A Prediction Model for Assessing Prognosis in Critically Ill Patients with Sepsis-associated Acute Kidney Injury

Abstract

Background: Sepsis-associated acute kidney injury (SA-AKI) is a common problem in critically ill patients and is associated with high morbidity and mortality. Early prediction of the survival of hospitalized patients with SA-AKI is necessary, but a reliable and valid prediction model is still lacking. Methods: We conducted a retrospective cohort analysis based on a training cohort of 2,066 patients enrolled from the Multiparameter Intelligent Monitoring in Intensive Care Database III (MIMIC III) and a validation cohort of 102 patients treated at Nanfang Hospital of Southern Medical University. Least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox regression analysis were used to identify predictors for survival. Areas under the ROC curves (AUC), the concordance index (C-index), and calibration curves were used to evaluate the efficiency of the prediction model (SAKI) in both cohorts. Results: The overall mortality of SA-AKI was approximately 18%. Age, admission type, liver disease, metastatic cancer, lactate, BUN/SCr, admission creatinine, positive culture, and AKI stage were independently associated with survival and combined in the SAKI model. The C-index in the training and validation cohorts was 0.73 and 0.72. The AUC in the training cohort was 0.77, 0.72, and 0.70 for the 7-day, 14-day, and 28-day probability of in-hospital survival, respectively, while in the external validation cohort, it was 0.83, 0.73, and 0.67. SAPSII and SOFA scores showed poorer performance. Calibration curves demonstrated a good consistency. Conclusions: Our SAKI model has predictive value for in-hospital mortality of SA-AKI in critically ill patients and outperforms generic scores. KEYWORDS-Acute kidney injury, critically ill patients, prediction model, sepsis ABBREVIATIONS-AKI-Acute kidney injury; AUC-Area under the receiver operator characteristic curves; BCR-Blood urea nitrogen/serum creatinine radio; CI-Confidence interval; C-index-Concordance index; HR-Hazard radio; ICU-Intensive care unit; IQR-Interquartile range; KDIGO-Kidney disease-improving global outcomes; LASSO-Least absolute shrinkage and selection operator; MAP-Mean arterial pressure; MIMIC-Multiparameter Intelligent Monitoring in Intensive Care Database III; RRT-Renal replacement therapy; SA-AKI-Sepsis-associated acute kidney injury; SAKI-The prediction model for assessing prognosis in critically ill patients with sepsis-associated acute kidney injury; SAPSII-Simplified Acute Physiology Score II; SCr-Serum creatinine; SOFA-Sequential Organ Failure Assessment

Development and Validation of Machine Learning Models for Real-Time Mortality Prediction in Critically Ill Patients With Sepsis-Associated Acute Kidney Injury

“…Generic scores, such as the SOFA score and SAPS-II, are widely used for outcome prediction in critical care settings. However, they have shown controversial results on predictive performance for mortality in AKI patients ( 7 , 8 , 30 – 32 ). Recently, several models have been proposed to predict AKI mortality in unselected ICU patients ( 31 , 32 ), but few have been validated in patients with SA-AKI.…”

“…The five-variable score showed better performance than the generic models, with a C-statistic of 0.82, but was limited to a single center and small sample size. In addition, Hu et al ( 8 ) established a prediction model for in-hospital mortality in critically ill patients with SA-AKI. However, the model included only static clinical variables and showed insufficient predictive power.…”

“…Studies have shown that widely-used severity scores, such as the Simplified Acute Physiology Score II (SAPS-II) and the Sequential Organ Failure Assessment (SOFA) score, exhibit insufficient power for outcome prediction in SA-AKI patients ( 7 , 8 ). A few prediction models for mortality in patients with SA-AKI have been established ( 7 , 8 ). However, they were limited to small sample size or inadequate predictive performance.…”

No sex differences in the incidence, risk factors and clinical impact of acute kidney injury in critically ill patients with sepsis

“…We did several sensitivity and subgroup analyses for the primary outcomes. First, given the considerable impact of age on sex hormones (7), we conducted a stratified analysis according to age in decades (≤40, (40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50), (50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60), (60-70], (70-80], (80-90] and>90 years). Second, to determine whether the illness severity of AKI would influence the results, we further compared patients without AKI to those with mild AKI or with severe AKI.…”

What's New in Shock, October 2021?

“…We desperately need a reliable prediction model for early prediction of mortality in hospitalized patients with SA-AKI. Hu and co-workers (19) approach this problem with a retrospective cohort analysis based on a training cohort of 2,066 patients enrolled from the Multiparameter Intelligent Monitoring in Intensive Care Database III (MIMIC III) and a validation cohort of 102 patients treated at Nanfang Hospital of Southern Medical University. Age, admission type, liver disease, metastatic cancer, lactate, BUN/SCr, admission creatinine, positive culture, and AKI stage were independently associated with survival and therefore combined in the SAKI model.…”