Development and Validation of a Machine Learning Algorithm to Predict the Risk of Blood Transfusion after Total Hip Replacement in Patients with Femoral Neck Fractures: A Multicenter Retrospective Cohort Study

Abstract

Objective: Total hip arthroplasty (THA) remains the primary treatment option for femoral neck fractures in elderly patients. This study aims to explore the risk factors associated with allogeneic blood transfusion after surgery and to develop a dynamic prediction model to predict post-operative blood transfusion requirements. This will provide more accurate guidance for perioperative humoral management and rational allocation of medical resources.

Methods: We retrospectively analyzed data from 829 patients who underwent total hip arthroplasty for femoral neck fractures at three third-class hospitals between January 2017 and August 2023. Patient data from one hospital were used for model development, whereas data from the other two hospitals were used for external validation. Logistic regression analysis was used to screen the characteristic subsets related to blood transfusion. Various machine learning algorithms, including logistic regression, SVA (support vector machine), K-NN (k-nearest neighbors), MLP (multilayer perceptron), naive Bayes, decision tree, random forest, and gradient boosting, were used to process the data and construct prediction models. A 10-fold cross-validation algorithm facilitated the comparison of the predictive performance of the models, resulting in the selection of the best-performing model for the development of an open-source computing program.

Results: BMI (body mass index), surgical duration, IBL (intraoperative blood loss), anticoagulant history, utilization rate of tranexamic acid, Pre-Hb, and Pre-ALB were included in the model as well as independent risk factors. The average area under curve (AUC) values for each model were as follows: logistic regression (0.98); SVA (0.91); k-NN (0.87) MLP, (0.96); naive Bayes (0.97); decision tree (0.87); random forest (0.96); and gradient boosting (0.97). A web calculator based on the best model is available at: (https://nomo99.shinyapps.io/dynnomapp/).

Conclusion: Utilizing a computer algorithm, a prediction model with a high discrimination accuracy (AUC > 0.5) was developed. The logistic regression model demonstrated superior differentiation and reliability, thereby successfully passing external validation. The model’s strong generalizability and applicability have significant implications for clinicians, aiding in the identification of patients at high risk for postoperative blood transfusion.

Keywords

Allogeneic transfusion / Femoral neck fracture / Machine learning / Prediction model / Total hip arthroplasty

References

[1] Levi N. Early mortality after cervical hip fractures. Injury. 1996; 27(8): 565–567.

[2] DeFrancesco CJ. CORR insights^®: how often do complications and mortality occur after operatively treated periprosthetic proximal and distal femoral fractures? A register-based study. Clin Orthop Relat Res. 2023; 481(10): 1950–1953.

[3] Li SG, Sun TS, Liu Z, Ren JX, Liu B, Gao Y. Factors influencing postoperative mortality one year after surgery for hip fracture in Chinese elderly population. Chin Med J. 2013; 126(14): 2715–2719.

[4] Hochreiter J, Hejkrlik W, Emmanuel K, Hitzl W, Ortmaier R. Blood loss and transfusion rate in short stem hip arthroplasty. A Comparative Study. Int Orthop. 2017; 41(7): 1347–1353.

[5] Menendez ME, Lu N, Huybrechts KF, Ring D, Barnes CL, Ladha K, et al. Variation in use of blood transfusion in primary Total hip and knee arthroplasties. J Arthroplasty. 2016; 31(12): 2757–2763.e2.

[6] Liu KC, Piple AS, Richardson MK, Mayer LW, Mayfield CK, Christ AB, et al. Increased risk of venous thromboembolism in patients with postoperative anemia after Total joint arthroplasty: are transfusions to blame? J Bone Jt Surg Am. 2023; 105(17): 1354–1361.

[7] Frisch NB, Wessell NM, Charters MA, Yu S, Jeffries JJ, Silverton CD. Predictors and complications of blood transfusion in total hip and knee arthroplasty. J Arthroplasty. 2014; 29(9 Suppl): 189–192.

[8] Jiang Q, Wang Y, Xie D, Wei J, Li X, Zeng C, et al. Trends, complications, and readmission of allogeneic red blood cell transfusion in primary total hip arthroplasty in China: a national retrospective cohort study. Arch Orthop Trauma Surg. 2024; 144(1): 483–491.

[9] Guerin S, Collins C, Kapoor H, McClean I, Collins D. Blood transfusion requirement prediction in patients undergoing primary total hip and knee arthroplasty. Transfus Med. 2007; 17(1): 37–43.

[10] Pagnussatt Neto E, Lopes da Costa PD, Gurgel SJT, Schmidt Azevedo P, Modolo NSP, do Nascimento Junior P. Plasma fibrinogen as a predictor of perioperative-blood-component transfusion in major-nontraumatic-orthopedic-surgery patients: a cohort study. Diagnostics. 2023; 13(5): 976.

[11] Pempe C, Werdehausen R, Pieroh P, Federbusch M, Petros S, Henschler R, et al. Predictors for blood loss and transfusion frequency to guide blood saving programs in primary knee-and hip-arthroplasty. Sci Rep. 2021; 11(1): 4386.

[12] Bian FC, Cheng XK, An YS. Preoperative risk factors for postoperative blood transfusion after hip fracture surgery: establishment of a nomogram. J Orthop Surg Res. 2021; 16(1): 406.

[13] Bao N, Zhou L, Cong Y, Guo T, Fan W, Chang Z, et al. Free fatty acids are responsible for the hidden blood loss in total hip and knee arthroplasty. Med Hypotheses. 2013; 81(1): 104–107.

[14] Yoshihara H, Yoneoka D. Predictors of allogeneic blood transfusion in total hip and knee arthroplasty in the United States, 2000–2009. J Arthroplasty. 2014; 29(9): 1736–1740.

[15] Hamet P, Tremblay J. Artificial intelligence in medicine. Metabolism. 2017; 69S: S36–S40.

[16] Grendas LN, Chiapella L, Rodante DE, Daray FM. Comparison of traditional model-based statistical methods with machine learning for the prediction of suicide behaviour. J Psychiatr Res. 2021; 145: 85–91.

[17] Sun H, Wu S, Li S, Jiang X. Which model is better in predicting the survival of laryngeal squamous cell carcinoma?: comparison of the random survival forest based on machine learning algorithms to cox regression: analyses based on SEER database. Medicine. 2023; 102(10): e33144.

[18] National Health Commission of the People’s Republic of China. WS/T796-2022 Guidelines for Perioperative Blood Management [S/OL]. Beijing: National Health Commission of the People’s Republic of China; 2022. https://www.nhc.gov.cn/wjw/s9493/202202/5e3bc1a664094da692bcb3e2e85efd34.shtml

[19] Gao FQ, Li ZJ, Zhang K, Sun W, Zhang H. Four methods for calculating blood-loss after Total knee arthroplasty. Chin Med J. 2015; 128(21): 2856–2860.

[20] Nadler SB, Hidalgo JH, Bloch T. Prediction of blood volume in normal human adults. Surgery. 1962; 51(2): 224–232.

[21] Investigators from Western Michigan University Released New Data on Support Vector Machines (Multiresolution Hierarchical Support Vector Machines for the Classification of Large Datasets). Robotics and Machine Learning Daily News; 2022.

[22] Shi X, Cui Y, Wang S, Pan Y, Wang B, Lei M. Development and validation of a web-based artificial intelligence prediction model to assess massive intraoperative blood loss for metastatic spinal disease using machine learning techniques. Spine J. 2024; 24(1): 146–160.

[23] Qin S, Sun S, Wang Y, Li C, Fu L, Wu M, et al. Immune, metabolic landscapes of prognostic signatures for lung adenocarcinoma based on a novel deep learning framework. Sci Rep. 2024; 14(1): 527.

[24] Yu W, Yu F, Li M, Yang F, Wang H, Song H, et al. Quantitative association between lead exposure and amyotrophic lateral sclerosis: a Bayesian network-based predictive study. Environ Health. 2024; 23(1): 2.

[25] Kaur I, Ahmad T. A cluster-based ensemble approach for congenital heart disease prediction. Comput Methods Programs Biomed. 2024; 243: 107922.

[26] Hennebelle A, Ismail L, Materwala H, Al Kaabi J, Ranjan P, Janardhanan R. Secure and privacy-preserving automated machine learning operations into end-to-end integrated IoT-edge-artificial intelligence-blockchain monitoring system for diabetes mellitus prediction. Comput Struct Biotechnol J. 2023; 23: 212–233.

[27] Xu B, Li H, Chen H, Wang W, Jia W, Gong L, et al. Identification and prediction of molecular subtypes of atherosclerosis based on m6A immune cell infiltration. Biochim Biophys Acta, Gen Subj. 2024; 1868(2): 130537.

[28] Nuttall GA, Santrach PJ, Oliver WC Jr, Horlocker TT, Shaughnessy WJ, Cabanela ME, et al. The predictors of red cell transfusions in total hip arthroplasties. Transfusion. 1996; 36(2): 144–149.

[29] Chang SS, Duong DT, Wells N, Cole EE, Smith JA Jr, Cookson MS. Predicting blood loss and transfusion requirements during radical prostatectomy: the significant negative impact of increasing body mass index. J Urol. 2004; 171(5): 1861–1865.

[30] Bashaireh K, Aljararhih O, Alawneh K. Impact of body mass index on hemoglobin level and blood transfusion in total knee arthroplasty: a retrospective case control study. Ann Med Surg. 2020; 28(55): 180–184.

[31] Frisch N, Wessell NM, Charters M, Peterson E, Cann B, Greenstein A, et al. Effect of body mass index on blood transfusion in Total hip and knee arthroplasty. Orthopedics. 2016; 39(5): e844–e849.

[32] Hrnack SA, Skeen N, Xu T, Rosenstein AD. Correlation of body mass index and blood loss during total knee and total hip arthroplasty. Am J Orthop. 2012; 41(10): 467–471.

[33] Walsh M, Preston C, Bong M, Patel V, Di Cesare PE. Relative risk factors for requirement of blood transfusion after total hip arthroplasty. J Arthroplasty. 2007; 22(8): 1162–1167.

[34] Salido JA, Marín LA, Gómez LA, Zorrilla P, Martínez C. Preoperative hemoglobin levels and the need for transfusion after prosthetic hip and knee surgery: analysis of predictive factors. J Bone Jt Surg Am. 2002; 84(2): 216–220.

[35] Chaudhry YP, MacMahon A, Rao SS, Mekkawy KL, Toci GR, Oni JK, et al. Predictors and outcomes of postoperative hemoglobin of <8 g/dL in Total joint arthroplasty. J Bone Joint Surg Am. 2022; 104(2): 166–171.

[36] Loppini M, Cannata R, Pisano A, Morenghi E, Grappiolo G. Incidence and predictors of blood transfusions in one-stage bilateral total hip arthroplasty: a single center prospective cohort study. Arch Orthop Trauma Surg. 2022; 142(11): 3549–3554.

[37] Smith GH, Tsang J, Molyneux SG, White TO. The hidden blood loss after hip fracture. Injury. 2011; 42(2): 133–135.

[38] Vochteloo AJ, Borger van der Burg BL, Mertens B, Niggebrugge AH, de Vries MR, Tuinebreijer WE, et al. Outcome in hip fracture patients related to anemia at admission and allogeneic blood transfusion: an analysis of 1262 surgically treated patients. BMC Musculoskelet Disord. 2011; 12: 262.

[39] Hill SS, Ottaviano KE, Palange DC, Chismark AD, Valerian BT, Canete JJ. Lee EC; NSQIP-IBD collaborative. Impact of preoperative factors in patients with IBD on postoperative length of stay: a National Surgical Quality Improvement Program-Inflammatory Bowel Disease Collaborative Analysis. Dis Colon Rectum. 2024; 67(1): 97–106.

[40] Wiedermann CJ. Controversies surrounding albumin use in sepsis: lessons from cirrhosis. Int J Mol Sci. 2023; 24(24): 17606.

[41] Newman JM, Sodhi N, Khlopas A, Piuzzi NS, Yakubek GA, Sultan AA, et al. Malnutrition increases the 30-day complication and re-operation rates in hip fracture patients treated with total hip arthroplasty. Hip Int. 2020; 30(5): 635–640.

[42] Moloney GB, Boakye LAT, Cluts LM, Palmeri C. Administration of prophylactic enoxaparin on the morning of surgery does not increase risk of blood transfusion or wound drainage following internal fixation of geriatric femur fractures. J Am Acad Orthop Surg. 2023; 31(6): 305–311.

[43] Borgen PO, Dahl OE, Reikeras O. Preoperative versus postoperative initiation of dalteparin thromboprophylaxis in THA. Hip Int. 2010; 20(3): 301–307.

[44] Sizer SC, Cherian JJ, Elmallah RD, Pierce TP, Beaver WB, Mont MA. Predicting blood loss in Total knee and hip arthroplasty. Orthop Clin North Am. 2015; 46(4): 445–459.

[45] Liu Z, Han N, Xu H, Fu Z, Zhang D, Wang T, et al. Incidence of venous thromboembolism and hemorrhage related safety studies of preoperative anticoagulation therapy in hip fracture patients undergoing surgical treatment: a case-control study. BMC Musculoskelet Disord. 2016; 17: 76.

[46] Raska A, Kálmán K, Egri B, Csikós P, Beinrohr L, Szabó L, et al. Synergism of red blood cells and tranexamic acid in the inhibition of fibrinolysis. J Thromb Haemost. 2023; 22(3): 794–804.

[47] Zheng C, Ma J, Xu J, Wu L, Wu Y, Liu Y, et al. The optimal regimen, efficacy and safety of tranexamic acid and aminocaproic acid to reduce bleeding for patients after total hip arthroplasty: a systematic review and Bayesian network meta-analysis. Thromb Res. 2023; 221: 120–129.

[48] Lewis SR, Pritchard MW, Estcourt LJ, Stanworth SJ, Griffin XL. Interventions for reducing red blood cell transfusion in adults undergoing hip fracture surgery: an overview of systematic reviews. Cochrane Database Syst Rev. 2023; 6(6): CD013737.

[49] Ling T, Zhao Z, Xu W, Ge W, Huang L. Effects of tranexamic acid on hemorrhage control and deep venous thrombosis rate after Total knee arthroplasty: a systematic review and network meta-analysis of randomized controlled trials. Front Pharmacol. 2021; 21(12): 639694.

[50] Yang YZ, Cheng QH, Zhang AR, Yang X, Zhang ZZ, Guo HZ. Efficacy and safety of single-and double-dose intravenous tranexamic acid in hip and knee arthroplasty: a systematic review and meta-analysis. J Orthop Surg Res. 2023; 18(1): 593.

[51] Knowlton LM, Arnow K, Trickey AW, Sauaia A, Knudson MM. Does tranexamic acid increase venous thromboembolism risk among trauma patients? A prospective multicenter analysis across 17 level I trauma centers. Injury. 2023; 54(11): 111008.

[52] Huang Z, Huang C, Xie J, Ma J, Cao G, Huang Q, et al. Analysis of a large data set to identify predictors of blood transfusion in primary total hip and knee arthroplasty. Transfusion. 2018; 58(8): 1855–1862.

[53] Surace P, Sultan AA, George J, Samuel LT, Khlopas A, Molloy RM, et al. The association between operative time and short-term complications in Total hip arthroplasty: An analysis of 89, 802 surgeries. J Arthroplasty. 2019; 34(3): 426–432.

[54] Shen HL, Li Z, Feng ML, Cao GL. Analysis on hidden blood loss of total knee arthroplasty in treating knee osteoarthritis. Chin Med J. 2011; 124(11): 1653–1656.

[55] Cai L, Chen L, Zhao C, Wang Q, Kang P. Influencing factors of hidden blood loss after primary total hip arthroplasty through the posterior approach: a retrospective study. BMC Musculoskelet Disord. 2023; 24(1): 582.

[56] Mohammed H, Huang Y, Memtsoudis S, Parks M, Huang Y, Ma Y. Utilization of machine learning methods for predicting surgical outcomes after total knee arthroplasty. PLoS One. 2022; 17(3): e0263897.

[57] Kolin DA, Lyman S, Della Valle AG, Ast MP, Landy DC, Chalmers BP. Predicting postoperative anemia and blood transfusion following Total knee arthroplasty. J Arthroplasty. 2023; 38(7): 1262–1266.e2.

[58] Buddhiraju A, Shimizu MR, Subih MA, Chen TL, Seo HH, Kwon YM. Validation of machine learning model performance in predicting blood transfusion after primary and revision Total hip arthroplasty. J Arthroplasty. 2023; 38(10): 1959–1966.

[59] Cohen-Levy WB, Klemt C, Tirumala V, Burns JC, Barghi A, Habibi Y, et al. Artificial neural networks for the prediction of transfusion rates in primary total hip arthroplasty. Arch Orthop Trauma Surg. 2023; 143(3): 1643–1650.

[60] FIBTEM as a predictor of intra-and postoperative blood loss in revision total hip arthroplasty: A prospective observational study. Medicine. 2018; 97(22): e10929.

[61] Orthopaedic Society of Chinese Medical Association. Guidelines for prevention of venous thromboembolism in major orthopedic surgery in China. Chin J Orthop. 2016; 36(2): 7.

RIGHTS & PERMISSIONS

2024 The Author(s). Orthopaedic Surgery published by Tianjin Hospital and John Wiley & Sons Australia, Ltd.