (10.2) 44 (18.3) 0.016 Recent exposures (previous month) Antibiotic therapy 239 (53.7) 106 (51.7) 133 (55.4) 0.434 Surgery 124 (27.9) 61 (29.8) 63 (26.3) 0.411 Parenteral nutrition 97 (21.8) 48 (23.4) 49 (20.4) 0.445 Infection acquisition Community 12 (2.7) 5 (2.4) 7 (2.9) 0.757 Healthcare-associated 69 (15.5) 28 (13.7) 41 (17) 0.320 Nosocomial 364 (81.8) 171 (83.4) 193 (80.4) 0.354 Previous length of hospitalization in days, median (IQR) 15 (9 - 27) 12 (5 – 21) 12 (3 – 24) 0.891 Hospital department of admission at onset Emergency 3 (0.7) 2 (1) 1 (0.4) 0.473 Intensive care 58 (13) 24 (11.7) 34 (14.2) 0.442 Page 63 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Surgery 123 (27.6) 69 (33.7) 54 (22.5) 0.009 Medicine 241 (54.2) 100 (48.8) 141 (58.8) 0.035 Others 20 (4.5) 10 (4.9) 10 (4.2) 0.718 Vascular access devices PVC 229 (51.5) 113 (55.1) 116 (48.3) 0.153 Short-term CVC 229 (51.5) 102 (49.8) 127 (52.9) 0.506 PICC 74 (16.6) 27 (13.2) 47 (19.6) 0.070 Tunneled CVC 12 (2.7) 0 12 (5) - Port-a-Cath 6 (1.4) 0 6 (2.5) - Other devices Pacemaker 13 (2.9) 0 13 (5.4) - Vascular prosthetic device 9 (2) 0 9 (3.8) - Prosthetic joint 15 (3.4) 0 15 (6.3) - Heart valvular prosthesis 16 (3.6) 0 16 (6.7) - Severity at onset SOFA score, median (IQR) 3 (1 - 6) 3 (1 - 5) 3 (1 - 6) 0.533 SOFA score ≥2 109 (24.5) 43 (21) 66 (27.5) 0.337 Pitt score, median (IQR) 0 (0-3) 1 (0 - 2) 1 (0 - 3) 0.828 Pitt score >3 87 (19.6) 39 (19) 48 (20) 0.815 Vasopressors support 27 (6.1) 8 (3.9) 19 (7.9) 0.087 Septic shock 33 (7.4) 9 (4.4) 24 (10) 0.024 Source of infection Catheter-related 336 (75.5) 142 (69.3) 194 (80.8) 0.005 Unknown 49 (11.0) 37 (18.1) 12 (5) <0.001 Endocarditis 19 (4.3) 0 19 (7.9) - Skin and soft tissues 14 (3.2) 9 (4.4) 5 (2.1) 0.165 Abdominal 7 (1.6) 4 (2) 3 (1.3) 0.533 Bone and joints 6 (1.4) 4 (2) 2 (0.8) 0.308 Others 14 (3.2) 9 (4.4) 5 (2.1) 0.264 CoNS species Staphylococcus epidermidis 332 (74.6) 145 (70.7) 187 (77.9) 0.083 Staphylococcus hominis 60 (13.5) 32 (15.6) 28 (11.7) 0.225 Page 64 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Staphylococcus haemolyticus 30 (6.7) 16 (7.8) 14 (5.8) 0.408 Staphylococcus capitis 16 (3.6) 8 (3.9) 8 (3.3) 0.748 Staphylococcus warneri 3 (0.7) 1 (0.5) 2 (0.8) 0.657 Staphylococcus schleiferi 2 (0.5) 1 (0.5) 1 (0.4) 0.279 Staphylococcus simulans 1 (0.2) 1 (0.5) 0 0.911 Staphylococcus auricularis 1 (0.2) 1 (0.5) 0 0.279 Oxacillin resistance 327 (73.5) 134 (65.4) 193 (80.4) <0.001 Follow-up blood cultures performed2 122 (27.4) 56 (27.3) 66 (27.5) 0.873 Positive follow-up blood cultures among those performed 44/122 (36.1) 0/56 (0) 44/66 (66.7) - Source control (only CRBSI) Early (<48 h) catheter removal 195/336 (58) 142/142 (100) 53/194 (27.3) - Days until catheter removal, median (IQR) 3 (2 – 8) 1 (0 – 3) 2 (2 – 7) 0.132 Antimicrobial treatment Empiric antibiotic therapy, yes 299 (67.2) 134 (65.4) 165 (68.8) 0.449 Active empiric therapy, yes 151/299 (50.5) 76/134 (56.7) 75/165 (45.4) 0.060 Empiric antibiotic therapy, no 146 (32.8) 71 (34.6) 75 (31.3) 0.714 Empiric no, targeted yes 129/146 (88.4) 59/71 (83.1) 70/75 (93.3) 0.054 Duration in days, median (IQR) 10 (7-17) 10 (6 – 16) 12 (7 – 18) 0.002 Outcomes 30-day mortality 77 (17.3) 24 (11.7) 53 (22.1) 0.004 ICU admission 90 (20.2) 40 (19.5) 50 (20.8) 0.999 Length of hospital stay after BSI in days, median (IQR) 13 (7 – 28) 11 (7 – 23) 17 (7 – 34) 0.079 Recurrence 8 (1.8) 3 (1.5) 5 (2.1) 0.459 COPD: chronic obstructive pulmonary disease; PVC: peripheral venous catheter; CVC: central venous catheter; PICC: peripherally inserted central catheter; CRBSI: catheter-related bloodstream infection; CoNS: coagulase-negative staphylococci; ICU: intensive care unit; BSI: bloodstream infection Page 65 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 1 Statistical analyses were not performed for variables that are criteria for complicated bacteremia; they are included solely to provide data for uncomplicated bacteremia cases. 2 Follow-up blood cultures were performed at the discretion of the attending physician, with a specific analysis of patient characteristics associated with obtaining them provided in the Supplementary Material. Page 66 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Table 2. Complicated bacteremia. Criteria for complicated bacteremia among the 240 patients with episodes classified as such. Patients may have had more than one criterion. Criteria for complicated bacteremia N (%) Absence of early (<48h) catheter removal in CRSBI 141 (58.8) Persistent fever ≥72h 77 (32.1) Implanted device 64 (26.7) Persistent bacteremia 44 (18.3) Metastatic complications 7 (2.9) Primary endocarditis 19 (7.9) Page 67 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Table 2. Complicated bacteremia. Criteria for complicated bacteremia among the 240 patients with episodes classified as such. Patients may have had more than one criterion. Criteria for complicated bacteremia N (%) Absence of early (<48h) catheter removal in CRSBI 141 (58.8) Persistent fever ≥72h 77 (32.1) Implanted device 64 (26.7) Persistent bacteremia 44 (18.3) Metastatic complications 7 (2.9) Primary endocarditis 19 (7.9) Page 68 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Table 3. Bivariate analysis of the association of different variables with 30-day mortality. Variables N (%) except where specified Alive N = 368 (82.7%) Dead N = 77 (17.3%) HR 95%CI p -value Age, median (IQR) 66 (53 – 75) 75 (62 – 81) 1.03 1.01 – 1.04 0.001 Male gender 222 (60.3) 46 (59.7) 1.02 0.64 – 1.61 0.943 Age-adjusted Charlson, median (IQR) 4 (2 - 6) 2 (1 - 5) 1.09 1.01 – 1.18 0.039 Age-adjusted Charlson >3 208 (56.5) 53 (68.8) 1.62 0.99 – 2.65 0.053 Comorbidities and conditions Diabetes mellitus 55 (15) 10 (13) 0.83 0.43 – 1.61 0.575 Chronic liver disease 33 (9) 6 (7.8) 0.86 0.38 – 1.99 0.732 Chronic kidney disease, moderate-severe 34 (9.2) 13 (16.9) 1.94 1.07 – 3.53 0.029 Hemodialysis 7 (1.9) 2 (2.6) 1.13 0.26 – 4.79 0.874 Cerebrovascular disease 34 (9.2) 16 (20.8) 2.22 1.28 – 3.85 0.005 COPD 31 (8.4) 9 (11.7) 1.42 0.71 – 2.84 0.326 Chronic heart failure 25 (6.8) 12 (15.6) 2.25 1.21 – 4.16 0.010 Solid cancer 91 (24.7) 24 (31.2) 1.31 0.81 – 2.13 0.269 Immunosuppressive therapy 48 (13) 17 (22.1) 1.70 0.99 – 2.92 0.054 Recent exposures (previous month) Antibiotic therapy 191 (51.9) 48 (62.3) 1.37 0.86 – 2.18 0.188 Surgery 107 (29.1) 17 (22.1) 0.68 0.39 – 1.17 0.167 Parenteral nutrition 82 (22.3) 15 (19.5) 0.82 0.47 – 1.45 0.497 Infection acquisition Community 11 (3) 1 (1.3) 0.45 0.06 – 3.21 0.422 Healthcare-associated 51 (13.9) 18 (23.4) 1.80 1.06 – 3.06 0.029 Nosocomial 305 (82.9) 58 (75.3) 0.65 0.39 – 1.09 0.107 Previous length of hospitalization in days, median (IQR) 12 (5 – 23) 11 (3 – 24) 1.01 0.99 – 1.01 0.492 Hospital department of admission at onset Emergency 3 (0.8) 0 - - 0.606 Intensive care 41 (11.1) 17 (22.1) 2.03 1.18 – 3.48 0.010 Page 69 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Surgery 111 (30.2) 12 (15.6) 0.45 0.25 – 0.84 0.012 Medicine 195 (53) 46 (59.7) 1.31 0.83 – 2.08 0.246 Others 18 (4.9) 2 (2.6) 0.35 0.05 – 2.51 0.295 Vascular access devices PVC 184 (50) 45 (58.4) 1.33 0.84 – 2.10 0.223 Short-term CVC 191 (51.9) 38 (49.4) 0.87 0.55 – 1.36 0.530 PICC 63 (17.1) 11 (14.3) 0.79 0.42 – 1.51 0.490 Tunneled CVC 8 (2.2) 4 (5.2) 2.22 0.81 – 6.09 0.119 Port-a-Cath 4 (1.1) 3 (3.9) 3.15 0.99 – 10.01 0.051 Other devices Pacemaker 8 (2.2) 5 (6.5) 3.18 1.29 – 7.89 0.012 Vascular prosthetic device 6 (1.6) 3 (3.9) 2.26 0.71 – 7.16 0.168 Prosthetic joint 11 (3) 4 (5.2) 1.60 0.59 – 4.38 0.359 Heart valvular prosthesis 11 (3) 5 (6.5) 2.05 0.83 – 5.07 0.122 Severity at onset SOFA score, median (IQR) 3 (1 - 5) 4 (1 - 8) 1.08 1.01 – 1.14 0.015 SOFA score ≥2 80 (21.7) 29 (37.7) 1.96 1.24 – 3.12 0.004 Pitt score, median (IQR) 1 (0 - 3) 0 (0 – 2) 0.99 0.91 – 1.09 0.893 Pitt score >3 74 (20.1) 13 (16.9) 0.86 0.47 – 1.56 0.619 Vasopressors support 16 (4.4) 11 (14.3) 2.29 0.97 – 5.39 0.059 Septic shock 16 (4.4) 17 (22.1) 5.09 2.96 – 8.75 <0.001 Source of infection Catheter-related 287 (78) 49 (63.6) 0.472 0.29 – 0.75 0.002 Unknown 40 (10.9) 9 (11.7) 1.38 0.69 – 2.76 0.369 Primary eEndocarditis 13 (3.5) 6 (7.8) 2.03 0.88 – 4.66 0.097 Skin and soft tissues 10 (2.7) 4 (5.2) 1.86 0.68 – 5.08 0.228 Abdominal 6 (1.6) 1 (1.3) 0.78 0.11 – 5.61 0.805 Bone and joints 5 (1.4) 1 (1.3) 0.99 0.14 – 7.18 0.999 Others 1 (0.3) 0 - - 0.766 CoNS species Staphylococcus epidermidis 278 (75.5) 54 (70.1) 0.75 0.46 – 1.22 0.246 Staphylococcus hominis 48 (13) 12 (15.6) 1.21 0.65 – 2.24 0.545 Page 70 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Staphylococcus haemolyticus 22 (6) 8 (10.4) 1.93 0.93 – 4.01 0.080 Staphylococcus capitis 14 (3.8) 2 (2.6) 0.72 0.18 – 2.95 0.652 Staphylococcus warneri 2 (0.5) 1 (1.3) 1.89 0.26 – 13.65 0.525 Staphylococcus schleiferi 2 (0.5) 0 - - 0.674 Staphylococcus simulans 1 (0.3) 0 - - 0.766 Staphylococcus auricularis 1 (0.3) 0 - - 0.766 Oxacillin resistance 266 (72.3) 61 (79.2) 1.42 0.81 – 2.49 0.228 Follow-up blood cultures performed 101 (27.5) 21 (27.3) 0.99 0.59 – 1.67 0.998 Positive follow-up blood cultures among those performed 29 (7.9) 15 (19.5) 2.34 1.36 – 4.20 0.003 Source control (only CRBSI) Early catheter removal 173/287 (60.3) 22/49 (44.9) 0.49 0.28 – 0.88 0.016 Days until catheter removal, median (IQR) 2 (1 – 6) 4 (1 – 8) 0.85 0.73 – 0.99 0.033 Complicated bacteremia 187 (50.8) 53 (68.8) 2.09 1.28 – 3.40 0.003 Absence of early (<48h) catheter removal in CRSBI 114/287 (39.7) 27/49 (55.1) 2.04 1.14 - 3.57 0.016 Persistent fever ≥72h 52 (14.1) 25 (32.5) 2.48 1.54 – 4.01 <0.001 Implanted device 44 (12) 20 (26) 2.44 1.47 – 4.07 <0.001 Persistent bacteremia 29 (7.9) 15 (19.5) 2.34 1.36 – 4.20 0.003 Metastatic complications 4 (1.1) 3 (3.9) 2.39 0.76 – 7.60 0.138 Antimicrobial treatment Empiric antibiotic therapy, yes 249 (67.7) 50 (64.9) 0.39 0.22 – 0.69 0.441 Active empiric therapy, yes 128/249 (51.4) 23/50 (46) 0.88 0.67 - 1.17 0.383 Empiric antibiotic therapy, no 119 (32.3) 27 (35.1) 1.21 0.75 - 194 0.549 Empiric no, targeted yes 104/119 (87.4) 25/27 (92.6) 0.98 0.23 – 4.14 0.976 Duration in days, median (IQR) 11 (7 – 18) 8 (5 – 13) 0.96 0.92 – 0.99 0.009 Outcomes ICU admission 66 (17.9) 24 (31.2) 1.89 1.17 – 3.07 0.010 Length of hospital stay after BSI in days, median (IQR) 14 (7 – 31) 12 (5 – 24) 0.99 0.99 – 1.00 0.219 Recurrence 5 (1.4) 3 (3.9) 2.39 0.75 – 7.58 0.140 Page 71 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review CKD: chronic kidney disease; COPD: chronic obstructive pulmonary disease; PICC: peripherally inserted central catheter; PVC: peripheral venous catheter; CVC: central venous catheter; CRBSI: catheter-related bloodstream infection. Page 72 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Table 4. Multivariate analysis of factors associated with 30-day mortality. Model 1 included complicated bloodstream infection, and Models 2 included the individual components of complicated bloodstream infection. In Model 2a, the “early catheter removal” variable was stratified into three categories to include all patients. Model 2b included only patients with CRBSI. Model 1 Variables HR 95% CI p -value Age 1.03 1.01 – 1.05 0.002 Cerebrovascular disease 2.58 1.45 – 4.58 0.001 Immunosuppressive therapy 2.16 1.22 – 3.84 0.008 SOFA score 1.09 1.03 – 1.16 0.005 Catheter-related bacteremia 0.49 0.30 – 0.80 0.004 Complicated bacteremia 2.14 1.29 – 3.53 0.003 Model 2a Variables HR 95% CI p -value Age 1.03 1.01 – 1.05 <0.001 Cerebrovascular disease 2.49 1.41 – 4.39 0.002 Immunosuppressive therapy 1.99 1.11 – 3.57 0.o020 SOFA score 1.09 1.03 – 1.16 0.005 Fever ≥72 hours 2.52 1.52 – 4.17 <0.001 Cather management CRBSI without early catheter removal Reference - - CRBSI with early catheter removal 0.47 0.26 – 0.83 0.010 CoNS BSI of non-catheter origin 1.14 0.66 – 1.98 0.638 Model 2b Page 79 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Variables HR 95% CI p -value Age 1.03 1.01 – 1.06 0.003 Cerebrovascular disease 2.92 1.49 – 5.70 0.002 Immunosuppressive therapy 2.22 1.09 – 4.52 0.028 SOFA score 1.10 1.02 – 1.18 0.019 Fever ≥72 hours 2.20 1.14 – 4.23 0.018 Early catheter removal 0.47 0.26 – 0.84 0.010 Page 80 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review Figure 1. Cumulative survival of patients with complicated (red line) and non-complicated (blue line) bacteremia due to coagulase-negative Staphylococcus. P value (log rank test) = 0.004 Day 0 5 10 15 20 25 30 NC 205 193 198 180 177 174 170 C 240 220 206 198 192 186 175 Number at risk (NC: non complicated; C: complicated) Page 81 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 1 1SUPPLEMENTARY MATERIAL 2 3 4Methods [1]. Study microbiological and clinical inclusion criteria for CoNS bloodstream infection 5episodes 6 7Episodes of CoNS BSI were included in the study if they met both microbiological and clinical criteria as follows: 8 91. Microbiological criteria. The same CoNS species with identical antibiotic susceptibility profiles was 10 isolated from at least two distinct sets of blood cultures drawn from separate sites. In the case of 11 catheter-related CoNS BSI: 12 -If the catheter was removed and cultured, a diagnosis of CoNS CRBSI was confirmed if the same 13 CoNS species with an identical susceptibility profile was isolated from at least one peripheral blood 14 culture and from a culture of the catheter tip, in either a semiquantitative (roll plate, with presence 15 of 15 CFU per plate or more) or quantitative (vortex with a count of at least 103 CFU/segment, or 16 sonication methods, with counts above 102 CFU/segment) techniques 17 -If the catheter was not removed, the same CoNS species had to be isolated from at least one 18 peripheral blood culture and from a sample drawn from the catheter hub. Diagnosis required either 19 a differential time to positivity (time to detected microbial growth) of at least 120 minutes lower in 20 the catheter sample compared to the sample from peripheral venipuncture; or quantitative blood 21 cultures, defined as a three-fold higher colony count in the blood sample obtained from the catheter 22 compared to the peripheral blood sample. 23 24 2. Clinical signs of infection. At least two of the following clinical signs had to be present: 25 -Fever (T >38ºC) or hypothermia (T <36ºC) 26 -Tachycardia (>90 beats per minute) 27 -Tachypnea (>20 breaths per minute) or PaCO₂ < 32 mmHg 28 -Leukocytosis (>12,000 cells/mm³) or leukopenia (<4,000 cells/mm³) 29 30 3. Other infection causes were excluded. For patients with an unusual source of bacteremia for CoNS 31 (i.e., pneumonia, intraabdominal infection, meningitis, unknown source in patients without catheters or 32 prosthetic material), the opinion of a second local investigator was sought. 33 Page 82 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 2 34 Methods [2]. Justification for the inclusion of “metastatic complications” as a criterion for “complicated 35 bacteremia” 36 37 Although metastatic complications occurred in fewer than 3% of cases, this was retained as a criterion for 38 complicated bacteremia due to the pre-defined nature of the definition. Notably, 5 of the 7 patients with 39 metastatic complications also met other criteria for "complicated bacteremia" as follows: 40 -1 patient was a vascular prosthetic device carrier 41 -1 had an articular device 42 -2 had persistent fever 43 -1 had a CRBSI without early catheter removal. 44 Only 2 patients did not meet any other "complicated criteria," and the definition of complicated bacteremia in 45 these cases was solely attributed to the presence of metastatic foci. 46 47 Page 83 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 3 49 Methods [3]. Multivariable regression model methodology 50 51 In the regression analysis, we employed a backward stepwise procedure. Variables were initially included based 52 on clinical relevance (e.g., their potential confounding or modifying effect on the association between 53 complicated bacteremia and mortality, or their direct association with mortality) and statistical significance (i.e., 54 p-value < 0.1 in bivariate analysis). The variables included were age, Charlson comorbidity index, chronic kidney 55 disease, cerebrovascular disease, chronic heart failure, immunosuppressive therapy, healthcare-associated 56 acquisition, ward of admission, presence of a port-a-cath or pacemaker, SOFA score, vasopressor use, septic 57 shock, catheter-related source, endocarditis, and complicated bacteremia. Collinearity was tested, and collinear 58 variables were removed based on clinical reasoning and variance inflation factor (VIF) values. Specifically, 59 vasopressor use, ICU admission at onset, and septic shock were excluded due to collinearity with the SOFA 60 score. The Charlson comorbidity index was replaced by individual comorbidities. All variables were treated as 61 categorical, except for age and SOFA score, which were treated as continuous variables. 62 63 In Model 1, the variables that constitute the criteria for complicated bacteremia were excluded, while 64 complicated bacteremia was retained in all steps. Variables were discarded sequentially based on the backward 65 stepwise procedure, eliminating the variable with the lowest influence on model performance, as determined by 66 the Akaike Information Criterion (AIC). The final model included 424 patients, with 21 excluded due to missing 67 data. 68 69 In Model 2, we included individual criteria for complicated bacteremia and excluded the composite variable. The 70 selection procedure for variables was identical to Model 1. Model 2a included 424 patients (21 missing data), 71 considering "early catheter removal" as a 3-category variable: catheter-related bacteremia without early catheter 72 removal (reference category), catheter-related bacteremia with early catheter removal, and non-catheter-related 73 bacteremia. Model 2b included 324 patients (12 missing values), focusing solely on patients with catheter74 related bacteremia. 75 76 Interactions between complicated bacteremia and the other variables in the final model were assessed, as well 77 as interactions between the variables and each individual criterion defining complicated bacteremia. 78 79 The predictive ability of the final models was assessed by calculating the area under the receiver operating 80 characteristic (AUROC) curve with 95% confidence intervals. 81 82 Page 84 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 4 83 Methods [4]:. Sensitivity analysis 84 85 For 21 patients, confirmation of survival at day 30 was lacking. Consequently, these patients were excluded 86 from the primary multivariable analysis for mortality predictors. All these patients were discharged alive. For 87 patients with complete data, 8 out of 379 patients who were discharged alive had died by day 30 (2%; 99% CI 88 0-5%). Based on this, an anticipation of 0-1 deaths among the 21 patients was established. Subsequently, two 89 scenarios were examined: 90 91 (a) assuming all 21 patients survived at day 30; a multivariate analysis for mortality predictors was performed 92 using logistic regression considering that all 21 patients were alive at day 30. 93 94 (b) considering the possibility of up to 2 deaths among them (two-fold the upper limit of 99% CI). Death was 95 randomly assigned to 2 patients from the 21 with 10 iterations, and multivariate logistic regression was 96 conducted for each iteration. 97 Page 85 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 5 98 Table S1. Features of patients according to whether follow-up blood cultures were performed 99 100 FUBC performed N=122 FUBC not performed N=323 OR (95% CI) p-value Fever >72h 27 (22.1) 50 (15.5) 1.55 (0.92 - 2.62) 0.098 Lack of early catheter removal (CRBSI only) 21/89 (23.6) 120/247 (48.6) 0.50 (0.33 - 0.77) <0.001 Endocarditis 12 (9.8) 7 (2.2) 4.93 (1.89 - 12.82) <0.001 Cardiac prosthetic valve 9 (7.4) 7 (2.2) 3.60 (1.31 – 9.88) 0.008 Pacemaker 8 (6.6) 5 (1.5) 4.46 (1.43 – 13.92) 0.005 Articular prosthetic device 6 (4.9) 9 (2.8) 1.81 (0.63 – 5.18) 0.266 Endovascular prosthesis 5 (4.1) 4 (1.2) 3.41 (0.90 – 12.91) 0.056 Metastatic foci 3 (2.5) 4 (1.2) 2.01 (0.44 – 9.12) 0.356 SOFA score – median (IQR) 3 (1 – 5) 3 (1 – 6) 0.97 (0.91 - 1.04) 0.407 101 Page 86 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 6 102 103 Table S21. Bivariate analysis of the association of different exposures with in-hospital mortality. 104 Variable OR 95% CI p-value Age, median (IQR) 1.03 1.01 – 1.05 <0.001 Age > 65 2.48 1.39 - 4.42 0.002 Male gender 1.26 0.74 – 2.15 0.402 Age-adjusted Charlson 1.07 0.90 – 1.26 0.442 Age-adjusted Charlson > 3 2.40 1.34 – 4.27 0.002 Diabetes mellitus 1.03 0.50 – 2.14 0.936 Chronic liver disease 0.82 0.31 – 2.17 0.683 Moderate-to-severe CKD 2.14 1.05 – 4.37 0.034 Hemodialysis 1.44 0.28 – 7.53 0.648 Cerebrovascular disease 2.79 1.43 – 5.45 0.002 COPD 1.47 0.64 – 3.34 0.359 Chronic heart failure 3.13 1.49 – 6.59 0.002 Rheumatologic disease 0.51 0.07 – 3.98 1.000 Solid malignancy 1.27 0.73 – 2.29 0.416 Immunosuppressive therapy 1.89 0.99 – 3.61 0.050 Recent antibiotic therapy (30 days) 1.48 0.87 – 2.52 0.145 Recent surgery (30 days) 0.53 0.27 – 1.03 0.049 Community-acquired 0.97 0.95 – 0.99 0.228 Hospital-acquired 0.61 0.33 – 1.13 0.112 Healthcare-associated 1.92 1.02 – 3.62 0.040 PICC 1.00 0.49 – 2.02 0.998 PVC 1.41 0.83 – 2.38 0.201 Short-term CVC 0.87 0.52 – 1.46 0.511 Tunneled catheter 3.42 0.97 – 12.02 0.068 Port-a-Cath® 0.96 0.11 – 8.06 1.000 Pacemaker 2.65 0.79 – 8.85 0.112 Vascular prosthetic device 1.66 0.34 – 8.15 0.628 Articular prosthesis 1.45 0.39 – 5.29 0.476 Cardiac valvular prosthesis 1.97 0.62 – 6.29 0.275 Parenteral nutrition 0.77 0.39 – 1.50 0.437 Septic shock 5.07 2.40 – 10.72 <0.001 SOFA score 1.20 1.07 – 1.34 0.002 SOFA score ≥ 2 1.92 1.11 – 3.35 0.019 Pitt score 1.26 1.12 – 1.43 <0.001 Pitt score > 3 3.57 1.57 – 8.12 0.004 ICU admission 1.75 0.97 – 3.15 0.072 Amine support 2.55 0.88 – 7.40 0.080 Oxacillin resistance 1.54 0.73 - 3.27 0.260 Catheter-related focus 0.55 0.31 – 0.96 0.034 Unknown focus 1.14 0.51 – 2.55 0.755 Urinary focus 0.99 0.98 – 0.99 0.348 Skin and soft tissues focus 1.59 0.43 – 5.87 0.480 Endocarditis focus 2.82 1.03 – 7.69 0.036 Respiratory focus 6.05 1.47 – 24.82 0.005 Abdominal focus 0.96 0.11 – 8.08 0.967 Bone and joints focus 1.15 0.13 – 10.01 0.899 No empiric therapy received 1.03 0.59 - 1.79 0.922 Adequacy of empiric therapy 0.79 0.42 – 1.50 0.468 Duration empiric, days 0.99 0.97 – 1.01 0.481 Duration targeted, days 0.99 0.99 – 1.01 0.353 Page 87 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 7 Duration of therapy overall, days 0.95 0.92 – 0.99 0.013 Catheter removal in CRBSI 0.59 0.31 - 1.11 0.101 Persistent fever at 72h 3.19 1.787 – 5.71 <0.001 Persistent bacteremia 2.74 1.350 – 5.57 0.005 Septic complications 4.45 0.974 – 20.37 0.036 Complicated BSI 2.19 1.25 – 3.85 0.005 Days to catheter removal 0.99 0.84 – 1.17 0.888 ICU stay, days 0.99 0.98 – 1.01 0.758 BSI onset to discharge, days 0.99 0.99 – 1.01 0.327 Admission to BSI, days 1.00 0.99 – 1.01 0.447 105 CKD: chronic kidney disease; COPD: chronic obstructive pulmonary disease; AIDS: acquired 106 immunodeficiency syndrome; PICC: peripherally inserted central catheter; PVC: peripheral venous catheter; 107 CVC: central venous catheter; ICU: intensive care unit; CRBSI: catheter-related bloodstream infection; BSI: 108 bloodstream infection 109 Page 88 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 14 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included 10111213 (b) Report category boundaries when continuous variables were categorized 12131011 Main results 16 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period - Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses 1113, S3 Discussion Key results 18 Summarize key results with reference to study objectives 11121314 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias 131415 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence 11141416 Generalizability 21 Discuss the generalizability (external validity) of the study results 1415 Other information Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based 1417 198 199 *Give information separately for exposed and unexposed groups. 200 201 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological 202 background and published examples of transparent reporting. The STROBE checklist is best used in 203 conjunction with this article (freely available on the Web sites of PLoS Medicine at 204 http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology at 205 http://www.epidem.com/). Information on the STROBE Initiative is available at http://www.strobe206 statement.org. Page 95 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 1 1SUPPLEMENTARY MATERIAL 2 3 4Methods [1]. Study microbiological and clinical inclusion criteria for CoNS bloodstream infection 5episodes 6 7Episodes of CoNS BSI were included in the study if they met both microbiological and clinical criteria as follows: 8 91. Microbiological criteria. The same CoNS species with identical antibiotic susceptibility profiles was 10 isolated from at least two distinct sets of blood cultures drawn from separate sites. In the case of 11 catheter-related CoNS BSI: 12 -If the catheter was removed and cultured, a diagnosis of CoNS CRBSI was confirmed if the same 13 CoNS species with an identical susceptibility profile was isolated from at least one peripheral blood 14 culture and from a culture of the catheter tip, in either a semiquantitative (roll plate, with presence 15 of 15 CFU per plate or more) or quantitative (vortex with a count of at least 103 CFU/segment, or 16 sonication methods, with counts above 102 CFU/segment) techniques 17 -If the catheter was not removed, the same CoNS species had to be isolated from at least one 18 peripheral blood culture and from a sample drawn from the catheter hub. Diagnosis required either 19 a differential time to positivity (time to detected microbial growth) of at least 120 minutes lower in 20 the catheter sample compared to the sample from peripheral venipuncture; or quantitative blood 21 cultures, defined as a three-fold higher colony count in the blood sample obtained from the catheter 22 compared to the peripheral blood sample. 23 24 2. Clinical signs of infection. At least two of the following clinical signs had to be present: 25 -Fever (T >38ºC) or hypothermia (T <36ºC) 26 -Tachycardia (>90 beats per minute) 27 -Tachypnea (>20 breaths per minute) or PaCO₂ < 32 mmHg 28 -Leukocytosis (>12,000 cells/mm³) or leukopenia (<4,000 cells/mm³) 29 30 3. Other infection causes were excluded. For patients with an unusual source of bacteremia for CoNS 31 (i.e., pneumonia, intraabdominal infection, meningitis, unknown source in patients without catheters or 32 prosthetic material), the opinion of a second local investigator was sought. 33 Page 96 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 2 34 Methods [2]. Justification for the inclusion of “metastatic complications” as a criterion for “complicated 35 bacteremia” 36 37 Although metastatic complications occurred in fewer than 3% of cases, this was retained as a criterion for 38 complicated bacteremia due to the pre-defined nature of the definition. Notably, 5 of the 7 patients with 39 metastatic complications also met other criteria for "complicated bacteremia" as follows: 40 -1 patient was a vascular prosthetic device carrier 41 -1 had an articular device 42 -2 had persistent fever 43 -1 had a CRBSI without early catheter removal. 44 Only 2 patients did not meet any other "complicated criteria," and the definition of complicated bacteremia in 45 these cases was solely attributed to the presence of metastatic foci. 46 47 Page 97 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 3 49 Methods [3]. Multivariable regression model methodology 50 51 In the regression analysis, we employed a backward stepwise procedure. Variables were initially included based 52 on clinical relevance (e.g., their potential confounding or modifying effect on the association between 53 complicated bacteremia and mortality, or their direct association with mortality) and statistical significance (i.e., 54 p-value < 0.1 in bivariate analysis). The variables included were age, Charlson comorbidity index, chronic kidney 55 disease, cerebrovascular disease, chronic heart failure, immunosuppressive therapy, healthcare-associated 56 acquisition, ward of admission, presence of a port-a-cath or pacemaker, SOFA score, vasopressor use, septic 57 shock, catheter-related source, endocarditis, and complicated bacteremia. Collinearity was tested, and collinear 58 variables were removed based on clinical reasoning and variance inflation factor (VIF) values. Specifically, 59 vasopressor use, ICU admission at onset, and septic shock were excluded due to collinearity with the SOFA 60 score. The Charlson comorbidity index was replaced by individual comorbidities. All variables were treated as 61 categorical, except for age and SOFA score, which were treated as continuous variables. 62 63 In Model 1, the variables that constitute the criteria for complicated bacteremia were excluded, while 64 complicated bacteremia was retained in all steps. Variables were discarded sequentially based on the backward 65 stepwise procedure, eliminating the variable with the lowest influence on model performance, as determined by 66 the Akaike Information Criterion (AIC). The final model included 424 patients, with 21 excluded due to missing 67 data. 68 69 In Model 2, we included individual criteria for complicated bacteremia and excluded the composite variable. The 70 selection procedure for variables was identical to Model 1. Model 2a included 424 patients (21 missing data), 71 considering "early catheter removal" as a 3-category variable: catheter-related bacteremia without early catheter 72 removal (reference category), catheter-related bacteremia with early catheter removal, and non-catheter-related 73 bacteremia. Model 2b included 324 patients (12 missing values), focusing solely on patients with catheter74 related bacteremia. 75 76 Interactions between complicated bacteremia and the other variables in the final model were assessed, as well 77 as interactions between the variables and each individual criterion defining complicated bacteremia. 78 79 The predictive ability of the final models was assessed by calculating the area under the receiver operating 80 characteristic (AUROC) curve with 95% confidence intervals. 81 82 Page 98 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 4 83 Methods [4]. Sensitivity analysis 84 85 For 21 patients, confirmation of survival at day 30 was lacking. Consequently, these patients were excluded 86 from the primary multivariable analysis for mortality predictors. All these patients were discharged alive. For 87 patients with complete data, 8 out of 379 patients who were discharged alive had died by day 30 (2%; 99% CI 88 0-5%). Based on this, an anticipation of 0-1 deaths among the 21 patients was established. Subsequently, two 89 scenarios were examined: 90 91 (a) assuming all 21 patients survived at day 30; a multivariate analysis for mortality predictors was performed 92 using logistic regression considering that all 21 patients were alive at day 30. 93 94 (b) considering the possibility of up to 2 deaths among them (two-fold the upper limit of 99% CI). Death was 95 randomly assigned to 2 patients from the 21 with 10 iterations, and multivariate logistic regression was 96 conducted for each iteration. 97 Page 99 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 5 98 Table S1. Features of patients according to whether follow-up blood cultures were performed 99 100 FUBC performed N=122 FUBC not performed N=323 OR (95% CI) p-value Fever >72h 27 (22.1) 50 (15.5) 1.55 (0.92 - 2.62) 0.098 Lack of early catheter removal (CRBSI only) 21/89 (23.6) 120/247 (48.6) 0.50 (0.33 - 0.77) <0.001 Endocarditis 12 (9.8) 7 (2.2) 4.93 (1.89 - 12.82) <0.001 Cardiac prosthetic valve 9 (7.4) 7 (2.2) 3.60 (1.31 – 9.88) 0.008 Pacemaker 8 (6.6) 5 (1.5) 4.46 (1.43 – 13.92) 0.005 Articular prosthetic device 6 (4.9) 9 (2.8) 1.81 (0.63 – 5.18) 0.266 Endovascular prosthesis 5 (4.1) 4 (1.2) 3.41 (0.90 – 12.91) 0.056 Metastatic foci 3 (2.5) 4 (1.2) 2.01 (0.44 – 9.12) 0.356 SOFA score – median (IQR) 3 (1 – 5) 3 (1 – 6) 0.97 (0.91 - 1.04) 0.407 101 Page 100 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 6 102 103 Table S2. Bivariate analysis of the association of different exposures with in-hospital mortality. 104 Variable OR 95% CI p-value Age, median (IQR) 1.03 1.01 – 1.05 <0.001 Age > 65 2.48 1.39 - 4.42 0.002 Male gender 1.26 0.74 – 2.15 0.402 Age-adjusted Charlson 1.07 0.90 – 1.26 0.442 Age-adjusted Charlson > 3 2.40 1.34 – 4.27 0.002 Diabetes mellitus 1.03 0.50 – 2.14 0.936 Chronic liver disease 0.82 0.31 – 2.17 0.683 Moderate-to-severe CKD 2.14 1.05 – 4.37 0.034 Hemodialysis 1.44 0.28 – 7.53 0.648 Cerebrovascular disease 2.79 1.43 – 5.45 0.002 COPD 1.47 0.64 – 3.34 0.359 Chronic heart failure 3.13 1.49 – 6.59 0.002 Rheumatologic disease 0.51 0.07 – 3.98 1.000 Solid malignancy 1.27 0.73 – 2.29 0.416 Immunosuppressive therapy 1.89 0.99 – 3.61 0.050 Recent antibiotic therapy (30 days) 1.48 0.87 – 2.52 0.145 Recent surgery (30 days) 0.53 0.27 – 1.03 0.049 Community-acquired 0.97 0.95 – 0.99 0.228 Hospital-acquired 0.61 0.33 – 1.13 0.112 Healthcare-associated 1.92 1.02 – 3.62 0.040 PICC 1.00 0.49 – 2.02 0.998 PVC 1.41 0.83 – 2.38 0.201 Short-term CVC 0.87 0.52 – 1.46 0.511 Tunneled catheter 3.42 0.97 – 12.02 0.068 Port-a-Cath 0.96 0.11 – 8.06 1.000 Pacemaker 2.65 0.79 – 8.85 0.112 Vascular prosthetic device 1.66 0.34 – 8.15 0.628 Articular prosthesis 1.45 0.39 – 5.29 0.476 Cardiac valvular prosthesis 1.97 0.62 – 6.29 0.275 Parenteral nutrition 0.77 0.39 – 1.50 0.437 Septic shock 5.07 2.40 – 10.72 <0.001 SOFA score 1.20 1.07 – 1.34 0.002 SOFA score ≥ 2 1.92 1.11 – 3.35 0.019 Pitt score 1.26 1.12 – 1.43 <0.001 Pitt score > 3 3.57 1.57 – 8.12 0.004 ICU admission 1.75 0.97 – 3.15 0.072 Amine support 2.55 0.88 – 7.40 0.080 Oxacillin resistance 1.54 0.73 - 3.27 0.260 Catheter-related focus 0.55 0.31 – 0.96 0.034 Unknown focus 1.14 0.51 – 2.55 0.755 Urinary focus 0.99 0.98 – 0.99 0.348 Skin and soft tissues focus 1.59 0.43 – 5.87 0.480 Endocarditis focus 2.82 1.03 – 7.69 0.036 Respiratory focus 6.05 1.47 – 24.82 0.005 Abdominal focus 0.96 0.11 – 8.08 0.967 Bone and joints focus 1.15 0.13 – 10.01 0.899 No empiric therapy received 1.03 0.59 - 1.79 0.922 Adequacy of empiric therapy 0.79 0.42 – 1.50 0.468 Duration empiric, days 0.99 0.97 – 1.01 0.481 Duration targeted, days 0.99 0.99 – 1.01 0.353 Page 101 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 7 Duration of therapy overall, days 0.95 0.92 – 0.99 0.013 Catheter removal in CRBSI 0.59 0.31 - 1.11 0.101 Persistent fever at 72h 3.19 1.787 – 5.71 <0.001 Persistent bacteremia 2.74 1.350 – 5.57 0.005 Septic complications 4.45 0.974 – 20.37 0.036 Complicated BSI 2.19 1.25 – 3.85 0.005 Days to catheter removal 0.99 0.84 – 1.17 0.888 ICU stay, days 0.99 0.98 – 1.01 0.758 BSI onset to discharge, days 0.99 0.99 – 1.01 0.327 Admission to BSI, days 1.00 0.99 – 1.01 0.447 105 CKD: chronic kidney disease; COPD: chronic obstructive pulmonary disease; AIDS: acquired 106 immunodeficiency syndrome; PICC: peripherally inserted central catheter; PVC: peripheral venous catheter; 107 CVC: central venous catheter; ICU: intensive care unit; CRBSI: catheter-related bloodstream infection; BSI: 108 bloodstream infection 109 Page 102 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 8 110 Table S3. Multivariate analysis of factors associated with in-hospital mortality. 111 112 Model 1 113 Variable OR 95% CI p-value Age > 65 years old 2.87 1.53 – 5.38 < 0.001 Cerebrovascular disease 3.55 1.72 – 7.35 < 0.001 Immunosuppressive therapy 3.24 1.55 – 6.79 0.002 Pitt score > 3 3.54 1.45 – 8.64 0.005 Complicated BSI 1.99 1.10 – 3.59 0.023 114 BSI: bloodstream infection 115 AUROC of 0.74 (95%CI 0.67 – 0.80) 116 117 Model 2 118 Variable OR 95% CI p-value Age > 65 years old 2.99 1.57 – 5.71 < 0.001 Cerebrovascular disease 3.59 1.71 – 7.53 < 0.001 Immunosuppressive therapy 2.98 1.34 - 6.41 0.005 Pitt score > 3 3.52 1.38 – 8.97 0.009 Fever > 72 hours 3.29 1.72 – 6.28 < 0.001 Early catheter removal in CRBSI 0.50 0.28 - 0.90 0.020 119 CRBSI: catheter-related bloodstream infection 120 AUROC of 0.76 (95%CI 0.69 – 0.82) 121 Page 103 of 108 Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For Peer Review 9 122 Table S4. Sensitivity analysis for 30-day mortality prediction. 123 124 125 (a) Considering that all 21 patients survived at day 30 126 Variable OR 95% CI p-value Age > 65 years 2.47 1.39 – 4.38 0.002 Cerebrovascular disease 3.07 1.53 – 6.19 0.002 Immunosuppressive therapy 2.92 1.46 – 5.85 0.003 Pitt score > 3 3.39 1.43 – 8.00 0.005 Complicated BSI 1.95 1.12 – 3.37 0.017 127 BSI: bloodstream infection 128 AUROC 0.72 (95%CI 0.66 – 0.78) 129 130 (b) Considering that up to 2 patients of them died (chosen randomly). We provide a summary of 131 the estimates in the 10 models. 132 1. Variable OR 95% CI p-value Age > 65 years 2.47 1.40 – 4.38 0.002 Cerebrovascular disease 3.07 1.53 – 6.19 0.002 Immunosuppressive therapy 2.92 1.46 – 5.85 0.003 Pitt score > 3 3.39 1.43 – 8.00 0.005 Complicated BSI 1.95 1.12 – 3.37 0.017 133 BSI: bloodstream infection 134 AUROC 0.73 (95%CI 0.66 – 0.78) 135 136 2. Variable OR 95% CI p-value Age > 65 years 2.58 1.46 – 4.57 0.001 Cerebrovascular disease 2.93 1.46 – 5.91 0.003 Immunosuppressive therapy 2.87 1.43 – 5.75 0.003 Pitt score > 3 3.90 1.67 – 9.12 0.002 Complicated BSI 1.90 1.10 – 3.25 0.021 137 BSI: bloodstream infection 138 AUROC 0.73 (95%CI 0.66 – 0.78) 139 3. 140 Variable OR 95% CI p-value Age > 65 years 2.59 1.47 – 4.59 0.001 Cerebrovascular disease 3.33 1.67 – 6.67 < 0.001 Immunosuppressive therapy 2.84 1.42 – 5.69 0.003 Pitt score > 3 3.22 1.36 – 7.63 0.009 Complicated BSI 2.01 1.20 – 3.59 0.009 141 BSI: bloodstream infection 142 AUROC 0.73 (95%CI 0.67 – 0.79) 143 144 4. 145 Variable OR 95% CI p-value Page 104 of 108Clinical Microbiology and Infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60