Premium
The impact of protocol design on MMSE changes in the screening period
Author(s) -
Kott Alan,
Miller David S.
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.043883
Subject(s) - medicine , dementia , baseline (sea) , clinical trial , protocol (science) , physical therapy , disease , pathology , oceanography , alternative medicine , geology
Background Anecdotally, we have observed subgroups of subjects who in the screening period of clinical trials in early Alzheimer’s disease exhibit unexpectedly large changes in MMSE scores. Such changes could be explained by a number of reasons, such as learning effects, regression to the mean, measurement error, or intentional score manipulation. If the latter was the case, differences in MMSE changes between screening and baseline should be observed between protocols where MMSE inclusion criteria are required to be met at screening only and protocols requiring the criteria be met at screening and baseline. We hypothesized that MMSE changes would be larger in protocols with criteria required only at screening. Method Data were pooled from multi‐national dementia clinical trials in early AD where MMSE was collected at screening and baseline. Subjects were categorized into 2 groups depending on whether the inclusion criteria were applied to screening alone or to both the screen and baseline visits. T‐test was used to compare the change between screening and baseline between the two groups of subjects. Additionally we compared the distribution of subjects changing by a) 3 and b) 5 MMSE points in either direction between screening and baseline between the two protocol groups using a chi‐square test. Result Data were collected from 1,409 subjects. The MMSE change in subjects with criteria required at screening only was ‐1.2 points compared to ‐.05 points in subjects with criteria required at screening and baseline (t(1,407) = 5.99, p < 0.001). MMSE changes of 3 or 5 points were seen significantly more often in studies where the requirement was applied at screen alone (chi2(1)=18.2, p < 0.001 and chi2(1)=13.1, p < 0.001, respectively). Conclusion Our analyses indicate that protocol design has a significant impact on the MMSE change observed in the screening period. Given the only difference between the two groups was the absence of inclusion criteria at baseline, we hypothesize that the MMSE score change observed differences can most likely be explained by score manipulation happening at either screening only or screening and baseline, depending on protocol type. We plan to replicate these results in larger datasets.