Identifying key covariates of clinical outcomes for critically ill patients with Parkinson's disease: analysis of the MIMIC-IV database

Abstract

Background: Parkinson's disease (PD) is a common chronic degenerative disease, and its exact pathological mechanism remains unclear. In this study, we identified covariates associated with the clinical outcomes of PD using patient data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, providing new references for the treatment of patients.

Methods: The data of patients with PD and relevant covariates were obtained from the MIMIC-IV database. The patients were categorized into the in-hospital death and in-hospital survival groups based on their survival status, and the relationship between their outcomes and covariates was investigated. Key covariates markedly associated with the clinical outcomes of PD were further screened using regression analysis. Finally, a nomogram for predicting the risk of in-hospital mortality in patients with PD was constructed and validated.

Results: A total of 143 patients with PD and 37 covariates were included in this study. Of the included patients, 25 were assigned to the in-hospital death group and 118 were assigned to the in-hospital survival group. Covariates such as the respiratory rate, mean arterial pressure (MAP), respiratory failure, anion gap, bicarbonate levels, blood urea nitrogen levels, and sequential organ failure assessment (SOFA) scores were markedly associated with in-hospital mortality in patients with PD. Subsequently, age [hazard ratio (HR) = 1.0565, 95% confidence interval (CI) = 1.0065-1.1090, p < 0.05], bicarbonate levels (HR = 0.8988, 95% CI = 0.8310-0.9722, p < 0.05), BUN levels (HR = 1.0292, 95% CI = 1.0084-1.0503, p < 0.05), and SOFA scores (HR = 1.1510, 95% CI = 1.0324-1.2831, p < 0.05) were identified as key covariates associated with in-hospital mortality. The nomogram incorporating these covariates exhibited favorable performance in predicting the risk of in-hospital mortality in patients with PD.

Conclusion: This study revealed four key covariates associated with the clinical outcomes of PD, providing new references for the treatment of patients.

Keywords: MIMIC IV database; Parkinson’s disease; covariate; in-hospital mortality; nomogram; prognostic factors.

Figure 1

Patient scheduling flowchart.

Figure 2

Kaplan–Meier curves of 143 patients.

Figure 3

(A) Simple linear regression analysis. (B) Least absolute shrinkage and selection operator regression analysis (LASSO) regression analysis. The horizontal axis represents the logarithm of lambda, whereas the vertical axis represents the model error. (C) Multivariate regression analysis.

Figure 4

(A) Nomogram. The horizontal axis represents the logarithm of lambda, whereas the vertical axis represents the model error. The nomogram is used to predict the risk of in-hospital mortality in patients with PD admitted to the ICU. The bumps in the figure represent subtle variations in the net benefit of model predictions across different threshold probabilities. (B) Calibration curve. The horizontal axis represents the predicted event rate, whereas the vertical axis represents the actual event rate. The straight line passing through the origin of the coordinate axes with a slope of 1 serves as the reference line indicating that the predicted probability of the nomogram is exactly the same as the true probability. The closer the predicted value is to the reference line, the more reliable the result is with both ranging from 0 to 1. The ideal prediction corresponds to the black dashed line. (C) ROC curve. (D) Decision curve analysis. The horizontal axis represents the risk threshold (Pt), whereas the vertical axis represents the net benefit (NB) after subtracting the harm from the benefit. The line segments in the graph represent the net benefit at each risk threshold. The black line parallel to the horizontal axis indicates that all samples are negative, meaning no intervention is applied to any patient. Therefore, the net benefit is 0. The sloping gray line represents the net benefit when all samples are positive, meaning everyone receives the intervention.