[Skip to Content]
[Skip to Content Landing]
November 2005

High Folic Acid Intake Is Not a Risk Factor for Cognitive Decline: Misinterpretation of Results—Reply

Arch Neurol. 2005;62(11):1786. doi:10.1001/archneur.62.11.1786-b

In reply

We strongly disagree with Dr Fridman’s assertion that our results are attributable to a difference among the folate exposure groups in our ability to detect cognitive decline. The letter by Dr Fridman speaks to the very important issue that the measure used to assess cognitive change must be able to cover the full range of cognitive ability in the study population and must also be able to detect cognitive change over time at all levels of ability. Dr Fridman’s concern was that there was a floor effect in the cognitive scores for persons in the lowest quintile of folate intake, meaning that the test scores for this group were the lowest possible at the baseline and therefore any further decline at subsequent testing periods could not be measured. Most single tests are insensitive to measuring change at 1 of the cognitive extremes. Therefore, we combined the 4 cognitive tests into a global measure, which served to minimize the error of the individual tests. We computed z scores (z = x−μ/SD, where x = actual score; μ = population mean, and SD = population standard deviation) for each test so that they received equal weighting in the global score. The mean global score for the analyzed cohort at the baseline was 0.18 (minimum, −3.50; maximum, 1.58). A comparison of the range of baseline scores for persons in the first quintile of folate intake (mean, 0.08; minimum, −2.49; maximum, 1.44) and the fifth quintile (mean, 0.34; minimum, −2.85; maximum, 1.44) indicates that there was a wide range of scores in both groups and the lowest scores for these 2 groups were far above the lowest scores in the total cohort. Further evidence of the sensitivity of our analysis to detect cognitive change in the lowest quintile group is a comparison of the crude slopes of change in cognitive score for persons in the first (mean slope, −0.047; SD, 0.18) and fifth (mean slope −0.052; SD, 0.18) quintiles of intake, which shows that there was similar variability in score change for the 2 groups.