Association Between the Hemoglobin-to-Red Blood Cell Distribution Width Ratio and Perioperative Mortality Risk in Myasthenia Gravis Patients with Sepsis: A Retrospective Cohort Study.

BackgroundThis study aims to explore the association between the Hemoglobin/Red Cell Distribution Width Ratio (HRR) and perioperative mortality in myasthenia gravis (MG) patients with sepsis based on the electronic ICU collaborative research database (eICU-CRD).MethodsWe retrospectively analyzed data of 151 patients from the eICU-CRD between 2014 and 2015. We performed multivariate logistic regression, stratified analysis with interaction terms, and restricted cubic spline regression to assess the linear relationship. Additionally, we applied the extreme gradient boosting (XGBoost) machine learning algorithm to evaluate the importance of selected variables.ResultsHRR, as a new marker of inflammation, were significantly lower in patients with myasthenia gravis complicated by sepsis who experienced perioperative mortality compared with the non-sepsis group [6.42 (5.47, 7.45) versus 8.16 (6.36, 9.29), P < .001].After further adjusting for gender, ethnicity, hypertension, heart rate, body temperature, blood urea nitrogen, and creatinine, the odds ratio (OR) for perioperative mortality in MG patients with sepsis was 0.578 (95% CI: 0.402-0.799, P = .0016). HRR is a relevant indicator of the mortality risk for MG patients with sepsis during the perioperative period. Quartile analysis of HRR revealed a significant increasing trend (Q4 vs Q1, OR: 0.039, 95% CI: 0.001-0.346, P = .017). The XGBoost machine learning model was employed to assess factor importance, revealing HRR as one of the top five variables influencing perioperative mortality in MG patients with sepsis. Subgroup analysis identified correlations between HRR and age, gender, ethnicity, and hypertension status. Additionally, HRR levels were associated with increased diastolic blood pressure (DBP), systolic blood pressure (SBP), and heart rate.ConclusionsA linear relationship was observed between baseline HRR levels and perioperative mortality in MG patients with sepsis. Furthermore, lower HRR levels were associated with higher perioperative mortality, indicating that HRR may serve as a potential clinical indicator for the risk of perioperative mortality in these patients.