Introduction
Sepsis is a systemic response to infection, which may progress to severe sepsis and septic shock [
1]. Severe sepsis and septic shock represent global problems with a high economic burden. In the United States, more than 750,000 people experience severe sepsis each year, with a short-term mortality of 20% to 30%, reaching up to 50% when shock is present [
2],[
3]. Therefore, numerous therapeutic strategies that aimed at reducing mortality in these patients have been investigated. However, most of them have not led to significant reductions in mortality [
4]. Goal-directed therapy (GDT) has been shown to substantially improve clinical outcomes in surgical patients [
5]. Two important aspects of a GDT protocol include early initiation of the therapeutic measures, together with specific (hemodynamic) targets. The Surviving Sepsis Campaign guidelines recommend GDT for the early resuscitation of patients with sepsis [
1], which is based largely upon the results of the Rivers
et al. trial, in which the researchers reported a 16% absolute reduction in mortality among patients with severe sepsis or septic shock who received early GDT compared to standard therapy [
6]. However, in a recent study, this approach was challenged [
7], with no benefit shown when GDT was compared to standard therapy. However, initiating therapy early rather than late in the course of critical illness remains a logical clinical goal. In the context of this situation, we systematically reviewed all trials of GDT in patients with sepsis and performed a meta-analysis, focusing on the early initiation of the protocol and its effect on mortality.
Methods
Ethical approval and patient consent were not required, because we conducted a meta-analysis of previously published studies. We followed the recommendations of the
Cochrane Handbook for Systematic Reviews of Interventions to carry out the study [
8] and follow the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement to report our meta-analysis [
9].
Search strategy
Electronic searches were conducted in the PubMed, Embase and Cochrane Central Register of Controlled Trials (CENTRAL) databases. Search terms included “goal directed,” “goal oriented,” “goal target,” “cardiac output,” “cardiac index,” “oxygen delivery,” “oxygen consumption,” “cardiac volume,” “stroke volume,” “fluid therapy,” “fluid loading,” “fluid administration,” “optimization,” “optimization,” “supranormal” and “sepsis,” “severe sepsis,” “septic shock,” “septicemia,” “septicaemia,” “pyohemia,” “pyaemia” and “pyemia.” There was no language restriction placed on the searches. Each database was searched from inception to April 2014. Additionally, reference lists in the articles chosen for inclusion, and the reference lists of previous reviews were screened to identify other potentially eligible trials.
Inclusion criteria
We included trials with the following characteristics:
1.
Population: Adult patients with one or more of the following characteristics were eligible for inclusion: sepsis, severe sepsis or septic shock. Adults were defined as being of a legal age for consent in the country where the trial was conducted. Studies that included sepsis secondary to noninfectious causes were excluded.
2.
Intervention: The intervention had to be GDT, defined as an explicit protocol encompassing the use of hemodynamic monitoring and manipulation of hemodynamic parameters to achieve predetermined hemodynamic endpoints.
3.
Control: The control group had to have received standard therapy or usual care.
4.
Outcomes: The overall mortality rate had to be the outcome measured.
5.
Type of study: The studies had to be randomized controlled trials (RCTs).
We included studies that randomized a mixed population of critically ill patients when a septic subpopulation that met our inclusion criteria was defined; that is, the patients with sepsis constituted a subgroup of the trial population.
We extracted data using a standardized data collection form. Discrepancies in collected data were addressed through team consensus. The following information was extracted from each trial: first author, year of publication, number of patients (GDT and control), study population, clinical setting, goals in GDT and control groups, timing of GDT, mortality endpoint, study design (patient selection and concealment) and outcome data (overall mortality and dobutamine use).
Outcomes and definitions
The prespecified primary outcome was overall mortality. If the study authors reported mortality at one time point, we used the only data used for analysis. If the study authors reported mortality at more than one time point, we used hospital mortality preferentially. Secondary outcomes included overall mortality in the early-initiated GDT (that is, within the first 6 hours) versus GDT and dobutamine use initiated later.
Assessment of risk of bias
We used the Cochrane Collaboration tool to assess the risk of bias of individual study and with bias domains across studies [
8],[
10]. Two investigators (WJG and FW) subjectively reviewed all studies and assigned a value of “high,” “low” or “unclear” to the following domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias. Trials with high risk of bias for any one or more key domains were considered to be at high risk of bias. Trials with low risk of bias for all key domains were considered to be at low risk of bias. Otherwise, they were considered to have an unclear risk of bias [
10].
Statistical analysis
We estimated the relative risk (RR) with 95% confidence interval (CI) for dichotomous outcomes. Statistical heterogeneity across studies was tested by using the
I2 statistic [
11],[
12]. Heterogeneity was suggested if the
P-value was ≤0.10.
I2 values of 0 to 24.9%, 25% to 49.9%, 50% to 74.9% and 75% to 100% were considered zero, low, moderate and high thresholds for statistical heterogeneity, respectively [
11],[
12]. Clinical heterogeneity could not be excluded, so the more conservative random-effects model [
13] (Mantel-Haenszel method) was used. Predefined subgroup analysis was conducted according to the timing of GDT for resuscitation (early being within the first 6 hours versus late or unclear being outside the first 6 hours or unclear timing). In addition, we performed
post hoc subgroup analyses according to risk of bias (low versus unclear), sample size (≥100 versus <100) and setting (emergency department versus intensive care unit). We further investigated the influence of a single study on the overall pooled estimate by omitting one study in each step. The potential for bias was assessed by inspection of a funnel plot and Egger’s test [
14]. The results were considered statistically significant at two-sided
P-values <0.05. All statistical analyses were performed using RevMan 5.2 software (The Nordic Cochrane Centre, Copenhagen, Denmark).
Quality of evidence
We evaluated the quality of the evidence by using the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach [
15]. In addition, the GRADEprofiler 3.6 software (The Nordic Cochrane Centre) was used to create the evidence profile. The GRADE Working Group grades of evidence used were as follows:
(1)
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
(2)
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
(3)
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
(4)
Very low quality: We are very uncertain about the estimate.
Discussion
Our meta-analysis of 13 RCTs showed that GDT was associated with a 17% RR reduction on overall mortality in patients with sepsis. This mortality benefit was present in studies in which GDT was started early, but not when initiated late or when the timing of GDT was unclear. In addition, GDT was significantly associated with dobutamine use.
Despite these findings, the effect of GDT remains a matter of debate, as the most recent and largest trial included in this meta-analysis [
7] did not show a difference in mortality, in contrast to many of the preceding studies. This could be due to the effect of the rapid acceptance of the principal interventions of the Rivers
et al. study [
6] and the subsequent Surviving Sepsis Campaign guidelines, which encompassed all elements of the Rivers study protocol. This is illustrated by the fact that, in the ProCESS (Protocolized Care for Early Septic Shock) trial, all groups received, on average, more than 2 L of fluid prior to randomization and more than 75% of patients received antibiotics before randomization. In addition, the mortality rate was much lower in the ProCESS trial than in preceding trials, possibly reflecting the effect of early diagnosis, fluid resuscitation and initiation of antibiotics on mortality. As the ProCESS trial, like the Rivers
et al. study, enrolled patients with sepsis in the emergency department, this effect may have been prominent [
27].
Although the current evidence supports the early use of GDT to improve outcomes in patients with sepsis, the optimal goals remain uncertain. Currently, the Surviving Sepsis Campaign guidelines recommend the use of central venous pressure (CVP), mean arterial pressure, urine output and central venous oxygen saturation (ScvO
2) as resuscitation goals. However, many of these recommendations have been questioned in recent studies. A recent study [
28] was designed to compare the use of lactate clearance to ScvO
2 as a goal of early (up to 6 hours) sepsis resuscitation. No significant difference in mortality was found (17% in the lactate clearance group versus 23% in ScvO
2 group). However, when both normalization of ScvO
2 and a rapid decrease in lactate levels were applied as therapeutic goals in the early resuscitation of a mixed group of critically ill patients, including a large subgroup of sepsis patients, mortality was significantly reduced [
29]. In addition, in a recent retrospective study, researchers questioned the CVP endpoint in sepsis resuscitation [
30].
Further research is needed before strong and definitive recommendations can be made regarding the effect of GDT for resuscitation of patients with sepsis. There are currently at least two ongoing RCTs of GDT in patients with sepsis: the Australasian Resuscitation in Sepsis Evaluation (ARISE) trial in Australia (ClinicalTrials.gov ID: NCT00975793) and the Protocolised Management in Sepsis (ProMISe) trial in the United Kingdom (Current Controlled Trials number: ISRCTN36307479) [
31]. The results of these ongoing trials should provide further guidance as to the effect of GDT for resuscitation of patients with sepsis.
A major strength of the present meta-analysis is its compliance with the Cochrane handbook methodology recommendations. We conducted an exhaustive literature search that included non-English-language articles. Two authors independently screened all references, included eligible trials, extracted data information, assessed risk of bias and performed statistical analyses. Moreover, we followed the PRISMA statement to report this meta-analysis and evaluated the quality of the evidence by using the GRADE approach.
Because early fluid resuscitation is vital in patients with sepsis, we performed predefined subgroup analyses according to the timing of GDT. We also performed post hoc subgroup analyses according to risk of bias, disease severity, sample size and publication date. These subgroup analyses based on assessment of bias and clinically relevant groups may help health care professionals in clinical decision-making.
Our analysis also has several limitations that must be taken into consideration when interpreting the results. First, most of the included trials had a high risk of bias (Figure
2). The potential importance of this issue is highlighted by the fact that predefined subgroup analysis comparing mortality estimates between trials with low versus unclear risk of bias suggested the mortality benefit is not clearly apparent among the trials with low risk of bias, although this subgroup difference was not statistically significant (
P =0.90). Second, there were some differences in the target populations and protocols of GDT of each study. These factors may have a potential impact on our results and may preclude firm conclusions. Third, different endpoints were used for mortality evaluation. Because this study was a study-level meta-analysis, individual patient data were not included in the analysis; thus, we could not adjust for patient-level confounders.
Authors’ contributions
WJG conceived of the study, participated in its design, collected the data, performed statistical analyses and drafted the manuscript. FW collected the data, performed statistical analyses and helped to draft the manuscript. JB participated in the study design, collected the data and helped to revise the manuscript critically for important intellectual content. LT collected the data, performed statistical analyses and helped to revise the manuscript critically for important intellectual content. JCL conceived of the study, participated in its design, collected the data and revised the manuscript critically for important intellectual content. All authors read and approved the final manuscript.
Authors’ details
WJG is now studying as a master course student and majors in anesthesiology. His major research interests include systematic reviews and meta-analysis, evidence-based medicine, critical care medicine and clinical anesthesiology. As the first author, he has published more than 20 papers in peer-reviewed journals such as
Chest,
European Respiratory Journal,
Critical Care Medicine,
Intensive Care Medicine and
Critical Care. He has authored chapters of a book on his research interests. He is one of section editors (Systematic Review and Meta-analysis) in the
Journal of Thoracic Disease. He also acts as a forum moderator in
http://i.dxy.cn/profile/200750383. FW is an anesthetist, and he has published more than 10 papers in peer-reviewed journals such as
Chest,
Critical Care and
American Journal of Emergency Medicine. JB is a professor of medicine and vice-chair of the Department of Intensive Care Adults at Erasmus Medical Centre in Rotterdam, the Netherlands, and is a professor of medicine at the Pontificia Universidad Católica de Chile. He is also currently a visiting professor at Columbia University Medical Center-New York Presbyterian Hospital in New York City. He has published over 160 articles and edited 19 books, and he has made presentations at more than 450 conferences. He is the editor of
Intensive Care Medicine and
Netherlands Journal of Critical Care. He is a member of the editorial boards of
Reista Brasileira de Terapia Intensivia (Brasil), the
Turkish Journal of Anaesthesiology and Reanimation,
Journal of Critical Care and
ICU Management. He is former treasurer and president of the Dutch Society of Intensive Care and currently is a member of the organizing committee of the annual International Symposium on Intensive Care and Emergency Medicine in Brussels, Belgium. He has organized a 5-day workshop on intensive care annually for 11 years and annual meetings for ICU nurses for the past 9 years. LT is a nurse anesthetist. JCL is a professor of anaesthesiology and acts as the Director of the Department of Anaesthesiology in the First Affiliated Hospital of Guangxi Medical University.
Competing interests
The authors declare that they have no competing interests.