The Application of Principal Component Regression in Modeling the Factors Associated with Mortality from COVID-19 during the Seventh Peak of the Pandemic

The seventh peak represented the latest surge of the coronavirus disease 2019 (COVID-19) pandemic in Iran, predominantly characterized by Omicron subvariants. Due to the complex interplay of various factors contributing to COVID-19 mortality, employing an advanced statistical technique such as principal component regression (PCR) allows for the categorization and evaluation of these variables while improving predictive accuracy and mitigating issues such as multicollinearity often encountered with traditional regression methods.

In this cross-sectional study, data from 8994 patients were extracted from the Medical Care Monitoring Center (MCMC) of hospitals affiliated with Mashhad University of Medical Sciences, Mashhad, Iran, covering the period from July to September 2022. Principal component logistic regression was employed to identify significant factors associated with patient mortality. Data analysis was done using SPSS software at a significance level of 0.05.

The mean age of participants was 50.87 ± 28.30 years. Statistically significant associations were found between several variables including drug use, COVID-19 test results, high fever, respiratory distress, decreased level of consciousness, gastrointestinal symptoms, intubation status, oxygen saturation (PO2) levels, chronic blood diseases, and histories of hypertension (HTN), cancer, and diabetes with patient mortality (P < 0.05). In the regression model, the components of respiratory factors and underlying factors increased the chance of death by 62% and 15%, respectively, with confidence intervals (CIs) of 1.41-1.86 and 1.01-1.30, respectively. Besides, the components of intubation and temperature increased the chance of death by 2.47 times with a CI of 2.10-2.89 (P < 0.05).

Identifying risk factors is essential for healthcare providers to recognize vulnerable patient subpopulations, enhance the quality of care, prioritize treatment interventions, and effectively allocate resources.