Mortality Prediction Using SaO2/FiO2 Ratio Based on eICU Database Analysis

Utility of Clinical Variables for Deciding Palliative Care in Paraquat Poisoning: A Retrospective Study

Background

Patients with paraquat poisoning (PP) have a mortality rate comparable to that of advanced malignancies, yet palliative care is seldom considered in these patients. This audit aimed to identify triggers for early palliative care referral in critically ill patients with PP.

Methods

Medical records of patients with PP were audited. Predictors of mortality within 48 hours of hospitalization and 24 hours of intensive care unit (ICU) admission were considered as triggers for palliative care referral.

Results

Among 108 patients, 84 complete records were analyzed, and 53 out of 84 (63.1%) expired. Within 48 hours after hospitalization, the lowest oxygen partial pressure in arterial blood to a fraction of inspired oxygen [the ratio of partial pressure of oxygen in arterial blood (PaO2) to the fraction of inspiratory oxygen concentration (FiO2) (PaO2/FiO2)] was the independent predictor of mortality, cut-off ≤ 197; the area under the curve (AUC), 0.924; sensitivity, 97%; specificity, 78%; p <0.001; and 95% confidence interval (CI): 0.878-0.978. Kaplan-Meier survival plot showed that the mean survival time of patients with the lowest PaO2/FiO2, ≤197, was 4.64 days vs 17.20 days with PaO2/FiO2 >197 (log-rank p < 0.001). Sequential organ failure assessment (SOFA) score within 24 hours of ICU admission had a cut-off ≥9; AUC, 0.980; p < 0.001; 95% CI: 0.955-1.000; 91% sensitivity; and 90% specificity for mortality prediction. Out of the total of 84 patients with PP analyzed, there were 11 patients admitted to the high dependency units (13.1%) and 73 patients admitted to the ICU (86.9%). Out of the total of 84 patients of PP in whom data was analyzed, 53 (63.1%) patients required ventilator support. All the 53 patients who required ventilator support due to worsening hypoxemia, eventually expired.

Conclusion

The lowest PaO2/FiO2 ≤ 197 within 48 hours of hospitalization, SOFA score ≥9 within 24 hours of ICU admission or need for mechanical ventilation are predictors of mortality in PP patients, who might benefit from early palliative care.

How to cite this article

Rao S, Maddani SS, Chaudhuri S, Bhatt MT, Karanth S, Damani A, et al. Utility of Clinical Variables for Deciding Palliative Care in Paraquat Poisoning: A Retrospective Study. Indian J Crit Care Med 2024;28(5):453-460.

Prognostic Accuracy of SpO2-based Respiratory Sequential Organ Failure Assessment for Predicting In-hospital Mortality

Introduction

In this study we aimed to investigate the prognostic accuracy for predicting in-hospital mortality using respiratory Sequential Organ Failure Assessment (SOFA) scores by the conventional method of missing-value imputation with normal partial pressure of oxygen (PaO2)- and oxygen saturation (SpO2)-based estimation methods.

Methods

This was a single-center, retrospective cohort study of patients with suspected infection in the emergency department. The primary outcome was in-hospital mortality. We compared the area under the receiver operating characteristics curve (AUROC) and calibration results of the conventional method (normal value imputation for missing PaO2) and six SpO2-based methods: using methods A, B, PaO2 is estimated by dividing SpO2 by a scale; with methods C and D, PaO2 was estimated by a mathematical model from a previous study; with methods E, F, respiratory SOFA scores was estimated by SpO2 thresholds and respiratory support use; with methods A, C, E are SpO2-based estimation for all PaO2 values, while methods B, D, F use such estimation only for missing PaO2 values.

Results

Among the 15,119 patients included in the study, the in-hospital mortality rate was 4.9%. The missing PaO2was 56.0%. The calibration plots were similar among all methods. Each method yielded AUROCs that ranged from 0.735-0.772. The AUROC for the conventional method was 0.755 (95% confidence interval [CI] 0.736-0.773). The AUROC for method C (0.772; 95% CI 0.754-0.790) was higher than that of the conventional method, which was an SpO2-based estimation for all PaO2 values. The AUROC for total SOFA score from method E (0.815; 95% CI 0.800-0.831) was higher than that from the conventional method (0.806; 95% CI 0.790-0.822), in which respiratory SOFA was calculated by the predefined SpO2 cut-offs and oxygen support.

Conclusion

In non-ICU settings, respiratory SOFA scores estimated by SpO2 might have acceptable prognostic accuracy for predicting in-hospital mortality. Our results suggest that SpO2-based respiratory SOFA score calculation might be an alternative for evaluating respiratory organ failure in the ED and clinical research settings.

Comparison of machine learning and logistic regression as predictive models for adverse maternal and neonatal outcomes of preeclampsia: A retrospective study

Introduction

Preeclampsia, one of the leading causes of maternal and fetal morbidity and mortality, demands accurate predictive models for the lack of effective treatment. Predictive models based on machine learning algorithms demonstrate promising potential, while there is a controversial discussion about whether machine learning methods should be recommended preferably, compared to traditional statistical models.

Methods

We employed both logistic regression and six machine learning methods as binary predictive models for a dataset containing 733 women diagnosed with preeclampsia. Participants were grouped by four different pregnancy outcomes. After the imputation of missing values, statistical description and comparison were conducted preliminarily to explore the characteristics of documented 73 variables. Sequentially, correlation analysis and feature selection were performed as preprocessing steps to filter contributing variables for developing models. The models were evaluated by multiple criteria.

Results

We first figured out that the influential variables screened by preprocessing steps did not overlap with those determined by statistical differences. Secondly, the most accurate imputation method is K-Nearest Neighbor, and the imputation process did not affect the performance of the developed models much. Finally, the performance of models was investigated. The random forest classifier, multi-layer perceptron, and support vector machine demonstrated better discriminative power for prediction evaluated by the area under the receiver operating characteristic curve, while the decision tree classifier, random forest, and logistic regression yielded better calibration ability verified, as by the calibration curve.

Conclusion

Machine learning algorithms can accomplish prediction modeling and demonstrate superior discrimination, while Logistic Regression can be calibrated well. Statistical analysis and machine learning are two scientific domains sharing similar themes. The predictive abilities of such developed models vary according to the characteristics of datasets, which still need larger sample sizes and more influential predictors to accumulate evidence.

Outcome Prediction in Critically-Ill Patients with Venous Thromboembolism and/or Cancer Using Machine Learning Algorithms: External Validation and Comparison with Scoring Systems

Intensive care unit (ICU) patients with venous thromboembolism (VTE) and/or cancer suffer from high mortality rates. Mortality prediction in the ICU has been a major medical challenge for which several scoring systems exist but lack in specificity. This study focuses on two target groups, namely patients with thrombosis or cancer. The main goal is to develop and validate interpretable machine learning (ML) models to predict early and late mortality, while exploiting all available data stored in the medical record. To this end, retrospective data from two freely accessible databases, MIMIC-III and eICU, were used. Well-established ML algorithms were implemented utilizing automated and purposely built ML frameworks for addressing class imbalance. Prediction of early mortality showed excellent performance in both disease categories, in terms of the area under the receiver operating characteristic curve (AUC–ROC): VTE-MIMIC-III 0.93, eICU 0.87, cancer-MIMIC-III 0.94. On the other hand, late mortality prediction showed lower performance, i.e., AUC–ROC: VTE 0.82, cancer 0.74–0.88. The predictive model of early mortality developed from 1651 VTE patients (MIMIC-III) ended up with a signature of 35 features and was externally validated in 2659 patients from the eICU dataset. Our model outperformed traditional scoring systems in predicting early as well as late mortality. Novel biomarkers, such as red cell distribution width, were identified.