Acidemia predicts mortality independently of lactate levels in patients after cardiac arrest

TL;DR

Low blood pH after cardiac arrest predicts higher mortality, even when accounting for lactate and CO2 levels, making it a useful early marker for risk assessment.

Contribution

Acidemia is shown to independently predict mortality after cardiac arrest, beyond lactate and paCO2 levels.

Findings

Admission pH levels predicted in-hospital mortality with an area-under-curve of 0.75.

Severe acidemia (pH ≤ 7.2) was significantly associated with higher mortality independent of lactate, paCO2, and CPR characteristics.

The association between acidemia and mortality persisted in subgroup analyses and external validation.

Abstract

We examined whether post-cardiac arrest acidemia is associated with 30-day mortality and neurological outcomes among hospital survivors, independent of lactate and partial arterial carbon dioxide pressure (paCO2) levels, in patients after cardiopulmonary resuscitation (CPR).

The predictive value of acidemia for in-hospital mortality was analyzed retrospectively in 742 non-traumatic cardiac arrest patients admitted to a German high-volume tertiary center using receiver operating characteristic (ROC) analysis. Patients were stratified using the ROC-derived (Youden-optimal) pH cut-off, and 30-day mortality was compared across strata. Cox regression assessed the association between severe acidemia (pH ≤ 7.2, binary) and mortality and examined its consistency across prespecified subgroups (age, sex, cardiac arrest type, lactate, and paCO2). Preliminary findings were externally validated in a database containing over 2000 patients from multiple intensive care units (eICU database).

Admission pH levels predicted in-hospital mortality (area-under-curve 0.75, p < 0.0001). The ROC-derived Youden-optimal threshold was pH 7.207; for clinical interpretability, this value was rounded to pH 7.2 and used as the cut-off for severe acidemia. Patients with severe acidemia exhibited substantially higher in-hospital and 30-day mortality. Among hospital survivors, no statistically significant association between admission pH and neurological outcome was observed. In Cox models, pH ≤ 7.20 remained significantly associated with mortality independent of lactate, paCO2, arrest type, age, and sex. In the validation eICU cohort, pH ≤ 7.2 accurately stratified cardiac-arrest patients with a higher mortality risk. This association persisted in Cox regression analyses of subgroups stratified by lactate and paCO2 levels (all p < 0.0001), as proxies for systemic hypoperfusion and ventilation.

Post-cardiac-arrest acidemia is associated with higher mortality independently of lactate, ventilation, or CPR characteristics. Among hospital survivors, admission pH was not significantly associated with neurological outcome. These findings support pH as an early marker for mortality risk stratification after cardiac arrest, to be interpreted in the context of multimodal prognostication.

Figures