Figure 1 shows a flow diagram for selection process. A total of 549 records were
initially identified from databases search. 212 records were excluded for duplicates, and 308 publications were excluded after screening the titles and abstracts. The remaining 29 full-text articles were assessed for eligibility, of 22 studies were further excluded. The remaining 7 RCTs (13-18, 22) were included in the final meta-analysis.
Characteristics of included studies
The characteristics of studies included in our meta-analysis are summarized in Table 1. The seven studies were published between 2013 and 2019, and sample sizes range from 41 to 154. Orbegozo-Cortes et al (22) and Morelli et al (18) reported the same clinical trial, but with different follow-up times. Six included RCTs (13, 15-18, 22) involve septic shock, wang et al (14) reported the effect of esmolol and milrinone on severe sepsis patients who randomly divided into control, milrinone, and milrinone-esmolol groups. Wang et al 2017(16) used isotonic saline in control group, while the remained studies (13, 15, 17, 18) used blank control. Overall, 232 patients included in esmolol group while 231 patients in control group. All studies focused on adults at the mean age of 34-67.2 years and 38.0-69 years respectively in intervention and control arms. Three studies (13, 15, 16) commenced at 0.05mg/kg/h esmolol continuous intravenous titrate, while four trials (14, 15, 17, 18) commenced at 25mg/h esmolol continuous intravenous infusion, and adjust the dosage according to heart rate until reach the predefined threshold rate.
Quality assessment
Risk-of-bias assessment of the included studies is presented in Figure 2. The included trials had some methodological strengths and limitations. All included trials were adjudicated to be low risk of bias in random sequence generation, blinding of outcome assessment, incomplete outcome data and other bias. The risk of bias regard to allocation concealment was low risk except Yang et al 2014(15) with high risk where their patients allocated by a random number table predisposing them to an elevated risk of bias. Six RCTs were high risk of bias in blinding of participants and personnel because all of them had no mechanisms in place for blinding except Liu et al 2019(17). Wang et al 2017(16) was high risk of bias in selective reporting because reporting bias was not addressed, while the other trials were low risk.
We were unable to assess the publication bias using a funnel plot due to the small number of studies (<10) included in this analysis. Therefore, publication bias cannot be excluded.
Heterogeneity and sensitivity analysis
No heterogeneity was observed in MAP, CVP, Lac (12 hours (h)), CI (72 h), WBC, and PO2/FiO2 (24, 48h), low heterogeneity in 28-day mortality, TnI (24, 48 h), SVI (24 h) and PO2/FiO2 (72, 96h). We found high heterogeneity in the length of ICU stay, HR, ScvO2, Lac (24, 48, 72, 96 h), TnI (72 h), CI (12, 24, 48 h), SVI (48, 72 h), IL-6 and TNF-α. A sensitivity analysis was performed to evaluate the stability of the results, by excluding each study one by one and recalculating the combined RR or SMD on the remaining studies. This analysis confirmed the stability of the results: the overall effects did not show statistically significant reversal, and recalculated pooled RR and SMD were consistent and without apparent fluctuation (data not shown).
Primary outcomes
28-day mortality Five trials (13, 14, 16-18) with 422 cases were included for meta-analysis, shown in Figure 3. Overall, there was significantly effective on esmolol decreased 28-day mortality compared with the control (RR = 0.68, 95% CI: 0.52-0.88, P = 0.004, I2 = 45%).
Secondary outcomes
Heart rate Five trials (13-17) evaluated the effect between esmolol and control group. Our pooled analyses included 309 adults and found esmolol can significantly decrease heart rate at 24, 48, and 72 hours (SMD = -1.83, 95% CI -2.95, -0.70, P = 0.001, I2 = 94%; SMD = -1.68, 95% CI -2.8, -0.56, P = 0.003, I2 = 94%; SMD = -1.91, 95% CI -3.23, -0.60, P = 0.004, I2 = 94%, respectively), shown in Figure 4.
Length of ICU stay Three trials (13, 17, 18) examined the length of ICU stay between esmolol and control, shown in Figure 5. The pooled analysis including 302 adults showed that there was no significant association with esmolol supplementation on septic shock treatment (SMD = -0.23, 95% CI: -0.96, 0.50, P = 0.54, I2 = 89%).
Mean arterial pressure Four trials (13-16) evaluated MAP between esmolol and control group, and 209 adults were included for meta-analysis. There was no significant difference at 12, 24, 48, and 72 hours (SMD = -0.21, 95% CI -0.52 to 0.10, P = 0.19; SMD = -0.26, 95% CI -0.53 to 0.02, P = 0.07; SMD = -0.02, 95% CI -0.29 to 0.25, P = 0.89; SMD = 0.07, 95% CI -0.25 to 0.39, P = 0.68, respectively), there were no heterogeneity detected, shown in Figure 6.
Lactic acid Six RCTs (13-18) including 463 patients showed that there was no significantly different between esmolol and control groups, and the pooled SMDs at 12, 24, 48, and 72 hours were 0.04 (95% CI: -0.27, 0.35; P = 0.79, I2 = 0%), 0.15 ( 95% CI: -0.48, 0.78; P = 0.64, I2 = 90%), -0.5 ( 95% CI: -1.01, 0.01; P = 0.06, I2 = 83%), -0.60( 95% CI: -1.24, 0.03; P = 0.06, I2 = 88%) and -0.40 ( 95% CI: -0.93, 0.12; P = 0.13, I2 = 80%) respectively, shown in Figure 7.
Stroke Volume Index Four studies (13-16) including 209 cases showed that there was no significantly different between esmolol and control groups, and the pooled SMDs at 24, 48 and 72 hours were 0.16 (95% CI: -0.12, 0.44; P = 0.27, I2 = 7%), 0.44 (95% CI: -0.27, 1.15; P = 0.23, I2 = 84%) and 0.43 (95% CI: -0.54, 1.41; P = 0.39, I2= 88%) respectively, shown in Figure 8.
Cardiac index Four trials (13-16) with 209 adults reported that esmolol can significantly decrease the CI at 72 hours (SMD = -0.4, 95% CI: -0.73, -0.07, P = 0.02, I2 = 0%) compared with control groups, but there were no difference at 12, 24 and 48 hours (SMD = -0.46, 95% CI: -1.52, 0.60, P = 0.39, I2 = 90%; SMD = -0.11, 95% CI: -0.74, 0.53, P = 0.74, I2 = 81%; SMD = -0.16, 95% CI: -0.76, 0.45, P = 0.61, I2 = 79%, respectively), shown in Figure 9.
Central venous pressure Three studies (13-15) with149 adults included for meta-analysis, showing no significant difference at 24, 48 and 72 hours between esmolol and control groups (SMD = 0.19, 95% CI: -0.13, 0.52, P = 0.24, I2 = 0%; SMD = -0.22, 95% CI: -0.55, 0.1, P = 0.17, I2 = 0; SMD = 0.03, 95% CI: -0.29, 0.36, P = 0.84, I2= 0%, respectively), shown in Figure 10.
Central venous oxygen saturation Two trials (13, 15) including 89 cases indicated that there was no significant difference at 24, 48 and 72 hours between esmolol and control groups (SMD = 0.86, 95% CI: -1.08, 2.79, P = 0.39, I2 = 94%; SMD = 1.43, 95% CI: -0.7, 3.56, P = 0.19, I2 = 95%; SMD = 1.87, 95% CI: -1.53, 5.26, P = 0.28, I2 = 97%, respectively), shown in figure 11.
Troponin I Two trials (14, 15) with 101 adults were included for meta-analysis, showing that esmolol can significantly decrease the level of TnI at 24, 48 and 72 hours (SMD = -0.59, 95% CI: -1.02, -0.16, P = 0.008, I2 = 13%; SMD = -0.97, 95% CI: -1.48, -0.45, P = 0.0002, I2 = 33%; SMD = -1.63, 95% CI: -2.54, -0.73, P = 0.0004, I2= 72%, respectively), shown in figure 12.
White Blood Cells Three studies (16-18) including 314 adults for meta-analysis, showing no significant difference between esmolol and control group (SMD = 0.86, 95% CI: -1.08, 2.79, P = 0.39, I2 = 94%), shown in Figure 13.
Interleukin 6 Two trials (14, 16) with 120 cases were included for meta-analysis. The pooled analysis showed no significant difference between esmolol and control group (SMD = -0.24, 95% CI: -1.03, 0.55, P = 0.54, I2 = 79%), shown in Figure 14.
Tumor necrosis factor-a Two studies (14, 16) including 120 patients showed that there was no significant difference between esmolol and control group (SMD = -0.42, 95% CI: -1.12, 0.27, P = 0.23, I2= 72%), shown in Figure 15.
PO2/FiO2 Two studies (14, 18) including 214 patients showed that there was no significant difference between esmolol and control group at 24, 48, 72 and 96 hours (SMD = 0.06, 95% CI: -0.21, 0.33, P = 0.66, I2= 0%; SMD = 0.06, 95% CI: -0.21, 0.32, P = 0.68, I2= 0%; SMD = 0.24, 95% CI: -0.15, 0.64, P = 0.22, I2= 46%; SMD = 0.24, 95% CI: -0.17, 0.66, P = 0.25, I2= 51%), shown in Figure 16.
Quality of evidence
We used the GRADE system to determine the quality of evidence in our meta-analysis. 28-day mortality and PaO2/FiO2 had “very low” quality with a serious risk of bias, inconsistency and indirectness. The length of ICU stay and ScvO2 had “very low” quality with risk of bias, inconsistency and imprecision. HR, MAP, CVP and TnI had “very low” quality with risk of bias, indirectness and imprecision. Lac, CI, SVI, TNF-a and IL-6 had “very low” quality with risk of bias, inconsistency, indirectness and imprecision. WBC had “low” quality with risk of bias and imprecision.