13 December 2023
I congratulate the authors of the study “A Mediterranean Diet and Walking Intervention to Reduce Cognitive Decline and Dementia Risk in Independently Living Older Australians: The MedWalk Randomized Controlled Trial Experimental Protocol, Including COVID-19 Related Modifications and Baseline Characteristics” published in the Journal of Alzheimer's Disease [1], for developing and contributing to the field of non-pharmacological interventions and lifestyle changes to reduce the risk of dementia and Alzheimer's disease in older adults. Given the limited success of all initiatives focused on finding pharmacological therapies, lifestyle modification interventions as secondary prevention measures are increasingly positioned as the first-line option, so it is increasingly necessary to perform randomized clinical trials to evaluate the effectiveness and efficacy of such non-pharmacological interventions.
Before mentioning the points of concern, I will mention what I consider one of the main strengths of this study, the inclusion of Cognitive behavioral therapy in the motivational interviewing protocol (MI-CBT). Behavioral modification is a key component of lifestyle modification, but it is often ignored. The inclusion of MI-CBT may contribute to intervention adherence, which is one of the most difficult aspects to sustain in this type of study and the authors addressed it consistently.
I will now turn to points of concern that I feel should be expanded upon by the investigators in this article. One of them is the representativeness of their sample. In this sense, we have a sample made up primarily of whites, which could make the study unrepresentative when it comes to extrapolating the results to the general Australian community, which has a large group of racial minorities and migrants of various ethnicities.
Another point of concern I find is the assessment of dementia or mild cognitive impairment (MCI). It has been suggested by previous research that the Mini-Mental State Examination (MMSE) is not the best measure of screening for dementia, furthermore a cut-off point of 24 points could lead to the inclusion of potential subjects with MCI or incipient dementia. If dementia or MCI was discarded just based on MMSE it could lead to a selection bias. According to the article, the MMSE < 24 was used as the cut-off point for the exclusion of potential candidates with one of the two conditions; however, it is not clear whether the investigators performed more specific assessments in the screening to determine the diagnosis of dementia or MCI. Furthermore, an MMSE with a 24-point cut-off could lead to a selection bias because it has the potential to include false negatives due to the ceiling effect reported for this screening test.
The next aspect to analyze is the randomization. As the authors explain they performed a cluster randomization due to the homogeneity of the villages where the study participants reside, however, due to the COVID-19 situation and the scarcity of subjects to recruit they ended up including subjects from the general community and randomizing them under the same cluster methodology, without explaining how they ensured that this community cluster did meet the homogeneity characteristics necessary to ensure the representativeness of the community cluster in the randomization. In this same sense, it will be very important to know the differences between the closed cluster made of the Villages residents and the community cluster, since these results may not be generalizable to the more diverse community that is not living in more homogeneous pockets such as the villas for the elderly.
Another major issue that needs revision is the significant reduction in sample size that they suffered due to the COVID-19 pandemic, this further impacted the attrition rate reducing it to 7%, which increases the risk of losing power due to the attrition rate expected from studies of this type which is expected to be around the 30%, the authors need to consider a very robust protocol of adherence to maintain this low attrition rate.
I look forward to the authors' response and any potential adjustments that may come from this dialogue. I believe that open and constructive discourse within the scientific community is instrumental in refining our understanding and advancing the field.
Thank you for your attention to these matters, and for the platform provided by the Journal of Alzheimer's Disease to foster such important discussions.
Lina Velilla
PhD Student
Department of Epidemiology
College of Public Health and Health Professions & College of Medicine
University of Florida
Comments
Response
We thank the author of the Letter to the Editor for commenting on our recently published paper, “A Mediterranean Diet and Walking Intervention to Reduce Cognitive Decline and Dementia Risk in Independently Living Older Australians: The MedWalk Randomized Controlled Trial Experimental Protocol, Including COVID-19 Related Modifications and Baseline Characteristics”. We provide responses to the comments made.
In relation to the “representativeness of their sample,” Table 3 shows basic baseline characteristics of our cohort, including ethnicity, which is representative of the participants volunteering for this trial through the 17 retirement villages and 4 community-dwelling clusters established for recruitment. We agree that this cohort is not representative of the general Australian community; a limitation we will discuss when the findings of this trial are reported. This is a common issue globally with the lack of representation from minority groups in dietary and lifestyle intervention trials highlighting a need for specific recruitment strategies to ensure there is representation from all populations [1].
Contrary to what has been suggested, we are not using the MMSE in "the assessment of dementia or mild cognitive impairment (MCI)." Rather we are using the MMSE to screen for cognitive impairment - to exclude any participants who may have dementia or MCI. The intended sample are independently living, cognitively healthy individuals. This MMSE screening is in addition to our exclusion criteria, of people who have already been diagnosed with dementia or other forms of cognitive impairment. We reported in Table 3 that MMSE is 28.9/30 with a standard deviation of 1.2, clearly above the MMSE threshold for cognitive impairment. Moreover, our primary outcome is cognition, using CANTAB cognitive tests that will be far more sensitive than any screening tools. In the unlikely event that an individual with dementia or MCI has been recruited, they are likely to be identified as an outlier in the data screening stage when analyzing these cognitive tests.
In relation to the comment made about retirement villages versus community clusters, as stated in our paper and detailed in the CONSERVE - spirit checklist, additional recruitment through four community clusters was undertaken because of COVID-related impacts. However, recruitment through the wider community utilized the same inclusion and exclusion criteria such as the requirement to be living alone or with their partner and, as was the case for retirement villages, couples were assigned to the same group.
We also stated that any significant differences between community/retirement villages will be investigated. Our approach will be as follows: i) A comparison of baseline values for the community and retirement village participants will be conducted for all the outcome measures and demographic characteristics. If no significant differences are found, we will treat as a combined sample. If not, we will control for a binary variable discriminating between community and retirement village participants (Community/Retirement Village Living) in all our outcome analyses. ii) In addition, we will test for a significant interaction effect for treatment*(Community/Retirement Village Living) in order to determine whether the treatment effect differs significantly for those living in the community and those living in retirement villages. If this interaction effect is found to be significant, marginal treatment means will be reported for community and Retirement Village Living participants. In addition, separate outcomes analyses will be performed for people living in retirement villages and people living in the community in order to validate the results.
Lastly, with regards to the attrition rate of 7% and statement that "the authors need to consider a very robust protocol of adherence to maintain this low attrition rate." To clarify, we are not setting ourselves a goal to reduce attrition from 30% to 7%, rather we had already achieved this at the time of publication of our protocol paper, with approximately three quarters of data collection completed. This much lower than expected attrition rate as well as other factors including: a reduced number of assessments per participant (5 to 3) and a reduced number of participants per cluster (13 to 7) meant that the trial was still adequately powered (see "Statistical Analysis" Section).
Andrew Pipingas, Karen J. Murphy, Courtney R. Davis, Catherine Itsiopoulos, Michael Kingsley, Andrew Scholey, Helen Macpherson, Leonie Segal, Jeff Breckon, Anne-Marie Minihane, Denny Meyer, Edward Ogden, Kathryn A. Dyer, Emily Eversteyn, Roy J. Hardman, Kaylass Poorun, Keri Justice, Maher Hana, Jonathan D. Buckley, David White, Kade Davison, Jessie S. Clark, Ella L. Bracci, and Greg Kennedy on behalf of MedWalk collaborative team
[1] Shaw AR, Perales-Puchalt J, Johnson E, Espinoza-Kissell P, Acosta-Rullan M, Frederick S, Lewis A, Chang H, Mahnken J, Vidoni ED (2022) Representation of racial and ethnic minority populations in dementia prevention trials: a systematic review. J Prev Alzheimers Dis 9, 113-118.
- Comment
|