The primary outcomes were neurocognitive measures of
memory function administered before and after the intervention.
The California Verbal Learning Test(40) was administered
to assess verbal learning and retention, and the Spatial Paired
Associate Learning Test(41) was used to evaluate non-verbal
memory. The California Verbal Learning Test is a listlearning
and recall task, and the Spatial Paired Associate
Learning Test assesses memory for visual-spatial information
that is not amenable to verbal encoding. Both list-learning and
paired associate tasks have been used in the context of cognitive
ageing and dementia and are among the more sensitive
measures of memory decline associated with neurodegeneration
(42 – 44). We also assessed mood as a potential covariate
of the cognitive measures with the Geriatric Depression
Scale(45). We performed weight and waist circumference
measures and obtained fasting blood samples for determination
of serum glucose and insulin values.
Analyses of covariance were performed for each outcome
factor to isolate effects of the intervention while controlling
for individual differences(46). The outcome score from the
final visit was the dependent measure and the corresponding
score from the baseline visit and the depressive symptom
score were covariate measures. We used eta squared values
to derive Cohen’s f effect size estimates, which are characterised
as small (0·10), medium (0·25) and large (0·40)(47).