Real-world evidence from a tertiary academic center: clinical characteristics, microbiology and outcomes of oncology patients with febrile neutropenia presenting to the emergency department Authors: Markus Kleinberger, Petar Popov, Anna Pfarrhofer, Sabina Pasalic, Sandra Mayer, Daniel Kogler, Filippo Cacioppo, Thorsten Fuereder, Markus Raderer, Thomas Brodowicz, Rupert Bartsch, Gerald Prager, Matthias Preusser, Christoph Minichsdorfer Published: 2026 ABSTRACT BackgroundFebrile neutropenia (FN) remains a frequent complication of cytotoxic chemotherapy despite G-CSF prophylaxis and is associated with substantial morbidity and mortality. This study characterized FN presentations to the emergency department (ED) and identified patient-, treatment- and microbiological factors associated with patient outcome.Methods and materialThis retrospective single-centre real-world cohort study included adult patients (≥18 years) with histologically confirmed malignancies who had received at least one line of systemic therapy and presented to the ED of the Medical University of Vienna with FN between August 2016 and December 2019. Demographic, clinical, microbiological, treatment-related and outcome data were extracted from electronic medical records. Univariable and multivariable logistic regression models were used to identify variables associated with in-hospital and 3-month mortality.ResultsNinety patients with FN were analysed (median age 65 years, range 18–87, 42% female). The most common malignancies were lymphoma (22%), hematologic malignancies (11%) and lung cancer (11%). In-hospital and 3-month mortality rates were 18% and 29%, respectively. Non-survivors more frequently presented with hepatic or pulmonary metastases (both p = 0.009), palliative treatment intent (p < 0.001) and lower rates of therapeutic G-CSF administration (p = 0.013). Elevated CRP on admission was associated with in-hospital mortality (p = 0.007), whereas higher LDH levels were associated with 3-month mortality (p = 0.016). Blood culture positivity was common (33/84; 39%) and associated with higher in-hospital mortality (p = 0.007) and reduced 3-month survival (p = 0.006). In the multivariable analysis, no single factor was significantly associated with in-hospital mortality, whereas elevated LDH (aOR 1.01 per U/L, p = 0.008), positive blood culture (aOR 6.71, p = 0.010) and curative treatment intent (aOR 0.085, p = 0.036) were associated with 3-month mortality.ConclusionsFN remains associated with substantial short-term mortality, particularly in patients with palliative treatment setting, elevated inflammatory markers or positive blood cultures. Therapeutic G-CSF use showed a protective association, whereas higher CRP and LDH levels were associated with adverse outcomes.Supplementary informationThe online version contains supplementary material available at 10.1186/s12879-026-13503-3. Introduction/Background Febrile neutropenia (FN) is a common, acute and potentially life-threatening complication in patients receiving cytotoxic chemotherapy for hematologic or solid malignancies. It is defined by an oral temperature of >38.3 °C or two consecutive readings of >38.0 °C within two hours, in combination with an absolute neutrophil count (ANC) <0.5 G/L or an anticipated decline below this threshold [1]. FN represents a medical emergency that requires prompt evaluation and initiation of empiric broad-spectrum antimicrobial therapy, ideally within one hour of clinical presentation [2]. Reported incidence and mortality rates vary across studies. According to the European Society for Medical Oncology (ESMO) guidelines, FN occurs in approximately 8 per 1,000 patients (0.8%), with in-hospital mortality rates of up to 10% [3]. More recent multicentre studies - while mostly conducted in resource-limited settings - have reported higher FN rates, with an incidence of approximately 9.8%, and particularly high rates were observed in gastrointestinal cancer patients (up to 25%) receiving platinum/taxane combination chemotherapy [4]. Prophylactic administration of granulocyte colony-stimulating factors (G-CSF) has been shown to effectively reduce the risk of FN in patients receiving myelosuppressive chemotherapy [5]. The decision to initiate prophylactic G-CSF should consider both patient-related factors, such as age over 65 years and comorbidities and the myelotoxic potential of the chemotherapy regimen [3]. In contrast, the therapeutic benefit of G-CSF once FN has occurred remains controversial. While meta-analyses have demonstrated a shorter duration of neutropenia with therapeutic G-CSF use, a consistent survival advantage has not been established [6]. FN occurs most frequently during the first cycle of chemotherapy, with reported incidence rates of FN around 5% in cycle 1, 3% in cycles 2–3 and 1% in cycles 4–6. Here, 59% of FN events were reported during the first cycle, which often lead to dose reductions or treatment delays and consequently may compromise oncologic efficacy [7]. These modifications can be primary (pre-emptive dose reduction to prevent FN or other severe toxicities) or secondary (following an FN episode or other high-grade adverse events). A comprehensive understanding of both patient- and treatment-related risk factors, as well as a high level of awareness for FN, is essential not only for haematologists and oncologists but also for physicians in emergency and general medical settings. The aim of this study is to analyse the prevalence and clinical characteristics of febrile neutropenia among cancer patients presenting to the emergency department of a tertiary centre within a high-income country. A particular focus was on identifying predictors of clinical outcomes, especially in-hospital and 3-month mortality to improve early risk stratification and management of FN in the emergency setting. Material & methods Study design This retrospective, observational cohort study included haematological and solid cancer patients undergoing systemic anticancer therapy who were admitted to the Department of Emergency Medicine (ED), Medical University of Vienna with febrile neutropenia between August 2016 and December 2019. The study was approved by the institutional ethics committee (ethic committee number 1688/2020) and conducted in accordance with the Declaration of Helsinki. As a descriptive, hypothesis-generating analysis, no randomization or formal sample size calculation was performed. Patient selection and data sources Eligible patients were adults (≥18 years) with histologically confirmed malignant disease of any type who had received at least one line of systemic anticancer therapy and presented at the ED with FN. Patients were identified through institutional databases, clinical registries and the oncology documentation system of the Medical University of Vienna. The cohort represents a consecutive, single-centre, real-world population of individuals with solid or hematologic malignancies managed in a tertiary academic oncology centre. G-CSF was prescribed in routine clinical practice according to international guidelines with primary prophylactic G-CSF administered in case of overall high risk for febrile neutropenia ≥ 20% or when an intermediate-risk regimen (10–20%) is accompanied by patient-related risk factors such as age > 65 years or relevant comorbidities [3]. Therapeutic G-CSF use during FN was based on the discretion of the treating oncologist. Demographic, clinical, and treatment-related data were extracted from electronic medical records, including tumour characteristics (diagnosis, stage), systemic anticancer treatment and laboratory findings at presentation. Variables and definitions Demographic and clinical variables included age, sex, body weight, height and body mass index (BMI). Disease-related parameters comprised cancer diagnosis, tumour histology, disease stage, presence and site of metastases (pulmonary, hepatic, osseous, cerebral, or peritoneal), treatment setting (curative, palliative) and systemic therapy regimen. Treatment-related factors included the use of granulocyte colony-stimulating factors in prophylactic or therapeutic settings and chemotherapy cycle (first versus subsequent). Emergency-related parameters comprised vital signs at admission (tympanic membrane temperature, systolic and diastolic blood pressure, heart rate, shock index) and Emergency Severity Index (ESI) triage category. Hospitalization-related information included data on admission and discharge, duration of hospital stays, inpatient versus outpatient management, admission to surgical versus medical ward, and transfer to intensive care if applicable. Outcome measures included in-hospital mortality and 3-month mortality. Laboratory parameters included haemoglobin, absolute leukocyte and neutrophil counts, platelet count, serum sodium, potassium, creatinine, lactate dehydrogenase (LDH) and C-reactive protein (CRP). Additionally, blood cultures were analysed which were obtained during index hospital stay. Statistical analysis Statistical analyses were performed to evaluate demographic, clinical, treatment-related, laboratory and microbiological characteristics of patients presenting to the ED with FN. Continuous variables were summarized as medians with interquartile ranges (IQR) and categorical variables as absolute frequencies and percentages. Group comparisons were conducted using the Wilcoxon rank-sum test for continuous variables and the χ2 or Fisher’s exact test for categorical variables, as appropriate. Univariable and multivariable logistic regression analyses were applied to identify parameters associated with 3-month mortality and in-hospital mortality. Candidate variables with a p-value < 0.10 in univariable analyses were included in the multivariable models if enough events were given. In case of limited number of outcome events Firth’s penalized logistic regression was used to obtain bias-reduced and stable parameter estimates. Additional exploratory analyses were conducted to assess associations between treatment characteristics (e.g., use of prophylactic or therapeutic G-CSF, chemotherapy regimen, or treatment setting) and clinical outcomes. All statistical tests were two-sided and p-values < 0.05 were considered statistically significant. Data management, statistical analyses and visualizations were performed using R (version 4.4.0; R Foundation for Statistical Computing). Results Patient demographics Between August 2016 and December 2019, a total of 5,216 emergency department visits of patients with active malignancy were screened at the ED. Among these, 1,552 visits involved patients receiving systemic anticancer therapy at the time of emergency presentation. Among those FN was present in 90 of 1,552 visits (5.8%), forming the final study cohort (Supplementary Figure 1). Within the final cohort of 90 patients the median age was 65 years (range 18–87, IQR 52–74) and 42% were female. The most common underlying malignancies were lymphoma (22%), hematologic malignancies (11%) and lung cancer (11%), followed by sarcoma (10%), breast cancer (6%) and colorectal cancer (4%). Metastatic disease was frequent, including pulmonary (14%), hepatic (14%) and bone metastases (18%). The median BMI was 23.3 kg/m2 (IQR 20.6–26.1). G-CSF prophylaxis was applied in 29% of patients, while therapeutic G-CSF was administered in 72%. Most patients were treated with chemotherapy (63%) and 30% received chemotherapy in combination with targeted therapy. Two-thirds of patients were treated with palliative intent (68%) and one-third in a curative setting (32%). The initial treatment setting prior to occurrence of FN was outpatient in 63% and inpatient in 38%. At presentation, patients had a median temperature of 38.8 °C (IQR 38.4–39.1), systolic blood pressure of 121 mmHg (IQR 109–131) and heart rate of 110 bpm (IQR 93–127), corresponding to a median shock index of 0.96 (IQR 0.75–1.13). Most patients were triaged as ESI category 2 (55%) or ESI category 3 (41%). The median duration of hospital admission was 9 days (IQR 5–13) and the median time from last therapy to admission was 9 days (IQR 6–11). Of all patients, 11/90 (12%) were treated at a surgical ward and 4/90 (4.4%) were admitted to an ICU. All patient characteristics are displayed in Table 1.Table 1Baseline characteristics of all patients with febrile neutropenia (overall cohort)Variablen = 901Age at Admission [years]65 (52–74)Sex Male52 (58%) Female38 (42%)BMI [kg/m2]23.3 (20.6–26.1) (missing)13Cancer Type Lymphoma20 (22%) Hematologic Malignancy10 (11%) Lung Cancer10 (11%) Sarcoma9 (10%) Head and Neck Cancer6 (7%) Breast Cancer5 (6%) Prostate Cancer5 (6%) Colorectal Cancer4 (4%) Gastroesophageal Cancer3 (3%) Pancreaticobiliary Cancer3 (3%) Gynaecologic Malignancy2 (2%) Urothelial Carcinoma2 (2%) Primary CNS Malignancy1 (1%) RCC1 (1%) CUP1 (1%) Multiple malignancies3 (3%) Other5 (6%)Pulmonary Metastases13 (14%)CNS Metastases2 (2%)Hepatic Metastases13 (14%)Bone Metastases16 (18%)Peritoneal Metastases7 (8%)G-CSF Prophylaxis21 (29%) (missing)17G-CSF Therapeutic59 (72%) (missing)8Therapy Type CHT57 (63%) CHT + targeted27 (30%) ICI + CHT3 (3%) ICI1 (1%) ICI + CHT + targeted1 (1%) Targeted1 (1%)Treatment Setting palliative61 (68%) curative29 (32%)Therapy Administration outpatient40 (63%) inpatient24 (38%) (missing)26Previous Dose Reduction8 (11%) (missing)15Cycle First administration13 (15%) Consecutive administration75 (85%) (missing)2Duration of admission [days]9 (5–13)ICU admission4 (4%)Treatment at a surgical ward11 (12%) (missing)1Time therapy to admission [days]9 (6–11) (missing)6ESI Category 12 (2%) 247 (55%) 335 (41%) 41 (1%) (missing)5Temperature [°C]38.8 (38.4–39.1)Systolic BP [mmHg]121 (109–131) (missing)10Diastolic BP [mmHg]68 (60–79) (missing)10Heart Rate [bpm]110 (93–127) (missing)15Shock Index [HR/SBP]0.96 (0.75–1.13) (missing)15Haemoglobin [g/dL]9.05 (8.00–10.58)WBC [G/L]0.73 (0.37–1.23)ANC [G/L]0.55 (0.18–0.80) (missing)82Sodium [mmol/L]133 (131–136) (missing)1Potassium [mmol/L]3.9 (3.6–4.2) (missing)3Creatinine [mg/dL]0.85 (0.67–1.19) (missing)1LDH [U/L]188 (145–237)(missing)2CRP [mg/L]9.19 (4.82–20.05)In-Hospital Mortality16 (18%)Cause of Mortality Probably infection-related12 (75%) Mixed oncology-/infection-related4 (25%)3-Month Mortality Alive64 (71%) Dead26 (29%)1Median (IQR); n (%)ANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CNS = central nervous system; CRP = C-reactive protein; CUP = cancer of unknown primary; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Table 1Baseline characteristics of all patients with febrile neutropenia (overall cohort)Variablen = 901Age at Admission [years]65 (52–74)Sex Male52 (58%) Female38 (42%)BMI [kg/m2]23.3 (20.6–26.1) (missing)13Cancer Type Lymphoma20 (22%) Hematologic Malignancy10 (11%) Lung Cancer10 (11%) Sarcoma9 (10%) Head and Neck Cancer6 (7%) Breast Cancer5 (6%) Prostate Cancer5 (6%) Colorectal Cancer4 (4%) Gastroesophageal Cancer3 (3%) Pancreaticobiliary Cancer3 (3%) Gynaecologic Malignancy2 (2%) Urothelial Carcinoma2 (2%) Primary CNS Malignancy1 (1%) RCC1 (1%) CUP1 (1%) Multiple malignancies3 (3%) Other5 (6%)Pulmonary Metastases13 (14%)CNS Metastases2 (2%)Hepatic Metastases13 (14%)Bone Metastases16 (18%)Peritoneal Metastases7 (8%)G-CSF Prophylaxis21 (29%) (missing)17G-CSF Therapeutic59 (72%) (missing)8Therapy Type CHT57 (63%) CHT + targeted27 (30%) ICI + CHT3 (3%) ICI1 (1%) ICI + CHT + targeted1 (1%) Targeted1 (1%)Treatment Setting palliative61 (68%) curative29 (32%)Therapy Administration outpatient40 (63%) inpatient24 (38%) (missing)26Previous Dose Reduction8 (11%) (missing)15Cycle First administration13 (15%) Consecutive administration75 (85%) (missing)2Duration of admission [days]9 (5–13)ICU admission4 (4%)Treatment at a surgical ward11 (12%) (missing)1Time therapy to admission [days]9 (6–11) (missing)6ESI Category 12 (2%) 247 (55%) 335 (41%) 41 (1%) (missing)5Temperature [°C]38.8 (38.4–39.1)Systolic BP [mmHg]121 (109–131) (missing)10Diastolic BP [mmHg]68 (60–79) (missing)10Heart Rate [bpm]110 (93–127) (missing)15Shock Index [HR/SBP]0.96 (0.75–1.13) (missing)15Haemoglobin [g/dL]9.05 (8.00–10.58)WBC [G/L]0.73 (0.37–1.23)ANC [G/L]0.55 (0.18–0.80) (missing)82Sodium [mmol/L]133 (131–136) (missing)1Potassium [mmol/L]3.9 (3.6–4.2) (missing)3Creatinine [mg/dL]0.85 (0.67–1.19) (missing)1LDH [U/L]188 (145–237)(missing)2CRP [mg/L]9.19 (4.82–20.05)In-Hospital Mortality16 (18%)Cause of Mortality Probably infection-related12 (75%) Mixed oncology-/infection-related4 (25%)3-Month Mortality Alive64 (71%) Dead26 (29%)1Median (IQR); n (%)ANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CNS = central nervous system; CRP = C-reactive protein; CUP = cancer of unknown primary; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Baseline characteristics of all patients with febrile neutropenia (overall cohort) 1Median (IQR); n (%) ANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CNS = central nervous system; CRP = C-reactive protein; CUP = cancer of unknown primary; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Blood culture sampling and microbiological findings Among the 90 patients presenting with FN, blood cultures were obtained in 84 patients (93.3%), while 6 patients (6.7%) did not undergo blood culture sampling. Of the 84 sampled patients 33 patients (39.3%) had positive blood cultures, and 51 patients (60.7%) had negative results (see Supplementary Table 1). Among the 33 patients with bloodstream infections, 29 (87.9%) had exclusively peripheral culture positivity, 3 (9.1%) had both peripheral and central positive cultures and 1 patient (3.0%) had a central-line positive culture only. Concerning pathogen burden, a single microorganism was cultured in 24 (72.7%) of patients; two microorganisms were cultured in six patients (18.2%); and, in one patient (3.0%), three microorganisms were cultured. The median time to positivity (shortest time to positivity per patient, TTP) was 11.9 hours (IQR 8.5–18.5). Overall, the combined total number of peripheral and central cultures was a median of two sets per patient (IQR 2–3). Across all patients with positive cultures Pseudomonas aeruginosa (n = 10, 30.3%) was the most common, followed by Escherichia coli (n = 9, 27.3%), Staphylococcus epidermidis (n = 5, 15.2%), Staphylococcus aureus (n = 2, 6.1%, including one MRSA) and Staphylococcus haemolyticus (n = 2, 6.1%). Other pathogens were rare. In total, 222 blood culture samples were analysed and detailed information is depicted in Supplementary Table 2 as well as resistance pattern in Supplementary Table 3. Determinants of in-hospital mortality During the index hospitalization due to FN, 16/90 (17.8%) patients deceased. Among these causes were categorized as probably infection-related in 12 patients (75%) and as mixed oncology-/infection-related in 4 patients (25%), acknowledging the inherent overlap between infectious complications and underlying oncologic disease. Patients who died were more often treated with palliative setting (94% vs 62%, p = 0.014) and were less likely to have received therapeutic G-CSF (43% vs 78%, p = 0.018). Moreover, differences in emergency severity index distribution were observed, with more severe triage categories being overrepresented among non-survivors (p = 0.043). CRP levels on admission were higher in patients who died (17.3 mg/L vs 7.2 mg/L, p = 0.007) while other laboratory parameters did not differ significantly between groups. Those who died tended to be older (median 68.5 years, IQR 62–71 vs 64.0, IQR 50–74, p = 0.3) and predominantly male (75% vs 54%), though differences were not statistically significant. Non-survivors were more likely to have positive blood cultures (69% vs 32%, p = 0.007). None of the deceased patients were treated on a surgical ward during their hospital stay compared with 15% among survivors, although this difference did not reach statistical significance (p = 0.2, see Table 2).Table 2Baseline characteristics of patients with febrile neutropenia by in-hospital mortalityDeath during hospital-stayNo (n = 74)1Yes (n = 16)1p-value2Age at Admission [years]64 (50–74)68 (62–71)0.3Sex0.12 Male40 (54%)12 (75%) Female34 (46%)4 (25%)BMI [kg/m2]23.3 (20.3–26.0)22.9 (21.9, 27.4)0.6 (missing)13-Cancer Type0.3 Breast Cancer5 (6.8%)- Colorectal Cancer2 (2.7%)2 (13%) CUP1 (1.4%)- Gastroesophageal Cancer2 (2.7%)1 (6.3%) Gynaecologic Malignancy2 (2.7%)- Head and Neck Cancer6 (8.1%)- Hematologic Malignancy8 (11%)2 (13%) Lung Cancer6 (8.1%)4 (25%) Lymphoma18 (24%)2 (13%) Multiple malignancies1 (1.4%)2 (13%) Other5 (6.8%)- Pancreaticobiliary Cancer2 (2.7%)1 (6.3%) Primary CNS Malignancy1 (1.4%)- Prostate Cancer4 (5.4%)1 (6.3%) RCC1 (1.4%)- Sarcoma8 (11%)1 (6.3%) Urothelial Carcinoma2 (2.7%)-Pulmonary Metastases9 (12%)4 (25%)0.2CNS Metastases1 (1%)1 (6%)0.3Hepatic Metastases8 (11%)5 (31%)0.050Bone Metastases11 (15%)5 (31%)0.2Peritoneal Metastases4 (5%)3 (19%)0.10G-CSF Prophylaxis19 (32%)2 (14%)0.3 (missing)152G-CSF Therapeutic53 (78%)6 (43%)0.018 (missing)62Therapy Type0.095 CHT48 (65%)9 (56%) CHT + targeted23 (31%)4 (25%) ICI1 (1%)- ICI + CHT1 (1%)2 (13%) ICI + CHT + targeted-1 (6%) Targeted1 (1%)-Treatment Setting0.014 palliative46 (62%)15 (94%) curative28 (38%)1 (6%)Therapy Administration0.2 outpatient30 (58%)10 (83%) inpatient22 (42%)2 (17%) (missing)224Previous Dose Reduction6 (10%)2 (14%)0.6 (missing)132Cycle0.7 First administration10 (14%)3 (19%) Consecutive administration62 (86%)13 (81%) (missing)2-ESI Category0.043 11 (1%)1 (7%) 238 (54%)9 (64%) 332 (45%)3 (21%) 4-1 (7%) (missing)32Temperature [°C]38.7 (38.4–39.0)38.9 (38.7–39.4)0.13Systolic BP [mmHg]121 (111–132)112 (98–123)0.2 (missing)73Diastolic BP [mmHg]69 (60–80)66 (62–68)0.5 (missing)73Heart Rate [bpm]115 (94–127)106 (90–112)0.3 (missing)123Shock Index [HR/SBP]0.96 (0.76–1.11)0.88 (0.75–1.23)0.7 (missing)123Haemoglobin [g/dL]9.05 (8.03–10.58)9.15 (7.93–10.53)>0.9WBC [G/L]0.82 (0.37–1.28)0.66 (0.20–1.10)0.3ANC [G/L]0.55 (0.18–0.80)- (missing)6616Sodium [mmol/L]133 (131–137)133 (129–135)0.3(missing)1-Potassium [mmol/L]3.9 (3.6–4.2)4.0 (3.8–4.1)0.7 (missing)21Creatinine [mg/dL]0.85 (0.67–1.18)0.91 (0.71–1.26)0.6 (missing)1-LDH [U/L]187 (145–227)268 (149–418)0.3 (missing)11CRP [mg/L]7.24 (4.29–18.04)17.33 (9.27–30.83)0.007Positive Blood Culture22 (32%)11 (69%)0.007 none obtained6-Treatment Surgical Ward11 (15%)-0.2 (missing)1-3-Month Mortality<0.001 Alive64 (86%)- Dead10 (14%)16 (100%)1Median (IQR); n (%)2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact testANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Table 2Baseline characteristics of patients with febrile neutropenia by in-hospital mortalityDeath during hospital-stayNo (n = 74)1Yes (n = 16)1p-value2Age at Admission [years]64 (50–74)68 (62–71)0.3Sex0.12 Male40 (54%)12 (75%) Female34 (46%)4 (25%)BMI [kg/m2]23.3 (20.3–26.0)22.9 (21.9, 27.4)0.6 (missing)13-Cancer Type0.3 Breast Cancer5 (6.8%)- Colorectal Cancer2 (2.7%)2 (13%) CUP1 (1.4%)- Gastroesophageal Cancer2 (2.7%)1 (6.3%) Gynaecologic Malignancy2 (2.7%)- Head and Neck Cancer6 (8.1%)- Hematologic Malignancy8 (11%)2 (13%) Lung Cancer6 (8.1%)4 (25%) Lymphoma18 (24%)2 (13%) Multiple malignancies1 (1.4%)2 (13%) Other5 (6.8%)- Pancreaticobiliary Cancer2 (2.7%)1 (6.3%) Primary CNS Malignancy1 (1.4%)- Prostate Cancer4 (5.4%)1 (6.3%) RCC1 (1.4%)- Sarcoma8 (11%)1 (6.3%) Urothelial Carcinoma2 (2.7%)-Pulmonary Metastases9 (12%)4 (25%)0.2CNS Metastases1 (1%)1 (6%)0.3Hepatic Metastases8 (11%)5 (31%)0.050Bone Metastases11 (15%)5 (31%)0.2Peritoneal Metastases4 (5%)3 (19%)0.10G-CSF Prophylaxis19 (32%)2 (14%)0.3 (missing)152G-CSF Therapeutic53 (78%)6 (43%)0.018 (missing)62Therapy Type0.095 CHT48 (65%)9 (56%) CHT + targeted23 (31%)4 (25%) ICI1 (1%)- ICI + CHT1 (1%)2 (13%) ICI + CHT + targeted-1 (6%) Targeted1 (1%)-Treatment Setting0.014 palliative46 (62%)15 (94%) curative28 (38%)1 (6%)Therapy Administration0.2 outpatient30 (58%)10 (83%) inpatient22 (42%)2 (17%) (missing)224Previous Dose Reduction6 (10%)2 (14%)0.6 (missing)132Cycle0.7 First administration10 (14%)3 (19%) Consecutive administration62 (86%)13 (81%) (missing)2-ESI Category0.043 11 (1%)1 (7%) 238 (54%)9 (64%) 332 (45%)3 (21%) 4-1 (7%) (missing)32Temperature [°C]38.7 (38.4–39.0)38.9 (38.7–39.4)0.13Systolic BP [mmHg]121 (111–132)112 (98–123)0.2 (missing)73Diastolic BP [mmHg]69 (60–80)66 (62–68)0.5 (missing)73Heart Rate [bpm]115 (94–127)106 (90–112)0.3 (missing)123Shock Index [HR/SBP]0.96 (0.76–1.11)0.88 (0.75–1.23)0.7 (missing)123Haemoglobin [g/dL]9.05 (8.03–10.58)9.15 (7.93–10.53)>0.9WBC [G/L]0.82 (0.37–1.28)0.66 (0.20–1.10)0.3ANC [G/L]0.55 (0.18–0.80)- (missing)6616Sodium [mmol/L]133 (131–137)133 (129–135)0.3(missing)1-Potassium [mmol/L]3.9 (3.6–4.2)4.0 (3.8–4.1)0.7 (missing)21Creatinine [mg/dL]0.85 (0.67–1.18)0.91 (0.71–1.26)0.6 (missing)1-LDH [U/L]187 (145–227)268 (149–418)0.3 (missing)11CRP [mg/L]7.24 (4.29–18.04)17.33 (9.27–30.83)0.007Positive Blood Culture22 (32%)11 (69%)0.007 none obtained6-Treatment Surgical Ward11 (15%)-0.2 (missing)1-3-Month Mortality<0.001 Alive64 (86%)- Dead10 (14%)16 (100%)1Median (IQR); n (%)2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact testANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Baseline characteristics of patients with febrile neutropenia by in-hospital mortality 1Median (IQR); n (%) 2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test ANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CHT = chemotherapy; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count In univariable logistic regression, several factors were associated with increased risk of in-hospital death: A positive blood culture showed strongest associations of mortality (OR 4.60, 95%CI 1.48–16.1, p = 0.011), while use of therapeutic G-CSF use was strongly associated with better outcome (OR 0.21, 95%CI 0.06–0.70, p = 0.012). Elevated CRP was also associated with higher odds of death (OR 1.05 per mg/L, 95%CI 1.01–1.10, p = 0.020) and hepatic metastases carried a threefold risk increase (OR 3.75, 95%CI 0.99–13.5, p = 0.044). Patients treated with curative setting experienced significantly lower in-hospital mortality (OR 0.11, 95%CI 0.01–0.59, p = 0.037). Older age showed a numerical, though not statistically significant, association with higher in-hospital mortality (OR 1.02 per year, 95%CI 0.99–1.07, p = 0.18). Similarly, prophylactic G-CSF use appeared less frequent among non-survivors and was associated with a lower odds ratio for in-hospital death (OR 0.35, 95%CI 0.05–1.46, p = 0.20), although this did not reach statistical significance. Comparative analysis of OR is depicted in Fig. 1A. Estimated odds of death within index hospital stay in the respective subgroup based on single-variable logistic regression models are shown in Supplementary Figure 2A.Fig. 1Forest plots summarizing univariable logistic regression analyses of patients presenting with FN. (A) In-hospital mortality and (B) 3-month mortality in patients presenting with febrile neutropenia. Odds ratios (ORs) with 95% confidence intervals (CIs) are shown for demographic, clinical, laboratory, metastatic, microbiological, and treatment-related subgroups. Metric variables are separated by median. ORs represent true pairwise comparisons (e.g., female vs male, older vs younger, positive vs negative). CIs are plotted on a log10 scale and visually truncated to 0.01–16.0 for readability. Subgroups with insufficient variation or samples numbers for model estimation are labelled “Not estimable” Forest plots summarizing univariable logistic regression analyses of patients presenting with FN. (A) In-hospital mortality and (B) 3-month mortality in patients presenting with febrile neutropenia. Odds ratios (ORs) with 95% confidence intervals (CIs) are shown for demographic, clinical, laboratory, metastatic, microbiological, and treatment-related subgroups. Metric variables are separated by median. ORs represent true pairwise comparisons (e.g., female vs male, older vs younger, positive vs negative). CIs are plotted on a log10 scale and visually truncated to 0.01–16.0 for readability. Subgroups with insufficient variation or samples numbers for model estimation are labelled “Not estimable” In the multivariable logistic regression model including therapeutic G-CSF use, CRP levels on admission and positive blood cultures, none of the parameters reached statistical significance, although each demonstrated trends. Therapeutic G-CSF showed a non-significant protective trend (aOR 0.34, 95%CI 0.09–1.31, p = 0.112), whereas a positive blood culture suggested increased odds of death (aOR 2.81, 95%CI 0.77–10.9, p = 0.120). Higher CRP levels showed numerically increased but non-significant trend towards increased mortality risk (aOR 1.03 per mg/L, 95%CI 0.99–1.09, p = 0.180). Given the limited number of in-hospital deaths (n = 16), the multivariable logistic regression model was intentionally restricted to the three clinically most relevant variables (therapeutic G-CSF, CRP and positive blood cultures) to avoid overfitting. To further account for limited data and potential separation, a penalized logistic regression using Firth’s method was performed, additionally including treatment setting. In this model, therapeutic G-CSF use showed lower odds of in-hospital mortality (OR 0.46, 95%CI 0.10–1.88), as did curative treatment intent (OR 0.25, 95%CI 0.05–1.20). Higher CRP levels were associated with increased odds of in-hospital mortality (OR 1.04 per mg/L, 95%CI 1.00–1.09), as were positive blood cultures (OR 2.31, 95%CI 0.50–9.71). Although no association reached statistical significance in the penalized model, the direction and magnitude of effect estimates were consistent with the standard logistic regression results. Results of univariable and multivariable analysis are shown in Supplementary Tables 4 & 5, respectively. Predictors of three-month mortality following febrile neutropenia At three months after admission, 26 patients (28.9%) had died. Compared to survivors, non-survivors were more likely to present with pulmonary (31% vs 8%) and hepatic metastases (31% vs 8%, p = 0.009 for both) and were almost exclusively treated within a palliative setting (96% vs 56%, p < 0.001). Moreover, they were less likely to have received therapeutic G-CSF (52% vs 80%, p = 0.013). Laboratory values on admission also differed, with higher LDH concentrations (200 vs 183 U/L, p = 0.043) and CRP levels (14.8 vs 6.6 mg/L, p = 0.016) observed among non-survivors. Non-survivors showed significantly higher rate of positive blood cultures (63% vs 30%, p = 0.006) Furthermore, treatment on a surgical ward was less common among patients who died (3.8% vs 16%), although this difference did not reach statistical significance (p = 0.2, see Table 3).Table 3Baseline characteristics of patients with febrile neutropenia by 3-month mortality3-Month-Mortality after FNAlive (n = 64)1Dead (n = 26)1p-value2Age at Admission [years]64 (49–74)68 (61–71)0.4Sex0.4 Male35 (55%)17 (65%) Female29 (45%)9 (35%)BMI [kg/m2]23.6 (20.5–25.9)22.5 (20.9–26.9)0.9 (missing)121Cancer Type0.12 Lymphoma17 (27%)3 (12%) Sarcoma7 (11%)2 (8%) Hematologic Malignancy6 (9%)4 (15%) Breast Cancer5 (8%)- Head and Neck Cancer4 (6%)2 (8%) Lung Cancer4 (6%)6 (23%) Prostate Cancer4 (6%)1 (4%) Colorectal Cancer2 (3%)2 (8%) Gynaecologic Malignancy2 (3%)- Pancreaticobiliary Cancer2 (3%)1 (4%) Urothelial Carcinoma2 (3%)- CUP1 (2%)- Gastroesophageal Cancer1 (2%)2 (8%) Primary CNS Malignancy1 (2%)- RCC-1 (4%) Multiple malignancies1 (2%)2 (8%) Other5 (8%)-Pulmonary Metastases5 (8%)8 (31%)0.009CNS Metastases-2 (8%)0.081Hepatic Metastases5 (8%)8 (31%)0.009Bone Metastases9 (14%)7 (27%)0.2Peritoneal Metastases4 (6%)3 (12%)0.4G-CSF Prophylaxis16 (31%)5 (24%)0.6 (missing)125G-CSF Therapeutic47 (80%)12 (52%)0.013 (missing)53Therapy Type0.003 CHT43 (67%)14 (54%) CHT + targeted21 (33%)6 (23%) ICI-1 (4%) ICI + CHT-3 (12%) ICI + CHT + targeted-1 (4%) Targeted-1 (4%)Treatment Setting<0.001 palliative36 (56%)25 (96%) curative28 (44%)1 (4%)Therapy Administration0.051 outpatient24 (55%)16 (80%) inpatient20 (45%)4 (20%) (missing)206Previous Dose Reduction6 (12%)2 (9%)>0.9 (missing)123Cycle0.5 First administration8 (13%)5 (19%) Consecutive administration54 (87%)21 (81%) (missing)2-ESI Category0.2 11 (2%)1 (4%) 232 (52%)15 (63%) 328 (46%)7 (29%) 4-1 (4%) (missing)32Temperature [°C]38.8 (38.4–39.0)38.8 (38.4–39.3)0.6Systolic BP [mmHg]120 (109–131)121 (107–133)>0.9 (missing)73Diastolic BP [mmHg]69 (60–80)66 (62–78)>0.9 (missing)73Heart Rate [bpm]117 (95–127)106 (89–123)0.2 (missing)114Shock Index [HR/SBP]1.00 (0.75–1.15)0.85 (0.77–1.06)0.6 (missing)114Haemoglobin [g/dL]9.15 (8.30–10.60)8.55 (7.70–10.48)0.3WBC [G/L]0.73 (0.37–1.29)0.78 (0.29–1.16)0.7ANC [G/L]0.30 (0.15–0.81)0.79 (0.79–0.79)>0.9 (missing)5725Sodium [mmol/L]134 (131–137)133 (131–136)0.14 (missing)1-Potassium [mmol/L]3.9 (3.6–4.1)4.0 (3.6–4.2)0.4 (missing)21Creatinine [mg/dL]0.88 (0.68–1.18)0.84 (0.68–1.23)>0.9 (missing)1-LDH [U/L]183 (142–227)200 (157–357)0.043 (missing)11CRP [mg/L]6.63 (4.17–19.03)14.77 (8.01–24.19)0.016Positive Blood Culture18 (30%)15 (63%)0.006 none obtained42Treatment Surgical Ward10 (16%)1 (3.8%)0.2 (missing)1-1Median (IQR); n (%)2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum exact testANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CI = confidence interval; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Table 3Baseline characteristics of patients with febrile neutropenia by 3-month mortality3-Month-Mortality after FNAlive (n = 64)1Dead (n = 26)1p-value2Age at Admission [years]64 (49–74)68 (61–71)0.4Sex0.4 Male35 (55%)17 (65%) Female29 (45%)9 (35%)BMI [kg/m2]23.6 (20.5–25.9)22.5 (20.9–26.9)0.9 (missing)121Cancer Type0.12 Lymphoma17 (27%)3 (12%) Sarcoma7 (11%)2 (8%) Hematologic Malignancy6 (9%)4 (15%) Breast Cancer5 (8%)- Head and Neck Cancer4 (6%)2 (8%) Lung Cancer4 (6%)6 (23%) Prostate Cancer4 (6%)1 (4%) Colorectal Cancer2 (3%)2 (8%) Gynaecologic Malignancy2 (3%)- Pancreaticobiliary Cancer2 (3%)1 (4%) Urothelial Carcinoma2 (3%)- CUP1 (2%)- Gastroesophageal Cancer1 (2%)2 (8%) Primary CNS Malignancy1 (2%)- RCC-1 (4%) Multiple malignancies1 (2%)2 (8%) Other5 (8%)-Pulmonary Metastases5 (8%)8 (31%)0.009CNS Metastases-2 (8%)0.081Hepatic Metastases5 (8%)8 (31%)0.009Bone Metastases9 (14%)7 (27%)0.2Peritoneal Metastases4 (6%)3 (12%)0.4G-CSF Prophylaxis16 (31%)5 (24%)0.6 (missing)125G-CSF Therapeutic47 (80%)12 (52%)0.013 (missing)53Therapy Type0.003 CHT43 (67%)14 (54%) CHT + targeted21 (33%)6 (23%) ICI-1 (4%) ICI + CHT-3 (12%) ICI + CHT + targeted-1 (4%) Targeted-1 (4%)Treatment Setting<0.001 palliative36 (56%)25 (96%) curative28 (44%)1 (4%)Therapy Administration0.051 outpatient24 (55%)16 (80%) inpatient20 (45%)4 (20%) (missing)206Previous Dose Reduction6 (12%)2 (9%)>0.9 (missing)123Cycle0.5 First administration8 (13%)5 (19%) Consecutive administration54 (87%)21 (81%) (missing)2-ESI Category0.2 11 (2%)1 (4%) 232 (52%)15 (63%) 328 (46%)7 (29%) 4-1 (4%) (missing)32Temperature [°C]38.8 (38.4–39.0)38.8 (38.4–39.3)0.6Systolic BP [mmHg]120 (109–131)121 (107–133)>0.9 (missing)73Diastolic BP [mmHg]69 (60–80)66 (62–78)>0.9 (missing)73Heart Rate [bpm]117 (95–127)106 (89–123)0.2 (missing)114Shock Index [HR/SBP]1.00 (0.75–1.15)0.85 (0.77–1.06)0.6 (missing)114Haemoglobin [g/dL]9.15 (8.30–10.60)8.55 (7.70–10.48)0.3WBC [G/L]0.73 (0.37–1.29)0.78 (0.29–1.16)0.7ANC [G/L]0.30 (0.15–0.81)0.79 (0.79–0.79)>0.9 (missing)5725Sodium [mmol/L]134 (131–137)133 (131–136)0.14 (missing)1-Potassium [mmol/L]3.9 (3.6–4.1)4.0 (3.6–4.2)0.4 (missing)21Creatinine [mg/dL]0.88 (0.68–1.18)0.84 (0.68–1.23)>0.9 (missing)1-LDH [U/L]183 (142–227)200 (157–357)0.043 (missing)11CRP [mg/L]6.63 (4.17–19.03)14.77 (8.01–24.19)0.016Positive Blood Culture18 (30%)15 (63%)0.006 none obtained42Treatment Surgical Ward10 (16%)1 (3.8%)0.2 (missing)1-1Median (IQR); n (%)2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum exact testANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CI = confidence interval; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count Baseline characteristics of patients with febrile neutropenia by 3-month mortality 1Median (IQR); n (%) 2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum exact test ANC = absolute neutrophil count; BMI = body mass index; BP = blood pressure; CI = confidence interval; CRP = C-reactive protein; ESI = Emergency Severity Index; FN = febrile neutropenia; G-CSF = granulocyte colony-stimulating factor; HR = heart rate; ICI = immune checkpoint inhibitor; IQR = interquartile range; LDH = lactate dehydrogenase; RCC = renal cell carcinoma; SBP = systolic blood pressure; WBC = white blood cell count In the univariable logistic regression analysis, several factors were associated with 3-month mortality. Patients treated with curative intent had markedly lower odds of death (OR 0.05, 95%CI 0.003–0.27, p = 0.005), while these odds were increased in those patients with elevated LDH levels (OR 1.01 per U/L, 95%CI 1.00–1.01, p = 0.006) or pulmonary or hepatic metastases (each OR 5.24, 95%CI 1.56–19.3, p = 0.009). A positive blood culture was also strongly associated to increased 3-month mortality (OR 3.89, 95%CI 1.47–10.9, p = 0.007). Conversely, therapeutic G-CSF administration was associated with a lower risk of death (OR 0.28, 95%CI 0.10–0.78, p = 0.015). Older age showed a numerical but non-significant association with increased 3-month mortality (OR 1.02 per year, 95%CI 0.99–1.06, p = 0.13). Similarly, prophylactic G-CSF use tended to be linked with lower odds of death at three months (OR 0.70, 95%CI 0.20–2.16, p = 0.55), although this association did not reach statistical significance. Comparative analysis of OR is depicted in Fig. 1B. Estimated odds of 3 months mortality within the respective subgroup based on single-variable logistic regression models are depicted in Supplementary Figure 2B. In the multivariable logistic regression model, higher LDH levels remained independently associated with an increased risk of 3-month mortality (aOR 1.01 per U/L, 95%CI 1.00–1.02, p = 0.008). Positive blood cultures were also strongly associated with higher mortality (aOR 6.71, 95%CI 1.72–32.4, p = 0.01). Curative treatment setting (aOR 0.085, 95%CI 0.004–0.591, p = 0.036) also showed lower odds of 3-months mortality, and a numerically lower risk was observed in patients with therapeutic G-CSF support as well (aOR 0.49, 95%CI 0.12–1.91, p = 0.299). With limited number of events (n = 26) and following the restriction to these clinically relevant predictors to minimize overfitting we conducted a penalized logistic regression using Firth’s correction. In this sensitivity analysis, the direction and magnitude of effects remained consistent with the standard multivariable model: higher LDH continued to be associated with increased mortality (penalized OR 1.008), while therapeutic G-CSF (penalized OR 0.52) and curative treatment setting (penalized OR 0.14) again showed protective associations. Positive blood cultures also remained associated with higher risk of 4-months mortality (penalized OR 5.51). The overall model fit was statistically significant (likelihood ratio test p < 0.001). Results of univariable and multivariable analysis are shown in Supplementary Tables 4 & 5, respectively. Kaplan-Meier survival analyses supported these findings (Fig. 2). Patients receiving therapeutic G-CSF showed significantly improved 3-month survival compared with those who did not (p = 0.006). Curative treatment setting was associated with superior survival relative to palliative management (p < 0.001). Elevated CRP levels and the presence of pulmonary or hepatic metastases were each linked to poorer survival (p = 0.004, p = 0.006 and p = 0.001, respectively). Furthermore, patients with a positive blood culture experienced significantly reduced 3-month survival compared with those without bloodstream infection (p = 0.003). Additional curves including age, sex, treatment at surgical ward, WBC count, peritoneal and bone metastases, LDH, creatinine and shock index are shown in Supplementary Figure 3.Fig. 2Kaplan-Meier curves for significant predictors of 3-month overall survival following febrile neutropenia. Kaplan-Meier survival analyses showing significant associations between clinical and treatment factors and 3-month mortality after FN. Shaded areas represent 95% confidence intervals. p-values are based on log-rank tests Kaplan-Meier curves for significant predictors of 3-month overall survival following febrile neutropenia. Kaplan-Meier survival analyses showing significant associations between clinical and treatment factors and 3-month mortality after FN. Shaded areas represent 95% confidence intervals. p-values are based on log-rank tests Discussion Emergencies represent an increasing challenge in modern oncology, as cancer incidence rises due to an aging population and improved life expectancy [8]. FN is among the most frequent treatment-related emergencies, with an estimated incidence of up to 50% among patients receiving cytotoxic chemotherapy [9–11]. Historically, mortality rates in FN exceeded 50% before the introduction of empiric antimicrobial therapy [12, 13]. Even today delayed initiation of appropriate antibiotics remains a key determinant of survival, each hour of delay in septic shock reduces survival by approximately 8% [2]. However, not all patients with febrile neutropenia present with septic shock and FN represents a clinically heterogeneous disease ranging from uncomplicated fever to severe sepsis and organ dysfunction. Consequently, while early antibiotic administration is universally recommended in FN, the magnitude of its impact on mortality may vary according to initial clinical severity. In this single-centre retrospective study of real-world oncology patients presenting with FN to the ED, both in-hospital and three-month mortality rates were higher than those reported in guideline-cited cohorts. The observed in-hospital mortality of 18% exceeds the approximately 10% reported in the ESMO Clinical Practice Guidelines [3]. Microbiology and blood stream infections The infection profile in FN has evolved over time, with gram-positive cocci (e.g., Staphylococcus spp., Streptococcus spp.) now predominating over gram-negative bacteria in many regions [14, 15]. In our cohort, bloodstream infections were common and associated with adverse outcomes. Among the 84 patients in whom blood cultures were obtained, 33 (39.3%) had microbiologically documented bacteriemia. Nearly all episodes involved peripheral-only positivity (87.9%) and most patients grew a single microorganism (72.7%), findings that mirror prior reports showing that FN typically yields only one positive culture set despite multiple draws and rarely observing occurrence of multiple pathogenese [4]. Time to positivity was rapid (median 11.9 h), consistent with literature demonstrating that > 90% of positive cultures in FN have a TTP < 24 h [16], underscoring the value of early microbiological diagnostics to optimally adapt antimicrobial regimen. The pathogen spectrum in our cohort was dominated by Gram-negative bacteria, most notably Pseudomonas aeruginosa (30.3%) and Escherichia coli (27.3%). This distribution is consistent with established FN epidemiology, in which Gram-negative pathogens are associated with early sepsis complications and higher mortality. The predominance of Gram-negative pathogens in our cohort likely reflects disease severity and emergency case mix rather than healthcare resource limitations and may partly explain the higher observed mortality compared with guideline-based cohorts. The predominance of Gram-negative pathogens in our cohort should be interpreted with caution. While this finding may partly reflect differences in case mix and clinical setting, the available variables do not allow a robust assessment of disease severity, and commonly used surrogates such as palliative treatment intent are of limited precision. In addition, selection effects related to the emergency department setting may have contributed to enrichment of patients with a higher likelihood of bloodstream infection. Pre-hospital antibiotic exposure, particularly in outpatient settings, may also have influenced both patient selection and microbiological yield. To provide further context, we included additional analyses describing the characteristics of patients with positive blood cultures. Importantly, blood culture positivity was strongly associated with worse clinical outcomes: patients with documented bacteraemia had significantly higher in-hospital mortality (p = 0.007) and poorer three-month survival (p = 0.006), consistent with prior evidence linking bloodstream infection in FN to increased mortality and complications, particularly among high-risk patients [17]. The relatively high rate of bloodstream infections observed in our cohort compared to commonly reported rates may have several explanations. First, our study reflects an emergency department population rather than all patients with febrile neutropenia, likely representing a clinically more severe subset with a higher pretest probability of bacteraemia. Second, blood cultures were obtained in a high proportion of patients with a low threshold and typically early at presentation, likely prior to antibiotic administration, which may have increased the diagnostic yield compared to cohorts with more selective or delayed sampling. Third, the underlying case mix, including a substantial proportion of patients with advanced disease and palliative treatment intent, may have contributed to higher observed rates of bloodstream infection. In addition, the applied definition of culture positivity may influence reported rates, particularly if typical skin commensals are included. Finally, the relatively small sample size may have contributed to variability in the observed proportion. In the context of febrile neutropenia, infections are most commonly of endogenous origin due to mucosal barrier injury and a distinct infectious focus is often not identifiable. Interpretation of microbiological findings should therefore consider this pathophysiological background [3, 18]. A more detailed characterization of infectious syndromes, antimicrobial therapy and neutropenia dynamics would further enhance clinical interpretability; however, this was limited by the retrospective design. In particular, duration of neutropenia could not be reliably determined due to variable timing of laboratory measurements, and severity grading showed limited variability within this cohort. Empiric antimicrobial therapy was initiated promptly in accordance with institutional standard operating procedures and sepsis guidelines [19], with patients typically triaged as high urgency and receiving early broad-spectrum antibiotics following blood culture sampling [20, 21]. Nevertheless, precise data on timing of antibiotic initiation were not consistently available, which represents an important limitation, as delays in antimicrobial therapy are known to be associated with increased mortality in septic patients and are a central component of current sepsis bundle recommendations. Patient- and treatment-related predictors Beyond microbiological factors, several patient- and treatment-related characteristics were associated with mortality in our cohort. Palliative treatment setting and absence of therapeutic G-CSF were independently associated with increased mortality. Importantly, palliative treatment intent likely influenced the intensity of supportive care, including decisions to withhold therapeutic G-CSF or other aggressive interventions in patients with limited prognosis. In contrast, prior chemotherapy dose reduction, the cycle of systemic therapy (first vs subsequent administration) and prophylactic G-CSF use were not significantly associated with either in-hospital or three-month mortality in our cohort. This lack of statistical significance likely reflects the limited sample size rather than a true absence of effect. Distinguishing between oncologic and infection-related drivers of mortality is clinically relevant but remains challenging in this retrospective setting, where multiple factors frequently contribute to patient outcomes. While precise attribution was not feasible, additional clinical context is provided to aid interpretation of mortality patterns. Moreover, it is plausible that patients who received prophylactic G-CSF were underrepresented in this emergency department cohort, as effective prophylaxis likely prevented the occurrence of febrile neutropenia and thus reduced the need for acute hospital presentation. Indeed, large meta-analyses have consistently demonstrated that prophylactic colony-stimulating factor administration reduces the incidence of FN and infection-related mortality by up to 50% [22, 23]. By comparison, the therapeutic benefit of G-CSF once FN has already occurred remains more controversial [6]. In our cohort, therapeutic G-CSF was administered less frequently among non-survivors, which may reflect clinical decision-making in patients with advanced disease or limited prognosis rather than a direct protective effect of the intervention. Potentially, here withholding therapeutic G-CSF was not leading to death, but rather withheld because death was imminent due to advanced disease and clinical deterioration. Accordingly, this association is likely influenced by confounding by indication and must be interpreted with caution, particularly given the small sample size and retrospective design. Prophylactic G-CSF was administered in fewer than one-third of patients in our cohort. Current guidelines recommend primary prophylaxis when the regimen-specific risk of FN is ≥20% in general, or ≥10% in patients with additional risk factors such as older age or relevant comorbidities [3, 24, 25]. In our real-world population, however, a precise comparison with these thresholds was not possible since a wide variety of systemic anticancer therapies, including complex combinations of conventional chemotherapy, targeted agents and immunotherapy were administered. Unfortunately, robust and harmonized FN-risk estimates are not consistently available, potentially underestimating the true FN rates of certain regimen. Nevertheless, the fact that FN occurred across many regimens suggests that at least a proportion of episodes might have been preventable with optimized, regimen-based risk assessment and prophylaxis. Early risk stratification using validated tools such as the Multinational Association for Supportive Care in Cancer (MASCC) score, the Clinical Index of Stable Febrile Neutropenia (CISNE) or the FEbrile Neutropenia after ChEmotherapy (FENCE) score may help to standardize decision-making, identify patients who would benefit from primary G-CSF prophylaxis and guide outpatient versus inpatient management [26, 27]. Interestingly, outcomes did not differ between patients being treated on surgical compared to medical wards, indicating that surgical services handled FN with similar effectiveness. To our knowledge, no prior studies have systematically examined ward-type-specific outcomes in FN. These findings may suggest that with adequate protocols and awareness FN can be safely and effectively managed across different inpatient specialties. However, this finding should be interpreted with caution, as ward allocation may have been influenced by underlying disease characteristics, treatment setting or metastatic burden (e.g. a higher proportion of patients with advanced or metastatic disease managed on medical oncology wards). Although no significant differences in mortality were observed, residual confounding cannot be excluded. Laboratory parameters such as CRP and LDH also carried prognostic information. Elevated CRP at presentation correlated with increased in-hospital mortality, reflecting systemic inflammation and possible sepsis [28]. Elevated LDH predicted three-month mortality, likely reflecting higher tumour burden or increased tissue turnover. Importantly, LDH may also act as a surrogate marker of underlying disease activity, particularly in hematological malignancies, and thus reflect more advanced disease stage rather than exerting an independent effect on mortality. This introduces the possibility of confounding, as the observed association between LDH and outcomes may be partly driven by the severity of the underlying cancer rather than LDH itself. The strong association between pulmonary and hepatic metastases and mortality further underscores the influence of disease stage and tumour biology on FN outcomes [29]. In addition, univariable analyses suggested heterogeneity in outcomes across tumor entities, reflecting known differences in underlying disease biology and prognosis. Patients with lymphoma and sarcoma exhibited lower in-hospital and three-month mortality compared with other malignancies - however, confidence intervals were wide, indicating limited precision and necessitating cautious interpretation. Further subdivision into more precises disease categories (e.g. lymphoma subtypes) was deliberately avoided, as smaller sample sizes would have resulted in unstable estimates and reduced statistical robustness. Older patients represent a particularly vulnerable subgroup. While age in our analysis was not significantly correlated with outcome, prior studies have identified advanced age, comorbidities and frailty as risk factors for FN-related complications and mortality [30–33]. Given that cancer incidence rises with population aging and that high-income countries are experiencing a marked demographic shift toward older populations [8], FN will increasingly contribute to emergency oncology caseloads. Since occurrence and consequently treatment of FN carries significant economic burden, consequently these will have increasing healthcare resource and financial implications. Prior studies have shown that FN-related ED visits and hospitalizations are associated with substantial costs, ranging from approximately $1,500 to $5,600 per episode, depending on country and healthcare system [11, 34]. Beyond financial burden, FN contributes to chemotherapy dose reductions and treatment delays, which can compromise long-term oncologic outcomes [10]. Limitations This study has several limitations that should be acknowledged. First, its retrospective design inherently carries the risk of selection bias, uncontrolled confounding, and incomplete documentation. Detailed data on the timing of antibiotic administration and the appropriateness of the initial empiric antimicrobial regimen were not consistently available due to the retrospective design. Given the well-established impact of early and adequate antibiotic therapy on FN outcomes, the inability to account for time-to-antibiotics and regimen adequacy represents an important limitation and may have influenced observed mortality rates. Nevertheless, key treatment-related parameters were systematically assessed and included in the statistical analysis. Second, the study was conducted at a single tertiary academic centre, which may limit generalizability to non-tertiary or community settings. However, the uniformity of clinical documentation, consistent diagnostic workup and centralized patient management represent important methodological strengths that enhance data reliability. Furthermore, established risk scores such as the MASCC score could not be reliably reconstructed due to incomplete documentation of key components, including burden of illness and dehydration status, which limits formal risk stratification. Third, while the sample size and observation period were limited, the cohort reflects a real-world patient population, allowing a comprehensive evaluation of both solid and hematologic malignancies. The pan-cancer design, though introducing clinical heterogeneity, provides insight into shared characteristics and outcome determinants of FN across tumour types and treatment regimens. However, the relatively small sample size and number of events limited the feasibility of more granular stratified analyses according to tumour type or treatment intensity without compromising statistical robustness. Fourth, at our institution, differential blood counts are not routinely performed when total white blood cell counts are below 1 G/L, resulting in a high proportion of missing absolute neutrophil counts in patients with severe leukopenia. This may have limited the ability to further characterize the depth of neutropenia in the cohort. Fifth, this study was designed as an exploratory, hypothesis-generating analysis. Given the number of subgroup and univariable analyses performed, no formal correction for multiple testing was applied, and the results should therefore be interpreted as descriptive and exploratory rather than confirmatory. Sixth, due to the retrospective design, we could not record all comorbidities, which may contribute to the higher observed mortality compared to guideline-reported cohorts. Seventh, the study period (2016 to 2019) predates the broader implementation of newer oncologic therapies such as antibody drug conjugates and CAR T-cell therapies. In addition, supportive care strategies, including the use of G-CSF, may have evolved, particularly in the post-COVID era, which may limit the generalizability of our findings to current clinical practice. Finally, the relatively small number of events, particularly in-hospital deaths (n = 16) and three-month mortality cases (n = 26), restricted the complexity of multivariable models and may have reduced statistical power for certain associations, although this was addressed using penalized analyses. Conclusions In this real-world cohort, febrile neutropenia was associated with substantial short-term mortality, underscoring its clinical relevance as an oncologic emergency. Bloodstream infections were identified in only approximately one-third of patients but were significantly associated with both in-hospital and three-month mortality, highlighting their prognostic importance and supporting systematic blood culture collection in suspected FN. Lack of therapeutic G-CSF administration after FN occurrence was associated with worse survival - however, this finding should be interpreted cautiously given the retrospective design and potential confounding by indication. Overall, these exploratory results generate hypotheses for future prospective studies and emphasize the need to further investigate management strategies, including time-to-antibiotic administration and supportive care interventions, in acute oncology settings. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1