ZB Oct 8 7:48pm

Suzuki et al. conducted a longitudinal (repeated-measures) observational study of Alzheimer’s patients, tracking changes in cognition and behavior over three years using their “Daily Cognition and Behavior for Alzheimer’s Disease” (DCB-AD) scale, alongside two other professionally used scales. The study design is a within-subjects longitudinal design, meaning that the same individuals were assessed repeatedly over time. This approach enables the tracking of individual changes, thereby minimizing the effects of between-subject variability (Frankfort-Nachmias, Leon-Guerrero, & Davis, 2020).

The authors used a one-way repeated-measures ANOVA to compare DCB-AD scores across different time points and to assess whether there was a systematic decline in cognitive and behavioral functioning over time. The ANOVA test is appropriate in this context as it allows for testing whether the mean scores differ significantly across repeated measures while accounting for the correlation between observations within the same subject (Frankfort-Nachmias et al., 2020). However, while ANOVA is commonly used for repeated-measures designs, it is essential to note that this test assumes sphericity, homogeneity of variance, and normality of the residuals. If these assumptions are violated, the results may be less reliable. In the context of Alzheimer’s disease, where cognitive decline is often non-linear and may vary in degree across individuals, a linear mixed-effects model or growth curve modeling would be more appropriate because these approaches better handle complex data structures, missing data, and non-constant variance (Frankfort-Nachmias et al., 2020).

Although the choice of ANOVA is reasonable, it is not without limitations. The results would benefit from a more robust statistical model that could account for individual differences in the rate of cognitive decline. Additionally, the authors displayed data in terms of mean scores over time, comparing DCB-AD results to those of the other scales. However, the absence of figures such as individual trajectories, confidence intervals, or error bars weakens the ability to assess variability in the data. Presenting the results in tabular form would provide a clearer picture of the variation between individuals and time points. The results, while statistically significant, do not “stand alone” because they lack crucial information on variability and individual patterns of change, which are essential for understanding the clinical significance of the findings.

Make recommendations for the design choice.

Explain whether you think that this is the appropriate test to use for the research question. Why or why not?

As a lay reader, were you able to understand the results and their implications? Why or why not