Nomogram to predict perioperative blood transfusion for hepatopancreaticobiliary and colorectal surgery

Systematic review highlights high risk of bias of clinical prediction models for blood transfusion in patients undergoing elective surgery

BACKGROUND

Blood transfusion can be a lifesaving intervention after perioperative blood loss. Many prediction models have been developed to identify patients most likely to require blood transfusion during elective surgery, but it is unclear whether any are suitable for clinical practice.

STUDY DESIGN AND SETTING

We conducted a systematic review, searching MEDLINE, Embase, PubMed, The Cochrane Library, Transfusion Evidence Library, Scopus, and Web of Science databases for studies reporting the development or validation of a blood transfusion prediction model in elective surgery patients between January 1, 2000 and June 30, 2021. We extracted study characteristics, discrimination performance (c-statistics) of final models, and data, which we used to perform risk of bias assessment using the Prediction model risk of bias assessment tool (PROBAST).

RESULTS

We reviewed 66 studies (72 developed and 48 externally validated models). Pooled c-statistics of externally validated models ranged from 0.67 to 0.78. Most developed and validated models were at high risk of bias due to handling of predictors, validation methods, and too small sample sizes.

CONCLUSION

Most blood transfusion prediction models are at high risk of bias and suffer from poor reporting and methodological quality, which must be addressed before they can be safely used in clinical practice.

Development and validation of a nomogram for blood transfusion during intracranial aneurysm clamping surgery: a retrospective analysis

PURPOSE

Intraoperative blood transfusion is associated with adverse events. We aimed to establish a machine learning model to predict the probability of intraoperative blood transfusion during intracranial aneurysm surgery.

METHODS

Patients, who underwent intracranial aneurysm surgery in our hospital between January 2019 and December 2021 were enrolled. Four machine learning models were benchmarked and the best learning model was used to establish the nomogram, before conducting a discriminative assessment.

RESULTS

A total of 375 patients were included for analysis in this model, among whom 108 received an intraoperative blood transfusion during the intracranial aneurysm surgery. The least absolute shrinkage selection operator identified six preoperative relative factors: hemoglobin, platelet, D-dimer, sex, white blood cell, and aneurysm rupture before surgery. Performance evaluation of the classification error demonstrated the following: K-nearest neighbor, 0.2903; logistic regression, 0.2290; ranger, 0.2518; and extremely gradient boosting model, 0.2632. A nomogram based on a logistic regression algorithm was established using the above six parameters. The AUC values of the nomogram were 0.828 (0.775, 0.881) and 0.796 (0.710, 0.882) in the development and validation groups, respectively.

CONCLUSIONS

Machine learning algorithms present a good performance evaluation of intraoperative blood transfusion. The nomogram established using a logistic regression algorithm showed a good discriminative ability to predict intraoperative blood transfusion during aneurysm surgery.

Contemporary practice and perception of autologous blood salvage in hepato-pancreatico-biliary operations: an international survey

BACKGROUND

This study aimed to assess contemporary knowledge, attitudes and behaviors around transfusion of intraoperative salvaged blood (sRBCt) during hepato-pancreatico-biliary (HPB) operations. Findings are meant to inform the design of future studies that address provider concerns to change behaviors and improve patient outcomes.

METHODS

A survey was designed and assessed for relevance, readability and content, and distributed to an international audience of surgeons performing HPB operations.

RESULTS

The 237 respondents were predominantly distributed across North America (37.55%), Europe (27.43%) and Asia (19.83%). Roughly one-half (52.74%) of respondents had used sRBCt in HPB surgery before. Transplantation surgeons were more likely than HPB surgeons to have previously used sRBCt [odds ratio = 5.18 (95% CI 1.89-14.20)]. More respondents believed sRBCt was safe for non-cancer versus cancer operations (68.57% vs. 24.17%, p < 0.0001). Less than half (45.71%) of respondents believed that sRBCt was safe in clean-contaminated fields. Most did not utilize preoperative strategies to avoid donor transfusion.

CONCLUSION

Practices related to sRBCt in HPB operations vary widely and there is no consensus on its use. Concerns seem primarily related to cancer-specific and infectious outcomes. While further studies are pursued, surgeons may increase their utilization of preoperative strategies to boost hemoglobin levels for at risk patients.

Development and validation of a nomogram for predicting post-operative abdominal infection in patients undergoing pancreaticoduodenectomy

AIM

The aim of this retrospective study is to develop and validate a predictive nomogram for predicting the risk of post-operative abdominal infection (PAI) in patients undergoing pancreaticoduodenectomy (PD).

METHODS

A total of 360 patients who underwent PD were enrolled into this research and randomly divided into the development and validation group. The clinical data of patients were statistically compared and the nomogram was constructed based on the results of multivariate logistic regression analysis and stepwise (stepAIC) selection. The nomogram was internally and crossly validated by the development and validation cohort. The discriminatory ability of the nomogram was estimated by AUC (Area Under the receiver operating characteristic Curve), calibration curve and decision curve analysis.

RESULTS

After PD, post-operative abdominal infection occurred in 33.89% (n = 122) of patients. The nomogram showed that preoperative biliary drainage and C-reactive protein (CRP), direct bilirubin (DB), alkaline phosphatase (AKP) levels on the 3rd postoperative day (POD3) were independent prognostic factors for abdominal infection after PD. The internal and cross validation of Receiver Operating Characteristic (ROC) curve was statistically significant (AUC = 0.723 and 0.786, respectively). The calibration curves showed good agreement between nomogram predictions and actual observations. The decision curves showed that the nomogram was of great clinical value.

CONCLUSION

A nomogram based on perioperative risk factors such as preoperative biliary drainage, CRP, DB and AKP could simply and accurately predict the risk degree of PAI in patients undergoing PD.

The impact of preoperative anemia on pancreatic resection outcomes

BACKGROUND

Preoperative anemia is a risk factor for blood transfusions and delayed postoperative recovery, but few data are available for pancreatic surgery. Aim of the study was to analyze the impact of preoperative anemia on outcomes after pancreatic resection.

METHODS

Retrospective review of 1107 patients resected at San Raffaele Hospital (2015-2018). Preoperative anemia was defined as hemoglobin lower than 130 g/L for men and 120 g/L for women. Primary outcome was 90-day comprehensive complication index (CCI). Analysis was stratified according to type of surgery; proximal resections (pancreaticoduodenectomy and total pancreatectomy) versus distal pancreatectomy.

RESULTS

In 776 proximal resection patients, preoperative anemia was associated with increased CCI (24 ± 25 vs. 19 ± 23, p = 0.018) and perioperative allogenic blood transfusions (n = 124, 46% vs. n = 129, 26%; p < 0.001). Multivariate analysis showed that anemia was associated with a 7% (95%CI 0.02-0.57 p = 0.047) increase in CCI, and was an independent factor associated with perioperative blood transfusion (OR 2.762, 95%CI 1.72-4.49, p < 0.001). In 331 distal pancreatectomies, anemia was not associated to increased morbidity but only to an increased risk of perioperative blood transfusion.

CONCLUSION

Preoperative anemia is an independent risk factor for increased complication severity and blood transfusion in patients undergoing major pancreatic resection.

Development and validation of a nomogram to predict intraoperative blood transfusion for gastric cancer surgery

OBJECTIVE

To construct and validate a nomogram composed of preoperative variables to predict intraoperative blood transfusion for gastric cancer surgery.

BACKGROUND

Intraoperative transfusion for gastric cancer surgery is a common medical procedure that is associated with increased postoperative complications.

METHODS

A total of 999 patients who underwent gastrectomy between January 2010 and June 2019 were randomly allocated into the primary and validation cohorts in a 2:1 ratio. In the primary cohort, logistic analyses were performed to identify independent predictors for transfusion. Using the Akaike information criterion, selected variables were incorporated to construct a nomogram. Validations of the nomogram were performed in the primary and validation cohorts. The discrimination ability of the nomogram was assessed by the concordance index (C-index), and calibration was assessed by calibration curves and the Hosmer-Lemeshow goodness-of-fit test.

RESULTS

The following risk factors for transfusion were identified and used to construct the nomogram: ASA status (III-IV vs I-II: odds ratio [OR] 1.74), comorbidities (yes vs no: OR 1.57), tumour location (diffuse vs lower: OR 4.05), cTNM stage (III vs I: OR 1.95), and a preoperative haemoglobin level less than 80 g/L (vs over 120 g/L: OR 35.30). The C-index was 0.859 and 0.850 in the primary and validation cohorts, respectively, which both indicated good discrimination of the nomogram. Additionally, both calibration curves and Hosmer-Lemeshow tests (p-value 0.184 and 0.887, respectively) demonstrated high agreement between the predictions and actual outcomes.

CONCLUSION

A nomogram composed of preoperative variables to predict blood transfusion for gastric cancer surgery was effectively developed and validated. This nomogram could be used to improve the utilisation of red blood cells for gastrectomy.

Nomogram based on albumin and neutrophil-to-lymphocyte ratio for predicting postoperative complications after pancreaticoduodenectomy

Background

The aim of this study was to identify a preoperative inflammatory marker with the most predictive value for postoperative complications after pancreaticoduodenectomy (PD). We then combined it with other perioperative variables to construct and validate a nomogram for complications after PD.

Methods

A total of 223 patients who received PD from January 2014 to July 2019 at a high-volume (>60 PDs/year) pancreatic centers in China were included in this retrospective study. All of the PDs were performed by the same surgeon who is beyond the learning curve with more than 100 PDs over the previous 3 years before 2014. 15 preoperative inflammatory markers were collected, including neutrophils, lymphocytes, high-sensitivity C-reactive protein and lactic dehydrogenase. The inflammatory markers' predicting abilities for complications were analyzed by calculating the values of an area under the curve (AUC). The complications included surgical complications (such as pancreatic fistula, delayed gastric emptying and bile leakage) and medical complications (such as sepsis, pneumonia, urinary tract infection, acute heart failure and acute liver failure) in this study. Univariable and multivariable logistic regression analyses were performed to investigate the perioperative features for independent risk factors for complications after PD. Nomograms with or without the most predictive inflammatory for complications were subsequently developed based on multivariable logistic regression using Akaike information criterion. Nomograms' performance was quantified and compared in terms of calibration and discrimination. We studied the utility of the nomograms using decision curve analysis.

Results

The albumin/ NLR score (ANS) exhibited the highest AUC value (0.616) for predicting postoperative complications. ANS and approach method were identified as independent risk factors for complications. The nomogram with ANS had higher C-index (0.725) and better calibration. The NRI compared between nomograms was 0.160 (95% CI: 0.023-0.296; P=0.022). By decision curve analysis, the model with ANS had higher clinical value.

Conclusions

The ANS is a useful predictor and an independent risk factor for postoperative complications after PD. The nomogram with ANS was constructed with better performance and more clinical benefit for predicting postoperative complications.

Comparison of three lymph node staging methods for predicting outcome in breast cancer patients with mastectomy

Background

Axillary lymph node (ALN) staging is essential in predicting the clinical outcome of breast cancer (BC) patients. Traditionally, it follows the tumor-node-metastasis (TNM) staging, but its accuracy needs further improvement.

Methods

A total of 9,616 BC patients from the Surveillance, Epidemiology, and End Results (SEER) database and 675 patients from the First Affiliated Hospital of China Medical University underwent mastectomy together with ALN dissection were reviewed. Univariate and multivariate logistic analyses were conducted to find the most meaningful factors relevant to prognosis.

Results

After univariate and multivariate analyses, age, race, primary site, radiation, chemotherapy, grade, T-stage, estrogen receptor (ER), progesterone receptor (PR), total number of positive lymph nodes (pN), positive lymph node ratio (LNR) and log odds of positive LNs (LODDS) were found to be significantly associated with overall survival (OS). Using these non-LN risk factors, we further compared the efficacy of three different ALN staging methods in prognosis via nomograms. Harrell's concordance index (C-index) and Akaike Information Criterion (AIC) were used to measure nomogram performance of the ALN staging methods: pN: C-index=0.687 (95% CI: 0.678-0.696), AIC =61,398.24; LNR: C-index =0.691 (95% CI: 0.683-0.701), AIC =61,313.56; and LODDS: C-index =0.691 (95% CI: 0.682-0.700), AIC =61,315.60. We found that the nomogram incorporating LODDS had better predictive ability compared with other two methods. Furthermore, an external validation revealed a C-index of 0.753 (95% CI: 0.690-0.816) for the Asian population, which indicates the nomogram based on LODDS may have universality for both Western and Asian populations.

Conclusions

Compared with pN and LNR, LODDS showed higher homeostasis in LN evaluation, and showed marked efficacy in evaluating survival differences among patients with negative LN staging. We constructed a BC prognosis model by incorporating highly relevant clinical pathological factors and a new method of LN staging, which may greatly aid in guiding postoperative treatment.

Establishment and assessment of a nomogram for predicting blood transfusion risk in posterior lumbar spinal fusion

Background

The aim of this study was to determine the risk factors and develop a nomogram for blood transfusions after posterior lumbar spinal fusion (PSL).

Methods

We conducted a retrospective, single-center study based on 885 patients receiving PSL, and data was obtained from May 2015 to September 2019. Univariable and multivariable logistics regression analysis were conducted to identify risk factors for blood transfusion, and a nomogram was constructed to individually evaluate the risk of blood transfusion. Discrimination, calibration, and clinical usefulness were validated by the receiver operating characteristics (ROC), C-index, calibration plot, and decision curve analysis, respectively. Bootstrapping validation was performed to assess the performance of the model.

Results

Of 885 patients, 885 were enrolled in the final study population, and 289 received blood transfusion. Statistical analyses showed that low preoperative hemoglobin (Hb), longer time to surgery, operative time, levels of fusion > 1, longer surgery duration, and higher total intraoperative blood loss (IBL) were the risk factors for transfusion. The C-index was 0.898 (95% CI 0.847–0.949) in this dataset and 0.895 in bootstrapping validation, respectively. Calibration curve showed satisfied discrimination and calibration of the nomogram. Decision curve analysis (DCA) shown that the nomogram was clinical utility.

Conclusions

In summary, we investigated the relationship between the blood transfusion requirement and predictors: levels of fusion, operative time, time to surgery, total intraoperative EBL, and preoperative Hb level. Our nomogram with a robust performance in the assessment of risk of transfusion can contribute to clinicians in making clinical decision. However, external validation is still needed in the further.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13018-020-02053-2.

Prognostic Impact of Pedicle Clamping during Liver Resection for Colorectal Metastases

Pedicle clamping (PC) during liver resection for colorectal metastases (CRLM) is used to reduce blood loss and allogeneic blood transfusion (ABT). The effect on long-term oncologic outcomes is still under debate. A retrospective analysis of the impact of PC on ABT-demand regarding overall (OS) and recurrence-free survival (RFS) in 336 patients undergoing curative resection for CRLM was carried out. Survival analysis was performed by both univariate and multivariate methods and propensity-score (PS) matching. PC was employed in 75 patients (22%). No increased postoperative morbidity was monitored. While the overall ABT-rate was comparable (35% vs. 37%, p = 0.786), a reduced demand for more than two ABT-units was observed (p = 0.046). PC-patients had better median OS (78 vs. 47 months, p = 0.005) and RFS (36 vs. 23 months, p = 0.006). Multivariate analysis revealed PC as an independent prognostic factor for OS (HR = 0.60; p = 0.009) and RFS (HR = 0.67; p = 0.017). For PC-patients, 1:2 PS-matching (N = 174) showed no differences in the overall ABT-rate compared to no-PC-patients (35% vs. 40%, p = 0.619), but a trend towards reduced transfusion requirement (>2 ABT-units: 9% vs. 21%, p = 0.052; >4 ABT-units: 2% vs. 11%, p = 0.037) and better survival (OS: 78 vs. 44 months, p = 0.088; RFS: 36 vs. 24 months; p = 0.029). Favorable long-term outcomes and lower rates of increased transfusion demand were observed in patients with PC undergoing resection for CRLM. Further prospective evaluation of potential oncologic benefits of PC in these patients may be meaningful.

Preoperative tranexamic acid does not reduce transfusion rates in major oncologic surgery: Results of a randomized, double-blind, and placebo-controlled trial

BACKGROUND AND OBJECTIVES

Allogeneic blood transfusions are associated with worse postoperative outcomes in oncologic surgery. The aim of this study was to introduce a preoperative intervention to reduce transfusion rates in this population.

METHODS

Adult patients undergoing major oncologic surgery in five categories with similar transfusion rates were recruited. Enrollees received a single preoperative intravenous dose of placebo or tranexamic acid (1000 mg). The primary outcome measure was perioperative transfusion rate. Secondary outcome measures included: estimated blood loss, thromboembolic events, morbidity, hospital length of stay, and readmission rate.

RESULTS

Seventy-six patients were enrolled, 39 in the tranexamic acid group and 37 in the placebo group, respectively. Demographics and surgery type were equivalent between groups. The transfusion rates were 8 out of 39 (20.5%) in the tranexamic acid group and 5 out of 37 (13.5%) in the placebo group, respectively (P = .418). Median estimated blood loss was 400 mL (interquartile range [IQR] = 150-600) in the tranexamic acid group compared with 300 mL (IQR = 150-800) in the placebo group (P = .983). There was one pulmonary embolism in each arm and no deep venous thrombosis (P > .999).

CONCLUSION

Preoperative administration of tranexamic acid at a 1000 mg intravenous dose does not decrease transfusion rates or estimated blood loss in patients undergoing major oncologic surgery.

Perioperative blood transfusions following hepatic lobectomy: A national analysis of academic medical centers in the modern era

BACKGROUND

The purpose of the study was to characterize the prevalence and impact of perioperative blood use for patients undergoing hepatic lobectomy at academic medical centers.

METHODS

The University HealthSystem Consortium (UHC) database was queried for hepatic lobectomies performed between 2011 and 2014 (n = 6476). Patients were grouped according to transfusion requirements into high (>5 units, 7%), medium (2-5 units, 6%), low (1 unit, 8%), and none (0 units, 79%) during hospital stay for comparison of outcomes.

RESULTS

Over 20% of patients undergoing hepatic lobectomy received blood perioperatively, of which 35% required more than 5 units. Patients with high transfusion requirements had increased severity of illness (p < 0.01). High transfusion requirements correlated with increased readmission rates (23.4% vs. 19.2% vs. 16.6% vs. 13.5%), total direct costs ($31,982 vs. $20,859 vs. $19,457 vs. $16,934), length of stay (9 days vs. 8 vs. 7 vs. 6), and in-hospital mortality (10.8% vs. 2.0% vs. 0.9% vs. 2.0%) compared to medium, low, and no transfusion amounts (all p < 0.01). Neither center nor surgeon volume were associated with transfusion use.

CONCLUSION

High transfusion requirements after hepatic lobectomy in the United States are associated with worse perioperative quality measures, but may not be influenced by center or surgeon volume.

A novel online prognostic tool to predict long-term survival after liver resection for intrahepatic cholangiocarcinoma: The "metro-ticket" paradigm

BACKGROUND

The aim of the current study was to develop an online calculator to predict survival after liver resection for intrahepatic cholangiocarcinoma (ICC) based on the "metro-ticket" paradigm.

METHODS

Between 1990 and 2016, patients who underwent liver resection for ICC were identified in an international multi-institutional database. The final multivariable model of survival was used to develop an online prognostic calculator of survival.

RESULTS

Among 643 patients, actual 5-year overall survival (OS) after resection for ICC was 42.7%. On multivariable analysis, CA19-9 > 200 (hazard ratio (HR), 2.62; 95% CI, 2.01-3.42), sum of the number and largest tumor size >7 (HR, 1.88; 95% CI, 1.46-2.42), N1 disease (HR, 2.87; 95% CI, 1.98-4.16), R1 resection (HR, 1.72; 95% CI, 1.21-2.46), poor/undifferentiated tumor grade (HR, 1.74; 95% CI, 1.25-2.44), major vascular invasion (HR, 1.47; 95% CI, 1.03-2.10), and adjuvant chemotherapy (HR, 0.64; 95% CI, 0.45-0.89) were significantly associated with survival and were included in the online calculator. The predictive accuracy of the model was good to very good as the C-statistics to predict 5-year OS was 0.696 in the training dataset and 0.672 with bootstrapping resamples (n = 5000) in the test dataset.

CONCLUSION

A novel, online calculator was developed to estimate the 5-year survival probability for patients undergoing resection for ICC. This tool could help provide useful information to guide treatment decision-making and inform conversations about prognosis.

Distinct risk factors for early and late blood transfusion following pancreaticoduodenectomy

BACKGROUND

The International Study Group of Pancreatic Surgery (ISGPS) has defined two periods of postpancreatectomy hemorrhage, early (<24 h) and late (>24 h). A previously published Blood Usage Risk Score (BURS) aimed to predict early and late blood transfusion. The primary aim of this study was to define risk factors for early and late blood transfusion after pancreaticoduodenectomy. Secondary aims were to assess the predictive accuracy of the BURS.

METHODS

In this retrospective observational study, multivariable analyses were used to identify independent risk factors for both early and late blood transfusion. The predictive ability of the BURS was then assessed using a receiver operating characteristic (ROC) curve analysis.

RESULTS

Among 628 patients, 99 (15.8%) and 144 (22.9%) received early and late blood transfusion, respectively. Risk factors for blood transfusion differed between early and late periods. Preoperative anemia and venous resection were associated with early blood transfusion whilst Whipple's resection (as opposed to pylorus preserving pancreaticoduodenectomy), lack of biliary stent and a narrow pancreatic duct were predictors of late blood transfusion. The BURS was significantly predictive of early blood transfusion, albeit with a modest degree of accuracy (AUROC: 0.700, P < 0.001), but not of late blood transfusion (AUROC: 0.525, P = 0.360). Late blood transfusion was independently associated with increasing severity of postoperative pancreatic fistula (POPF) (OR: 1.85, 3.18 and 9.97 for biochemical, types B and C POPF, respectively, relative to no POPF).

CONCLUSIONS

Two largely different sets of variables are related to early and late blood transfusion following pancreaticoduodenectomy. The BURS was significantly associated with early, albeit with modest predictive accuracy, but not late blood transfusion. An understanding of POPF risk allows assessment of the need for late blood transfusion.

Postoperative bleeding risk prediction for patients undergoing colorectal surgery

BACKGROUND

There is limited consensus regarding risk factors for postoperative bleeding. The objective of this work was to investigate the capability of machine learning techniques in combination with practice-based longitudinal electronic medical record data for identifying potential new risk factors for postoperative bleeding and predicting patients at high risk of postoperative bleeding.

METHODS

A retrospective study was conducted for patients who underwent colorectal surgery 1998-2015 at a single tertiary referral center. Various predictors were extracted from electronic medical record. The outcome of interest was the occurrence of postoperative bleeding within 7 days of surgery. Logistic regression and gradient boosting machine models were trained. Area under the receiver operating curve and area under the precision recall curve were used to evaluate the performance to different models.

RESULTS

Of 13,399 cases undergoing colorectal resection, 1,680 (12.5%) experienced postoperative bleeding. A total of 299 variables were evaluated. Logistic regression and gradient boosting machine models returned an area under the receiver operating curve of 0.735 and 0.822 and area under the precision recall curve of 0.287 and 0.423, respectively. In addition to well-known risk factors for postoperative bleeding, nutrition (ranked third), weakness (ranked fifth), patient mobility (ranked sixth), and activity level (ranked eighth) were found to be novel predictors in the gradient boosting machine model based on permutation importance.

CONCLUSION

The study identified measures of functional capacity of patient as novel predictors of postoperative bleeding. The study found that risk of postoperative bleeding can be assessed, allowing for better use of human resources in addressing this important adverse event after surgery.