Risks and benefits of hypotensive resuscitation in patients with traumatic hemorrhagic shock: a meta-analysis

Background Damage control strategies play an important role in trauma patient management. One such strategy, hypotensive resuscitation, is being increasingly employed. Although several randomized controlled trials have reported its benefits, the mortality benefit of hypotensive resuscitation has not yet been systematically reviewed. Objectives To conduct a meta-analysis of the efficacy of hypotensive resuscitation in traumatic hemorrhagic shock patients relative to mortality as the primary outcome, with acute respiratory distress syndrome (ARDS), acute kidney injury (AKI), and multiple organ dysfunction as the secondary outcomes. Methods PubMed, Medline-Ovid, Scopus, Science Direct, EMBASE, and CNKI database searches were conducted. An additional search of relevant primary literature and review articles was also performed. Randomized controlled trials and cohort studies reporting the mortality rate associated with hypotensive resuscitation or limited fluid resuscitation were selected. The random-effects model was used to estimate mortality and onset of other complications. Results Of 2114 studies, 30 were selected for this meta-analysis. A statistically significant decrease in mortality was observed in the hypotensive resuscitation group (risk ratio [RR]: 0.50; 95% confidence interval [CI]: 0.40–0.61). Heterogeneity was observed in the included literature (I2: 27%; degrees of freedom: 23; p = 0.11). Less usage of packed red cell transfusions and fluid resuscitations was also demonstrated. No significant difference between groups was observed for AKI; however, a protective effect was observed relative to both multiple organ dysfunction and ARDS. Conclusions This meta-analysis revealed significant benefits of hypotensive resuscitation relative to mortality in traumatic hemorrhagic shock patients. It not only reduced the need for blood transfusions and the incidences of ARDS and multiple organ dysfunction, but it caused a non-significant AKI incidence.


Introduction
Hemorrhagic shock is one of the most common causes of death in trauma or traumatized patients [1]. This is due to the fact that hemorrhagic shock sets in motion a vicious cycle of outcomes, consisting of hypothermia, acidosis, and coagulopathy-otherwise known as the lethal triad. To mitigate these effects, damage control strategies have been proposed, including the early control of bleeding and adequate fluid resuscitation. The aim of hypotensive resuscitation is to maintain systolic blood pressure (or mean arterial pressure) in order to sustain organ perfusion [2,3]. It was believed that induced intraoperative hypotension could lead to reduced blood loss and fewer transfusions [4][5][6]. However, much recent data have shown that there is a significant relationship between hypotensive resuscitation and postoperative renal injuries [7][8][9].
Fluid resuscitation with the rapid administration of intravenous fluids until the blood pressure is normalized is the traditional fluid-resuscitation strategy. A number of randomized controlled trials have demonstrated some improvement in survival and mortality using a liberal fluid hypotensive resuscitation. Nevertheless, other studies have concluded that a more conservative hypotensive strategy-in which minimal amounts of fluids are given until the severe bleeding has been controlledis a more efficacious resuscitation strategy [10][11][12][13]. Although the volume of data on hypotensive resuscitation continues to grow, conflicting findings are being reported regarding the efficacy of this traumamitigating strategy. Accordingly, the aim of this study was to conduct a meta-analysis of the efficacy and drawbacks of hypotensive resuscitation in traumatic hemorrhagic shock patients relative to mortality as the primary outcome, with acute respiratory distress syndrome, acute kidney injury, and multiple organ dysfunction as the secondary outcomes.

Inclusion criteria
This meta-analysis included only randomized controlled trials (RCTs) and cohort studies that evaluated adult patients aged older than 18 years who had traumatic hemorrhagic shock and a systolic blood pressure below 90 mmHg.
The intervention assessed was conventional fluid resuscitation with normotension (liberal fluid resuscitation) versus hypotensive resuscitation (limited fluid resuscitation). Conventional fluid resuscitation was defined as liberal fluid resuscitation until the systolic blood pressure exceeded 90 mmHg (normal blood pressure). Hypotensive resuscitation was defined as limited fluid resuscitation to maintain adequate organ perfusion, with a systolic blood pressure of~70-80 mmHg or a mean arterial pressure of~50 mmHg.
Excluded were studies of patients who were pregnant or had traumatic brain injuries, research with insufficient mortality data, and investigations that had not received ethical approval.

Outcomes assessed
The primary outcome was all-cause mortality, as reported by the authors of the included studies. The secondary outcomes included the rates of the following morbidities: acute respiratory distress syndrome (ARDS), acute kidney injury (AKI), and multiple organ dysfunction (MODS). Other secondary outcomes included the fluid resuscitation volume and the transfusion of packed red cells.

Search methods
PubMed, Medline-Ovid, Scopus, Science Direct, EMBASE, and CNKI database searches were conducted for articles published before January 31, 2018. An additional search of relevant primary literature and review articles was also performed. The references from identified studies that appeared to be germane to the topic of study were hand-searched. The medical subject headings (MeSHs) used in our searches included "hypotensive resuscitation" or "limited fluid resuscitation" and "trauma" or "trauma*". There was no language restriction.

Data collection
Two reviewers (N.O. and T.B.) independently inspected each article identified during the search, scanned the full text of relevant articles, applied the inclusion and exclusion criteria, and extracted and  recorded the data. Disagreements relating to any aspect of the data extraction process were resolved by discussion with a third reviewer (K.C.), with the final decision made by consensus.

Quality assessment
The quality of included studies was evaluated using the Jadad quality assessment scale for randomized controlled studies. Our subsequent analysis only included those studies that scored ≥2 on the scale (which indicated that their results were valid) [14].
To assess the quality of nonrandomized studies, the Newcastle-Ottawa Scale (NOS) was used. This 3-item scoring system evaluates the selection of the participants; the comparability between the groups; and the ascertainment of exposure for case-control studies, and the outcome of interest in the case of cohort studies [15].

Statistical analysis
All statistical analyses were performed using Review Manager 5.3 software from the Cochrane Collaboration (London, United Kingdom). We extracted the proportions and 95% confidence intervals from each study and pooled them using the random effects model. Cochran's Q test was performed and quantified using the I 2 statistic to evaluate the statistical heterogeneity among the included studies. The I 2 -values were categorized as follows: 0-25% indicated insignificant heterogeneity; 26-50%, low heterogeneity; > 50% to ≤75%, moderate heterogeneity; and > 75%, high heterogeneity [16]. The fixed effects model was used for analyses that evaluated data with no significant heterogeneity. Funnel plots were used to evaluate for publication bias. P-values less than 0.05 were considered statistically significant.

Results
Search results and study characteristics were prospective cohort studies. The baseline characteristics described in Table 1 comprise country, article type, the quality assessment score for each study, and the intervention definition.

Assessment of reporting bias
A funnel plot of the reporting bias shows asymmetry on visual inspection (Fig. 2 [20], and Wang Fengyong, 2016 [27]). These studies give the plot an asymmetrical appearance, with a gap at a bottom corner of the graph. In this setting, the effect calculated in a meta-analysis will tend to overestimate the intervention effect. However, after reviewing the study design and evaluating the outcomes, we still decided to include them in our analysis. The Egger test found a strong significance in publication bias (P = 0.000) Therefore, the trim and fill method was conducted to adjust the publication bias by removing the smaller studies causing asymmetry and replacing the omitted studies. This method has been claimed to improve the effect size and confidence intervals [36]. As a result, the present study drew upon 11 hypothesized studies. The new RR was 0.65 and CI (0.52-0.80), which showed significant difference in mortality benefit (Fig. 3).

Synthesis of primary outcome
A pooled analysis was performed of the 24 studies using a random-effects model, with findings reported as RR and 95% CI (n = 1473; RR: 0.50; 95% CI: 0.40-0.61). A mild heterogeneity among these 24 studies was observed (Q test: 0.11, which is greater than 0.1; I 2 : 27%). Given the degree of homogeneity among the studies, a random-effects model was subsequently applied (Fig. 4). As described in Fig. 4, some overtime bias was observed among the studies. Although the first published study on this topic (by Bickell WH et al. [2]) was published in 1994, we still included it due to its sound methodology and overall high level of quality.

Secondary outcomes
The hypotensive resuscitation group had a lower amount of fluid resuscitation and packed red cell transfusion than the normotensive resuscitation group (Figs. 5 and 6). The mean difference in fluid resuscitation between groups was 1233 ml (95% CI: -1576, -890) and red cell transfusion was 132 ml (95% CI: -203, -61).

Discussion
During the last decade, two differing approaches to damage control have been used for severe trauma patients. The traditional liberal fluid approach has, however, come under increasing attention due to the possible complications associated with administering more than 2 l of fluid before surgery. In contrast, hypotensive resuscitation is being increasingly emp loyed because many animal studies have found its use is associated with reduced effects of the lethal triad. The concept of this strategy is to restrict the amount of fluid resuscitation to maintain a low enough blood pressure for adequate cerebral perfusion. To avoid further bleeding due to dilution coagulopathy and to dislodge hemostatic blood clots, hypotensive resuscitation has been adopted as a part of damage control resuscitation for trauma patients [37,38]. Bickell WH et al. were the first to study and publish the role of hypotensive resuscitation in penetrating trauma patients. They showed that hypotensive resuscitation before surgery could significantly decrease mortality. Not only did the fatality rate decrease, but the hypotensive resuscitation also appeared to correlate with less coagulopathy and other complications, including less cardiovascular failure, respiratory failure, and acute kidney injury (AKI). Despite the sizeable number of studies on hypotensive resuscitation in animal models and in humans, there is only one Cochrane-based review, and that review pertains only to the timing and volume of fluid administration in patients with bleeding. Moreover, in the present study, all types of hemorrhagic shock were included, including gastrointestinal bleeding. The results of this meta-analysis showed that there is no evidence for or against an early or larger volume of intravenous fluid administration in cases of uncontrolled hemorrhage.
Given that there is no meta-analysis related to studies comparing hypotensive to liberal fluid resuscitation, we decided to perform such an investigation. It is noteworthy that a large majority of the present literature had to be excluded from further analysis. This was mainly due to either the presence of irrelevant data and/or the lack of sufficient data. This highlights and supports the validity of the Cochrane approach.
Our analysis revealed that hypotensive resuscitation (i.e., limited fluid resuscitation) has beneficial effects on survival in traumatic hemorrhagic shock. Even in Fig. 6 Forest plot of association between hypotensive resuscitation and normal resuscitation, relative to transfusion of packed red cells the blood product based resuscitation era, all the literature still shows significant survival benefits [10][11][12][13]. Hypotensive resuscitation is also associated with a significantly lower amount of fluid resuscitation and packed red cell transfusion, and a significantly lower incidence of acute respiratory distress syndrome and multiple organ dysfunction. In contrast, no significant difference was observed between resuscitation methods relative to the incidence of acute kidney injury. Moreover, we were not able to identify any report of renal replacement therapy and/or long-term dialysis in traumatic hemorrhagic shock patients.
The subgroup analysis that we performed showed a mortality benefit of hypotensive resuscitation in traumatic hemorrhagic shock with coexisting traumatic brain injury. This might alert readers to carefully interpret the reported findings from Gong Hong-chang et al. and LIU Wei-zheng et al., whose studies both maintained low normal blood pressure (SBP 90 mmHg or MAP 60-70 mmHg). Meanwhile, LIU Yu et al. maintained SBP 70 mmHg, which truly met the definition of hypotensive resuscitation. However, the data is not compelling enough to strongly recommend hypotensive resuscitation in traumatic hemorrhagic shock patients with traumatic brain injury, and there were no reported data on the functional outcomes after this strategy.
This study supports the use of hypotensive resuscitation, not only in penetrating wound patients, but also in other types of traumatic injuries. This significant finding was based on low heterogeneity. However, the limitations of this analysis include the fact that some clinical and methodological heterogeneities between the studies cannot be ruled out, and there may be some overtime bias. Despite significant publication bias shown by an asymmetrical funnel plot and Egger test, the trim and fill method also showed a statistically significant mortality benefit from the hypotensive strategy. Although a large, multi-center, randomized controlled trial should be conducted, the likelihood of such an approach succeeding given the present literature (including the present review) seems unlikely.

Conclusions
The results of this meta-analysis revealed a significant benefit of hypotensive resuscitation relative to mortality in traumatic hemorrhagic shock patients. Moreover, hypotensive resuscitation was found to be a significantly more effective strategy than traditional fluid resuscitation in terms of acute respiratory distress syndrome and multiple organ dysfunction.