Literature selection process
A total of 2324 records were generated through the literature search, from which 1434 studies were selected for further evaluation after excluding 890 duplicates. Then 1371 studies were excluded by reading the titles and abstracts, and 63 studies were potentially eligible for evaluation by reading the full text. Finally, 24 studies were subjected to final meta-analysis. Figure 1 shows the process of literature selection and the reasons for study exclusion.
The characteristics and quality of the included studies
A total of 24 randomized clinical trials [15, 16, 19–40] were eligible for inclusion in this study, involving 2210 septic shock patients who received vasopressin or its analogs, while 1870 patients received catecholamine alone. The characteristics of the included studies are summarized in Table 1. It is worth noting that only three [23, 27, 37] of the included studies focused on children, with a total of 163 children included in these studies. Patients in 10 studies [19, 21, 24, 27, 30–32, 36, 37, 40] received vasopressin and 13 studies received vasopressin’s analogues (selepressin 2[35, 39], terlipressin 11[15, 16, 20, 22, 23, 25, 28, 29, 33, 34, 38]), respectively.
Table 1
Characteristics of included randomized clinical trials comparing vasopressin or its analogs vs catecholamines alone in patients with septic shock.
Study ID | Design | Setting | No. Of patients | Patients | Intervention Groups | Comparison Groups | Outcome |
Acevedo 2009 | Open-label | Single center | 24 | Cirrhotic patients with septic shock | Terlipressin 1–2 mg/4 h plus alpha-adrenergic drugs | Alpha-adrenergic drugs alone | ICU mortality |
Albanese 2005 | Open-label | Single center | 20 | Septic shock | Terlipressin: one bolus of 1 mg (0.03-0.04UI/min) and a second bolus of 1 mg was given if the MAP < 65 mmHg after 20 mins | NE was started at a dose of 0.3µg/kg/min, followed by 0.3µg/kg/min increments at 4-min intervals to raise MAP to 65 to 75 mmHg | Hospital mortality |
Barzegar 2014 | Open-label | Single center | 30 | Septic shock and 2 or more organ dysfunction | Vasopressin: 0.03u/min | NE: infusion adjusted to MAP ≥ 65 mmHg | 28d-mortality |
Choong 2009 | Double blinding | Multi-center | 65 | Pediatric vasodilatory shock | Vasopressin: infused at a starting dose of 0.0005 U/kg/min in addition to the open-label vasoactive infusions and titrated every 5 minutes up to 0.002 U/kg/min (0.05 U/min maximum total dose) to maintain a target MAP for age. | Open label vasoactive drugs | 28d-mortality |
Choudhury 2016 | Open-label | Single center | 84 | Liver cirrhosis with septic shock | Terlipressin:1.3to 5.2µg/min over 24h | NE:7.5-60ug/min | 28d-mortality |
Gondon 2016 | Double blinding | Multi-center | 421 | Septic shock | Vasopressin: up to 0.06 U/min with target MAP between 65–75 mm Hg, or at physician’s discretion | NE: up to 12µg/min with target MAP between 65–75 mm Hg, or at physician’s discretion 28d Lowa | 28d-mortality |
Hajjar 2019 | Double blinding | Single center | 250 | Cancer with septic shock | Vasopressin: 0.01 to 0.06 IU/min | NE: 10–60 µg/min. | 28d-mortality |
Hammond 2018 | Open-label | Single center | 82 | Septic shock | Vasopressin and NE: 0.05–0.5µg/kg/min of NE and 0.04 U/min of vasopressin to maintain MAP between 65–75 mm Hg | NE: 0.05–0.5µg/kg/min to maintain MAP between 65–75 mm Hg | Hospital mortality |
Hua 2013 | Open-label | Single center | 32 | Septic shock and ARDS | Terlipressin: 1.3 mg/kg/h | DA: 20 mg/kg/min | 28d-mortality |
James 2018 | - | Single center | 40 | Pediatric septic shock | Vasopressin:0.02 IU/kg/h | NE: 0.02 mcg/kg/min | 28d-mortality |
Laterre 2019 | Double blinding | Multi-center | 828 | Septic shock | Selepressin: 1.7、2.5 和 3.5 ng/kg/min | 5µg/min or more of norepinephrine | 90d-mortality |
Lauzier 2006 | Open-label | Multi-center | 23 | Septic shock | Vasopressin:0.04–0.20 U /min | NE: 0.1–2.8 µg/kg/min | ICU mortality |
Liu 2018 | Double blinding | Multi-center | 617 | Septic shock | Terlipressin: 20–160µg/h | NE: 4–30µg/min | 28d-mortality |
Malay 1999 | Double blinding | Single center | 10 | Vasodilatory septic shock | Vasopressin:0.04 U/min | Placebo | 24h-mortality |
Morelli 2008 | Open-label | Single center | 59 | Septic shock | Terlipressin: a single bolus dose of terlipressin 1 mg and a titrated infusion of dobutamine plus norepinephrine. | NE: infusion at incremental doses | ICU mortality |
Morelli 2009 | Open-label | Single center | 45 | Septic shock | Terlipressin 1.3µg/kg/h; Vasopressin 0.03 U/min | NE: 15µg/min | ICU mortality |
Oliveira 2014 | Double blinding | Single center | 387 | Septic shock | Vasopressin:0.01 to 0.03 U/minute with low doses of norepinephrine | NE: 0.05-2.0µg/kg/minute | 28d-mortality |
Prakash 2017 | Open-label | Single center | 184 | Cirrhotic patients with septic shock | Terlipressin and NE: 2 mg/24 h fixed dose infusion of terlipressin and 3.75-30µg/min of NE as needed to maintain MAP > 65 mm Hg | NE: 7.5–60 ug/min, | 30d-mortality |
Russell 2008 | Double blinding | Multi-center | 802 | Septic shock | Vasopressin: the blinded vasopressin infusion was started at 0.01 U/minute and titrated to a maximum of 0.03 U/min | NE: infusion was started at 5µg/min and titrated to a maximum of 15µg/min | Hospital mortality |
Russell 2017 | Double blinding | Multi-center | 53 | Septic shock | Selepressin 1.25, 2.5, and 3.75 ng/kg/min until shock resolution or a maximum of 7 days | Open label NE | 28d-mortality |
Sahoo 2022 | Open-label | Single center | 50 | Septic shock | Terlipressin 0.02 µg/kg/min (fixed dose) infusion and injection + norepinephrine (0.01–3) µg/kg/min infusion | NE: (0.01–3) µg/kg/min infusion | ICU mortality |
Svoboda 2012 | Open-label | Single center | 32 | Septic shock | Terlipressin 4mg/24h for 72 hours in addition to open label norepinephrine | NE: >0.6microg/kg/min for more than 24h | 28d-mortality |
Wang 2022 | Open-label | Single center | 22 | Septic shock | Terlipressin:1.3ug/kg/h | Usual care | 28d-mortality |
Yildizdas 2007 | Open-label | Single center | 58 | Pediatric septic shock | Terlipressin: administered unblinded as intravenous bolus doses of 20µg/kg every 6 h for a maximum of 96h | Usual care | ICU mortality |
The results of the methodological quality of the including studies using the Rob2 assessment tool are shown in Fig. 2. Specifically, ten studies [15, 16, 19, 23, 24, 27, 32, 34, 39, 40] were considered as having a low risk of bias. And ten studies [20, 21, 25, 26, 28, 29, 31, 34–36] were assessed as having high risk of bias: a high risk of both performance and detection bias was existed in three studies on account of the lack of blinding. There was a selection bias in Hammond’s study [36] because the allocation method was not based on the randomization principle. The study of Liu [38] may exit a high risk of other potential sources of bias because of the early termination of schedule. Anotherfour studies [ 22, 30, 33, 37] were assessed as unclear risk, because of “abstract only and no description”, “not blinded, but objective outcomes”, “no registered protocol, but standard outcome”, and “selection of the reported results category”.
Meta-analysis
The primary outcome: mortality
Mortalities at 4 different time-points (28/90-day, ICU, hospital mortality) data were presented in the Fig. 3. Additionally, there was only one study [19] that focused on 24-hour mortality. Comparison with catecholamine group, there was no advantage in reducing mortality for the intervention group, either for the combined results of individual time points or for the total results of all the 24 included studies (n = 4080; RR, 0.95; 95% CI 0.88–1.02).
The results of the TSA indicated that the optimal information size for mortality was determined to be 7109 patients. However, it was concluded that more high-quality RCTs are still required. It is worth noting that although the Z curve had crossed the general boundary line, it did not cross any adjusted boundary line favoring either the intervention group or the control group. The adjusted RR was calculated to be 0.95, based on a 10% relative risk reduction (RRR) from a baseline event rate of 30.0% [2, 4] (Fig. 4).
Post hoc sensitive and subgroup analysis
Firstly, the combined RR for the three studies of septic shock in children was 1.14 (95% CI 0.83–1.57, P = 0.28, I2 = 21%), and for the remaining literature in adults, the combined RR was 0.94 (95% CI 0.87–1.01, P = 0.10, I2 = 0%). Secondly, in subgroup analyses for patients with underlying disease, there was a slight improvement in mortality in the vasopressor group compared with the control group (RR 0.86; 95% CI 0.75–0.99, P = 0.04). However, those without underlying disease did not show a significant difference between the two groups. Subgroup analyses were also performed to assess the use of VP or its analogs as a first-line option, whether they were combined with catecholamine therapy, and whether they were added early to the intervening medication: no statistically significant differences were found in these subgroup analyses (Fig. 5).
Furthermore, another subgroup analysis was conducted based on the different types of intervening drugs (Table 2): the combined RR was 0.94 (95% CI 0.84–1.05, P = 0.26, I2 = 32%) for patients with septic shock using VP, 0.90 (95% CI 0.81–1.01, P = 0.09, I2 = 0%) for patients using telipressin. We further categorized the population using TP according to the presence or absence of underlying disease. Our analysis revealed that using terlipressin resulted in a reduced mortality rate among individuals with underlying disease (RR 0.74, 95% CI 0.62 ~ 0.89, P < 0.01, I2 = 0%) [25, 29, 33, 34]. However, there were no statistically significant findings in individuals without underlying disease (RR 1.00, 95% CI 0.87 ~ 1.16, P = 0.98, I2 = 0%) [15, 16, 20, 22, 23, 28, 38]. Only two publications [35, 39] used selepressin, and no further subgroup analysis was performed.
Table 2
Subgroup analysis based on medication type
| VP | | TP |
| RR/MD | 95%CI | I2 | | RR/MD | 95%CI | I2 |
Mortality | 0.94 | 0.84 ~ 1.05 | 32% | | 0.90 | 0.81 ~ 1.01 | 0% |
ICU LOS (days) | -0.91 | -2.43 ~ 0.62 | 79% | | -1.02 | -2.78 ~ 0.74 | 55% |
MV LOS (days) | -0.05 | -2.44 ~ 0.34 | 0% | | -0.62 | -1.22~-0.03 | 0% |
Total adverse events | 1.07 | 0.87 ~ 1.32 | 0% | | 1.39 | 0.73 ~ 2.66 | 68% |
Digital ischemia | 2.72 | 1.25 ~ 5.93 | 0% | | 3.03 | 1.04 ~ 8.89 | 61% |
Arrythmia | 0.88 | 0.63 ~ 1.22 | 10% | | 0.83 | 0.62 ~ 1.10 | 5% |
Renal failure/injury | 0.96 | 0.88 ~ 1.05 | 0% | | 1.00 | 0.87 ~ 1.16 | 45% |
Sensitivity analysis and assessment of publication bias
The sensitivity analysis revealed that none of the individual studies exerted a notable impact on the combined RR, suggesting the stability and reliability of the findings (primary outcome: mortality) (Additional file 1: Figure S1). The funnel plot, depicted in additional file 1: eAppendix Figure (a)), did not indicate any substantial publication bias for the primary outcome in the pooled studies. This finding was further supported by the statistical tests (Egger’s test, P = 0.842; Additional file 1: eAppendix Figure (b)).
The secondary outcome: Effective indicators
ICU length of stay
Twelve studies [23, 24, 26, 27, 29, 31–33, 36, 38–40] reported ICU length of stay (ICU LOS). The results showed there were no effects of vasopressin receptor agonists on ICU LOS (MD -0.82, 95% CI -1.80-0.16, P = 0.10, I2 = 71%) (Fig. 6a). A sensitivity analysis was conducted by excluding one study at a time, and this exclusion of studies one by one demonstrated that the results were not robust (Additional file 1: Figure S2). Subgroup analyses based on VP and TP classification did not demonstrate statistical significance (VP: MD -0.91, 95% CI -2.43-0.62, P = 0.24, I2 = 79%; TP: MD -1.02, 95% CI -2.78-0.74, P = 0.26, I2 = 55%) (Table 2).
Duration of mechanical ventilation
Six studies [23, 29, 31, 36, 38, 39] were included in the analysis of the duration of mechanical ventilation (MV LOS). The combined mean difference (MD) was − 0.60 (95% CI -1.10 to -0.11, P = 0.02, I2 = 0%) (Fig. 6a). These findings suggest that vasopressin receptor agonists have the potential to significantly decrease the duration of mechanical ventilation. Further subgroup analyses were conducted based on the type of intervening medication. The TP group showed a significant reduction in the MV LOS (MD -0.62, 95% CI -1.22–0.03, P = 0.04, I2 = 0%) [23, 29, 38], while the VP group was no longer statistically significant (MD -0.05, 95% CI -2.44-0.34, P = 0.14, I2 = 0%) [31, 36]. (Table 2)
SOFA score change
A meta-analysis was conducted using data from five studies [16, 21, 32, 38, 40] to analyze the change in SOFA score. The results showed that there was no statistically significant decrease in the vasopressin group (MD -0.73, 95% CI -1.73-0.26, P = 0.15, I2 = 72%) (Fig. 6a). The sensitivity analysis confirmed the robustness of the model (Additional file 1: Figure S3). Furthermore, when the data from Liu [38] (7-day SOFA score change) and Sohoo [16] (12-hour SOFA score change) were excluded, the combined results of the remaining studies also showed no statistically significant difference in the 24-hour SOFA score change (MD -0.17, 95% CI -1.38-1.04, P = 0.06, I2 = 64%).
Lactate change
Seven studies [15, 16, 21, 26, 27, 32, 40] reported the lactate change, and the combined MD was − 0.48 (95% CI -1.05-0.09, P = 0.10, I2 = 78%). Sensitivity analysis further supported this result (Additional file 1: Figure S4).
NE doses change
There were six studies [15, 16, 21, 26, 32, 38] reported the norepinephrine (NE) doses change. Norepinephrine levels were significantly reduced in the vasopressin group compared to the catecholamine group (MD -10.81, 95% CI -16.78–4.83, P < 0.01, I2 = 92%). This finding was further supported by a sensitivity analysis (Additional file 1: Figure S5). After removing the two papers [16, 38] that reported the amount of change over 7 days and 12 hours, similar results were still obtained.
Time to shock reversal
Compared to the catecholamine group, the vasopressin group demonstrated an increase in the time required for shock reversal (MD 13.06, 95% CI 4.18–21.94, P < 0.01, I2 = 46%). This finding was based on the analysis of two studies [27, 31].
The secondary outcome: Safety indicators
Total adverse events
The analysis included a total of twelve studies [15, 21, 23, 24, 26, 27, 31, 33, 35, 36, 38, 40] to examine the occurrence of total adverse events. The combined RR was 1.24 (95% CI 0.88–1.74, P = 0.22, I2 = 64%) (Fig. 6b). The sensitivity analysis confirmed the robustness of the model (Additional file 1: Figure S6). Subgroup analyses based on VP and TP yielded the similar negative results. (Table 2)
Renal failure or injury
Renal failure or injury was not observed as a significant result of vasopressin receptor agonists in the studies analyzed [15, 24, 25, 31, 32, 38, 40]. The combined RR or this outcome was 0.97 (95% CI 0.90–1.05, P = 0.47, I2 = 0%) (Fig. 6b). Subgroup analyses based on VP and TP yielded the consistant results (Table 2).
Renal replacement therapy
A total of five studies [16, 27, 31, 32, 36] were included in the analysis of renal replacement therapy (RRT), and the combined RR was 0.71 (95% CI 0.55–0.93, P = 0.01, I2 = 12%) (Fig. 6b).
Arrhythmia
No effects of vasopressin receptor agonists on arrhythmia were found in the studies analyzed [15, 16, 24, 26–28, 31, 33, 35, 36, 38, 40]. The combined RR was 0.86 (95% CI 0.69–1.07, P = 0.18, I2 = 2%) (Fig. 6b). Subgroup analyses based on VP and TP yielded the same results (Table 2).
Digital ischemia
Twelve studies [15, 16, 23, 24, 27, 28, 31–33, 35, 38, 40] were included in the analysis. The combined RR was 3.85 (95% CI 2.45–6.06, P < 0.01, I2 = 27%) (Fig. 6b), a significant association between the use of vasopressin receptor agonists and an increased risk of digital ischemia events.
Mesenteric ischemia
No significant effects of vasopressin receptor agonists on mesenteric ischemia were observed [24, 31, 38]. The combined RR was 0.88 (95% CI 0.46–1.67, P = 0.69, I2 = 0%).