Sex differences in neuropsychological tests

A brief summary of my paper 'Maximizing utility of neuropsychological measures in sex-specific predictive models of incident Alzheimer’s disease in the Framingham Heart Study'

It is known that slight differences occur in the performance of selected Neuropsychological (NP) tests, but could we use such differences to create new and more predictive tests? To answer this question one would need a massive dataset, with thousands of participants, with incidence of dementia recorded over a long period of time. Reader, this dream dataset exists and it has been collected in the frame of the Framingham Heart Study, Boston University School of Medicine (FHS).

The cohort used

FHS was initiated in 1948 and recruited an initial cohort (Gen 1), followed by Gen 2 in 1971. Amazing enough, 99% of participants regularly return to follow-ups, including extensive NP assessment. In this paper we combined Gen 1 and 2 participants who underwent NP (for a total of 1787 men and 2228 women).

Sex differences in performances and processing

In spite of higher years of education in men, women showed the well known advantage in a number of NP tests involving logical memory and verbal fluency. Men in contrast showed better performance in finger tapping and visual reproduction.

In addition to scoring for correct responses, FHS also collected incorrect or extraneous responses (things such as confabulations, perseveration and intrusions). For the first time to my knowledge, we report sex differences in such scores, highlighting the impact of sex on processing of classical NP tests, not only in their performance.

How about predictivity?

The differences are highly significant due to high n, but with tiny effect sizes. Yet,

they might become relevant when looking at correlation with incidence of Alzheimer's disease (AD) later on in life. With a mean follow up time of 9 years, we assessed exactly that and also created sex-specific optimal NP profiles for AD dementia prediction. It was really interesting to see that performance in specific tests can be more predictive for a given sex (for instance, paired associated learning could identify preclinical female better); also, the optimal algorithm for incident AD prediction features different items for men and women.

?Conclusions

These results suggest that sex differences could be leveraged for increasing the predictivity of current tests we use in AD diagnosis, especially in early stages. In addition, these insights could inform the creation of new wave of tests in the context of precision medicine.

With a big thank you to the whole team and in particular to Rhoda Au for believing in this project since day 0, and Ting Fang Alvin Ang and Huitong Ding for the amazing collaboration.

https://alz-journals.onlinelibrary.wiley.com/doi/epdf/10.1002/alz.13500

?