Risk Prediction in Clinical Practice: A Practical Guide for Cardiothoracic Surgeons

Advanced statistical methods for hazard modeling in cardiothoracic surgery: a comprehensive review of techniques and approaches

Hazard modeling in cardiothoracic surgery, crucial for understanding patient outcomes, utilizes survival analysis like the Cox proportional hazards model. Kaplan-Meier curves are employed in survival analysis to represent the probability of survival over time. While Cox assumes proportional hazards, the Fine-Gray model deals with competing risks. Parametric models (e.g., Weibull) specify survival distributions, unlike Cox. Bayesian analysis integrates prior knowledge with data. Machine learning, including decision trees and support vector machines, enhances risk prediction by analyzing extensive datasets. However, it is important to note that whatever new approaches one may adopt will enhance the quality of risk assessment and not the risk assessment as such. Preprocessing is vital for data quality in complex cardiovascular datasets, alongside robust validation methods like cross-validation for model reliability across patient cohorts.

Risk prediction algorithms for prolonged postoperative ileus: A systematic review

AIM

Prolonged postoperative ileus (PPOI) is common and is associated with a significant healthcare burden. Previous studies have attempted to predict PPOI clinically using risk prediction algorithms. The aim of this work was to systematically review and compare risk prediction algorithms for PPOI following colorectal surgery.

METHOD

A systematic literature search was conducted using MEDLINE, Embase, Web of Science and CINAHL Plus. Studies that developed and/or validated a risk prediction algorithm for PPOI in adults following colorectal surgery were included. Data were collected on study design, population and operative characteristics, the definition of PPOI used and risk prediction algorithm design and performance. Quality appraisal was assessed using the PROBAST tool.

RESULTS

Eleven studies with 87 549 participants were included in our review. Most were retrospective, single-centre analyses (6/11, 55%) and rates of PPOI varied from 10% to 28%. The most commonly used variables were sex (8/11, 73%), age (6/11, 55%) and surgical approach (5/11, 45%). Area under the curve ranged from 0.68-0.78, and only three models were validated. However, there was significant variation in the definition of PPOI used. No study reported sensitivity, specificity or positive/negative predictive values.

CONCLUSION

Currently available risk prediction algorithms for PPOI appear to discriminate moderately well, although there is a lack of validation data. Future studies should aim to use a standardized definition of PPOI, comprehensively report model performance and validate their findings using internal and external methodologies.

External validation of the pediatric International IgA Nephropathy Prediction Tool in a central China cohort

BACKGROUND

This study aimed to externally validate the pediatric International IgA Nephropathy (IgAN) Prediction Tool updated from the adult IgAN Prediction Tool.

METHODS

439 children with biopsy-confirmed idiopathic IgAN were enrolled in this external validation study. The primary outcome was a 30% decline in eGFR or end-stage kidney disease. We evaluated the discrimination using Harrell's C-index, the receiver operating characteristic (ROC) curve, and Kaplan-Meier curves for four risk groups (< 16th [low risk], ∼16 to < 50th [intermediate risk], ∼50 to < 84th [high risk], and ≥ 84th percentiles [highest risk] of linear predictor). Calibration was assessed using calibration plots.

RESULTS

The median follow-up time of the 439 patients was 4.5 (2.7-6.8) years, and 27 patients reached the primary outcome. Compared with the reported cohorts, our cohort was more contemporary, with milder proteinuria at biopsy, and had lower proportions of S1 and T1 lesions. Harrell's C-index and area under the ROC curve at 5 years were < 0.7 for both the models with and without race. The Kaplan-Meier curves of the risk groups were not well separated for the two models, only separated completely between the highest-risk group and the others for the model without race. The two models generally overestimated the risk of the primary outcome, CONCLUSION: The model without race could accurately distinguish the highest-risk patients from patients with low, intermediate, and high risk for kidney progression. Discrimination and calibration for the full model with or without race were unsatisfactory in this contemporary cohort in central China.