Random-effects meta-analyses were performed using Comprehensive Meta-Analysis Version 2.2 (Biostat, Englewood, NJ).
The effect size reported here was OR. Since a ‘pure’ cognitive task rarely exists, cognitive tasks categorized into different domains may still recruit similar cognitive processes, eg, attention, and their measures may be correlated.
Thus, for studies which used multiple cognitive measures, we performed intra-study meta-analyses [40,41] to combine all the ORs to form an independent effect size of the overall impact of self-reported short or long sleep on cognitive performance.
For studies that used multiple tasks for the same cognitive domain, we meta-analyzed the results to generate an independent effect size for this domain. For those studies that used multiple statistical models, results from the most restricted model, ie, the largest number of covariates, were used to provide an estimate of the unique contribution of sleep to cognitive functions.