We conducted a multicenter retrospective cohort study, collecting and analyzing data from HIV-positive patients with fractures treated at three Chinese healthcare centers: Beijing Ditan Hospital, Changsha First Hospital, and Chengdu Public Health Medical Center. The study protocol was reviewed and approved by the Ethics Committee of Beijing Ditan Hospital, Capital Medical University. All procedures involving human participants were carried out in compliance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

The inclusion criteria for this study were as follows: (1) Patients diagnosed with HIV infection and fractures; (2) Patients who underwent surgical treatment for their fractures; (3) Patients who received follow-up care within 3 to 12 months after surgery; and (4) Patients aged between 19 and 85 years.

The exclusion criteria were: (1) Patients who declined to participate in the study (N = 3); (2) Patients with incomplete data on neutrophil, lymphocyte, and platelet counts (N = 6); (3) Patients presenting with multiple fractures (N = 11); (4) Patients with pathological or old fractures (N = 6); and (5) Patients with serious complications, such as Pneumocystis pneumonia, tuberculosis, toxoplasmosis, Candida albicans infection, Kaposi’s sarcoma, or other severe comorbidities (N = 11). The initial cohort consisted of 375 HIV-positive patients with fractures, of whom 37 were excluded based on the aforementioned criteria (Figure 1). The final study cohort comprised 338 patients.

Fasting blood samples were obtained from study participants by venipuncture, using EDTA-containing Vacutainer tubes were used for routine complete blood count testing and flow cytometry for T-cell subset measurement. Serum samples were allowed to clot at room temperature for 45 minutes before being centrifuged at 3000 g for 10 minutes. Following centrifugation, serum aliquots were stored at -80°C until further analysis.

HIV viral load in plasma was quantified using the Abbott RealTime HIV viral load assay (m2000sp, Abbott Molecular, IL, USA), with a lower detection limit of 40 copies/mL. Absolute CD4 cell counts in whole blood were determined by standard flow cytometry using a Beckman Coulter Navios instrument (Beckman, San Jose, CA, USA). Flow cytometry was used to measure CD4+ and CD8+ T-cell counts as part of routine HIV monitoring. T cell subsets were identified using fluorochrome-conjugated monoclonal antibodies procured from BD Biosciences, San Jose, CA: anti-CD4 FITC (clone RPA-T4, RRID: AB_2562052) for T helper cells and anti-CD8 AF700 (clone RPA-T8, RRID: AB_396953) for cytotoxic T cells. Cells were incubated with the antibodies for 15 minutes at room temperature in the dark, washed, and then analyzed using a Beckman Coulter Navios flow cytometer (Beckman, San Jose, CA, USA). The percentages of T helper and cytotoxic T cells were determined based on their positive surface expression of CD4 and CD8, respectively, and were reported as a proportion of the total gated lymphocyte population.

Preoperative complete blood count (CBC) parameters were obtained from peripheral venous blood samples collected within 24 hours before surgery as part of routine clinical care. Neutrophil (Neu), lymphocyte (Lym), and platelet (PLT) counts were extracted from the hospital information system and used to calculate the Systemic Immune-Inflammation Index (SII) according to the standard formula:

Importantly, SII was calculated exclusively based on routine CBC parameters and was not derived from cell sorting or flow cytometry (FACS) assays.

This calculation utilized counts of neutrophils (Neu), lymphocytes (Lym), and platelets (PLT) obtained from peripheral blood samples, reflecting the patients’ immune and inflammatory status. However, we found that the SII distribution was right-skewed (Figure 2A); after log2 transformation, the distribution converted to normal (Figure 2B).

In this study, the primary outcome variable was the occurrence of surgical site infections (SSIs). SSIs were defined and classified according to the criteria established by the American Academy of Orthopaedic Surgeons (AAOS) (Berríos-Torres et al., 2017; McLaren and Lundy, 2019). SSI was considered to be any infection related to the surgical procedure that occurred within 30 days after surgery in the absence of an implant, or within 1 year after surgery when an implant was present. Implants included, but were not limited to, fracture internal fixation systems, vertebral pedicle screws, intervertebral fusion devices, artificial intervertebral discs, and prostheses. The classification of SSIs encompassed superficial incisional infections, deep incisional infections, and organ/space infections. Superficial incisional SSIs involved only the skin and subcutaneous tissue, while deep incisional SSIs affected deeper soft tissues, and organ/space SSIs extended beyond the surgical incision to involve any part of the anatomy manipulated during the procedure.

The consistent application of this well-defined criteria across all participating centers ensured the reliable identification of SSIs and enabled an accurate evaluation of the predictive value of the Systemic Immune-Inflammation Index (SII). The diagnosis of SSIs was made by the treating surgeons based on clinical signs and symptoms, such as localized pain, tenderness, swelling, redness, warmth, or purulent discharge from the surgical site. Microbiological culture results, when available, were used to support the diagnosis and guide antimicrobial therapy.

To comprehensively assess the potential risk factors for surgical site infections (SSIs) in HIV-positive patients, we collected a wide range of clinical data, including patient demographics, HIV-related health indicators, comorbidities, and surgical characteristics. Demographic information included age (categorized as<30 and ≥30 years), sex (male and female), duration of HIV infection (<2 and ≥2 years), antiretroviral therapy (ART) status (yes or no), and presence of comorbidities such as hepatitis and diabetes mellitus. Preoperative laboratory parameters were also recorded, including hemoglobin levels (Hb, <120 and ≥120 g/L), serum albumin concentrations (<40 and ≥40 g/L), CD4+ T cell counts (<200 and ≥200 cells/µl), CD8+ T cell counts (<400 and ≥400 cells/µl), HIV viral loads (categorized as undetectable, <20, 000, and ≥20, 000 copies/ml), as well as neutrophil, lymphocyte, and platelet counts. Surgical factors were also considered, such as fracture site (upper limbs, lower limbs, or spine), fracture type (closed or open), anesthesia type (general or local), and estimated blood loss during the procedure (<400 and ≥400 ml). T lymphocyte subpopulations were quantified using flow cytometry, a technique that allows for the direct measurement of absolute CD4+ T cell counts and the determination of the relative proportion of CD4+ T cells within the total lymphocyte population. This method provides a reliable assessment of the patient’s immune status and has been widely used in HIV monitoring and research.

All clinical data were extracted from the hospital information system (HIS) medical records by trained research staff. To ensure data accuracy and reliability, each data point was independently reviewed and verified by at least two resident physicians involved in the study. In cases of discrepancies, the medical records were re-examined, and consensus was reached through discussion among the research team. By gathering such a comprehensive set of clinical data, we aimed to identify the most relevant risk factors for SSIs in our HIV-positive patient cohort and to develop a robust predictive model based on the Systemic Immune-Inflammation Index (SII) and other key variables.

In this study, all patients received prophylactic antibiotics (cefazolin or cefuroxime) intravenously during surgery and for 48 hours postoperatively, according to institutional guidelines. These first-generation cephalosporins were chosen for their broad-spectrum activity against common pathogens associated with surgical site infections (SSIs) in orthopedic procedures, particularly Gram-positive bacteria like Staphylococcus aureus and Staphylococcus epidermidis. The dosing regimen consisted of 1.5 grams administered intraoperatively, followed by 1.5 grams every 8 hours for 48 hours postoperatively. This schedule was designed to maintain therapeutic antibiotic concentrations at the surgical site throughout the perioperative period. The 48-hour duration was based on evidence showing no additional benefits in reducing SSI rates with prolonged prophylaxis, while minimizing the risk of adverse effects and antibiotic resistance. Adherence to the prophylaxis protocol was closely monitored by surgical and nursing staff, and any deviations were reviewed by the infection control team to ensure optimal patient care.

Data analysis was performed using R version 4.1.3 software (https://www.R-project.org), with a P-value less than 0.05 considered statistically significant. Continuous variables were presented as mean ± standard deviation (SD) or interquartile range (IQR), and differences between groups were assessed using one-way ANOVA or Kruskal-Wallis U test, as appropriate. Categorical variables were expressed as percentages (n, %), and differences between groups were compared using the Chi-squared test or Fisher’s exact test. The difference in SII between patients with and without SSI was assessed using an independent t-test for normally distributed data. If data were not normally distributed, a Mann-Whitney U test was performed. The variance in SII between the groups was substantial, indicating heterogeneity in inflammatory responses.

To identify potential risk factors for SSIs in the study cohort, univariate binary logistic regression models were employed, incorporating SII and other clinical variables such as HIV viral load, CD4 count, and the presence of comorbidities. This comparison was designed to evaluate whether SII provides additional discriminatory information compared with a conventional immune-status marker in HIV care, rather than to benchmark all established clinical or surgical risk factors.

To further investigate the relationship between SII and SSIs while controlling for potential confounders, we constructed a series of multivariable logistic regression models with increasing levels of adjustment. The crude model did not adjust for any covariates, while model 1 adjusted for age and gender. Model 2 additionally adjusted for hepatitis (HBV and/or HCV), diabetes, HIV viral load, duration of HIV infection, and ART status, while model 3 further included surgical site, fracture type, anesthesia type, and estimated blood loss as covariates. Finally, model 4 adjusted for all the covariates in model 3 plus serum albumin, hemoglobin, CD4, and CD8 counts.

Subgroup analyses were performed to assess the stability of the findings across different patient subsets and to explore potential interactions among various predictors. These analyses provided insights into the generalizability of the results and helped identify any effect modification by key clinical characteristics. To examine the potential nonlinear relationship between SII and SSIs, we employed restricted cubic splines (RCS) and smooth curve fitting techniques. These methods allowed for a flexible modeling of the association, capturing any potential nonlinearities or threshold effects. The smooth curve fitting was performed using generalized additive models (GAMs) with a penalized spline function to estimate the nonlinear relationship. All statistical analyses were conducted using appropriate R packages, such as ‘glm’ for logistic regression, ‘pROC’ for ROC analysis, and ‘rms’ and ‘mgcv’ for RCS and GAM fitting, respectively.

No additional cell sorting or FACS experiments were performed specifically for SII, as SII is a composite index derived from routine CBC parameters.

In our study, 338 HIV-infected patients diagnosed with fractures were included. In total, 338 HIV-infected patients with fractures were included: 226 (63.9%) from Beijing Ditan Hospital, 86 (25.4%) from Chengdu Public Health Clinical Medical Center, and 36 (10.6%) from Changsha First Hospital (Table 1). Among the 36 patients who developed SSI, 23 (63.89%) were from Beijing Ditan Hospital, 9 (25.00%) from Chengdu Public Health Clinical Medical Center, and 4 (11.11%) from Changsha First Hospital (Table 1). The distribution of SSI cases across hospitals was comparable to that of non-SSI cases, and no significant difference was observed across centers (P = 0.994).

The study flow diagram, including patient screening, exclusion criteria, and the final analytic cohort, is presented in Figure 1. 36 (10.65%) developed surgical site infections (SSIs) during the perioperative period. The SII of the group without surgical site infection was 607.5 ± 460.2, while the SII of the group with surgical site infection was 1317.8 ± 837.1; As shown in Figure 3A, SII levels were significantly higher in patients with surgical site infection than in those without surgical site infection. Upon reviewing the baseline demographics and clinical characteristics, no significant differences were noted in age and gender distribution between patients who developed SSIs and not. However, a notable distinction was observed in the Neu levels, Lym levels, HIV viral load, and CD4 counts between the two groups (Table 1).

There was a significant difference in SII between the groups with and without SSI; however, the variance within both groups was relatively large (460.2 for the non-SSI group, and 837.1 for the SSI group), suggesting considerable heterogeneity in inflammatory responses across individuals. This variability may be due to the diverse factors influencing immune function and inflammation in HIV-infected patients. As shown in Figure 2A, the preoperative SII values were right-skewed. Therefore, SII was log2-transformed to improve normality, and the transformed distribution approximated normality (Figure 2B).

The univariate logistic regression analysis identified several factors associated with an increased risk of SSIs (Table 2). These included a higher HIV viral load (odds ratio [OR] = 5.51, 95% CI = 2.3-13.2), the presence of open fractures (OR = 4.62, 95% CI = 1.65-11.9), lower albumin levels (OR = 0.4, 95% CI = 0.2-0.8), reduced CD4 counts (OR = 0.18, 95% CI = 0.08-0.37), lower CD4/CD8 ratios (OR = 0.26, 95% CI = 0.08-0.74), and higher SII values (OR = 3.09, 95% CI = 2.15-4.6). These initial findings underscored the potential relevance of various clinical parameters, including SII, in identifying patients at heightened risk for SSIs.

Furthermore, we evaluated the discriminatory performance of SII in predicting SSIs and compared it with the CD4/CD8 ratio, a commonly used immune-status marker in HIV-related clinical practice. The ROC analysis yielded an AUC of 0.810 for SII, compared with 0.689 for the CD4/CD8 ratio (Figure 3B). The difference between the two AUCs was statistically significant (Z = 2.1573, p=0.031), suggesting that SII may offer improved discrimination over the CD4/CD8 ratio in this cohort.

After performing multivariate binary logistic regression analysis (As shown in Table 3), our results indicate that a higher SII score is associated with an increased risk of developing surgery site infection. This association was significant in our crude model (OR = 3.086; 95%CI=2.149-4.602, p<0.001), model 1 (OR = 3.078; 95%CI=2.142-4.591, p<0.001), model 2 (OR = 3.247; 95%CI=2.191-5.065, p<0.001) and model 3 (OR = 3.381; 95%CI=2.235-5.409, p<0.001). In the fully adjusted model, the positive association between SII and SSI remained stable (OR = 3.276; 95%CI=2.067-5.549, p<0.001), indicating that for every unit increase in log-SII levels, the risk of developing SSI increased by 3.276-fold.

We further transformed the SII from a continuous variable into a categorical variable (lower SII vs higher SII) for sensitivity analysis (Table 3). Compared with the lower SII, the risk of developing SSI in the higher SSI increased by 9.635-fold (OR = 9.635; 95%CI=3.71-32.945, p<0.001) in the crude model and 10.8-fold (OR = 10.801; 95%CI=3.594-41.974, p<0.001). That is, patients with higher SII levels exhibited a markedly increased risk of SSIs compared to those with lower SII levels, reinforcing the value of SII in preoperative risk stratification.

Subgroup analysis provided additional insights, demonstrating the stability of the relationship between SII levels and SSI risk across different age (<30 and ≥30 years), gender (male and female), hepatitis (Yes and No), and diabetes mellitus (Yes and No), type of anesthesia (general or local), Hb (<120 and ≥120 g/L), albumin (<40 and ≥40 g/L), and CD4+ counts (<200 and ≥200 cells/ul) subgroups (Table 4).

Moreover, as shown in Figure 4, a linear correlation between SII and SSIs was observed (Figure 4A), regardless of variations in CD4 count (<200 and ≥200 cells/ul, Figure 4B) and HIV viral load (undetectable, <20, 000, and ≥20, 000 copes/mL, Figure 4C), indicating a consistent association between systemic immune inflammation and SSI risk among HIV-positive patients with fractures. This consistency across various patient demographics and clinical profiles underscores the reliability and applicability of SII as a predictive marker for SSIs.