Do non-traumatic stressful life events and ageing negatively impact working memory performance and do they interact to further impair working memory performance?

Stress and normal ageing produce allostatic load, which may lead to difficulties with cognition thereby degrading quality of life. The current study’s objective was to assess whether ageing and cumulative stress interact to accelerate cognitive decline. With 60 participants, Marshall et al. found that ageing and cumulative stress interact significantly to impair working memory performance in older adults, suggesting vulnerability to the cumulative effects of life events beyond 60 years old. To replicate and extend this finding, we increased the sample size by conducting 3 independent studies with 156 participants and improved the statistical methods by conducting an iterative Bayesian meta-analysis with Bayes factors. Bayes factors deliver a more comprehensive result because they provide evidence for either the null hypothesis (H0), the alternative hypothesis (H1) or for neither hypothesis due to evidence not being sufficiently sensitive. Young (18–35 yrs) and older (60–85 yrs) healthy adults were categorised as high or low stress based on their life events score derived from the Life Events Scale for Students or Social Readjustment Rating Scale, respectively. We measured accuracy and reaction time on a 2-back working memory task to provide: a) Bayes factors and b) Bayesian meta-analysis, which iteratively added each study’s effect sizes to evaluate the overall strength of evidence that ageing, cumulative stress and/or the combination of the two detrimentally affect working memory performance. Using a larger sample (N = 156 vs. N = 60) and a more powerful statistical approach, we did not replicate the robust age by cumulative stress interaction effect found by Marshall et al.. The effects of ageing and cumulative stress also fell within the anecdotal range (⅓

1 S3 Appendix 2. Additional tasks administered during the study (description and results)

Brief Resilience Scale
We wanted to extend Marshall's study protocol by adding resilience as a factor to enhance our understanding of how cumulative stress might affect cognition, given participants' ability to recover from stressful events.To this end, we administered the Brief Resilience Scale (BRS) [65].The BRS has good internal consistency (Cronbach α <.95 > .70) and test-retest validity (interclass correlation coefficient .69 to .62) with a range of populations [e.g.66, 67] and was found to be well-suited to stress-related contexts [68].
Participants were asked to self-report the extent to which they agreed with 6 statements on a scale of 1 ('Strongly Disagree') to 5 ('Strongly Agree').Three of the statements were worded positively (items 1,3,5) and 3 negatively (items 2,4,6).Scores were derived by reverse-scoring items 2, 4 and 6 and then calculating the mean of all items.A higher mean score indicates greater resilience; previous research has shown that the BRS is negatively associated with physical symptoms and negative affect (e.g.irritability and distress) [65].

Subjective Sleep Quality
We also added the Pittsburgh Sleep Quality Index (PSQI) [69] to the study protocol to assess sleep quality, asking participants to report their sleep quality over the past month.
Sleep quality has been consistently linked to variability in cognitive performance, stress, anxiety and illness [69][70][71][72][73][74].Individual items in the PSQI yielded a Cronbach α of 0.83, indicating a high degree of internal consistency.Test-retest reliability revealed coefficient of .85 and there was good discriminant validity between clinical (depressed, disorders initiating and maintaining sleep, disorders of excessive somnolence) and control groups (p < .001).
We used only questions 5 and 6 to keep the experiment short to reduce fatigue.Both questions were rated on a 4-point scale (score range: 0 to 3).Question 5a, in this study, 2 provided an index for 'sleep latency' and was rated as: 'Not during the past month' = 0 to 'Three or more times a week' = 3. Question 5 b-j comprises 10 questions assessing 'sleep disturbances' rated as per Q5a above.These values were summed for each participant.
A global score, which had a score range of 0 -9, was computed by summing the 3 aforementioned components, namely 'sleep latency', 'sleep disturbances' and 'subjective sleep quality'.Note that these methods are adapted from the original PSQI which yields a global score of 0 -21, based on 7 components.

Appendix Table 1. Descriptive statistics and p-values for self-reported resilience and sleep quality by age, by stress group for each study.
a Mean (SE).Standard error obtained via BCa Bootstrap with 1000 samples.bIndependentsamples t-test (low vs high stress) were performed by age group.cAdditional Mann-Whitney U test were performed with similar outcomes.* significant at < 0.05 ** significant at < 0.01