Benefits of cognitive-motor intervention in MCI and mild to moderate Alzheimer disease
Harish C.Kavirajan, UCLA Geffen School of Medicine, 950 South Coast Drive, Suite 202, Costa Mesa, CA 92626hkaviraj@yahoo.com
Submitted March 08, 2005
We read the article by Olazaran et al with interest. [1] There are several problems with study design and data analysis:
1. The quality of the blinding is questionable. Most patients had MCI or mild dementia, and therefore presumably fair recall of recent events/activities. The patients themselves, not only caregivers, could have divulged group assignment to the blinded evaluators.
2. The control group lacked a social stimulation intervention so the modest benefits of CMI could be ascribed to nonspecific social stimulation effects, rather than specific properties of the CMI.
3. The CMI group was responsible for paying for treatment and attending frequent training sessions. Accordingly, differences in financial or psychosocial resources may have impacted differences in outcome.
4. “Mood responders” were defined as patients who maintain or improve their baseline GDS score. However, GDS depression scores were low in both groups, with means well below the cutoff for even mild depression. Lack of change or improvement in such subjects would not qualify as a response.
5. Instead using of a more conventional paired t-test, the authors treated data from each group at different points in time as though from different groups, suggesting that large within-subject variance may have reduced statistical significance with analysis of within-subject change. The unreliability of the MMSE for measuring short-term change is well known. [2]
6. Failure to adjust for multiple statistical analyses increased the probability of a type I error. The only justification offered is that such a correction would have increased the probability of a type II error.
7. The conclusions regarding behavioral benefits were based on comparisons of the mean scores on the ADRQL and NPI only at study endpoint, without controlling for baseline data. Hence, the differences between groups at endpoint may simply have reflected baseline differences in the two groups.
8. The authors failed to describe CMI program attendance. Besides demonstrating tolerability of the intervention, such data might suggest a dose-effect of treatment, which would bolster the case for an effect of CMI.
9. The explanation of the negative effect of education is not clear. The logic of the cognitive reserve hypothesis is inconsistent with the notion that higher levels of education could explain poorer performance on cognitive measures.
References
1. Olazaran J, Muniz R, Reisberg, B, et al. Benefits of cognitive-motor intervention in MCI and mild to moderate Alzheimer’s disease. Neurology 2004;63:2348-2353.
2. Clark, CM, Sheppard L, Fillenbaum GG, et al. Variability in Annual Mini-Mental State Examinations Score in Patients With Probable Alzheimer Disease. Arch Neurol. 1999;56:857-62.
We read the article by Olazaran et al with interest. [1] There are several problems with study design and data analysis:
1. The quality of the blinding is questionable. Most patients had MCI or mild dementia, and therefore presumably fair recall of recent events/activities. The patients themselves, not only caregivers, could have divulged group assignment to the blinded evaluators.
2. The control group lacked a social stimulation intervention so the modest benefits of CMI could be ascribed to nonspecific social stimulation effects, rather than specific properties of the CMI.
3. The CMI group was responsible for paying for treatment and attending frequent training sessions. Accordingly, differences in financial or psychosocial resources may have impacted differences in outcome.
4. “Mood responders” were defined as patients who maintain or improve their baseline GDS score. However, GDS depression scores were low in both groups, with means well below the cutoff for even mild depression. Lack of change or improvement in such subjects would not qualify as a response.
5. Instead using of a more conventional paired t-test, the authors treated data from each group at different points in time as though from different groups, suggesting that large within-subject variance may have reduced statistical significance with analysis of within-subject change. The unreliability of the MMSE for measuring short-term change is well known. [2]
6. Failure to adjust for multiple statistical analyses increased the probability of a type I error. The only justification offered is that such a correction would have increased the probability of a type II error.
7. The conclusions regarding behavioral benefits were based on comparisons of the mean scores on the ADRQL and NPI only at study endpoint, without controlling for baseline data. Hence, the differences between groups at endpoint may simply have reflected baseline differences in the two groups.
8. The authors failed to describe CMI program attendance. Besides demonstrating tolerability of the intervention, such data might suggest a dose-effect of treatment, which would bolster the case for an effect of CMI.
9. The explanation of the negative effect of education is not clear. The logic of the cognitive reserve hypothesis is inconsistent with the notion that higher levels of education could explain poorer performance on cognitive measures.
References
1. Olazaran J, Muniz R, Reisberg, B, et al. Benefits of cognitive-motor intervention in MCI and mild to moderate Alzheimer’s disease. Neurology 2004;63:2348-2353.
2. Clark, CM, Sheppard L, Fillenbaum GG, et al. Variability in Annual Mini-Mental State Examinations Score in Patients With Probable Alzheimer Disease. Arch Neurol. 1999;56:857-62.