Instrumental activities of daily living (IADLs) refer to functional competencies that support independent functioning, including communicating via phone or email, managing finances, or completing household chores. Much research suggests that executive functioning (EF) (i.e., a set of higher-order cognitive processes that subserve planning, selection, and execution of goal-directed behaviors; Lezak et al., Reference Lezak, Howieson, Bigler and Tranel2012; Stuss & Knight, Reference Stuss and Knight2002; Suchy, Reference Suchy2015) is necessary for successful execution of IADLs (e.g., Bell-McGinty, Podell, Franzen, Baird, & Williams, Reference Bell-McGinty, Podell, Franzen, Baird and Williams2002; Boyle et al., Reference Boyle, Malloy, Salloway, Cahn-Weiner, Cohen and Cummings2003; Jefferson, Paul, Ozonoff, & Cohen, Reference Jefferson, Paul, Ozonoff and Cohen2006), due to its role in (a) reducing commission errors (Giovannetti, Mis, Hackett, Simone, & Ungrady, Reference Giovannetti, Mis, Hackett, Simone and Ungrady2021), and (b) faciliating successful “meta-tasking” (i.e, interleaving of tasks; Suchy, Reference Suchy2015). Consequently, EF represents an important aspect of neuropsychological evalution.
However, EF is not a completely stable trait. For example, Gamaldo and Allaire (Reference Gamaldo and Allaire2016) found that approximately 50% of older adults’ cognitive performance was not explained by between-person variability, but rather within-person variability. Although increases in intra-individual variability (i.e., greater cognitive fluctuations) can be indicative of cognitive declines (Gamaldo & Allaire, Reference Gamaldo and Allaire2016), even healthy older adults experience greater EF fluctuations than younger adults (Strauss, MacDonald, Hunter, Moll, & Hultsch, Reference Strauss, Macdonald, Hunter, Moll and Hultsch2019). Such fluctuations can be exacerbated by a variety of contextual factors, such as experience of pain, quality of sleep, and engagement in burdensome emotion regulation (e.g., Franchow & Suchy, Reference Franchow and Suchy2017; Higgins, Martin, Baker, Vasterling, & Risbrough, Reference Higgins, Martin, Baker, Vasterling and Risbrough2017; Holanda, de Almondes, & Almondes, Reference Holanda, de Almondes and Almondes2016; Niermeyer & Suchy, Reference Niermeyer and Suchy2020a).
Contextual Factors, Fluctuations in EF, and IADLs
Since EF is known to predict IADLs, it follows that factors that lead to EF fluctuations might also lead to fluctuations in IADLs. Indeed, considerable research supports this notion. For example, among older adults, chronic, acute, or fluctuating pain has been associated with poorer IADL performance (e.g., Eggermont et al., Reference Eggermont, Leveille, Shi, Kiely, Shmerling, Jones and Bean2014; Williams et al., Reference Williams, Tinetti, Kasl and Peduzzi2006) and future IADL declines (Thakral et al., Reference Thakral, Shi, Foust, Patel, Shmerling, Bean and Leveille2019). Similarly, sleep disorders (Hanewinckel et al., Reference Hanewinckel, Maksimovic, Verlinden, van der Geest, Hofman, van Doorn and Ikram2015; Spira, Chin-Edinborough, Wu, & Yaffe, Reference Spira, Stone, Rebok, Punjabi, Redline, Ancoli-Israel and Yaffe2014) and poor sleep quality have been related to poor IADL performance (e.g., Fung et al., Reference Fung, Martin, Chung, Fiorentino, Mitchell, Josephson and Alessia2012; Holfeld & Ruthig, Reference Holfeld and Ruthig2012) and future IADL impairment (e.g., Park, Buchman, Lim, Leurgans, & Bennett, Reference Park, Buchman, Lim, Leurgans and Bennett2014; Spira et al., Reference Spira, Covinsky, Rebok, Punjabi, Stone, Hillier and Yaffe2012).
In contrast to the relatively extensive research on the association of pain and sleep with daily functioning, only two studies to our knowledge examined the impact of burdensome emotion regulation on functionality, and both of these specifically focused on an emotion regulation strategy known as expressive suppression (ES; Niermeyer & Suchy, Reference Niermeyer and Suchy2020b; Suchy, Niermeyer, Franchow, & Ziemnik, Reference Suchy, Niermeyer, Franchow and Ziemnik2019). ES is used commonly and involves deliberate inhibition of emotional expressions (e.g., laughter, frowning) to facilitate socially appropriate conduct (Gross, Reference Gross1998). In contrast to other emotion regulation strategies like cognitive reappraisal, ES is thought to be particularly burdensome, partly because it typically occurs after an emotional response has been activated (Cutuli, Reference Cutuli2014). Additionally, while cognitive reappraisal dampens both the subjective experience of emotion and the concomitant physiologic arousal, ES accomplishes neither and may, at times, even heighten both (e.g., John & Gross, Reference John and Gross2004; Li et al., Reference Li, Wang, Fan, Zhu, Li, Zhang and Luo2017). Importantly, excessive recent engagement in ES can temporarily impact subsequent EF performance (e.g., Baumeister, Reference Baumeister2002; Baumeister, Schmeichel, & Vohs, Reference Baumeister, Schmeichel, Vohs, Kruglanski and Higgins2007; Szczygieł & Maruszewski, Reference Szczygieł and Maruszewski2015), but does not seem to impact other cognitive domains, as evidenced by experimental (e.g., Franchow & Suchy, Reference Franchow and Suchy2017; Schmeichel, Reference Schmeichel2007) and correlational research (Franchow & Suchy, Reference Franchow and Suchy2015; Niermeyer, Ziemnik, Franchow, Barron, & Suchy, Reference Niermeyer, Ziemnik, Franchow, Barron and Suchy2019; Niermeyer & Suchy, Reference Niermeyer and Suchy2020a, Reference Niermeyer and Suchy2020b; Suchy, Brothers, Mullen, & Niermeyer, Reference Suchy, Brothers, Mullen and Niermeyer2020). Additionally, similar to the impact of sleep and pain on IADLs, self-reported recent burden of ES is associated with subsequent slower dual-task walking (Niermeyer & Suchy, Reference Niermeyer and Suchy2020b) and less accurate performance on subsequently-administered office-based IADL tasks (Suchy et al., Reference Suchy, Ziemnik, Niermeyer and Brothers2019). Notably, in both studies, the association between ES and functionality was fully mediated by EF.
In summary, past research suggests (a) EF plays a role in performance of IADLs, (b) engagement in ES leads to subsequent temporary declines in EF, which over time present as EF fluctuations, and (c) engagement in ES impacts future IADL performance, presumably due to its impact on EF. Taken together, these findings imply that daily fluctuations in EF should lead to more frequent lapses in IADLs. Such an association was found in a recent study by Schmitter-Edgecombe, Sumida, and Cook (Reference Schmitter-Edgecombe, Sumida and Cook2020). This study utilized ecological momentary assessment (EMA)Footnote 1 to assess EF four times daily for 7 days and used the EMA performance as a predictor of (a) an office-based objective measure of functional status and (b) self-reported cognitive/functional lapses in daily life. Because EMA assessments are completed at home and are embedded within the context of other daily activities, the study captured not only the level of EF (i.e., the average across assessments), but also the daily EF fluctuations (i.e., the SD across assessments). The results showed that older adults with more fluctuations in EF self-reported experiencing higher levels of functional lapses in daily life. In contrast, those with higher average EF scores tended to score better on an objective lab-based measure of functional status; there was no association between average EMA scores of EF and self-reported functional lapses. These findings suggest that variability in EF, as compared to average EF performance, may play a more significant role in executing IADLs in daily life.
Limitations of Past Research
Prior studies have one salient limitation in that they relied on two methods of IADL assessment that likely do not adequately capture at-home functioning. Specifically, Schmitter-Edgecombe et al. (Reference Schmitter-Edgecombe, Sumida and Cook2020) and Suchy et al. (Reference Suchy, Ziemnik, Niermeyer and Brothers2019) assessed IADLs in controlled laboratory conditions, allowing for optimal performance under ideal circumstances. Such an approach fails to capture the complexity of daily life that typically requires the interleaving of daily tasks and demands. To address this limitation, Schmitter-Edgecombe et al. (Reference Schmitter-Edgecombe, Sumida and Cook2020) supplemented their office-based assessment with self-reports, which reflect individuals’ perception of their functioning at home. While this method may better capture the impact of interruptions and interfering state factors (e.g., Schmitter-Edgecombe et al., Reference Schmitter-Edgecombe, Sumida and Cook2020), self-reported IADLs have been shown to be inaccurate for at least a substantial subset of older adults (e.g., Suchy, Kraybill, & Franchow, Reference Suchy, Kraybill and Franchow2011).
The Present Study
The purpose of the present study was to address the main limitations of past research by replacing self-report and office-based IADL assessments with real-world daily IADL tasks completed at home over the course of three weeks. Using this highly ecological IADL measure, we aimed to (1) test the Schmitter-Edgecombe et al. (Reference Schmitter-Edgecombe, Sumida and Cook2020) assertion that daily EF variability is a better predictor of daily functioning than average EF performance, and (2) validate, in the home environment, our prior laboratory-based findings that variability in ES is an indirect contributor to IADL lapses, and that this effect is mediated by EF (Suchy et al, Reference Suchy, Ziemnik, Niermeyer and Brothers2019).
To accomplish these aims, we recruited older adults who agreed to complete IADL-like tasks (e.g., rescheduling an appointment, grocery shopping, filling out a reimbursement form) in the context of their daily lives across three weeks. Participants concurrently completed daily EMA to assess their average daily ES and EF and their daily variability in ES and EF. The reason for following our participants for three weeks was to capture daily fluctuations and associated real-world functional lapses. Based on past research, we hypothesized (1) greater variability in daily EF would relate to poorer performance on at-home IADL tasks, above and beyond average level of daily EF; (2) greater variability in ES would be associated with poorer performance on at-home IADL tasks; and (3) any significant association between ES variability and performance on at-home IADL tasks would be fully mediated by daily EF variability.
Method
Participants
Participants were 54 independently living older adults recruited as part of a larger study through the University of Utah’s Center on Aging database, the Osher Lifelong Learning Institute, and a Senior Expo. Inclusion criteria included being 60 years or older, having at least an 8th grade education, and being able to read/understand English.
Potential participants were excluded if they had (a)uncorrected hearing, vision, or motor impairments that would affect task performance; (b) a self-reported diagnosis of dementia, mild cognitive impairment, moderate to severe traumatic brain injury, stroke, seizure disorder, or other significant psychiatric or neurological disorder, as we are interested in how ES impacts EF and IADLs in neurologically and cognitively healthy older adults; or (c) missing data (a failure to complete baseline assessment; n = 1; a failure to complete daily tasks; n = 1). The final sample consisted of 52 older adults aged 60 to 95 (M = 69.5, SD = 6.35), primarily Caucasian (98%) and female (65.4%). See Table 1 for additional sample characteristics.
Demographic, cognitive, and affective characteristics of the sample
Note. * = z-score, where mean = 0; DRS = Dementia Rating Scale – 2nd Edition; TOPF = Test of Premorbid Functioning; GDS = Geriatric Depression Scale; SD = Standard Deviation; Baseline EF speed = composite of speeded scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery at baseline; Baseline EF errs = composite of error scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery. EMA EF Avg = average executive functioning performance across 18 days of assessment at home; EMA EF SD = standard deviation of executive functioning performance across 18 days of assessment at home; Baseline ES = self-reported expressive suppression on the Burden of State Emotion Regulation Questionnaire (B-SERQ) at baseline in lab; EMA ES Avg = average self-reported expressive suppression burden across 18 days of assessment at home; EMA ES SD = standard deviation of self-reported expressive suppression burden across 18 days of assessment at home; DAILIES = score based on performance of daily IADL tasks over 18 days.
Procedure
The study was approved by the University of Utah Institutional Review Board and complied with institutional research standards for human research in accordance with the Helsinki Declaration. Participants underwent a phone screening to determine eligibility. Those who qualified were invited for a baseline assessment at the University of Utah. Participants were consented before completing measures of EF and ES, and additional cognitive/psychological/psychophysiological testing as part of a larger study. At the end of the session, participants were given detailed instructions and sent home with daily tasks to complete six days a week over the next three weeks. Participants received $60 for the 6-hour baseline testing session, $4 for every daily task and $4 for every daily survey, for up to an additional $144. Interested participants received brief feedback on their personality, cognition, and psychiatric symptoms as assessed by screening measures, and community resources.
Measures
Characterizing the Sample
General cognitive status and premorbid intelligence were assessed using the Dementia Rating Scale – Second Edition (DRS-2; Jurica, Leitten, & Mattis, Reference Jurica, Leitten, Mattis and Lutz2001) and the Test of Premorbid Functioning (TOPF; Pearson Assessment, 2009). Depression was assessed using the 30-item self-report Geriatric Depression Scale (GDS; Yesavage, Reference Yesavage1988). See Table 1.
Daily Functioning
Daily IADLs were assessed using the Daily Assessment of Independent Living and Executive Skills (DAILIES) protocol, developed in our laboratory. The DAILIES consists of 18 tasks (e.g., rescheduling an appointment, filling out a reimbursement form; see supplementary material Appendix 1) for participants to complete over the course of three weeks. This is the first study to include the DAILIES protocol.
Participants were instructed to complete one task daily and provide their answer using a specific method of communication (e.g., participants were instructed to either leave a voicemail on the lab phone, email the lab account, or respond via postal mail). A variety of methods were chosen to increase ecological validity, though the method was standardized for each task. Participants were instructed to submit their responses during a specified time of day, thereby having to incorporate the completion of the DAILIES tasks within the rest of their daily routine. During baseline assessment in the laboratory, participants completed two practice tasks to ensure they understood the procedure. Most tasks took approximately 5 minutes to complete.
Daily tasks were scored based on timeliness and accuracy. Participants received one point each for completing the task on the correct day and the correct time of day (as evident per time stamps of phone messages, emails, postal stamp, etc.). Additional points were awarded based on performance accuracy, ranging from 1 to 7 points depending on task complexity. The sum of total scores earned on each task, comprising both timeliness and accuracy, was used in analyses and is referred to as “DAILIES” hereafter.
Ecological Momentary Assessment (EMA)
To assess daily fluctuations in ES and EF, participants completed a daily EMA survey on their smart phone or tablet. EMA has been shown to be appropriate and reliable with older adults (e.g., Schweitzer et al., Reference Schweitzer, Husky, Allard, Amieva, Péres and Foubert-Samier2017). Participants received a link to a Qualtrics survey Monday through Saturday at 7:00 PM (to be completed before bed), inquiring about the effort expended on ES over the past 24 hours. Participants were asked two questions about how much they needed to suppress during the day and how much effort they exerted to do so (Appendix 2). From the daily ES data, we calculated two overarching variables: (a) each participant’s average daily ES burden across the three weeks (Cronbach’s alpha = .863), and (b) variability in daily ES (i.e., standard deviation) across the three weeks. Hereafter, these variables are referred to as EMA “ESAvg” and “ESSD,” respectively.
After survey completion, participants completed two EF tasks, utilizing Stroop (Stroop, Reference Stroop1935) and digit span paradigms (Wechsler, Reference Wechsler2008). The Stroop task was comprised of names of common furniture (baseline condition) or colors (inhibition condition) written in red, blue, or yellow ink. Below each word, participants selected (from three available colors) the color of ink the word was written in. The baseline condition was administered first and lasted 20 seconds, followed by a 20-second inhibition condition. The difference between baseline and inhibition trials was computed to isolate inhibition, both for the total number of correct responses and the number of errors daily. We then computed the means and SDs of these variables across the 18 days. To combine the numbers of responses and errors into one variable, we converted these means and SDs to z-scores to place them on the same metric, generating Stroop mean and Stroop SD scores. The digit span task was comprised of 12 strings of numbers increasing in length by one digit every 2 trials (ranging from 3 to 8). Participants had a standard amount of time to view each string (adjusted for number of digits) and then typed the digits in reverse order. The number of items answered correctly each day was recorded and an 18-day mean and SD were computed. To quantify participants’ overall EF performance, Stroop and digit span means and SDs were converted to z-scores (based on the sample) to place them on the same metric and then combined into a single EMA “EFAvg” and “EFSD” for use in analyses. Correlation coefficients between EF composite scores and corresponding Stroop and digit span scores all exceeded .84, suggesting good internal consistency.
Validation of EMA Assessment
To validate the EMA EF tasks, we used a widely accepted battery of EF tests, the Delis-Kaplan Executive Function System (D-KEFS; Delis, Kaplan, & Kramer, Reference Delis, Kaplan and Kramer2001). Consistent with past research (e.g., Franchow & Suchy, Reference Franchow and Suchy2017; Suchy et al., Reference Suchy, Brothers, Mullen and Niermeyer2020), we used the Letter-Number Switching, three Design Fluency conditions, three Verbal Fluency conditions, and the Color-Word Interference and Interference-Switching conditions to generate two EF composite scores (i.e., means across subtests), one based on speed and one based on accuracy. Raw scores from relevant conditions in all subtests were converted to scaled scores based on the norms for adults aged 60-69; using this single age-band for normative comparison allowed us to generate clinically interpretable composite scores while not correcting for age. These composites are referred to as Baseline “EFspeed” and “EFerrs.”
To validate the EMA assessment of ES, we used the Burden of State Emotion Regulation Questionnaire (B-SERQ) at baseline. The B-SERQ is a 14-item self-report questionnaire that assesses the burden of naturally-occurring ES over the 24 hours prior to testing. It has been validated by prior research, with Cronbach’s alphas ranging from .81 to .91 (Franchow & Suchy, Reference Franchow and Suchy2015; Niermeyer et al., Reference Niermeyer, Ziemnik, Franchow, Barron and Suchy2019; Niermeyer & Suchy, Reference Niermeyer and Suchy2020a; Suchy et al., Reference Suchy, Ziemnik, Niermeyer and Brothers2019, Reference Suchy, Brothers, Mullen and Niermeyer2020). Due to administration error, three participants received a version of the B-SERQ that asks about ES burden “today” instead of “in the past 24 hours.” The sample means for “24 hours” and “today” were similar (M = 9.67 and 8.78, respectively) and results of the analyses were unchanged when the three “today” participants were removed. Thus, all participants were used in analyses. We refer to the total B-SERQ scores as “Baseline ES.”
Results
Preliminary Analyses
Data Distribution
The data were examined for normality using Q-Q plots, skewness, and kurtosis. Three variables were slightly skewed (skewness > 1.0 and < 2.5), with 1-2 outliers identified for each variable. Because these outliers reflected performance in the low average range, they were not excluded from analyses. Instead, we used the “boot” package in R to use bootstrapping, which minimizes the impact of such outliers while still including them. Additionally, all cognitive measures evidenced a steady practice effect across all 18 days. The practice effect was not factored into any analyses.
Descriptive Statistics and Zero-Order Correlations
See Table 1 for descriptive statistics on the demographics, as well as dependent and independent variables. Zero-order correlations between demographics and dependent and independent variables are presented in Table 2.
Correlations among lab-based and home-based EF and ES variables and demographic variables, using bootstrapping
Note. * = 95% confidence intervals generated with bootstrapping (n = 1000) that do not cross 0, indicating significance; GDS = Geriatric Depression Scale total raw score; Baseline EF speed = composite of speeded scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery at baseline; Baseline EF errs = composite of error scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery; Baseline ES = self-reported expressive suppression on the Burden of State Emotion Regulation Questionnaire (B-SERQ) at baseline in lab; EMA EF Avg = average executive functioning performance across 18 days of assessment at home; EMA EF SD = standard deviation of executive functioning performance across 18 days of assessment at home; EMA ES Avg = average self-reported expressive suppression burden across 18 days of assessment at home; EMA ES SD = standard deviation of self-reported expressive suppression burden across 18 days of assessment at home; DAILIES = score based on performance of daily IADL tasks over 18 days.
Validation of EMA Variables
As a prerequisite to principal analyses, we examined whether the EMA variables (i.e., daily measurements of ES and EF) represented the intended constructs by examining correlations between the EMA variables assessed at home and the relevant baseline variables assessed in the lab. As seen in Table 3, EMA EF Avg was associated with both baseline EF speed and baseline EF errs , which suggests that the level of EF performance at home does in fact tap into the EF construct as assessed via previously validated measures. In other words, individuals who performed better on measures of EF in the lab were likely to perform better, on average, at home. Additionally, EMA EF SD was associated with baseline EF errs , suggesting that individuals who make more errors under ideal conditions (i.e., within the structure of office-based assessment) are more likely to demonstrate increased variability in their EF performance when confronted with the complexities of daily life.
Correlations among lab-based and home-based EF and ES variables, using bootstrapping
Note. * = confidence intervals generated with bootstrapping (n = 1000) that do not cross 0, indicating significance
a = for the speed composite, higher scores reflect better performance
b = for the Error composite, lower scores mean more errors
Baseline EF speed = composite of speeded scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery at baseline
Baseline EF errs = composite of error scores on four timed subtests of the Delis-Kaplan Executive Function System (D-KEFS) battery at baseline
Baseline ES = self-reported expressive suppression on the Burden of State Emotion Regulation Questionnaire (B-SERQ) at baseline in lab
EMA EF Avg = average executive functioning performance across 18 days of assessment at home
EMA EF SD = standard deviation of executive functioning performance across 18 days of assessment at home
EMA ES Avg = average self-reported expressive suppression burden across 18 days of assessment at home
EMA ES SD = standard deviation of self-reported expressive suppression burden across 18 days of assessment at home
As also seen in Table 3, EMA ES SD correlated with baseline ES, suggesting that individuals who reported engaging in less ES in the 24 hours prior to baseline visit had less variability in ES at home. Conversely, EMA ES Avg was not significantly associated with baseline ES, consistent with the notion that reports of acute ES in the past 24 hours are not reflecting stable long-term ES levels. Beyond correlations with baseline ES, the relationships between the EMA ES and EF variables were consistent with prior research. Specifically, EMA ES SD was significantly associated with EMA EF SD and EF Avg suggesting that variability in ES at home is related to both variability and level of EF performance at home. This mirrors prior research in which recent ES was associated with EF errors in the lab (Suchy et al., Reference Suchy, Brothers, Mullen and Niermeyer2020), a relationship that was also replicated in the present study. Since both baseline EF errs and baseline EF speed were associated with age, we also ran partial correlations controlling for age and the results did not change.
Principal Analyses
To address the first hypothesis that variability in EF at home (i.e., EMA EF SD ) predicts daily task performance (i.e., DAILIES) beyond the level of EF performance at home (i.e., EMA EF Avg ), we ran a hierarchical linear regression with bootstrapping (n = 1000) in RStudio Version 1.1.462 using the “boot” package. DAILIES was entered as the dependent variable, with EMA EF Avg and EMA EF SD entered as predictors on Steps 1 and 2, respectively. As seen in Table 4 Model 1, contrary to expectation, EMA EF SD did not predict daily IADL performance (i.e., DAILIES) beyond EMA EF Avg . When reversing the order of variable entry (Table 4 Model 2), EMA EF SD predicted DAILIES by itself in Step 1, with EMA EF Avg predicting DAILIES above and beyond EMA EF SD. These findings suggest that the level of daily EF performance is a stronger predictor of daily IADL functioning than is EF variability, contrary to prediction.
Hierarchical linear regressions of daily EF and ES predicting daily IADL performance (DAILIES)
Note.** = significance at the p < .01 level
* = significance at the p < .05 level
EMA EF Avg = average executive functioning performance across 18 days of assessment at home
EMA EF SD = standard deviation of executive functioning performance across 18 days of assessment at home
EMA ES Avg = average self-reported expressive suppression burden across 18 days of assessment at home
EMA ES SD = standard deviation of self-reported expressive suppression burden across 18 days of assessment at home
To address the second hypothesis that variability in ES is related to poorer performance of IADLs at home, we again ran a hierarchical linear regression with bootstrapping (n = 1000), with DAILIES as the dependent variable. Paralleling our approach to the first hypothesis, we entered EMA ES Avg and EMA ES SD in Steps 1 and 2, respectively (Table 4 Model 3). The results confirmed that EMA ES SD was not only a significant predictor of DAILIES, but that it predicted DAILIES above and beyond EMA ES Avg , which was not significant. These results suggest that variability in ES is more important in the execution of daily tasks than average level of ES.
The third hypothesis was that any significant associations between ES variability and performance on daily tasks would be mediated by EF variability. However, since EF level was a stronger predictor of DAILIES than was EF variability, we altered our initially planned model, examining instead whether the relationship between EMA ES SD and DAILIES was mediated by EMA EF Avg . To do so, the “mediation” package in R was used in addition to the bootstrap function (n = 1000). One model was created where DAILIES was regressed onto EMA ES SD , another model had EMA EF Avg regressed onto EMA ES SD , and finally a model where DAILIES was regressed onto EMA ES SD and EMA EF Avg . See Figure 1. Per the results of the first model, EMA ES SD significantly predicted DAILIES. However, once EMA EF Avg was added into the model, it alone significantly predicted DAILIES, while EMA ES SD was no longer significant. These findings, along with the significant indirect effect (CI = -.315, −.02), are consistent with EMA EF Avg fully mediating the relationship between daily ES variability and daily IADL performance.
Mediation model analysis. EMA ES SD and EMA EF Avg were both significantly related to each other and to DAILIES (p < .05). However, once EMA EF Avg was added into the model, EMA ES SD was no longer as significant predictor of DAILIES; further analysis revealed that EMA EF Avg fully and significantly mediated the relationship.
DAILIES = score based on performance of daily IADL tasks.
EMA EF Avg = average executive functioning performance across 18 days of assessment at home.
EMA ES SD = standard deviation of self-reported expressive suppression burden across 18 days of assessment at home.
Discussion
The present study examined whether variability in ES and EF would be associated with real-world daily IADL functioning among older adults. The key findings are (a) average daily level of EF was a better predictor of daily IADL performance than daily EF variability, contrary to prediction; (b) daily ES variability was significantly negatively associated with daily IADL performance; and (c) the association between daily ES variability and daily IADL performance was fully mediated by average daily level of EF. These findings extend prior research by demonstrating that daily variability in ES has an impact on actual daily functioning at home, both in terms of cognition and IADLs. The present study was the first to examine the associations among these variables using EMA methodology in conjunction with real-world daily IADL performance.
Our finding that average EF was more predictive of daily task performance than EF variability was inconsistent with prior research (Schmitter-Edgecombe et al., Reference Schmitter-Edgecombe, Sumida and Cook2020). There are several possible explanations for this discrepancy. First, the Schmitter-Edgecombe study examined self-reported functional lapses instead of objective daily performance. It could be participants are more aware of lapses in functioning than consistently poor functioning. Consequently, individuals who experience more variability likely self-report more difficulties than do individuals who perform more consistently, regardless of their actual level of performance. Second, given that average EF (i.e., EMA EF Avg ) and EF variability (i.e., EMA EF SD ) are moderately negatively correlated (Table 3), one could conceptualize average EF as encompassing both the variability in EF and one’s overall EF ability. Since individuals with increased variability in EF tend to have lower average EF (Table 3), and those with poorer EF ability generally have poorer IADL performance (e.g., Jefferson et al., Reference Jefferson, Paul, Ozonoff and Cohen2006), it follows that individuals who have both greater variability in EF and poorer EF capacity would have the lowest mean EF performance when assessed across days. These individuals would then be at the greatest risk of poorer IADL performance.
The present study was also the first to confirm that ES variability across days relates to daily variability in EF and to daily IADL performance. The association between ES and EF, in more general terms, has been well established. For example, previous research has established that ES shares neural underpinnings with EF (most notably frontal-lobe networks; e.g., Hermann, Bieber, Keck, Vaitl, & Stark, Reference Hermann, Bieber, Keck, Vaitl and Stark2014), and that those who perform better on measures of EF also suppress emotions more effectively (Gyurak, Goodkind, Kramer, Miller, & Levenson, Reference Gyurak, Goodkind, Kramer, Miller and Levenson2012). Additionally, like EF, ES is important for successful navigation of daily social situations, such as regulating negative emotions during conflict (Thomson, Overall, Cameron, & Low, Reference Thomson, Overall, Cameron and Low2018). However, likely due to the overlap between the two constructs, ES also appears to transiently deplete EF and lead to transiently poorer EF performance (e.g., Baumeister et al., Reference Baumeister, Schmeichel, Vohs, Kruglanski and Higgins2007; Franchow & Suchy, Reference Franchow and Suchy2017). Thus, our findings that variability in ES is correlated with both mean EF and the variability in EF are consistent with prior research. Notably, we cannot establish causation from the present results. Daily fluctuations in ES engagement may have a corresponding effect on EF variability, thereby also having an impact on overall level of EF. Alternatively, individuals who have lower mean EF may be more likely to exhibit fluctuations in EF and concomitant lapses in emotion regulation. Indeed, poorer EF is often associated with less frequent and effective use of cognitive reappraisal (McRae, Jacobs, Ray, John, & Gross, Reference McRae, Jacobs, Ray, John and Gross2012), necessitating greater reliance on ES. This lack of clear causality also has ramifications for the interpretation of mediation results, since definitive conclusions about mediation can be made only if the mediator (in this case, EF) is measured after the predictor (in this case, ES). Thus, future research should examine these associations either in an experimental design or by temporally separating the measurement of ES and EF.
The present study took a novel approach to measuring real-world IADL functioning by having participants complete IADL tasks at home as part of their daily routine. While there has been interest in at-home IADL evaluations for clinical purposes (e.g., Zilbershlag & Josman, Reference Zilbershlag and Josman2019), typical IADL research primarily utilizes self-report questionnaires, or lab-based measures with discrete IADL tasks. Although such measures resemble tasks people complete in their homes, there is evidence that real-world IADL functioning is actually better predicted by measures of EF than by lab-based IADL tasks (Ziemnik & Suchy, Reference Ziemnik and Suchy2019). To address these limitations, a few studies have attempted to target real-world functioning. For example, one study presented participants with a list of tasks (e.g., making lunch, changing a light bulb, demonstrating their medication routine) to complete at home during a brief observation (less than 2 hours; Weakley & Schmitter-Edgecombe, Reference Weakley, Weakley and Schmitter-Edgecombe2019). Although this approach allows participants to complete tasks they would complete regardless of being in a study, the single observation period precludes measurement of daily variation or the capacity to spontaneously interleave study tasks with other daily routines.
To facilitate assessment of participants’ ability to complete IADLs in a natural and representative manner, researchers have recently begun looking towards utilizing Smart Home technology, such as sensors and videos that capture older adults’ daily functioning (Jekel, Damian, Storf, Hausner, & Frolich, Reference Jekel, Damian, Storf, Hausner and Frolich2016). However, this approach is costly and invasive, making it impractical for many researchers and unattractive to many participants. Additionally, there may be variation in which IADLs a participant chooses to complete, particularly if they have a partner (e.g., one partner manages finances, the other prepares meals). To avoid inconsistency across participants, some studies have instead focused on specific IADLs, such as actual at-home medication management (e.g., Pasina et al., Reference Pasina, Brucato, Falcone, Cucchi, Bresciani, Sottocorno and Nobili2014; Suchy et al., Reference Suchy, Ziemnik, Niermeyer and Brothers2019; Thiruchselvam et al., Reference Thiruchselvam, Naglie, Moineddin, Charles, Orlando, Jaglal and Tierney2012). Obviously, however, such studies are limited in that they cannot necessarily comment on performance of other IADLs. The present study, by requiring participants to complete a variety of IADL tasks that encompass multiple IADL categories (e.g., shopping, communication, transportation) and to do so in the context of their daily lives, offers a unique insight into factors that impact participants’ completion of various IADLs without being overly invasive or cost-prohibitive.
Limitations
First, the present sample was fairly homogenous (e.g., healthy, well-educated, Caucasian). Since factors like education are often related to EF (Dorbath, Hasselhorn, & Titz, Reference Dorbath, Hasselhorn and Titz2013), it is unclear how the results would translate to more heterogenous samples with less education. Additionally, it would be important to examine whether present results would replicate among individuals with cognitive impairment or poor physical or mental health. Furthermore, there are cultural considerations. For example, ES is used commonly in some Asian cultures and does not seem to have the same negative effect on cognition, perhaps due to extensive practice (e.g., Chen, Burton, & Bonanno, Reference Chen, Burton and Bonanno2020; Soto, Perez, Kim, Lee, & Minnick, Reference Soto, Perez, Kim, Lee and Minnick2011). Consequently, ES engagement may not impact EF or IADL performance in such populations, and replication with more diverse samples is warranted.
Second, while we took care to develop tasks that resemble typical IADLs in daily life, this is the first study to use the DAILIES protocol. Normative data for the DAILIES do not exist, nor has it been validated with other measures of daily functioning. Consequently, it is not clear what constitutes normatively “good” or “bad” performance. Additionally, although we instructed participants to complete tasks independently, participants may have received help from partners or family. Alternatively, participants may have had difficulties successfully integrating our tasks into their normal daily routines or may have not put forth the effort they would have on tasks with real-world ramifications. That said, participants received $4 per completed task, making the benefit relatively comparable to the financial benefits experienced in daily life, such as canceling appointments to avoid penalties or filling out rebate forms to receive discounts. Still, further validation and norming of the DAILIES is needed.
Third, our EMA measures of EF and ES have not been previously validated. However, we used well-validated EF paradigms, i.e., the Stroop task and digit span (Delis et al., Reference Delis, Kramer, Kaplan and Holdnack2004; Wechsler, Reference Wechsler2008) and questions from a well-validated ES measure (Franchow & Suchy, Reference Franchow and Suchy2015). Importantly, we found expected associations between the EMA EF variables and previously validated laboratory measures. Additionally, unlike typical EMA research, our EMA sampling was done only once a day, precluding detection of fluctuations during the day. Finally, while our sample size was larger than similar studies using EMA data (Schmitter-Edgecombe et al., Reference Schmitter-Edgecombe, Sumida and Cook2020), it was still modest, which limited statistical power. However, repeated EMA assessment across 18 days served to strengthen the reliability.
Future Directions
Future research should examine temporal associations between ES, EF, and IADLs to allow better causal determinations, as well as to determine whether ES on a given day increases the risk for lapses that day, or in the subsequent days. Differences in patterns across time should also be examined to determine what types of variability are most detrimental (e.g., consistent variability throughout the study vs. variability on days or weeks vs. variability through the day). Finally, other factors known to deplete EF and cognitive functions (e.g., pain, sleep; see Weizenbaum, Torous, & Fulford, Reference Weizenbaum, Torous and Fulford2020, for a review), and cognitive fatigue (Blain, Hollard, & Pessiglione, Reference Blain, Hollard and Pessiglione2016) should also be examined to determine what impact they may have on daily fluctuations in EF and IADLs.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1355617721001156
Acknowledgements
The authors wish to acknowledge Ty McKinney for his extensive contributions to the development and implementation of the EMA tasks.
Financial support
This work was supported by the first author’s professional development fund from the University of Utah.
Conflicts of Interest
None.
