et al. 1 attempted to establish an association between glucose metabolism and memory performance. They concluded that higher blood glucose levels negatively influenced cognition. We believe that the design and methodology is inadequate to draw these conclusions.The bivariate approaches do not consider that confounding factors might be related to both glucose metabolism and memory (e.g., age). The hippocampus as a potential mediator is stated to lose significance upon adjustment for even the most basic confounders. Presenting raw and adjusted models would allow the assessment of the role of confounders.In addition, we could not interpret the standardized regression coefficients for sex as a dichotomous trait. Measures of precision (e.g., confidence intervals [CIs]) for associations were not reported. To translate the authors' results to real-world quantities, we do not see how to use many significant test results without adjustment for multiple testing. We were not able to derive a meaningful adjusted measure of association between HbA1c and the presented memory tests.The authors should provide a comprehensible magnitude adjusted for potential confounders, for example: "5 mmol/mol higher HbA1c levels were associated with an average of 1.5 (0.5-3.0) fewer words remembered in the delayed recall test."Author Response: Lucia Kerti, A. Veronica Witte, Ulrike Grittner, Agnes Floeel, Berlin: The authors thank Drs. Grabenhenrich and Roll for their comments. We reported significant associations of markers of glucose metabolism with memory performance in 141 older adults.1 In addition to bivariate analyses, we reported adjusted multiple regression models, which were controlled for confounding factors including age and sex using a stepwise selection method.We have conducted alternative models using the "enter" method for age and sex variables and obtained similar results to those we reported, thus supporting our original conclusions. HbA1c remained significantly associated with the primary outcome (delayed recall) and the other 2 subtests of memory performance (p , 0.05). If a Bonferroni correction for multiple testing is applied, the associations between HbA1c and delayed recall and between HbA1c and learning ability remained significant. According to the regression model, a difference in HbA1c levels of 5 mmol/mol was associated with a reduction of 1.4 remembered words (95% CI 0.4-2.3) in the delayed recall task.In our study, we outlined strengths and weaknesses of a cross-sectional design and highlighted the necessity of future longitudinal trials. A recent epidemiologic survey of glucose levels and dementia revealed that higher glucose levels may be a risk factor for dementia, even in those without diabetes.