Authors: Wanyun Tang, Yudong Hou, Chenglin Bai, Yaohua Shang
Categories: Research, Hip fracture, Postoperative pneumonia, Geriatric, Prediction model, Nomogram
Source: European Journal of Medical Research
Authors: Wanyun Tang, Yudong Hou, Chenglin Bai, Yaohua Shang
Hip fracture is a common and serious injury in older adults, with an increasing incidence as the population ages. Postoperative pneumonia (POP) is a common complication after hip fracture surgery, and is associated with increased mortality, morbidity, and length of hospital stay. The purpose of this study was to develop and validate a nomogram to predict the risk of POP in geriatric patients with hip fracture.
We conducted a retrospective cohort study of geriatric patients who underwent hip fracture surgery in our hospital between Feb 2015 and Feb 2024. The cohort was randomly divided into training (n = 908) and validation (n = 388) sets. We used least absolute shrinkage and selection operator (LASSO) regression and logistic regression to identify independent risk factors for POP in the training set and constructed a nomogram using these predictors. The performance of the nomogram was evaluated by the C-index, Receiver Operating Characteristic (ROC) analysis, calibration plots, and decision curve analysis.
A total of 1296 patients were included in the study. The incidence of POP was 9.1%. The nomogram included the following age, sex, Chronic Obstructive Pulmonary Disease (COPD), postoperative Intensive Care Unit (ICU), American Society of Anesthesiologists (ASA) classification, blood glucose, album. The nomogram for predicting POP achieved an area under the curve (AUC) of 0.874 (training set) and 0.840 (validation set). Furthermore, C-index, calibration plots and decision curve analysis demonstrated the adequate performance of the nomogram in both the training and validation sets.
The nomogram is a well-calibrated and discriminative tool that can be used to predict the risk of POP in geriatric patients with hip fracture. The nomogram may be useful for identifying patients who may benefit from preventive measures.
The online version contains supplementary material available at 10.1186/s40001-025-02710-4.
Hip fractures are a major health concern for the elderly population, posing significant risks to mobility, independence, and overall well-being [1–3]. With the rising incidence of fractures due to increased life expectancy and osteoporosis, the burden on healthcare systems is growing [4–6]. Currently, early surgical intervention is advocated [7, 8]. However, surgery can also have numerous complications, postoperative pneumonia (POP) stands out as a particularly serious threat [9, 10].
The reported incidence of POP after hip fracture surgery in elderly patients varies considerably, with a wide range of 4.9% to 15.2% [11–16]. Development of postoperative pneumonia results in longer hospital stays, increased costs, and higher mortality [9, 17–21]. Thus, in recent years, there has been an increased clinical focus on the prevention, early diagnosis, and management of POP in elderly patients with hip fractures. Several risk factors for postoperative pneumonia after hip fracture surgery have been identified, including advanced age, poor nutritional status, and pre-existing comorbidities [22, 23]. However, there is currently no accurate and convenient tool to predict an individual patient's risk of developing postoperative pneumonia after hip fracture surgery.
Nomograms are user-friendly graphical tools that allow prediction of a numerical probability or outcome. They can incorporate multiple risk factors to provide individualized risk estimation. Although there have been studies using nomograms to predict postoperative pneumonia in elderly hip fracture patients, the results have been inconsistent and further research is needed.
Therefore, the aim of this study was to develop and validate a nomogram to predict personalized risk of postoperative pneumonia in geriatric hip fracture surgery patients. This nomogram is expected to assist clinicians in identifying high-risk elderly patients and inform preventive strategies to reduce this major surgical complication.
The study protocol was approved and obtained informed consent exemption by institutional ethics committee of Dandong Central Hospital and complied with the principles outlined in the Helsinki Declaration. We retrospectively analyzed the medical records of geriatric patients (≥ 60 years old) diagnosed with hip fractures at the Affiliated Dandong Central Hospital of China Medical University between Feb 2015 and Feb 2024. The studies were conducted by the local legislation and institutional requirements. The need for informed consent from all subjects and/or their legal guardian(s) was waived by the ethical committee board of Dandong Central Hospital due to retrospective nature of the study.
The following exclusion criteria were (1) age < 60 years; (2) patients with a preoperative diagnosis of pneumonia; (3) patients who died within 48 h after surgery; (4) Patients with incomplete medical records; (5) patients with multiple fractures; (6) patients who received conservative treatment. Based on these exclusion criteria, a total of 356 patients were excluded, resulting in a retrospective cohort study involving 1296 patients.
The main outcome was the occurrence of POP for hip fracture. Patients with POP was defined as new infiltrates on chest X-ray postoperatively in patients who did not have pneumonia before surgery. One or more of the following criteria needs to be met [24–26]: (1) new onset or worsening respiratory symptoms such as cough and purulent sputum production; (2) abnormal body temperature, either fever (> 38 °C) or hypothermia (< 36.0 °C); (3) lung abnormalities detected on physical exam such as consolidations or crackles; (4) abnormal white blood cell count indicating leukocytosis (> 10 × 109/L) or leukopenia (< 4 × 109/L); (5) identification of relevant pathogens from sputum or blood cultures.
Please note that this is a simplified overview and specific diagnostic protocols may vary depending on the clinical setting and individual patient presentation. Two senior respiratory physicians made the definitive diagnoses of pelvic organ prolapse by examining the clinical presentations and assessing radiographic imaging of the lungs including X-rays or CT scans.
Our study examined 36 potential risk factors linked to postoperative pneumonias (POP) by analyzing relevant published literature and clinical data from our healthcare facility. The following variables were extracted from patient medical records. Demographic variables encompassed age, gender, smoking habits, alcohol consumption. We considered comorbidities such as hypertension, diabetes, COPD, cardiovascular disease, stroke, dementia, intracerebral hemorrhage, chronic liver disease, chronic kidney disease. Injury-related factors included fracture type, surgical procedure, duration from admission to surgery, intraoperative blood loss, transfusion, postoperative ICU, bedridden time (time from injury to hospital admission), surgery time, and ASA classification. Laboratory parameters obtained within 24 h of admission comprised red blood cell count, white blood cell count, platelet count, hemoglobin, blood glucose level, blood urea nitrogen, creatinine, uric acid, alanine aminotransferase, aspartate aminotransferase, albumin, high-density lipoprotein, triglyceride, fibrinogen and d-dimer.
Categorical variables were expressed as percentages (%) and compared utilizing chi-square tests. Continuous variables were shown as mean ± standard deviation and analyzed by independent sample t-tests.
In the training dataset, LASSO regression was used for variable selection, followed by univariate and multivariate logistic regression analyses to identify independent risk factors for POP. Variables with p < 0.05 in univariate analysis were included in the multivariate logistic regression analysis to pinpoint potential predictors. A predictive nomogram for assessing POP risk was developed utilizing R software, incorporating variables from the multivariate logistic regression model derived from the training dataset. The LASSO regression analysis was implemented using the"glmnet"package in R. Nomogram construction and visualization were conducted using the"rms"package for static nomograms and the"regplot"package for dynamic nomograms. Calibration plots were generated via the"rms"package, while ROC curves and AUC calculations were performed using the"pROC"package. DCA was executed with the"rmda"package. Bootstrap resampling (1000 iterations) for internal validation and bias correction in calibration plots utilized the"boot"package. All graphical outputs, including scatterplots and distribution visualizations, were finalized using"ggplot2". The scatterplot of predicted probabilities visualizes overlap between POP and non-POP groups, revealing the model's discrimination ability.
The predictive nomogram's performance was assessed by generating a receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC) to evaluate sensitivity and specificity. Optimal cutoff values were determined by the Youden index. The cutoff value from the ROC curve of the predictive model was used to divide hospitalized patients into high-risk and low-risk groups to assess the effectiveness of the model in stratifying patients according to risk.
Variance inflation factors (VIFs) were calculated for the multivariate model to evaluate multicollinearity. Calibration plots were inspected by the Hosmer–Lemeshow test to analyze the nomogram’s accuracy. Decision curve analysis (DCA) determined if the predictive model improves net benefit predictions, assessing its clinical utility. To comprehensively assess model performance, ROC analysis, calibration plots, and decision curve analysis were applied in both the training set and validation set.
Statistical analysis used SPSS 27.0 (StatsCorp, Britain) and R 4.1.3 (RStudio Inc., Germany) for nomogram construction.
This study enrolled 1296 geriatric patients with hip fractures, of whom 118 (9.1%) developed postoperative pneumonia (POP). We randomly allocated the patients using a 3 ratio, assigning 908 patients to the training set and 388 patients to the validation set. The training and validation cohorts demonstrated similar baseline characteristics (Table 1). Supplementary eTable 1 presents baseline clinical and demographic characteristics of patients without and with POP in the training set (n = 908) and validation set (n = 388). In the training set, patients with POP (n = 75) were significantly older than patients without POP (n = 833) (81.9 vs 74.0 years, p < 0.001). A greater proportion of patients with POP were male compared to without POP (64.0 vs 39.6%, p < 0.001). Patients with POP also had higher rates of comorbidities like COPD, cardiovascular disease, stroke, chronic liver and kidney disease (all p < 0.05). Similar trends were noted in the validation cohort.Table 1Baseline clinical and demographic characteristics of the training and validation setVariablesTraining set (n = 908)Validation set (n = 388)p valueDemographicAge, × year [Mean (SD)]74.70 (9.54)75.12 (9.92)0.477Male gender (n,%)378 (41.6)161 (41.5)0.964Smoking (n,%)152 (16.7)69 (17.8)0.647Alcohol (n,%)104 (11.5)46 (11.9)0.836ComorbiditiesHypertension (n,%)465 (51.2)182 (46.9)0.156Diabetes (n,%)212 (23.3)88 (22.7)0.794COPD (n,%)106 (11.7)46 (11.9)0.926Cardiovascular disease (n,%)273 (30.1)124 (32.0)0.498Prior stroke (n,%)236 (26.0)99 (25.5)0.858Dementia (n,%)33 (3.6)15 (3.9)0.740Intracerebral hemorrhage (n,%)50 (5.5)22 (5.7)0.906Chronic liver disease(n,%)39 (4.3)17 (4.4)0.944Chronic kidney disease(n,%)50 (5.5)13 (3.4)0.098Type of fractureFemoral neck fracture (n,%)499 (55.0)205 (52.8)0.662Intertrochanteric fracture (n,%)360 (39.6)158 (40.7)Subtrochanteric fracture (n,%)49 (5.4)25 (6.4)TreatmentTotal hip arthroplasty (n,%)121 (13.3)47 (12.1)0.708Hemiarthroplasty (n,%)239 (26.3)92 (23.7)Intramedullary nail (n,%)287 (31.6)126 (32.5)Plate/screw (n,%)114 (12.6)57 (14.7)Multiple screws (n,%)147 (16.2)66 (17.0)Surgical blood loss, × ml [Mean (SD)]171.00 (145.58)184.45 (164.75)0.143Transfusion (n,%)146 (16.1)66 (17.0)0.678Postoperative ICU (n,%)36 (4.1)22 (5.9)0.859Bedridden time, × day [Mean (SD)]5.79 (4.13)6.09 (3.79)0.218Surgery time, × hour [Mean (SD)]1.67 (0.81)1.66 (0.76)0.753ASAIII–V (n,%)443 (48.8)186 (47.9)0.779I–II (n,%)465 (51.2)202 (52.1)Laboratory findingsRBC level, × 10^9^/L [Mean (SD)]3.94 (0.66)3.91 (0.71)0.495WBC level, × 10^9^/L [Mean (SD)]8.92 (2.91)9.20 (3.14)0.128PLT count, × 10^9^/L [Mean (SD)]208.73 (84.59)204.59 (79.28)0.411HGB level, × g/L [Mean (SD)]120.33 (20.26)119.15 (21.22)0.345Blood glucose, × mmol/L [Mean (SD)]6.99 (2.81)6.79 (2.55)0.222BUN, × mmol/L [Mean (SD)]7.29 (4.40)7.53 (5.53)0.454Cr, × μmol/L [Mean (SD)]73.87 (69.53)68.48 (47.61)0.107UA, × mmol/L [Mean (SD)]290.39 (102.73)286.34 (108.54)0.523ALT, × U/L [Mean (SD)]21.50 (41.65)20.89 (48.73)0.818AST, × U/L [Mean (SD)]24.03 (22.56)23.70 (20.98)0.806ALB, × g/L [Mean (SD)]37.94 (4.58)38.04 (4.78)0.705HDL, × mmol/L [Mean (SD)]1.28 (0.45)1.24 (0.41)0.145TG, × mmol/L [Mean (SD)]1.35 (0.78)1.23 (0.89)0.939FIB, × g/L [Mean (SD)]3.69 (1.16)3.56 (1.03)0.052D-Dimer, × mg/L [Mean (SD)]4.80 (5.01)4.87 (5.16)0.812p < 0.05: statistically significant differenceSD standard deviation, COPD chronic obstructive pulmonary disease, ICU Intensive Care Unit, ASA American Society of Anesthesiologists physical status classification, RBC red blood cell count, WBC white blood cell count, PLT platelet count, HGB hemoglobin, Blood glucose blood glucose level, BUN blood urea nitrogen, Cr creatinine, UA uric acid, ALT alanine aminotransferase, AST aspartate aminotransferase, ALB albumin, HDL high-density lipoprotein, TG Triglyceride, FIB fibrinogen
We conducted LASSO regression to identify 27 potential risk factors from the 36 candidates in the training set (Fig. 1). Univariate and multivariate logistic regression analyses were then conducted to further evaluate these factors. After controlling for potential confounding factors, seven factors were identified as independent predictors of POP: age, sex, COPD, postoperative ICU, ASA classification, blood glucose, album. These factors remained significant even after adjusting for other variables (Supplementary eTable 2). Table 2 presents the results of the multivariate logistic regression analysis, including the intercept, β coefficient, and odds ratio. The variance inflation factors (VIFs) for all risk factors ranged from 1.03 to 1.34, indicating no multicollinearity.Fig. 1Scatter plot of the predicted values of the POP group and the non-POP groupTable 2Multivariate analysis of POP in geriatric patients with hip fractures in the training setRisk factorβS.EWaldOR95%CIp valueAge0.0550.01710.0071.061.02–1.090.002Sex (male vs. female)1.1460.59415.1763.151.77–5.60 < 0.001COPD1.5200.31223.8014.572.48–8.42 < 0.001Postoperative ICU1.0180.4205.8722.771.22–6.300.015ASA classification (III–V vs. I–II)1.0930.34510.0292.981.52–5.860.002Blood glucose0.1550.04114.1421.171.08–1.27 < 0.001ALB−0.0920.0385.8650.810.85–0.980.015β, beta; SE, standard error; Wald: Wald statistic; OR, odds ratio; CI, confidence interval; COPD, chronic obstructive pulmonary disease; ICU, Intensive Care Unit; ALB: Albumin; The p-value is used to determine whether the relationship between each independent variable and POP is statistically significantLogit(p) = −6.487 + 0.055Age + 1.146 Sex + 1.520COPD + 1.018 Postoperative.ICU + 1.093* ASA.classification + 0.155* Blood.glucose −0.092*ALB
We developed a nomogram model to predict the risk of POP in elderly hip fracture patients. The model was developed based on a multivariate logistic regression analysis and includes 7 independent risk factors (Fig. 2a, b). Each factor is assigned a score, and the total score reflects the predicted POP risk. For example, a 95-year-old (104 points) male (50 points) patient with a serum albumin level of 15 (83 points), fasting blood glucose of 12 (103 points), ASA grade of 3 (50 points), ICU monitoring (55 points), and no history of COPD (33 points) would have a total score of 478 points, corresponding to a predicted POP risk of 89.6%. Furthermore, a predictive formula was created based on regression coefficients and Logit(p) = −6.487 + 0.055Age + 1.146 Sex + 1.520COPD + 1.018 Postoperative.ICU + 1.093* ASA.classification + 0.155* Blood.glucose − 0.092*ALB. Probability predicted by the nomogram = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \frac{1}{1+{e}^{-Logit(P)}}
### Nomogram model validation We constructed a Nomogram model of POP and employed the variance inflation factors (VIFs) to assess the presence of multicollinearity. The results demonstrated that the VIFs values for all risk factors ranged from 1.03 to 1.34, indicating the absence of multicollinearity in the model. The scatter plot of the new variable generated by the above expression is shown in Fig. 3A, B. The scatterplots of the new variable showed well-separated, non-overlapping distributions in both the training set (POP −1.39 ± 1.43 vs. non-POP −3.73 ± 1.42, p < 0.001) and validation set (POP −1.82 ± 1.52 vs. non-POP −3.79 ± 1.45, p < 0.001). The results suggest that the nomogram exhibits good discrimination ability. The nomogram achieved an area under the curve (AUC) of 0.874 and 0.840 in the training and validation sets, respectively (Fig. 4A, B). Notably, the nomogram's ROC curve surpassed those of individual risk factors (0.602–0.729), as depicted in Fig. 4C. Additionally, based on the cutoff value of the ROC curve, we divided hospitalized patients into high-risk and low-risk groups. The high-risk group had a significantly higher incidence of POP compared to the low-risk group in both the training set (29.3 vs 2.4%, p < 0.001; Fig. 5A; Table 3) and validation set (32.9 vs 5.5%, p < 0.001; Fig. 5B).Fig. 3Nomogram for predicting POP in geriatric patients with hip fracture. **A** Seven variables were included in the nomogram prediction model, age, sex, COPD, postoperative ICU, ASA classification, blood glucose and album.** B** Dynamic nomogram as an example. For example, a 95-year-old (104 points) male (50 points) patient with a serum albumin level of 15 (83 points), fasting blood glucose of 12 (103 points), ASA grade of 3 (50 points), ICU monitoring (55 points), and no history of COPD (33 points) would have a total score of 478 points, corresponding to a predicted POP risk of 89.6%Fig. 4The ROC of the training set (**A**) and validation set (**B**). **C** The ROC curve of seven independent risk factorsFig. 5POP incidence comparison between low-risk and high-risk groups by the cut-off value in the training set (**A**) and validation set (**B**)Table 3POP incidence comparison between low-risk and high-risk groups in training and validation setPOPTraining set (n = 908)p valueValidation set (n = 388)p valueLow-risk group(n = 710)High-risk group(n = 198)Low-risk group(n = 309)High-risk group(n = 79)Yes17 (2.4%)58 (29.3%) < 0.00117 (5.5%)26 (32.9%) < 0.001No693 (97.6%)140 (70.7%)292 (94.5%)53 (67.1%)p-value is from the Chi-Squared Test to indicate significant differentiation (p < 0.05 means significant differentiation) Calibration plots demonstrated close agreement between predicted and observed probabilities (Fig. 6A, B). The Hosmer–Lemeshow goodness-of-fit test revealed no significant deviation from perfect fit for the models in both the training (χ2 = 6.04, df = 8, p = 0.643) and validation sets (χ2 = 4.79, df = 8, p = 0.779), solidifying their calibration. Decision curve analysis further highlighted the consistent advantage of the nomogram in terms of net benefit compared to no assessment across a wide range of threshold probabilities (training 1–81%, validation 2–72%) (Fig. 7A, B).Fig. 6Calibration plot of the nomogram in the training (**A**) and validation (**B**) set. Predictions generated from the model are plotted against actual patient outcomes. The dotted line represents the perfect model calibration. The red line (apparent) indicates calibration when the model is applied to each set, and the green line (bias-corrected) indicates calibration when the model is applied to the bootstrap setFig. 7Decision curve analysis of the nomogram in the training set (**A**) and validation set (**B**). The blue line displays the net benefit of our model. The red line assumes that all patients develop POP. The green line assumes that no patients develop POP Overall, these findings demonstrate the nomogram's ability to accurately predict POP risk in elderly hip fracture patients. Its strong discriminatory power, robust prediction performance, and lack of multicollinearity suggest its potential value as a reliable clinical tool. ## Discussion ### Key findings This study successfully developed and validated a nomogram to predict the risk of postoperative pneumonia (POP) in geriatric patients undergoing hip fracture surgery. The nomogram incorporates seven independent risk age, sex, COPD, postoperative ICU, ASA classification, blood glucose level, and albumin level. The nomogram demonstrated good accuracy and stability in predicting POPs, as evidenced by internal and external validation. The nomogram's predictive accuracy was further verified through calibration plots and decision curve analysis. Importantly, by calculating a total risk score, the nomogram enables individualized risk assessment for a given patient based on their unique combination of characteristics. ### Comparison with existing literature Several studies [13, 27, 28] have attempted to develop nomograms for POP prediction in hip fracture patients, but with varying results and limited validation (Table 4). This study addresses these limitations by including a larger sample size (n = 1296), employing robust statistical methods, and rigorously validating the nomogram's performance in an independent dataset. Additionally, our nomogram incorporates a broader range of risk factors compared to previous studies, potentially enhancing its accuracy and generalizability.Table 4Summarizes the key features of our nomogram in comparison with existing studiesStudySample sizePOPRisk factorsType of validationValidation toolAUCVIFsCurrent study12968.4%Age, sex, COPD, postoperative ICU, ASA classification, blood glucose, albuminExternal validationVIFs valuesC-indexScatterplots of the new variableRisk stratificationROC curveDCA curveHosmer–Lemeshow testCalibration curve0.874 (training), 0.840 (validation)1.03–1.34Zhang et al. (2022)12855.4%COPD, number of comorbidities, ASA classification > 2, preoperative dependent functional status and cognitive impairmentinternal validationC-indexROCDCA curveHosmer–Lemeshow testCalibration curve0.840NAXiang et al. (2020)111314.9%body mass index, serum albumin, c-reactive protein, functional status and time to surgeryinternal validationROC curveDCA curveHosmer–Lemeshow testCalibration curve0.905NAHuang et al. (2023)10088.6%age > 73, time from fracture to surgery (d) > 4 days, smoking, ASA ≥ III level, COPD, hypoproteinemia, red cell distribution width > 14.8%, mechanical ventilation time > 180 min, and stay in the ICUExternal validationVIFs valuesROC curveHosmer–Lemeshow test0.891 (training), 0.881 and 0.843 (validation)1.0–1.2 ### Age Age was identified as an independent risk factor for postoperative pneumonia in the nomogram model in this study. This is in line with previous studies, which have consistently reported an increased risk of POP in geriatric patients. A recent study reported that age was an independent risk factor for POP in this population (OR = 4.95, 95% CI 3.22–6.69, p < 0.0001) [29]. Similar findings were reported by Kyun-Ho Shin et al., who found that age (OR, 1.062; 95% CI 1.022–1.103; p = 0.002) was an independent risk factor for postoperative pneumonia [30]. A retrospective study by Seong-Eun Byun et al. further demonstrated that the mean age of patients with POP after hip fracture was significantly higher than that of the control group (83.7 vs. 78.6 years, p < 0.001) [19]. There are several potential reasons for this Advancing age leads to declines in respiratory defenses, resulting in impaired clearance of aspirated pathogens and increased susceptibility to pulmonary infections in the elderly [31]. Additionally, chronic conditions like COPD and diabetes that commonly accompany old age compromise immunity and predispose to postoperative pneumonia [32]. The trauma of surgery and poorer neuromuscular function in the elderly can cause inadequate ventilation, delayed cough reflex, and weak cough postoperatively, leading to retained aspirate and subsequent infection [33]. Age-related malnutrition, anemia, and hypoproteinemia also weaken immune responses [34]. Prolonged postoperative immobility in older patients further elevates pneumonia risk [35]. With these multifaceted effects of aging weakening host defenses, the elderly are at greater risk of developing postoperative pneumonia compared to younger patients. ### Sex Our study revealed that male patients faced a significantly higher risk of POP compared to females. This finding is consistent with previous research, which has identified sex as an independent risk factor for POP [36]. For example, A retrospective cohort study by Salarbaks et al. found that the incidence of postoperative pulmonary infection in elderly male patients with hip fracture was 2.21 times higher than that in females [14]. This finding is further supported by a retrospective analysis of 984 elderly patients by Zhao et al., which demonstrated a significantly higher incidence of POP in male patients (OR = 2.22, 95% CI 1.28–3.86) [37]. Weaker immune function in men compared to women, stemming from differences in sex hormones. Male hormones like testosterone have immunosuppressive effects, while female hormones like estrogen enhance immune defenses [38, 39]. This results in men having poorer clearance of lung pathogens. In addition, behavioral factors more prevalent among elderly men that impair respiratory health, especially smoking and alcohol consumption. Smoking damages airway cilia and alcohol suppresses immunity, leading to increased infection susceptibility [40, 41]. ### COPD COPD emerged as an independent risk factor for postoperative pneumonia in our study's predictive model (OR = 4.57, 95% CI 2.48–8.42, p < 0.001). Our findings align with previous studies [42–44]. For example, a meta-analysis found COPD was associated with 2.42-fold higher odds of developing pneumonia after hip fracture surgery (OR: 2.42, 95% CI 1.82–3.24) [45]. Han et al. reported a high greater incidence of POP in COPD group compared to non- COPD group (OR: 2.05; 95% CI 1.43–2.94; p < 0.01) [46]. The mechanistic basis for COPD increasing postoperative respiratory infection susceptibility is multifactorial. The chronic airway inflammation and remodeling in COPD leads to impaired mucociliary clearance [47, 48]. COPD patients also exhibit reduced cough reflexes, dysfunctional immune responses, and increased bacterial colonization, which hinder clearing of aspirated pathogens [49, 50]. Coexisting nutritional deficiencies and muscle dysfunction further weaken their defenses [51]. Oxygen desaturation and inadequate lung expansion due to poor respiratory mechanics also raise risks [52]. ### Postoperative ICU Our study identified postoperative ICU stay as an independent predictor of POP in elderly hip fracture patients (OR = 2.77, 95% CI 1.22–6.30, p < 0.001). Prior research corroborates postoperative ICU monitoring as a risk factor for pneumonia across various surgeries. A retrospective review found admission to ICU after hip fracture surgery was associated with 3.25-fold higher odds of POP (OR = 3.25, 95% CI 1.77–5.96) [53]. Huang et al. in elderly patients with hip fractures also showed postoperative ICU were independent risk factors for POP (OR = 3.343, 95% CI 1.803–6.199, p < 0.001) [28]. Preventive measures like lung-protective ventilation, oral decontamination, and early mobilization are especially pertinent for postoperative ICU patients [54]. ICU stay as a key modifiable risk to address through proper protocols and pneumonia bundles. Reducing ICU duration and avoiding intubation when possible can provide benefit [55]. The ICU environment and related care factors increase vulnerability through several mechanisms. Prolonged immobilization in ICU promotes atelectasis and impaired cough, leading to aspiration [56, 57]. Endotracheal intubation and mechanical ventilation also elevate risks of ventilator-associated pneumonia [58]. Additionally, exposure to multidrug resistant bacteria is heightened in ICUs, while invasive devices breach defenses [59]. ICU patients experience stress and compromised nutrition which further weaken immunity [60]. ### ASA classification Our study demonstrated higher ASA classification as an independent predictor of postoperative pneumonia in elderly hip fracture patients (OR = 2.98, 95% CI 1.52–5.86, p = 0.002). The American Society of Anesthesiologists (ASA) physical status classification system is a widely used tool for predicting the risk of postoperative complications, including pneumonia [61]. ASA grade represents the patient's preoperative physical status and comorbidities. Higher grades indicate greater systemic disease burden, which raises postoperative complications through several mechanisms [62]. Multiple chronic illnesses like heart disease, obesity, and diabetes impair immunity and respiratory mechanics. Poorer baseline functional capacity also hinders recovery [63]. Additionally, some comorbidities directly increase aspiration risks due to effects on mentation, gag reflexes, and gastric emptying [64]. Prior studies on hip fracture patients align with our findings. A nationwide cohort study by Meyer et al. demonstrated that a high American Society of Anesthesiologists (ASA) physical status classification (III/IV) is associated with an increased risk of postoperative pneumonia (POP). Specifically, individuals with ASA IV classification had a nearly three-fold increased risk of developing pneumonia compared to the control group (OR = 2.87, 95% CI 2.19–3.78) [65]. A review cohort study of 1008 elderly hip fracture patients found ASA ≥ III versus I-II was associated with 2.054 times higher odds of postoperative pneumonia (OR = 2.054,95% CI 1.36–33.096, p = 0.001) [28]. Another investigation identified ASA classification over 2 as an independent risk factor. Patients with postoperative pneumonia (POP) had a significantly higher proportion of ASA classification above grade 2 compared to those without POP (61.4 vs. 26.9%, p < 0.001) [66]. Our results reaffirm the value of preoperative ASA assessment as a simple predictor of postoperative pneumonia after hip fracture surgery. ASA grade can guide targeted preventive interventions in high-risk patients. ### Blood glucose Our study identified elevated preoperative blood glucose as an independent predictor of postoperative pneumonia in elderly hip fracture patients. Hyperglycemia impairs immune defenses against infection through multiple mechanisms, including reduced neutrophil function, depressed humoral immunity, and impaired complement activation [67, 68]. High blood glucose also increases risk of aspiration pneumonia by delaying gastric emptying and decreasing gag reflexes [69]. While most studies have identified diabetes as an independent risk factor for pneumonia after hip fracture surgery, few have specifically examined the impact of blood glucose levels on this outcome. Prior few research in hip fracture populations also found hyperglycemia associated with heightened postoperative pneumonia risk. For example, a propensity score-matched study conducted by Tang et al. found admission hyperglycemia in elderly hip fracture patients increases the risk of postoperative pneumonia (OR = 2.090, 95% CI 1.135–3.846, p = 0.016) [70]. Chang et al. revealed hyperglycemia (> 200 mg/dL) were identified as risk factors for pneumonia (OR = 24.75, p < 0.001) [17]. Another investigation noted poor perioperative glycemic control was an independent predictor of respiratory infections [71]. A population-based case–control study revealed type 1 and type 2 diabetes are risk factors for a pneumonia-related hospitalization and poor long-term glycemic control among patients with diabetes clearly increases the risk of hospitalization with pneumonia [72]. ### Albumin Our study identified preoperative hypoalbuminemia as an independent predictor of postoperative pneumonia in elderly hip fracture patients. Reduced serum albumin levels impair host defense through multiple pathophysiological First, albumin exerts antioxidant effects and binds pathogens/toxins [73, 74]. Second, by maintaining colloid osmotic pressure, albumin prevents pulmonary interstitial edema; its depletion disrupts alveolar-capillary barrier integrity and impairs respiratory clearance mechanisms [75]. Third, as a critical transporter of immunomodulatory molecules including vitamins, hormones, and medications hypoalbuminemia induces multidimensional immune dysregulation [76]. These synergistic mechanisms collectively establish the biological basis for increased pneumonia susceptibility. Previous studies in hip fracture populations also found associations between preoperative albumin and postoperative pneumonia. A study by Xiang et al. demonstrated that hypoalbuminemia is associated with an increased risk of postoperative pneumonia (POP) (OR = 0.86, 95% CI 0.79–0.93, p < 0.001) [77]. Wang et al. and Bohl et al. observed that the low albumin levels (OR = 10.16, p = 0.001) were independent risk factors for POP in hip fracture patients [78, 79]. ### Limitations While nomograms are gaining traction as noninvasive tools for clinical prediction, this study presents limitations to consider. Firstly, its single-center design potentially restricts the generalizability of findings to diverse populations and healthcare settings. Secondly, the retrospective nature introduces potential bias and limits causal inference between identified risk factors and POP occurrence. Thirdly, while the nomogram demonstrates good performance, its external validation was limited to populations from the same geographic region. Further validation on diverse populations is needed to confirm generalizability. Furthermore, the nomogram's utility should be evaluated by measuring the effectiveness of risk-based interventions. Other promising predictive variables like frailty could be incorporated to potentially improve performance. Overall, this study provides preliminary evidence supporting a nomogram for individualized POP risk assessment after hip fracture surgery. Following external validation, the model can potentially be implemented clinically to guide care of this vulnerable population. ### Future directions Future studies should prospectively evaluate the impact of the nomogram on clinical practice and patient outcomes. Additionally, research could explore the cost-effectiveness of using this nomogram in routine clinical care. Integrating the nomogram into electronic health records could further facilitate its implementation and improve its clinical utility. ## Conclusion In this study, we developed and validated a nomogram to predict individual risk of pneumonia after hip fracture surgery in elderly patients. This nomogram incorporates seven independent risk factors including age, sex, COPD, postoperative ICU, ASA classification, blood glucose and album which accurately discriminated risk in both the development and validation cohorts. It has the potential to aid clinical decision-making through identifying high-risk patients who may benefit from preventive interventions. ## Supplementary Information Additional file 1.