The association between social support and executive function in mid- to late-life within the context of attrition: Results from the Canadian Longitudinal Study on Aging

Abstract

Background. As the population over the age of 65 increases, a greater proportion of Canadians will be impacted by the effects of cognitive decline, necessitating research on risk factors that can influence cognitive outcomes. Social support has been identified as a potential area of intervention for the cognitive health of the aging population and may have a particularly strong association with key cognitive domains such as executive function. Executive function is an important cognitive domain in successful aging given its role in the tasks required for daily independent living. While there is largely consistent evidence for a positive cross-sectional association between social support and executive function, longitudinal studies have produced more mixed findings. As both social support and executive function play a role in attrition (i.e., participants dropping out of studies over time), longitudinal studies investigating the association between the two may be at greatest risk of increasingly selective, and possibly biased, samples. Developing a better understanding of how social support impacts executive function within the context of attrition is essential to maximizing public health efforts aimed at supporting the cognitive health of the aging population. Aim. The overall aim of this thesis was to investigate the association between functional social support and executive function in mid- to late-life and to explore how attrition may be impacting our knowledge of this relationship. Methods. Modified Poisson and logistic regression analyses utilized baseline (T1; 2012-2015), first (T2; 2015-2018) and second (T3; 2018-2021) follow-up data from the Comprehensive cohort of the Canadian Longitudinal Study on Aging (CLSA). Attrition was assessed based on participation at baseline only (T1), both T1 and T2, and at all timepoints (T1, T2, T3). As data collection for follow-up 2 (T3) occurred both pre- and post-implementation of COVID-19-related restrictions in Canada, this time point was divided into two cohorts (i.e., those who participated pre vs post COVID-19 restrictions). The 19-item Medical Outcomes Study-Social Support Survey was used to assess perceived availability of social support (overall and subtypes: emotional/informational, tangible, affectionate, positive social interactions). Executive function was calculated by averaging standardized scores across multiple cognitive tests. Final models were adjusted for sociodemographic, health, lifestyle, and additional social variables. Results. Study 1 (Chapter 4) investigated if baseline social support and executive function were associated with attrition and if the association between social support and executive function varied across those participating at one (T1), two (T1, T2), or three (T1, T2, T3) timepoints. After accounting for covariates, those with low social support, low executive function, or those missing data on any of the tests of executive function were significantly more likely to drop out of the study after baseline (T1) assessment and were significantly less likely to complete all three waves (T1, T2, T3) of the study. The cross-sectional association between low social support and low executive function at baseline was statistically significant for those who participated at baseline-only but not for those who participated at both T1 and T2 or at all three timepoints. This suggests that the association between functional social support and executive function at baseline varies in a meaningful way for those lost to attrition vs those who remained in the study, and that those with the strongest association between social support and executive function may be those most likely to be lost to follow-up. Attrition in longitudinal studies can lead to underestimating or overestimating the impact of risk factors given the cognitively healthier and more socially engaged participants who remain in the study over time. Study 2 (Chapter 5) investigated the longitudinal role of functional social support (T1) and its subtypes on executive function (T3) while exploring how attrition and individual characteristics may impact the results and conclusions generated from these analyses. The longitudinal association between low social support and low executive function was not statistically significant. Taken at face value, these results would seem to add to the conflicting, non-significant, findings regarding the long-term association between social support and cognitive function. Yet, when those who completed all assessments were compared cross-sectionally (T1) to those who had dropped out of the study, only those lost after baseline had a significant positive association between low social support and low executive function. These findings indicate that attrition may be introducing bias into our longitudinal results and, if we as researchers are not careful in our interpretation, into our conclusions about the value of functional social support for cognitive health. Study 3 (Chapter 6) investigated how the association between social support and executive function differs pre- vs post-COVID-19 restrictions (T3), with half of the sample completing their assessments in-person, pre-COVID-19, and the latter half via telephone interviews post-COVID-19 restrictions. While the cross-sectional (T3) association between social support and executive function pre-COVID-19 restrictions was only statistically significant for one subtype of support (i.e., emotional/informational), post-COVID-19 restrictions, overall social support, as well as positive social interactions, emotional/informational and affectionate social support were statistically significantly associated with executive function, such that higher social support was associated with higher executive function. It is unclear whether the significant relationship between social support and executive function post-COVID-19 restrictions reflects the changes in data collection procedures or the greater psychological and cognitive impacts of the COVID-19 pandemic and related changes to daily life. The COVID-19 pandemic will continue to have an undeniable impact on data analyses for the foreseeable future and analyses considering the long-term impact of social exposures on cognitive health should explicitly account for how the COVID-19 pandemic and related restrictions may impact our understanding of this association. Contributions. Understanding how different aspects of social support impact cognitive function—and, by proxy, independence and adaptability— is essential to public health efforts aimed at improving health outcomes for older adults through evidence-based social programs and policies. The results of this dissertation reveal that the longitudinal relationship between social support and executive function is complex and that conclusions in this research area may be influenced by attrition. This dissertation demonstrated that those with the strongest association between social support and executive function were also those most likely to be lost to follow-up. While the long-term impact of social support on cognitive health is best investigated using longitudinal studies, these samples are vulnerable to biases which can lead to generalizations that do not reflect the greater population. Of greatest concern is that those who are lost to these studies over time may be those who would benefit most greatly from cognitive research, given their poorer health. While what is demonstrated in this dissertation discusses the association between social support and executive function, the limitations of the data represented here have wide-spread implications for longitudinal research on aging. Future research considering the long-term impact of exposures on cognitive health over the lifespan should actively investigate how attrition may impact their samples and findings.