Risk factors and predictive nomogram of human serum albumin infusion after total hip arthroplasty for femoral head necrosis Authors: Haixiang Miao, Han Gong, Weishi Hong, Zhiwei Peng, Jihang Dai Published: 2025 ABSTRACT PurposePatients with femoral head necrosis (FHN) represent a distinct subgroup undergoing total hip arthroplasty (THA), often characterized by younger age, specific etiologies, and potentially higher nutritional risks. This study aimed to develop and validate a nomogram for predicting human serum albumin (HSA) infusion after THA in this FHN population.Patients and methodsA retrospective analysis was conducted on 584 patients who underwent primary THA due to femoral head necrosis at Jiangsu Provincial Subei People’s Hospital from January 2018 to April 2022. Univariate and multivariable logistic regression analyses, along with LASSO regression, were employed to determine independent risk factors. A predictive model incorporating nomogram was constructed and its performance evaluated using ROC curves, calibration curves, and decision curve analyses.ResultsAmong the 584 enrolled patients, 107 (18.3%) received human serum albumin infusion postoperatively. This cohort was then randomly split into a modeling set (n = 409) for model development and a validation set (n = 175) for internal validation. Seven independent risk factors were identified: age, BMI, surgical duration, preoperative PTINR, preoperative RBC count, preoperative HCT, and preoperative serum albumin levels. The model demonstrated high accuracy with an AUC of 0.854 (95% CI 0.804–0.903) in the modeling group and 0.790 (95% CI 0.700–0.881) in the validation group. Calibration curves indicated good agreement between predicted and observed outcomes.ConclusionThe proposed nomogram, based on seven independent risk factors, effectively predicts the risk of post-THA hypoalbuminemia requiring albumin infusion in patients with femoral head necrosis. Utilizing this tool to mitigate hypoalbuminemia is potentially clinically significant for lowering healthcare costs, minimizing transfusion-related complications, and facilitating enhanced recovery.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40001-025-03500-8. Introduction Femoral head necrosis is a debilitating hip condition, often caused by factors such as excessive alcohol consumption, hormonal use, and trauma [1–3]. Total hip arthroplasty (THA) is an effective treatment for its end stages, proven to relieve pain, restore function, and improve quality of life [4]. However, a significant clinical concern following THA in these patients is postoperative hypoalbuminemia (plasma albumin < 35 g/L), which has been reported to occur in a substantial proportion of cases [5, 6]. Postoperative hypoalbuminemia is a well-established independent risk factor for complications across surgical disciplines, and its prevention is a key goal of perioperative nutritional optimization [7, 8]. Postoperative hypoalbuminemia is associated with an increased susceptibility to infections, prolonged recovery, and a higher rate of complications, including severe sequelae and mortality [9–12]. Consequently, it has garnered significant clinical attention [13]. Human serum albumin (HSA) infusion is commonly used to correct severe hypoalbuminemia, such as at levels of ≤ 30 g/L. However, its application is not without controversy, involving considerations of cost-effectiveness and varying institutional guidelines [14]. The use of HSA not only increases hospitalization costs but may also reflect and contribute to poorer patient outcomes [15]. Nomogram is an intuitive and easy-to-understand data visualization tool that performs complex multifactor calculations, combining multiple predictors into one. Currently, more preoperative hypoalbuminemia complications have been studied, and fewer studies on risk factors for HSA infusion after THA for hip necrosis. Therefore, the aim of this study was to find the incidence and risk factors of HSA infusion in the short term after undergoing total hip arthroplasty for femoral head necrosis by retrospective analysis, which was assessed by establishing a prediction model of the nomogram and validated. In this way, patients with high postoperative risk can be screened out early, and a more rational treatment plan can be formulated for them to reduce the use of HSA in the postoperative period and to avoid the negative effects caused by the above complications. Material and methods Patients The present investigation collated data from 603 patients who underwent primary hip arthroplasty as a remedial measure for femoral head necrosis between January 2018 and April 2022 at Subei People’s Hospital in Jiangsu Province. Inclusion criteria included (1) patients diagnosed with aseptic necrosis of the femoral head with indications for surgery by comprehensive history and imaging; (2) no history of previous surgery on the affected hip; (3) no serious cardiovascular or cerebral vascular problems in the six months prior to the operation; and (4) true and complete information on medical records. Exclusion criteria were (1) preoperative albumin infusion; (2) severe hepatic and renal insufficiency and other diseases affecting metabolism; (3) perioperative contraindications; and (4) missing clinical data information. This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Yangzhou University Subei People’s Hospital (acceptance number No. 2022KY212). After applying the exclusion criteria and removing cases with missing data, a final cohort of 584 patients were included in this study. The participant selection process is detailed in Fig. 1.Fig. 1Flow chart Flow chart Data collection A plethora of variables were scrutinized, including but not limited to, age, gender, body mass index (BMI), hypertension, diabetes mellitus, preoperative red blood cell count (RBC), preoperative hemoglobin levels (HB), preoperative platelet counts (PLT), preoperative prothrombin time (PT), preoperative activated partial thrombin time (APTT), and preoperative blood pressure (BP). Further, additional coagulation parameters such as international normalized ratio (INR), thrombin time (TT), fibrinogen (FIB), D-dimer (DD), and preoperative albumin levels were evaluated alongside the American Society of Anesthesiologists (ASA) score, operative duration, intraoperative hemostatic metrics, and postoperative albumin levels. For patients with preexisting hypertension and diabetes mellitus, pharmaceutical interventions were employed to achieve optimal preoperative blood pressure (< 140/90 mmHg) and fasting blood glucose levels (< 7.3 mmHg). Data were meticulously collated by two independent surgeons via the Hospital Information System (HIS) system. Any discordant data points were reconciled through consensus between the initial two surgeons and, if necessary, adjudicated by a third surgical consultant. All surgical procedures were executed by a singular senior attending surgeon and his surgical consortium. Uniform postoperative clinical management protocols were instituted for all patients. Protein management In this study, the administration of human serum albumin (HSA) was guided by a standardized clinical approach aligned with Enhanced Recovery After Surgery (ERAS) principles. The decision to infuse HSA was not based solely on a biochemical threshold but was reserved for patients with a postoperative albumin level ≤ 30 g/L who concurrently exhibited specific clinical barriers to recovery. These objective indications included (1) severe tissue edema impairing early ambulation, (2) excessive wound exudate threatening healing, (3) a planned re-operation, or (4) hemodynamic instability. This protocol ensured a consistent and reproducible decision-making process across our institution. Patients meeting both the biochemical and clinical criteria constituted the HSA infusion cohort, while all others were assigned to the control cohort. The demographic and clinical characteristics of these cohorts are elucidated in Table 1. Table 1Demographic and clinical parameters among postoperative patients receiving protein infusion versus those devoid of protein infusionParameterNo albumin infusion (n = 477)Albumin infusion (n = 107)Age [year, X ± SD]61.71 ± 11.6665.89 ± 10.35BMI [kg/m2, X ± SD]24.64 ± 3.3223.59 ± 3.44Operation time [min, X ± SD]79.56 ± 27.14101.39 ± 35.52Preoperative RBC [*10^12/L, X ± SD]4.56 ± 0.594.10 ± 0.71Preoperative HGB [g/L, X ± SD]136.02 ± 15.93131.19 ± 18.61Preoperative HCT [%, X ± SD]40.54 ± 4.4338.46 ± 5.10Preoperative PLT [*10^9/L, X ± SD]210.52 ± 70.21204.94 ± 62.04Preoperative PT [s, X ± SD]12.74 ± 0.9212.90 ± 0.73Preoperative PTINR [1 ± SD]0.96 ± 0.090.99 ± 0.07Preoperative APTT [s X ± SD]37.17 ± 4.3536.66 ± 4.69Preoperative TT [s, X ± SD]17.18 ± 1.1617.15 ± 1.18Preoperative FIB [g/L, X ± SD]3.60 ± 0.763.92 ± 1.10Preoperative DD polymer [ug/ml, X ± SD]0.91 ± 1.201.38 ± 1.91Preoperative serum calcium [mmol/L, X ± SD]2.37 ± 0.102.34 ± 0.11Preoperative CRP [mg/L, X ± SD]5.20 ± 8.8510.08 ± 21.39Preoperative ESR [mm/h, X ± SD]19.85 ± 18.7227.45 ± 27.06Postoperative HGB [g/L, X ± SD]109.12 ± 15.69105.20 ± 17.56Postoperative HCT [%, X ± SD]33.20 ± 4.4632.16 ± 4.98Postoperative serum calcium [mmol/L, X ± SD]2.12 ± 0.102.10 ± 0.12Transfusion volume [u, X ± SD]0.59 ± 1.290.99 ± 1.56Intraoperative bleeding [ml, X ± SD]180.40 ± 98.18197.84 ± 213.67Preoperative systolic blood pressure [mmHg, X ± SD]117.68 ± 33.24107.87 ± 38.94Preoperative diastolic blood pressure [mmHg, X ± SD]85.31 ± 26.6791.76 ± 28.46Postoperative systolic blood pressure [mmHg, X ± SD]105.97 ± 30.9996.70 ± 33.12Postoperative diastolic blood pressure [mmHg, X ± SD]48.56 ± 29.3337.80 ± 30.33Intraoperative fluid infusion [ml, X ± SD]953.41 ± 396.08881.49 ± 424.26Additional fluid infusion on the same day [ml, X ± SD]614.24 ± 590.23560.54 ± 623.65Preoperative albumin[g/L, X ± SD]43.38 ± 5.9041.53 ± 3.16Gender [number, (%)] Male211(44.2)45(42.1) Female266(55.8)62(57.9)Surgical side [number, (%)] Left229(48.0)54(50.5) Right248(52.0)53(49.5)Hypertension [number, (%)] Yes198(41.5)50(46.7) No279(58.5)57(53.3)Diabetes [number, (%)] Yes47(9.9)6(5.6) No430(90.1)101(94.4)Cardiovascular [number, (%)] Yes48(10.1)20(18.7) No429(89.9)87(81.3)Drinking history [number, (%)] Yes36(7.5)3(2.8) No441(92.5)104(97.2)ASA [number, (%)] ≥ 3179(37.5)58(54.2) < 3298(62.5)49(45.8)Gastrointestinal drug use [number, (%)] Yes124(26.0)29(27.1) No353(74.0)78(72.9) Table 1Demographic and clinical parameters among postoperative patients receiving protein infusion versus those devoid of protein infusionParameterNo albumin infusion (n = 477)Albumin infusion (n = 107)Age [year, X ± SD]61.71 ± 11.6665.89 ± 10.35BMI [kg/m2, X ± SD]24.64 ± 3.3223.59 ± 3.44Operation time [min, X ± SD]79.56 ± 27.14101.39 ± 35.52Preoperative RBC [*10^12/L, X ± SD]4.56 ± 0.594.10 ± 0.71Preoperative HGB [g/L, X ± SD]136.02 ± 15.93131.19 ± 18.61Preoperative HCT [%, X ± SD]40.54 ± 4.4338.46 ± 5.10Preoperative PLT [*10^9/L, X ± SD]210.52 ± 70.21204.94 ± 62.04Preoperative PT [s, X ± SD]12.74 ± 0.9212.90 ± 0.73Preoperative PTINR [1 ± SD]0.96 ± 0.090.99 ± 0.07Preoperative APTT [s X ± SD]37.17 ± 4.3536.66 ± 4.69Preoperative TT [s, X ± SD]17.18 ± 1.1617.15 ± 1.18Preoperative FIB [g/L, X ± SD]3.60 ± 0.763.92 ± 1.10Preoperative DD polymer [ug/ml, X ± SD]0.91 ± 1.201.38 ± 1.91Preoperative serum calcium [mmol/L, X ± SD]2.37 ± 0.102.34 ± 0.11Preoperative CRP [mg/L, X ± SD]5.20 ± 8.8510.08 ± 21.39Preoperative ESR [mm/h, X ± SD]19.85 ± 18.7227.45 ± 27.06Postoperative HGB [g/L, X ± SD]109.12 ± 15.69105.20 ± 17.56Postoperative HCT [%, X ± SD]33.20 ± 4.4632.16 ± 4.98Postoperative serum calcium [mmol/L, X ± SD]2.12 ± 0.102.10 ± 0.12Transfusion volume [u, X ± SD]0.59 ± 1.290.99 ± 1.56Intraoperative bleeding [ml, X ± SD]180.40 ± 98.18197.84 ± 213.67Preoperative systolic blood pressure [mmHg, X ± SD]117.68 ± 33.24107.87 ± 38.94Preoperative diastolic blood pressure [mmHg, X ± SD]85.31 ± 26.6791.76 ± 28.46Postoperative systolic blood pressure [mmHg, X ± SD]105.97 ± 30.9996.70 ± 33.12Postoperative diastolic blood pressure [mmHg, X ± SD]48.56 ± 29.3337.80 ± 30.33Intraoperative fluid infusion [ml, X ± SD]953.41 ± 396.08881.49 ± 424.26Additional fluid infusion on the same day [ml, X ± SD]614.24 ± 590.23560.54 ± 623.65Preoperative albumin[g/L, X ± SD]43.38 ± 5.9041.53 ± 3.16Gender [number, (%)] Male211(44.2)45(42.1) Female266(55.8)62(57.9)Surgical side [number, (%)] Left229(48.0)54(50.5) Right248(52.0)53(49.5)Hypertension [number, (%)] Yes198(41.5)50(46.7) No279(58.5)57(53.3)Diabetes [number, (%)] Yes47(9.9)6(5.6) No430(90.1)101(94.4)Cardiovascular [number, (%)] Yes48(10.1)20(18.7) No429(89.9)87(81.3)Drinking history [number, (%)] Yes36(7.5)3(2.8) No441(92.5)104(97.2)ASA [number, (%)] ≥ 3179(37.5)58(54.2) < 3298(62.5)49(45.8)Gastrointestinal drug use [number, (%)] Yes124(26.0)29(27.1) No353(74.0)78(72.9) Demographic and clinical parameters among postoperative patients receiving protein infusion versus those devoid of protein infusion Statistical methods Statistical computations were undertaken utilizing SPSS (Version 26, IBM Corp, USA) and R statistical software (Version 4.2.1, R Foundation for Statistical Computing, Austria). Continuous variables underwent analysis via the independent samples t-test, while categorical variables were examined through the chi-square test. The assumption of linearity for continuous variables in the logistic regression model was verified using the Box-Tidwell test. Continuous variables are denoted as mean ± standard deviation (M ± SD), and categorical variables are represented by counts and proportions [N (%)]. In the current study, a p-value of < 0.05 (bilateral) was deemed to manifest statistical significance. The complete dataset was allocated to a modeling subset and a validation subset in a 7:3 ratio, respectively, by computer-generated random numbers in R software (version 4.2.1 R Foundation for Statistical Computing, Austria) to ensure analytical robustness. Data in the modeling subset were deployed for the derivation of the predictive model, whereas the validation subset functioned to corroborate the model’s predictive validity. Variables demonstrating statistical significance (p < 0.05) in the univariate analysis were subsequently incorporated into the Least Absolute Shrinkage and Selection Operator (LASSO) regression to mitigate multicollinearity and refine variable selection. Multivariate logistic regression was then implemented to isolate independent risk factors germane to human serum albumin (HSA) infusion. Columnar line graphs were generated predicated on multivariate logistic regression outcomes. Moreover, Receiver Operating Characteristic (ROC) curves, calibration plots, and Decision Curve Analysis (DCA) were executed for both modeling and validation subsets, facilitated by R software. Results Baseline data This study encompassed a total of 584 participants, of which 477 (81.7%) did not undergo human serum albumin (HSA) infusion, and 107 (18.3%) were administered HSA. The cohort included 256 males (43.84%) and 328 females (56.16%), with ages ranging from 23 to 92 years (Table 1). For the purposes of analytical rigidity, 409 subjects were allocated to the modeling subset, while 175 comprised the validation subset (Table 2). Table 2Comparison of clinical data between modeling and validation groupsVariantModeling group (n = 409)Validation team (n = 175)t/x2PAge [X ± SD]62.59 ± 11.7262.22 ± 11.120.3490.727BMI [kg/m2, X ± SD]24.34 ± 3.3324.70 ± 3.45−1.1750.241Operation time [min, X ± SD]84.79 ± 31.5080.68 ± 26.181.5160.130Preoperative RBC [*10^12/L, X ± SD]4.47 ± 0.634.49 ± 0.66−0.4400.660Preoperative HGB [g/L, X ± SD]134.85 ± 16.82135.80 ± 15.89−0.6350.526Preoperative HCT [%, X ± SD]40.14 ± 4.6540.21 ± 4.58−0.1830.855Preoperative PLT [*10^9/L, X ± SD]207.71 ± 69.45213.66 ± 67.18−0.9580.339Preoperative PT [s, X ± SD]12.77 ± 0.7212.77 ± 1.200.0070.995Preoperative PTINR [1, X ± SD]0.96 ± 0.070.97 ± 0.12−0.7510.453Preoperative APTT [s, X ± SD]37.25 ± 4.4836.66 ± 4.231.4860.138Preoperative TT [s, X ± SD]17.13 ± 1.1017.26 ± 1.29−1.2430.214Preoperative FIB[g/L, X ± SD]3.67 ± 0.843.62 ± 0.850.6000.549Preoperative DD polymer [ug/ml, X ± SD]1.03 ± 1.340.93 ± 1.430.8260.409Postoperative serum calcium [mmol/L, X ± SD]2.36 ± 0.102.37 ± 0.10−0.4760.634Preoperative CRP [mg/L, X ± SD]6.56 ± 13.884.98 ± 7.211.8020.072Preoperative ESR [mm/h, X ± SD]21.68 ± 21.1220.21 ± 19.660.7900.430Postoperative HGB [g/L, X ± SD]107.99 ± 16.22109.37 ± 15.85−0.9490.343Postoperative HCT [%, X ± SD]32.89 ± 4.5733.294 ± 4.56−0.9900.322Postoperative serum calcium [mmol/L, X ± SD]2.11 ± 0.112.12 ± 0.09−1.2320.219Transfusion volume [ml, X ± SD]0.68 ± 1.370.63 ± 1.320.4310.666Intraoperative bleeding [ml, X ± SD]205.43 ± 151.02193.77 ± 113.800.9160.360Preoperative systolic blood pressure [mmHg, X ± SD]133.88 ± 21.72135.27 ± 22.82−0.7000.484Preoperative diastolic blood pressure [mmHg, X ± SD]72.29 ± 12.8371.58 ± 12.660.6180.537Postoperative systolic blood pressure [mmHg, X ± SD]120.93 ± 20.17121.04 ± 20.82−0.0620.951Postoperative diastolic blood pressure [mmHg, X ± SD]65.29 ± 11.9964.16 ± 12.341.0310.303Intraoperative fluid infusion [ml, X ± SD]1000.22 ± 322.65965.14 ± 318.111.2090.227Additional fluid infusion on the same day [ml, X ± SD]858.71 ± 546.39833.75 ± 524.120.5120.609Length of stay [day, X ± SD]8.42 ± 2.807.78 ± 2.352.8460.005Preoperative albumin [g/L, X ± SD]42.58 ± 6.0247.16 ± 3.93−.3.622 < 0.001Gender [number, (%)] Male177(43.3)79(45.1)0.1730.677 Female232(56.7)96(54.9)Surgical side [number, (%)] Left201(49.1)82(46.9)0.2570.612 Right208(50.9)93(53.1)Hypertension [number, (%)] Yes169(41.3)79(45.1)0.7330.392 No240(58.7)96(54.9)Diabetes [number, (%)] Yes36(8.8)17(9.7)0.1240.725 No373(91.2)158(90.3)Cardiovascular [number, (%)] Yes51(12.5)17(9.7)0.9040.343 No358(87.5)158(90.3)Drinking history [number, (%)]Yes24(5.9)15(8.6)1.4370.231No385(94.1)160(91.4)ASA [number, (%)] ≥ 3168(41.1)69(39.4)0.1380.710 < 3241(58.9)106(60.6)Gastrointestinal drug use [number, (%)] Yes106(25.9)47(26.9)0.0560.813 No303(74.1)128(73.1) Table 2Comparison of clinical data between modeling and validation groupsVariantModeling group (n = 409)Validation team (n = 175)t/x2PAge [X ± SD]62.59 ± 11.7262.22 ± 11.120.3490.727BMI [kg/m2, X ± SD]24.34 ± 3.3324.70 ± 3.45−1.1750.241Operation time [min, X ± SD]84.79 ± 31.5080.68 ± 26.181.5160.130Preoperative RBC [*10^12/L, X ± SD]4.47 ± 0.634.49 ± 0.66−0.4400.660Preoperative HGB [g/L, X ± SD]134.85 ± 16.82135.80 ± 15.89−0.6350.526Preoperative HCT [%, X ± SD]40.14 ± 4.6540.21 ± 4.58−0.1830.855Preoperative PLT [*10^9/L, X ± SD]207.71 ± 69.45213.66 ± 67.18−0.9580.339Preoperative PT [s, X ± SD]12.77 ± 0.7212.77 ± 1.200.0070.995Preoperative PTINR [1, X ± SD]0.96 ± 0.070.97 ± 0.12−0.7510.453Preoperative APTT [s, X ± SD]37.25 ± 4.4836.66 ± 4.231.4860.138Preoperative TT [s, X ± SD]17.13 ± 1.1017.26 ± 1.29−1.2430.214Preoperative FIB[g/L, X ± SD]3.67 ± 0.843.62 ± 0.850.6000.549Preoperative DD polymer [ug/ml, X ± SD]1.03 ± 1.340.93 ± 1.430.8260.409Postoperative serum calcium [mmol/L, X ± SD]2.36 ± 0.102.37 ± 0.10−0.4760.634Preoperative CRP [mg/L, X ± SD]6.56 ± 13.884.98 ± 7.211.8020.072Preoperative ESR [mm/h, X ± SD]21.68 ± 21.1220.21 ± 19.660.7900.430Postoperative HGB [g/L, X ± SD]107.99 ± 16.22109.37 ± 15.85−0.9490.343Postoperative HCT [%, X ± SD]32.89 ± 4.5733.294 ± 4.56−0.9900.322Postoperative serum calcium [mmol/L, X ± SD]2.11 ± 0.112.12 ± 0.09−1.2320.219Transfusion volume [ml, X ± SD]0.68 ± 1.370.63 ± 1.320.4310.666Intraoperative bleeding [ml, X ± SD]205.43 ± 151.02193.77 ± 113.800.9160.360Preoperative systolic blood pressure [mmHg, X ± SD]133.88 ± 21.72135.27 ± 22.82−0.7000.484Preoperative diastolic blood pressure [mmHg, X ± SD]72.29 ± 12.8371.58 ± 12.660.6180.537Postoperative systolic blood pressure [mmHg, X ± SD]120.93 ± 20.17121.04 ± 20.82−0.0620.951Postoperative diastolic blood pressure [mmHg, X ± SD]65.29 ± 11.9964.16 ± 12.341.0310.303Intraoperative fluid infusion [ml, X ± SD]1000.22 ± 322.65965.14 ± 318.111.2090.227Additional fluid infusion on the same day [ml, X ± SD]858.71 ± 546.39833.75 ± 524.120.5120.609Length of stay [day, X ± SD]8.42 ± 2.807.78 ± 2.352.8460.005Preoperative albumin [g/L, X ± SD]42.58 ± 6.0247.16 ± 3.93−.3.622 < 0.001Gender [number, (%)] Male177(43.3)79(45.1)0.1730.677 Female232(56.7)96(54.9)Surgical side [number, (%)] Left201(49.1)82(46.9)0.2570.612 Right208(50.9)93(53.1)Hypertension [number, (%)] Yes169(41.3)79(45.1)0.7330.392 No240(58.7)96(54.9)Diabetes [number, (%)] Yes36(8.8)17(9.7)0.1240.725 No373(91.2)158(90.3)Cardiovascular [number, (%)] Yes51(12.5)17(9.7)0.9040.343 No358(87.5)158(90.3)Drinking history [number, (%)]Yes24(5.9)15(8.6)1.4370.231No385(94.1)160(91.4)ASA [number, (%)] ≥ 3168(41.1)69(39.4)0.1380.710 < 3241(58.9)106(60.6)Gastrointestinal drug use [number, (%)] Yes106(25.9)47(26.9)0.0560.813 No303(74.1)128(73.1) Comparison of clinical data between modeling and validation groups Analysis of relevant and independent risk factors in the modeling set Within the modeling cohort of 409 patients, univariate analyses revealed 17 variables that attained statistical significance (p < 0.05). These variables encompassed age, body mass index (BMI), history of cardiovascular disease, duration of surgery, preoperative red blood cell (RBC), hematocrit (HCT), preoperative prothrombin time (PT), prothrombin time international normalized ratio (PTINR), fibrinogen (FIB), D-dimer (DD), calcium, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) level, postoperative calcium, blood transfusion, American Society of Anesthesiologists (ASA) classification and preoperative albumin (Tables 3, S1). Following initial variable screening with LASSO regression (Fig. 2), all selected variables were included. Optimal subsets were ascertained through R language screening (P < 0.05). Multivariable logistic regression ultimately discerned seven autonomous risk factors: age, BMI, duration of surgery, preoperative PTINR, preoperative RBC, preoperative HCT, and preoperative albumin (Tables 4, S2). Table 3One-way regression analysisβS.EWardOR95% CIPAge0.0380.0129.7911.0391.014 ~ 1.0640.002BMI−0.1060.0416.660.8990.829 ~ 0.9750.01Cardiovascular and cerebrovascular diseases0.810.345.6692.2471.154 ~ 4.3760.017Operation time0.0240.00432.5531.0251.016 ~ 1.033 < 0.001Preoperative RBC−1.2760.25824.4660.2790.168 ~ 0.463 < 0.001Preoperative HCT−0.0590.0294.1070.9430.89 ~ 0.9980.043Preoperative PT0.4480.1796.2651.5651.102 ~ 2.2230.012Preoperative PTINR5.4541.8628.58233.7256.078 ~ 8987.5870.003Preoperative FIB0.520.14413.0061.6831.268 ~ 2.233 < 0.001Preoperative DD polymer0.2360.096.9011.2661.062 ~ 1.5090.009Preoperative serum calcium−3.0251.2925.4780.0490.004 ~ 0.6120.019Preoperative CRP0.0250.0097.9731.0251.008 ~ 1.0420.005Preoperative ESR0.0160.0058.4521.0161.005 ~ 1.0270.004Postoperative serum calcium−2.271.1373.9860.1030.011 ~ 0.9590.046Transfusion volume0.2220.0846.9881.2481.059 ~ 1.4710.008ASA0.2880.1265.1791.3341.041 ~ 1.7090.023Preoperative albumin−0.0880.02216.2890.9160.878 ~ 0.956 < 0.001Fig. 2LASSO Analysis for Identification of Significant Variables from Univariate Analysis. a Lasso Coefficient Trajectories; b Lasso Deviance ProfilesTable 4Multifactor regression screening setBS.EWardOR95% CIPAge0.050.0169.6811.0511.019 ~ 1.0840.002BMI−0.1790.0512.8830.8360.758 ~ 0.922 < 0.001Operation time0.0330.00539.6491.0341.023 ~ 1.044 < 0.001Preoperative RBC−1.8460.44717.0340.1580.066 ~ 0.379 < 0.001Preoperative HCT0.1520.0547.7991.1641.046 ~ 1.2950.005Preoperative PTINR5.1792.2915.111177.4381.992 ~ 15,806.260.024Preoperative albumin−0.0750.0278.0270.9280.881 ~ 0.9770.005Constant−3.4823.1121.25210.263 Table 3One-way regression analysisβS.EWardOR95% CIPAge0.0380.0129.7911.0391.014 ~ 1.0640.002BMI−0.1060.0416.660.8990.829 ~ 0.9750.01Cardiovascular and cerebrovascular diseases0.810.345.6692.2471.154 ~ 4.3760.017Operation time0.0240.00432.5531.0251.016 ~ 1.033 < 0.001Preoperative RBC−1.2760.25824.4660.2790.168 ~ 0.463 < 0.001Preoperative HCT−0.0590.0294.1070.9430.89 ~ 0.9980.043Preoperative PT0.4480.1796.2651.5651.102 ~ 2.2230.012Preoperative PTINR5.4541.8628.58233.7256.078 ~ 8987.5870.003Preoperative FIB0.520.14413.0061.6831.268 ~ 2.233 < 0.001Preoperative DD polymer0.2360.096.9011.2661.062 ~ 1.5090.009Preoperative serum calcium−3.0251.2925.4780.0490.004 ~ 0.6120.019Preoperative CRP0.0250.0097.9731.0251.008 ~ 1.0420.005Preoperative ESR0.0160.0058.4521.0161.005 ~ 1.0270.004Postoperative serum calcium−2.271.1373.9860.1030.011 ~ 0.9590.046Transfusion volume0.2220.0846.9881.2481.059 ~ 1.4710.008ASA0.2880.1265.1791.3341.041 ~ 1.7090.023Preoperative albumin−0.0880.02216.2890.9160.878 ~ 0.956 < 0.001 One-way regression analysis LASSO Analysis for Identification of Significant Variables from Univariate Analysis. a Lasso Coefficient Trajectories; b Lasso Deviance Profiles Table 4Multifactor regression screening setBS.EWardOR95% CIPAge0.050.0169.6811.0511.019 ~ 1.0840.002BMI−0.1790.0512.8830.8360.758 ~ 0.922 < 0.001Operation time0.0330.00539.6491.0341.023 ~ 1.044 < 0.001Preoperative RBC−1.8460.44717.0340.1580.066 ~ 0.379 < 0.001Preoperative HCT0.1520.0547.7991.1641.046 ~ 1.2950.005Preoperative PTINR5.1792.2915.111177.4381.992 ~ 15,806.260.024Preoperative albumin−0.0750.0278.0270.9280.881 ~ 0.9770.005Constant−3.4823.1121.25210.263 Multifactor regression screening set Development and validation of a column chart to predict the risk of postoperative hypoalbuminemia in patients with femoral head necrosis A prognostic column chart was meticulously constructed, incorporating seven autonomous predictors: age, Body Mass Index (BMI), surgical duration, preoperative Prothrombin Time International Normalized Ratio (PTINR), preoperative Red Blood Cell (RBC) count, preoperative Hematocrit (HCT), and preoperative serum albumin concentrations (Fig. 3). Within the modeling cohort, the Area Under the Receiver Operating Characteristic Curve (AUC-ROC) registered at 0.854 (95% Confidence Interval [CI] 0.804–0.903), elucidating a high predictive acumen for the requirement of Human Serum Albumin (HSA) infusion following aseptic necrosis of the femoral head (Fig. 4a). In concordance, the calibration curves manifested a commendable congruence between the predicted and empirically observed outcomes (Fig. 4b). Decision Curve Analysis (DCA) corroborated an augmented net benefit for the prediction of postoperative HSA infusion risk using the aforementioned column chart (Fig. 4c).Fig. 3Nomogram for predicting postoperative human serum albumin (HSA) infusion riskFig. 4Performance of the predictive model in the modeling group. a ROC curve; b Calibration plot; c Decision curve analysis Nomogram for predicting postoperative human serum albumin (HSA) infusion risk Performance of the predictive model in the modeling group. a ROC curve; b Calibration plot; c Decision curve analysis For the external validation cohort, the AUC-ROC was quantified to be 0.790 (95% CI 0.700–0.881), substantiating the model’s robustness in the precise prediction of postoperative HSA infusion in patients presenting with femoral head necrosis (Fig. 5a). Moreover, calibration curves delineated an elevated fidelity between the predicted and actual observed outcomes (Fig. 5b). DCA re-affirmed the superior net benefit of employing the prognostic column chart for the risk stratification of postoperative HSA infusion in the clinical setting of femoral head necrosis (Fig. 5c).Fig. 5Performance of the predictive model in the validation team. a ROC curve; b Calibration plot; c Decision curve analysis Performance of the predictive model in the validation team. a ROC curve; b Calibration plot; c Decision curve analysis Discussion Total hip arthroplasty (THA) remains the standard surgical treatment for end-stage femoral head necrosis [16]. While preoperative hypoalbuminemia is widely recognized as a risk factor for postoperative complications, consensus is lacking regarding postoperative albumin infusion criteria [17–20]. In contrast to previous studies on hypoalbuminemia, our study is specifically focused on patients with femoral head necrosis and predicts the clinical decision for HSA infusion rather than merely the laboratory abnormality [21, 22]. This study developed a predictive model for human serum albumin (HSA) infusion following THA in this specific patient population. Our multivariate analysis identified seven independent predictors: age, BMI, operation time, preoperative PTINR, RBC count, hematocrit, and serum albumin. Based on these factors, we constructed and validated a nomogram for clinical use. Our analysis identified advanced age as a significant predictor for postoperative hypoalbuminemia requiring HSA infusion, with patients aged 95 years demonstrating approximately 40 points higher on the nomogram compared to 20-year-old patients. This finding is consistent with existing literature documenting the association between advanced age and increased perioperative risk [23–25]. The substantial point difference in our nomogram underscores the importance of recognizing advanced age as a key risk factor in clinical assessment and preoperative planning. Furthermore, our study identified BMI as another salient risk variable for postoperative hypoalbuminemia. Extant literature posits BMI as a reliable nutritional index; a decrement in BMI ostensibly indicates a compromised physiological state, often characterized by reduced muscle mass, internal milieu perturbations, and a negative nitrogen balance. This, in turn, leads to elevated proteolysis at the metabolic level [26]. Thus, heightened clinical vigilance is warranted, particularly among older patients and those with aberrant BMI values. Mitigating postoperative hypoalbuminemia could be effectively managed through preoperative dietary optimization, bolstered by protein-rich foods, and judicious postoperative nutritional supplementation. The temporal extent of surgical intervention serves as a significant determinant in the manifestation of hypoalbuminemia. This association may be attributable to the corollary that an elongated surgical duration induces more extensive surgical trauma, thereby escalating intraoperative albumin depletion in tandem with blood loss [27]. Concurrently, protracted surgical time potentiates the severity of the postoperative inflammatory milieu. Previous research underscores that endothelial gaps that form during inflammatory episodes, catalyzed by the abundant secretion of inflammatory mediators, augment capillary permeability, thereby precipitating albumin efflux into the interstitial compartments and hastening the onset of postoperative hypoproteinemia [28–30]. Our study corroborates this assertion. Each incremental 120-min prolongation in operative time correspondingly elevated the column chart by approximately 42 points. As a corollary, meticulous preoperative consultations and individualized surgical schemas can ameliorate surgical duration, thereby mitigating the incidence of postoperative albumin diminution. Prothrombin Time International Normalized Ratio (PTINR) functions as a globally recognized metric for hemostatic efficacy. Existing clinical evidence postulates that elongated PTINR durations usually portend attenuated coagulative function, thus amplifying the hemorrhagic risk profile in patients [31]. In this investigation, a heightened PTINR was observed among patients who received albumin transfusions as opposed to their non-transfused counterparts, thereby potentially catalyzing augmented intraoperative blood loss and drainage, often concomitant with escalated albumin attrition. Erythrocytic count serves as an indicator of anemia and generally encapsulates the nutritional state of the patient’s physiological system. A paucity in erythrocytic count generally heralds protein insufficiency and metabolic dysregulation [32]. A decrement of 0.5 × 10^12/L in erythrocytic count led to an approximate 10-point ascension in the specific gravity of the column-line graph. An incremental 5% surge in Hematocrit (HCT) was linked to an approximate 8-point increase in postoperative hypoalbuminemia, a relationship potentially rooted in heightened hemoconcentration and a concomitant decrement in blood colloid content during the preoperative phase. This may culminate in reduced capacity for fluid retention and may be exacerbated by episodes of emesis and diarrhea, thereby reflecting malnutrition and metabolic derangement preoperatively [33]. The paucity of preoperative serum albumin emerges as another salient risk factor for postoperative hypoalbuminemia. Human Serum Albumin (HSA), synthesized hepatically, fulfills critical roles in nutrient transport and osmoregulation. A decrement in preoperative protein concentration compromises capillary filtration pressure, thereby facilitating tissue fluid egress, resulting in wound edema and delayed recuperative processes [18]. Additionally, the metabolic half-life of albumin ranges from 15 to 19 days. Patients manifesting low preoperative albumin concentrations generally exhibit nutritional inadequacy or muscular enfeeblement, rendering them less resilient to surgical interventions and predisposing them to augmented postoperative complications [34–36]. In the scope of this study, a diminution of 5 g/L in preoperative albumin augmented the weight in the column-line graph by an approximate four-point increment, underscoring the potential benefits of preemptive nutritional fortification through protein supplementation in minimizing the likelihood of postoperative hypoalbuminemia. This computational model significantly augments both the predictive and diagnostic accuracy for postoperative hypoalbuminemia subsequent to Total Hip Arthroplasty (THA) for femoral head necrosis. Utilizing this model as a framework, tailored therapeutic regimens can be devised for individual patients, thereby mitigating the incidence of short-term postoperative hypoalbuminemia and concomitant complications. However, this investigation is not without limitations. Primarily, its retrospective nature inherently confines its scope; data sourced from the hospital’s medical record systems were selectively curated, thereby engendering a potential selection bias. Secondly, this study’s single-center design, along with its lack of consideration for operator-related factors such as surgeon volume, limits the generalizability of our findings, as the results reflect the clinical practices of a single institution and require external validation to establish broader applicability. Thirdly, an intrinsic bias may exist in variables such as surgical techniques, infrastructure, and pharmacological agents used, thereby excluding potentially significant variables. Consequently, there exists an exigency to expand both the sample size and conduct multicentric research endeavors to bolster the model’s robustness and generalizability. Conclusion To encapsulate, we have developed a predictive model based on seven readily available preoperative variables that effectively stratifies the risk of postoperative Human Serum Albumin (HSA) infusion following THA in patients with femoral head necrosis. The nomogram, developed using the R computational environment, demonstrates commendable differentiation and accuracy. For clinical implementation, this model could be integrated into electronic health record systems as a preoperative risk calculator, providing real-time risk assessment during preoperative evaluation. This clinical tool enables preoperative identification of high-risk candidates, facilitating proactive interventions such as nutritional optimization and individualized surgical planning. Implementation of this model in clinical practice holds promise for reducing the incidence of postoperative hypoalbuminemia and corresponding HSA administration requirements. Supplementary Information Supplementary material 1 Supplementary material 1