A Simple Scoring System Based on CRP/Alb Ratio, Fasting Glucose, and Surgery Duration for Predicting Postoperative Infection in Older Adult Patients with Intertrochanteric Femoral Fracture: A Retrospective Cohort Study Authors: Jinyang Li, Hongmin Yuan, Yanli Niu Published: 2026 ABSTRACT BackgroundPostoperative infection is a serious complication in older adults undergoing surgery for intertrochanteric femoral fracture (IFF). This study aimed to develop a simple scoring system and nomogram for predicting postoperative infection using routinely available preoperative indicators.MethodsA retrospective analysis was conducted on 286 older adult IFF patients (25 with postoperative infection, 261 without) treated at a single tertiary hospital. Optimal cut-off values were determined using the Youden index. Univariate and multivariate logistic regression identified independent risk factors. A nomogram and a simple scoring system (0–3 points) were constructed. Model performance was assessed by AUC, calibration curve, and subgroup analyses. Bootstrap resampling (500 iterations) was used for internal validation.ResultsMultivariate analysis confirmed the CRP/albumin ratio (OR = 2582.84, 95% CI: 201.44–67,929.9, p < 0.001), fasting glucose (OR = 2.82, 95% CI: 1.32–6.73, p = 0.011), and surgery duration (OR = 1.06, 95% CI: 1.01–1.12, p = 0.020) as independent predictors. Optimal cut-offs were 1.85 (CRP/albumin), 7.6 mmol/L (glucose), and 89 minutes (surgery). The scoring system showed infection rates of 0%, 5.6%, 40.9%, and 100% for scores 0, 1, 2, and 3, respectively, with an AUC of 0.932. The full model achieved an AUC of 0.951, with Bootstrap-corrected AUC of 0.945 (95% CI: 0.923–0.951). Calibration was satisfactory (Hosmer-Lemeshow, p = 0.327). Subgroup analyses demonstrated robust performance across gender, BMI, and age groups, with AUCs ranging from 0.922 to 0.982.ConclusionIn this single-center retrospective cohort, the CRP/albumin ratio, fasting glucose, and surgery duration were independent predictors of postoperative infection in older adults with IFF. The proposed simple scoring system and nomogram provide practical tools for risk stratification; however, external validation in multicenter prospective cohorts is warranted before widespread clinical implementation. Introduction Intertrochanteric femoral fracture (IFF) is one of the most common hip fractures in the older adult population, typically resulting from low-energy trauma such as a fall from standing height.1–3 With the accelerating aging of the global population, the incidence of IFF has been steadily increasing, posing significant challenges to healthcare systems worldwide.4,5 These fractures are associated with substantial morbidity, mortality, and healthcare costs, making them a major public health concern. Surgical intervention, including internal fixation (eg, proximal femoral nail anti-rotation or dynamic hip screw) and joint replacement, is the standard treatment for IFF.6 Early surgery facilitates prompt mobilization, reduces complications associated with prolonged bed rest, and improves functional recovery.7 Despite advances in surgical techniques and perioperative care, postoperative infection remains a serious complication, occurring in approximately 5–10% of older adults IFF patients.8 Postoperative infection not only prolongs hospital stay and increases medical costs but also significantly elevates the risk of reoperation and mortality.9 In orthopedic surgery, it is important to distinguish between surgical-site infection (SSI) and fracture-related infection (FRI). SSI is defined as infection occurring at or near the surgical incision within 30 days postoperatively (or within 90 days if a prosthetic implant is placed), according to CDC/NHSN criteria.10 FRI is a broader concept defined by international consensus as infection associated with a bone fracture and its management, which can present at various time points and is often related to the presence of implants.11 In this study, we focused specifically on early SSI occurring within two weeks post-surgery. This timeframe was selected based on the Chinese Expert Consensus on the Diagnosis and Treatment of Postoperative Infection after Fracture Internal Fixation (2018), which defines this period as the primary surveillance window for early SSI in orthopedic surgery.12 Among these infections, surgical-site infection (SSI) is of particular concern, as it often requires prolonged antibiotic therapy, may necessitate implant removal or revision surgery, and is associated with substantially worse functional outcomes compared to other postoperative complications. In recent years, several prediction models have been developed for postoperative complications in older adult hip fracture patients. Wei et al identified the modified 5-item frailty index (mFI-5) and the preoperative C-reactive protein to albumin ratio (CAR) as independent predictors for severe complications in older adult hip fracture patients, with their nomogram achieving an AUC of 0.963.13 Li et al established a nomogram for predicting noninfectious fever after IFF surgery, identifying factors including male sex, general anesthesia, longer operation time, and low postoperative albumin.14 Feng et al constructed a nomogram for postoperative incision infection in older adult patients with traumatic limb fractures, identifying CAR as an independent risk factor.15 Despite these advances, several gaps remain. First, most existing prediction models for hip fracture patients focus on “severe complications” (a composite outcome) or “noninfectious fever”, rather than specifically on surgical site infection (SSI).13 Second, few studies have developed simple bedside clinical tools that do not require complex calculations. Third, the specific combination of the CRP/Alb ratio, fasting glucose, and surgery duration for predicting SSI in older adult IFF patients has not been systematically investigated. Fourth, although fasting glucose is a well-established risk factor for postoperative infection,16,17 it has rarely been integrated with the CRP/Alb ratio in prediction models for this specific population. The C-reactive protein to albumin ratio (CRP/Alb) has emerged as a promising biomarker for predicting postoperative complications across various surgical fields.18–20 CRP is an acute-phase reactant that reflects systemic inflammation, whereas albumin is a marker of nutritional status and immune function. The combination of these two indicators into a single ratio may provide a more comprehensive assessment of a patient’s inflammatory and nutritional status than either marker alone.21 Fasting glucose is another readily available preoperative indicator; hyperglycemia is known to impair immune function, including neutrophil chemotaxis, phagocytosis, and bacterial killing.22,23 Similarly, surgery duration is a modifiable risk factor; prolonged operative time increases wound exposure to airborne bacteria and induces greater physiological stress.24,25 The combination of these three predictors is uniquely comprehensive: CRP/Alb captures the interplay between systemic inflammation and nutritional status, fasting glucose reflects perioperative metabolic dysregulation, and surgery duration accounts for procedural complexity and exposure risk. While each factor has been individually associated with postoperative complications, their combined predictive value for SSI in older adults with IFF has not been systematically evaluated. Given these considerations, this study aimed to: (1) determine the optimal cut-off values for the CRP/Alb ratio, fasting glucose, and surgery duration for predicting postoperative infection in older adult IFF patients; (2) develop a nomogram and a simple scoring system based on these factors for individualized risk assessment; (3) evaluate and compare the predictive performance of different models; and (4) assess the stability of the model across different patient subgroups (gender, BMI, and age). Unlike previous studies that focused on composite outcomes or required multiple variables, our study provides a simple, practical tool that can be easily implemented in clinical settings. Methods Study Design and Population This retrospective cohort study included 286 older adult patients with intertrochanteric femoral fractures (IFF) who were treated at a single tertiary hospital between August 2021 and August 2024. The inclusion criteria were as follows: (1) age ≥ 60 years; (2) confirmed diagnosis of IFF based on imaging findings; (3) receipt of surgical treatment (internal fixation or joint replacement); (4) availability of complete preoperative laboratory data. The exclusion criteria were as follows: (1) concomitant severe trauma or fractures at other sites; (2) pre-existing infectious diseases; (3) long-term use of glucocorticoids or immunosuppressants; (4) pathological or old fractures; (5) history of lower limb surgery; (6) contraindications to internal fixation or joint replacement; (7) presence of malignant tumors or hematologic diseases; (8) severe organ dysfunction; (9) incomplete medical records. Postoperative infection was defined as surgical-site infection (SSI) occurring within two weeks after surgery. Surgical-site infection refers to infection occurring at or near the surgical incision, which can be superficial incisional, deep incisional, or organ/space. Fracture-related infection (FRI) is a broader concept that includes infections related to the fracture healing process and can present at various time points, often associated with the presence of implants. In this study, we specifically focused on early SSI, as the two-week surveillance window aligns with the Chinese Expert Consensus on the Diagnosis and Treatment of Infection after Fracture Fixation,23 which defines this period as the primary surveillance timeframe for early SSI in orthopedic surgery. This timeframe also coincides with the typical inpatient hospital stay, ensuring complete and accurate data collection. Ethics Approval and Informed Consent This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Huaihe Hospital of Henan University (approval number: 2026138). Patient data were kept strictly confidential. All participants had provided written informed consent upon hospital admission, which explicitly authorized the use of their de-identified clinical data for future academic research. Given the retrospective, non-interventional design of this study and the use of fully anonymized data, the ethics committee waived the requirement for additional study-specific informed consent. Data Collection The following data were extracted from medical records: demographic characteristics (age, sex, and body mass index [BMI]); admission data (fasting glucose and blood pressure); preoperative laboratory data (serum C-reactive protein [CRP] and serum albumin [Alb]); surgical data (surgery duration, surgery type, intraoperative blood loss, and reduction quality); and postoperative outcomes (infection status). The CRP/Alb ratio was calculated by dividing serum CRP by serum Alb. Surgery duration was recorded as an intraoperative variable. It can serve as a risk update variable for intraoperative or postoperative decision-making, such as intensifying surveillance or adjusting antibiotic prophylaxis in patients with prolonged surgery. Statistical Analysis All statistical analyses were conducted using R software (version 4.2.0). Continuous variables were presented as mean ± standard deviation and compared using independent t-tests. Categorical variables were expressed as frequencies and percentages and compared using chi-square tests. Determination of Optimal Cut-off Values The Youden index (sensitivity + specificity − 1) was employed to identify the optimal cut-off values for C-reactive protein (CRP), albumin, the CRP-to-albumin ratio, fasting glucose, and surgery duration based on receiver operating characteristic (ROC) curves. The Youden index maximizes the difference between the true positive rate and the false positive rate, thereby achieving an optimal balance between sensitivity and specificity. Univariate and Multivariate Logistic Regression Analysis Univariate logistic regression was first performed for each potential predictor. Variables with a p value < 0.05 in the univariate analysis were subsequently entered into a multivariate logistic regression model using a forward stepwise selection method. Results were reported as odds ratios (OR) with 95% confidence intervals (CI). Due to collinearity among C-reactive protein (CRP), albumin, and the CRP/albumin ratio (CRP/Alb = CRP ÷ Alb), only the CRP/albumin ratio was included in the final multivariate model, along with fasting glucose and surgery duration. Nomogram A nomogram was constructed based on the multivariate logistic regression model. The nomogram converted each variable into a point scale, and the sum of the points (total points) corresponded to a predicted probability of postoperative infection. The nomogram was visualized using the “rms” package in R. Simple Scoring System Based on the optimal cut-off values determined by the Youden index, each risk factor was assigned 1 point if its value exceeded the cut-off threshold. The total score was calculated as the sum of these points, ranging from 0 to 3. The infection rate for each total score was then calculated. Model Comparison Three predictive models were compared: Model 1 (CRP/albumin ratio alone), Model 2 (CRP/albumin ratio + fasting glucose), and Model 3 (CRP/albumin ratio + fasting glucose + surgery duration). The area under the receiver operating characteristic curve (AUC) was calculated for each model, and pairwise comparisons were performed using the DeLong test. Subgroup Analysis Subgroup analyses were conducted to evaluate the stability of Model 3 across different patient populations. Subgroups were defined by gender (male vs female), body mass index (BMI; ≥24 vs <24 kg/m2), and age (≥75 vs <75 years). The AUC with 95% CI was calculated for each subgroup. Internal Validation and Calibration Bootstrap resampling with 500 iterations was used to calculate the optimism-corrected AUC of Model 3. The calibration of Model 3 was assessed using a calibration curve and the Hosmer–Lemeshow goodness-of-fit test (p > 0.05 indicated good calibration). To minimize the risk of overfitting given the limited number of infection events (n = 25), only variables with p < 0.05 in univariate analysis were considered for inclusion in the multivariate model. Collinearity was assessed using variance inflation factors (VIF). Model calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test. All tests were two-tailed, and a p value < 0.05 was considered statistically significant. Results Baseline Characteristics A total of 286 older adult patients with intertrochanteric femoral fracture (IFF) were included in this study, of whom 25 (8.7%) developed postoperative infection within two weeks after surgery. Table 1 presents the baseline characteristics of the study population stratified by infection status.Table 1Baseline Characteristics of the Study PopulationVariableInfection Group (n = 25)Non-Infection Group (n = 261)p valueDemographic and Clinical CharacteristicsAge (years)73.04 ±5.7872.49 ±5.740.65BMI (kg/m2)23.72 ±1.3623.80 ±1.410.785Gender (Male)13 (52.0%)141 (54.0%)1Vital signsDiastolic BP (mmHg)96.20 ±6.7897.02 ±6.840.566Systolic BP (mmHg)148.64 ±12.56148.02 ±13.040.819Laboratory parametersFasting glucose (mmol/L)8.25 ±1.247.25 ±0.85<0.001CRP (mg/L)63.06 ±16.5448.36 ±12.60<0.001Albumin (g/L)30.70 ±6.5837.45 ±6.44<0.001CRP/Alb ratio2.05 ±0.321.48 ±0.23<0.001Surgical and hospitalization parametersTime to surgery (h)8.49 ±3.168.46 ±3.200.961Surgery duration (min)87.38 ±16.3972.47 ±12.54<0.001Intraoperative blood loss (mL)186.54 ±27.39188.02 ±28.120.802Hospital stay (days)11.00 ±1.3210.85 ±1.300.595Classification and surgery typeASA grade (II vs I)14 (56.0%)141 (54.0%)1AO classification0.747A12 (8.0%)12 (4.6%)A214 (56.0%)148 (56.7%)A39 (36.0%)101 (38.7%)Surgery type (Internal fixation)17 (68.0%)197 (75.5%)0.561Fracture side (Left)14 (56.0%)134 (51.3%)0.814Injury cause0.92Fall12 (48.0%)132 (50.6%)Traffic accident10 (40.0%)104 (39.8%)High fall3 (12.0%)25 (9.6%)Note: Data are presented as mean ± SD or n (%). Table 1Baseline Characteristics of the Study PopulationVariableInfection Group (n = 25)Non-Infection Group (n = 261)p valueDemographic and Clinical CharacteristicsAge (years)73.04 ±5.7872.49 ±5.740.65BMI (kg/m2)23.72 ±1.3623.80 ±1.410.785Gender (Male)13 (52.0%)141 (54.0%)1Vital signsDiastolic BP (mmHg)96.20 ±6.7897.02 ±6.840.566Systolic BP (mmHg)148.64 ±12.56148.02 ±13.040.819Laboratory parametersFasting glucose (mmol/L)8.25 ±1.247.25 ±0.85<0.001CRP (mg/L)63.06 ±16.5448.36 ±12.60<0.001Albumin (g/L)30.70 ±6.5837.45 ±6.44<0.001CRP/Alb ratio2.05 ±0.321.48 ±0.23<0.001Surgical and hospitalization parametersTime to surgery (h)8.49 ±3.168.46 ±3.200.961Surgery duration (min)87.38 ±16.3972.47 ±12.54<0.001Intraoperative blood loss (mL)186.54 ±27.39188.02 ±28.120.802Hospital stay (days)11.00 ±1.3210.85 ±1.300.595Classification and surgery typeASA grade (II vs I)14 (56.0%)141 (54.0%)1AO classification0.747A12 (8.0%)12 (4.6%)A214 (56.0%)148 (56.7%)A39 (36.0%)101 (38.7%)Surgery type (Internal fixation)17 (68.0%)197 (75.5%)0.561Fracture side (Left)14 (56.0%)134 (51.3%)0.814Injury cause0.92Fall12 (48.0%)132 (50.6%)Traffic accident10 (40.0%)104 (39.8%)High fall3 (12.0%)25 (9.6%)Note: Data are presented as mean ± SD or n (%). Baseline Characteristics of the Study Population Note: Data are presented as mean ± SD or n (%). The infection and non-infection groups were comparable with respect to demographic characteristics. The mean age was 73.04 ± 5.78 years in the infection group and 72.49 ±5.74 years in the non-infection group, with no significant difference (p = 0.650). Similarly, there were no significant differences in BMI (23.72 ±1.36 vs 23.80 ±1.41 kg/m2, p = 0.785) or gender distribution (52.0% male in the infection group vs 54.0% in the non-infection group, p = 1.000). Regarding vital signs and laboratory parameters, the infection group had significantly higher fasting glucose levels (8.25 ±1.24 vs 7.25 ±0.85 mmol/L, p < 0.001), higher CRP levels (63.06 ±16.54 vs 48.36 ±12.60 mg/L, p < 0.001), and higher CRP/Alb ratios (2.05 ±0.32 vs 1.48 ±0.23, p < 0.001). Conversely, albumin levels were significantly lower in the infection group (30.70 ±6.58 vs 37.45 ±6.44 g/L, p < 0.001). No significant differences were observed in diastolic or systolic blood pressure (both p > 0.05). With respect to surgical and hospitalization parameters, surgery duration was significantly longer in the infection group (87.38 ±16.39 vs 72.47 ±12.54 minutes, p < 0.001). However, there were no significant differences in time from injury to surgery (8.49 ±3.16 vs 8.46 ±3.20 hours, p = 0.961), intraoperative blood loss (186.54 ±27.39 vs 188.02 ±28.12 mL, p = 0.802), or hospital stay (11.00 ±1.32 vs 10.85 ±1.30 days, p = 0.595). Regarding classification and surgery type, no significant differences were found between the two groups in ASA grade (p = 1.000), AO classification (p = 0.747), surgery type (internal fixation vs arthroplasty, p = 0.561), fracture side (p = 0.814), or injury cause (p = 0.920). These findings indicate that the two groups were well-balanced for most baseline variables, and the differences in infection risk were primarily attributable to the laboratory and surgical parameters of interest. Univariate and Multivariate Analysis Table 2 presents the results of univariate and multivariate logistic regression analyses for postoperative infection.Table 2Univariate and Multivariate Logistic Regression Analysis for Postoperative InfectionVariableUnivariate AnalysisMultivariate AnalysisOR (95% CI)p valueOR (95% CI)p valueAge (years)1.02 (0.95–1.09)0.649——BMI (kg/m2)0.96 (0.72–1.29)0.784——CRP (mg/L)1.09 (1.05–1.12)<0.001——Albumin (g/L)0.86 (0.81–0.92)<0.001——Fasting glucose (mmol/L)2.93 (1.81–4.74)<0.0012.82 (1.32–6.73)0.011CRP/Alb ratio4150.82 (269.53–63,923.19)<0.0012582.84 (201.44–67,929.9)<0.001Surgery duration (min)1.09 (1.05–1.13)<0.0011.06 (1.01–1.12)0.02 Table 2Univariate and Multivariate Logistic Regression Analysis for Postoperative InfectionVariableUnivariate AnalysisMultivariate AnalysisOR (95% CI)p valueOR (95% CI)p valueAge (years)1.02 (0.95–1.09)0.649——BMI (kg/m2)0.96 (0.72–1.29)0.784——CRP (mg/L)1.09 (1.05–1.12)<0.001——Albumin (g/L)0.86 (0.81–0.92)<0.001——Fasting glucose (mmol/L)2.93 (1.81–4.74)<0.0012.82 (1.32–6.73)0.011CRP/Alb ratio4150.82 (269.53–63,923.19)<0.0012582.84 (201.44–67,929.9)<0.001Surgery duration (min)1.09 (1.05–1.13)<0.0011.06 (1.01–1.12)0.02 Univariate and Multivariate Logistic Regression Analysis for Postoperative Infection Univariate analysis revealed that several variables were significantly associated with postoperative infection. Fasting glucose showed a strong association, with an odds ratio (OR) of 2.93 (95% CI: 1.81–4.74, p < 0.001). C-reactive protein (CRP) also demonstrated a significant association, with an OR of 1.09 (95% CI: 1.05–1.12, p < 0.001), indicating that each 1 mg/L increase in CRP was associated with a 9% increase in infection risk. Conversely, albumin was inversely associated with infection risk, with an OR of 0.86 (95% CI: 0.81–0.92, p < 0.001), meaning that each 1 g/L increase in albumin was associated with a 14% reduction in infection risk. The CRP/albumin ratio showed the strongest association, with an OR of 4150.82 (95% CI: 269.53–63,923.19, p < 0.001), reflecting the combined effect of elevated CRP and reduced albumin. Surgery duration was also significantly associated, with an OR of 1.09 (95% CI: 1.05–1.13, p < 0.001), indicating a 9% increase in infection risk per minute of surgery. In contrast, age (p = 0.649) and BMI (p = 0.784) were not significantly associated with infection risk. Multivariate analysis was then performed, including variables with p < 0.05 from the univariate analysis. Due to collinearity among CRP, albumin, and the CRP/albumin ratio (CRP/albumin = CRP ÷ albumin), only the CRP/albumin ratio was retained in the final model, along with fasting glucose and surgery duration. The multivariate model identified three independent risk factors for postoperative infection. The CRP/albumin ratio remained the strongest predictor, with an OR of 2582.84 (95% CI: 201.44–67,929.9, p < 0.001). Fasting glucose was also an independent predictor, with an OR of 2.82 (95% CI: 1.32–6.73, p = 0.011). Surgery duration showed a smaller but significant association, with an OR of 1.06 (95% CI: 1.01–1.12, p = 0.020), indicating a 6% increase in infection risk per additional minute of surgery after adjusting for other factors. The high OR for the CRP/Alb ratio (2582.84) reflects the narrow range of this ratio (typically 1–3 in this population); clinically, a 0.1 increase in CRP/Alb corresponds to approximately a 2.2-fold increase in infection risk (exp(0.1 × 7.856) ≈ 2.2). Nomogram Based on the three independent risk factors identified in the multivariate analysis (CRP/Alb ratio, fasting glucose, and surgery duration), a nomogram was constructed to provide individualized risk assessment for postoperative infection (Figure 1). The nomogram includes three variables: CRP/Alb ratio, fasting glucose (mmol/L), and surgery duration (minutes). For each patient, the points corresponding to each variable are summed to obtain the total points, which are then converted into the predicted probability of infection. The nomogram offers clinicians a visual and accurate tool for estimating infection risk at the individual level, thereby facilitating risk stratification and supporting shared decision-making with patients and their families. Figure 1Nomogram for predicting postoperative infection in older adult patients with intertrochanteric femoral fracture. The nomogram incorporates three predictors: CRP/albumin ratio, fasting glucose (mmol/L), and surgery duration (minutes). To use the nomogram, locate each variable value on its corresponding axis, draw a line upward to the “Points” axis to obtain the points for that variable, sum all points to obtain the “Total Points”, then draw a line downward from the “Total Points” axis to the “Infection Risk” axis to read the predicted probability of postoperative infection.Nomogram with six aligned horizontal scales used to convert three predictors into total points and infection risk. Top scale label, Points, has tick labels from 0 to 100 in steps of 10. Second scale label, CRP/Alb ratio, ranges from 0 to 4.5 with tick labels at 0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, and 4.5. Third scale label, Fasting glucose (millimoles per liter), ranges from 5 to 12 with tick labels at 5, 5.5, 6, 6.5, 7, 7.5, 8, 8.5, 9, 9.5, 10, 10.5, 11, 11.5, and 12. Fourth scale label, Surgery duration (minutes), ranges from 30 to 120 with tick labels at 30, 40, 50, 60, 70, 80, 90, 100, 110, and 120. Fifth scale label, Total Points, ranges from 0 to 180 with tick labels at 0, 20, 40, 60, 80, 100, 120, 140, 160, and 180. Bottom scale label, Infection Risk, shows tick labels 0.01, 0.05, 0.1, 0.2, 0.4, 0.6, 0.8, and 0.9. To use: locate each variable on its axis, draw a vertical line upward to the Points scale, sum to obtain Total Points, then draw a vertical line downward to the Infection Risk scale to read the predicted probability.Nomogram relating CRP/Alb ratio, fasting glucose, and surgery duration to infection risk in older adult patients with intertrochanteric femoral fracture. Nomogram for predicting postoperative infection in older adult patients with intertrochanteric femoral fracture. The nomogram incorporates three predictors: CRP/albumin ratio, fasting glucose (mmol/L), and surgery duration (minutes). To use the nomogram, locate each variable value on its corresponding axis, draw a line upward to the “Points” axis to obtain the points for that variable, sum all points to obtain the “Total Points”, then draw a line downward from the “Total Points” axis to the “Infection Risk” axis to read the predicted probability of postoperative infection. Simple Scoring System Based on the optimal cut-off values determined by the Youden index (CRP/Alb ratio > 1.85, fasting glucose > 7.6 mmol/L, surgery duration > 89 minutes), a simple scoring system was developed. Each risk factor was assigned 1 point if its value exceeded the corresponding cut-off, and the total score was calculated as the sum of these points, ranging from 0 to 3, while surgery duration is assessed intraoperatively, the scoring system can be applied postoperatively to update risk estimates and guide surveillance strategies. Table 3 presents the infection rates according to the total score. Among the 286 patients, 141 (49.3%) had a score of 0, and none of these patients developed infection (infection rate: 0%). A total of 113 patients (39.5%) had a score of 1, of whom 6 developed infection, resulting in an infection rate of 5.6%. Twenty-two patients (7.7%) had a score of 2, with 9 infections (infection rate: 40.9%). Finally, 10 patients (3.5%) had a score of 3, and all 10 developed infection (infection rate: 100%). The scoring system demonstrated excellent discriminative ability, with an AUC of 0.932 (95% CI: 0.894–0.970). Figure 2 presents a bar chart visualizing the infection rates across the four score categories, clearly illustrating the progressive increase in infection risk with higher scores.Table 3Infection Rates by Simple Scoring SystemScoreNInfectedInfection RateRisk Level014100%Very low111365.60%Low222940.90%High31010100%Very high Figure 2Infection rates by the simplified scoring system. The simplified scoring system assigns one point for each risk factor present: CRP/Alb ratio > 1.85, fasting glucose > 7.6 mmol/L, and surgery duration > 89 minutes. Total scores range from 0 to 3 points. The bar chart shows the infection rates for each score category: 0% for score 0 (very low risk), 5.6% for score 1 (low risk), 40.9% for score 2 (high risk), and 100% for score 3 (very high risk). The scoring system demonstrated excellent discriminative ability with an AUC of 0.932.Bar chart showing infection rates by simplified scoring system score (0–3); rates rise from 0% at score 0 to 100% at score 3 (AUC 0.932). Table 3Infection Rates by Simple Scoring SystemScoreNInfectedInfection RateRisk Level014100%Very low111365.60%Low222940.90%High31010100%Very high Infection Rates by Simple Scoring System Infection rates by the simplified scoring system. The simplified scoring system assigns one point for each risk factor present: CRP/Alb ratio > 1.85, fasting glucose > 7.6 mmol/L, and surgery duration > 89 minutes. Total scores range from 0 to 3 points. The bar chart shows the infection rates for each score category: 0% for score 0 (very low risk), 5.6% for score 1 (low risk), 40.9% for score 2 (high risk), and 100% for score 3 (very high risk). The scoring system demonstrated excellent discriminative ability with an AUC of 0.932. Diagnostic Performance Table 4 summarizes the diagnostic performance of the simple scoring system and the three combined models. At a cut-off of ≥1 point, the simple scoring system demonstrated a sensitivity of 0.880, a specificity of 0.881, and a Youden index of 0.761. Among the individual predictors, the CRP/Alb ratio exhibited the best performance, with an AUC of 0.916 (95% CI: 0.85–0.98), a sensitivity of 0.840, a specificity of 0.943, and a Youden index of 0.783 at a cut-off of 1.85. Model 2 (CRP/Alb + fasting glucose) achieved an AUC of 0.943 (95% CI: 0.89–0.99), with a sensitivity of 0.880, a specificity of 0.877, and a Youden index of 0.757 at a predicted probability cut-off of 0.0806. Model 3 (CRP/Alb + fasting glucose + surgery duration) attained the highest AUC of 0.951 (95% CI: 0.90–1.00), with a sensitivity of 0.840, a specificity of 0.923, and a Youden index of 0.763 at a predicted probability cut-off of 0.108.Table 4Diagnostic Performance of the Simple Scoring System and Combined ModelsModel/ToolAUC95% CISensitivitySpecificityYouden IndexCut-offSimple scoring system (0–3 points)0.932(0.894–0.970)0.880.8810.761≥1 pointModel 1 (CRP/Alb)0.916(0.85–0.98)0.840.9430.7831.85Model 2 (+ Fasting glucose)0.943(0.89–0.99)0.880.8770.7570.0806*Model 3 (+ Surgery duration)0.951(0.90–1.00)0.840.9230.7630.108*Note: *Cut-off values for Models 2 and 3 represent predicted probability thresholds. Table 4Diagnostic Performance of the Simple Scoring System and Combined ModelsModel/ToolAUC95% CISensitivitySpecificityYouden IndexCut-offSimple scoring system (0–3 points)0.932(0.894–0.970)0.880.8810.761≥1 pointModel 1 (CRP/Alb)0.916(0.85–0.98)0.840.9430.7831.85Model 2 (+ Fasting glucose)0.943(0.89–0.99)0.880.8770.7570.0806*Model 3 (+ Surgery duration)0.951(0.90–1.00)0.840.9230.7630.108*Note: *Cut-off values for Models 2 and 3 represent predicted probability thresholds. Diagnostic Performance of the Simple Scoring System and Combined Models Note: *Cut-off values for Models 2 and 3 represent predicted probability thresholds. Figure 3 presents the ROC curves of all four models for visual comparison. The curves indicate that Model 3 achieved the highest AUC, followed by Model 2, the simple scoring system, and Model 1. However, DeLong tests revealed no statistically significant differences among the models (Model 2 vs Model 1: p = 0.140; Model 3 vs Model 2: p = 0.508), suggesting that even the simplest model (CRP/Alb alone) already provides excellent predictive performance. Figure 3ROC curves for predicting postoperative infection. Receiver operating characteristic (ROC) curves of four predictive models: Model 1 (CRP/Alb ratio alone, AUC = 0.916), Model 2 (CRP/Alb ratio + fasting glucose, AUC = 0.943), Model 3 (CRP/Alb ratio + fasting glucose + surgery duration, AUC = 0.951), and the simplified scoring system (AUC = 0.932). The diagonal line represents the reference line (AUC = 0.5). Model 3 achieved the highest AUC, followed by Model 2, the scoring system, and Model 1. DeLong tests showed no statistically significant differences between the models (all p > 0.05), indicating that even the simplest model (CRP/Alb alone) provides excellent predictive performance.ROC curves comparing four predictive models for postoperative infection; Model 3 achieved the highest AUC (0.951). ROC curves for predicting postoperative infection. Receiver operating characteristic (ROC) curves of four predictive models: Model 1 (CRP/Alb ratio alone, AUC = 0.916), Model 2 (CRP/Alb ratio + fasting glucose, AUC = 0.943), Model 3 (CRP/Alb ratio + fasting glucose + surgery duration, AUC = 0.951), and the simplified scoring system (AUC = 0.932). The diagonal line represents the reference line (AUC = 0.5). Model 3 achieved the highest AUC, followed by Model 2, the scoring system, and Model 1. DeLong tests showed no statistically significant differences between the models (all p > 0.05), indicating that even the simplest model (CRP/Alb alone) provides excellent predictive performance. Subgroup Analysis To assess the stability of the full model (Model 3), subgroup analyses were conducted based on gender, body mass index (BMI), and age. Table 5 presents the area under the curve (AUC) values for each subgroup. The model demonstrated consistent performance across all subgroups, with AUCs ranging from 0.922 to 0.982. In the gender subgroups, the AUC was 0.953 (95% CI: 0.901–1.000) for males and 0.951 (95% CI: 0.891–1.000) for females, indicating no notable difference between sexes. Within the BMI subgroups, the model performed best among patients with a BMI of 24 kg/m2 or higher (AUC = 0.982, 95% CI: 0.947–1.000) compared to those with a BMI below 24 kg/m2 (AUC = 0.922, 95% CI: 0.852–0.992). Regarding age subgroups, the AUC was 0.958 (95% CI: 0.905–1.000) for patients aged 75 years or older and 0.947 (95% CI: 0.894–1.000) for those younger than 75 years. Figure 4 provides a forest plot illustrating the AUCs and 95% confidence intervals for each subgroup, highlighting the model’s consistent performance across diverse patient populations.Table 5Subgroup Analysis of Model 3 PerformanceSubgroupAUC95% CINEventsMale0.953(0.901–1.000)15413Female0.951(0.891–1.000)13212BMI ≥ 240.982(0.947–1.000)12011BMI < 240.922(0.852–0.992)16614Age ≥ 75 years0.958(0.905–1.000)1037Age < 75 years0.947(0.894–1.000)18318 Figure 4Forest plot of subgroup analysis for Model 3 performance. Forest plot showing the area under the curve (AUC) with 95% confidence intervals for Model 3 across different patient subgroups. Subgroups were defined by gender (male vs female), body mass index (BMI ≥ 24 vs < 24 kg/m2), and age (≥ 75 vs < 75 years). The vertical dashed line indicates an AUC of 0.95. The model performed consistently well across all subgroups, with AUCs ranging from 0.922 to 0.982. The size of each data point is proportional to the number of infection events in each subgroup. The highest AUC was observed in the BMI ≥ 24 subgroup (0.982), while the lowest was in the BMI < 24 subgroup (0.922). The results demonstrate the excellent stability and generalizability of the model across different patient populations.A forest plot with six horizontal confidence interval bars and circular markers for Model 3. The x-axis label is “AUC (95% CI)” with tick labels 0.85, 0.90, 0.95, and 1.00. A vertical dashed reference line is drawn at 0.95. Subgroup labels at left are: BMI < 24 (N = 166, E = 14); Age < 75 years (N = 183, E = 18); Female (N = 132, E = 12); Male (N = 154, E = 13); Age ≥ 75 years (N = 103, E = 7); BMI ≥ 24 (N = 120, E = 11). Numeric AUC values printed at right are, in the same order: 0.922, 0.947, 0.951, 0.953, 0.958, 0.982. Each row shows a horizontal 95% confidence interval spanning roughly from just below 0.90 up to 1.00, except BMI < 24 which extends left to about 0.85. Marker sizes vary by subgroup. A legend below reads “Infection Events”, with five marker sizes labeled 7.5, 10.0, 12.5, 15.0, and 17.5.A forest plot of Model 3 area under the curve across patient subgroups, with values clustered near 0.95. Table 5Subgroup Analysis of Model 3 PerformanceSubgroupAUC95% CINEventsMale0.953(0.901–1.000)15413Female0.951(0.891–1.000)13212BMI ≥ 240.982(0.947–1.000)12011BMI < 240.922(0.852–0.992)16614Age ≥ 75 years0.958(0.905–1.000)1037Age < 75 years0.947(0.894–1.000)18318 Subgroup Analysis of Model 3 Performance Forest plot of subgroup analysis for Model 3 performance. Forest plot showing the area under the curve (AUC) with 95% confidence intervals for Model 3 across different patient subgroups. Subgroups were defined by gender (male vs female), body mass index (BMI ≥ 24 vs < 24 kg/m2), and age (≥ 75 vs < 75 years). The vertical dashed line indicates an AUC of 0.95. The model performed consistently well across all subgroups, with AUCs ranging from 0.922 to 0.982. The size of each data point is proportional to the number of infection events in each subgroup. The highest AUC was observed in the BMI ≥ 24 subgroup (0.982), while the lowest was in the BMI < 24 subgroup (0.922). The results demonstrate the excellent stability and generalizability of the model across different patient populations. Internal Validation and Calibration Bootstrap resampling with 500 iterations was performed to evaluate the optimism-corrected area under the curve (AUC) of Model 3. The corrected AUC was 0.945 (95% CI: 0.923–0.951), which closely approximated the original AUC of 0.951, suggesting satisfactory internal validity without substantial overfitting. Figure 5 presents the calibration curve of Model 3. The predicted probabilities closely matched the observed infection rates across the entire range of predicted risk. The Hosmer–Lemeshow goodness-of-fit test yielded a non-significant result (χ2 = 3.45, p = 0.327), indicating that the model is well-calibrated and that no significant discrepancy exists between predicted and observed outcomes. Figure 5Calibration curve for Model 3. Calibration curve of Model 3 (CRP/Alb ratio + fasting glucose + surgery duration). The x-axis represents the predicted probability of postoperative infection, and the y-axis represents the observed infection rate. The diagonal dashed line (“Ideal”) represents perfect calibration where predicted probabilities equal observed outcomes. The solid line (“Apparent”) shows the model’s calibration on the original data, and the dotted line (“Bias-corrected”) shows the calibration after Bootstrap correction (500 iterations). The close proximity of the bias-corrected line to the ideal line indicates good calibration. The Hosmer-Lemeshow test yielded a non-significant result (χ2 = 3.45, p = 0.327), confirming no significant difference between predicted and observed infection rates.Calibration curve of Model 3 for postoperative infection in older adult patients with intertrochanteric femoral fracture, showing the bias-corrected line closely following the ideal diagonal. The Hosmer-Lemeshow test indicates good calibration (χ2 = 3.45, p = 0.327). Calibration curve for Model 3. Calibration curve of Model 3 (CRP/Alb ratio + fasting glucose + surgery duration). The x-axis represents the predicted probability of postoperative infection, and the y-axis represents the observed infection rate. The diagonal dashed line (“Ideal”) represents perfect calibration where predicted probabilities equal observed outcomes. The solid line (“Apparent”) shows the model’s calibration on the original data, and the dotted line (“Bias-corrected”) shows the calibration after Bootstrap correction (500 iterations). The close proximity of the bias-corrected line to the ideal line indicates good calibration. The Hosmer-Lemeshow test yielded a non-significant result (χ2 = 3.45, p = 0.327), confirming no significant difference between predicted and observed infection rates. Discussion In this single-center retrospective cohort study of 286 older adult patients with IFF, we identified the preoperative CRP/Alb ratio, fasting glucose, and intraoperative surgery duration as independent predictors of postoperative infection. To our knowledge, this is the first study to combine these three specific variables into a practical bedside scoring tool for this population. The simple scoring system demonstrated a clear risk gradient from 0% to 100% infection rates (AUC = 0.932), and the nomogram provided individualized risk estimation. However, we acknowledge that the novelty of this study is modest and our findings should be considered preliminary until external validation is performed in independent cohorts. Comparison with Previous Studies The CRP/Alb ratio emerged as the strongest predictor in our study AUC = 0.916. Although the OR for the CRP/Alb ratio appears extremely high (2582.84), this reflects the narrow range of the ratio itself (1–3). A 0.1 increase in the ratio, which is clinically meaningful, corresponds to a 2.2-fold increase in infection risk. This interpretation makes the clinical significance of the CRP/Alb ratio more accessible for clinicians. This is consistent with previous studies demonstrating the predictive value of CRP/Alb in orthopedic populations. Wei et al identified the modified 5-item frailty index (mFI-5) and the preoperative C-reactive protein to albumin ratio (CAR) as independent predictors for severe complications in older adults hip fracture patients, with their nomogram achieving an AUC of 0.963.13 Li et al established a nomogram for predicting noninfectious fever after IFF surgery, identifying factors including male sex, general anesthesia, longer operation time, and low postoperative albumin.14 Feng et al confirmed CRP/Alb as an independent risk factor for incision infection in older adult patients with traumatic limb fractures.15 Our study extends these observations in several important ways. First, while previous studies often used composite outcomes (eg, “severe complications” including pneumonia, cardiovascular events, and death), we focused specifically on SSI, which has different risk factor profiles and clinical implications. Second, we are the first to integrate fasting glucose with CRP/Alb in a prediction model for IFF patients. Our multivariate analysis showed that fasting glucose remained significant even after adjusting for CRP/Alb and surgery duration OR = 2.82, p = 0.011, indicating its independent contribution to infection risk beyond inflammation and surgical factors. Third, unlike most previous models that require multiple variables (eg, Wei’s model included mFI-5 and CAR as independent predictors13), our scoring system uses only three readily available variables and can be calculated at the bedside in under one minute. Fasting glucose was identified as an independent predictor, with an optimal cut-off value of 7.6 mmol/L. This finding aligns with well-established evidence that hyperglycemia impairs immune function, including neutrophil chemotaxis, phagocytosis, and bacterial killing.22,23 Elevated glucose levels also create a favorable environment for bacterial growth.26 Our cut-off value is similar to those reported in previous studies of orthopedic surgery, which have indicated that preoperative glucose levels above 7.0–8.0 mmol/L are associated with increased infection risk.27,28 Surgery duration was the third independent predictor, with each minute increase associated with a 6% increase in infection risk OR=1.06 and an optimal cut-off of 89 minutes. This finding is consistent with a large body of evidence linking prolonged operative time to SSI. Cheng et al reported that each 15-minute increase in surgery duration was associated with a 13% increase in infection risk.24 Li et al also identified longer operation time as an independent risk factor for noninfectious fever after IFF surgery.14 Our cut-off of 89 minutes provides a practical benchmark for surgical teams. Clinical Utility of the Simple Scoring System The simple scoring system developed in this study offers a substantial practical contribution. By assigning one point for each of the three risk factors (CRP/Alb > 1.85, fasting glucose > 7.6 mmol/L, and surgery duration > 89 minutes), clinicians can rapidly stratify patients into four distinct risk categories: score 0 (0% infection, very low risk), score 1 (5.6% infection, low risk), score 2 (40.9% infection, high risk), and score 3 (100% infection, very high risk). This clear gradient carries direct clinical implications. For patients with a score of 0, simplified perioperative protocols may be considered, potentially reducing unnecessary antibiotic use and hospital stay. For patients with a score of 1, the infection rate was 5.6% (6/113), indicating that most patients in this category (107/113, 94.7%) did not develop infection. This suggests that a single risk factor confers modestly increased risk, but routine perioperative management may be sufficient without the need for aggressive interventions. For patients with a score of 2, the infection rate increased substantially to 40.9% (9/22), indicating a high-risk group that requires enhanced preventive measures. For these patients, clinicians should consider strict glycemic control, nutritional support, minimization of surgery duration, and intensified postoperative monitoring. For patients with a score of 3, all 10 patients developed postoperative infection (100% infection rate), suggesting near-perfect positive predictive value for identifying the highest-risk patients. For these patients, aggressive perioperative interventions should be strongly considered, including strict glycemic control, prophylactic antibiotic adjustment, nutritional optimization, minimization of surgical time, and enhanced postoperative surveillance. However, given that this observation is based on only 10 patients, this finding should be interpreted with caution and requires validation in larger cohorts before being translated into routine clinical practice. Unlike previous nomograms that require complex calculations or specialized software, the proposed scoring system can be used at the bedside without any computational aid. Model Performance and Comparison In the comparison of different predictive models, Model 1 (CRP/Alb alone) achieved an AUC of 0.916, Model 2 (+ fasting glucose) achieved 0.943, and Model 3 (+ surgery duration) achieved 0.951. Although the full model yielded the highest AUC, DeLong tests indicated no statistically significant differences among the models (Model 2 vs Model 1: p = 0.140; Model 3 vs Model 2: p = 0.508). These results suggest that CRP/Alb alone already provides excellent predictive performance, and the inclusion of fasting glucose and surgery duration may not be necessary in all clinical contexts. This finding holds clinical significance, as it implies that even when fasting glucose or precise surgery duration data are unavailable, clinicians can still rely on CRP/Alb to assess infection risk with confidence. Subgroup Stability Our subgroup analyses indicated that the model performed consistently well across different patient populations. The AUC values ranged from 0.922 to 0.982 across subgroups defined by gender, BMI, and age, demonstrating excellent stability and generalizability. Notably, the model achieved the highest performance in the BMI ≥ 24 subgroup AUC = 0.982 and relatively lower performance in the BMI < 24 subgroup AUC = 0.922. This difference may be attributed to the more pronounced inflammatory and metabolic responses observed in overweight patients, which could enhance the discriminative power of the predictive indicators. To the best of our knowledge, this is the first study to conduct such comprehensive subgroup analyses for an SSI prediction model in patients with IFF, providing evidence of the model’s robustness across diverse patient characteristics. Limitations Several limitations of this study should be acknowledged. First, this was a single-center retrospective study with a relatively small sample size, particularly for the infection group (n = 25), which limited the number of variables that could be included in the multivariate model and resulted in wide confidence intervals for some estimates. The model may be susceptible to overfitting, although bootstrap internal validation (corrected AUC = 0.945) suggested acceptable stability. External validation in independent cohorts is essential before generalization. Second, the optimal cut-off values were derived from the same dataset used for model development, which may lead to overestimation of performance despite bootstrap internal validation (corrected AUC = 0.945). Third, the study only assessed short-term infections within two weeks post-surgery; the predictive value for late-onset infections remains unknown. Fourth, due to the retrospective design, we were unable to collect data on potentially important confounders, including diabetes status, HbA1c, smoking, comorbidity, surgical procedure, antibiotic prophylaxis, and perioperative care, which may have contributed to residual confounding. Future Directions Future research should focus on the following areas: (1) external validation of the present findings in multicenter prospective cohorts, which is currently being planned by our research team through a funded research grant; (2) evaluation of the impact of the scoring system on clinical outcomes and cost-effectiveness; (3) exploration of additional biomarkers—such as procalcitonin, interleukin-6, or the neutrophil-to- lymphocyte ratio—that may further enhance predictive performance; and (4) development of a mobile application or online calculator based on the nomogram to facilitate clinical application. Conclusions In this single-center retrospective cohort, the CRP/Alb ratio, fasting glucose, and surgery duration were identified as independent predictors of postoperative infection in older adults with IFF. A simple scoring system (0–3 points) and a nomogram were developed and internally validated in this cohort (n = 286, 25 events). The scoring system demonstrated a clear risk gradient from 0% to 100% across scores, with an AUC of 0.932. However, the absolute infection rates and optimal cut-off values reported in this study were derived from a single-center cohort and should be considered preliminary. External validation in multicenter prospective cohorts is warranted before these thresholds can be recommended for routine clinical practice. The scoring system can be applied in clinical practice as follows: patients with a score of 0 may be candidates for simplified perioperative protocols; those with a score of 1 require only standard preventive measures; those with a score of 2 warrant enhanced interventions including glycemic control, nutritional support, and intensified monitoring; and those with a score of 3 should be considered for aggressive perioperative management. The scoring system is currently best suited for identifying high-risk patients for enhanced monitoring and for guiding the design of future validation studies, rather than for definitive clinical decision-making.