A Systematic Review and Meta-Analysis on the Use of Preemptive Hemodynamic Intervention to Improve Postoperative Outcomes in Moderate and High-Risk Surgical Patients

Abstract

BACKGROUND: 

Complications from major surgery are undesirable, common, and potentially avoidable. The long-term consequences of short-term surgical complications have recently been recognized to have a profound influence on longevity and quality of life in survivors. In the past 30 years, there have been a number of studies conducted attempting to reduce surgical mortality and morbidity by deliberately and preemptively manipulating perioperative hemodynamics. Early studies had a high control-group mortality rate and were criticized for this as being unrepresentative of current practice and raised opposition to its implementation as routine care. We performed this review to update this body of literature and to examine the effect of changes in current practice and quality of care to see whether the conclusions from previous quantitative analyses of this field remain valid.

METHODS: 

Randomized clinical trials evaluating the use of preemptive hemodynamic intervention to improve surgical outcome were identified using multiple methods. Electronic databases (MEDLINE, EMBASE, and the Cochrane Controlled Clinical Trials register) were screened for potential trials, reference lists of identified trials were examined, and additional sources were sought from experts and industry representatives. Identified studies that fulfilled the entry criteria were examined in full and subjected to quantifiable analysis, subgroup analysis, and sensitivity analysis where possible.

RESULTS: 

There were 29 studies identified, 23 of which reported surgical complications. In total, the 29 trials involved 4805 patients with an overall mortality of 7.6%. The use of preemptive hemodynamic intervention significantly reduced mortality (pooled odds ratio [95% confidence interval] of 0.48 [0.33–0.78]; P = 0.0002) and surgical complications (odds ratio 0.43 [0.34–0.53]; P < 0.0001). Subgroup analysis showed significant reductions in mortality for studies using a pulmonary artery catheter, supranormal resuscitation targets, studies using cardiac index or oxygen delivery as goals, and the use of fluids and inotropes as opposed to fluids alone. By contrast, there was a significant reduction in morbidity for each of the 4 subgroups analyzed.

CONCLUSION: 

The use of a preemptive strategy of hemodynamic monitoring and coupled therapy reduces surgical mortality and morbidity.

In Brief

Published ahead of print October 21, 2010

Major surgery is associated with a significant and quantifiable rate of both morbidity and mortality.^1,2 This risk of adverse events is increased in groups of patients with certain clinical criteria, for instance, emergency surgery or surgery in a patient with limited cardiovascular or respiratory reserve. The mitigation of these risks is important both for the individual patient who then has a better outcome and for health care planners and managers who are able to provide a higher quality of care for a reduced overall cost.^1,3

Over the last 30 years, many authors have described how the use of flow-based hemodynamic monitoring combined with hemodynamic manipulations in the perioperative period can reduce the incidence of both morbidity, and in some studies, mortality.^4–32 For a variety of reasons, this practice has not been developed as a routine standard of care and includes the fact that many of these studies have been performed on small sample sizes from single centers. Very few studies have been conducted in a multicentric manner. Previous systematic reviews and meta-analyses have demonstrated benefits with this approach for major surgical patients^33–36; however, there is a need for these to be updated given the recent plethora of studies published since the generation of newer and less-invasive hemodynamic monitors and improvement in the overall quality of care delivered.

This systematic review and meta-analysis was designed to explore whether a preemptive strategy of hemodynamic monitoring and manipulation in the perioperative period for moderate- and high-risk surgical patients can improve postoperative outcome.

METHODS

Search Strategy

Three electronic databases (MEDLINE, EMBASE, and the Cochrane Controlled Clinical Trials register) were searched with the following keywords: hemodynamic monitoring, cardiac output, stroke volume, oxygen delivery, goal-directed therapy, dobutamine, dopexamine, surgery, and randomized controlled trial. The search strategy was run from 1985 and closed on January 24, 2010. Articles were restricted to English language and human studies only. In addition to electronic searching, industry representatives were contacted for additional material, and personal archives and communications were searched. All identified review articles and evidence-based guidelines were hand-searched for additional references, and reference lists for identified studies were snowballed for additional articles. The title and abstracts identified from the search strategy were then screened for potential articles by 2 investigators. After this primary exclusion, full articles were obtained and examined for suitability.

Study Inclusion Criteria

All randomized controlled clinical trials evaluating a preemptive hemodynamic monitored approach to cardiovascular management were considered and reviewed. All studies had to be properly randomized to control for selection bias and had to report hospital mortality as an outcome on an intention-to-treat basis. Studies were excluded from the analysis if the hemodynamic monitoring was only used differently between the control and protocol groups before randomization, because these studies tended to be fixed-dose drug studies that did not fit our selection criteria.^24,37,38 Only peer-reviewed papers were included. Abstracts from scientific meetings were not screened; previous studies in this field have shown only low yield to this process, despite attempting to counter for potential reporting bias.

We defined a hemodynamic intervention as the proactive use of hemodynamic monitoring and therapies to manipulate hemodynamics in the perioperative period. Therapies could be classified as IV fluids and/or additional inotropic support. The hemodynamic intervention had to be preemptively started in the perioperative period, which was defined as 24 hours preoperatively, intraoperatively, or up to 24 hours into the postoperative period. Previous meta-analyses have included heterogeneous groups of patients that were not restricted to moderate- and high-risk groups of surgical patients. We therefore aimed only to assess studies that were assessing the impact of these interventions on a moderate- to high-risk group of patients. We defined this group according to criteria previously published by Shoemaker et al.¹¹ and later modified by Pearse et al.¹⁵

Methodological Quality of Included Studies

Eligible studies were graded using the systems described by Jadad et al.³⁹ Nonrandomized studies were excluded. This scale is used to describe study quality by scoring 5 elements of randomization, application, and blinding with a score range of 1 to 5.

Analysis of Outcomes

The primary outcome was hospital mortality. As an outcome measure, it is discrete, well defined, and reported in the majority of articles. Our secondary outcome measure was the number of patients with complications after surgery. Although this outcome measure is less easy to define, it is frequently reported and provides a description of what is happening to the patient population. We chose not to report length of stay because it is often used as a marker of process and reported differently across institutions and countries. A number of a priori subgroup and sensitivity analyses were planned: (a) the type of monitoring used, (b) therapy used (fluids versus fluids and inotropes), (c) therapeutic goals, and (d) resuscitation target (normal versus supranormal). Supranormal was defined as any study that aimed to achieve an oxygen delivery index of ≥600 mL/min/m² in 1 of the trial arms. Data were extracted from each original article by 2 authors and cross-checked for reliability; disputed data were resolved by the third author by a majority decision on reference to the original text.

A sensitivity analysis was performed on both the primary and secondary outcomes. This consisted of a correction for quality using the Jadad score, with a score >3 classified as a higher quality study.³⁹ In addition, a time-dependent analysis was performed to examine the influence of change in care and underlying event rates in the last 35 years by decade.

Statistical Analysis

The meta-analysis was performed using the Review Manager, version 5 software (The Cochrane Collaboration, Oxford, UK), with a random effects model. The results are presented as an odds ratio (OR) for dichotomous data with its 95% confidence intervals (CIs). All results were checked for statistical heterogeneity using the I² methodology. Significance was set at a P value <0.05.

RESULTS

Included Trials

Our search strategy retrieved 4974 titles suitable for further review. Screening of these titles and abstracts produced 91 potential articles that were examined in detail against the predefined eligibility criteria. Further examination led to the exclusion of 68 studies from the analysis because they did not meet high-risk criteria, lacked randomization or nonprospective study design, or did not fulfill our moderate- and high-risk inclusion criteria (Fig. 1). This resulted in 23 articles that were included from electronic databases. Through snowballing of references, hand-searching, and contacting experts and industry representatives, 6 more articles were added. Therefore, 29 articles were included in the final analysis.

Figure 1: Flow chart showing the number of abstracts and articles identified and evaluated during the review process.

Description of Studies

The 29 identified studies are described in detail in Table 1. All reported mortality as an end point. Three studies had a zero rate of mortality in both protocol and control groups. These 3 studies were reexamined to ensure that they met moderate- to high-risk criteria. Twenty-three studies reported the number of patients in whom complications developed during the course of their stay. The reporting of complications was variable across the studies as were the definitions of complications in use. Two studies did use standardized methods of outcome reporting such as the postoperative morbidity survey, but the majority were diffuse.^6,40

Table 1: Randomized Clinical Trials of Preemptive Hemodynamic Intervention to Improve Postoperative Outcome

Mortality

All 29 studies reported mortality as an end point. The overall effect when combining the studies was a significant reduction in mortality for the intervention group (pooled OR of 0.48 [0.33–0.78]; P = 0.0002) (Fig. 2). Subgroup analysis of the mortality end point revealed that mortality was reduced in those studies using a pulmonary artery catheter (OR 0.35 [0.19–0.65]; P = 0.001), for fluids and inotropes as opposed to IV fluids alone (OR 0.47 [0.29–0.76]; P = 0.002), cardiac index or oxygen delivery as the end point (OR 0.38 [0.21–0.68]; P = 0.001), and studies using a supranormal resuscitation target (OR 0.29 [0.18–0.47]; P = 0.00001) (Table 2).

Figure 2: Effects of preemptive hemodynamic intervention in protocol group versus control on mortality rate. M-H = Mantel-Haenszel.

Table 2: Subgroup Analysis for Mortality

Morbidity

Twenty-three of the 29 studies reported the number of patients with complications as an end point. Meta-analysis of these studies (Fig. 3) demonstrated a significant reduction in the overall complication rate (OR 0.43 [0.34–0.53]; P < 0.00,001) and a significant reduction across all of the 4 subgroups assessed (Table 3).

Figure 3: Effects of preemptive hemodynamic intervention in protocol group versus control on complication rate. M-H = Mantel-Haenszel.

Table 3: Subgroup Analysis for Number of Patients with Complications

Trial Quality

The quality of the individual studies as assessed by the Jadad score is presented in Table 4. It is apparent that very few of the studies were performed in a double-blind manner and nearly all were done in a single center. Figure 4 shows an OR plot of mortality split by quality. The higher quality studies (with a Jadad score ≥3) fail to show a significant reduction in mortality, as opposed to lower quality studies that do. The effect of quality on morbidity is shown in Figure 5. In contrast to mortality, there is a significant reduction in morbidity irrespective of trial quality. The point estimate of effect is similar for the 2 groups but the CI for the lower quality studies is wider.

Table 4: Quality Assessment of Included Randomized Clinical Trials

Figure 4: Effects of preemptive hemodynamic intervention in protocol group versus control on mortality rate, grouped by quality of the study as assessed by a Jadad score of more than or less than 3. M-H = Mantel-Haenszel.

Figure 5: Effects of preemptive hemodynamic intervention in protocol group versus control on complication rate, grouped by quality of the study as assessed by a Jadad score of more than or less than 3. M-H = Mantel-Haenszel.

Time-Dependent Analysis

Figure 6 shows a graph of the apparent decline in control-group mortality over time, with recent studies demonstrating lower mortality rates. Figure 7 shows an approximate halving of mortality rates in the control group every decade (29.5%, 13.5%, 7%). It can be seen, however, that although the mortality is reduced over time, the complication rate remains consistent, with approximately one-third of patients experiencing complications (Fig. 8).

Figure 6: Graphical representation of control-group mortality in identified trials over time.

Figure 7: Effects of preemptive hemodynamic intervention in protocol group versus control on mortality rate, grouped by the decade the study was performed. M-H = Mantel-Haenszel.

Figure 8: Effects of preemptive hemodynamic intervention in protocol group versus control on complication rate, grouped by the decade the study was performed. M-H = Mantel-Haenszel.

DISCUSSION

This systematic review and meta-analysis has demonstrated that preemptive hemodynamically targeted therapy in the perioperative period can reduce both morbidity and mortality after surgery. Although over time the control-group mortality decreased, suggesting a lowering of the threshold for the performance of these techniques, the impact of this therapy remains even for the lower risk categories of patients. Although mortality was not proven to be reduced in the lower risk group, the effects on reducing morbidity were still valid, confirming the assumption that the technique of targeted hemodynamic intervention is beneficial across risk profile groups and across monitoring technologies.

There are a number of reasons why the control mortality may have decreased over time. These include the possibilities of better overall care thus decreasing mortality for similar patients, clinicians learning from previous early published studies and therefore drifting their practice toward lower risk groups, and also the likelihood that as technology has improved and become less invasive, the technique has gained more credibility. This can be seen especially in the way the pulmonary artery catheter, with all of its incumbent controversies,^41,42 has now been largely superseded by less-invasive hemodynamic monitoring techniques such as esophageal Doppler-based systems and arterial pressure analysis.⁴³ It is of note that, although the debate surrounding the pulmonary artery catheter focused on an inability to prove a significant beneficial effect to patients,^44–46 this study has demonstrated a highly significant reduction in both morbidity and mortality with the use of this technique for these patients. The same is also true for the newer generation of monitoring modalities.

The burden of complications and mortality for surgical patients is becoming increasingly understood.^1,3 Many authors have now demonstrated that the rate of complications is related to a number of factors that include the type of surgery performed, the skill of the operating team, the overall “fitness” of the patient, and also the provision of a number of techniques that have shown to reduce the risk.^47–50

This study confirms that hemodynamic targeted therapy can reduce this risk. Khuri et al.¹ demonstrated that this reduction in postoperative complications can have long-lasting effects on the survival of these patients, outside of the remit of a short-term follow-up period in these studies. If this hypothesis is correct, then the upfront costs of this relatively inexpensive technique are easily outweighed by the longer-term benefits. Irrespective of this, the prevention of complications is in itself a mechanism for saving significant amounts of health care resources, because it is often these complications that prolong hospital length of stay and result in multiple costly interventions. Recent work published in the New England Journal of Medicine has also raised the possibility that survival is related to the identification and then immediate and appropriate management of these complications.³ It remains a possibility that patients being studied in trials such as ours have a lower than normal complication and mortality rate for this very reason. By participating in a study, they are frequently assessed and probably offered a quality of care that is above the standard approach.

This study has a number of limitations. We made no attempt to correct for the type or quantity of fluids or inotropes given, because they are inconsistently reported in the literature and have a demonstrable wide variability in their dosing across studies. Also, a number of grouped studies rather than individual patient data were meta-analyzed. Some authors would suggest that this would be a more robust methodology, although obtaining the original data is often not possible, especially over such a long period such as this. It also has to be recognized that very few of the studies that we identified were performed in a high-quality design. It is almost impossible to have a properly double-blind study when the 2 groups need to have therapy targeted to different protocols. It is also of note that the majority of the trials were single-centered and performed on a limited sample size. The heterogeneity of this analysis is therefore relatively high, although the results remain consistent across a number of subgroups and sensitivity analyses, thereby helping to affirm our assumptions. We have also reported on studies that describe the incidence of postoperative complications. It has to be recognized that the reporting of complications is not consistent and that the definitions used can differ, limiting the applicability of some of our findings.

CONCLUSIONS

This meta-analysis suggests that a preemptive targeted approach to the management of hemodynamics in the perioperative period may reduce morbidity and mortality for high-risk surgical patients.

AUTHOR CONTRIBUTIONS

All authors were involved and helped in the design, execution, analysis, and writing of the manuscript.

DISCLOSURE

Andrew Rhodes, LiDCO (lecturing fees and grant), Edwards Lifesciences (lecturing fees), Hutchinson Medical (research support), PulseCor (research support), Cheetah (research support), and Abbott (consulting fees); Mark A. Hamilton, Deltex Medical (lecturing fees), Edwards Lifesciences (lecturing fees), and Hutchinson Medical (research support); and Maurizio Cecconi, LiDCO (lecturing fees and grant), Edwards Lifesciences (lecturing fees), Hutchinson Medical (research support), PulseCor (research support), and Chhetah (research support).