Positive end-expiratory pressure titration in COVID-19 acute respiratory failure: electrical impedance tomography vs. PEEP/FiO2 tables

To the Editor, Hypoxemic acute respiratory failure (hARF) secondary to COVID-19 presents with heterogeneous features depending on several determinants, such as the extent of intravascular microthrombosis, superinfections, and other complications [1, 2]. The easiest approach for setting positive end-expiratory pressure (PEEP) and inspiratory oxygen fraction (FiO2) is using PEEP/FiO2 tables [3, 4]. However, because the magnitude of lung recruitability is variable, personalizing PEEP would be desirable [1]. Electrical impedance tomography (EIT) offers this opportunity by bedside estimating both alveolar collapse and lung overdistension throughout a decremental PEEP trial [5]. This investigation (Ethics Committee approval: Ref: 4853/AO/20-AOP2012) aims to assess the agreement between EIT-based PEEP values and those recommended by the higher and lower PEEP/FiO2 tables [6] in a series of consecutive intubated COVID-19 hARF patients, admitted to intensive care unit at our institution. Written informed consent was obtained from all patients. We performed 15 decremental PEEP trials through a dedicated device (Pulmovista500, Dr ger-Medical, Germany) and subsequently analyzed pulmonary perfusion distribution [5]. Five patients were evaluated in a prone position. EIT optimal PEEP (PEEPEIT) was defined as the best compromise between lung collapse and overdistension [5]. All patients were deeply sedated without spontaneous breathing efforts and ventilated in volume control mode with lung-protective settings [3]. PEEPEIT was compared with PEEP from higher and lower PEEP/ FiO2 tables [6]. Data, expressed as median and interquartile ranges or 95% confidence interval (CI), were analyzed with the Mann–Whitney test for comparisons and Spearman rank test for correlations, considering p values < 0.05 significant. The Bland–Alman analysis was also performed. Patients had received invasive ventilation for 12.0 (10.0–14.5) days. Patients’ age was 63 (56–78) years, while body mass index (BMI) was 26.2 (25.4–30.9) kg/ m. Pulmonary shunt and dead space, as assessed by EIT [5], were 4% (2–6%) and 27% (23–36%), respectively. Ddimer was increased [759 (591–1208) mcg/L], while procalcitonin blood concentration was nearly normal [0.53 (0.34–0.70) mcg/L]. PEEPEIT was 12 (10–14) cmH2O and was significantly different from PEEP values of both higher [17 (16–20) cmH2O, p < 0.001] and lower [9 (8– 10) cmH2O, p = 0.049] PEEP/FiO2 tables. The Bland–

To the Editor, Hypoxemic acute respiratory failure (hARF) secondary to COVID-19 presents with heterogeneous features depending on several determinants, such as the extent of intravascular microthrombosis, superinfections, and other complications [1,2]. The easiest approach for setting positive end-expiratory pressure (PEEP) and inspiratory oxygen fraction (FiO 2 ) is using PEEP/FiO 2 tables [3,4]. However, because the magnitude of lung recruitability is variable, personalizing PEEP would be desirable [1]. Electrical impedance tomography (EIT) offers this opportunity by bedside estimating both alveolar collapse and lung overdistension throughout a decremental PEEP trial [5].
This investigation (Ethics Committee approval: Ref: 4853/AO/20-AOP2012) aims to assess the agreement between EIT-based PEEP values and those recommended by the higher and lower PEEP/FiO 2 tables [6] in a series of consecutive intubated COVID-19 hARF patients, admitted to intensive care unit at our institution. Written informed consent was obtained from all patients.
We performed 15 decremental PEEP trials through a dedicated device (Pulmovista500, Dr ger-Medical, Germany) and subsequently analyzed pulmonary perfusion distribution [5]. Five patients were evaluated in a prone position. EIT optimal PEEP (PEEP EIT ) was defined as the best compromise between lung collapse and overdistension [5]. All patients were deeply sedated without spontaneous breathing efforts and ventilated in volume control mode with lung-protective settings [3]. PEEP EIT was compared with PEEP from higher and lower PEEP/ FiO 2 tables [6]. Data, expressed as median and interquartile ranges or 95% confidence interval (CI), were analyzed with the Mann-Whitney test for comparisons and Spearman rank test for correlations, considering p values < 0.05 significant. The Bland-Alman analysis was also performed.
Patients had received invasive ventilation for 12.0 (10.0-14.5) days. Patients' age was 63 (56-78) years, while body mass index (BMI) was 26.2 (25.4-30.9) kg/ m 2 . Pulmonary shunt and dead space, as assessed by EIT [5], were 4% (2-6%) and 27% (23-36%), respectively.  (Fig. 1). No correlation was found between PEEP EIT and FiO 2 (p = 0.789) (Fig. 2) [4]. These differences are partly explained by the different criteria for PEEP EIT selection, which in that study was set above the value indicated by the built-in algorithm corresponding to the least lung collapse and overdistension [4]. Also, compared to our study, they enrolled more obese patients, as indicated by the higher BMI [30.0 (27.0-34.0) kg/m 2 ] [4]. Not reported in that study [4], our patients showed increased D-dimer and high fraction of pulmonary dead space, while shunt fraction and procalcitonin were nearly normal, suggesting predominant lung vascular disruption.
In conclusion, we confirm the rationale for individualized PEEP setting in COVID-19 patients intubated for Fig. 1 Bland-Altman plot, evaluating the agreement between PEEP EIT and the PEEP values proposed by the higher (a) and lower (b) PEEP/FiO 2 tables from the ALVEOLI trial [6]. X-axis: average of paired measurements. Y-axis: difference between paired measurements. The blue line and blue shaded area: bias and 95% confidence interval of the bias between PEEP EIT and the PEEP values suggested by PEEP/FiO 2 tables. Red lines: upper and lower limits of agreement between methods hARF. Whether EIT is the best technique for this purpose and the overall influence of personalizing PEEP on clinical outcome remain to be determined.

Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.