A roadmap to optimize intraoperative mechanical ventilation

Low tidal volume ventilation has been a critical care mainstay for almost a quarter century and represents a triumph of modern medical science. From the first clinical descriptions of acute respiratory distress syndrome (ARDS) in 1967¹ to the recognition that lungs with ARDS are physiologically “smaller” due to heterogeneous injury (and thus prone to over inflation),² and publication of randomized clinical trials demonstrating better outcomes in patients that received lower tidal volumes,³ the chain of evidence linking large tidal volumes to worsened lung injury in patients with ARDS is as clear as any in critical care medicine.⁴

Whether similar mechanical ventilation strategies improve outcomes in perioperative patients is a logical next question and one that deserves an empiric test. Unlike patients with ARDS, those undergoing surgery generally begin with uninjured lungs, are less likely to have risk factors for acute lung injury, and are exposed to positive pressure ventilation for shorter periods. In light of currently understood mechanisms of volume-induced injury,⁴ these factors should result in a smaller potential benefit of low tidal volume ventilation in patients undergoing surgery than in those with ARDS.

Although the clinical research question (“Do intraoperative ventilator settings affect perioperative outcomes?”) seems straightforward, the answer has been surprisingly elusive. Conflicting clinical trials^{5, 6} and guarded editorials⁷ dot the literature. Nevertheless, declining real-world intraoperative tidal volumes,⁸ recent reviews,⁹ and consensus guideline recommendations¹⁰ suggest increasing agreement that targets for mechanical ventilation can affect perioperative outcomes. Accurately characterizing the benefits (if any) of specific intraoperative mechanical ventilation strategies is highly relevant as an estimated 230 million anesthetics are administered worldwide each year.¹¹

In this issue of Anesthesia & Analgesia, Bolther et al. present a comprehensive meta-analysis of trials of intraoperative mechanical ventilation “targets and strategies.”¹² From a database of 534 articles, the authors identify 63 trials of intraoperative ventilator targets and perform meta-analyses of the relationships between delivered tidal volumes, PEEP, lung protective strategies (a combination of higher PEEP and low tidal volume), recruitment maneuvers, ventilator modes, end-tidal carbon dioxide levels (ETCO₂) and I:E ratios and postoperative pulmonary complications (PPCs). The work includes GRADE (grading of recommendations, assessment, development, and evaluations) level analysis of certainty, subgroup analyses for different surgical subtypes, and assessments of the risks of study and publication bias.

The results are comprehensive, detailed, and sobering in what they suggest about the state of knowledge regarding intraoperative mechanical ventilation in 2022. The authors find no reduction in PPCs with trials of tidal volumes or PEEP alone but do observe a decrease in PPCs with trials that studied different combinations of tidal volume and PEEP. They find less atelectasis with recruitment maneuvers, insufficient data to inform the use of specific ventilator modes, ETCO₂ and I:E ratios, and low to moderate GRADE-rated certainty of evidence for all comparisons.

Should anesthesiologists target specific intraoperative ventilation endpoints to optimize perioperative outcomes in their patients? Unfortunately, Bolther et al. (appropriately) conclude that existing literature can identify no meaningful ventilatory targets. For ventilator mode, ETCO2, and I:E ratio, the authors find little evidence of benefit and low confidence in those findings due to inadequate numbers of trials and patients.

The study’s findings regarding the more widely studied effects of PEEP and low tidal volume are more difficult to interpret. For example, Bolther et al. found that strategies varying both PEEP and tidal volume dramatically reduced pulmonary complications, but that varying either PEEP or tidal volumes separately had no effect, suggesting an extremely potent synergy between the two interventions. That the authors also found large effects of varying both PEEP and tidal volume on pulmonary complications, atelectasis, and postoperative mechanical ventilation but not on mortality or hospital length of stay also raises questions about the relevance of a perioperative complication that does not meaningfully change the hospital course.

In their discussion, the authors identify several aspects of their meta-analysis that explain their low confidence in their results. The small sample size in most trials (median = 66 for all trials) increased the risk of bias,¹³ particularly delta inflation bias,¹⁴ in which overestimations of treatment effect during study planning leads to underestimated sample sizes and underpowered studies. Small sample sizes in most studies also exaggerated the relative impacts of the few larger studies included in the analysis. For example, the 2020 Karalapillai trial⁵ found no benefit of 6 cc/kg tidal volumes (vs. 10 cc/kg) in 1,236 patients. The trial was included in the “tidal volume” comparison but not in the “lung protective” one because PEEP was the same in both treatment and control arms. As one of the largest in the cohort, including this trial in either comparison would have biased those results towards the null. While it is possible that adding a PEEP intervention to the Karalapillai trial might have unlocked the 70% reduction in pulmonary complications suggested by this meta-analysis for combined PEEP and tidal volume interventions, existing evidence¹⁵ would argue that such a possibility is unlikely. While not explicitly addressed by the authors, another significant methodological challenge was considerable variation in the types of patients, procedures, and ventilator strategies used. For, example, trials in the PEEP group included both fixed and titrated comparisons.

Despite 18 years, 63 trials, and thousands of enrolled patients, the optimal approach to perioperative mechanical ventilation thus remains uncertain. Although an initial reflex might be to conduct more trials (and more meta-analyses) it is unclear whether the wide variety of ventilator interventions, measured outcomes, and low targeted enrollments (most less than 100) in the 99 planned, ongoing, and unpublished trials of intraoperative mechanical ventilation identified by the study authors will add certainty or further obfuscate the question.

How might we optimize clinical research to clarify the role of perioperative mechanical ventilatory targets going forward? By shining a light on some of the challenges encountered in past trials, Bolther et al. also suggest ways forward. First, peer reviewers’ and journal editors’ adherence to trial reporting guidelines may address design, conduct, and reporting challenges highlighted by the authors’ low level of confidence in the trials they analyzed. For trialists, unmeasured heterogeneity in patients’ risk of studied outcomes threatens the validity of individual study results. Better defining the study population is one way to reduce the impact of study heterogeneity. Because risk factors for postoperative pulmonary complications are well characterized,¹⁶ investigators can use these factors to stratify randomization or analyses among patients with similar baseline risks. Furthermore, heterogeneity in interventions studied and outcomes measured complicates the conduct of future meta-analyses. The use of consistently defined patient populations and interventions may also help, as would explicit recognition of the need to generate consistent, comparable literature. Consensus definitions of intraoperative ‘lung protective ventilation’ may also be developed to facilitate comparisons between studies and address specific perioperative scenarios, such as single lung ventilation, abdominal insufflation, and use of neuromuscular blocking drugs. With respect to outcomes, wider use of consensus definitions of postoperative pulmonary complications¹⁷ would facilitate comparisons between trials. One complementary approach is to use prespecified hierarchical outcomes that weight, for example, the need for postoperative mechanical ventilation as more impactful than radiographic evidence of atelectasis.

Second, clinical trialists can also adopt innovative trial design and analysis methods translated from other specialties and disciplines. Determining which combination of ventilatory targets impact outcomes (such as PEEP and tidal volume) requires a factorial trial design in which patients are randomized to multiple interventions alone and in combination. Adaptive trial designs, which use prespecified interim analyses to inform protocol changes (such as stopping enrollment in a treatment arm unlikely to be effective), may improve trial efficiency.

Last, research leaders, funding agencies, and professional societies might consider whether the current clinical research infrastructure is optimally designed to answer this deceptively complex clinical question. Bolther et al. observe that the current approach, which involves individual researchers testing multiple hypotheses, has generated a large number of incompatible research results that, taken together, are insufficient to inform practice. We note that history provides a similar lesson: multiple smaller trials of smaller tidal volumes in ARDS with conflicting results^18–20 had been published before the multicenter ARDSNet collaboration produced the pivotal prospective randomized trial in 2000.³ That group, now the NIH NHLBI-funded PETAL Network, has become the paradigm organization for conducting multicenter trials of interventions for ARDS in the United States. Such coordinated trial networks bring both state-of-the-art expertise in study design and analysis and operational resources to optimize study conduct. An example of a successful large-scale perioperative research collaboration is the Multicenter Perioperative Outcomes Group. More recently, this group has proposed the development of a clinical trials network for perioperative interventions supported by key institutional stakeholders (IARS, FAER, AUA).

Despite biological plausibility and strong evidence of efficacy in patients with ARDS, the relevance of intraoperative mechanical ventilatory targets has remained frustratingly unclear. Bolther and colleagues’ comprehensive assessment of the state of the evidence suggests that, despite two decades of work and real-world adoption by clinicians, available research remains insufficient to inform practice. As these questions persist into their third decade, we look forward to innovative approaches to study design and conduct and research coordination and support that can shed more light on these questions while laying the groundwork for high-quality clinical research in anesthesiology.

Glossary of terms:

ARDS

Acute Respiratory Distress Syndrome

PPC

Postoperative Pulmonary Complications

PEEP

Positive End-expiratory pressure

ETCO2

End-tidal carbon dioxide

I:E

Inspiratory to expiratory

GRADE

Grading of recommendations, assessment, development, and evaluations

NIH

National Institutes of Health

NHLBI

National Heart, Lung, and Blood Institute

PETAL

Prevention and Treatment of Acute Lung Injury

IARS

International Anesthesia Research Society

FAER

Foundation for Anesthesia Education and Research

AUA

Association of University Anesthesiologists