**3. Results**

Participants in this sample had a mean age of 60 years (SD = 8.5 years) and were reasonably well balanced in regard to sex (55% female). Table 3 provides average levels of cognitive performance over the course of the study as well as mean change in cognitive function for each domain. Equivalence of the groups at baseline was confirmed via no statistically significant differences between treatment arms for any measure (all *p* ≥ 0.42). Descriptive statistics regarding changes in cognition show that it was generally neutral or positive.

**Table 3.** Means (± standard error of the mean) for cognitive variables at baseline, endpoint and for overall change.


Analysis of data using mixed effects models showed significant overall time effects for reasoning speed, verbal working memory and numeric working memory (all *p* ≤ 0.001). In addition, there were significant time by treatment interaction effects observed for reasoning speed (*p* = 0.04) and reaction time (*p* = 0.02). In both cases, performance increased over the duration of the intervention with a significantly greater increase observed during the SPI compared to WPI treatment. Additional three way interactions (time by treatment by sex) showed that these effects were only evident in females during SPI treatment (both *p* < 0.05). As shown in Figure 2, there was generally no change in performance for males during either treatment for either cognitive domain but significant direct changes were observed for females during SPI treatment only.

The results of this intervention on various biochemical scores have been reported in full previously [31]. However, given the presence of prior studies suggesting a relationship between alternate markers of B12 status and cognition, additional analyses were conducted in order to examine changes in markers of B12 status during the active WPI treatment phase only and their associations with observed changes in cognitive function during the same treatment period. Briefly, the results of the intervention on the parameters of interest are provided in Table 4. WPI treatment lead to significant direct changes in active B12 and serum folate, however changes in MMA did not quite reach significance. Correlations between changes in these biochemical measures and cognitive function during both WPI and SPI intervention periods can be seen in Table 5. These correlations were fully adjusted for age and sex covariates. As can be seen, 5/6 significant correlations emerged for the WPI intervention. Increases in in active B12 were associated with improvements in processing speed. Similarly, increased serum folate was associated with improvements in delayed word memory. Reductions (i.e., improvements) in homocysteine levels were related to improvements in processing speed, numeric working memory and verbal working memory. In the SPI condition, homocysteine changes were related to reasoning speed with higher levels equating to reduced performance. **(A)**

We conducted the same analyses using combined B12 (cB12), a combined indicator of corrected B12 status [31] and the pattern of correlations was the same as for active B12.

**(B)** 

**Figure 2.** Mean change scores and 95% Confidence Intervals for Treatment × Gender interactions for Delta (**A**) Reasoning Speed and (**B**) Reaction Time. \*\**p* < 0.01.


**Table 4.** Baseline scores (Mean ±SE) and percentage change observed due to treatment.

a Comparison of percentage change between treatments. ↑ Increase, ↓ decrease.

**Table 5.** Correlations between changes in biometric measures and changes in cognitive function during each treatment period; adjusted for sex and age covariates.


MMA = Methylmalonic Acid; tHcy = total homocysteine. \* *p* < 0.05, \*\* *p* < 0.01 (one-tailed).
