Goligher et al 2014: Oxygenation Response to PEEP Predicts Mortality in ARDS

Goligher EC, et al. Oxygenation Response to Positive End-Expiratory Pressure Predicts Mortality in Acute Respiratory Distress Syndrome. A Secondary Analysis of the LOVS and ExPress Trials. Am J Respir Crit Care Med. 2014 Jul 1;190(1):70-6. doi: 10.1164/rccm.201404-0688OC. PubMed PMID: 24919111.

• In ARDS patients, is a there an association between the physiological or oxygenation response to increased PEEP and patient mortality?

TYPE OF STUDY

• Secondary analysis of Lung Open Ventilation Study (LOVS) data set with corroboration from the ExPress trial data set; these were two MCRCTs comparing higher levels of PEEP with standard ventilation strategies
• LOVS (2008): n = 983
  • 30 centers in Canada, Australia and Saudi Arabia
  • recruitment 2000-2006
• ExPress (2008): n = 767
  • 37 centers in France
  • recruitment 2002-2005

POPULATION

• ARDS patients recruited to the LOVS and Express trials

From the LOVS trial the population characteristics were:

• mean age: 55.7y
• female sex: 40.1%
• mean APACHE II score: 25.4
• sepsis diagnosis: 47.0%
• ARDS severity: mild 15%, moderate 64%, severe 21%

INTERVENTIONS/ COMPARISONS

• There was no intervention as this was a secondary analysis.
• Goligher et al extracted information from the existing data sets to see if there was an association between change in PEEP (specifically increased PEEP) and it’s effect on oxygenation (∆ P/F), with mortality.
• To understand the secondary analysis, an understanding of the preceding LOVS and ExPress interventions is necessary. In both trials, patients were randomised to either a control or intervention ventilation strategy (outlined below).

• Goligher et al compared patients in LOVS based on the direction of the initial PEEP change after randomisation (NOT the ventilation strategy vs the control group), resulting in significant crossover between the groups (24% of LOVS intervention patients had an initial decrease in PEEP and approximately 20% of control patients had in initial increase in PEEP).
• 63 patients from LOVS were also excluded from analysis due to missing data, resulting in:
  • Increased PEEP: n = 431 (461)
  • Decreased or unchanged PEEP: n = 489 (513)
• Goligher et al also hypothesised that the association between physiological response to ∆ PEEP and mortality would be stronger in patients with more severe ARDS, and thus further subdivided their analysis for patients with PaO2/FiO2 < 150 mmHg.
• Sensitivity analyses were conducted on subsets of the data, model effects were unchanged.
• Similar analyses were conducted on the ExPress data sets.

OUTCOMES

• Oxygenation response to increased PEEP was extremely variable and only weakly associated with the baseline severity of hypoxaemia and the magnitude of the PEEP change.
• Based on the LOVS data, after increase in PEEP, an increase in the P/F ratio of >25 mmHg was strongly associated with a lower mortality (OR 0.80, 95% CI 0.72-0.89). This association was stronger in patients with more severe ARDS (OR 0.73, 95% CI 0.63 – 0.84).
• This result was corroborated in the ExPress trial; (OR 0.92, 95% CI 0.86-1.00) and this association was also stronger in patients with more severe ARDS (OR 0.86, 95% CI 0.78 – 0.96). In ExPress the P/F response after PEEP reduction was also associated with lower mortality, particularly in patients with less severe ARDS (OR 0.80, 95% CI 0.65 – 0.98).
• In analysis of the combined data sets (n = 1732), the oxygenation response after PEEP modification was associated with mortality only in the subgroup of patients with more severe ARDS who were subjected to an increase in PEEP (OR 0.80, 95%CI 0.73 – 0.88, p value = 0.03)

Commentary

• Baseline severity of hypoxaemia and magnitude of PEEP change were only weakly associated with positive oxygenation response – therefore have limited utility as predictors of responders vs non responders. This makes sense given the heterogeneity of ARDS patients.
• Patients with more severe ARDS appeared to benefit more from the higher PEEP strategies as identified in both studies and in the combined data analysis.
  • The improved oxygenation in response to increased PEEP is a marker for lung recruitability; and identifies those patients (‘responders’) that will go on to accrue a mortality benefit from higher PEEP ventilation strategies (causal hypothesis).
  • Alternatively the patients who showed a positive oxygenation response to increased PEEP, represent those who would have better outcomes no matter what ventilation strategy was employed (prognostic hypothesis).
• Goligher et al argues in support of the causal hypothesis stating that patients with less severe ARDS as shown in ExPress (which included patients with the more liberal cut off of P/F <300), had a mortality benefit from reduced or unchanged PEEP – suggesting that increased PEEP is not necessary and potentially harmful in those with more normal lungs.
• Strong points of this study:
  • Produces statistically significant results that can be used as a platform or component of further research into this area.
  • results were corroborated with second data set and sensitivity analyses.
  • Draws from a large data set.
  • Patient groups were similar at baseline in terms of major confounders, despite cross over between the groups.
  • Identifies it’s own limitations in supporting but by no means confirming their proposed hypothesis.

Criticisms

• Post hoc analysis of existing data sets are merely hypothesis generating – not a prospective analysis of an intervention; the outcome may be due to chance.
• No sample size analysis undertaken
• Data was missing and patients exclude from analysis
• Looking at the crude mortality data for LOVS (31% vs 54%), the lower mortality is arguably not better than standard care with low tidal volumes.
• Many of the 95% confidence intervals approach 1.0 (particularly in the ExPress trial), so the magnitude of the benefit many be very small.
• Patients were not analysed in the groups they were randomised to, so the study doesn’t provide information on which ventilation strategy is better, or if recruitment maneuvers are of benefit.
• Oxygenation is influenced by many factors unrelated to lung recruitability (cardiac output, oxygen consumption, extrapulmonary shunt, efficiency of hypoxic pulmonary vasoconstriction and the FiO2 setting for example) – therefore it is open to confounding.
• No comment on change in lung compliance as a marker of lung recruitment. From the PHARLAP pilot study – there was a 2-3 day delay in improvement of lung compliance following institution of higher PEEP.
• No comment on driving pressures used, an area of current research. Given the plateau of <40cm H2O in LOVS interventional arm, were they using higher driving pressures to achieve tidal volumes and what impact did this have on outcomes?
• Doesn’t provide information on adverse effects/safety of the intervention.

• There is an association between the oxygenation response to PEEP and hospital mortality, but it is uncertain if this because those pateints would do better any or if it is causally related to higher PEEP settings..
• In clinical practice, if a patient fails to show an improvement in oxygenation following an increase in PEEP this is a poor prognostic sign. On the other hand, those who respond with an improvement in oxygenation may be more likely to respond to higher PEEP ventilation strategies.