Prognostic Nutritional Index Is a Better Predictor of Perioperative Adverse Outcomes in Patients With Femoral Fractures: a Retrospective Study

Background: Surgery is usually the best treatment for patients with femoral fractures. However, the incidence of perioperative adverse outcomes in such cases is quite high. Nutrition has a major inuence on fracture healing, and malnutrition is associated with higher complication rates, higher mortality rates, and longer hospitalisation periods. In this study, we aimed to identify independent risk factors and assess the predictive value of the prognostic nutritional index (PNI) for perioperative adverse outcomes in patients with femoral fractures. Methods: This retrospective observational study included 343 patients who underwent surgery for a single femur fracture at the Aliated Hospital of Zunyi Medical University in 2018. Binary logistic regression analysis was applied to identify signicant independent risk factors. The discriminatory ability of independent predictors was assessed using the receiver operating characteristic curve analysis, and DeLong's test was used to compare the area under the curve (AUC). Results: In total, 159 patients (46.4%) had perioperative adverse outcomes. PNI (OR: 0.819, 95% CI: 0.754–0.889, P < 0.001), age (OR: 1.042, 95% CI: 1.020–1.066, P < 0.001), time to admission (OR: 1.404, 95% CI: 1.117–1.765, P = 0.004), hypertension (OR: 1.912, 95% CI: 1.049–3.488, P = 0.034), combined injures (OR: 2.739, 95% CI: 1.338–5.607, P = 0.006), and operation types (OR: 3.696, 95% CI: 1.913–7.138, P < 0.001) were independent factors for perioperative adverse outcomes. Based on the AUC (PNI: 0.772, 95% CI: 0.723–0.821, P < 0.001; age: 0.678, 95% CI: 0.622–0.734, P < 0.001; time to admission: 0.585, 95% CI: 0.525–0.646, P 0.006), the PNI had the optimal discrimination ability, indicating its superiority over other independent predictors (age vs. PNI, P = 0.002; time to admission vs. PNI, P < 0.001). Conclusions: This study showed that the PNI was a better and effective independent predictor of perioperative adverse outcomes in patients with femoral fractures. Our ndings suggest that nutritional assessment at admission and appropriate intervention strategies are necessary for patients with femoral fractures. Categorical parameters are presented as frequencies and associated percentages. A Student's t-test was used to analyse continuous variables with normal distribution, and a Mann-Whitney U test was used to analyse continuous variables with non-normal distribution and ordinal variables (ASA, surgical grades, and time to admission). A chi-squared or Fisher's exact test was employed to analyse categorical variables. In these analyses, variables with unadjusted P < 0.10 were identied as confounding factors and were included in multivariate regression analyses to determine independent predictors of perioperative adverse outcomes. The results are expressed as odds ratios (ORs) and 95% condence intervals (95% CIs). The discriminatory ability of independent predictors was assessed using the receiver operating characteristic (ROC) curve analysis. Optimal cut-off values were obtained using the Youden index, and DeLong's test was used to compare the area under the curve (AUC) with MedCalc statistical software version 19.3.1 (MedCalc Software Ltd., Ostend, Belgium). A P value of < 0.05 was considered statistically signicant. All tests were two-sided. All statistical analyses were conducted using Statistical Package for Social Sciences version 17.0 (IBM SPSS Statistics NY,

injuries, accounting for approximately 2% of all femoral injuries (10), and often develop as a result of vehicular trauma or sports activities with varus or valgus impact at the knee (10).
Surgery is usually the best treatment, and it is often performed for patients with femoral fractures.
However, the incidence of perioperative adverse outcomes is quite high, including lower limb vein thrombus or pulmonary embolism, pneumonia, incision disunion or infection, acute exacerbation of underlying chronic diseases, transfer to the intensive care unit (ICU), and even death. Age, trauma, stress, surgery, anaemia, bleeding, infection, pain, activity limitation, and a bedridden state are commonly considered to be the causes. Such patients can be at risk of protein catabolism and malnutrition.
Nutrition has a major in uence on fracture healing, and fracture healing impairment has been observed in the malnourished and undernourished population (2,11,12). Protein-depleted patients with a hip fracture have shown higher complication rates and longer hospitalisation periods (11). Notably, Hughes et al. showed that nutritional improvement led to increased muscle mass in the leg and greater bone mineral density in the fractured callus in protein-malnourished rats with femoral fractures (2). Nevertheless, only a few studies have investigated the impact of nutrition on adverse outcomes in patients with femoral fractures, speci cally during the perioperative stage.
The prognostic nutritional index (PNI), initially proposed by Buzby et al. (13), is a comprehensive index for evaluating the preoperative nutritional status of surgical patients (14,15) and for predicting the risk of postoperative complications (14)(15)(16)(17). Currently, a low PNI, as a proxy of subpar perioperative nutritional status, is reported to be a signi cant predictor of poor postoperative outcomes and increased mortality in various malignancies (16,17). However, no studies on PNI have focused on perioperative adverse outcomes in patients undergoing surgery for femoral fractures. Therefore, in this retrospective study, we aimed to determine the independent risk factors and evaluate the predictive value of the PNI for perioperative adverse outcomes in patients with femoral fractures.

Data source
The data for this retrospective observational study were extracted from the Hospital Information System (HIS; TianJian Technology Co., Ltd., Beijing, China) and Anesthesia Information Management System (AIMS; Medical System Technology Co., Ltd., Suzhou, Jiangsu, China). The HIS and the AIMS, which maintain a complete record of healthcare services, are electronic medical record management systems for hospitals in China.

Patients
A retrospective review was performed using data from a database of 446 patients who underwent surgery for a single femur fracture during hospitalisation between January 2018 and December 2018 at the A liated Hospital of Zunyi Medical University. The case de nition of femur fracture was based on speci c diagnosis codes from the International Classi cation of Diseases, Tenth Revision (ICD-10, S72).

Statistical analysis
Continuous data are presented as mean with standard deviation or median (interquartile range) according to statistical distribution (assumption of normality assessed using the Kolmogorov-Smirnov test). Categorical parameters are presented as frequencies and associated percentages. A Student's t-test was used to analyse continuous variables with normal distribution, and a Mann-Whitney U test was used to analyse continuous variables with non-normal distribution and ordinal variables (ASA, surgical grades, and time to admission). A chi-squared or Fisher's exact test was employed to analyse categorical variables. In these analyses, variables with unadjusted P < 0.10 were identi ed as confounding factors and were included in multivariate regression analyses to determine independent predictors of perioperative adverse outcomes. The results are expressed as odds ratios (ORs) and 95% con dence intervals (95% CIs). The discriminatory ability of independent predictors was assessed using the receiver

Confounding and independent factors
As shown in Table 1, age; hypertension; combined injuries; time to admission; ASA classi cation; operative types; the ratio of blood transfusion; intraoperative blood loss; ALB, PAb, and HB levels; LYM counts; NEUT counts; and PNI were associated with adverse outcomes (all P values < 0.10). All the above confounding factors, except ALB levels (which showed collinearity with PNI), were included in multivariate regression analyses to determine independent predictors. As shown in

Discussion
This study found that PNI, age, time to admission, hypertension, combined injuries, and operation types were independent factors for perioperative adverse outcomes in patients with femoral fractures, and PNI was a better independent predictor than age and time to admission. Our ndings suggest that preoperative nutritional status cannot be underestimated and that interventions should be considered for patients with femoral fractures, where necessary.
The PNI uses the combined effects of hypoalbuminemia and lymphocytopenia to assess the immunologic and nutritional aspects of surgical patients (13,17), and it is a pre-treatment nutritional risk strati cation tool (18). Accumulating evidence indicates that approximately 20-40% of the patients show an acute, prolonged, and profound decrease in serum albumin levels after surgery (17), and this is even more serious in patients with femoral fractures; lower serum albumin levels and total lymphocyte counts are important risk factors for predicting the 1-year mortality of elderly patients with intertrochanteric fractures (19). Consistently, in this study, multivariate regression analysis showed that the nutritional status at admission was negatively correlated with perioperative adverse outcomes in patients with femoral fractures. Although age (P < 0.001), time to admission (P = 0.004), hypertension (P = 0.034), combined injures (P = 0.006), and operative types (P < 0.001) were also independent predictors, they are factors that are not necessarily controllable. Notably, the ROC curve analysis showed that PNI could provide more accurate prediction than other independent predictors in this study, including age and time to admission. Because the ALB level is one main measure included in the PNI (in addition to lymphocyte concentration), the ALB level might also be associated with adverse outcomes. Nevertheless, the PNI was superior to ALB in predicting perioperative adverse outcomes in patients with femoral fractures (P = 0.038). Collectively, PNI was a better and regulatable predictor of perioperative adverse outcomes. Therefore, nutritional assessment at admission and appropriate intervention strategies are necessary for patients with femoral fractures.
The conservative therapy of traditional Chinese orthopaedics is easily accepted by patients (particularly the elderly population) in China, especially in minority nationality regions (Guizhou is a multi-ethnic province of China). Some patients might choose conservative treatment using Chinese medicine in Chinese medical clinics or at home before a surgical operation. Therefore, it is reasonable that some patients were admitted to the hospital after sustaining the fracture for > 22 days in this study. Although traditional Chinese medicine has a long history and has curative effects on fracture rehabilitation, our ndings suggest that early admission after trauma (within 2 days) for surgical treatment is bene cial to patients with femoral fractures, especially elderly patients.
Hypertension is closely related to vascular endothelial cell injury and is often accompanied by dyslipidaemia, and both vascular endothelial cell injury and dyslipidaemia are associated with the formation of venous thrombus. In this study, we observed that thrombus accounted for 77.4% (123/159) of perioperative adverse outcomes. This nding suggests that hypertension is an independent risk factor for perioperative adverse outcomes.
Operation types were classi ed into only two primary categories in this study: hemi/total hip replacement and internal xation (mainly consisting of intramedullary nailing, cannulated-screw, and plate-screw internal xation). The former is primarily performed in elderly patients with proximal femoral fractures, and the latter is performed commonly in younger or non-hip fracture patients. There are differences in incision, operating time, degree of ache, blood loss, and hospital stay among patients treated by different surgical methods. This present study comprehensively evaluated the above-mentioned surgery-related factors using univariate and multivariate analyses. The results indicated that the number of patients with femoral fractures who underwent internal xation was 3.7 times as many as that of patients with femoral fractures who underwent hemi/total-hip replacement. The possible reasons were more severe pain, bleeding, in ammation, activity limitation, and a longer bedridden period in internal xation-treated patients than in hemi/total-hip replacement patients.
There are several limitations of this study. First, the body mass index (BMI) was not evaluated in this study. The BMI is an indicator for the assessment of nutritional status and a good predictor of morbidity and mortality (20); however, the height values were not documented in this study, mainly because patients with femoral fractures were unable to stand up to provide an accurate height measurement. Second, the lipid pro le was not measured in most enrolled patients. Further studies are needed to evaluate the lipid pro le (total cholesterol, triglycerides, and lipoprotein levels) because the lipid pro le is associated with the risk of venous thrombus (21)(22)(23). Third, the duration that a patient is bedridden is closely associated with lower limb vein thrombus, hypostatic pneumonia, and bedsores, and it was not assessed in this retrospective study. Fourth, we did not observe the long-term complications and mortality. Further studies are needed to evaluate the aforementioned issues and explore the effects of preoperative nutritional treatment on perioperative adverse outcomes in patients with femoral fractures.

Conclusions
In conclusion, this study showed that age, hypertension, combined injuries, and internal xation were independent risk factors for perioperative adverse outcomes in patients with femoral fractures, and earlier admission to hospital for treatment and improved preoperative nutrition were associated with a signi cant decrease in the incidence of perioperative adverse outcomes. Notably, the PNI was a better and regulatable independent predictor of outcomes. Our ndings suggest that preoperative nutritional interventions should be considered for patients with femoral fractures. Availability of data and materials

List Of Abbreviations
The datasets used and analysed during the current study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported by a grant from Chengdu Municipal Health Commission (NO. 2020134).
Authors' contributions MH and QF designed and oversaw the study, prepared the statistical analysis plan, conducted the data analysis, and wrote the manuscript; YZ, DL, XL, SX, JP, and ZZ contributed to the design; ZZ revised the paper and had primary responsibility for the nal content. All authors read and approved the nal manuscript.