# Early metoprolol use in ICU patients with congestive heart failure is associated with increased 30-day mortality: a causal machine learning study

## TL;DR

Using metoprolol early in ICU patients with heart failure may increase 30-day mortality, especially in high-risk patients.

## Contribution

Applies causal machine learning to assess metoprolol's impact on ICU patients with heart failure, revealing mortality risks and treatment heterogeneity.

## Key findings

- Early metoprolol use increased 30-day ICU mortality by 2%.
- Adverse effects were most pronounced in patients with high APSIII, age >78, elevated SOFA, WBC, and BUN.

## Abstract

The risk-benefit balance of early metoprolol administration in critically ill patients with congestive heart failure (CHF) admitted to the intensive care unit (ICU) remains controversial. However, high-quality causality clinical evidence is still lacking. This study utilized real-world data and causal forest algorithm to investigate the average treatment effect (ATE) of early metoprolol use on 30-day mortality in ICU patients with CHF. We also explored the heterogeneity of individual treatment effects (ITE) to identify clinical characteristics across different treatment-effect populations.

Study population and their clinical characteristics were obtained from the MIMIC-IV database (version 3.1). Propensity score matching (PSM) was performed to balance baseline confounders. Causal forest models estimated the ATE and ITE. Variable importance, conditional average treatment effect (CATE), and ITE-based quartile analysis were performed to identify heterogeneity and clinical characteristics of patients.

A total of 5,758 patients were included in the initial cohort, with 3,934 patients entering the final analysis after PSM. The causal forest model indicated that early metoprolol use increased the risk of 30-day ICU mortality by 2% compared with no use group (ATE = 0.02, 95% CI: 0.0004–0.0396, P = 0.045). APSIII, age, and SOFA were the top three variables contributing to the ATE. The causal forest model estimated the conditional average treatment effects for every variable. Individual treatment effects exhibited a right-skewed distribution. The highest ITE percentile was associated with APSIII >64, age >78 years, SOFA >8, WBC >13.95 × 109/L, and BUN >43.06 mg/dL.

Early metoprolol use within 24 h of ICU admission in CHF patients is associated with increased 30-day mortality. Therapeutic effects were heterogeneous, with the adverse effects being particularly pronounced in patients with high APSIII, advanced age, elevated SOFA, high white blood cell counts, and elevated blood urea nitrogen levels. These findings suggest careful consideration should be given to the early use of metoprolol in high-risk patients.