Catheter-related bloodstream infections (CRBSIs) are a common and serious form of hospital-acquired infection,

particularly among critically ill patients in intensive care units (ICUs) and nephrology wards, and have become a major focus of infection prevention worldwide (1–3). CRBSIs usually result from microbial colonization of central venous catheters during insertion or maintenance, leading to bloodstream invasion, systemic inflammatory response, sepsis and even death (4, 5). Their clinical presentation is often insidious, and early diagnosis is challenging, increasing the risk of delayed treatment and disease progression (6). Although the implementation of aseptic techniques and catheter-care bundles has contributed to an overall decline in CRBSI incidence (7), these infections still occur in high-risk units, prolong hospital stay, increase healthcare costs and are associated with substantial short-term mortality, especially in patients with significant comorbidities (8–11).

In recent years, the rising burden of multidrug-resistant organisms (MDROs) in CRBSIs has attracted increasing attention (12). Carbapenem-resistant Acinetobacter baumannii (CR-AB) and carbapenem-resistant Klebsiella pneumoniae (CR-KP) spread rapidly in hospital settings, severely constraining therapeutic options and markedly worsening outcomes (13–16). MDROs frequently exhibit enhanced virulence and a pronounced capacity for biofilm formation; biofilms on catheter surfaces substantially increase bacterial tolerance to antibiotics and host immune clearance, rendering infections difficult to eradicate and prone to relapse, and predisposing to uncontrolled systemic infection and multi-organ failure (17–19). These features are particularly relevant in CRBSIs, in which catheter-associated biofilm serves as a reservoir for MDRO colonization and recurrent bloodstream seeding. By contrast, some studies suggest that CRBSIs caused by Gram-positive organisms may be associated with relatively more favorable outcomes at the population level (20). However, MRSA represents a notable exception: owing to its multidrug resistance and high virulence, MRSA bacteremia is consistently linked to increased mortality and cannot be considered prognostically benign (21, 22). In parallel, several epidemiological studies have reported a year-on-year increase in infections due to resistant Gram-negative bacteria, indicating a dynamic shift in the pathogen spectrum (23). This trend also applies to CRBSIs, where carbapenem-resistant Enterobacterales, Pseudomonas aeruginosa and A. baumannii are increasingly represented and now constitute a central challenge for prevention and treatment (24, 25). These dynamic changes demand timely adaptation of empirical therapy, with early coverage of high-risk resistant Gram-negative organisms in appropriately selected patients, to mitigate treatment failure and poor outcomes.

Beyond microbiological characteristics, host-related factors play a crucial role in determining CRBSI prognosis (26). Disease severity scores such as Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA) and Pitt bacteremia score are core components of prognostic assessment in severe infections, with higher scores consistently associated with increased short-term mortality (27–31). Specific MDROs, including MRSA, CR-KP and carbapenem-resistant P. aeruginosa (CRPA), have been identified as independent predictors of death in bloodstream infections (32, 33). Host-related parameters such as advanced age, hyperglycemia and hypoalbuminemia reflect baseline physiological reserve and nutritional status and have been incorporated as prognostic items in previous CRBSI-related scoring systems (34, 35). Dynamic changes in inflammatory biomarkers, including C-reactive protein (CRP) and procalcitonin (PCT), provide objective information on infection severity and treatment response (36, 37). In addition, the timeliness of key interventions—particularly catheter removal after onset of infection and the prompt initiation of targeted antimicrobial therapy according to susceptibility results—has been shown to significantly influence mortality, with delayed source control or optimization of antibiotics markedly increasing the risk of death (38, 39). Despite these insights, multivariable mortality prediction models that simultaneously incorporate disease severity, pathogen profile, host condition and intervention timing remain underdeveloped in the field of CRBSIs, and practical, user-friendly visual tools are still lacking (40).

Existing epidemiological data on CRBSIs also show marked geographical imbalance within China. Prior work has indicated that reports are disproportionately concentrated in northern, eastern and southwestern regions, whereas studies from southern China remain relatively scarce, partly due to uneven distribution of clinical research resources (41). In particular, regional data systematically linking different pathogen categories, including MDROs, to mortality risk in CRBSIs are limited.

Against this background, the present study focuses on patients with CRBSIs in a large tertiary hospital in Nanchang, southern China. We aimed to characterize the local epidemiology and pathogen distribution, to delineate the impact of different causative organisms—especially multidrug-resistant Gram-negative bacteria—on 28-day mortality, and to clarify the independent prognostic contributions of APACHE II score, cardiovascular disease (CVD) and other host-related factors. Furthermore, we sought to develop a visual, clinically applicable risk prediction model integrating multiple variables, with the goal of providing a regionally relevant tool to support early risk stratification, inform infection control strategies and guide individualized antimicrobial therapy in CRBSIs.

This study, conducted in a large tertiary referral hospital in Nanchang, analyzed 172 patients with CRBSIs, systematically characterizing their epidemiological features, microbiological profiles and independent predictors of 28-day mortality, thereby providing local evidence to support regional CRBSI management and prognostic assessment. We observed a year-on-year decline in CRBSI incidence from 2021 to 2024, a trend likely linked to the implementation of catheter care bundles, reinforced hand hygiene and antimicrobial stewardship at national and institutional levels (47). Despite the overall decline, CRBSIs remained clustered in nephrology, the ICU and gastroenterology. This pattern is consistent with previous work identifying the ICU as a primary site of CRBSI (48), while also underscoring that other catheter-intensive wards should not be overlooked. In nephrology, catheter utilization is extremely high among hemodialysis patients, with more than 85% initiating dialysis via central venous catheters and a reported CRBSI incidence of approximately 0.78/1,000 catheters (49). In the ICU, central venous catheters are almost routine for life support and hemodynamic monitoring, with a CRBSI incidence of about 9.63% (50). Patients in gastroenterology and gastrointestinal surgery similarly require long-term vascular access for chemotherapy or perioperative support, with reported CRBSI incidences of 10.82%–12.94% and 8.0%, respectively (51, 52). Together, these data extend the traditional ICU-centric view of CRBSI risk and highlight the need to prioritize strengthened catheter management strategies across all high-utilization departments.

With respect to microbiology, Gram-positive bacteria remained the predominant pathogens, accounting for more than half of all isolates, with S. aureus as the most common species. This finding is in line with most domestic studies and reinforces the central role of Gram-positive organisms in bloodstream infections (53, 54). Compared with classic CRBSI series in which CoNS predominate (55), our cohort showed a higher proportion of S. aureus and a lower proportion of CoNS. This discrepancy is plausibly explained by the strict CDC-based diagnostic criteria applied in this study, which likely excluded a larger proportion of CoNS-related blood culture contaminants and therefore yielded a pathogen spectrum more reflective of true infection. In addition, the high proportion of surgical and hemodialysis patients in our cohort, with frequent disruption of skin barriers, may favor bloodstream invasion by high-virulence S. aureus. Notably, we also observed a progressive increase in Gram-negative pathogens—particularly E. coli and K. pneumoniae—over the study period, a trend consistent with longitudinal data from large urban hospitals (56, 57). Intensive use of broad-spectrum agents, especially third-generation cephalosporins, can deplete susceptible intestinal flora and select for resistant Enterobacterales (e.g., ESBL-producing strains), promoting their colonization (58). In the context of increasing invasive procedures and a growing population of critically ill patients, more frequent disruption of physiological barriers facilitates endogenous translocation of these colonizing resistant Gram-negative bacilli into the bloodstream, thereby increasing their contribution to the CRBSI pathogen spectrum (12). In this setting, effective infection control requires parallel surveillance of both Gram-positive and Gram-negative trends to strengthen early warning systems and refine empiric antimicrobial strategies.

In this cohort, the 28-day all-cause mortality of CRBSIs was 19.77%, slightly higher than some domestic reports but still acceptable for a high-risk tertiary population. Further analysis demonstrated that the APACHE II score was the strongest independent predictor of mortality, underscoring the central role of baseline disease severity, consistent with its established prognostic value in critically ill patients (27–29). Unlike previous studies that typically evaluated APACHE II in isolation, we incorporated APACHE II, cardiovascular comorbidity and microbiological variables into a nomogram, and confirmed its predictive performance by ROC analysis (C-index = 0.979). This approach not only reaffirms the clinical utility of APACHE II, but also provides a more intuitive and practical tool for bedside risk stratification in specific infectious populations.

Beyond global severity, we identified CVD as another independent risk factor for 28-day mortality, addressing a gap in prior CRBSI prognostic models that have rarely included CVD (59). Although chronic comorbidities have been broadly associated with adverse infection outcomes (60, 61), our findings emphasize the particular impact of CVD in the context of CRBSIs: infection-related systemic inflammation and hemodynamic instability can exacerbate pre-existing cardiac dysfunction, precipitate multi-organ injury and ultimately increase mortality. This suggests that, in patients with CRBSIs and CVD, management should shift from a narrow focus on source control and antimicrobial therapy toward an integrated strategy that balances “infection control and cardiovascular protection,” with meticulous hemodynamic management, cardioprotective care and early cardiology involvement, especially in elderly patients with limited cardiac reserve.

A clinically informative observation from our multivariable analysis was that Gram-positive infection, considered as a composite category, was inversely associated with mortality and appeared as a “protective” factor. This does not imply that Gram-positive bacteremia is benign, nor that it is associated with shorter time-to-positivity or easier localization of the infectious focus. Rather, the association likely reflects differences in therapeutic accessibility and pathogen biology. Many common Gram-positive pathogens, such as methicillin-susceptible S. aureus (MSSA) and some CoNS, remain susceptible to β-lactams and glycopeptides, which are widely available and well-characterized pharmacodynamically, enabling more predictable treatment responses (62, 63). In addition, several Gram-positive species (e.g., S. epidermidis) are intrinsically less virulent, whereas Gram-negative bacilli release endotoxin that can trigger more intense systemic inflammation and septic shock, leading to more severe organ damage (64–66). Our results are therefore consistent with reports suggesting that, when recognized early and appropriately covered empirically, Gram-positive infections may be relatively more manageable (67). At the same time, substantial heterogeneity exists within the Gram-positive group: S. aureus (including MRSA and MSSA) produces potent virulence factors such as coagulase and often causes highly invasive disease (68), and MRSA’s broad β-lactam resistance increases therapeutic complexity and has been associated with worse outcomes than MSSA (69, 70). CoNS, although generally less virulent, possess strong biofilm-forming capacity and are prototypical pathogens of device-related infections (71). Owing to limited event numbers and sample size, we were unable to perform adequately powered stratified analyses comparing MRSA, MSSA and individual CoNS species; this important question warrants further investigation in larger cohorts.

The decline in MRSA detection observed in our study is likely driven by multiple concurrent interventions. Strengthened infection control measures, such as improved hand hygiene, contact precautions and environmental cleaning, can reduce nosocomial MRSA transmission, while antimicrobial stewardship programmes mitigate selection pressure by limiting unnecessary anti-staphylococcal antibiotic use. In addition, MRSA screening and decolonization strategies implemented in some institutions have been shown to reduce MRSA infection burden (72–74). The convergence of these measures may underlie the downward trend in MRSA-related CRBSIs at our center and illustrates how coordinated infection prevention and antimicrobial stewardship can reshape pathogen ecology over time.

In contrast to the relatively favorable outcomes associated with Gram-positive infections overall, CR-KP and CR-AB were clearly identified as independent risk factors for 28-day mortality, reflecting the high lethality of these MDROs. CR-KP and CR-AB frequently harbor multiple resistance mechanisms, dramatically narrowing the spectrum of effective agents, and often exhibit strong biofilm-forming capacity, facilitating persistent colonization of catheters and hospital environments and leading to protracted, difficult-to-eradicate infections (75, 76). The rising trend of CR-KP in Nanchang parallels that reported in coastal cities such as Guangzhou and Fuzhou (77), suggesting a risk of regional dissemination of resistance. In this context, our findings support a pathogen-specific management strategy: for CRBSIs caused by CR-KP or CR-AB, prompt catheter removal and thorough source control should be prioritized; individualized combination antibiotic regimens should be initiated as early as possible based on susceptibility testing; and strict contact precautions should be implemented to prevent nosocomial spread of MDROs (77–79). More broadly, our data indicate that clinicians should pay particular attention to patients with high APACHE II scores, concomitant CVD and CR-GNB infection, especially in nephrology and ICU settings where catheter density and MDRO pressure are highest.

This study has several limitations. First, this was a single-center retrospective cohort conducted in a tertiary referral hospital with a high proportion of critically ill and transferred patients; therefore, selection bias is unavoidable and the generalizability of our findings may be greatest in similar high-risk settings rather than in community hospitals or general wards. Second, although key catheter-related variables were available, more granular information (e.g., insertion site, number of lumens, catheter care bundle adherence, timing of catheter removal/replacement, and other procedural details) was not consistently recorded in the electronic charts. In addition, several time-sensitive diagnostic and therapeutic parameters, such as differential time to positivity between catheter-drawn and peripheral blood cultures, semi-quantitative catheter-tip cultures, and the timing/appropriateness of initial empirical antimicrobial therapy, were incompletely documented, which may have introduced residual confounding and potential misclassification, particularly for CoNS-associated episodes. Third, the non-survival group was relatively small (34 deaths), which limited the statistical power for exploring rare pathogens and finer subgroups (e.g., MRSA vs. MSSA vs. CoNS) and may have affected the stability of regression estimates. Finally, although the nomogram showed high discrimination (C-index = 0.979), it was only internally validated within the same cohort and may be subject to optimism and overfitting; external validation and calibration in larger multicentre prospective cohorts are required before broad clinical implementation.

This study systematically characterized the clinical features and prognostic determinants of CRBSIs, providing critical evidence for risk stratification and therapeutic decision-making. We identified the APACHE II score, underlying CVD and infection with CR-KP or CR-AB as independent predictors of 28-day mortality, and showed that, at the grouped level, Gram-positive infections were associated with comparatively more favorable outcomes. These findings translate into several concrete clinical implications: clinicians should routinely calculate APACHE II at CRBSI onset and use the proposed nomogram to identify high-risk patients early; patients with concomitant CVD warrant intensified hemodynamic monitoring and cardiology co-management; episodes caused by CR-KP or CR-AB should prompt immediate catheter removal, early initiation of active (often combination) antimicrobial regimens and strict contact precautions; and catheter care bundles together with targeted surveillance for carbapenem-resistant Gram-negative organisms should be prioritized in high-utilization wards such as nephrology and the ICU. Implementation of these measures may help to optimize antimicrobial use, focus resources on patients at greatest risk and ultimately improve the short-term prognosis of CRBSI.