BBoB: Blood Glucose Levels Before and After Cognitive Testing in Diabetes Mellitus

This month's BBoB was an article examining the consumption of glucose by diabetics during cognitive testing. Diabetics have difficulty utilizing the glucose in their blood, which might have a cognitive impact The researchers measured glucose levels before and after cognitive testing and subtracted to create a delta score. The researchers hypothesis states that those diabetics that consume glucose (have a large negative delta number) will also perform better on the cognitive tests compared to those diabetics that do not consume glucose (have a low negative, or positive delta number).

The authors used a statistical technique that I think is kind of shady. The first thing the authors did was to throw away 33% of their data. They seperated the partisipants into three groups based on their delta (glucose consumption) score. Low consumers, middle, and High consumers, and then they discarded all the middle consumers from any analysis. I understand why you would do that (to make your comparison groups more extreme) but it is an arbitrary choice and I think it reduces the validity of the results. Ultimately I think the data they have (two scale variables) would be better analyzed with regression. It seemed to me that they were trying to ask the question, "do changes in blood glucose level predict performance on cognitive tasks?" If that is the case then regression could be used to answer that question. Without having to discard the middle portion of data, AND without having to reduce the other cases into two groups.

That said, the study seemed also to deal with a relatively small and homogeneous group and while there may be something to find there, I think they went about it in the wrong way. I would love to see this study replicated in a larger sample with more representation, with regression as the statistical analysis.

They also used a depression inventory as a screening tool that I hadn't heard of before: the Zung depression test. I looked it up and it seems fine as such things go, but there are a lot of physiological questions (are you constipated, etc.) that I thought might possibly be confounded by diabetes (e.g. Constipation interferes with some diabetes treatments). That said the Zung is what a couple of other diabetes studies had used, so I assume that's why it was used here, and it was only a screener anyway.

Natalie Galanina, Vijaya Surampudi, Daniela Ciltea, Sant Singh, Lawrence Perlmuter (2008). Blood Glucose Levels Before and After Cognitive Testing in Diabetes Mellitus Experimental Aging Research, 34 (2), 152-161 DOI: 10.1080/03610730701876979