Corrigendum to “Effect of acute hypoxia on cognition: a systematic review and meta-regression analysis” (Effect of acute hypoxia on cognition: A systematic review and meta-regression analysis (2017) 74(PA) (225–232), (S0149763416307072), (10.1016/j.neubiorev.2017.01.019))

The authors regret that in the original article, several studies (Asmaro et al., 2013; Hayashi et al., 2005; Pavlicek et al., 2005; Shukitt-Hale et al., 1998; Tsarouchas et al., 2008; Wesensten et al., 1993; Wu et al., 1998) were allocated to the normobaric hypoxia (NH) condition when in fact they were hypobaric hypoxia studies (HH). This has no effect on the primary aspects of the study, namely the overall effect of hypoxia on cognition and comparison of effects on central executive, perception/attention and short-term memory tasks. However, it was necessary to re-calculate the meta-regression analyses for the moderators, moderate versus low arterial partial pressure of oxygen (PaO ₂ ) level and NH versus HH, and the interaction between the two. The meta-regression for the PaO ₂ variable, with low PaO ₂ as the reference category, showed that this was a significant, moderate moderator (R ² =.47, Q(1) = 14.90, p <.001, B = 0.81). This was not affected by the re-calculation, as the NH versus HH data were not included at this stage. For the NH versus HH variable, with NH as the reference category, the re-calculation showed that there was no significant effect of the moderator (R ² <.01, Q(1) = 0.20, p = >.05, B = 0.13). This differs slightly from the original which demonstrated a small moderator of borderline significance (R ² =.29, Q(1) = 3.99, p =.046: B = 0.50). The re-calculated interaction model (R ² =.62, Q(2) = 21.92, p <.001) showed that PaO ₂ was by far the better predictor of effect size (B = 0.92, p <.001), with the normobaric versus hypobaric variable adding nothing significant to the model (B = −0.33, p >.05). This differs only slightly to the original (R ² =.45, Q(2) = 14.72, p <.001), which also found that PaO ₂ was by far the better predictor of effect size (B = 0.75, p <.002), with the normobaric versus hypobaric variable adding nothing significant to the model (B = 0.14, p >.05). The authors apologize for this mistake but the re-analysis does not affect the scientific discussion and conclusions of the article in any way. The authors would like to apologise for any inconvenience caused.