The Effects and Safety of Vasopressin Receptor Agonists in Patients With Septic Shock

A Meta-analysis and Trial Sequential Analysis

Libing Jiang; Yi Sheng; Xia Feng; Jing Wu

Disclosures

Crit Care. 2019;23(91) 

In This Article

Results

Literature Selection Process

After excluding 230 duplicates, 1887 studies from 2117 hits were chosen for further evaluation. Through reading title and abstract, 1855 studies were excluded and 32 studies were potentially eligible for evaluation by reading the full text. Finally, 20 studies were included for meta-analysis.[12–31] Figure 1 shows the process of literature selection and reasons for study exclusion. Detailed information of excluded studies and ongoing studies is presented in Additional file 1: Table S2, S3.

Figure 1.

Flow chart of literature selection

The Characteristics and Quality of the Included Studies

Twenty studies[12–31] with 2250 septic shock patients who received vasopressin receptor agonists and 2281 septic shock patients who received catecholamine alone were eligible. Patients in 9 studies and 11 studies received vasopressin[12,17,19,20,21,28–31] and vasopressin's analogues (pituitrin 1,[14] selepressin 1,[23] terlipressin 9[13,15,16,18,22,24–27]), respectively. Among them, four studies[12,17,18,21] were published in abstract and relevant data were obtained from the study by McIntyre et al..[32] Detailed information is showed in Table 1, and quality evaluation of all included studies is showed in Additional file 1: Figures S1, S2.

Meta-analysis

The primary endpoint: mortality. Twenty studies were included for mortality analysis,[12–31] and the combined RR was 0.92 (95% confidence interval (CI) 0.84 to 0.99, P = 0.03, I 2 = 0%) (Figure 2). The quality of evidence is presented in Additional file 1: Table S4. The results of TSA indicated the optimal information size was 4103 patients for mortality and more high-quality RCTs are needed, although z curve had crossed the general boundary line, but it did not cross any adjusted boundary line favoring the intervention group or control group. And the adjusted RR was 0.92 (95% CI 0.84 to 0.99, P = 0.03, I 2 = 0%), based on 10% RRR (from a baseline event rate of 43%) (Figure 3).

Figure 2.

Forest plot for effects of vasopressin or its analogues on 28/30-day mortality (mortality rate within 30 was equal)

Figure 3.

Trial sequential analysis for effects of vasopressin or its analogues on 28/30-day mortality. The diversity-adjusted required information size (4103 participants) was based on a relative risk reduction of 10%, an alpha of 5%, a beta of 20%, and an event proportion of 43% in the control arm. The blue cumulative z curve was constructed using a fixed effects model

Post hoc sensitive and subgroup analysis. Firstly, the combined RR was 0.95 (95% CI 0.86–1.05, P = 0.30, I2 = 0%)[13–16,19,20,22–31] for studies published in full text and 0.85 (95% CI 0.74–0.98, P = 0.02, I2 = 23%)[12,17,18,21] for studies published in abstract. In addition, after removing three studies that did not report 28/30-day mortality, the combined RR was 0.92 (95% CI 0.83–1.00, P = 0.06, I 2 = 0%).[12–15,17–19,21–29,31] Thirdly, the combined RR was 0.94 (95% CI 0.84–1.05, P = 0.70, I 2 = 0%)[12,17,19–21,28–31] for patients who received vasopressin and 0.88 (95% CI 0.78–0.99, P = 0.04, I 2 = 0%)[13–16,18,22–27] for patients who received its analogues (Figure 2). Finally, we performed another subgroup analysis based on different diagnoses. In patients with cirrhosis, the combined RR was 0.73 (95% CI 0.61–0.88, P = 0.001, I 2 = 23%), and in other patients, the combined RR was 0.95 (95% CI 0.87–1.04, P = 0.24, I 2 = 0%).

The Secondary Endpoints

ICU length of stay. Ten studies reported ICU length of stay.[12,14–16,19,22,24,27,28,31] The results showed there were no effects of vasopressin receptor agonists on ICU length of stay (MD − 0.08, 95% CI − 0.68–0.52, P = 0.79, I 2 = 0%) (Figure 4).

Figure 4.

Forest plot for effects of vasopressin or its analogues on intensive care unit length of stay

Duration of mechanical ventilation. Five studies were eligible for analysis of duration of mechanical ventilation.[14,15,19,24,27] The combined MD was − 0.58 (95% CI − 1.47–0.31, P = 0.20, I2 = 57%) (Figure 5). The results showed vasopressin receptor agonist administration did not significantly affect the duration of mechanical ventilation. The result of the study by Han was different from the other studies.[14] A sensitive analysis was performed by removing the study by Han; the combined MD was − 1.05 (95% CI − 1.77 to − 0.32, P = 0.005, I 2 = 0%).

Figure 5.

Forest plot for effects of vasopressin or its analogues on the duration of mechanical ventilation

Adverse Events

Total adverse events. Eleven studies were included in the analysis of total adverse events.[16,19–23,27–31] The combined RR was 1.28 (95% CI 0.87–1.90, P = 0.21, I2 = 57%) (Figure 6). The results were driven by the study by Liu et al.,[27] which carried 18.3% of the weight. A sensitive analysis was performed by removing the study by Liu et al., and the combined RR was 1.11 (95% CI 0.86–1.43, P = 0.44, I 2 = 10%).

Figure 6.

Forest plot for effects of vasopressin or its analogues on total adverse events

Digital ischemia. Eight studies and 1964 patients were eligible for the analysis.[19,22,23,25,27–29,31] The combined RR was 4.85 (95% CI 2.81–8.39, P < 0.001, I2 = 26%) (Figure 7), indicating that the use of vasopressin receptor agonists was associated with more digital ischemia events. A sensitive analysis was performed by removing the study by Liu et al.,[27] due to its results that were significantly different from the other studies. And the combined RR was 2.79 (95% CI 1.54–5.05, P < 0.001, I 2 = 0%), supporting the original conclusion.

Figure 7.

Forest plot for effects of vasopressin or its analogues on digital ischemia

Other adverse events. There were no effects of vasopressin receptor agonists on cardiovascular events,[19,23,27,28,30,31] arrhythmia,[16,19,21–23,27,28,31] mesenteric ischemia,[19,23,27,31] diarrhea,[23,27,31] cerebrovascular events,[23,31] and hyponatremia.[27,28,31] And the combined RR was 0.91 (95% CI 0.53–1.57, P = 0.73, I 2 = 1%) (Additional file 1: Figure S3), 0.77 (95% CI 0.48–1.23, P = 0.28, I 2 = 23%) (Additional file 1: Figure S4), 0.83 (95% CI 0.44–1.55, P = 0.55, I 2 = 0%) (Additional file 1: Figure S5), 2.47 (95% CI 0.77–7.96, P = 0.13, I 2 = 49%) (Additional file 1: Figure S6), 1.36 (95% CI 0.18–10.54, P = 0.77, I 2 = 0%) (Additional file 1: Figure S7), and 1.47 (95% CI 0.84–2.55, P = 0.18, I 2 = 0%) (Additional file 1: Figure S8), respectively. Additional subgroup analyses are showed in Table 2.

Publication bias. Publication bias was assessed via funnel plots and Egger's test (Additional file 1: Figure S9). The results of Egger's test indicated there was no significant publication bias among the included studies (P = 0.39).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....