18
|
Editorials
€ ck J, Enlund M, Hedenstierna G. 10. Edmark L, Auner U, Lindba Post-operative atelectasis – a randomised trial investigating a ventilatory strategy and low oxygen fraction during recovery. Acta Anaesthesiol Scand 2014; 58: 681–8 11. Pagano A, Barazzone-Argiroffo C. Alveolar cell death in hyperoxia-induced lung injury. Ann N Y Acad Sci 2003; 1010: 405–16 12. Girardis M, Busani S, Damiani E, et al. Effect of conservative vs conventional oxygen therapy on mortality among patients in an intensive care unit: the oxygen-ICU Randomized Clinical Trial. JAMA 2016; 316: 1583–9 13. Berger MM, Grocott MPW. Facing acute hypoxia: from the mountains to critical care medicine. Br J Anaesth 2017; 118: 283–6 14. The LAS VEGAS study Investigators for the PROVE Network. Epidemiology, practice of ventilation and outcome for patients at increased risk of postoperative pulmonary complications (LAS VEGAS) – an Observational Study in 29 Countries. Eur J Anaesthesiol (accepted) 15. PROVE Network Investigators for the Clinical Trial Network of the European Society of Anaesthesiology, Hemmes SNT, Gama de Abreu M, Pelosi P, Schultz MJ. High versus low positive end-expiratory pressure during general anaesthesia for open abdominal surgery (PROVHILO trial): a multicentre randomised controlled trial. Lancet 2014; 384: 495–503 16. Ball L, Sutherasan Y, Pelosi P. Monitoring respiration: what the clinician needs to know. Best Pract Res Clin Anaesthesiol 2013; 27: 209–23 17. Rice TW, Wheeler AP, Bernard GR, Hayden DL, Schoenfeld DA, Ware LB. Comparison of the Spo2/Fio2 ratio and the Pao2/ Fio2 ratio in patients with acute lung injury or ARDS. Chest 2007; 132: 410–7 18. Allegranzi B, Zayed B, Bischoff P, et al. New WHO recommendations on intraoperative and postoperative measures for surgical site infection prevention: an evidence-based global perspective. Lancet Infect Dis 2016; 16: e288–303
19. Hedenstierna G, Perchiazzi G, Meyhoff CS, Larsson A. Who can make sense of the WHO guidelines to prevent surgical site infection? Anesthesiology 2017; 126: 771–3 20. Myles PS, Kurz A. Supplemental oxygen and surgical site infection: getting to the truth. Br J Anaesth 2017; 119: 13–6 21. Myles PS, Leslie K, Chan MTV, et al. Avoidance of nitrous oxide for patients undergoing major surgery: a randomized controlled trial. Anesthesiology 2007; 107: 221–31 22. Kurz A, Fleischmann E, Sessler DI, Buggy DJ, Apfel C, Akc¸a O. Effects of supplemental oxygen and dexamethasone on surgical site infection: a factorial randomized trial. Br J Anaesth 2015; 115: 434–43 23. Canet J, Gallart L, Gomar C, et al. Prediction of postoperative pulmonary complications in a population-based surgical cohort. Anesthesiology 2010; 113: 1338–50 24. Mazo V, Sabate´ S, Canet J, et al. Prospective external validation of a predictive score for postoperative pulmonary complications. Anesthesiology 2014; 121: 219–31 25. Staehr-Rye AK, Meyhoff CS, Scheffenbichler FT, et al. High intraoperative inspiratory oxygen fraction and risk of major respiratory complications. Br J Anaesth 2017; 119: 140–9 26. Gupta S, Saint S, Detsky AS. Hiding in plain sight— resurrecting the power of inspecting the patient. JAMA Intern Med 2017; 177: 757–8 27. Levin MA, McCormick PJ, Lin HM, Hosseinian L, Fischer GW. Low intraoperative tidal volume ventilation with minimal PEEP is associated with increased mortality. Br J Anaesth 2014; 113: 97–108 28. Gueldner A, Kiss T, Serpa Neto A, et al. Intraoperative protective mechanical ventilation for prevention of postoperative pulmonary complications: a comprehensive review of the role of tidal volume, positive end-expiratory pressure, and lung recruitment maneuvers. Anesthesiology 2015; 123: 692–713
British Journal of Anaesthesia 119 (1): 18–21 (2017) doi:10.1093/bja/aex206
Methodology in systematic reviews of goal-directed therapy: improving but not perfect S. T. Vistisen1,2,3, E. Keus4 and T. W. L. Scheeren3,* 1
Research Centre for Emergency Medicine, Institute of Clinical Medicine, Aarhus University, Denmark, 2Department of Anaesthesiology & Intensive Care, Aarhus University Hospital, Denmark, 3University of Groningen, University Medical Center Groningen, Department of Anaesthesiology, Groningen, The Netherlands and 4University of Groningen, University Medical Center Groningen, Department of Critical Care, Groningen, The Netherlands
*Corresponding author. E-mail:
[email protected]
There has been a recent tsunami of articles on goal-directed (fluid) therapy, haemodynamic optimization and validation of cardiovascular monitoring devices. This has been followed by a wave of systematic reviews, in particular over the last five years, trying to summarize and derive conclusions and recommendations from many of these studies.1–19 Terminology for systematic
reviews and meta-analyses is frequently used incorrectly. A systematic review refers to a rigorous scientific process of reviewing relevant literature whereas meta-analysis refers to a statistical method of pooling data from multiple studies to derive a summary effect estimate. Any well-conducted scientific process needs to comply with quality standards. Likewise, systematic
Editorials
reviews need to meet quality criteria before qualifying for the highest level of evidence (http://handbook.cochrane.org/, accessed May 29, 2017), including a sufficiently detailed published protocol, full search strategy in multiple databases, outcome selection following GRADE20 (www.gradeworkinggroup.org, accessed May 29, 2017), assessment of risks of bias, assessment of risks of random errors, and reporting results following PRISMA21 and GRADEpro. Producing pooled estimates is always tempting whereas frequently not pooling data because of large heterogeneity might be much wiser. Eventually, only a few systematic reviews qualify for Level 1A evidence. Over the years, the quality of systematic reviews conducted in the field of anaesthesia has improved,22 and in this issue of the British Journal of Anaesthesia, Michard and colleagues23 provide a thoroughly prepared meta-analysis of the clinical impact of perioperative goal-directed therapy when uncalibrated pulse-contour cardiac output monitors are used to titrate treatment. In line with most systematic reviews on goal-directed therapy,1–19 Michard and colleagues23 conclude that goal-directed therapy reduces postoperative complications. In addition, they showed that total fluid volumes were not different between experimental and control groups, although giving fluid is an intervention. While the study’s conclusions should be interpreted in the context of the authors’ insightful statistical presentation and interpretation of the data and the completeness of the data, the study’s conclusions must also be interpreted in the context of possible gaps for a systematic review to qualify for Level 1A evidence. We wish to highlight the findings of Michard and colleagues23 but also discuss methodological issues related to systematic reviews in general.
Fixed-effect vs random-effects models First, and foremost, the authors chose to provide us with results from both fixed-effect and random-effects models, giving us an opportunity to evaluate the impact of choosing either method. The Cochrane Handbook of Systematic Reviews of Interventions clearly states that ‘choosing between a fixed-effect or random-effects model should never be made on the basis of a statistical test for heterogeneity.’24 Fixed-effect models assume one underlying true population effect whereas random-effects models assume an underlying distribution of intervention effects. Randomeffects models inherently down weight large trials in their statistical model (in the present case, the OPTIMISE trial provides 40% of all patients, but contributes 13% weight) and usually have more conservative confidence intervals whereas fixed-effects models have confidence intervals that are usually artificially narrow as they do not include a between-trial variance component. Fixed-effect models are preferred if one or two trials dominate the evidence. Yet, both models are complementary: conclusions can be considered more robust if both models agree. Having both models’ results presented for interpretation is therefore an important strength of the present systematic review, and both models suggest fewer postoperative complications associated with the use of goal-directed therapy. The authors managed to get nearly complete data sets for administered fluid volumes by contacting the investigators of the randomized controlled trials (RCTs) included in the metaanalysis. This effort reduces the possibility of selective reporting regarding infused fluid volumes, because most RCTs evaluated fluid volumes as secondary outcomes and therefore might not have extensively reported on this (e.g. as crystalloid, colloid and total volumes infused). The authors’ effort in retrieving the necessary data has reinforced the conclusion that goal-directed
|
19
therapy does not increase the total amount of fluids infused when using flow-related treatment goals and applying uncalibrated pulse-contour analysis cardiac output monitors.
Risk of bias The authors report statistically significant odds ratios varying from 0.28 to 0.62 for their primary outcome of morbidity (irrespective the risk of bias of the trials). Such large intervention effects are fantastic and compare with milestone discoveries, but could arguably be questioned. Risk of bias is associated with an overestimation of benefit and an underestimation of harm.25 Therefore, conclusions ought to be derived from pooled estimates of trials with low risk of bias with trial sequential analysis (TSA)-adjusted confidence intervals.26 The authors accept only five or six adequate domains to qualify for a trial to have a low risk of bias. They rightfully argue that goal-directed therapy cannot be blinded for the caregiver, but this does not mean that this risk of bias is absent. The analysis shows us what the available data indicate despite some risks of bias, which is related to the quality of the individual RCTs. Still, qualification for the highest level of evidence requires at least phrasing of conclusions in the perspective of the risks of bias of all domains.
Outcome selection Selection and grading of outcomes in systematic reviews should originate from the patient perspective following GRADE.20 Michard and colleagues23 chose morbidity as primary outcome whereas mortality, hospital stay, amounts of fluid and fluid variability were chosen as secondary outcomes. Taking the patient perspective, mortality should always come first, independent of incidence proportions and the power to detect statistically significant differences. It is, however, unrealistic to expect impact on mortality without many more adequately powered trials such as OPTIMISE 2 (http://optimi seii.org was accessed May 29, 2017) and others.27 Amounts of fluid and fluid variability (and even hospital stay) are surrogate outcomes. Morbidity is clearly important from the patient’s perspective, but there is variability in definitions and measurements across the included studies. The COMET initiative (http://www.comet-initiative. org, accessed May 29, 2017) has been launched to agree on a standard minimum set of clearly defined outcomes for specific conditions and interventions to facilitate exchange of information.
Power of meta-analysis estimates Meta-analyses increase the power and precision of estimated intervention effects. Still, the best available evidence might not be synonymous with sufficient evidence. A sample size calculation of a randomized trial is typically based on a hypothesis, specified by a control event rate, a relative risk reduction (RRR), and maximum type I and II errors. Any pooled effect of a metaanalysis should be interpreted in the perspective of such a hypothesis, and confidence intervals should additionally be adjusted (i.e. increased) for heterogeneity. Both positive and neutral (or negative) findings can be because of chance as conventional meta-analyses do not take into account the amount of evidence in relation to a hypothesis. TSA combines information size estimation for a meta-analysis (cumulated sample size of included trials) with an adjusted threshold for statistical significance in the cumulative meta-analysis, and is able to assess risks of type I and type II errors. In TSA, the addition of each trial in a cumulative meta-analysis is regarded as an interim
20
|
Editorials
meta-analysis and helps to clarify whether additional trials are needed or not. The control event rate and the RRR of the hypothesis need to be realistic. Use of TSA is a major strength of the analysis by Michard and colleagues,23 but TSA varies with the conditions and including only low risk of bias trials with a smaller (more realistic) RRR would be more informative.
Heterogeneity By pooling RCTs with patients having emergency surgery and RCTs with patients having elective surgery, the authors by definition allowed for clinical heterogeneity in their data. This is reflected by a wide range of morbidity proportions in both the control group (ranging from 12 to 100%) and the intervention group (from 8 to 80%). While the amount of statistical heterogeneity should never influence the choice for the pooling model, perhaps the authors should have considered not pooling data in the presence of 65% (or even 99%) heterogeneity in the forest plots. Regardless of these methodological considerations, the current systematic review has important clinical implications. It provides us with the best currently available evidence on the topic, as follows: 1. Although the accuracy of uncalibrated pulse-contour cardiac output monitors has been questioned, this systematic review shows that they are clinically useful when used in combination with a goal-directed therapy protocol to provide a benefit for patients. 2. Although it has been suggested that stroke volume and cardiac output optimization lead to excessive fluid administration, this systematic review shows that this is not the case. 3. It has been assumed that goal-directed fluid therapy reduces variability in fluid management; however, this systematic review shows that this variability is not reduced.
Authors’ contributions Wrote the first draft: S.T.V. Initial discussion on the paper’s content, developed and agreed on the final version: all authors.
Declaration of interest The authors declare no conflicts of interest.
Funding This work was not supported financially.
References 1.
2.
3.
Arulkumaran N, Corredor C, Hamilton MA, et al. Cardiac complications associated with goal-directed therapy in high-risk surgical patients: a meta-analysis. Br J Anaesth 2014; 112: 648–59 Aya HD, Cecconi M, Hamilton M, Rhodes A. Goal-directed therapy in cardiac surgery: a systematic review and metaanalysis. Br J Anaesth 2013; 110: 510–7 Benes J, Giglio M, Brienza N, Michard F. The effects of goaldirected fluid therapy based on dynamic parameters on post-surgical outcome: a meta-analysis of randomized controlled trials. Crit Care 2014; 18: 584
4.
5.
6.
7.
8.
Berger MM, Gradwohl-Matis I, Brunauer A, Ulmer H, Dunser MW. Targets of perioperative fluid therapy and their effects on postoperative outcome: a systematic review and metaanalysis. Minerva Anestesiol 2015; 81: 794–808 Cecconi M, Corredor C, Arulkumaran N, et al. Clinical review: goal-directed therapy-what is the evidence in surgical patients? the effect on different risk groups. Crit Care 2013; 17: 209 Corcoran T, Rhodes JE, Clarke S, Myles PS, Ho KM. Perioperative fluid management strategies in major surgery: a stratified meta-analysis. Anesth Analg 2012; 114: 640–51 Egal M, Erler NS, de Geus HR, van Bommel J, Groeneveld AB. Targeting oliguria reversal in goal-directed hemodynamic management does not reduce renal dysfunction in perioperative and critically ill patients: a systematic review and meta-analysis. Anesth Analg 2016; 122: 173–85 Giglio M, Dalfino L, Puntillo F, Rubino G, Marucci M, Brienza N. Haemodynamic goal-directed therapy in cardiac and vascular surgery. A systematic review and meta-analysis. Interact Cardiovasc Thorac Surg 2012; 15: 878–87
9.
Giglio M, Manca F, Dalfino L, Brienza N. Perioperative hemodynamic goal-directed therapy and mortality: a systematic review and meta-analysis with meta-regression. Minerva Anestesiol 2016; 82: 1199–213 10. Grocott MP, Dushianthan A, Hamilton MA, et al. Perioperative increase in global blood flow to explicit defined goals and outcomes after surgery: a cochrane systematic review. Br J Anaesth 2013; 111: 535–48 11. Li P, Qu LP, Qi D, et al. Significance of perioperative goal-directed hemodynamic approach in preventing postoperative complications in patients after cardiac surgery: a meta-analysis and systematic review. Ann Med 2017; 49: 343–51 12. Osawa EA, Rhodes A, Landoni G, et al. Effect of perioperative goal-directed hemodynamic resuscitation therapy on outcomes following cardiac surgery: a randomized clinical trial and systematic review. Crit Care Med 2016; 44: 724–33 13. Ripolles-Melchor J, Espinosa A, Martinez-Hurtado E, et al. Perioperative goal-directed hemodynamic therapy in noncardiac surgery: a systematic review and meta-analysis. J Clin Anesth 2016; 28: 105–15 14. Rollins KE, Lobo DN. Intraoperative goal-directed fluid therapy in elective major abdominal surgery: a meta-analysis of randomized controlled trials. Ann Surg 2016; 263: 465–76 15. Som A, Maitra S, Bhattacharjee S, Baidya DK. Goal directed fluid therapy decreases postoperative morbidity but not mortality in major non-cardiac surgery: a meta-analysis and trial sequential analysis of randomized controlled trials. J Anesth 2017; 31: 66–81 16. van der Zee EN, Egal M, Gommers D, Groeneveld AB. Targeting urine output and 30-day mortality in goal-directed therapy: a systematic review with meta-analysis and metaregression. BMC Anesthesiol 2017; 17: 22 17. Wilms H, Mittal A, Haydock MD, van den Heever M, Devaud M, Windsor JA. A systematic review of goal directed fluid therapy: rating of evidence for goals and monitoring methods. J Crit Care 2014; 29: 204–9 18. Yuan J, Sun Y, Pan C, Li T. Goal-directed fluid therapy for reducing risk of surgical site infections following abdominal surgery—a systematic review and meta-analysis of randomized controlled trials. Int J Surg 2017; 39: 74–87 19. Pearse RM, Harrison DA, MacDonald N, et al. Effect of a perioperative, cardiac output-guided hemodynamic therapy algorithm on outcomes following major gastrointestinal
Editorials
20.
21.
22.
23.
surgery: a randomized clinical trial and systematic review. JAMA 2014; 311: 2181–90 Guyatt GH, Oxman AD, Kunz R, et al. What is “quality of evidence” and why is it important to clinicians? BMJ 2008; 336: 995–8 Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement. Int J Surg 2010; 8: 336–41 Hall AM, Lee S, Zurakowski D. Quality assessment of metaanalyses published in leading anesthesiology journals from 2005 to 2014. Anesth Analg 2017; 124: 2063–7 Michard F, Giglio M, Brienza N. Perioperative goal directed therapy with uncalibrated pulse contour methods: impact on fluid management and post-operative outcome. Br J Anaesth 2017; 119: 22–30
|
21
24. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 2011. Available from www.handbook.cochrane.org (accessed 29 May 2017) 25. Savovic J, Jones HE, Altman DG, et al. Influence of reported study design characteristics on intervention effect estimates from randomized, controlled trials. Ann Intern Med 2012; 157: 429–38 26. Brok J, Thorlund K, Gluud C, Wetterslev J. Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol 2008; 61: 763–9 27. Montenij L, de Waal E, Frank M, et al. Influence of early goaldirected therapy using arterial waveform analysis on major complications after high-risk abdominal surgery: Study protocol for a multicenter randomized controlled superiority trial. Trials 2014; 15: 360
