An early warning model to predict acute kidney injury in sepsis patients with prior hypertension

Abstract

Background: In the context of sepsis patients, hypertension has a significant impact on the likelihood of developing sepsis-associated acute kidney injury (S-AKI), leading to a considerable burden. Moreover, sepsis is responsible for over 50 % of cases of acute kidney injuries (AKI) and is linked to an increased likelihood of death during hospitalization. The objective of this research is to develop a dependable and strong nomogram framework, utilizing the variables accessible within the first 24 h of admission, for the anticipation of S-AKI in sepsis patients who have hypertension.

Methods: In this study that looked back, a total of 462 patients with sepsis and high blood pressure were identified from Nanfang Hospital. These patients were then split into a training set (consisting of 347 patients) and a validation set (consisting of 115 patients). A multivariate logistic regression analysis and a univariate logistic regression analysis were performed to identify the factors that independently predict S-AKI. Based on these independent predictors, the model was constructed. To evaluate the efficacy of the designed nomogram, several analyses were conducted, including calibration curves, receiver operating characteristics curves, and decision curve analysis.

Results: The findings of this research indicated that diabetes, prothrombin time activity (PTA), thrombin time (TT), cystatin C, creatinine (Cr), and procalcitonin (PCT) were autonomous prognosticators for S-AKI in sepsis individuals with hypertension. The nomogram model, built using these predictors, demonstrated satisfactory discrimination in both the training (AUC = 0.823) and validation (AUC = 0.929) groups. The S-AKI nomogram demonstrated superior predictive ability in assessing S-AKI within the hypertension grade I (AUC = 0.901) set, surpassing the hypertension grade II (AUC = 0.816) and III (AUC = 0.810) sets. The nomogram exhibited satisfactory calibration and clinical utility based on the calibration curve and decision curve analysis.

Conclusion: In patients with sepsis and high blood pressure, the nomogram that was created offers a dependable and strong evaluation for predicting S-AKI. This evaluation provides valuable insights to enhance individualized treatment, ultimately resulting in improved clinical outcomes.

Keywords: Acute kidney injury; Early warning; Hypertension; Nomogram; Sepsis.

Fig. 1

The flowchart of patient selection.

Fig. 2

A nomogram for assessing the likelihood of S-AKI in septic patients who have high blood pressure. To utilize the nomogram, our initial step involves sketching a vertical line in an upward direction towards the “Points” axis. Add up the scores for each predictor and find the ultimate value on the 'Total Points' axis. In conclusion, simply draw a vertical line to the axis labeled “Probability of S-AKI” to ascertain the likelihood of AKI. Sepsis-induced acute kidney injury (S-AKI) is characterized by elevated serum procalcitonin (PCT) levels, prolonged thrombin time (TT), decreased prothrombin time activity (PTA), and increased creatinine (Cr) levels.

Fig. 3

The receiver operating characteristic curve for the nomogram predicting S-AKI in sepsis patients with hypertension. AUC refers to the area under the curve of receiver operating characteristics. The nomogram achieved an AUC of 0.823 in the training set (a) and 0.829 in the validation set (b) for predicting sepsis-induced acute kidney injury (S-AKI) in patients with hypertension.

Fig. 4

Calibration curves for the predicted nomogram were obtained in both the training set (a) and validation set (b). The nomogram calculates the predicted probability on the x-axis, while the observed actual probability of S-AKI is represented on the y-axis. The clinodiagonal is the perfect prediction of an ideal model. Consequently, the initial cohort is depicted by the solid curve while the bias-corrected curve, obtained through bootstrapping with 1000 repetitions (B = 1000), demonstrates the performance of the predicted model.

Fig. 5

DCA was performed on the nomogram in the training set (a) and the validation set (b). No patients develop S-AKI as indicated by the horizontal line, and patients develop S-AKI as indicated by the gray oblique line. The risk nomogram for S-AKI is depicted by the solid red line. Across a range of threshold probabilities, the nomogram in DCA demonstrates a higher advantage compared to both complete treatment and no treatment.

Fig. 6

The receiver operating characteristic curve for the nomogram predicting S-AKI in sepsis patients with different hypertension grades. AUC refers to the area under the curve of receiver operating characteristics. In sepsis patients with hypertension, the nomogram achieved an AUC of 0.901 in the grade I hypertension set (a), 0.818 in the grade II hypertension set (b), and 0.810 in the grade III hypertension set (c) for predicting S-AKI.