Estimating the causal effect of dexamethasone versus hydrocortisone on the neutrophil-lymphocyte ratio in critically ill COVID-19 patients from Tygerberg Hospital ICU using targeted maximum likelihood estimation method

Abstract

Abstract Background Causal inference from observational studies is a study topic that has advanced fast over the years, as have methods for causal effect estimation. Among them, Targeted Maximum Likelihood estimation (TMLE) possesses the most outstanding statistical properties, and with no outright treatment for COVID-19, there was an opportunity to estimate the causal effect of dexamethasone versus hydrocortisone upon the neutrophil-lymphocyte ratio (NLR), a vital indicator for disease progression among critically ill COVID-19 patients. This study is designed to show the application of TMLE variations to estimate the causal effect of dexamethasone versus hydrocortisone on the neutrophil-lymphocyte ratio in critically ill COVID-19 patients. Methods We retrospectively analysed data from the first and second COVID-19 waves, including critically ill COVID-19 patients. TMLE variations were used in the analysis and Super Learner (SL), Bayesian Additive Regression Trees (BART) and parametric regression (PAR) were implemented to estimate the average treatment effect (ATE) and its 95%CI Statistical analysis was carried out with ltmle package in R-software. Result presented in graph and tables. Results The study had 168 participants, 128 on dexamethasone and 40 on hydrocortisone. The mean causal difference in NLR on day 5; ATE [95% CI]: from SL-TMLE was − 2.28[-6.811, 2.246], BART-TMLE − 2.10[-6.464, 2.262] and PAR-TMLE − 2.16[-5.710, 1.397]. The ATE of dexamethasone versus hydrocortisone on NLR was not statistically significant since the confidence interval included zero. Conclusion In critically ill COVID-19 patients admitted to ICU, the effect of dexamethasone on NLR was not significantly different from that of hydrocortisone. This means that the variation in NLR impact between the two medicines may be attributable to random chance. However, TMLE remains an excellent tool for causal analysis of observational research, with the ability to be supplemented by numerous prediction approaches.