My concerns with evidence supporting neuromechanical inhibition (NMI) as the cause of mechanical ventilation (MV)-associated apnea were not moderated after reading Madison's article. I must, however, address two other issues:

First, partial inhibition during synchronized MV (e.g. [1]) received much emphasis (weren't we debating apnea?). There was an ≈ 50% reduction in inspiratory output during MV (1). Given that reducing Pco ₂ ≈ 2 mm Hg abolishes inspiratory output during sleep, this inhibition is equivalent to reducing Pco ₂ ≈ 1 mm Hg. Most inhibitory artifacts listed in the Con article apply also to synchronized MV, albeit to a lesser extent. Partial inhibition could well be from the same artifacts. Additional inhibition results from conventional NMI that operates on a breath-by-breath basis; an increase in flow/volume, in phase with inspiration, reduces inspiratory time (Ti) and, hence, peak activity (2, 3).

Second, several publications originating outside Madison were cited to support Madison's version of NMI. I have two comments here. First, the same concerns regarding Madison's data apply to the external data. Second, Madison's use of this external evidence exemplifies beautifully how two allegedly seasoned scientists can interpret the same data in diametrically opposite ways. Two examples follow: (a) Morrell's results (4) are cited as supporting NMI; diaphragm EMG decreased in sleeping subjects placed on PSV despite no “significant” reduction in Pet _CO2 . These results, clearly, cannot support NMI as neither Vt nor RR increased. Pet _CO2 decreased in all subjects, but the average change (−2 mm Hg) just missed significance (p = 0.06) because of small n. Rather than supporting NMI, this study reflects exquisite sensitivity to CO₂. (b) Madison cites Lawson's study (5) as demonstrating delayed inhibitory effects after stimulating pulmonary afferents in the vagus. However, important poststimulus inhibition occurred only after laryngeal, but not vagal, simulation.

Demonstration of neuromechanical reflexes with slow dynamics is extremely difficult in humans as one must ensure that other sources of drive remain absolutely constant over a long period or, if changes must occur, that they are identical in the “control” and experimental periods. Comparing respiratory output with and without MV during total cardiopulmonary bypass, or during rebreathing, exemplifies what is required. Only one study met these criteria (6) and, here, no NMI was found.