Highly significant differences were observed between baseline and after stages for both congruent (air: Z = −4.675, p < 0.001; heliox: Z = −2.985, p = 0.003; trimix: Z = −3.650, p < 0.001) and incongruent trials (air: Z = −4.200, p < 0.001; heliox: Z = −3.299, p < 0.001; trimix: Z = −3.154, p = 0.002).
Pairwise comparisons revealed that results obtained at 4 ATA were significantly higher than initial values for flicker (Z = −2.352, p = 0.019) and average CFFF (Z = −2.433, p = 0.015), but not for fusion, where only a statistical trend was observed (Z = −1.922, p = 0.055).
Furthermore, when comparing the 4 ATA stage to the final stage, a significant decline was observed for fusion (Z = −3.990, p < 0.001) and average (Z = −3.613, p < 0.001), whereas flicker did not show a statistically significant change, although the result approached significance (Z = −1.901, p = 0.057).