TCTR20160808002 in Thai Clinical Trials Registry ( https://www.thaiclinicaltrials.org/export/pdf/TCTR20160808002), Prospectively registered on 3 ^rd August 2016.

Sepsis and septic shock potentially cause high mortality and burden for patients and health care system. ^{1 , 2} Rivers, et al. introduced early goal-directed therapy as the mainstay treatment of sepsis, which has been able to reduce the mortality rate of patients with sepsis or septic shock for more than two decades. ³ The Surviving Sepsis Campaign Bundle 2018 Update recommends to initiate rapid administration of intravenous crystalloid solution, at least 30 milliliters per kilogram (mL⋅kg ⁻¹), in the first hour for resuscitation in the early phase of sepsis ⁴ because most patients with sepsis are affected by hypovolemic status. Adequate cardiac preload is an essential component for hemodynamic restoration in patients with sepsis. Nevertheless, excessive fluid resuscitation may occur and harm patients with sepsis. ^{5 – 11} Several physiologic studies have shown that dynamic variables, e.g., pulse pressure variation, ^{12 , 13} stroke volume variation, ^{14 – 16} respiratory change in aortic blood velocity, ^{17 , 18} vena cava diameter variation, ^{19 – 22} etc. were more accurate for assessment of fluid responsiveness, ²³ while systematic reviews found that static central venous pressure (CVP) was not highly reliable for assessment of fluid responsiveness. ^{24 , 25} Surviving Sepsis Campaign Guidelines 2016 suggest using dynamic variables over static variables to guide fluid resuscitation, ²⁶ but currently, dynamic variables are not yet standard for prediction of fluid responsiveness.

This study aimed to determine the effects of dynamic inferior vena cava (IVC)-guided versus static CVP-guided fluid resuscitation in patients with sepsis on mortality and other clinical outcomes, i.e., targets of resuscitation, duration of shock, amount of fluid and vasopressor administration, invasive respiratory support, length of stay (LOS) in hospital and intensive care unit (ICU), and adverse events.

A total of 995 patients with sepsis or septic shock treated at the emergency department or medical wards in TUH between August 2016 and April 2020 were screened for eligibility, and 124 patients were included. After stratification by APACHE II of <25 or ≥25, 62 patients were randomized into each intervention group. Individual patients were followed until shock recovery, transfer or death within 30 days. One patient in the dynamic IVC measurement group withdrew after randomization because of discomfort for data collection, leaving 123 patients included in the modified intention-to-treat analysis. One additional patient in the static CVP measurement group was excluded from analysis for the 30-day mortality because they were referred to another hospital and their survival status could not be tracked. The study process from enrollment to analysis is shown as a Consolidated Standards of Reporting Trials (CONSORT) flow diagram in Figure 3. The study was ended after completion of patient recruitment and follow-up.

The baseline characteristics and severity of sepsis, including ICU admission, were not significantly different between the two intervention groups, as shown in Table 1. Of 111 patients (90.2%) who required NE, the number of patients in the dynamic IVC measurement group (52, 85.3%) was lower than that in the static CVP measurement group (59, 95.2%), but not statistically different (p = 0.064). The primary outcome and most secondary outcomes were not significantly different between both groups, i.e., mortality rates in 30 days, achievement of macrovascular target(s) (MAP ≥65 mmHg or urine output ≥0.5 mL⋅kg ⁻¹⋅h ⁻¹ or both) in 6 h or achievement of microvascular target(s) (ScvO ₂ ≥70% or lactate clearance ≥10% or both) in 6 h, as well as total volume of fluid and total volume of fluid per kg of BW administered in the first 72 h of sepsis, ETT with MV, duration of MV, hospital LOS and ICU LOS. However, the dynamic IVC measurement group demonstrated significantly shorter median (IQR) duration of shock, considerably lower median (IQR) accumulated dose of NE and lower median (IQR) accumulated dose of NE per kg of BW. Rates of adverse events were not different between the two intervention groups. All outcomes in detail are shown in Table 2. Thirty-day mortality rates and significantly different secondary outcomes (i.e., duration of shock, accumulated dose of NE and accumulated dose of NE per kg of BW) are presented in Extended data Figure S1., S2., S3. and S4. All findings were based on analyses of the dataset from this study. ³⁴

Our study showed that the baseline characteristics were not different between the two intervention groups. The 30-day mortality rates were not different between both groups, nor were achievement of macrovascular and microvascular targets in 6 h, total volume of fluid administration and total volume of fluid administration per kg of BW in the first 72 h of sepsis, rate of ETT with MV, duration of MV, hospital LOS, ICU LOS and rates of adverse events. However, patients in the dynamic IVC measurement group experienced shorter duration of shock, lower accumulated dose of NE and lower accumulated dose of NE per kg of BW, compared with those in the static CVP measurement group.

To our knowledge, this is the first study attempting to determine the clinical effects of fluid resuscitation guided by feasible and practical hemodynamic parameter for prediction of fluid responsiveness in patients with sepsis or septic shock in Thailand. There was no difference in the 30-day mortality rates between the two groups, which was consistent with the 28-day mortality rate of septic shock patients in a study from Richard, et al. ³³ Although the 30-day mortality rates in the two groups were not statistically different, the percentage of mortality in the dynamic parameter group was somewhat lower (absolute difference about 10%) than in the static parameter group in our study, and was even considerably lower (absolute difference about 20%) in that study. ³³ This may imply important clinical significance in certain clinical situations, especially for the disease with high prevalence like sepsis, because the absolute number of deaths would be magnified by the prevalence of the disease. The mortality rates (at 30 days, ^{35 , 36} 90 days, ³⁷ 180 days, ³⁵ and hospital mortality ³⁸ ) of surgical ICU patients between fluid therapy guided by dynamic variables and standard care (fluid therapy guided by static variables or clinical examination) were also not different in other studies. However, a recent systematic review and meta-analysis including studies of critically ill adult patients requiring acute fluid resuscitation demonstrated that dynamic variable-guided fluid therapy decreased the mortality risk (risk ratio = 0.59, 95%CI = 0.42-0.83, p = 0.002) compared with standard care. ³⁹ The difference between our findings and the systematic review findings in terms of mortality may be partially explained by the different types of patients and therefore different pathophysiologies of diseases. The majority of patients in our study were medical patients while most patients in that systematic review were surgical patients. Medical patients usually have more comorbidities, which poses a higher risk of complications and/or death, compared with surgical patients, and this factor may interfere with the effect of intervention on mortality.

Shorter duration of shock and lower accumulated dose of NE (either total dose or total dose per kg of BW) in the dynamic IVC measurement group in our study suggests that dynamic IVC diameter variation may be able to predict fluid responsiveness better than static CVP value, so the patients received more adequate volume of fluid administration. This assumption corresponds to the study findings from Lanspa, et al., which showed that IVC collapsibility index threshold of ≥50% had a fair positive predictive value (75%) and a good negative predictive value (80%) for prediction of fluid responsiveness, although this did not reach statistical significance (p = 0.09). ³¹ Theoretically, patients with adequate fluid volume administration can likely maintain macrovascular and microvascular targets, and minimize duration of shock without excessive dose of vasopressors, resulting in fewer complications from prolonged shock or high dose of vasopressors. ^{40 – 42} However, patients in the dynamic IVC measurement group did not achieve more macrovascular or microvascular target(s) in 6 h than those in the other group in our study. The reason for this may be the fact that the timing of vasopressor initiation was not controlled because it was decided based on the individual attending physician’s discretion. Given that a higher proportion of patients who received vasopressor earlier could possibly restore macrovascular and microvascular targets as demonstrated in a study from Permpikul, et al. (CENSER trial), ⁴³ this may confound the effect of fluid administration guidance method on achievement of macrovascular and microvascular targets.

In our study, total fluid volume and total fluid volume per kg of BW administered in the first 72 h were not different between the two intervention groups, which was consistent with a study from Trof, et al. ⁴⁴ but different from a study by Richard, et al. which showed lower daily intravenous fluid volume administration in a dynamic variable-guided fluid therapy group. ³³ The reason for different findings can be partially explained by different time-frames of fluid administration and also different dynamic parameters guiding fluid resuscitation between our study and that study. ³³ Moreover, our study protocol did not control fluid administration after shock recovery occurring within 72 h, which affected total fluid volume and total fluid volume per kg of BW in the first 72 h. Another study by Douglas, et al. demonstrated less positive fluid balance at 72 h or ICU discharge in septic patients who received dynamic stroke volume-guided fluid resuscitation, compared to those with standard usual care, ⁴⁵ but this outcome cannot be directly compared to total fluid volume in 72 h of our study.

Our study found no difference in rates of ETT and MV between the two intervention groups. This was a reasonable finding because indications of intubation and MV are broader than only unstable hemodynamics. Durations of MV were also not different between the two intervention groups in our study, which was the same as studies of Richard, et al., ³³ Douglas, et al. ⁴⁵ and Trof, et al., specifically for septic shock patients. ⁴⁴

Hospital LOS in both intervention groups were not different, which was similar to some studies ^{35 , 45 – 47} but different from others (shorter hospital LOS in dynamic variable-guided fluid therapy group). ^{36 , 48 – 52} ICU LOS were not different between our two intervention groups, which also was consistent with several studies ^{33 , 37 , 45 , 46 , 49} but different from others. ^{47 , 51 , 52} The heterogeneous findings in hospital LOS and ICU LOS were likely caused by the different nature of diseases, given that most patients in these studies were various types of surgical patients, and there were specific indications for ICU admission for certain types of surgical patients (e.g., postoperative observation).

There are some limitations in this study. Firstly, the sample size was quite small and possibly underpowered to be able to detect the differences of some secondary outcomes between the two intervention groups. A study with a larger sample size is warranted to clarify whether such outcomes are different. However, our sample size was comparable or even greater than other studies assessing the effects on clinical outcomes between dynamic and static hemodynamic parameters. ^{33 , 44 , 53} Second, treatment of sepsis for each patient was based on individual attending physicians, except for fluid resuscitation and intensive monitoring in the early phase of sepsis. This might have interfered with the effects of intervention, but clinical practices were standardized by following the Surviving Sepsis Campaign Guidelines ^{4 , 26 , 28} and standard of care at TUH under supervision of specialists. Third, the US measurement process of respiratory variation of IVC diameter is operator-dependent. Accordingly, all US measurements of IVC diameter variation were done by the same well-trained physician and the same US machine to eradicate interobserver variability and standardize the findings of US measurement. Fourth, the parameters for prediction of fluid responsiveness in our study were derived from right-sided cardiac preload, which indirectly reflects the variation of stroke volume or cardiac output of the left ventricle after fluid administration. Nevertheless, right-sided cardiac preload-derived parameters were selected based on availability and practicability. Fifth, this was a single-center study, which may not be able to perfectly generalize the findings to other institutes or other hospitals. Physicians need to carefully consider the context of individual situations before adoption of our study findings for clinical practices. Strengths of this study include study design (randomized controlled trial), which could balance baseline characteristics and severity of sepsis in patients between the two intervention groups in order to minimize confounding factors affecting the outcomes, and feasibility for limited-resource hospitals.

Dynamic IVC measurement-guided fluid resuscitation in patients with sepsis or septic shock does not affect mortality, macrovascular and microvascular targets, volume of fluid administration, ETT with MV, duration of MV, hospital LOS, ICU LOS and adverse events. Nevertheless, such intervention may decrease the duration of shock and accumulated dose of NE, compared with static CVP measurement-guided fluid resuscitation.

Conceptualization: Sricharoenchai T. and Saisirivechakun P.; Data curation: Sricharoenchai T. and Saisirivechakun P.; Formal analysis: Sricharoenchai T. and Saisirivechakun P.; Investigation: Sricharoenchai T. and Saisirivechakun P.; Methodology: Sricharoenchai T. and Saisirivechakun P.; Project administration: Sricharoenchai T. and Saisirivechakun P.; Resources: Sricharoenchai T. and Saisirivechakun P.; Software: Sricharoenchai T. and Saisirivechakun P.; Supervision: Sricharoenchai T.; Validation: Sricharoenchai T.; Visualization: Sricharoenchai T. and Saisirivechakun P.; Writing − Original Draft Preparation: Sricharoenchai T.; Writing − Review & Editing: Sricharoenchai T. and Saisirivechakun P.