Sepsis is a life-threatening organ dysfunction that is a result of the dysfunction of the response to infection of the host. It remains a serious and costly complication after surgery, such as minimal-invasive urological operations like ureteroscopic lithotripsy (URSL) (Singer et al., 2016; Bonkat et al., 2019). Even though URSL is regarded as minimally invasive, there are other factors including high-pressure irrigation, manipulation of instruments and the existence of infectious stones that might cause bacteremia or endotoxin release which can eventually result to sepsis (Wagenlehner et al., 2017; Pietropaolo et al., 2021). Early recognition of the high-risk patients is extremely important to timely implementation of intervention strategies and enhancement of treatment results. This mainly relies on the precision of the risk prediction devices (Laih et al., 2022; Dai et al., 2025).

Existing initiatives aim to develop clinical predicting models that would take into account patient demographics, comorbidities, rountine laboratory results and relevant factors during surgery (Hu et al., 2018; Sugizaki et al., 2025). Nevertheless, these classical indicators have a tendency to be not sensitive or specific enough since they cannot fully represent the essence of sepsis pathophysiologic characteristic of systemic immune inflammatory imbalance (Gao et al., 2025; Saavedra-Torres et al., 2025; Valsamaki et al., 2025). Strong and unbalanced immune response, including the activation, migration and functional modifications of neutrophils, lymphocytes, and monocytes and the formation of a cytokine storm characterizes sepsis, along with severe infection (Bonavia et al., 2023; Tian et al., 2025). Thus, theoretically, combined biomarkers that reflect this systemic immune-inflammatory condition directly might offer more predictive value than conventional markers.

Over the past several years, the combined immune-inflammatory markers of standard peripheral blood cell counts have attracted significant attention because of their low cost-effectiveness and simplicity of access, as well as the abundance of pathophysiological information. For instance, the Systemic Immune-Inflammation Index (SII) (Chen et al., 2025; Chung et al., 2025), Systemic Inflammation Response Index (SIRI) (Agar et al., 2025; Li et al., 2025), and Pan-Immune Inflammation Value (PIV) (Liu et al., 2025; Onal Kalkan et al., 2025) are strongly prognostically valuable both in oncology and in infectious diseases. These markers comprehensively reflect the status of the pro-inflammatory state (neutrophil, monocyte), the state of immunosuppression (lymphocyte reduction), and host responses (platelet), and therefore they offer a broad view of the immune-inflammation network. Nevertheless, the predictive performance of these indicators in specific situations following the occurrence of sepsis, especially when combined with existing clinical models, has not yet been systematically compared and verified.

The novel index, Systemic Immune-Inflammatory Complex Index (SIICI), is calculated as (Neutrophil count × Monocyte count × 1000)/(Platelet count × Lymphocyte count) (Wang et al., 2025). The formula purposefully puts myeloid derived inflammatory cells (neutrophils and monocytes) at the numerator, indicating the hyperinflammatory state (Cheng et al., 2025; Duan et al., 2025), and lymphocytes and platelets at the denominator, indicating their function as the immune competencies and thrombotic regulation respectively (Cilloniz et al., 2021). This integrated ratio is hypothesized to capture the concurrent “myeloid overactivation” and “lymphocyte exhaustion/immunoparalysis” that characterize sepsis, while also incorporating the role of platelets in sepsis-associated immunothrombosis. (Venkata et al., 2013). Hence, SIICI might provide a more comprehensive picture of the dysregulated host responses than the indices which reflect some aspects of these pathways only.

Therefore, this study aims to rigorously evaluate the incremental value of four immune-inflammatory parameters (one of them is SIICI) towards improving the baseline clinical model performance in the prediction of sepsis after URSL. Beyond improvements in model discrimination (AUC) (Hanley and McNeil, 1982), we will use net reclassification improvement (NRI), integrated discrimination improvement (Pencina et al., 2008) as well as likelihood ratio tests and decision curve analysis (DCA) (Vickers and Elkin, 2006)to judge both statistical significance and clinical utility. Besides, we pre-specified subgroup analyses according to preoperative urine culture status in order to clarify the applicability of this model to different patient populations defined in regard to microbiological evidence. The purpose of this analysis is to find out if SIICI predictive utility is conditioned by the presence of culturable bacteriuria or, whether it remains strong by capturing the intrinsic nature of the host immune-inflammatory susceptibility, regardless of the absence of conventional infection signs. We predict that SIICI will show the greatest incremental predictive ability. Finally, based on the optimal model, we will develop and validate a visual nomogram (Balachandran et al., 2015) and an online calculator (Ahn et al., 2025), which will be helpful to clinicians as a powerful and user-friendly tool to classify and predict risks of post-URSL sepsis at an early stage.

In this retrospective cohort study, we systematically developed and validated a preoperative predictive model for sepsis following ureteroscopic lithotripsy (URSL), with a core focus on evaluating the incremental predictive value of a novel systemic immune-inflammatory complex index (SIICI) relative to established immune-inflammatory indices (SII, SIRI, PIV). The key finding of our study is that the integration of SIICI into a baseline clinical model significantly enhanced predictive performance across multiple metrics (discrimination, reclassification, clinical utility), outperforming SII, SIRI, and PIV by a substantial margin. We further developed and validated a visual nomogram and a user-friendly online calculator based on the Base + SIICI model, which exhibits robust calibration and discrimination (AUC = 0.863, bootstrap-corrected AUC = 0.849) and maintains excellent performance even in patients with negative preoperative urine cultures. These findings not only identify SIICI as a clinically valuable biomarker for post-URSL sepsis but also provide a practical, individualized risk stratification tool for urologists, addressing the unmet clinical need for accurate prediction of this life-threatening complication in the context of minimally invasive urological surgery.

A critical mechanistic explanation for the superior performance of SIICI lies in its unique formula design, which is precisely tailored to capture the core pathophysiological characteristics of sepsis: concurrent hyperinflammation and immune suppression, combined with immunothrombosis. Sepsis is defined by a dysregulated host immune response to infection, characterized by the overactivation of myeloid-derived inflammatory cells (neutrophils and monocytes) that drive a pro-inflammatory cytokine storm, alongside the exhaustion of lymphocytes that mediate adaptive immunity and the dysfunction of platelets that regulate thrombosis and immune responses (Gao et al., 2025; Valsamaki et al., 2025). Unlike existing indices, SIICI is calculated as (Neutrophil count × Monocyte count × 1000)/(Platelet count × Lymphocyte count), which intentionally places hyperactivated myeloid cells in the numerator and immune-competent/thromboregulatory cells (lymphocytes, platelets) in the denominator. This design allows SIICI to quantitatively integrate three interrelated pathological processes of sepsis in a single index: myeloid overactivation, lymphocytic exhaustion, and platelet dysfunction (Samuelsen et al., 2024). In contrast, SII only incorporates neutrophils, platelets, and lymphocytes (excluding monocytes, a key mediator of the pro-inflammatory response), SIRI omits platelets (a critical player in sepsis-associated immunothrombosis), and PIV simply multiplies all four cell types without distinguishing their opposing roles in sepsis pathophysiology (Agar et al., 2025; Chung et al., 2025; Onal Kalkan et al., 2025).Our baseline data (Table 1) further support this mechanistic rationale: the sepsis group exhibited marked elevations in neutrophils and monocytes, and significant reductions in lymphocytes and platelets-changes that directly amplify the SIICI value and make it a more sensitive and specific reflection of sepsis-related immune-inflammatory dysregulation than the other indices. This mechanistic alignment between SIICI and sepsis pathophysiology is the fundamental reason for its superior incremental predictive value, and it also explains why SIICI can capture subtle immune-inflammatory perturbations that are not detected by single biomarkers or traditional composite indices.

The comprehensive statistical evaluation of SIICI further confirms its robust predictive value, going beyond simple improvements in AUC to demonstrate meaningful clinical utility. The Base + SIICI model achieved a significant NRI (0.133, p=0.001) and IDI (0.058, p=0.002), indicating that SIICI not only enhances the model’s ability to distinguish between sepsis and non-sepsis patients but also improves the accuracy of individual risk stratification—a critical outcome for clinical practice, as it allows urologists to more precisely identify patients who require proactive intervention rather than just distinguishing high- and low-risk groups. The non-significant Hosmer-Lemeshow test (p=0.662) and well-calibrated calibration curve confirm that the model’s predicted risk probabilities closely align with actual sepsis incidence across the entire risk spectrum, eliminating systematic miscalibration that plagues many clinical prediction models and ensuring reliable risk estimates for individual patients. Most importantly, decision curve analysis (DCA) and clinical impact curve analysis demonstrated that the Base + SIICI model provides a higher net clinical benefit across a wide range of probability thresholds and the highest efficiency ratio (0.291) among all models. This means that the SIICI-enhanced model minimizes unnecessary interventions while maximizing the identification of true high-risk patients—a key consideration in clinical practice, where over-stratification leads to avoidable medical costs and patient anxiety, and under-stratification results in missed opportunities for preemptive management of sepsis.

A pivotal and novel finding of our study is the superior discriminative ability of the SIICI-based model in patients with negative preoperative urine cultures (AUC = 0.850), which even exceeds its performance in culture-positive patients (AUC = 0.807). This finding extends the traditional clinical paradigm, which has relied on microbiological evidence (e.g., positive urine culture), by providing an additional means to identify patients at risk of post-URSL sepsis. Conventional urine culture only detects culturable bacteria, and negative results do not rule out the presence of non-culturable pathogens, bacterial biofilms, or endotoxin release—all of which are common in patients with ureteral stones and can trigger sepsis by inducing host immune-inflammatory dysregulation (Dai et al., 2025; Sugizaki et al., 2025). Our results strongly suggest that SIICI is not merely a surrogate for bacterial burden but a sensitive measure of the host’s intrinsic immune-inflammatory susceptibility. In urine culture-negative patients, the risk of post-URSL sepsis is driven not by overt bacterial infection but by pre-existing immune-inflammatory dysregulation that renders the host vulnerable to infection-induced organ dysfunction; SIICI quantifies this dysregulation and thus identifies the “hidden high-risk host” that would otherwise be missed by standard microbiological and clinical assessments. This is a critical clinical advance, as approximately 88.3% of patients in our cohort had negative preoperative urine cultures (Table 1)—a proportion consistent with real-world clinical practice—and our model enables targeted risk stratification and proactive management for this large, previously under-recognized high-risk subgroup.

When contextualizing our findings with previous research on post-URSL sepsis prediction, the novelty and advantages of our study become clear. Prior studies have developed prediction models for post-URSL sepsis based on clinical factors, single inflammatory biomarkers, or machine learning algorithms (Hu et al., 2018; Pietropaolo et al., 2021; Dai et al., 2025; Sugizaki et al., 2025). For example, Pietropaolo et al. developed a machine learning model for post-URSL urosepsis requiring ICU admission, but it relied on a multicenter European cohort and lacked validation in Asian populations (Pietropaolo et al., 2021); Hu et al. developed a nomogram for ureteral calculi-associated urosepsis, but it did not incorporate composite immune-inflammatory indices (Hu et al., 2018). In contrast, our study is the first to evaluate the incremental value of SIICI in post-URSL sepsis prediction and to directly compare it with three widely used immune-inflammatory indices (SII, SIRI, PIV) in a single-center Asian cohort with rigorous statistical methods (AUC, NRI, IDI, LRT, DCA). Our SIICI-based nomogram also has the advantage of simplicity and clinical feasibility: all variables in the model (CRP, albumin, hydronephrosis, positive leukocyte esterase, positive urine culture, SIICI) are readily available from routine preoperative clinical and laboratory assessments, without the need for specialized testing or complex machine learning algorithms. The development of an online calculator (https://sepsisriskprediction.shinyapps.io/DynNom/) further bridges the gap between research and clinical practice, allowing urologists to obtain real-time, individualized sepsis risk estimates at the point of care—an essential feature for translating predictive models into daily clinical practice (Ahn et al., 2025).

The clinical translational potential of our SIICI-based nomogram and online calculator is substantial, with direct applications in preoperative risk stratification and personalized patient management for URSL. For urologists, the model allows for quantitative, evidence-based preoperative counseling of patients, enabling clear communication of sepsis risk and shared decision-making regarding surgical planning (e.g., timing of URSL, preoperative prophylactic antibiotics, postoperative monitoring intensity). For high-risk patients identified by the model (e.g., those with high SIICI values, hydronephrosis, and positive leukocyte esterase), proactive interventions such as extended preoperative antibiotic prophylaxis, intraoperative low-pressure irrigation, and postoperative intensive care unit monitoring can be implemented to mitigate sepsis risk—interventions that have been shown to reduce the incidence and severity of post-URSL sepsis (Wagenlehner et al., 2017; Bonkat et al., 2019). For low-risk patients, the model avoids unnecessary interventions and reduces healthcare costs, aligning with the principles of value-based medicine. Beyond URSL, our findings also raise the possibility that SIICI may be a valuable predictive biomarker for sepsis in other surgical and clinical settings (e.g., percutaneous nephrolithotomy, abdominal surgery), where immune-inflammatory dysregulation is a key driver of sepsis; future research can explore this broader applicability.

Several limitations of our study warrant acknowledgment. First, its single-center, retrospective design inherently carries risks of selection bias and limits the generalizability of our findings. External validation in multi-center, prospective cohorts is essential to confirm the robustness and transportability of our model. Second, while our model demonstrates strong predictive performance, the absolute number of sepsis events was relatively small, which may affect the stability of the estimates. Larger studies are needed to refine the model coefficients. Third, our study is subject to the inherent challenge of class imbalance, with only 57 sepsis events (7.1%) among 803 patients. While this proportion reflects the true, low incidence of post-URSL sepsis in clinical practice, it may affect the precision and stability of the logistic regression coefficients and increase the risk of model overfitting. Although we employed bootstrap validation and imbalance-resistant metrics (AUC, NRI, IDI), future studies with larger, multicenter cohorts or those employing advanced sampling techniques (e.g., oversampling) are warranted to further validate and refine the model. Fourth, our model is based on logistic regression. While this provides high clinical interpretability, we acknowledge that other advanced machine learning algorithms were not compared in this study. Future research could involve head-to-head comparisons between logistic regression and various machine learning models to determine the optimal predictive framework for post-URSL sepsis. Finally, the model is based on preoperative variables; incorporating dynamic postoperative changes in laboratory values or clinical status might further improve predictive accuracy but was beyond the scope of this initial investigation.

In conclusion, this study identifies the SIICI index as a powerful and clinically valuable biomarker that significantly improves the prediction of sepsis following URSL. By comprehensively capturing the underlying immune-inflammatory dysregulation, SIICI provides incremental value over both a base clinical model and other immuno-inflammatory indices. The validated nomogram and online calculator we have developed offer a practical means to leverage this information, paving the way for improved risk stratification and personalized patient management in urological surgery. Crucially, its exceptional performance in patients with negative urine cultures highlights its unique role in identifying the “hidden high-risk host” based on immune phenotype rather than relying solely on microbiological evidence. Given the retrospective nature of the study, the requirement for informed consent was waived.