Infect Dis Ther (2022) 11:1505–1519 https://doi.org/10.1007/s40121-022-00657-1 ORIGINAL RESEARCH Duration of Treatment for Pseudomonas aeruginosa Bacteremia: a Retrospective Study Tanya Babich . Pontus Naucler . John Karlsson Valik . Christian G. Giske . Natividad Benito . Ruben Cardona . Alba Rivera . Celine Pulcini . Manal Abdel Fattah . Justine Haquin . Alasdair Macgowan . Sally Grier . Bibiana Chazan . Anna Yanovskay . Ronen Ben Ami . Michal Landes . Lior Nesher . Adi Zaidman-Shimshovitz . Kate McCarthy . David L. Paterson . Evelina Tacconelli . Michael Buhl . Susanna Mauer . Jesús Rodrı́guez-Baño . Marina de Cueto . Antonio Oliver . Enrique Ruiz de Gopegui . Angela Cano . Isabel Machuca . Monica Gozalo-Marguello . Luis Martinez-Martinez . Eva M. Gonzalez-Barbera . Iris Gomez Alfaro . Miguel Salavert . Bojana Beovic . Andreja Saje . Manica Mueller–Premru . Leonardo Pagani . Virginie Vitrat . Diamantis Kofteridis . Maria Zacharioudaki . Sofia Maraki . Yulia Weissman . Mical Paul . Yaakov Dickstein . Leonard Leibovici . Dafna Yahav Received: March 30, 2022 / Accepted: May 9, 2022 / Published online: May 25, 2022 Ó The Author(s) 2022 ABSTRACT Methods: We present a retrospective multicen- ter study including patients with P. aeruginosa Introduction: There is no consensus regarding bacteremia during 2009–2015. We evaluated optimal duration of antibiotic therapy for outcomes of patients treated with short (6–- Pseudomonas aeruginosa bacteremia. We aimed 10 days) versus long (11–15 days) antibiotic to evaluate the impact of short antibiotic courses. The primary outcome was a composite course. of 30-day mortality or bacteremia recurrence and/or persistence. Univariate and inverse probability treatment-weighted (IPTW) adjusted Supplementary Information The online version multivariate analysis for the primary outcome contains supplementary material available at https:// doi.org/10.1007/s40121-022-00657-1. T. Babich  R. B. Ami  Y. Weissman N. Benito Sackler Faculty of Medicine, Tel Aviv University, Tel Infectious Diseases Unit, Department of Internal Aviv, Israel Medicine, Hospital de la Santa Creu i Sant Pau - Institut d’Investigació Biomèdica Sant Pau, T. Babich  L. Leibovici Universitat Autònoma de Barcelona, Barcelona, Rabin Medical Center, Research Authority, Spain Beilinson Hospital, Petah-Tikva, Israel R. Cardona P. Naucler  J. K. Valik Department of Internal Medicine, Hospital de la Division of Infectious Diseases, Department of Santa Creu i Sant Pau, Barcelona, Spain Medicine Solna, Karolinska Institutet, Department of Infectious Diseases, Karolinska University A. Rivera Hospital, Stockholm, Sweden Department of Microbiology, Hospital de la Santa Creu i Sant Pau - Institut d’Investigació Biomèdica C. G. Giske Sant Pau, Barcelona, Spain Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden C. Pulcini Université de Lorraine, APEMAC, 54000 Nancy, France 1506 Infect Dis Ther (2022) 11:1505–1519 was performed. To avoid immortal time bias, Keywords: Pseudomonas aeruginosa; Bac- the landmark method was used. teremia; Antibiotics; Duration; Antimicrobial Results: We included 657 patients; 273 stewardship received a short antibiotic course and 384 a long course. There was no significant difference in Key Summary Points baseline characteristics of patients. The com- posite primary outcome occurred in 61/384 Pseudomonas aeruginosa bacteremia is a patients in the long-treatment group (16%) severe infection, often treated with long- versus 32/273 in the short-treatment group course (* 14 days) antibiotics. (12%) (p = 0.131). Mortality accounted for 41/384 (11%) versus 25/273 (9%) of cases, We aimed to assess whether 6–10 days of respectively. Length of hospital stay was signif- antibiotics would be as effective as icantly shorter in the short group [median 11–15 days for this infection. 13 days, interquartile range (IQR) 9–21 days, versus median 15 days, IQR 11–26 days, Data from 657 patients with P. aeruginosa p = 0.002]. Ten patients in the long group dis- bacteremia collected retrospectively continued antibiotic therapy owing to adverse demonstrated no association between events, compared with none in the short group. duration of therapy and mortality or On univariate and multivariate analyses, dura- bacteremia recurrence. tion of therapy was not associated with the Short therapy was associated with less drug primary outcome. discontinuation and shorter length of stay. Conclusions: In this retrospective study, 6–- 10 days of antibiotic course for P. aeruginosa Short course of antibiotics (6–10 days) bacteremia were as effective as longer courses in may be considered for P. aeruginosa terms of survival and recurrence. Shorter ther- bacteremia. apy was associated with reduced length of stay and less drug discontinuation. C. Pulcini  M. A. Fattah  J. Haquin E. Tacconelli  M. Buhl  S. Mauer Infectious Diseases Department, Université de Division of Infectious Diseases, Tuebingen Lorraine, CHRU-Nancy, 54000 Nancy, France University Hospital, Tuebingen, Germany A. Macgowan  S. Grier J. Rodrı́guez-Baño  M. de Cueto Department of Infection Sciences, Pathology Departamentos de Medicina y Microbiologı́a, Sciences Building, Southmead Hospital, Bristol, UK Unidad Clı́nica de Enfermedades Infecciosas and Microbiologı́a, Hospital Universitario Virgen B. Chazan  A. Yanovskay Macarena, Universidad de Sevilla, Instituto de Infectious Diseases Unit, Rappaport Faculty of Biomedicina de Sevilla (IBiS), Seville, Spain Medicine, Technion, Emek Medical Center, Afula, Haifa, Israel A. Oliver  E. R. de Gopegui Servicio de Microbiologı́a and Unidad de R. B. Ami  M. Landes Investigación, Hospital Universitario Son Espases, Infectious Diseases Unit, Sourasky Medical Center, Instituto de Investigación Illes Balears (IdISBa), Tel-Aviv, Israel Palma, Spain L. Nesher  A. Zaidman-Shimshovitz A. Cano  I. Machuca Infectious Disease Institute, Soroka Medical Center, Infectious Diseases Unit, Maimonides Biomedical Ben-Gurion University of the Negev, Beer Sheba, Research Institute of Cordoba (IMIBIC), Reina Sofia Israel University Hospital, University of Cordoba, Cordoba, Spain K. McCarthy  D. L. Paterson UQ Centre for Clinical Research, The University of M. Gozalo-Marguello  L. Martinez-Martinez Queensland, Brisbane, Australia Microbiology Service, University Hospital Marqués de Valdecilla-IDIVAL, Santander, Spain Infect Dis Ther (2022) 11:1505–1519 1507 use prolonged (C 14 days) antibiotic course to INTRODUCTION treat P. aeruginosa bacteremia [14]. Few data are available to guide the duration of antibiotic In recent years, evidence has accumulated to treatment for P. aeruginosa bacteremia. Two support short duration of antibiotic therapy for previous retrospective studies including * 540 Gram-negative bacteremia in general. Several patients in total demonstrated no difference in nonrandomized studies have demonstrated no mortality or recurrence using 7–11 days of difference in mortality or other outcomes antibiotics compared with longer treatment of between short (6–10 days) and longer therapy 12–21 days [11, 15]. for Gram-negative bacteremia [1–6]. Three ran- We aimed to further investigate whether domized controlled trials have also demon- short-duration antibiotics is non-inferior to strated noninferiority of 7 days antibiotic longer duration in cases of P. aeruginosa treatment compared with 14 days. The vast bacteremia. majority of these studies, however, included Enterobacterales [7–9] Patients with Pseudomonas aeruginosa bac- METHODS teremia are more likely to be neutropenic or have another immunosuppressive condition, Data Collection and Patient Inclusion and more likely to have hospital-acquired bac- teremia compared with those with Enterobac- We used a large database of P. aeruginosa bac- terales bacteremia. The most common sources teremia collected during 2009–2015 from nine of bacteremia are pulmonary, central line-asso- countries, 25 centers [16]. The collection of data ciated, urinary tract, and unknown source, with was approved by the medical ethical commit- pulmonary source carrying worse prognosis tees of each participating center. STROBE [10–12]. Intrinsic resistance to some antibiotics, guidelines for reporting in epidemiological enhanced by tremendous ability to acquire studies were followed for the reporting of the resistance to others, in addition to paucity of current study. available oral agents, makes this pathogen dif- We included consecutive adult patients ficult to treat [13]. Hence, some recommend to E. M. Gonzalez-Barbera  I. G. Alfaro M. Paul  Y. Dickstein Microbiology Department, La Fe University Infectious Diseases Unit, Rambam Health Care Hospital, Valencia, Spain Campus, Haifa, Israel M. Salavert Infectious Diseases Unit, Hospital Universitario y D. Yahav (&) Politécnico La Fe, Valencia, Spain Infectious Diseases Unit, Sheba Medical Center, 2 Sheba Road, 52621 Ramat-Gan, Israel B. Beovic  A. Saje e-mail: dafna.yahav@gmail.com Department of Infectious Diseases, University Medical Centre, Ljubljana, Slovenia C. G. Giske Department of Clinical Microbiology, Karolinska M. Mueller–Premru University Hospital, Stockholm, Sweden Institute of Microbiology and Immunology, Faculty of Medicine, University of Ljubljana, Ljubljana, J. Rodrı́guez-Baño  M. de Cueto Slovenia Centro de Investigación Biomédica en Red en Enfermedades Infecciosas (CIBERINFEC), Seville, L. Pagani  V. Vitrat Spain Infectious Diseases Unit, Annecy-Genevois Hospital Center (CHANGE), Annecy, France B. Beovic  A. Saje Faculty of Medicine, University of Ljubljana, D. Kofteridis  M. Zacharioudaki  S. Maraki Ljubljana, Slovenia Infectious Disease Unit, Department of Internal Medicine, University Hospital of Heraklion, Heraklion, Crete, Greece 1508 Infect Dis Ther (2022) 11:1505–1519 (age C 18 years) hospitalized with P. aeruginosa occurring upon completion of antibiotic bacteremia. Further details on patients’ identi- treatment. fication and data collection are published else- Secondary outcomes included individual where [12, 16, 17]. For the current analysis, and components of the primary outcome: length of considering previous studies [11, 15], we inclu- hospital stay, Clostridioides difficile-associated ded patients who received at least 6 days of diarrhea, and discontinuation of antibiotics due antibiotics and less than 16 days, assuming that to adverse events. Detailed definitions of these patients with prolonged treatments probably outcomes were previously described [16]. had a complicated course, with confirmed/sus- pected metastatic foci of infection, and Statistical Analysis endovascular or osteoarticular infection. Patients receiving either monotherapy or com- Categorical data are expressed as numbers and bination therapy were included, excluding percentages. Continuous data are expressed as monotherapy with aminoglycosides, polymyx- means with standard deviation (SD) or medians ins, antipseudomonal penicillin, or initial oral with interquartile range (IQR), as appropriate. fluoroquinolone treatment. Patients with step- The t-test was used for comparison of normally down to oral fluoroquinolones following any distributed variables, and Mann–Whitney test duration of initial intravenous therapy were for abnormally distributed variables. Compar- also included. We dichotomized the compar- ison of categorical variables was conducted ison to short (6–10 days) and longer therapy using v2 test or Fisher’s exact test, as appropri- (11–15 days), considering any in vitro antibiotic ate. A propensity score for duration of therapy courses (as above), administered in hospital and was calculated for each patient by adjusting for post-discharge. all pretreatment variables potentially con- We documented baseline patients charac- founding the duration of treatment. Variables teristics (demographics, comorbidities), infec- used to estimate the propensity score are tion-related parameters (source, severity, detailed in Supplementary Material. Assuming susceptibility data), and management data—the baseline differences between groups, we used latter included documentation of appropriate- inverse probability weighting to control for ness of empirical treatment. Appropriate was these differences. Standardized mean difference defined as in vitro antibiotic courses adminis- was used to check whether balance between the tered withing 24 or 48 h (both were docu- baseline characteristics was achieved; less than mented). Type of main definitive antibiotic 10% difference was considered a reasonable therapy (i.e., administered according to sus- balance. Weighted subjects were entered in the ceptibility testing) and duration were docu- generalized estimating equation (GEE) binary mented as well. Outcomes were also collected, logistics. To avoid immortal time bias, the as below. landmark method was used. Landmark was set at day 11 to ensure no deaths occurred in the Outcomes short treatment group (6–10 days). Missing val- ues of covariates were handled by multiple The primary outcome was a composite of mor- imputation. The center was introduced as a tality or recurrence/persistence of bloodstream random effect variable using GEE models. For infection within 30 days of positive blood cul- model selection, we used the quasi-likelihood ture, defined as: recurrence—positive blood under the independence model criterion (QIC). cultures for Pseudomonas after negative ones; Six models with varied subset of covariates were and persistence—positive blood cultures for tested, and the model with the lowest QIC was Pseudomonas after more than 72 h of treatment, selected. as documented in patient records. Recurrence/ Variables in the analyses were considered persistence was considered an outcome if statistically significant if they had a Infect Dis Ther (2022) 11:1505–1519 1509 Table 1 Patient characteristics by duration of treatment group (short/long) Variables Short treatment Long treatment All cohort p value N = 273 N = 384 N = 657 Age (years), N = 655 (median, 25–75%) 68 (57.5–79) 67 (54–76) 67 (55–77) 0.094 Weight (kg), N = 369 (median, 25–75%) 72.5 (60–82.25) 71 (60–85) 72 (60–83.8) 0.984 Height (m), N = 326 (mean ± SD) 1.69 ± 0.09 1.68 ± 0.09 1.69 ± 0.09 0.556 Gender (female) 102 (37.4%) 133 (34.6%) 235 (35.8%) 0.472 Department of hospitalization N = 273 N = 383 N = 656 0.727 Non-ICU 233 (85.3%) 318 (83%) 551 (84%) ICU 40 (14.7%) 65 (17%) 105 (16%) Daily living activities N = 215 N = 339 N = 554 Completely able 102 (18.4%) 189 (34.1%) 291 (52.5%) 0.221 Limited activities 56 (26%) 79 (23.3%) 135 (24.4%) Need help with activities of daily living 47 (21.9%) 55 (16.2%) 102 (18.4%) Bedridden 10 (4.7%) 16 (4.7%) 26 (4.7%) Arrival from N = 272 N = 381 N = 653 0.226 Home 225 (82.7%) 330 (86.6%) 555 (85%) Nursing home/LTCF 20 (7.4%) 27 (7.1%) 47 (7.2%) Another hospital 27 (9.9%) 24 (6.3%) 51 (7.8%) Previous hospitalization (90 days) 155 (58.1%) 206 (55.4%) 361 (56.5%) 0.501 Endotracheal 36 (13.2%) 51 (13.3%) 87 (13.2%) 0.972 Arterial line 38 (13.9%) 62 (16.2%) 100 (15.3%) 0.418 Central venous line 127 (46.5%) 183/382 (47.9%) 310 (47.3%) 0.726 Urinary catheter 112/271 (41.3%) 142/382 (37.2%) 254/653 (38.9%) 0.283 Any foreign body 52/271 (19.2%) 88/382 (23%) 140/653 (21.4%) 0.238 Chemotherapy previous 30 days 79 (28.9%) 111 (28.9%) 190 (28.9%) 0.993 Corticosteroids 69/272 (25.4%) 75/384 (19.5%) 144/656 (22%) 0.075 Previous surgery (90 days) 68 (24.9%) 103 (26.8%) 171 (26%) 0.582 Chronic dialysis 9 (3.3%) 17 (4.4%) 26 (4%) 0.481 Intravenous drug use 5/271(1.8%) 3 (0.8%) 8/655 (1.2%) 0.286 Neutropenia 56 (20.5%) 86 (22.4%) 142 (21.6%) 0.563 Charlson score (median, 25–75%) 4 (1–6) 4 (1.25–6) 4 (1–6) 0.499 1510 Infect Dis Ther (2022) 11:1505–1519 Table 1 continued Variables Short treatment Long treatment All cohort p value N = 273 N = 384 N = 657 Charlson score CHF 40 (14.7%) 49 (12.8%) 89 (13.5%) 0.485 Solid metastatic 34 (12.5%) 58 (15.1%) 92 (14%) 0.335 Hemiplegia 14 (5.1%) 28 (7.3%) 42 (6.4%) 0.264 Diabetes with end organ damage 36 (13.2%) 37 (9.6%) 73 (11.1%) 0.153 Dementia 19 (7%) 21 (5.5%) 40 (6.1%) 0.431 Moderate-to-severe renal disease 25 (9.2%) 49 (12.8%) 74 (11.3%) 0.150 Chronic pulmonary disease 42 (15.4%) 73 (19%) 115 (17.5%) 0.228 Mild liver disease 12 (4.4%) 13 (3.4%) 25 (3.8%) 0.505 Moderate-to-severe liver disease 6 (2.2%) 12 (3.1%) 18 (2.9%) 0.473 Rheumatic 22 (8.1%) 36 (9.4%) 58 (8.8%) 0.558 Any malignancy 115 (42.1%) 162 (42.2%) 277 (42.2%) 0.987 Place of acquisition 0.532 Unknown 13 (4.8%) 12 (3.1%) 25 (3.8%) Community 35 (12.8%) 43 (11.2%) 78 (11.9%) Healthcare associated* 74 (27.1%) 118 (30.7%) 192 (29.2%) Nosocomial 151 (55.3%) 211 (54.9%) 362 (55.1%) MDR Pseudomonas 20 (7.3%) 39 (10.2%) 59 (9%) 0.211 Aminoglycoside resistance 33 (12.1%) 55 (14.3%) 88 (13.4%) 0.407 Fluoroquinolone resistance 63 (23.1%) 77 (20.1%) 140 (21.3%) 0.346 Sepsis presentation Source of bacteremia 0.277 Unknown 68 (24.9%) 79 (20.6%) 147 (22.4%) Abdominal 17 (6.2%) 28 (7.3%) 45 (6.8%) Line associated 64 (23.4%) 104 (27.1%) 168 (25.6%) Pulmonary 40 (14.7%) 62 (16.1%) 102 (15.5%) Skin, soft tissue, bone, and joints 16 (5.9%) 25 (6.5%) 41 (6.2%) Urinary 67 (24.5%) 78 (20.8%) 145 (22.1%) Other 1 (0.4%) 8 (2.1%) 9 (1.4%) SOFA score, N = 571 (median, 25–75%) 3 (1–5) 3 (1–5) 3 (1–5) 0.172 Temperature, N = 634 (mean ± SD) 38.5 ± 0.94 38.5 ± 0.94 38.5 ± 0.94 0.562 Systolic blood pressure, N = 616 (mean ± SD) 103.47 ± 24.3 104.37 ± 22.7 104 ± 23.4 0.637 Infect Dis Ther (2022) 11:1505–1519 1511 Table 1 continued Variables Short treatment Long treatment All cohort p value N = 273 N = 384 N = 657 Diastolic blood pressure, N = 517 (median, 25–75%) 55 (49–65) 58 (49–64) 56 (49–64.5) 0.382 Leukocytes, N = 641 (median, 25–75%) 11 (2.25–17.7) 8.7 (2–15.3) 9.5 (2.03–15.8) 0.170 Neutrophils, N = 566 (median, 25–75%) 8.34 (1.02–13.8) 6.8 (1.1–13) 7.6 (1.09–13.2) 0.385 Creatinine, N = 633 (median, 25–75%) 1.06 (0.71–1.6) 1.01 (0.72–1.6) 1.03 (0.71–1.6) 0.887 Sepsis management Main treatment Antipseudomonal cephalosporin 81 (29.7%) 116 (30.2%) 197 (30%) 0.331 Antipseudomonal penicillin 129 (47.3%) 162 (42.2%) 291 (44.3%) Carbapenem 63 (23.1%) 106 (27.6%) 169 (25.7%) Appropriate empirical treatment (\ 48 h) 207/247 313/360 520/607 0.278 (83.8%) (86.9%) (85.7%) Appropriate empirical treatment (\ 24 h) 179/247 97/360 (73.1%) 442/607 0.873 (72.5%) (72.8%) Combination therapy 50 (18.3%) 69 (18%) 119 (18.1%) 0.910 Treatment duration (median, 25–75%) 8 (7–10) 13 (12–14) 11 (9–14) \ 0.001 ICU intensive care unit, LTCF long-term care facility, CHF congestive heart failure, MDR multidrug resistant *Healthcare-associated bacteremia was defined as any patient coming from long-term care facility center, home intravenous or wound care previous 30 days, hemodialysis patients, previous C 2 days hospitalization during last 90 days; nosocomial bacteremia was defined as onset after C 48 h of hospitalization p value \ 0.05. All statistical analyses were median duration being 8 days (7–10 days) and conducted using SPSS program. the long arm 13 days (12–14 days, p \ 0.001). Baseline characteristics of patients by duration of therapy are presented in Table 1. Variables RESULTS included in the propensity score are detailed in Supplementary Material, with standardized In the original database, 2396 patients were mean difference of the original and weighted included. Of these, 657 patients were treated for cohort. 6–15 days with an in vitro course of antipseu- No significant differences were documented domonal therapy that fulfilled our inclusion in either demographic data, infection charac- criteria above (see Supplemental Fig. 1 for flow teristics, or treatment between patients receiv- chart of patient selection). Median age was ing short or long therapy. Sources of bacteremia 67 years [interquartile range (IQR) 55–77 years], were also similar between groups (Table 1). A and 234 (35.8%) were females. Of these trend for older age and more frequent use of patients, 273 received antipseudomonal regi- corticosteroids was observed in the short-dura- men for 6–10 days (short course), and 384 tion group, though nonsignificant [median age received 11–15 days (long course). Overall, 68 years, interquartile range (IQR) 57.5–79 years median treatment duration in the entire cohort versus 67 years (54–76 years), p = 0.075; corti- was 11 days (IQR 9–14 days), with the short-arm costeroids 69/272, 25.4% versus 75/384, 19.5%, 1512 Infect Dis Ther (2022) 11:1505–1519 Table 2 Outcomes by duration of treatment group (short/long) Variable Short Long All cohort p value treatment treatment N = 657 N = 273 N = 384 30-Day mortality* 25 (9.2%) 41 (10.7%) 66 (10%) 0.523 Days of hospital stay—entire cohort, N = 544 (median, 13 (9–21) 15 (11–26) 15 (10–23) 0.002 25–75%) Fever duration, N = 562 (median, 25–75%) 2 (1–3) 1 (1–3) 2 (1–3) 0.280 Recurrent/persistent bacteremia* 8/264 (3%) 21/375 (5.6%) 29/639 0.124 (4.5%) Adverse events 0.398 Renal failure N = 265 N = 372 N = 637 No 240 (90.6%) 328 (88.2%) 568 (89.2%) Risk 12 (4.5%) 23 (6.2%) 35 (5.5%) Injury 4 (1.5%) 11 (3%) 15 (2.4%) Failure 6 (2.3%) 3 (0.8%) 9 (1.4%) Loss 2 (0.8%) 5 (1.3%) 7 (1.1%) End stage kidney disease 1 (0.4%) 2 (0.5%) 3 (0.5%) Clostridioides difficile 10/270 (3.7%) 9/379 (2.4%) 19/649 0.322 (2.9%) Drug discontinuation d/t adverse events 0 10 (2.8%) 10 (1.6%) 0.006 GNR Gram-negative rods * Mortality and recurrence/persistence do not sum up because one patient in each group died following a case of recurrence/ persistence p = 0.094]. Multidrug-resistant (MDR) P. aerugi- 69/384 (18%) in the long treatment group and nosa rates were similar between groups (short 20 36/273 (13.2%) in the short arm (p = 0.11). (7.3%) versus long 39 (10.2%), p = 0.211) Positive repeated cultures, indicating recur- (Table 1). rence/persistence, were demonstrated among The primary outcome of mortality or recur- 21/375 (5.6%) long-arm patients versus 8/264 rence/persistence occurred overall in 93 (3%) short-arm patients (p = 0.124). Length of patients, 61/384 in the long-duration group hospital stay was significantly shorter in the (15.9%) and 32/273 in the short-duration group short-duration group (median 13 days, IQR 9–- (11.7%) (p = 0.131). Components of the pri- 21 days, versus 15 days, IQR 11–26 days, mary outcome were without significant differ- p = 0.002). Drug discontinuation due to adverse ence between groups (Table 2). All-cause events was significantly more common in the mortality at 30 days was 41 (10.7%) among long-duration group, with ten cases compared those treated with longer duration, compared with none in the short treatment group with 25 (9.2%) among patients treated with (p = 0.006). These ten cases included five cases short antibiotic course (p = 0.523). Repeated of discontinuation of b-lactams and five blood cultures were obtained from 105 patients, Infect Dis Ther (2022) 11:1505–1519 1513 quinolones. Similar rates of C. difficile were outcome of mortality or recurrence in multi- observed between the groups (Table 2). variate analysis included baseline metastatic Univariate analysis for risk factors for the malignancy, and a more severe clinical presen- composite outcome of mortality or recurrence is tation of infection, represented by a higher described in Table 3. Arrival to hospital from SOFA score, and probably more severe baseline another institution (nursing home or another condition, represented by a nasogastric tube. hospital); metastatic malignancy; previous hos- Duration of treatment (short or long) was not a pitalization within 90 days; nasogastric tube or significant predictor. urinary catheter at bacteremia onset; higher These findings are in accordance with previ- sequential organ failure assessment (SOFA) ous results of two smaller, retrospective, single- score; and lower systolic blood pressure were country studies from the USA and South Korea. significantly associated with the primary out- Fabre et al. included 249 patients with P. aerug- come. MDR P. aeruginosa rates, early appropriate inosa bacteremia, 69 of whom received short- empirical antibiotic therapy, and combination duration therapy of 7–11 days and 180 of whom therapy rates were nonsignificantly different. received longer regimens of 12–21 days. No Duration of therapy (short versus long) was not difference in rates of death or recurrent infec- associated with this outcome (Table 3). Multi- tion within 30 days were demonstrated between variate analysis demonstrated baseline meta- short and long therapy in this study. Length of static malignancy [odds ratio (OR) 2.5, 95% hospital stay was significantly shorter in the confidence interval (CI) 1.36–4.6]; higher SOFA short-duration group, similar to our findings. score (OR 1.1, 95% CI 1.01–1.2); and presence of [11] Similarly, Bae et al. evaluated 290 patients nasogastric tube (OR 2.11, 95% CI 1.18–3.74) as with P. aeruginosa bacteremia, 97 of whom predictors of mortality or recurrence. Duration received short-course therapy (7–11 days) and of treatment was not significantly associated 193 of whom received 12–21 days of therapy. with the primary outcome in this analysis (OR This study also found no significant difference 1.66, 95% CI 0.94–2.95) [Table 4; odds ratio [ 1 in mortality or recurrence between groups. represents higher risk for the composite out- In recent years, shortening the duration of come with long-duration therapy (not statisti- antibiotic therapy has become an important cally significant)]. backbone of antibiotic stewardship. Potential benefits are reduction in adverse events, cost, superinfections including fungal infection and DISCUSSION C. difficile, and prevention of resistance devel- opment. Shorter antibiotic courses have been In this multinational, multicenter study demonstrated effective in randomized con- including 657 patients with P. aeruginosa bac- trolled trials for the treatment of various respi- teremia, we did not find a strong association ratory infections, urinary tract infections, between duration of therapy and the composite intraabdominal infections, and others [18]. outcome of mortality or recurrence/persistence Specifically for uncomplicated Gram-negative of infection among patients treated with 6–- bacteremia, three recent randomized controlled 10 days antibiotic course compared with a trials have demonstrated noninferiority of longer course of 11–15 days. Mortality at 7 days antibiotic course compared with 14 days 30 days was similar (9.2% with short antibiotic [7–9]. However, only one of these trials included course and 10.7% with longer duration, nonfermenters, with limited number of p = 0.523). Duration of hospital stay was sig- P. aeruginosa cases [7]; and immunocompro- nificantly shorter in the short-duration group, mised patients were excluded from one trial [8] and discontinuation of antibiotics due to and constituted less than one-fourth of patients adverse events was significantly more common in the other trials [7, 9]. P. aeruginosa bacteremia in the long-duration group. No significant dif- is frequently hospital acquired, and occurs in ferences were observed for C. difficile-associated patients who are immunocompromised, criti- diarrhea rates. Predictors of the composite cally ill, or with chronic underlying medical 1514 Infect Dis Ther (2022) 11:1505–1519 Table 3 Univariate analysis for risk factors for the primary outcome—mortality/recurrence Variable Mortality/recurrence Mortality/recurrence p value (no) (yes) N = 552 N = 105 Age (years), N = 655 (median, 25–75%) 67 (55–78) 68 (55–76) 0.951 Weight (kg), N = 356 (median, 25–75%) 72 (60–84.1) 70 (61–82) 0.613 Height (m), N = 326 (mean ± SD) 1.68 ± 0.09 1.7 ± 0.08 0.272 Gender (female) 199 (36.1%) 36 (34.3%) 0.729 Department of hospitalization N = 551 N = 105 0.931 Non-ICU 464 (84.2%) 87 (82.9%) ICU 87 (15.8%) 18 (17.1%) Daily living activities N = 462 N = 92 0.337 Completely able 248 (53.7%) 43 (46.7%) Limited activities 113 (24.5%) 22 (23.9%) Need help with activities of daily living 82 (17.7%) 20 (21.7%) Bedridden 19 (4.1%) 7 (7.6%) Arrival from N = 550 N = 103 0.141 Home 473 (86%) 82 (79.6%) Nursing home/LTCF 35 (6.4%) 12 (11.7%) Another hospital 42 (7.6%) 9 (8.7%) Residency—home 473 (85.7%) 82 (78.1%) 0.049 Previous hospitalization (90 days) 296/539 (54.9%) 65/100 (65%) 0.062 Endotracheal 69 (12.5%) 18 (17.1%) 0.198 Arterial line 84/550 (15.3%) 16 (15.2%) 0.993 Central venous line 251/550 (45.6%) 59 (56.2%) 0.047 Urinary catheter 205/549 (37.3%) 49/104 (47.1%) 0.061 Any foreign body 117/549 (21.3%) 23/104 (22.1%) 0.855 Chemotherapy previous 30 days 156 (28.3%) 34 (32.4%) 0.393 Steroids 120/551 (21.8%) 24 (22.9%) 0.807 Previous surgery (90 days) 139 (25.2%) 32 (30.5%) 0.257 Chronic dialysis 23 (4.2%) 3 (2.9%) 0.784 Intravenous drug use 5 (0.9%) 3 (2.9%) 0.119 Neutropenia 119 (21.6%) 23 (21.9%) 0.937 Charlson score (median, 25–75%) 4 (1–6) 4 (2–6) 0.328 Infect Dis Ther (2022) 11:1505–1519 1515 Table 3 continued Variable Mortality/recurrence Mortality/recurrence p value (no) (yes) N = 552 N = 105 Charlson score CHF 72 (13%) 17 (16.2%) 0.388 Solid metastatic 67 (12.1%) 25 (23.8%) 0.002 Hemiplegia 38 (6.9%) 4 (3.8%) 0.283 Diabetes with end organ damage 64 (11.6%) 9 (8.6%) 0.366 Dementia 32 (5.8%) 8 (7.6%) 0.474 Moderate-to-severe renal disease 63 (11.4%) 11 (10.5%) 0.781 Chronic pulmonary disease 101 (18.3%) 14 (13.3%) 0.220 Mild liver disease 19 (3.4%) 6 (5.7%) 0.265 Moderate-to-severe liver disease 15 (2.7%) 3 (2.9%) 1 Rheumatic 51 (9.2%) 7 (6.7%) 0.394 Any malignancy 229 (41.5%) 48 (45.7%) 0.421 Place of acquisition 0.239 Unknown 22 (4%) 3 (2.9%) Community 71 (12.9%) 7 (6.7%) Healthcare associated 162 (29.3%) 30 (28.6%) Nosocomial 297 (53.8%) 65 (61.9%) Place of acquisition—healthcare/nosocomial 459 (83.2%) 95 (90.5%) 0.058 MDR Pseudomonas 49 (8.9%) 10 (9.5%) 0.832 Aminoglycosides Resistance 72 (13%) 16 (15.2%) 0.545 Sepsis presentation Source of bacteremia 0.404 Unknown 120 (21.7%) 27 (25.7%) Abdominal 38 (6.9%) 7 (6.7%) Line associated 141 (25.5%) 27 (25.7%) Pulmonary 81 (14.7%) 21 (20%) Skin, soft tissue, bone, and joints 38 (6.9%) 3 (2.9%) Urinary 127 (23%) 18 (17.1%) Other 7 (1.3%) 2 (1.9%) SOFA score, N = 571 (median, 25–75%) 3 (1–5) 4 (2–6) 0.002 Temperature, N = 634 (mean ± SD) 38.5 ± 0.95 38.4 ± 0.86 0.758 1516 Infect Dis Ther (2022) 11:1505–1519 Table 3 continued Variable Mortality/recurrence Mortality/recurrence p value (no) (yes) N = 552 N = 105 Systolic blood pressure, N = 616 (mean ± SD) 105.2 ± 23.6 97.9 ± 21.6 0.004 Diastolic blood pressure, N = 517 (median, 25–75%) 58 (50–65) 53 (45–64) 0.129 Leukocytes, N = 641 (median, 25–75%) 9.5 (2.08–16.1) 9.2 (1.8–15.4) 0.913 Neutrophils, N = 566 (median, 25–75%) 7.6 (1.08–13.2) 7 (1.12–13.4) 0.898 Creatinine, N = 633 (median, 25–75%) 1.02 (0.71–1.6) 1.07 (0.74–1.76) 0.681 Treatment Duration of treatment—long 318 (57.6%) 66 (62.9%) 0.317 Appropriate empirical treatment (\ 24 h) 365/505 (72.3%) 77/102 (75.5%) 0.506 Appropriate empirical treatment (\ 48 h) 430/505 (85.1%) 90/102 (88.2%) 0.417 Combination therapy 104 (18.8%) 15 (14.3) 0.267 ICU intensive care unit, LTCF long-term care facility, CHF congestive heart failure, MDR multidrug resistant Table 4 Risk factors for a composite outcome; multivariate generalized linear models (generalized estimating equation), adjusted by inverse propensity score weighting Risk factor Univariate analysis OR (95% CI) Multivariate logistic regression analysis OR (95% CI) Propensity score – 0.33 (0.08–1.4) Duration of treatment—long 1.42 (0.89–2.25) 1.66 (0.94–2.95) Residency—home 0.62 (0.36–1.07) 0.67 (0.35–1.26) Solid metastatic 2.35 (1.38–4.02) 2.5 (1.36–4.6) Previous hospitalization 1.52 (0.95–2.43) 1.47 (0.86–2.5) SOFA score 1.14 (1.05–1.23) 1.1 (1.01–1.2) Nasogastric tube 2.26 (1.32–3.86) 2.11 (1.18–3.74) Systolic blood pressure 1.14 (1.05–1.23) 0.99 (0.98–1.0) N = 657, QIC 940.85, constant b = 0.043 As nasogastric tube and urinary catheter at presentation probably represent similarly baseline patient condition, only one was selected for multivariable analysis Odds ratio [ 1 correlates with higher risk for the composite outcome OR odds ratio, CI confidence interval conditions. Antibiotic resistance is common, accompanied by previous studies, support the and mortality is high [19]. Consequently, tra- use of shorter duration of therapy. ditional duration of therapy is considered by Rac et al. proposed using a combination of some to be 14 days [11, 15]. Our results, several vital signs and laboratory values at Infect Dis Ther (2022) 11:1505–1519 1517 72–96 h after Gram-negative bacteremia onset to test our findings, and are the optimal design as early clinical failure criteria. These were sug- to solve this clinical question. gested as a tool to guide treatment duration in a personalized approach [20]. Considering this, and our results, supporting the safety of shorter CONCLUSIONS therapy, further studies may test this approach for P. aeruginosa bacteremia. In this retrospective study, for patients with Limitations of our study are its retrospective P. aeruginosa bacteremia, we did not demon- design, with data collected until 2015. Though strate a difference in clinical outcomes between propensity score weighting was used to adjust patients treated with 6–10 days antibiotics for differences between groups, residual bias compared with longer courses. Duration of cannot be excluded. In addition, patients trea- hospital stay was shorter and adverse events ted for more than 15 days were excluded, leading to drug discontinuation were less com- assuming a complicated infection, without mon among patients treated with a shorter detailed documentation of these complications, course. Prospective studies, ideally randomized owing to the retrospective nature of this study. controlled trials, should be designed to test The comparison of duration ranges (6–10 days shorter antibiotic therapy for P. aeruginosa bac- versus 11–15 days), rather than a fixed duration teremia, and define subgroups of patients who (e.g., 7 or 14 days), limits our ability to provide a would benefit most from this approach. specific recommended duration that is safe and effective. It also selects for a specific population, possibly limiting the generalizability of our ACKNOWLEDGEMENTS results for any patient with P. aeruginosa bac- teremia. Since the study was retrospective, the outcome of recurrent and/or persistent bac- Funding. No funding or sponsorship was teremia was not based on a preplanned culture received for this study or publication of this collection schedule, rather the practice used in article. each center according to the patient’s clinical condition. In addition, data regarding source Authorship. All named authors meet the control are lacking. Since we excluded patients International Committee of Medical Journal treated more than 15 days, we assume that most Editors (ICMJE) criteria for authorship for this cases without appropriate source control were article, take responsibility for the integrity of not included in this analysis, though this could the work as a whole, and have given their not be verified. Another limitation lies in the approval for this version to be published. use of the landmark method, which disregards patients who died/were censored before the Author Contributions. Dafna Yahav, Tanya landmark and restricts implementation of Babich, and Leonard Leibovici contributed to results to included patients. This inclusion only the study conception and all authors partici- of patients who survived at least 11 days may pated in the study’s protocol writing and study produce a bias favoring the short therapy. The design. Material preparation and data collection propensity score analysis for this study on were performed by all authors. Dafna Yahav, duration may have better used variables of days Tanya Babich, and Leonard Leibovici performed 6–7, trying to mimic a randomized controlled the analysis and wrote the first draft of the trial recruiting at that point. However, these manuscript. All authors commented on previ- variables were missing for a considerable por- ous versions of the manuscript, and read and tion of patients as well as, when available, approved the final manuscript. probably reflecting sicker patients still hospi- talized and tested. Considering these limita- Disclosures. Tanya Babich, Pontus Naucler, tions, randomized controlled trials are needed John Karlsson Valik, Christian G. Giske, Natividad Benito, Ruben Cardona, Alba Rivera, 1518 Infect Dis Ther (2022) 11:1505–1519 Celine Pulcini, Manal Abdel Fattah, Justine REFERENCES Haquin, Alasdair Macgowan, Sally Grier, Bib- iana Chazan, Anna Yanovskay, Ronen Ben Ami, 1. Chotiprasitsakul D, Han JH, Cosgrove SE, Harris Michal Landes, Lior Nesher, Adi Zaidman- AD, Lautenbach E, Conley AT, et al. Comparing the Shimshovitz, Kate McCarthy, David L Paterson, outcomes of adults with Enterobacteriaceae bac- Evelina Tacconelli, Michael Buhl, Susanna teremia receiving short-course versus prolonged- Mauer, Jesús Rodrı́guez-Baño, Marina de Cueto, course antibiotic therapy in a multicentre, propen- sity score-matched cohort. Clin Infect Dis. 2018;66: Antonio Oliver, Enrique Ruiz de Gopegui, 172–7. https://doi.org/10.1093/cid/cix767. Angela Cano, Isabel Machuca, Monica Gozalo- Marguello, Luis Martinez-Martinez, Eva M 2. Sousa A, Pérez-Rodrı́guez MT, Suárez M, Val N, Gonzalez-Barbera, Iris Gomez Alfaro, Miguel Martı́nez-Lamas L, Nodar A, et al. Short- versus long-course therapy in Gram-negative bacilli Salavert, Bojana Beovic, Andreja Saje, Manica bloodstream infections. Eur J Clin Microbiol Infect Mueller–Premru, Leonardo Pagani, Virginie Dis. 2019;38:851–7. https://doi.org/10.1007/ Vitrat, Diamantis Kofteridis, Maria Zachari- s10096-019-03467-5. oudaki, Sofia Maraki, Yulia Weissman, Mical 3. Giannella M, Pascale R, Toschi A, Ferraro G, Gra- Paul, Yaakov Dickstein, Leonard Leibovici and ziano E, Furii F, et al. Treatment duration for Dafna Yahav all have nothing to disclose. Escherichia coli bloodstream infection and out- comes: retrospective single-centre study. Clin Compliance with Ethics Guidelines. The Microbiol Infect. 2018;24:1077–83. https://doi.org/ study was approved by the local ethical com- 10.1016/j.cmi.2018.01.013. mittees of each participating center, which 4. Doi A, Morimoto T, Iwata K. Shorter duration of waved informed consent due to the retrospec- antibiotic treatment for acute bacteraemic cholan- tive nature of the study. gitis with successful biliary drainage: a retrospective cohort study. Clin Microbiol Infect. 2018;24: 1184–9. https://doi.org/10.1016/j.cmi.2018.01.021. Data Availability. The datasets generated during and/or analyzed during the current 5. Uno S, Hase R, Kobayashi M, Shiratori T, Nakaji S, study are available from the corresponding Hirata N, et al. Short-course antimicrobial treat- author on reasonable request, and with agree- ment for acute cholangitis with Gram-negative bacillary bacteremia. Int J Infect Dis. 2017;55:81–5. ment of all contributing authors. https://doi.org/10.1016/j.ijid.2016.12.018. Open Access. This article is licensed under a 6. Nelson AN, Justo JA, Bookstaver PB, Kohn J, Creative Commons Attribution-Non- Albrecht H, Al-Hasan MN. Optimal duration of Commercial 4.0 International License, which antimicrobial therapy for uncomplicated Gram- negative bloodstream infections. Infection. permits any non-commercial use, sharing, 2017;45:613–20. https://doi.org/10.1007/s15010- adaptation, distribution and reproduction in 017-1020-5. any medium or format, as long as you give appropriate credit to the original author(s) and 7. Yahav D, Franceschini E, Koppel F, Turjeman A, Babich T, Bitterman R, et al. Seven versus fourteen the source, provide a link to the Creative days of antibiotic therapy for uncomplicated Gram- Commons licence, and indicate if changes were negative bacteremia: a non-inferiority randomized made. The images or other third party material controlled trial. Clin Infect Dis. 2018. https://doi. in this article are included in the article’s org/10.1093/cid/ciy1054. Creative Commons licence, unless indicated 8. von Dach E, Albrich WC, Brunel A-S, Prendki V, otherwise in a credit line to the material. If Cuvelier C, Flury D, et al. Effect of C-reactive pro- material is not included in the article’s Creative tein-guided antibiotic treatment duration, 7-day Commons licence and your intended use is not treatment, or 14-day treatment on 30-day clinical failure rate in patients with uncomplicated Gram- permitted by statutory regulation or exceeds the negative bacteremia: a randomized clinical trial. permitted use, you will need to obtain permis- JAMA. 2020;323:2160–9. https://doi.org/10.1001/ sion directly from the copyright holder. To view jama.2020.6348. a copy of this licence, visit http:// 9. Molina J, Montero-Mateos E, Praena-Segovia J, creativecommons.org/licenses/by-nc/4.0/. León-Jiménez E, Natera C, López-Cortés LE, et al. Infect Dis Ther (2022) 11:1505–1519 1519 Seven-versus 14-day course of antibiotics for the 16. Babich T, Naucler P, Valik JK, Giske CG, Benito N, treatment of bloodstream infections by Enterobac- Cardona R, et al. Ceftazidime, carbapenems, or terales: a randomized, controlled trial. Clin Micro- piperacillin–tazobactam as single definitive therapy biol Infect. 2021. https://doi.org/10.1016/j.cmi. for Pseudomonas aeruginosa bloodstream infection— 2021.09.001. a multi-site retrospective study. Clin Infect Dis. 2019. https://doi.org/10.1093/cid/ciz668. 10. Choi Y, Paik JH, Kim JH, Han SB, Durey A. Clinical predictors of Pseudomonas aeruginosa bacteremia in 17. Babich T, Naucler P, Valik JK, Giske CG, Benito N, emergency department. Emerg Med Int. 2018;2018: Cardona R, et al. Combination versus monotherapy 7581036. https://doi.org/10.1155/2018/7581036. as definitive treatment for Pseudomonas aeruginosa bacteraemia: a multicentre retrospective observa- 11. Fabre V, Amoah J, Cosgrove SE, Tamma PD. tional cohort study. J Antimicrob Chemother. Antibiotic therapy for Pseudomonas aeruginosa 2021;76:2172–81. https://doi.org/10.1093/jac/ bloodstream infections: how long is long enough? dkab134. Clin Infect Dis. 2019. https://doi.org/10.1093/cid/ ciz223. 18. Spellberg B, Rice LB. Duration of antibiotic therapy: shorter is better. Ann Intern Med. 2019;171:210–1. 12. Babich T, Naucler P, Valik JK, Giske CG, Benito N, https://doi.org/10.7326/M19-1509. Cardona R, et al. Risk factors for mortality among patients with Pseudomonas aeruginosa bacteraemia: a 19. Bassetti M, Vena A, Croxatto A, Righi E, Guery B. retrospective multicentre study. Int J Antimicrob How to manage Pseudomonas aeruginosa infections. Agents. 2019. https://doi.org/10.1016/j.ijantimicag. Drugs Context. 2018;7: 212527. https://doi.org/10. 2019.11.004. 7573/dic.212527. 13. Herrera S, Bodro M, Soriano A. Predictors of mul- 20. Rac H, Gould AP, Bookstaver PB, Justo JA, Kohn J, tidrug resistant Pseudomonas aeruginosa involve- Al-Hasan MN. Evaluation of early clinical failure ment in bloodstream infections. Curr Opin Infect criteria for Gram-negative bloodstream infections. Dis. 2021;34:686–92. https://doi.org/10.1097/QCO. Clin Microbiol Infect. 2020;26:73–7. https://doi. 0000000000000768. org/10.1016/j.cmi.2019.05.017. 14 Kanj S, Sexston D. Pseudomonas aeruginosa bac- teremia and endocarditis. Waltham: Up To Date; 2021. Publisher’s Note 15. Bae M, Jeong Y, Bae S, Kim MJ, Chong YP, Kim S-H, Springer Nature remains neutral with regard to jurisdic- et al. Short versus prolonged courses of antimicro- tional claims in published maps and institutional bial therapy for patients with uncomplicated Pseu- affiliations. domonas aeruginosa bloodstream infection: a retrospective study. J Antimicrob Chemother. 2021. https://doi.org/10.1093/jac/dkab358.