10 Gender differences in cognitive task performance and neurophysiology in a normative database
Introduction and aim: Behavioral and neurophysiological outcome measures of a visual-continuous-performance-task (VCPT) as well as characterization using electroencephalography (EEG) resting-state data can be used to complement clinical diagnosis of mental disorders like ADHD. The main aim of this study is to assess the necessity to apply sex-matching in the comparison of individual patient's data to a normative database.
Methods: Analyses have been performed with the data of the Swiss subjects from the HBimed database (n=754 subjects, 58% male, 42% female), compromising an age range of 6 to 80 years. Age-adjusted differences between male and female subjects were obtained from multiple linear regression models with sex and age, age category, and the interaction of the latter two, as independent variables. Logarithmic transformation of the dependent variable was applied if appropriate.
Results: For the behavioral measure of the VCPT, we found evidence for an increased number of commission errors in female subjects compared to males (t(747)=2.72, p<0.02, Cohen's d=0.21). We did not find evidence for gender differences in the number of omission errors, reaction time, and reaction time variability. For the neurophysiological measures of the VCPT, namely amplitudes and latencies of event-related potentials (ERPs), we found evidence for attenuated GoP3 amplitudes in females compared to males (t(747)=-4.22, p<0.01, d=-0.32), but not for NoGP3, NoGoN2, and CNV amplitude or GoP3, NoGP3 or NoGoN2 latency (all p>0.05). In the outcome measures deriving from resting state EEG we did find evidence for increased theta-to-beta ratio (eyes-closed: t(707)=2.42, p=0.02, d=0.19, eyes-open: t(716)=4.11, p<0.01, d=0.32) and decreased arousal during eyes-open (t(716)=-2.35, p=0.02, d=-0.18) in females compared to males.
Conclusion: Gender differences were evident in resting-state EEG markers, task performance, and ERPs measures within healthy subjects. Although the effects were small (d<0.5) we recommend sex-matching to avoid gender-biased diagnosis, if normative databases are used.