Bibliography Card for Dementia Intervention Study

Huntley, J. D., Gould, R. L., Liu, K., Smith, M., & Howard, R. J. (2015). Do cognitive interventions improve general cognition in dementia? A meta-analysis and meta-regression. BMJ Open, 5(4), e005247. Content Summary: This article reports that there are three main approaches in the use of cognitive interventions for people with dementia: cognitive training (CT), cognitive stimulation (CS), and cognitive rehabilitation; however the studies reporting the results of these interventions have been mixed (pg.2).

The researchers attempted to use meta-analytic procedures to evaluate the efficiency of CT and CS for dementia that used active and non-active control groups, to examine their effects on commonly outcome measures in the research (e.g., MMSE, ADAS-Cog), and to determine associations between effect sizes and variables that influence the efficiency of these interventions by using meta- regression analyses (pg. 2). The researchers did an extensive search of online literature databases and trial registers using stringent inclusion criteria to pare down 420 relevant studies to 33 studies for meta-analysis (pg. 4).

Results of the meta-analyses indicated small to moderate effect sizes for CS when the MMSE was used as the outcome measure with higher effect sizes for non-active controls (g = 0.51) compared to active controls ( g = 0.35). Smaller effect sizes for the ADAS-Cog were obtained (g = 0.51). CT was not found to be effective for treating dementia and combinations of CS and CT were not more effective than CS alone (pg. 5).

Meta-regression analyses revealed no association between the effect sizes and the control group, setting (inpatient or outpatient), interventions in a group or an individual setting, intervention length or intensity, or the severity of the participant's dementia (pg. 8). Evaluation/Implications: Given the effect sizes the authors concluded that CS and CT do not result in significant cognitive changes for dementia.

However, even the authors admit that statistically significant effect sizes and not the same thing as clinically significant effects (pg 9) and quite frankly it is very difficult to find studies that define clinically significant outcomes for this patient group and even more difficult to generate agreement on what a clinically significant outcome would be. The studies do not make good use of blinding (for experimenters) or the use of placebos and of course RCTs are the gold standard in this type of research.

It could be that significant findings are biased by experimenter expectations (pg. 9). Moreover, the researchers that the studies used here are limited by the outcome measures (pg 10). They do not discuss the limitations of these screening measures however. The MMSE is a very brief screen and is not an appropriate measure to determine changes in cognition but is best used to determine if further cognitive testing is needed. For example, the MMSE memory component has little potential for variation.