Respiratory Drive Measurements Do Not Signify Conjectural Patient Self-inflicted Lung Injury.

To the Editor:

We read with interest the editorial by Gattinoni and colleagues on the role of respiratory drive in the application of mechanical ventilation in patients with coronavirus disease (COVID-19) (1).

Gattinoni and colleagues conclude that measurements of airway occlusion pressure (P0.1) and maximal deflection in airway pressure during end-expiratory airway occlusion (ΔPocc) “displayed good prognostic performance in predicting respiratory deterioration at 24 hours” and can forewarn against patient self-inflicted lung injury (P-SILI).

Apart from the arbitrary definition of respiratory deterioration, the scatter plot in Figure 1 of the article by Esnault and colleagues (2) shows that the vast majority of data overlap between the two groups: 66% of P0.1 values in nine “Relapse” patients overlap with 74% of P0.1 values in 19 “Non-Relapse” patients. Only one value of ΔPocc in the “Relapse” group does not overlap with values in the “Non-Relapse” group. Gattinoni and colleagues fail to point out that the “break-points” were selected in a post hoc manner—a step known to markedly overestimate the accuracy of predictive indexes (3). No conclusions about reliability of a predictive index can be reached without the threshold being prospectively tested with a validation data set.

In addition to methodological problems, there is no justification for judging P0.1 4 cm H2O as a worrisome high value. Such values are seen in patients with stable chronic obstructive pulmonary disease and in patients successfully weaned from mechanical ventilation (4). There is no biological rationale for proposing that this level of respiratory motor output likely causes structural injury of the lung or respiratory muscles. The claim by Gattinoni and colleagues that P0.1 ≥4 cm H2O “portends subsequent worsening of respiratory function” constitutes major overinterpretation of the data.

Gattinoni and colleagues convey that P0.1 provides a reliable measure of respiratory motor output in individual patients. For decades, it has been known that numerous difficult-to-control factors alter the relationship between P0.1 and inspiratory muscle pressure output (4). Moreover, P0.1 exhibits a coefficient of variation as high as 38% in critically ill patients.

Gattinoni and colleagues claim that P0.1 and ΔPocc “correlate well with relatively more precise methods for effort estimation.” On the contrary, P0.1 ∼4 cm H2O is associated with a wide range of pressure–time product: ∼110 to ∼420 cm H2O · s · min−1 (Figure 3H of Reference 5). P0.1 ∼1 cm H2O is associated with a wide range of peak electrical activity of the diaphragm: ∼5 to ∼20 μV · s−1 (Figure 3B of Reference 5). ΔPocc of approximately −9 cm H2O is associated with a wide range of pressure–time product: ∼2.5 to ∼10 cm H2O · s · breath−1 (Figure E1 in the online supplement of Reference 6). Investigators excluded 30 of 82 recordings because the ratio of ΔPocc to change in esophageal pressure fell outside the range of 0.7–1.3. Basing decisions on P0.1 and ΔPocc regarding mechanical ventilation in individual patients is perilous.

Gattinoni and colleagues draw conclusions based on observed rapid shallow breathing index of 49 breaths/min/L. It has been known for decades that measurements of rapid shallow breathing index in the presence of un-estimated levels of respiratory work—inevitable with pressure support ranging between <4 and >11 cm H2O and positive end-expiratory pressure <10 to >14 cm H2O—are uninterpretable (3).

Gattinoni and colleagues continue to claim that the study by Tonelli and colleagues supports the existence of P-SILI (7). If inspiratory efforts were causing P-SILI, one would expect a decrease in Vt-to-transpulmonary pressure swing ratio—a surrogate of lung compliance; yet, Vt-to-transpulmonary pressure swing ratio remained constant across 24 hours of noninvasive ventilation. Chest radiography cannot be linked mechanistically to P-SILI because radiologists were not blinded.

Mechanical ventilation plays a crucial role in the management of patients with COVID-19. Conducting rigorous research is vital to enlighten clinicians at the bedside. A pandemic is no time to engage in speculation and broad generalizations based on dubious interpretations of small data sets. On the contrary, ventilator research in COVID-19 needs to aspire to the highest internal validity.