Predictors Of Blood Transfusion Research Articles

Development of machine learning models to predict perioperative blood transfusion in hip surgery

BackgroundAllogeneic Blood transfusion is common in hip surgery but is associated with increased morbidity. Accurate prediction of transfusion risk is necessary for minimizing blood product waste and preoperative decision-making. The study aimed to develop machine learning models for predicting perioperative blood transfusion in hip surgery and identify significant risk factors.MethodsData of patients undergoing hip surgery between January 2013 and October 2021 in the Peking Union Medical College Hospital were collected to train and test predictive models. The primary outcome was perioperative red blood cell (RBC) transfusion within 72 h of surgery. Fourteen machine learning algorithms were established to predict blood transfusion risk incorporating patient demographic characteristics, preoperative laboratory tests, and surgical information. Discrimination, calibration, and decision curve analysis were used to evaluate machine learning models. SHapley Additive exPlanations (SHAP) was performed to interpret models.ResultsIn this study, 2431 hip surgeries were included. The Ridge Classifier performed the best with an AUC = 0.85 (95% CI, 0.81 to 0.88) and a Brier score = 0.21. Patient-related risk factors included lower preoperative hemoglobin, American Society of Anesthesiologists (ASA) Physical Status > 2, anemia, lower preoperative fibrinogen, and lower preoperative albumin. Surgery-related risk factors included longer operation time, total hip arthroplasty, and autotransfusion.ConclusionsThe machine learning model developed in this study achieved high predictive performance using available variables for perioperative blood transfusion in hip surgery. The predictors identified could be helpful for risk stratification, preoperative optimization, and outcomes improvement.

Predicting blood transfusion following traumatic injury using machine learning models: A systematic review and narrative synthesis.

Haemorrhage is a leading cause of preventable death in trauma. Accurately predicting a patient's blood transfusion requirement is essential but can be difficult. Machine learning (ML) is a field of artificial intelligence that is emerging within medicine for accurate prediction modelling. This systematic review aimed to identify and evaluate all ML models that predict blood transfusion in trauma. This systematic review was registered on The International Prospective register of Systematic Reviews (CRD4202237110). MEDLINE, Embase and the Cochrane Central Register of Controlled Trials were systematically searched. Publications reporting a ML model that predicted blood transfusion in injured adult patients were included. Data extraction and risk of bias assessment was performed using validated frameworks. Data was synthesised narratively due to significant heterogeneity. Twenty-five ML models for blood transfusion prediction in trauma were identified. Models incorporated diverse predictors and varied ML methodologies. Predictive performance was variable but eight models achieved excellent discrimination (AUROC >0.9) and nine models achieved good discrimination (AUROC >0.8) in internal validation. Only two models reported measures of calibration. Four models have been externally validated in prospective cohorts: the Bleeding Risk Index, Compensatory Reserve Index, the Marsden model and the Mina model. All studies were considered at high risk of bias often due to retrospective datasets, small sample size and lack of external validation. This review identified twenty-five ML models developed to predict blood transfusion requirement after injury. Seventeen ML models demonstrated good to excellent performance in-silico but only four models were externally validated. To date ML models demonstrate the potential for early and individualised blood transfusion prediction but further research is critically required to narrow the gap between ML model development and clinical application. Systematic Review Without Meta-Analysis, Level IV.

Shots fired: evaluation of vascular injury, compartment syndrome, and transfusion rates among civilian ballistic orthopaedic fracture patients presenting to two Level I trauma centres.

This study investigates baseline patient demographics and predictors of vascular injury, blood transfusion, and compartment syndrome in patients with orthopaedic fractures secondary to GSWs at two high-volume Level I trauma centres. A retrospective chart review of all GSW-related trauma patients at two Level I trauma centres between July 2019 and September 2021 was conducted. Chi-squared and two-tailed independent t tests were used for data analysis, and logistic regression with odds ratios (OR) determined predictors of primary outcomes. Among 478 GSW patients, 94 (19.7%) sustained 130 orthopaedic fractures, most commonly at the lower extremity (77.7%). Orthopaedic fracture patients showed significantly higher rates of vascular injury (29.8 vs. 4.7%, p < 0.001), transfusion (27.7 vs. 12.8%, p = 0.006), and compartment syndrome (3.2 vs. 0.3%, p = 0.011) compared to non-orthopaedic injury patients. Univariable analysis identified ankle (OR = 47.50, p < 0.001) and hip/femur fractures (OR = 5.31, p < 0.001) as predictors of vascular injury. Multivariable logistic regression revealed lower extremity vascular injury (OR = 54.69, p = 0.006) and anatomic fracture sites of the humerus (OR = 15.17, p = 0.008), clavicle/scapula (OR = 11.30, p = 0.009), and acetabulum/pelvis (OR = 7.17, p = 0.025) as predictors of blood transfusion. Univariable analysis showed lower extremity vascular injury (OR = 30.14, p = 0.007) as a predictor of compartment syndrome. These findings underscore the importance of diagnosing and managing vascular injuries and compartment syndrome in GSW-related orthopaedic fractures, emphasizing the necessity for targeted transfusion strategies in such cases.

Predictors of Blood Transfusion in Patients Undergoing Cytoreductive Surgeries for Ovarian Malignancy.

The aims of this study were to describe the baseline estimated blood loss (EBL) in surgery and transfusion rate in patients undergoing cytoreductive surgeries for ovarian malignancy, and identify perioperative variables associated with blood loss and transfusion. A retrospective cohort study at a single institution was performed that included patients with known or suspected ovarian malignancy undergoing cytoreductive surgery between 2016 and 2021. t tests, χ2 tests, and multiple logistic regression analyses were used. Among 44 patients meeting inclusion criteria, 61% received perioperative blood transfusion. There were significant differences in EBL and preoperative hemoglobin levels between patients who did and did not receive transfusion (EBL 442.6 vs 236.8 mL, P = 0.0008; preoperative hemoglobin 10.2 vs 11.2 g/dL, P = 0.049). After adjusting for preoperative hemoglobin, the risk of transfusion increased for each additional 200 mL of EBL (odds ratio [OR] 3.8, 95% confidence interval [CI] 1.5-9.5). Stratified by race, the association between EBL and transfusion risk remained statistically significant only for non-Latinx White patients (OR 6.1, 95% CI 1.7-21.9), who made up 77% of the study population, but not for patients of other races and ethnicities (OR 1.0, 95% CI 0.16-6.42). Perioperative blood transfusion is common in patients undergoing cytoreductive surgery. In this study, EBL and preoperative hemoglobin levels were significantly associated with transfusion receipt. Clinicians should optimize hemoglobin levels and intraoperative blood conservation strategies to reduce the need for transfusion. The results also highlight the importance of considering racial and ethnic differences when developing strategies to reduce transfusion risk.

STONE Score: A Predictor for Need of Blood Transfusion in Percutaneous Nephrolithotomy.

To determine if the STONE score is a predictor of blood transfusion and if patient-related factors, i.e., the presence of comorbidities such as urinary tract infection and obesity, can predict blood transfusion post-percutaneous nephrolithotomy. A cross-sectional descriptive study. Place and Duration of the Study: Department of Urology, The Aga Khan University Hospital, Karachi, Pakistan, between March 2022 and 2023. All patients admitted for percutaneous nephrolithotomy (PNCL) were included in the study. STONE score and patient related factors were assessed. Chi-square or Fisher's exact test was applied to check the association between the dependent variables (blood transfusion) and the independent variables. Logistic regression analysis was applied to compare the variables responsible for the outcome. Results: During the study period, 150 patients underwent PCNL. After exclusion, 89 patients were included in the study. The mean STONE score was 7.87, and the total number of transfusions was 8 (8.9%). BMI (body mass index) >25kg/m2 and STONE score were found to be significant factors predicting the need for transfusion with p-values of 0.02 and 0.03, respectively. On multivariate analysis, only BMI was found to be a significant contributing factor for blood transfusion. High BMI and STONE score are significant predictive factors for blood transfusion post-PCNL. Blood product arrangements should be restricted to obese patients. STONE score, Body mass index, Blood transfusion.

Comparison of Risk Scores for Predicting Adverse Outcomes in Acute Lower Gastrointestinal Bleeding.

Acute lower gastrointestinal bleeding (ALGIB) is a common emergency in gastroenterology. Currently, there is insufficient information to predict adverse outcomes in patients with acute lower gastrointestinal bleeding. Our study is aimed at comparing the effectiveness of the clinical risk scores currently utilized and their ability to predict significant outcomes in lower gastrointestinal bleeding. We conducted a retrospective observational study of patients who were admitted to ALGIB and underwent colonoscopy or angiography at a single center between January 2018 and December 2022. Adverse outcomes associated with ALGIB included rebleeding, blood transfusion, hemostatic interventions, and in-hospital death. We calculated six risk scores at admission (Oakland, Birmingham, SHA2PE, Ramaekers, SALGIB, and CNUH-5). We measured the accuracy of these scores using the area under the receiver operating characteristic curve (AUC) and compared them with DeLong's test. 123 patients with confirmed LGIB (aged 65 years, 55-75) were finally included. The most common diagnoses were colorectal cancer (25%) and hemorrhoids (14%). All scores demonstrated sufficient and comparable effectiveness for hemostatic intervention but no discrimination for rebleeding. The Oakland and SALGIB scores were superior to the other scores in predicting blood transfusion (AUC: 0.97 and 0.95, respectively; p = 0.208) and any adverse outcomes (AUC: 0.78 and 0.78, respectively; p = 0.854). The Oakland and SALGIB scores outperform the other scores in predicting the requirement for blood transfusion in ALGIB patients, but no single prediction tool had the best ability across all outcomes. Novel risk stratification scores with higher performance are needed for better risk stratification in ALGIB.

Predicting the Need for Blood Transfusions in Cardiac Surgery: A Comparison between Machine Learning Algorithms and Established Risk Scores in the Brazilian Population.

Blood transfusion is a common practice in cardiac surgery, despite its well-known negative effects. To mitigate blood transfusion-associated risks, identifying patients who are at higher risk of needing this procedure is crucial. Widely used risk scores to predict the need for blood transfusions have yielded unsatisfactory results when validated for the Brazilian population. In this retrospective study, machine learning (ML) algorithms were compared to predict the need for blood transfusions in a cohort of 495 cardiac surgery patients treated at a Brazilian reference service between 2019 and 2021. The performance of the models was evaluated using various metrics, including the area under the curve (AUC), and compared to the commonly used Transfusion Risk and Clinical Knowledge (TRACK) and Transfusion Risk Understanding Scoring Tool (TRUST) scoring systems. The study found that the model had the highest performance, achieving an AUC of 0.7350 (confidence interval [CI]: 0.7203 to 0.7497). Importantly, all ML algorithms performed significantly better than the commonly used TRACK and TRUST scoring systems. TRACK had an AUC of 0.6757 (CI: 0.6609 to 0.6906), while TRUST had an AUC of 0.6622 (CI: 0.6473 to 0.6906). The findings of this study suggest that ML algorithms may offer a more accurate prediction of the need for blood transfusions than the traditional scoring systems and could enhance the accuracy of predicting blood transfusion requirements in cardiac surgery patients. Further research could focus on optimizing and refining ML algorithms to improve their accuracy and make them more suitable for clinical use.

Validation of Machine Learning Model Performance in Predicting Blood Transfusion After Primary and Revision Total Hip Arthroplasty

BackgroundThe rates of blood transfusion following primary and revision total hip arthroplasty (THA) remain as high as 9% and 18%, respectively, contributing to patient morbidity and healthcare costs. Existing predictive tools are limited to specific populations, thereby diminishing their clinical applicability. This study aimed to externally validate our previous institutionally developed machine learning (ML) algorithms to predict the risk of postoperative blood transfusion following primary and revision THA using national inpatient data. MethodsFive ML algorithms were trained and validated using data from 101,266 primary THA and 8,594 revision THA patients from a large national database to predict postoperative transfusion risk after primary and revision THA. Models were assessed and compared based on discrimination, calibration, and decision curve analysis. ResultsThe most important predictors of transfusion following primary and revision THA were preoperative hematocrit (<39.4%) and operation time (>157 minutes), respectively. All ML models demonstrated excellent discrimination (area under the curve (AUC) >0.8) in primary and revision THA patients, with artificial neural network (AUC = 0.84, slope = 1.11, intercept = −0.04, Brier score = 0.04), and elastic-net-penalized logistic regression (AUC = 0.85, slope = 1.08, intercept = −0.01, and Brier score = 0.12) performing best, respectively. On decision curve analysis, all 5 models demonstrated a higher net benefit than the conventional strategy of intervening for all or no patients in both patient cohorts. ConclusionsThis study successfully validated our previous institutionally developed ML algorithms for the prediction of blood transfusion following primary and revision THA. Our findings highlight the potential generalizability of predictive ML tools developed using nationally representative data in THA patients.

An update on red blood cell transfusion in non-cardiac thoracic surgery.

The aim of this study was to evaluate risk factors for red blood cell (RBC) transfusion in non-cardiac thoracic surgery. All patients undergoing non-cardiac thoracic surgery in a single tertiary referral center between January and December 2021 were eligible for this study. Data on blood requests and perioperative RBC transfusion were retrospectively analyzed. A total of 379 patients were included, of whom 275 (72.6%) underwent elective surgery. The overall RBC transfusion rate was 7.4% (elective cases: 2.5%, non-elective cases: 20.2%). Patients with lung resections required transfusion in 2.4% of the cases versus 44.7% in patients undergoing surgery for empyema. In multivariable analysis, empyema (P=0.001), open surgery (P<0.001), low preoperative hemoglobin (P=0.001), and old age (P=0.013) were independent risk factors for RBC transfusion. The best predictor of blood transfusion was preoperative hemoglobin with a cut-off value <10.4 g/dL (sensitivity 82.1%, specificity 86.3%, area under the curve 0.882). The rate of RBC transfusion in current non-cardiac thoracic surgery is low, especially in elective lung resections. In urgent cases and open surgery, transfusion rates remain high, particularly in empyema cases. Preoperative requesting of RBC units should be tailored to patient-specific risk factors.

300 Intraoperative Thresholds for Transfusion During Burn Surgery

Abstract Introduction Burn injury is associated with significant morbidity. Deep partial thickness or full thickness burns require surgical excision and skin grafting. This process results in significant blood loss and frequently requires blood transfusions. There is scarcity in published literature addressing intraoperative blood transfusion requirements. The relationship between blood product volume and patient characteristics has not been clearly defined. The enhanced ability to predict blood transfusion requirements would allow for improved allocation of blood products and safer overall operations. Methods This is a retrospective review conducted at a large academic burn center. Inclusion criteria are adult patients admitted between December 2019 and August 2020 with burns requiring surgical excision. Prisoners, children, and non-operative burns were excluded. Operative cases requiring blood product administration were compared to cases that did not require transfusion. Two-tailed student’s t-tests were conducted with p &amp;lt; 0.05 considered significant. Results There were 103 total patients who underwent a total of 217 operations. Forty-eight percent (n=104) of cases required blood transfusion. As detailed in the table, the transfusion group had significantly increased percent total body surface area (TBSA) burn, along with longer hospital and ICU lengths of stay (LOS). Operatively, this group underwent more tissue excision (approximately 15% TBSA compared to 5%) with longer case lengths, coupled with higher estimated blood loss and a lower preoperative hemoglobin. Interestingly, the median time for the initiation of packed red blood cell transfusion was six minutes after incision. Conclusions The average estimated blood loss in cases resulting in transfusion was 907±462mL (with a median of 800mL) compared to 467±398mL estimated blood loss in the non-transfused group. Expected blood loss along with parameters such as planned area of excision (which was approximately three times greater in transfused patients) and severity of burn injury can guide the surgeon in anticipating perioperative administration of blood products. Future directions include stratification of blood volume administration by %TBSA, operative time, LOS, and percent excision. Applicability of Research to Practice The ability to predict blood product requirements preoperatively coupled with the incorporation of perioperative parameters can assist in the judicious and appropriate use of blood products during burn operations.

Development of machine learning and multivariable models for predicting blood transfusion in head and neck microvascular reconstruction for risk‐stratified patient blood management

Although blood transfusions have adverse consequences for microvascular head and neck reconstruction, they are frequently administered. Pre-identifying patients would allow risk-stratified patient blood management. Development of machine learning (ML) and logistic regression (LR) models based on retrospective inclusion of 657 patients from 2011 to 2021. Internal validation and comparison with models from the literature by external validation. Development of a web application and a score chart. Our models achieved an area under the receiver operating characteristic curve (ROC-AUC) of up to 0.825, significantly outperforming LR models from the literature. Preoperative hemoglobin, blood volume, duration of surgery and flap type/size were strong predictors. The use of additional variables improves the prediction for blood transfusion, while models seems to have good generalizability due to surgical standardization and underlying physiological mechanism. The ML models developed showed comparable predictive performance to an LR model. However, ML models face legal hurdles, whereas score charts based on LR could be used after further validation.

Plasma Fibrinogen as a Predictor of Perioperative-Blood-Component Transfusion in Major-Nontraumatic-Orthopedic-Surgery Patients: A Cohort Study.

There is a trend towards increased perioperative bleeding in patients with plasma fibrinogen levels < 200 mg/dL-1. This study aimed to assess whether there is an association between preoperative fibrinogen levels and perioperative blood-product transfusion up to 48 h after major orthopedic surgery. This cohort study included 195 patients who underwent primary or revision hip arthroplasty for nontraumatic etiologies. Plasma fibrinogen, blood count, coagulation tests, and platelet count were measured preoperatively. A plasma fibrinogen level of 200 mg/dL-1 was the cutoff value used to predict blood transfusion. The mean (SD) plasma fibrinogen level was 325 (83) mg/dL-1. Only thirteen patients had levels < 200 mg/dL-1, and only one of them received a blood transfusion, with an absolute risk of 7.69% (1/13; 95%CI: 1.37-33.31%). Preoperative plasma fibrinogen levels were not associated with the need for blood transfusion (p = 0.745). The sensitivity and the positive predictive value of plasma fibrinogen < 200 mg/dL-1 as a predictor of blood transfusion were 4.17% (95%CI: 0.11-21.12%) and 7.69% (95%CI: 1.12-37.99%), respectively. Test accuracy was 82.05% (95%CI: 75.93-87.17%), but positive and negative likelihood ratios were poor. Therefore, preoperative plasma fibrinogen level in hip-arthroplasty patients was not associated with the need for blood-product transfusion.

Validation of Blood Transfusion Risk Scores (TRACK and TRUST) in a Cardiac Surgery Service in Brazil.

Transfusion of red blood cells is recurrent in cardiac surgery despite the well-established deleterious effects. Identifying patients with higher chances of requiring blood transfusion is essential to apply strategic preventive measures to reduce such chances, considering the restricted availability of this product. The most used risk scores to predict blood transfusion are the Transfusion Risk and Clinical Knowledge (TRACK) and Transfusion Risk Understanding Scoring Tool (TRUST). However, these scores were not validated for the Brazilian population. The objective of this study was to assess the accuracy of TRACK and TRUST scores in estimating the need for postoperative transfusion of red blood cell concentrates (TRBCC) after cardiac surgery. A clinical retrospective study was conducted using the database of a Brazilian reference service composed of patients operated between November 2019 and September 2021. Scores were compared using Mann-Whitney U test. Hosmer-Lemeshow goodness of fit test assessed calibration of the scores. Accuracy was assessed using the area under the receiver operating characteristic curve (AUC). All analyses considered a level of significance of 5%. The study was approved by the research ethics committee (CAAE 55577421.4.0000.5201). This study assessed 498 patients. Only the TRACK score presented good calibration (P=0.238; TRUST P=0.034). AUC of TRACK was 0.678 (95% confidence interval 0.63 to 0.73; P<0.001), showing a significant accuracy. Between the scores analyzed, only the TRACK score showed a good calibration, but low accuracy, to predict postoperative TRBCC after cardiac surgery.

Validation of Oakland score for predicting adverse outcomes of acute lower gastrointestinal bleeding

Background: Gastrointestinal bleeding is a common emergency in medical facilities. In clinical practice, there are many scales used to evaluate the risk of poor prognosis (Rockall, Glasgow, Blatch Ford) and mainly applied in upper gastrointestinal tract. However, the prognosis of lower gastrointestinal tract diseases remains limited. The appropriate selection and the application of scales for assessing the risks of poor prognosis in acute lower gastrointestinal bleeding (ALGIB) cases are essential, assist the physician to direct the intensive management attitude as well as the waste of human resources and health economy. Oakland score has been considered as a simple, convenient score to assist clinicians quickly, accurately stratify the risk of ALGIB patients. This study aimed to validate the Oakland score for predicting adverse outcomes of ALGIB patients. Methods: A cross-sectional study was conducted at Cho Ray Hospital from Jan 2021 to May 2021, including patients aged ≥ 16 years who had indication of admission due to symptoms suggesting ALGIB and underwent lower gastrointestinal endoscopy. These patients were then followed-up to observe the adverse outcomes, including blood transfusion, endoscopic intervention, surgery and death. Area under the receiver operating characteristic curve (AUC) and the best cut-off value of Oakland score for predicting adverse outcomes of ALGIB patients were analyzed using SPSS software. Results: A total of 70 patients with mean age 59.5 ± 16.4, male/female ratio (1.5 : 1) was recruited in this study. 27/70 (38.6%) stopped bleeding spontaneously without any interventions. The rate of clinical outcomes was blood transfusion (33/43, 76.7%), endoscopic intervention (5/43, 11.6%), blood transfusion plus endoscopic intervention (2/43, 4.7%), blood transfusion plus surgery (2/43, 4.7%), blood transfusion plus DSA plus surgery (1/43, 2.3%), respectively. AUC for predicting blood transfusion and adverse outcomes of ALGIB patients were 0.95 (0.91 - 0.99) and 0.91 (0.84 - 0.98), respectively. Oaklad score threshold of 21 showed that sensitivity and specificity for predicting blood transfusion were 94.7% and 78.1%, sensitivity and specificity for predicting adverse outcomes of ALGIB patients were 81.5% and 90.7% respectively. Conclusions: Oakland is an excellent score in order to predict clinical outcomes on ALGIB patients. Key words: Oakland; acute lower gastrointestinal bleeding; colonoscopy

The Emergency Surgery Score as a Predictor for Blood Transfusion: NSQIP 2018 Data

The Emergency Surgery Score as a Predictor for Blood Transfusion: NSQIP 2018 Data

Does the transfusion rate following reverse total shoulder arthroplasty warrant aggressive perioperative blood management? An indications-based analysis

HypothesisWe hypothesized that certain operative conditions are more likely to require transfusion after reverse total shoulder arthroplasty (RTSA), but found no data comparing transfusion rates for traumatic, post-traumatic, and non-traumatic indications. MethodsRetrospective review identified 1083 patients with RTSA. After exclusion of revision and tumor cases, 783 patients were divided into 3 groups as follows: traumatic (acute fractures, fracture-dislocations), post-traumatic (nonunions, malunions, chronic dislocations), or non-traumatic (chronic rotator cuff insufficiency, glenohumeral arthritis with advanced glenoid wear). Demographics, operative indications, preoperative hematocrit, postoperative hematocrit, change in hematocrit, estimated blood loss, operative time, and blood transfusion rates were compared. We constructed a logistic regression model to determine the risk factors predicting blood transfusion. ResultsIndications were acute trauma in 83 patients, post-traumatic sequelae in 65, and non-traumatic conditions in 635. There were no statistically significant differences regarding body mass index (P = .5) or American Society of Anesthesiologists score (P = .11). There were more females in the traumatic and post-traumatic cohorts than in the non-traumatic cohort (P < .001). Surgical age was younger in the post-traumatic cases (P = .002). The overall frequency of blood transfusion was 4.6% (36/783): traumatic, 19% (16/83); post-traumatic, 17% (11/65); and non-traumatic, 1.4% (9/635). Both the traumatic and post-traumatic groups had significantly higher rates of blood transfusion than non-traumatic cases (P < .001). Preoperative hematocrit was lower in the traumatic group and estimated blood loss was higher in the traumatic and post-traumatic groups (P = .001). Post-traumatic patients experienced the greatest drop in postoperative hematocrit from preoperative levels. The multivariate logistic regression model showed that traumatic and post-traumatic indications for surgery were each an independent risk factor for blood transfusion, as was change in hematocrit, body mass index, surgical age, estimated blood loss, and preoperative hematocrit. ConclusionsOur results suggest that acute trauma as the indication for RTSA is the strongest independent risk factor for blood transfusion followed by post-traumatic sequelae. Surgeons should be aware of the markedly increased risk of transfusion based purely on these 2 surgical indications especially when in combination with other predictors of transfusion.

Validation of 5 models predicting transfusion, bleeding, and mortality in liver transplantation: an observational cohort study.

Historically, orthotopic liver transplantation (OLT) has been associated with massive blood loss, blood transfusion and morbidity. In order to predict such outcomes five nomograms have been published relating to transfusions and morbidity associated with OLTs. These nomograms, developed on the basis of three cohorts of patients consisting of 406, 750, and 800 having undergone OLTs, aimed to predict a transfusion of ≥1 red blood cell unit (RBC), a transfusion of >2 RBC units, a blood loss of >900ml, as well as one-month and one-year survival rates. The aim of this study was to validate these five nomograms in a contemporary, independent cohort of patients. Five nomograms were previously developed based on 406, 750, and 800 OLTs. In this study we performed a temporal validation of these nomograms on contemporary patients that consisted of three cohorts of 800, 250, and 200 OLTs. Logistic regression coefficients from the historic development cohorts were applied to the three contemporary temporal validation cohorts. The most accurate nomogram was able to predict transfusion of ≥1 RBC units with an area under the curve (AUC) was 0.91. The second-best nomogram was able to predict bleeding of >900ml with an AUC of 0.70. T he AUC of the third nomogram (transfusion of >2 RBC units) was 0.70. However, is temporal validation was suboptimal, due to a low prevalence of OLTs transfused with >2 RBC units. The last 2 nomograms exhibited clearly suboptimal AUC values of 0.54 and 0.61. Two of the five nomograms predict blood transfusion and blood loss with excellent accuracy. Transfusion of ≥1 RBC unit and blood loss of >900ml can be predicted on the basis of these nomograms. However, these nomograms are not accurate to predict one-month and one-year survival rates. These results should be further cross-validated, ideally prospectively, in additional external independent cohorts.

Temporal trends, predictors of blood transfusion and in-hospital outcomes among patients with severe burn injury in the United States-A national database-based analysis.

Severe burn can be accompanied by life-threatening bleeding on some occasions, thus, blood transfusion is often required in these patients during their hospitalization. Therefore, we aimed to examine temporal trends, predictors, and in-hospital outcomes of blood transfusion in these patients in the United States. The National Inpatient Sample wasused to identify severe burn patients between January 2010 and September 2017 in the United States. Trends in the utilization of blood transfusion were analyzed using the Cochran-Armitage trend test. Moreover, propensity score matching (PSM) was employed, and then in-hospital outcomes were compared between these two groups in the matched cohort. Multivariable logistic regressions were further used to validate the results of PSM. Among 27,260 severe burn patients identified during the study period, 2120 patients (7.18%) received blood transfusion. Blood transfusion rates decreased significantly from 9.52% in 2010 to 5.02% in 2017 (p for trend <.001). In the propensity-matched cohort (2120 pairs with and without transfusion), patients transfused were at increased risk of in-hospital mortality (13.3% vs 8.77%, p < .001), overall postoperative complications (88.3% vs 72.59%, p < .001), longer hospital stays (defined as > median hospital stays=5d) (73.8% vs 50.6%, p< .001) and increased overall cost (defined as > median overall costs=30,746) (81.6% vs 57.3%, p< .001). This was also the case for the multivariable analysis. Blood transfusion following severe burn injury may be associated with worse clinical outcomes. The utility for blood transfusion in burn patients warrants further prospective exploration.

Development and validation of a Bayesian belief network predicting the probability of blood transfusion after pediatric injury.

Early recognition and intervention of hemorrhage are associated with decreased morbidity in children. Triage models have been developed to aid in the recognition of hemorrhagic shock after injury but require complete data and have limited accuracy. To address these limitations, we developed a Bayesian belief network, a machine learning model that represents the joint probability distribution for a set of observed or unobserved independent variables, to predict blood transfusion after injury in children and adolescents. We abstracted patient, injury, and resuscitation characteristics of injured children and adolescents (age 1 to 18 years) from the 2017 to 2019 Trauma Quality Improvement Project database. We trained a Bayesian belief network to predict blood transfusion within 4 hours after arrival to the hospital following injury using data from 2017 and recalibrated the model using data from 2018. We validated our model on a subset of patients from the 2019 Trauma Quality Improvement Project. We evaluated model performance using the area under the receiver operating characteristic curve and calibration curves and compared performance with pediatric age-adjusted shock index (SIPA) and reverse shock index with Glasgow Coma Scale (rSIG) using sensitivity, specificity, accuracy, and Matthew's correlation coefficient (MCC). The final model included 14 predictor variables and had excellent discrimination and calibration. The model achieved an area under the receiver operating characteristic curve of 0.92 using emergency department data. When used as a binary predictor at an optimal threshold probability, the model had similar sensitivity, specificity, accuracy, and MCC compared with SIPA when only age, systolic blood pressure, and heart rate were observed. With the addition of the Glasgow Coma Scale score, the model has a higher accuracy and MCC than SIPA and rSIG. A Bayesian belief network predicted blood transfusion after injury in children and adolescents better than SIPA and rSIG. This probabilistic model may allow clinicians to stratify hemorrhagic control interventions based upon risk. Prognostic and Epidemiologic; Level III.

Selective type & screen for elective colectomy based on a transfusion risk score may generate substantial cost savings.

Preoperative type and screen are currently recommended for all patients undergoing colectomy. We aimed to identify risk factors for transfusion and define a low-risk cohort of patients undergoing colectomy in whom type and screen may be safely avoided. We identified all patients undergoing elective colectomy in the National Surgical Quality Improvement Project-Targeted Colectomy files from 2012 to 2016. Patients transfused preoperatively and those undergoing other concurrent major abdominal procedures were excluded. We compared patients who received blood transfusion on the day of surgery to those who did not. Half of the cohort was randomly selected for development of a points-based model predicting blood transfusion on the day of surgery. This model was then validated using the remaining patients. Of 61,964 patients undergoing colectomy, 3128 (5%) patients were transfused with 1290 (2.1%) occurring on the day of surgery. Preoperative anemia was the strongest predictor of blood transfusion on the day of surgery. Among patients with hematocrit > 35%, day of surgery transfusion risk was 0.8%; 99% of patients with hematocrit > 35% had a score 20 or less. Selective type and screen for patients with score ≤ 20 or hematocrit > 35% would avoid type and screen in 91% and 81% of patients, respectively. Transfusion following elective colectomy is rare and can be accurately predicted by preoperative patient characteristics. Selective type and screen based on these parameters have the potential to prevent operative delays and lower cost.