Retrospective cohort study based on the MIMIC-IV database: analysis of factors influencing all-cause mortality at 30 days, 90 days, 1 year, and 3 years in patients with different types of stroke

Abstract

Objective: This study aims to evaluate key factors influencing the short-term and long-term prognosis of stroke patients, with a particular focus on variables such as body weight, hemoglobin, electrolytes, kidney function, organ function scores, and comorbidities. Stroke poses a significant global health burden, and understanding its prognostic factors is crucial for clinical management. Methods: This is a retrospective cohort study based on data from the MIMIC-IV database, including stroke patients from 2010 to 2020. A total of 5,110 patients aged 18 and older were included in the study. The exposure variables included body weight and hemoglobin levels, while the outcome variables were the 30-day, 90-day, 1-year, and 3-year mortality risks. Covariates included electrolyte levels, kidney function, organ function scores, and comorbidities. Random forest and gradient boosting tree models were employed for data analysis to assess mortality risk. Results: Kaplan–Meier survival analysis showed that ischemic stroke patients had the highest 30-day mortality rate at 8.5%, with only 20% 1-year survival. Traumatic subarachnoid hemorrhage patients had the best prognosis, with a 1-year survival rate of 60%. Multivariable Cox regression analysis revealed that each 1-point increase in the Charlson Comorbidity Index raised the 1-year and 3-year mortality risks by 1.39 times (95% CI: 1.10–1.56) and 1.44 times, respectively. Each 1-point increase in the SOFA score increased the 30-day, 90-day, 1-year, and 3-year mortality risks by 2.11 times, 2.03 times, and 1.84 times, respectively. Additionally, lower hemoglobin levels were significantly associated with increased mortality, with 30-day, 90-day, and 1-year mortality risks increasing by 3.33 times, 3.34 times, and 4.16 times, respectively (p < 0.005). Age ≥ 71 years, longer hospital stays, and organ dysfunction were also significant factors affecting mortality. Conclusion: This study highlights the critical role of stroke type, comorbidity index, SOFA score, hemoglobin levels, and length of hospital stay in stroke prognosis. These findings provide valuable insights for clinical risk assessment and the development of individualized treatment strategies, which may improve the management and outcomes of stroke patients. The predictive model constructed effectively assesses mortality risks in stroke patients, offering support for future clinical practice.