Introduction
It is undisputed that age adversely impacts cognitive functioning, yet cognitive outcomes in later life vary widely across individuals (Agrigoroaei and Lachman
2011). Efforts have been made to account for this variance, yielding evidence for explanatory effects of many modifiable lifestyle factors, such as exercise, smoking cessation, management of hearing loss (Livingston et al.
2017) and social activity engagement (Fratiglioni et al.
2004; McHugh Power et al.
2016; Zunzunegui et al.
2003; Kelly et al.
2017). Modification of such factors may be particularly critical among older adults with non-reversible risk factors for cognitive decline, such as APOE-ε4 genotype (Henderson et al.
1995). Social activity engagement in particular has a strong evidence base connecting it to improvement in (or maintenance of) cognitive function in later life. As a corollary, social isolation is robustly associated with cognitive decline and dementia among older cohorts (Cacioppo and Hawkley
2009; Yu et al.
2020; Kuiper et al.
2015).
Another such non-modifiable risk factor may be exposure to psychological trauma during the lifespan. Such traumatic exposure in the general population is common, with cumulative prevalence estimated at 70% (Benjet et al.
2016). From a developmental systems perspective, traumatic exposures represent environmental contexts, and can account for individual variation in outcomes such as health and cognitive functioning (Molenaar et al.
2013). Empirical evidence exists for an association between traumatic exposure and cognitive dysfunction or decline, e.g. in first responders (Levy-Gigi et al.
2014), and among specific clinical populations such as older adults with anxiety disorders, and breast cancer survivors, and those with a history of childhood traumatic exposure (Petkus et al.
2018; Kamen et al.
2017). Exposure to the Holocaust was also found to be associated with worse cognitive functioning (Barel et al.
2010) HPA axis dysfunction (which could in turn cause cognitive dysfunction) (Yehuda et al.
2005), and increased incidence of dementia in later life (Kodesh et al.
2019), a finding interpreted with recourse to the vulnerability hypothesis–that exposure to genocide had left participants more susceptible to later-life morbidity.
These studies describe a negative impact of traumatic exposure upon cognitive outcomes in later life. However, others have argued that those who survive early life traumatic events represent a “survivor” cohort, and may possess trait resilience, which could be protective against future morbidities (Barel et al.
2010). Findings from the Irish Longitudinal Study on Ageing (TILDA) show that individuals who experienced childhood sexual abuse have superior cognitive functioning relative to those without such history (Feeney et al.
2013), and in the face of worse mental health. Similarly, Tedeschi and Calhoun’s post-traumatic growth concept (Tedeschi and Calhoun
2004) suggests that positive mental changes can occur as a consequence of trauma, and this concept may offset the deleterious impact of trauma on physical health (Greenblatt Kimron et al.
2019).
The ageing Northern Irish population have collectively been exposed to a higher-than-average number of traumatic experiences, due to the historical period of the Troubles, a period of 30 years of civil unrest which impacted almost everybody in the state (Muldoon et al.
2005). Traumatic exposure due to the Troubles are likely to have occurred during early adulthood for the majority of the current older population of Northern Ireland (NI), which may be a ‘critical period’, in the language of life-course epidemiology (Kuh et al.
2003), more proximally linked to cognitive decline in later life than childhood traumatic experience (Burri et al.
2013). The legacy of the Troubles may in part explain the increased rates of mental health difficulties found in NI relative to the Republic of Ireland (RoI) (O’Reilly and Stevenson
2003).
However, little empirical evaluation is available of the hypothesis that experience of the Troubles has an association with health outcomes. Additionally, little evidence is available in relation to the impact of traumatic events in later life (Cook and Simiola
2018), which is troublesome since older adults may be more likely to experience traumatic events (such as bereavement or serious illness) than younger adults (Ogle et al.
2014). However, it has also been suggested that time moderates the impact of trauma on aspects of cognitive function, such as autobiographical memory (Wittekind et al.
2017), which may mean that current older adults in NI are relatively unaffected by the traumatic experiences of the Troubles, which would have peaked more than 40 years ago.
To offset adverse associations between traumatic exposure and dementia risk, protective habits such as social and leisure activity engagement have been advised (Kodesh et al.
2019). Since traumatic exposure is a non-modifiable risk factor for dementia, encouraging those who have been exposed to trauma to engage in protective activities may help to lower their dementia risk. Social and leisure activity engagement has been linked to lower risk of dementia (Fratiglioni et al.
2004) and better cognitive function in later life (McHugh Power et al.
2016), and as such social activity interventions have been successfully implemented to improve cognitive function in later life (Pitkala et al.
2011). To corroborate this recommendation specifically for populations with traumatic exposure, it is necessary to explore potential moderating effects of social activity engagement on the association between traumatic exposure and outcomes of interest. Thus the current analysis tests the following hypotheses: that retrospective reporting of exposures to the Troubles is positively associated with later-life memory functioning, and, following recommendations made by Kodesh (Kodesh et al.
2019), that this hypothesised association is moderated by social activity engagement. Previous findings indicate that traumatic exposure is associated with domains such as processing speed and executive function (Petkus et al.
2018). We focus our analysis on memory functioning since this is a cognitive domain with a strong ageing component, and since it is also a key marker of early dementia.
Discussion
The current study investigated whether exposure to the Troubles would be associated with memory functioning in older NI adults, and whether this association would be moderated by social activity engagement. Better performance in memory tasks was associated with higher levels of Troubles Exposure, and with higher levels of social activity engagement. The association between Troubles Exposure and memory was also moderated by social activity engagement, such that at low levels of social activity engagement, those with higher levels of Troubles Exposure did better on memory, while at higher levels of social activity engagement there was little association between Troubles Exposure and memory.
The pattern of main effects found is compatible with previous findings. Social activities may constitute a cognitively stimulating pastime (Scarmeas and Stern
2003), which could bolster cognitive function in the ageing brain, according to the “use it or lose it” hypothesis of cognitive functioning (Hultsch et al.
1999). Our (positive) main effect of Troubles exposure on memory functioning is also compatible with findings reported by Feeney and colleagues based on data from TILDA (Feeney et al.
2013) which showed that retrospective reporting of childhood sexual abuse was associated with superior cognitive functioning performance in later life. This is also compatible with Tedeschi and Calhoun’s theory of post-traumatic growth, that positive changes can occur after trauma as well as negative ones (Tedeschi and Calhoun
2004). Feeney and colleagues had controlled for social activity engagement and found no effect–in our findings we report a main effect (positive) of social activity engagement and Memory, but also that the effect of Troubles Exposure is different at high and low levels of social activity engagement.
The finding that participants with high levels of Troubles exposure had good outcomes in relation to memory functioning is at odds with other findings in the literature, however, which points to traumatic exposure predicting cognitive decline (Kamen et al.
2017; Barel et al.
2010; Petkus et al.
2017). There are two possible, interesting explanations to this discrepancy. It should be noted that these previous studies explored associations between childhood trauma and later cognitive functioning, while in the current study (where participants currently have a median age of 71, and experienced the height of the Troubles around 40 years ago) we describe an effect of adulthood trauma on later memory functioning. It is possible, as previously suggested (Burri et al.
2013), that adulthood and childhood trauma exert effects of differing magnitude on later life memory functioning. Alternatively, the nature of the traumatising phenomenon may be different–the childhood trauma and genocide trauma described in these previous studies is inarguably of an entirely different nature and higher degree than the Troubles. It is possible that the degree of traumatisation yielded by the Troubles was sufficient to confer an inoculation effect without yielding a deleterious impact on the brain health of survivors. Childhood trauma may directly impact brain health by exerting a deleterious impact on the developing brain, as outlined in the traumagenic neurodevelopmental model of schizophrenia (Read et al.
2001). This explanation fails to account for the pattern of results reported by Feeney and colleagues, however (Feeney et al.
2013). Further research into the impact of traumatising events at different lifestages on later-life brain health is warranted to explain these inconsistencies. It is worth considering that participants had to have capacity to provide consent to be included in NICOLA so those with serious cognitive impairment were by design excluded from the outset, biasing study results. However given that scores on the Mini-Mental State Examination included a cognitively impaired range it is unlikely that NICOLA failed to capture those with cognitive impairment.
We also considered whether the effect of Troubles exposure and age on memory are confounded in our retrospective yet cross-sectional analysis because of “critical” or “sensitive” period effects”. We controlled for demographic, health, and social variables in the current analysis, and inspected levels of multicollinearity (which was nowhere found to be a problem), and the effects of Troubles variables on memory outcomes is nonetheless observed. This association could be explained with recourse to the resilient survivor cohort effect described previously (Barel et al.
2010)–i.e. that those who survive traumatic exposure (and be more likely to be recruited into cohorts) are higher in trait resilience, which mitigates later risk of morbidity, a form of collider stratification bias (Richiardi et al.
2019). These findings may also be driven, in part, by pre-existing differences in memory function; people with better memory may recall more events from their lives, just as they recall more words in a memory test.
We focus on traumatic exposure in the current analysis, rather than the psychological effects of such traumatic exposure, which are also associated with cognitive decline (Cohen et al.,
2013). Post-traumatic Stress Disorder (PTSD) patients underperform in a range of cognitive domains, most notably memory, which seems to be driven by hippocampal volume differences between PTSD patients and healthy controls (Bremner and Narayan
1998). Disentangling the effects of traumatic exposure and related psychological impacts thereof upon cognitive functioning has proven difficult, historically, and longitudinal data would be required to comprehensively understand their relative impacts, something that can be done once later (planned) waves of NICOLA are completed. It will also be clarifying to consider the potential epigenetic mediators of the association between traumatic exposure and cognitive outcomes in the Northern Irish population, as is suggested by work in other populations (Cecil et al.
2016; Marzi et al.
2018).
At high levels of social activity engagement, Troubles Exposure exerted little effect on Memory, but at low levels of social activity engagement, those with higher levels of Troubles exposure outperformed those with lower levels of Troubles exposure. Kodesh had previously suggested that social activity engagement might offset the negative impact of traumatic exposure on dementia risk (Kodesh et al.
2019), but instead here we find that it there is no such negative impact to offset. Instead, it seems that for individuals who faced high levels of exposure to the Troubles, and thus high exposure to traumatic events during adulthood, memory functioning was preserved, possibly due to a resilience effect. This may have created a ceiling effect such that less variance was left for social activity engagement to account for–in individuals with lower levels of exposure to the Troubles, having a high level of social activity engagement mattered such that it was associated with better memory. Parsing the chain of risk events leading to positive outcomes in later life would help to determine the reality of the health impact of the Troubles, using a lifecourse epidemiological perspective to do so (Kuh et al.
2003).
Further inspiration can be taken from the life-course epidemiological perspective in relation to the “critical” or possibly “sensitive” period for cognitive decline; understanding precisely when events such as traumatic exposures can yield the largest effect on later-life cognitive functioning, by group comparison based on age at the peak of the Troubles, would also contribute to our understanding of this association. We had the data available to conduct this. Here our hypothesis was that traumatisation at different age groups (measured as age at self-rated worst point of the Troubles) would differentially impact memory functioning, yielding an effect of age at self-rated worst point of the Troubles upon memory functioning later in life. If, for instance, young adulthood is a critical or sensitive period for cognitive decline, then those participants who were young adults at the (self-rated) peak of the Troubles would fare worst in their later-life memory functioning. IF, on the other hand, middle adulthood were the critical or sensitive period, then those who were middle-aged during the self-rated peak of the Troubles would fare the worst.The results of our analyses (see Supplementary Materials 2) did not indicate that age at the self-rated worst point of the Troubles was predictive of memory functioning, nor did it moderate the impact of Troubles Exposure on memory functioning. Further research may be necessary to identify precisely the critical period for traumatic exposure during adulthood in relation to later-life cognitive functioning–from the current results this critical period does not appear to be during early/middle adulthood (our participants would have been aged 19–46 during the peak of the Troubles).
There are notable strengths worth highlighting in relation to the current study. NICOLA is a population representative study, with a large sample size and considerable sampling effort. A flexible statistical methodology was used which optimised the use of data despite the incomplete nature of the dataset. Data pertaining to a large, nationally representative group who were largely exposed to a period of civil unrest is potentially of use to life-course epidemiologists interested in broadening their frame of reference from childhood to early adulthood adversity. Relatedly, a strength of the current study is the exploration of traumatic exposure in a sample of non-military participants. This type of research is valuable to increase the heterogeneity of trauma research, which predominantly focuses on veterans and prisoners of war, meaning that women are under-represented (Schuitevoerder et al.
2013).
A limitation of the study is the cross-sectional nature of the data (although because of the retrospective nature of the questions, it is clear that Troubles Exposure temporally precedes social activity engagement and later-life memory functioning). Further, we cannot comment on the temporal nature of the relationship between social activity engagement and memory functioning, since these two were measured cross-sectionally, and it is possible that some reciprocal causation is indeed in effect here. The NICOLA study is a prospective cohort study so longitudinal data will be available in the near future for this sample. Re-exploration of our current research question with longitudinal data will be critical, since the impact of trauma over time may have a cumulative effect on cognitive function in later life (Schuitevoerder et al.
2013). Taking into account other potential moderators of the relationship between experience in the Troubles and health outcomes, such as conflict-related deprivation (O'Neill et al.
2015; O'Connor and O'Neill
2015), is also necessary.
To conclude, we report a moderating effect of social activity engagement and Troubles exposure on memory functioning in later life among those living in NI. Overall, those with a higher level of Troubles exposure did best on memory functioning, although there was no such effect among those with high levels of social activity engagement. Further research is required to fully elucidate the Troubles legacy as it relates to decline in later life in this cohort, and to identify particularly at-risk groups therein.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.