Early PEEP Linked to Mortality in Sepsis-Associated Myocardial Injury

Key Takeaways

• In a retrospective cohort of 3,421 mechanically ventilated adults with sepsis-associated myocardial injury, higher early TWA PEEP was associated with higher intensive care unit and in-hospital mortality.
• A 285-patient external cohort showed the same directional pattern between higher early TWA PEEP and mortality.
• Spline analyses suggested a generally increasing association without a clear threshold or plateau, although relatively few patients had early TWA PEEP above 15 cmH2O, so estimates at the upper end were less precise.
• Prespecified subgroup and sensitivity analyses did not materially change the direction of the mortality association.

In sepsis-associated myocardial injury, ventilatory support and hemodynamic vulnerability can pull in opposite directions because positive end-expiratory pressure (PEEP) may improve oxygenation while also altering preload and afterload in patients with limited myocardial reserve. For mechanically ventilated adults early in critical illness, the unresolved question is whether greater early PEEP exposure tracks with worse outcomes in this subgroup or mainly reflects the severity of concurrent lung and circulatory failure. To examine that question, early airway pressure exposure was quantified soon after invasive ventilation began and related to short-term mortality.

In a retrospective observational cohort study, investigators used the Medical Information Mart for Intensive Care IV (MIMIC-IV) database for derivation and externally validated the findings at Liuzhou Workers’ Hospital in Ge et al. in Frontiers in Cellular and Infection Microbiology. Adults had sepsis-associated myocardial injury and received invasive mechanical ventilation (IMV) within 24 hours of intensive care unit (ICU) admission. Myocardial injury was defined by the first cardiac troponin T (cTnT) measured within that window above the assay-specific 99th percentile upper reference limit, set at more than 0.01 ng/mL in MIMIC-IV and more than 14 pg/mL in the external cohort. Exposure was early time-weighted average (TWA) PEEP during the first 48 hours after IMV initiation, estimated from intermittently charted ventilator settings using last observation carried forward. The derivation cohort included 3,421 patients and the external cohort 285; the primary outcome was ICU mortality, the secondary outcome was in-hospital mortality, and the main analysis used generalized propensity score weighting with covariate-adjusted logistic regression.

Higher early TWA PEEP was associated with higher mortality in the derivation cohort: each 1 cmH2O increase was linked to intensive care unit mortality odds ratio 1.14 (95% CI 1.10-1.18; P<0.001) and in-hospital mortality odds ratio 1.11 (95% CI 1.07-1.15; P<0.001). Restricted cubic spline analysis suggested a generally increasing risk pattern across the observed PEEP range without a clear threshold or plateau, although relatively few patients had early TWA PEEP above 15 cmH2O, so estimates at the upper end were less precise. The external cohort was directionally consistent, with adjusted odds ratios of 1.31 (95% CI 1.08-1.59; P=0.006) for ICU mortality and 1.29 (95% CI 1.07-1.57; P=0.009) for in-hospital mortality. Prespecified subgroup analyses showed no significant interaction, and alternate exposure windows and model specifications remained directionally consistent.

Because the analysis was observational, residual confounding and confounding by indication remained possible after adjustment. Higher early TWA PEEP may have tracked with more severe hypoxemia, poorer recruitability, or greater cardiopulmonary instability rather than isolating a direct harmful effect of airway pressure itself. The exposure measure was reconstructed from intermittently charted PEEP values with last observation carried forward, so TWA PEEP should be interpreted as an approximation of cumulative early exposure rather than a continuous physiologic trace. Several physiologic variables were not uniformly available, including echocardiography, respiratory compliance, recruitment maneuvers, fluid balance, and dynamic volume responsiveness. The external cohort was smaller and less precise, and sepsis there was identified by International Classification of Diseases (ICD) codes rather than the derivation cohort’s clinical definition. For North American readers, the signal comes from a U.S.-derived critical care database paired with a Chinese hospital cohort, which broadens relevance without establishing universal applicability.

Among adults with sepsis-associated myocardial injury receiving IMV, higher early TWA PEEP was consistently associated with higher ICU and in-hospital mortality in both the derivation and external validation cohorts. The authors framed these findings as associative rather than causal and called for prospective, physiology-informed studies to test whether different PEEP strategies change outcomes in this population.

Clinician Questions

How was early PEEP exposure defined in ventilated patients with sepsis-associated myocardial injury?

Early exposure was defined as the time-weighted average PEEP during the first 48 hours after IMV initiation, calculated from charted ventilator settings with last observation carried forward between measurements and through the end of the 48-hour window for patients who were extubated or died earlier. The measure was intended to capture cumulative early airway pressure exposure rather than a continuously recorded tracing.

Which patients do these PEEP-mortality findings apply to?

These findings apply to adults in the ICU with sepsis-associated myocardial injury who started IMV within 24 hours of ICU admission and had an early cTnT value above the assay-specific 99th percentile upper reference limit, after excluding major alternative causes of troponin elevation or primary cardiac disease such as acute coronary syndrome, cardiac arrest, cardiomyopathy, valvular or inflammatory cardiac disease, cardiac surgery, aortic dissection, and end-stage renal disease. The cohorts came from MIMIC-IV and Liuzhou Workers’ Hospital, so the signal pertains to this mechanically ventilated population rather than to all patients with sepsis or all ventilated ICU patients.

Did the study identify a PEEP threshold at which mortality risk changed abruptly in sepsis-associated myocardial injury?

No clear threshold emerged. The spline analyses showed a generally increasing association across the observed PEEP range without a plateau or abrupt inflection, although the in-hospital mortality curve showed modest curvature.

What uncertainty remained after adjustment for severity and validation in a second cohort?

The association remained directionally consistent after weighting, subgroup analyses, and multiple sensitivity analyses, but the design could not eliminate residual confounding or confounding by indication. Exposure was estimated from intermittent charting rather than continuous measurement, and the smaller external cohort supported direction more than precise replication of effect size.