Prediction of the severity of acute kidney injury after on-pump cardiac surgery

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

BACKGROUND

Allogeneic 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.

METHODS

Data 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.

RESULTS

In 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.

CONCLUSIONS

The 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.

A systematic review of cardiac surgery clinical prediction models that include intra-operative variables

BACKGROUND

Most cardiac surgery clinical prediction models (CPMs) are developed using pre-operative variables to predict post-operative outcomes. Some CPMs are developed with intra-operative variables, but none are widely used. The objective of this systematic review was to identify CPMs with intra-operative variables that predict short-term outcomes following adult cardiac surgery.

METHODS

Ovid MEDLINE and EMBASE databases were searched from inception to December 2022, for studies developing a CPM with at least one intra-operative variable. Data were extracted using a critical appraisal framework and bias assessment tool. Model performance was analysed using discrimination and calibration measures.

RESULTS

A total of 24 models were identified. Frequent predicted outcomes were acute kidney injury (9/24 studies) and peri-operative mortality (6/24 studies). Frequent pre-operative variables were age (18/24 studies) and creatinine/eGFR (18/24 studies). Common intra-operative variables were cardiopulmonary bypass time (16/24 studies) and transfusion (13/24 studies). Model discrimination was acceptable for all internally validated models (AUC 0.69-0.91). Calibration was poor (15/24 studies) or unreported (8/24 studies). Most CPMs were at a high or indeterminate risk of bias (23/24 models). The added value of intra-operative variables was assessed in six studies with statistically significantly improved discrimination demonstrated in two.

CONCLUSION

Weak reporting and methodological limitations may restrict wider applicability and adoption of existing CPMs that include intra-operative variables. There is some evidence that CPM discrimination is improved with the addition of intra-operative variables. Further work is required to understand the role of intra-operative CPMs in the management of cardiac surgery patients.

Association between intraoperative hyperglycemia/hyperlactatemia and acute kidney injury following on-pump cardiac surgery: a retrospective cohort study

Background

Despite the long-lasting notion about the substantial contribution of intraoperative un-stabilization of homeostasis factors on the incidence on acute kidney injury (AKI), the possible influence of intraoperative glucose or lactate management, as a modifiable factor, on the development of AKI remains inconclusive.

Objectives

To investigated the relationship between intraoperative hyperglycemia, hyperlactatemia, and postoperative AKI in cardiac surgery.

Methods

A retrospective cohort study was conducted among 4,435 adult patients who underwent on-pump cardiac surgery from July 2019 to March 2022. Intraoperative hyperglycemia and hyperlactatemia were defined as blood glucose levels >10 mmol/L and lactate levels >2 mmol/L, respectively. The primary outcome was the incidence of AKI. All statistical analyses, including t tests, Wilcoxon rank sum tests, chi-square tests, Fisher's exact test, Kolmogorov-Smirnov test, logistic regression models, subgroup analyses, collinearity analysis, and receiver operating characteristic analysis, were performed using the statistical software program R version 4.1.1.

Results

Among the 4,435 patients in the final analysis, a total of 734 (16.55%) patients developed AKI after on-pump cardiac surgery. All studied intraoperative metabolic disorders was associated with increased AKI risk, with most pronounced odds ratio (OR) noted for both hyperglycemia and hyperlactatemia were present intraoperatively [adjusted OR 3.69, 95% confidence intervals (CI) 2.68-5.13, p < 0.001]. Even when hyperglycemia or hyperlactatemia was present alone, the risk of postoperative AKI remained elevated (adjusted OR 1.97, 95% CI 1.50-2.60, p < 0.001).

Conclusion

The presence of intraoperative hyperglycemia and hyperlactatemia may be associated with postoperative acute kidney injury (AKI) in patients undergoing on-pump cardiac surgery. Proper and timely interventions for these metabolic disorders are crucially important in mitigating the risk of AKI.

Interpretable machine learning models for early prediction of acute kidney injury after cardiac surgery

OBJECTIVE

Postoperative acute kidney injury (PO-AKI) is a common complication after cardiac surgery. We aimed to evaluate whether machine learning algorithms could significantly improve the risk prediction of PO-AKI.

METHODS

The retrospective cohort study included 2310 adult patients undergoing cardiac surgery in a tertiary teaching hospital, China. Postoperative AKI and severe AKI were identified by the modified KDIGO definition. The sample was randomly divided into a derivation set and a validation set based on a ratio of 4:1. Exploiting conventional logistic regression (LR) and five ML algorithms including decision tree, random forest, gradient boosting classifier (GBC), Gaussian Naive Bayes and multilayer perceptron, we developed and validated the prediction models of PO-AKI. We implemented the interpretation of models using SHapley Additive exPlanation (SHAP) analysis.

RESULTS

Postoperative AKI and severe AKI occurred in 1020 (44.2%) and 286 (12.4%) patients, respectively. Compared with the five ML models, LR model for PO-AKI exhibited the largest AUC (0.812, 95%CI: 0.756, 0.860, all P < 0.05), sensitivity (0.774, 95%CI: 0.719, 0.813), accuracy (0.753, 95%CI: 0.719, 0.781) and Youden index (0.513, 95%CI: 0.451, 0.573). Regarding severe AKI, GBC algorithm showed a significantly higher AUC than the other four ML models (all P < 0.05). Although no significant difference (P = 0.173) was observed in AUCs between GBC (0.86, 95%CI: 0.808, 0.902) and conventional logistic regression (0.803, 95%CI: 0.746, 0.852), GBC achieved greater sensitivity, accuracy and Youden index than conventional LR. Notably, SHAP analyses showed that preoperative serum creatinine, hyperlipidemia, lipid-lowering agents and assisted ventilation time were consistently among the top five important predictors for both postoperative AKI and severe AKI.

CONCLUSION

Logistic regression and GBC algorithm demonstrated moderate to good discrimination and superior performance in predicting PO-AKI and severe AKI, respectively. Interpretation of the models identified the key contributors to the predictions, which could potentially inform clinical interventions.

Postoperative acute kidney injury after on-pump cardiac surgery in patients with connective tissue disease

Objective

Patients with connective tissue disease have a poor prognosis after receiving cardiac surgery. This study described the clinical scenarios and investigated factors correlated with acute kidney injury (AKI) after on-pump cardiac surgery in patients with systemic lupus erythematosus (SLE) or vasculitis.

Methods

Patients with SLE or vasculitis who underwent on-pump cardiac surgery from March 2002 to March 2022 were enrolled, while patients with preoperative renal dysfunction were excluded. AKI was defined according to the Kidney Disease: Improving Global Outcomes (KDIGO) criteria. Uni- and multivariable analyses were performed to identify potential factors associated with postoperative AKI.

Results

Among 123 patients enrolled, 39 (31.7%) developed AKI within seven days after receiving on-pump cardiac surgery. Four patients died in the hospital, resulting in an overall in-hospital mortality of 3.3%, and all deaths occurred in the AKI group. Patients in the AKI group also had longer ICU stays (median difference 3.0 day, 95% CI: 1.0-4.0, P < 0.001) and extubation time (median difference 1.0 days, 95% CI: 0-2.0, P < 0.001) than those in the non-AKI group. Multivariable logistic regression revealed that BMI over 24 kg/m2 (OR: 3.00, 95% CI: 1.24-7.28) and comorbid SLE (OR: 4.73, 95% CI: 1.73-12.93) were independently correlated with postoperative AKI.

Conclusion

Factors potentially correlated with AKI following on-pump cardiac surgery in patients with connective tissue disease were explored. Clinicians should pay more attention to preoperative evaluation and intraoperative management in patients with risk factors.

A novel model for predicting a composite outcome of major complications after valve surgery

Background

On-pump valve surgeries are associated with high morbidity and mortality. The present study aimed to reliably predict a composite outcome of postoperative complications using a minimum of easily accessible clinical parameters.

Methods

A total of 7,441 patients who underwent valve surgery were retrospectively analyzed. Data for 6,220 patients at West China Hospital of Sichuan University were used to develop a predictive model, which was validated using data from 1,221 patients at the Second Affiliated Hospital of Zhejiang University School of Medicine. The primary outcome was a composite of major complications: all-cause death in hospital, stroke, myocardial infarction, and severe acute kidney injury. The predictive model was constructed using the least absolute shrinkage and selection operator as well as multivariable logistic regression. The model was assessed in terms of the areas under receiver operating characteristic curves, calibration, and decision curve analysis.

Results

The primary outcome occurred in 129 patients (2.1%) in the development cohort and 71 (5.8%) in the validation cohort. Six variables were retained in the predictive model: New York Heart Association class, diabetes, glucose, blood urea nitrogen, operation time, and red blood cell transfusion during surgery. The C-statistics were 0.735 (95% CI, 0.686-0.784) in the development cohort and 0.761 (95% CI, 0.694-0.828) in the validation cohort. For both cohorts, calibration plots showed good agreement between predicted and actual observations, and ecision curve analysis showed clinical usefulness. In contrast, the well-established SinoSCORE did not accurately predict the primary outcome in either cohort.

Conclusions

This predictive nomogram based on six easily accessible variables may serve as an "early warning" system to identify patients at high risk of major complications after valve surgery.

Clinical Trial Registration

[www.ClinicalTrials.gov], identifier [NCT04476134].

Perioperative parameters-based prediction model for acute kidney injury in Chinese population following valvular surgery

Background

Acute kidney injury (AKI) is a relevant complication after cardiac surgery and is associated with significant morbidity and mortality. Existing risk prediction tools have certain limitations and perform poorly in the Chinese population. We aimed to develop prediction models for AKI after valvular cardiac surgery in the Chinese population.

Methods

Models were developed from a retrospective cohort of patients undergoing valve surgery from December 2013 to November 2018. Three models were developed to predict all-stage, or moderate to severe AKI, as diagnosed according to Kidney Disease: Improving Global Outcomes (KDIGO) based on patient characteristics and perioperative variables. Models were developed based on lasso logistics regression (LLR), random forest (RF), and extreme gradient boosting (XGboost). The accuracy was compared among three models and against the previously published reference AKICS score.

Results

A total of 3,392 patients (mean [SD] age, 50.1 [11.3] years; 1787 [52.7%] male) were identified during the study period. The development of AKI was recorded in 50.5% of patients undergoing valve surgery. In the internal validation testing set, the LLR model marginally improved discrimination (C statistic, 0.7; 95% CI, 0.66-0.73) compared with two machine learning models, RF (C statistic, 0.69; 95% CI, 0.65-0.72) and XGBoost (C statistic, 0.66; 95% CI, 0.63-0.70). A better calibration was also found in the LLR, with a greater net benefit, especially for the higher probabilities as indicated in the decision curve analysis. All three newly developed models outperformed the reference AKICS score.

Conclusion

Among the Chinese population undergoing CPB-assisted valvular cardiac surgery, prediction models based on perioperative variables were developed. The LLR model demonstrated the best predictive performance was selected for predicting all-stage AKI after surgery.

Clinical trial registration

Trial registration: Clinicaltrials.gov, NCT04237636.