In Search of the Optimal PEEP Strategy

Aaron B. Holley, MD


August 26, 2022

Since the advent of invasive positive pressure ventilation, intensivists have been arguing over how to optimize settings. For patients with moderate to- severe acute respiratory distress syndrome (ARDS), the lung is particularly fragile, so this group is the subject of much study. The ARDSNet trial resulted in low tidal volume ventilation becoming the standard of care for ARDS more than two decades ago. Since then, the academic critical care medicine community has successfully designed multiple complex randomized controlled trials (RCTs) to study the effects of mechanical ventilation settings on outcomes. Many have focused on the approach to setting positive end-expiratory pressure (PEEP).

Unfortunately, PEEP management in patients with ARDS remains a muddled picture. We know the physiology, and if you've worked in a critical care unit, you've heard it explained before — avoid ventilator-induced lung injury; stay away from upper and lower inflection points on the pressure-volume curve; keep the driving and plateau pressures less than 30 and 15 cm H2O, respectively; and so on. Intensivists can talk a good physiologic game, but can we translate that knowledge to setting PEEP at the bedside and improving outcomes?

Sort of.

Older trials implied that a "higher" PEEP was superior, but meta-analyses were inconsistent in their conclusions. More recently, RCTs have failed to show benefit from using recruitment maneuvers or esophageal manometry. In fact, recruitment maneuvers were found to be harmful and are all but dead as a clinical concept. Admittedly, I'm biased here as I never used them in my practice anyway.

Enter Network Meta-analysis

If knowledge of physiology, use of esophageal manometry, and excellent RCTs with meta-analyses don't get us an answer, maybe we're looking in the wrong place. Specifically, we're asking the wrong people. We don't need clinicians, physiologists, or clinical trialists; we need a statistician and a librarian to do a network meta-analysis (NMA).

A recent NMA concluded that a higher PEEP strategy without recruitment maneuvers has a 99% posterior probability of improving mortality compared with a lower PEEP strategy. Several additional pairwise comparisons were also evaluated, including different combinations of high and low PEEP strategies with and without recruitment maneuvers and PEEP guided by esophageal manometry. The headline, though, was that higher PEEP without recruitment maneuvers is the way to go. Perhaps a co-headline is that recruitment maneuvers are harmful, with prolonged recruitment maneuvers being particularly dangerous.

Have the investigators solved the "optimal PEEP" mystery? They claim they were able to isolate the positive impact of a high PEEP strategy from the negative effects from recruitment maneuvers. Many of the RCTs that looked at high PEEP in patients with ARDS included recruitment maneuvers as part of their strategy. The authors of the new NMA evaluated the high PEEP strategy without recruitment maneuvers and found a mortality benefit.

How did they do this? And what is NMA? I'm glad you asked. It's kind of a standard meta-analysis on steroids. A standard meta-analysis combines arms from RCTs (or other study types depending on design and inclusion criteria) and compares them with each other directly. So, experimental treatment A is compared with control B in every RCT analyzed, then all the As are combined and compared with all the Bs to optimize sample size. It's a proven approach that represents high-level evidence when it's done carefully and presented well.

For an NMA, the authors incorporate indirect as well as direct comparisons. If treatment A is compared with control B in a study and treatment C is compared with control B in a different study, NMA allows for a comparison of A vs C. Oftentimes comparisons are both direct and indirect because the opportunities for both exist and combining them will optimize sample size. This is all accepted practice. NMA represents a powerful and important tool.

A Messy Business

It's not perfect, though. As a mentor of mine told me when we published the equivalent of an NMA more than 15 years ago, "Meta-analysis is a messy business." Although fancy statistical tests for bias, inhomogeneity, transitivity, and inconsistency are used to assure us that the methods are valid, they're not perfect. Important design and qualitative differences between studies can skew findings and interpretation even when all the statistics indicate that they shouldn't. This was highlighted in a recent editorial in the CHEST journal, and an editorial accompanying the PEEP NMA demonstrates this phenomenon nicely.

In fairness, the PEEP NMA is incredibly well done and I believe it moves the debate forward. The authors do an outstanding job of presenting the data and disclosing limitations. Despite the reservations of the editorialists, I will be moving toward a higher PEEP (without recruitment maneuvers) for patients with moderate to severe ARDS.

The higher PEEP message is all the more important given that the LUNGSAFE investigators and others have found that the default PEEP for ARDS is actually low.

The PEEP NMA editorial talks eloquently about personalizing PEEP, which is great, but that's easier said than done at the bedside. So no, it'll take more than a good librarian and statistician to identify optimal PEEP for your patient. However, all things being equal, you should probably be aiming higher than you are.

As for meta-analyses and NMA, they will continue to be important tools. They require rigid reporting, though, and outside those published in mainstream journals, I'd be skeptical of any meta-analysis or NMA with findings that overpromise or buck the conventional wisdom. At the risk of overgeneralizing, I'd advise caution before changing clinical management based on a meta-analysis published in a lower-tier journal. At a minimum, make sure the meta-analysis/NMA you're reading mentions the PRISMA guidelines.

Aaron B. Holley, MD, is an associate professor of medicine at Uniformed Services University and program director of pulmonary and critical care medicine at Walter Reed National Military Medical Center. He covers a wide range of topics in pulmonary, critical care, and sleep medicine.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.