Sepsis resulting from systemic inflammation and exaggerated host immune responses to infection is a leading cause of death in hospitalized patients, with 30–40% mortality (Levy et al., 2012; Evans et al., 2021). Appropriate management in the initial stages of sepsis development improves outcomes (Evans et al., 2021). The current therapeutics for sepsis mainly focus on early empiric antibiotics, fluid resuscitation, and the use of vasoactive drugs, as well as temporary organ support such as mechanical ventilation and renal replacement therapy. Given the lack of high-quality evidence, many treatments based on dysregulation of the host immune response are not routinely applied in clinical practice (Donovan et al., 2021).

Sustained sympathetic stimulation is an integral part of the physiological response to sepsis, resulting in the downregulation of myocardial beta-adrenergic receptors and tachycardia (Hollenberg and Singer, 2021). Even after adequate fluid resuscitation, patients with sepsis who remain tachycardic carry a poor prognosis (Leibovici et al., 2007; Domizi et al., 2020). Therefore, adequate heart rate control may help improve outcomes in managing sepsis. Treatment with β-blockers for sepsis to relieve the sympathetic stress response has been previously investigated (Liu et al., 2018). β-blockers were regarded as a relative contraindication for septic shock because of the cardiac suppression. The effect of β-blocker treatment on the clinical outcome of sepsis and septic shock is still unclear. Previous clinical studies have mainly focused on the roles of short-acting β-blockers (esmolol or landiolol) in septic patients with tachycardia because these agents are recognized as relatively safer than other longer-acting β-blockers (Kakihana et al., 2020; Hasegawa et al., 2021). However, almost all those studies were limited by their small sample sizes (less than 155), and significant heterogeneity in the baseline characteristics of patients, leading to inconsistency in the findings. In addition, previous reviews have mainly focused on short-acting β-blockers leaving out studies on long-acting β-blockers. Propensity score matching (PSM) can largely reduce baseline differences between groups in observational studies and is referred to as retrospective randomization (Jupiter, 2017). Therefore, in the current study, we aimed to investigate whether early use of long-acting or short-acting β-blockers is associated with better clinical outcomes in patients with sepsis using the method of PSM in a huge sample.

The results of our study suggest that age, tachycardia, AKI, cancer, SOFA score, lactate, and gram-positive bacteria were independently associated with greater mortality in sepsis. Furthermore, long-acting β-blocker therapy may improve short- and long-term survival for patients with sepsis and septic shock. However, short-acting β-blockers (esmolol) treatment did not reduce the mortality in sepsis.

In patients with sepsis and septic shock, the adrenergic system is the initial adaptive response to maintain cellular and tissue homeostasis. However, catecholamine release can cause impaired autonomic control of the heart, such as persistent tachycardia, which worsens the prognosis of patients with sepsis (Domizi et al., 2020; Carrara et al., 2021). Our study found that tachycardia was an independent risk factor for the 28-day and 90-day mortality in the univariate and multivariable Cox regression models. This is consistent with previously reported studies (Leibovici et al., 2007; Guz et al., 2021).

β-blockers, specifically the selective β1 blockers, have been shown to have a favorable effect on heart failure because they prevent catecholamine release and sympathetic activity, diminish heart rate and improve diastolic function (Kubon et al., 2011). Animal models of sepsis treated with an esmolol infusion have shown improved outcomes, including cardiac function, heart rate, blood pressure and inflammatory cytokines (Suzuki et al., 2005). In 1969, Berk et al. (Berk et al., 1969), in the first study on the β-adrenergic effects in sepsis, found that dogs that received endotoxin followed by propranolol had lower mortality and less hyperemia in the lungs and other internal organs than those that received endotoxin alone.

The first randomized controlled study to report on the effects of esmolol infusion in patients with septic shock was conducted in 2013 by Morelli et al. (Morelli et al., 2013). The study randomly included 77 patients to receive esmolol infusion and 77 to receive standard treatment. The results showed that esmolol infusion in patients with septic shock was associated with a reduced heart rate to target levels without increased adverse events compared with patients who received the standard care. But the 28-day mortality was not reduced in the esmolol group. Hasegawa et al.¹¹ conducted a meta-analysis which included seven clinical trials with a pooled sample size of 613 patients on the effect of ultrashort-acting β-blockers on mortality in patients of sepsis with tachycardia after initial resuscitation. Two studies (Xinqiang et al., 2015; Morelli et al., 2016) showed that esmolol infusion improved 28-day mortality and cardiovascular efficiency in septic shock. Four studies (Wang et al., 2015; Wang et al., 2017; Liu et al., 2019; Kakihana et al., 2020) did not reveal statistically significant differences in the 28-day mortality between patients with sepsis or septic shock receiving short-acting β-blockers and controls. Results of this meta-analysis demonstrated that the use of ultrashort-acting β-blockers (esmolol and landiolol) was associated with improved 28-day mortality in patients with sepsis with persistent tachycardia after initial resuscitation. However, the small sample size and variation in baseline characteristics in these studies might have caused a bias in the results. To minimize such a selection bias, we used the PSM method and confirmed that baseline covariates were balanced between the two groups. In addition, the multivariate Cox regression method was also used to reduce the possible confounder bias in our study. We finally included 264 patients who received esmolol infusion within 48 hours of admission and 264 patients who did not receive esmolol infusion. Our results illustrated that esmolol infusion did not significantly change the 28-day and 90-day mortality after PSM, which conflicts with the meta-analysis findings. Baseline patient characteristics and sample size may partially explain this discrepancy.

The role of long-acting β-blockers on the mortality of patients with sepsis has not been fully understood. A systematic review (Tan et al., 2019) conducted in 2019 showed that β-blockers exposure before sepsis was associated with reduced mortality. However, the review did not provide enough information to perform a meta-analysis. A recent single-center observational study with 320 cases included after PSM by Guz et al. (Guz et al., 2021) found that long-acting β-blockers were associated with decreased mortality in patients with sepsis and relative tachycardia. Although the results were promising, the study still had some limitations. First, there was still an obvious imbalance remained in baseline characteristics between groups after PSM, and essential confounders, including age, heart rate at admission, and the use of angiotensin receptor blockers (ARBs). Second, the sample size was too small to generalize the findings. To address these issues, we included 3415 cases (β-blocker group) after PSM and perfectly balanced the baseline characteristics ( Figure S2 ). Our findings agreed with those of previous research that showed long-acting β-blocker therapy was associated with significantly lower mortality rates in patients with sepsis.

In addition, we performed multiple subgroup analyses after PSM and found that β-blocker therapy also benefited 28-day mortality in patients with septic shock. P for interaction in subgroup analyses increased the robustness of the findings ( Figure 3 ). Therefore, septic shock is not an absolute contraindication for using β-blockers.

Our study has many strengths and limitations that merit consideration. The sample size of 6830 cases in this study is the largest in the literature on the association of β-blocker therapy and sepsis. Moreover, we eliminated the potential confounding effect of baseline factors using PSM to balance the baseline characteristics among the two groups. In addition, our studies are the first to suggest that short- and long-acting β-blockers have different therapeutic profiles on sepsis.

However, our study has several limitations. Firstly, this study is based on a clinical database containing missing variables. To reduce potential bias, multiple imputations were used to minimize the risk of deviation resulting from missing values. Secondly, we did not report the doses of β-blockers in this study. The impact of the dose of β-blockers on mortality is not clear. Thirdly, due to the inherent limitation of the retrospective research, a cause-effect relationship between the β-blocker therapy and mortality cannot be established. In addition, patients who were discharged or died within 48 hours of ICU admission were excluded. This exclusion criteria exerts a selection bias on the study population. Among the 12884 sepsis cases, we observed a small proportion of deaths (100/12884) within the first 48 hours of admission, and only 11% of the deaths used beta blockers. Finally, our study did not report the adverse events of β-blocker treatment. Since extracting adverse events from the MIMIC database is difficult, further research in prospective trials is needed.

In conclusion, the results of our study suggest that β-blockers were associated with improved 28- and 90-day mortality in patients with sepsis and septic shock. Long-acting β-blocker therapy may have a protective role in patients with sepsis, reducing the 28-day and 90-day mortality. However, short-acting β-blocker (esmolol) treatment did not reduce the mortality in sepsis. The results need to be validated in future randomized controlled trials.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

The study was an analysis of a third-party anonymized publicly available database with pre-existing institutional review board (IRB) approval. The Institutional review boards at the Beth Israel Deaconess Medical Center (protocol 2001-P-001699/14) and Massachusetts Institute of Technology (protocol 0403000206) have approved the data collection and the use of MIMIC-III for research purposes and granted waiver of informed consent. All methods were carried out in accordance with relevant guidelines and regulations.

C-LG performed the data analysis and wrote the manuscript. Q-YP designed the research. YZ, Z-WY and WC collected the data. L-NZ and Y-HA reviewed the manuscript. All the authors reviewed and approved the manuscript. All authors contributed to the article and approved the submitted version.