Clinical Characteristics and Predictors of Mortality in Elderly Patients with Bloodstream Infections: A 12-Year Retrospective Cohort Study

Abstract

Background

Bloodstream infections (BSIs) in elderly patients are associated with heightened mortality risks. This study aimed to characterize clinical profiles and identify predictors of mortality in this at-risk population.

Methods

In this 12 years retrospective cohort study, elderly patients (≥65 years) with BSI (mean age 89.9 ± 8.5 years), admitted to a tertiary geriatric hospital in Beijing, China from December 2011 to February 2024, were enrolled. Comprehensive analyses of clinical characteristics, microbiological profiles, and laboratory parameters were performed. Multivariable logistic regression analysis was performed to assess mortality-associated risk factors. Additionally, receiver operating characteristic (ROC) curve analysis was used to assess the diagnostic utility of various laboratory biomarkers.

Results

A total of 527 elderly patients (mean age 89.9 ± 8.5 years) were included in the final analysis, among whom 440 (83.5%) survived and 87 (16.5%) died. Multivariate analysis revealed that prolonged mechanical ventilation was strongly associated with the highest risk of all-cause mortality, with escalating risks observed in duration-dependent patterns:1-30 days (odds ratio [OR], 12.70; 95% confidence interval [CI], 5.86–29.00; P < 0.001) and 31–90 days (OR, 3.21; 95% CI, 1.53–6.89; P = 0.002). Additional independent risk factors for all-cause mortality included: the use of central venous catheterization (1–60 days) (OR, 2.56; 95% CI, 1.05–6.72; P = 0.045) and urinary catheterization (1–60 days) (OR, 2.28; 95% CI, 1.13–4.62; P = 0.021), chemotherapy or radiotherapy (OR, 3.25; 95% CI, 1.42–7.43; P = 0.005). ROC curve analysis demonstrated platelet count as the most discriminative prognostic marker, with an area under the curve (AUC) of 0.79, a Youden index of 0.53, sensitivity of 58.3%, and specificity of 94.2%.

Conclusions

Prolonged invasive medical interventions, including mechanical ventilation, central venous catheterization, and urinary catheterization, as well as chemotherapy or radiotherapy, were independently associated with higher mortality in elderly BSI patients. These findings highlight the need for risk-stratified management strategies to optimize outcomes in these elderly patients.

Introduction

Bloodstream infection (BSI) is a life-threatening infection caused by the invasion of pathogenic microorganisms (eg, bacteria and fungi) into the bloodstream, where they release toxins and trigger systemic inflammatory responses.¹ The global incidence of BSI continues to rise, driven by widespread use of broad-spectrum antimicrobials and glucocorticoids, growing application of invasive medical procedures, and the global antimicrobial resistance (AMR) crisis, which collectively compromise host immune barriers and impair treatment efficacy.^2–4 Furthermore, BSI frequently progresses to severe complicationsincluding septic shock and multiple organ dysfunction syndrome (MODS), resulting in poor clinical outcomes and high mortality.^5,6

According to World Health Organization projections, the population aged ≥65 years is anticipated to escalate from 524 million in 2010 to approximately 1.46 billion by the year 2050,⁷ imposing unprecedented pressure on healthcare infrastructures worldwide. Elderly patients are uniquely susceptible to BSI due to immunosenescence, physiological decline, multimorbidity, frailty, and functional dependence, which together impair pathogen clearance and worsen prognosis.^8,9 Epidemiological surveillance data reveal a 2.3- to 5.1-fold increase in 30-day mortality among elderly BSI patients relative to their younger counterparts, with persistent survival disparities extending beyond 1-year follow-up.¹⁰

Although several prognostic tools including the SOFA score, qSOFA score and Charlson Comorbidity Index are widely used for risk stratification in BSI, their performance is attenuated in very elderly populations.^5,6 Atypical inflammatory responses, high baseline comorbidity burden, and frailty-related physiological reserve decline reduce the discriminative value of conventional scores in this group. Furthermore, most existing prognostic studies have focused on general adult cohorts, with limited evidence specific to hospitalized very old patients with long-term healthcare exposure.¹¹ Healthcare-associated factors such as prolonged invasive devices, which are prevalent in geriatric inpatient settings, are also under-characterized as prognostic determinants.

To address this evidence gap, we implemented a 12-year retrospective cohort study enrolling consecutive patients ≥65 years with culture-proven BSIs at a tertiary geriatric hospital. This study aimed to characterize the clinical and microbiological profiles and identify independent predictors of all-cause mortality in this high-risk population, to inform risk-stratified management strategies.

Methods

Study Design and Setting

This single-center, retrospective study included elderly patients (≥65 years) diagnosed with BSI at a tertiary geriatric hospital in Beijing, China, from December 2011 to February 2024. Written informed consent was waived for this study due to its retrospective observational design and the absence of identifiable private patient information (Figure 1). Ethical approval was granted by the Chinese PLA Hospital Ethics Committee (Approval No.: S2024-359-02).

Figure 1.

Flow chart of participants included in this study.

Patient Selection

Elderly patients aged ≥65 years with positive blood cultures during the study period were screened for eligibility. BSIs were defined as the first positive blood culture obtained at least 48 hours after admission, with no evidence of infection present at the time of admission.^12,13

A total of 925 patients with positive blood cultures were identified during the study period. After excluding 242 patients aged < 65 years, 683 eligible elderly patients with BSI remained. Among them, 156 patients were excluded due to incomplete clinical information (n=64) or missing laboratory test data (n=92), yielding a final analytic cohort of 527 participants. The participant selection process is summarized in Figure 1.

Data Collection

Data on HAIs among hospitalized patients were gathered using a real-time nosocomial infection surveillance system (RT-NISS), which also helped identify inpatients who developed BSI. Several previous studies have reported that RT-NISS can accurately collect and assess HAI-related information in patients.^14–16 Pre-infection length of hospital stay was defined as the interval from hospital admission to the date of the first positive blood culture (BSI onset). Post-infection length of hospital stay was defined as the interval from BSI onset to hospital discharge or in-hospital death. Demographic characteristics, comorbidities, and clinical parameters were extracted from electronic health records of elderly BSI patients. Pathogen data, including blood culture results and microbial profiles, were provided by the hospital’s microbiology laboratories. The present analysis was restricted to single-pathogen BSI episodes. Polymicrobial BSIs were excluded from the final analytic cohort owing to the limited sample size and incomplete clinical documentation. All microbial specimens were collected aseptically by clinical staff and processed in the hospital’s accredited microbiology laboratory according to standard culture and identification protocols. Laboratory biomarkers such as neutrophil-to-lymphocyte ratio (NLR), hemoglobin, red blood cell count (RBC), white blood cell count (WBC), platelet count, D-dimer, C-reactive protein (CRP), glucose, and albumin were recorded on the day of symptom onset. Healthcare-associated BSIs were defined per established criteria as the first positive blood culture collected ≥48 hours post-admission in patients without signs of infection at hospitalization.^12,13,17 Combination of drug was defined as the simultaneous administration of two or more antibacterial agents for the management of BSI.

Definition of Study Outcome

The outcome of this study was in-hospital mortality, defined as death from any cause during hospitalization. Whether patients died during admission was determined using the electronic medical record system. Patients who survived to hospital discharge were categorized into the survival group, and those who died before discharge were categorized into the non-survival group.

Statistical Analysis

Normally distributed continuous variables were reported as mean ± standard deviation (SD) and analyzed using Student’s t-test; non-normally distributed variables were expressed as median (interquartile range, IQR) and compared via Mann–Whitney U-test. Categorical variables were presented as frequencies (%) and assessed using χ² or Fisher’s exact tests. To identify mortality predictors in elderly BSI patients, univariate logistic regression was performed comparing survivors and non-survivors, with variables showing P < 0.1 incorporated into a multivariate logistic regression model. Multicollinearity among covariates was assessed using the variance inflation factor (VIF), with VIF < 2.0 considered to indicate no significant multicollinearity. Model calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, and model discrimination was quantified using the C-statistic. Results were reported as adjusted odds ratios (aOR) with 95% confidence intervals (CI) and corresponding P-values. Diagnostic performance of laboratory markers was evaluated using receiver operating characteristic (ROC) curves, with area under the curve (AUC), sensitivity, specificity, and Youden index (sensitivity + specificity − 1) calculated. Statistical significance was set at P < 0.05 (two-tailed). All statistical analyses were conducted using R Statistical Software (version 4.1.3).

Results

Clinical and Epidemiological Characteristics of BSI in Elderly Patients

A total of 527 elderly patients (mean age 89.9 ± 8.5 years; 97.2% male) with BSIs were analyzed. Most patients (89.4%, n=471) were admitted to internal medicine departments, with 21.8% (n=115) undergoing one surgical procedure and 13.7% (n=72) requiring ≥2 operations. Polypharmacy (concurrent use of ≥3 medications) was documented in 85.8% (n=452) of cases. The median hospital stay was 90.0 days (IQR: 64.3–96.3), comprising pre-infection (38.2 days, IQR: 17.6–62.6) and post-infection (37.4 days, IQR: 15.4–62.4) periods. Baseline characteristics are detailed in Table 1.

Table 1.

Characteristics of the Elderly Patients with Bloodstream Infections

Characteristic	Overall (n = 527)
Age (year), mean±SD	89.9±8.5
Gender, n (%)
Female	15 (2.8)
Male	512 (97.2)
Department, n (%)
Internal medicine	471 (89.4)
Surgery	56 (10.6)
Number of operations, n (%)
0	340 (64.5)
1	115 (21.8)
≥2	72 (13.7)
Combination of drug, n (%)
No	75 (14.2)
Yes	452 (85.8)
Bacteria, n (%)
Gram-negative bacteria	290 (55.0)
Gram-positive bacteria	207 (39.3)
Fungi	30 (5.7)
Length of hospital stay (median [IQR])	90.0 (64.3, 96.3)
Length of hospital stay before infection (median [IQR])	38.2 (17.6, 62.6)
Length of hospital stay after infection (median [IQR])	37.4 (15.4, 62.4)

Abbreviation: IQR, interquartile range.

Pathogen Distribution in Elderly BSI Patients

Among 527 confirmed pathogens isolated from elderly patients with BSI, gram-negative bacteria predominated (55.0%, n=290), followed by gram-positive bacteria (39.3%, n=207) and fungi (5.7%, n=30). The most prevalent isolates were Escherichia coli (18.6%), Klebsiella pneumoniae (16.7%), Enterococcus faecium (7.4%), Staphylococcus hominis (7.4%), and Pseudomonas aeruginosa (6.5%) (Table 2).

Table 2.

Distribution and Rank Order of the Reported Pathogens for Bloodstream Infections in the Elderly Patients

Pathogen	No. (%) Pathogens	Rank
Gram-negative bacteria	290 (55.0)
Escherichia coli	98 (18.6)	1
Klebsiella pneumoniae	88 (16.7)	2
Pseudomonas aeruginosa	34 (6.5)	5
Burkholderia cepacia	12 (2.3)	11
Enterobacter aerogenes	10 (1.9)	12
Serratia marcescens	9 (1.7)	13
Stenotrophomonas maltophilia	9 (1.7)	13
Acinetobacter baumannii	9 (1.7)	13
Others	21 (4.0)
Gram-positive bacteria	207 (39.3)
Enterococcus faecium	39 (7.4)	3
Staphylococcus hominis	39 (7.4)	3
Staphylococcus capitis	27 (5.1)	6
Enterococcus faecalis	25 (4.7)	7
Staphylococcus epidermidis	25 (4.7)	7
Staphylococcus aureus	19 (3.6)	9
Staphylococcus haemolyticus	9 (1.7)	13
Others	24 (4.6)
Fungi	30 (5.7)
Candida parapsilosis	17 (3.2)	10
Candida tropicalis	7 (1.3)	17
Candida albicans	6 (1.1)	18
Total	527 (100)

Comparative Analysis of Survival and Non-Survival Groups

Patients were stratified into survivors (n=440, 83.5%) and non-survivors (n=87, 16.5%). No significant intergroup differences were observed in age (survivors: 90.1 ± 8.1 vs non-survivors: 88.8 ± 10.1 years), sex, comorbidities (diabetes, hypertension, coronary disease), or procedural factors (surgeries, dialysis, transfusions) (all P > 0.05). However, non-survivors exhibited significantly prolonged mechanical ventilation duration (P < 0.001) and central venous catheterization (P < 0.05) Urinary catheterization (P < 0.05). Additionally, non-survivors had higher rates of chemotherapy/radiotherapy (19.5% vs 8.9%, P < 0.05) and combination antimicrobial therapy (P < 0.05) (Table 3).

Table 3.

Clinical Characteristics of the Survival and Non-Survival Groups of the Elderly Patients with Bloodstream Infections

Characteristic	Survival (n = 440)	Non-Survival (n = 87)	Effect Size	P value
Age (year), mean±SD	90.1±8.1	88.8±10.1	0.15	0.166
Gender, n (%)
Female	11 (2.5)	4 (4.6)	0.05	0.47
Male	429 (97.5)	83 (95.4)
Department, n (%)
Internal medicine	394 (89.5)	77 (88.5)	0.01	0.923
Surgery	46 (10.5)	10 (11.5)
Smoking, n (%)
No	270 (61.4)	52 (59.8)	0.01	0.874
Yes	170 (38.6)	35 (40.2)
Diabetes, n (%)
No	259 (58.9)	42 (48.3)	0.08	0.088
Yes	181 (41.1)	45 (51.7)
Hypertension, n (%)
No	108 (24.5)	21 (24.1)	0.00	1
Yes	332 (75.5)	66 (75.9)
Coronary disease, n (%)
No	209 (47.5)	42 (48.3)	0.01	0.988
Yes	231 (52.5)	45 (51.7)
Number of operations, n (%)
0	278 (63.2)	62 (71.3)	0.08	0.224
1	102 (23.2)	13 (14.9)
≥2	60 (13.6)	12 (13.8)
Dialysis, n (%)
No	424 (96.4)	83 (95.4)	0.02	0.903
Yes	16 (3.6)	4 (4.6)
Days of ventilator, n (%)
0	234 (53.2)	24 (27.6)	0.48	<0.001
1–30	23 (5.2)	37 (42.5)
31–90	82 (18.6)	25 (28.7)
>90	101 (23.0)	1 (1.1)
Days of central venous catheter, n (%)
0	80 (18.2)	9 (10.3)	0.35	<0.001
1–60	81 (18.4)	50 (57.5)
61–90	117 (26.6)	21 (24.1)
>90	162 (36.8)	7 (8.0)
Days of urinary catheter, n (%)
0	171 (38.9)	26 (29.9)	0.35	<0.001
1–60	58 (13.2)	42 (48.3)
61–90	76 (17.3)	14 (16.1)
>90	135 (30.7)	5 (5.7)
Chemotherapy or radiotherapy, n (%)
No	401 (91.1)	70 (80.5)	0.13	0.006
Yes	39 (8.9)	17 (19.5)
Blood transfusion, n (%)
No	273 (62.0)	48 (55.2)	0.05	0.28
Yes	167 (38.0)	39 (44.8)
Combination of drug, n (%)
No	71 (16.1)	4 (4.6)	0.12	0.008
Yes	369 (83.9)	83 (95.4)

Multivariable Predictors of All-Cause Mortality

The variables with P values <0.1 in the comparison of clinical features between the survival and non-survival groups were included in the multivariate logistic regression model. Multicollinearity among candidate covariates was evaluated using the VIF. All VIF values were below 2.0, indicating no significant multicollinearity. The Hosmer-Lemeshow goodness-of-fit test confirmed good calibration of the logistic model (P = 0.512). The C-statistic for the full multivariate prediction model was 0.81 (95% CI: 0.75–0.87), demonstrating good discriminative ability in distinguishing survivors from non-survivors. Key independent risk factors for mortality in the elderly patients with BSI included: mechanical ventilation duration, invasive device exposure (central venou catheter and urinary catheter) and oncological therapies (chemotherapy/radiotherapy) (Figure 2). Compared to elderly BSI patients without ventilation, patients with 1–30 days of ventilation and 31–90 days of ventilation had an OR of 12.70 (95% CI: 5.86–29.00, P<0.001) and 3.21 (95% CI: 1.53–6.89, P=0.002) respectively. The OR for all-cause mortality among elder BSI patients with 1–60 days of central venous catheter and 1–60 days of urinary catheter compared to patients without invasive device exposure was 2.56 (95% CI, 1.05–6.72, P=0.045) and 2.28 (95% CI, 1.13–4.62, P=0.021) respectively. Chemotherapy or radiotherapy was associated with a higher risk of all-cause mortality in the elderly patients with bloodstream infections (OR: 3.25, 95% CI: 1.42–7.43, P=0.005).

Figure 2.

Forest plot of multivariable logistic analysis of all-cause mortality in the elderly patients with bloodstream infections in the multivariable logistic regression model, the reference categories were defined as follows: no mechanical ventilation (ventilation duration = 0 days) for Days of ventilator; no central venous catheterization (central venous catheter duration = 0 days) for Days of central venous catheter; and no urinary catheterization (urinary catheter duration = 0 days) for Days of urinary catheter. All other variables were treated as binary variables, with “no” as the reference category. Odds ratios are presented with 95% confidence intervals, and exact P values are shown alongside the corresponding odds ratios.

Abbreviations: CI, confidence interval; OR, odds ratio.

Diagnostic Performance of Laboratory Biomarkers

To evaluate the diagnostic performance of laboratory indicators in elderly patients with BSIs, we conducted a ROC curve analysis. The results showed that the levels of NLR, CRP, RBC, platelet count, D-dimer, glucose, hemoglobin and albumin were significantly different between the survival and non-survial groups (Table 4 and Figure 3). Notably, the platelet count had the highest AUC at 0.793 (95% CI, 0.692–0.895), with an optimal cut-off value of 98×10⁹/L. At this cut-off, the sensitivity was 58.3% (95% CI: 47.2–68.8%) and the specificity was 94.2% (95% CI: 91.3–6.3%). The corresponding Youden index was 0.526. This performance indicates moderate discriminative value for mortality prediction (Figure 4).

Table 4.

Diagnostic Ability of Laboratory Test Indicators for Mortality in the Elderly Patients with Bloodstream Infections

Parameter	Survive (n = 440)	Non-Survival (n = 87)	P value	AUC (95% CI)	Youden Index	Sensitivity (%)	Specificity (%)	Optimal Cut-off
NLR (median [IQR])	3.2 (2.2, 5.2)	6.6 (4.1, 15.4)	<0.001	0.778 (0.688, 0.868)	0.438	73.9	69.9	4.5
Hemoglobin (median [IQR])	104.0 (92.0, 116.0)	91.0 (85.5, 100.5)	0.002	0.693 (0.581, 0.805)	0.350	82.6	52.4	102
RBC (median [IQR])	3.4 (3.0, 3.8)	2.9 (2.7, 3.3)	0.002	0.690 (0.582, 0.799)	0.344	79.2	55.3	3.3
WBC (median [IQR])	6.6 (5.3, 8.1)	8.4 (4.8, 11.6)	0.081	0.606 (0.450, 0.762)	0.310	50.0	81.0	8.4
Platelet (median [IQR])	164.0 (116.2, 217.8)	52.0 (35.5, 138.2)	<0.001	0.793 (0.692, 0.895)	0.526	58.3	94.2	98
D-dimer (median [IQR])	1.7 (1.0, 2.9)	3.5 (2.5, 8.8)	<0.001	0.771 (0.670, 0.872)	0.490	87.0	62.1	2.3
CRP (median [IQR])	1.7 (0.7, 3.9)	6.4 (2.2, 15.5)	<0.001	0.741 (0.624, 0.858)	0.428	45.8	97.0	8.0
Glucose (median [IQR])	5.8 (5.0, 7.6)	9.1 (5.9, 11.3)	0.001	0.711 (0.598, 0.825)	0.381	54.2	84.0	8.3
Albumin (median [IQR])	33.3 (31.0, 35.8)	30.4 (27.6, 32.6)	<0.001	0.735 (0.632, 0.838)	0.419	65.2	76.7	31.2

Notes: Youden index were calculated as follows: Youden index = sensitivity + specificity − 1.

Abbreviations: IQR, interquartile range; OR, odds ratio; CI, confidence interval; AUC, area under the curve; NLR, neutrophil to lymphocyte ratio; RBC, red blood cell count; WBC, white blood cell count; CRP, C-reactive protein.

Figure 3.

Comparison of biomarkers level between survival and non-survival in older patients with bloodstream infections. (A) Boxplot of NLR in survival and non-survival group. (B) Boxplot of hemoglobin in survival and non-survival group. (C) Boxplot of RBC in survival and non-survival group. (D) Boxplot of WBC in survival and non-survival group. (E) Boxplot of platelet in survival and non-survival group. (F) Boxplot of D-dimer in survival and non-survival group. (G) Boxplot of CRP in survival and non-survival group. (H) Boxplot of glucose in survival and non-survival group. (I) Boxplot of albumin in survival and non-survival group. Sample sizes were n = 440 for survivors and n = 87 for non-survivors. **P <0.01, ***P <0.001.

Abbreviations: NLR, neutrophil to lymphocyte ratio; RBC, red blood cell count; WBC, white blood cell count; CRP, C-reactive protein; NS, no significance.

Figure 4.

ROC curve analysis for biomarkers to predict the all-cause mortality in elderly patients with bloodstream infections.

Abbreviations: ROC, Receiver operating characteristic; NLR, neutrophil to lymphocyte ratio; RBC, red blood cell count; WBC, white blood cell count; CRP, C-reactive protein.

Discussion

Key Findings and Contributions

Prognostic factors for bloodstream infection have been extensively characterized in general adult populations. The principal contribution of this study is not the identification of entirely novel predictors. Instead, it provides systematic validation and quantification of established mortality-associated factors in a distinct cohort of very elderly hospitalized patients. This cohort has several notable features that distinguish it from most existing studies. It was conducted at a specialized tertiary geriatric hospital over a 12-year period. The enrolled patients had a mean age of 89.9 years, which represents one of the oldest BSI cohorts reported to date. This population remains underrepresented in most prior prognostic research. Our findings therefore extend current evidence on BSI prognosis to this high-risk and understudied age group, and provide real-world data to support risk stratification in geriatric inpatient care settings.

Pathogen Epidemiology and Demographic Trends

In this 12 years retrospective cohort study, Escherichia coli emerged as the predominant pathogen detected in elderly BSI patients. This finding is consistent with a large population-based study across all age groups in the UK, which identified Escherichia coli as the primary cause of BSI, indicating that the predominance of Escherichia coli in BSI is broadly consistent across age strata.^18,19 Also aligning with prior reports of its rising prevalence in China from 16.4% (2015) to 23.1% (2022) per the China Antimicrobial Surveillance Network.²⁰ Elderly patients are disproportionately affected by BSI, with a higher risk of infection-related mortality compared to younger patients, and accounted for approximately half of all Escherichia coli BSI cases.²¹ Given China’s rapidly aging population, prioritizing preventive strategies against E. coli BSIs in geriatric care settings is both clinically urgent and demographically imperative.

Mechanistic Insights: Invasive Devices and Mortality

This study demonstrated that mechanical ventilation, central venous catheterization and urinary catheterization were all significantly associated with increased mortality risk in elderly BSI patients (P < 0.01). This association likely arises from two underlying mechanisms. Biofilm formation on device surfaces sustains persistent bacteremia and drives secondary infections. Catheter-related disruption of mucosal barriers also enables invasion by secondary pathogens. The link between invasive medical devices and clinical outcomes is well-established, with numerous studies highlighting the increased risk of complications and poor prognosis associated with invasive manipulations.^22–24 Elderly patients, who often require invasive devices due to underlying comorbidities and complex medical conditions, are particularly vulnerable to the risks associated with these devices. The introduction of microorganisms through invasive device catheters can lead to BSI, and prolonged devices use can hinder timely sterilization and maintenance, ultimately contributing to a poorer prognosis consistent with evidence that delayed sterilization permits microbial colonization.²⁵

The gradient of mortality risk across increasing durations of mechanical ventilation requires cautious interpretation. The markedly higher odds ratio observed in the 1 to 30 day ventilation group relative to the 31 to 90 day group is consistent with survival bias. Patients who die in the acute phase of infection only accrue short durations of ventilatory support. Patients who survive the initial high-risk period are able to receive longer courses of ventilation. This pattern accounts for the apparent reduction in mortality risk with longer ventilation exposure. A similar survival bias may also affect risk estimates for prolonged catheter use. These associations should be interpreted as prognostic markers of clinical trajectory rather than direct dose-dependent causal effects.

In summary, these findings emphasize the need for strict adherence to catheter maintenance protocols and daily assessment for early device removal especially in elderly patients with BSI. Unnecessary device use should be minimized and rigorous infection prevention and control measures should be implemented to mitigate the risk of adverse outcomes. Notably, these invasive device exposures may also act as proxies for greater underlying illness severity, and their independent contribution to mortality requires further validation in prospective studies adjusted for disease severity.

Immunosuppression and Therapeutic Risks

This study found that chemotherapy/radiotherapy recipients exhibited a higher all-cause mortality risk in elderly patients with BSIs, attributable to therapy-induced neutropenia and impaired pathogen clearance. Immunosuppressive effects of these therapies may impair pathogen clearance and amplify sepsis severity. Consistent with previous literature, chemotherapy has been identified as an independent risk factor for BSI.^25,26 Additionally, patients receiving chemotherapy or radiotherapy were at a higher risk of developing severe neutropenia, which can lead to subsequent infectious complications. Notably, 58% of mortality events in this subgroup occurred within 14 days post-BSI diagnosis, suggesting a critical window for immunomodulatory interventions.^27,28

Prognostic Biomarker

Platelet count demonstrated superior prognostic performance among all tested biomarkers with an AUC of 0.793. This discriminative ability is moderate and should not be interpreted as excellent clinical prognostic utility as a standalone marker. This finding is consistent with a previous ICU-based study, where monitoring platelet-related parameters was shown to aid in the timely assessment of prognosis for patients with BSI.²⁹ The mechanism underlying this association may involve the role of platelets in modulating leukocyte function and endothelial cell activation, leading to impaired vascular integrity and an increased risk of mortality in patients with BSI.³⁰ Notably, this analysis assessed each laboratory biomarker in isolation. No direct comparison was performed against well established clinical prediction models or multivariable prognostic scores (eg, Charlson Comorbidity Index, frailty measures). The incremental prognostic value of these biomarkers beyond standard clinical scoring systems remains to be clarified in future studies. In addition, the levels of inflammatory factors, such as NLR and CRP, were found to be of clinical value in the determination of mortality in elderly patients with bloodstream infections, which is similar to the result of previous study.³¹ The inflammatory response and the immune status of the body are closely related to the disease prognosis of bloodstream infections in elderly patients. Blood indices such as NLR, WBC, and CRP reflect the level of inflammation, whereas albumin level reflects the nutritional status of the body and organ function, and therefore have some value in reflecting the poor prognosis of elderly patients with bloodstream infections.³²

Clinical Implications

These findings have several practical implications for BSI management in geriatric inpatient settings. The strong and consistent associations between invasive device exposure and higher mortality underscore the need for strict implementation of standardized care bundles for vascular catheters, urinary catheters, and mechanical ventilation. Regular daily assessment of device necessity is particularly important in this population, as it can minimize unnecessary exposure and reduce infection-related risk. The prognostic markers identified in this study, including platelet count and NLR, offer moderate discriminative value for mortality risk and can serve as adjunct tools to complement routine clinical assessment and support early risk stratification. Integrating these factors can help identify high-risk elderly patients with BSI at an early stage, allowing targeted enhanced monitoring and supportive care to be allocated to those at greatest risk of adverse outcomes. These biomarkers should always be interpreted alongside full clinical evaluation rather than used as standalone decision-making tools.

Limitations and Future Directions

Several limitations of this study should be acknowledged. First, the retrospective observational design precludes definitive causal inferences. Invasive interventions such as mechanical ventilation and catheterization may partially reflect underlying disease severity, and residual confounding from unmeasured variables cannot be fully excluded. Second, several key clinical confounders were unavailable for adjustment, including formal frailty scores, precise timing of initial antibiotic therapy, and detailed antimicrobial resistance profiles. Temporal shifts in pathogen epidemiology and treatment practices over the 12-year study period were also not accounted for, which may introduce additional confounding. Third, the single-center setting, predominantly male cohort, and very advanced mean age (89.9 years) limit the generalizability of the findings to other populations and care settings. Fourth, stratified analyses by pathogen type, infection setting, and ICU status were not performed due to limited statistical power in small subgroups. Additionally, the study was restricted to single-pathogen healthcare-associated BSIs, excluding polymicrobial and multiple-site infections, which further limits generalizability.

Future prospective multicenter studies with larger sample sizes, incorporating comprehensive geriatric assessment and granular treatment data, are warranted to validate these prognostic predictors, clarify causal pathways, and externally validate the identified risk factors in independent elderly cohorts. Developing and validating multivariable prognostic models tailored to very elderly patients with BSI, together with stratified analyses by pathogen type, infection source, and ICU status, would further improve risk stratification and clinical applicability.

Conclusions

Escherichia coli was identified as the predominant pathogen in elderly patients with BSIs, reflecting its escalating prevalence in healthcare-associated infections. Invasive interventions such as mechanical ventilation, central venous catheterization, and urinary catheterization and immunosuppressive therapies (chemotherapy/radiotherapy) were independently associated with higher mortality in elderly BSI patients. Laboratory biomarkers, particularly platelet count, demonstrated moderate discriminative value for mortality risk and could serve as adjunct tools to complement clinical assessment for early risk stratification.

These findings support two potential areas for clinical consideration and future research. First, targeted surveillance for Escherichia coli may have value in geriatric care settings. Second, routine daily assessment of invasive device necessity may help minimize unnecessary risk exposure. The prognostic utility of platelet-related biomarkers and the potential role of immunomodulatory support during cytotoxic therapies remain unvalidated and require evaluation in prospective clinical studies. This evidence establishes a reference framework for clinical risk stratification and prioritized monitoring of this vulnerable geriatric population. Further prospective research is required to determine whether interventions targeting these factors can reduce mortality in this high-risk population.

Funding Statement

This retrospective study was supported by National Clinical Research Center for Geriatric Diseases Open Project: NCRCG-PLAGH-2022012 and NCRCG-PLAGH-2023004; Beijing Natural Science Foundation: L222014 and 7242029.

Data Sharing Statement

The datasets used and analyzed during the current study are available from the corresponding author (Yingzhen Du) upon reasonable request.

Ethics Approval

This study was ethically approved by the Ethical Committee of Chinese PLA General Hospital (S2024-359-02), and was exempt from informed consent as it was a retrospective analysis and did not involve patients’ private information. The study was conducted in accordance with the Declaration of Helsinki.

Author Contributions

Guogang Xu and Yingzhen Du are senior authors who contributed equally to this manuscript. All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors have no competing interests to declare.