My bipolar sample has undertaken the CANTAB assessment over 8-year follow-up. Since I do not have a control group, I consider to use Z-scores to compare the raw scores with their norm group according to age and gender.
Based on previous literature on neurocognition, I have not found consistent evidence regarding cut-off levels for Z-scores. Which Z-scores should be considered for cognitive deficits (Z
usually, z-scores below 1.0 SD indicate "borderline" cognitive impairment; 1.5 SD indicates minor cognitive impairment (where cognitive deficits start to be evident) and 2.0 SD indicates major cognitive impairment. So it really depends on how stringent you want your analysis to be. Hope you find it useful.
I second Xin Xu on this. The standard deviation of z scores is 1.00; that on the Wechsler deviation IQ scale (used on many standardized tests of cognitive functions) is 15. That's why the usual cutoff for "Intellectual Disability" (formerly mental retardation) is 70: it's set at two standard deviations below the mean. But if you are concerned with identifying patients who have probably sustained a "hit" to their intellectual abilities, a one standard deviation cutoff makes sense. After all, that's enough to move an "average" person to about the 14th percentile!
It sounds like you won't be able to discern the extent to which the intellectual limitations were always present vs. occurred with the onset of Bipolar Disorder. That's a pity, since of course it makes a difference - does lower IQ make you more vulnerable, is there a reduction in intellectual abilities present from early childhood? Or is there "damage" occurring in the early stage of the disorder?
On the bright side, with an 8-year follow-up period you will be able to discern whether there is progressive deterioration as opposed to a static deficit. That's especially true of the subset of patients who were newly diagnosed/had only recently developed the disorder at the beginning of your study. (The chronic cases wouldn't necessarily change much over 8 years, though I believe there is evidence of long-term deterioration in many cases of Bipolar Disorder.)
Thank you for your prompt responses!!!
Indeed, I will performed a linear mixed model with repeated measures. So, I am very interesting in determining the longitudinal course as you indicate. I hope I will obtain interesting results!
Regards
From a statistical standpoint, using a cut-off of 1.645 would give you an equivalent p-value of p = .05, indicating a significant change in scores (or 1.96 if two-tailed). Try this website to calculate your scores:
https://www.easycalculation.com/statistics/p-value-for-z-score.php
It also depends on the number of tests in your cognitive battery - see attached publication.
Hi Alvaro,
If it is a longitudinal study is it that you are attributing cognitive decline due to Bipolar Disorder? If so would you therefore be looking at mean differences in scores, from baseline to the 8th year? Just wondering how you are going to derive the norms. If the CANTAB does not have meaningful, clinical cut-offs points on the scale (and across x domains of CANTAB-measured intelligence) I would suggest you include the disclaimer that the meanings you have attached to the z-score distances are from the general neurocog. literature.
Xin Xu could you please send me the reference? I need it in order to use it for an article! Thank you
- Badges
- Science topic
- Similar topics
- Psychology