Statement of Research Significance
Research Question(s) or Topic(s): To determine the extent to which cognition mediates the relationship between depression severity and everyday functioning in older adults with treatment-resistant late-life depression (TR-LLD). Main Findings: We found that depression severity in older adults with TR-LLD was related to greater impairment in immediate memory, attention, and executive functioning as well as worse everyday functioning. Attention and executive functioning partially mediated the relationship between depression severity and self-reported everyday functioning, while immediate memory partially mediated the relationship between depression severity and performance-based everyday functioning. Study Contributions: This study expands on an existing body of literature demonstrating that cognition mediates the relationship between depression severity and everyday functioning by using a sample of older adults with higher levels of depression severity. We also examined the relationship using multiple methods of assessing everyday functioning, further clarifying the nature of the relationship.
Introduction
Late-life depression (LLD), defined as depression occurring in adults over the age of 60 regardless of age of symptom onset, is one of the most common conditions among geriatric patients. Approximately 28% of older adults report experiencing at least some depressive symptoms while 13% meet criteria for major depressive disorder globally (Abdoli et al., Reference Abdoli, Salari, Darvishi, Jafarpour, Solaymani, Mohammadi and Shohaimi2022; Hu et al., Reference Hu, Zhao, Wu, Li, Luo, Yang and Yang2022). Of those with LLD, 33% have treatment-resistant LLD (TR-LLD), meaning their mood has not improved to remission after two or more adequate trials (i.e., use of a dose considered to be effective for treating depression and long enough duration to determine efficacy) of antidepressant medications (Göke et al., Reference Göke, McClintock, Mah, Rajji, Lee, Nestor, Downar, Noda and Blumberger2024). As the leading cause of psychiatric morbidity, the impact of LLD is immense, with cascading negative impacts on cognition, suicide risk, functioning, and response to medical treatments; all of which are exacerbated in those with TR-LLD (Göke et al., Reference Göke, McClintock, Mah, Rajji, Lee, Nestor, Downar, Noda and Blumberger2024; Valiengo et al., Reference Valiengo, Stella and Forlenza2016).
Lower levels of everyday functioning, including difficulty completing instrumental activities of daily living (IADL) such as managing medications and finances, is associated with higher depressive symptoms and LLD (Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Numbers et al., Reference Numbers, Jang, Brodaty, Sachdev, Draper and Reppermund2022). This association has been found using both self- and informant-reported ratings of IADL (Bombin et al., Reference Bombin, Santiago-Ramajo, Garolera, Vega-González, Cerulla, Caracuel, Cifuentes, Te Bascarán and Bobes2012; Brown et al., Reference Brown, Liu, Sneed, Pimontel, Devanand and Roose2013; Burton et al., Reference Burton, O’Connell and Morgan2018; Divers et al., Reference Divers, Robinson, Miller, Davis, Reed and Calamia2022; Hybels et al., Reference Hybels, Pieper and Blazer2009; Numbers et al., Reference Numbers, Jang, Brodaty, Sachdev, Draper and Reppermund2022), which are believed to tap into one’s beliefs about functional ability based on cumulative experiences, as well as performance-based measures of IADL functioning (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011), which assess capacity to apply problem solving skills in the moment to complete a task (Schmitter-Edgecombe et al., Reference Schmitter-Edgecombe, Parsey and Cook2011). These findings highlight that depression impacts both subjective evaluation of everyday functioning as well as capacity to carry out activities.
LLD/TR-LLD is also associated with cognition cross-sectionally, with worse cognitive functioning in the domains of executive functioning, visuospatial functioning, and processing speed being most prominent (Butters et al., Reference Butters, Whyte, Nebes, Begley, Dew, Mulsant, Zmuda, Bhalla, Meltzer, Pollock, Reynolds and Becker2004; Göke et al., Reference Göke, McClintock, Mah, Rajji, Lee, Nestor, Downar, Noda and Blumberger2024; Shimada et al., Reference Shimada, Park, Makizako, Doi, Lee and Suzuki2014). LLD, particularly among those with onset of first episode occurring after the age of 60 (late-onset LLD), has been longitudinally associated with steeper declines in verbal abilities and delayed memory (Ly et al., Reference Ly, Karim, Becker, Lopez, Anderson, Aizenstein, Reynolds, Zmuda and Butters2021). Although a consensus has yet to be reached on the temporal relationship between depression and cognitive impairment, a growing body of literature suggests depression precedes cognitive decline in older adults (Butters et al., Reference Butters, Young, Lopez, Aizenstein, Mulsant, Reynolds III, DeKosky and Becker2008; Zahodne et al., Reference Zahodne, Stern and Manly2014). Some previous work has even posited that LLD may represent a prodromal stage of Alzheimer’s disease and related dementias (ADRD) rather than a simple risk factor (Panza et al., Reference Panza, Frisardi, Capurso, D’Introno, Colacicco, Imbimbo, Santamato, Vendemiale, Seripa, Pilotto, Capurso and Solfrizzi2010). Based on the growing body of literature supporting depressive symptoms manifesting prior to cognitive decline, it is plausible that cognition mediates the performance between depression and everyday functioning (Mcalister et al., Reference Mcalister, Schmitter-Edgecombe and Lamb2016). In fact, several previous studies have found memory, executive functioning, and processing speed to fully mediate the relationship between depression and performance-based everyday functioning (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011).
A primary limitation of the aforementioned studies is that their samples, on average, experienced low levels of depressive symptoms. Average scores on the Center for Epidemiologic Studies Depression Scale (CES-D) ranged from 5.2 to 7.6 (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011), with a cutoff score of 16 on the CES-D being indicative of clinically significant but mild depressive symptoms (Lewinsohn et al., Reference Lewinsohn, Seeley, Roberts and Allen1997). Additionally, they operationalized everyday functioning solely through performance-based measures without use of self-/informant-reported ratings of functioning (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011). The aim of the present study was to examine the extent to which the relationship between depression severity and everyday functioning is mediated by cognition (i.e., executive function, attention, language, memory, and visuospatial domains) in a sample of older adults with TR-LLD using both performance-based and self/informant-reported ratings of everyday functioning. Moderate to severe depressive symptoms are associated with additional non-cognitive factors such as pain (Lin et al., Reference Lin, Yen, Chen and Chen2014) and physical disability (Lenze et al., Reference Lenze, Rogers, Martire, Mulsant, Rollman, Dew, Schulz and Reynolds2001) that may influence the relationship between depression and everyday functioning. Given the increased severity of depressive symptoms in the current sample, we hypothesized that cognition will only partially mediate the relationship between depression severity and everyday functioning as measured by both self/informant-rated and performance-based measures of everyday functioning.
Methods
Participants
The current study uses baseline data from the Optimizing Depression Treatment in Older Adults (OPTIMUM) study, which was a Patient-Centered Outcomes Research Institute (PCORI) funded multi-center clinical trial (Lenze et al., Reference Lenze, Mulsant, Roose, Lavretsky, Reynolds, Blumberger, Brown, Cristancho, Flint, Gebara, Gettinger, Lenard, Miller, Nicol, Oughli, Pham, Rollman, Yang and Karp2023), and the “OPTIMUM-Neuro” study which was an NIMH funded observational study that recruited from OPTIMUM, conducting a more extensive neurocognitive battery as well a series of biomarker assessments to monitor cognitive trajectories of individuals with a history of TR-LLD. Participants were recruited at one of five data collection sites including the University of Pittsburgh, Washington University in St Louis, the University of California – Los Angeles, the Centre for Addiction and Mental Health in Toronto, and Columbia University. Inclusion criteria for OPTIMUM included age 60 years or older, a diagnosis of Major Depressive Disorder with a current major depressive episode and a failure to remit following a minimum of two adequate antidepressant trials. Exclusion criteria were severe mental illness, dementia at the time of enrollment, unstable physical illness, uncorrectable sensory impairment, and high suicide risk at the time of enrollment. Participants were recruited directly from OPTIMUM to OPTIMUM-Neuro with no additional inclusion or exclusion criteria. Of note, given that some participants achieved remission from their depressive symptoms after completing some or all of the treatment arms in OPTIMUM, a current major depressive episode was not required for enrollment in OPTIMUM-Neuro.
Protocol approvals and participant consent
Both the OPTIMUM and OPTIMUM-Neuro protocols were approved by the Institutional Review Boards at the University of Pittsburgh, Washington University in St Louis, the University of California – Los Angeles, the Centre for Addiction and Mental Health in Toronto, and Columbia University and were completed in accordance with the Helsinki Declaration. Participants were provided with a detailed description of protocol procedures prior to study staff obtaining their written informed consent for participation.
Assessment of depression
Following study enrollment, depression was assessed using the Montgomery–Åsberg Depression Rating Scale (MADRS), a 10-item interview administered by a trained assessor designed to be sensitive to changes in depressive symptoms over time (Montgomery & Asberg, Reference Montgomery and Asberg1979). Items are rated from 0 – 6 with a total possible score of 60, with higher scores indicating higher levels of depressive symptoms. Total scores ranging from 7 – 19 indicate mild depression, 20 – 34 moderate depression, and 35 and higher severe depression (Müller et al., Reference Müller, Himmerich, Kienzle and Szegedi2003). The MADRS possesses strong psychometric properties including good internal reliability, test–retest reliability, and inter-rater reliability (Galinowski & Lehert, Reference Galinowski and Lehert1995; Montgomery & Asberg, Reference Montgomery and Asberg1979).
Assessment of everyday functioning
Everyday functioning was assessed using the Performance Assessment of Self-Care Skills (PASS; Holm & Rogers, Reference Holm, Rogers and Hemphill-Pearson2008) and the Everyday Cognition Scale – 12 item (ECog-12) self- and informant-rated questionnaires (Tomaszewski Farias et al., Reference Tomaszewski Farias, Mungas, Harvey, Simmons, Reed and DeCarli2011). The PASS is a standardized performance-based assessment of capacity to complete activities in four primary functional domains: functional mobility, personal self-care, IADL, and IADL with cognitive emphasis (C-IADL). The current study utilizes subtests from the C-IADL domain including shopping, bill paying, checkbook balancing, mailing bills, and medication management to create a total overall score. Participants were observed completing each subtest by a trained assessor, who provided cued assistance as needed. The assessor rates the level of preclinical disability using a standardized scoring system, with more cues equating to worse everyday functioning.
The ECog-12 is a self- and informant-rated questionnaire that measures change over time in overall functional ability. Participants and their informants are asked to rate the degree of change in the participant’s everyday functioning over the past 10 years. Degree of change is rated on a scale from 1 (better performance or no change) to 4 (consistently much worse performance) for each of the 12 items. Participants and their informants may also indicate if an item is not applicable to them or if they don’t know about the degree of change. An overall mean score is calculated using only items rated 1 – 4, with a higher score denoting a greater level of decline in everyday functioning (Tomaszewski Farias et al., Reference Tomaszewski Farias, Mungas, Harvey, Simmons, Reed and DeCarli2011).
Assessment of cognition
Baseline cognition was measured as part of the OPTIMUM-Neuro protocol. Premorbid cognitive functioning was measured using the Wide Range Achievement Test – Fourth Edition (WRAT-4) Reading subtest (Wilkinson & Robinson, Reference Wilkinson and Robertson2006). The cognitive domains of immediate memory, attention, visuospatial functioning, language, and delayed memory were measured using index scores derived from the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS; Randolph et al., Reference Randolph, Tierney, Mohr and Chase1998). Executive Functioning was measured using select subtests from the Delis–Kaplan Executive Function Systems (D-KEFS; Delis, Kaplan, & Kramer, Reference Delis, Kaplan and Kramer2001) including Verbal Fluency, Trail Making Test (TMT), and Color Word Interference (CWI).
Statistical analyses
Of the 496 participants who completed baseline assessments as part of OPTIMUM-Neuro, 99 participants from the Columbia recruitment site were excluded due to an ongoing data freeze at the time this manuscript was prepared and 14 were removed due to having missing MADRS scores. A total of 383 participants were included in the current analyses. Data were examined for normality and outliers. PASS scores worse than three standard deviations above the sample mean were identified in three cases. In order to reduce their influence on analyses while retaining the cases, winsorization was applied to bring these values within three standard deviations. In instances where participants had data missing at random for one of the cognitive assessments or everyday functioning measures, pairwise deletion was applied in order to maximize the number of participants with data in each model. A confirmatory factor analysis using the D-KEFS subtest scores was performed to derive an executive function composite factor score. Of note, the effect of processing speed on the D-KEFS subtests included in the composite score was not able to be factored out in the current study. The RBANS index scores were converted to Z-scores to allow for direct comparison of cognitive measures. All analyses were conducted in MPlus (Muthen et al., Reference Muthen, Muthen and Muthén2017) and R version 4.3.2 (Core Team R, 2023).
To examine the extent to which cognition mediates the relationship between depression and everyday functioning, correlations among the cognitive composite scores, PASS, ECog-12 self- and informant ratings, and MADRS were examined. Simple linear regression models were run with each of the everyday functioning variables and cognitive composite scores as the dependent variable, the MADRS score as the independent variable, and age, sex, and education as the covariates. In cases where depression scores predicted both cognitive composite scores and one of the everyday functioning scores, mediation models with individual cognitive composite scores as mediators between depression and everyday functioning would be run using PROCESS Macro Version 4.3.1, model 4 (Hayes, Reference Hayes2022). Models were conducted with 95% bias-corrected confidence intervals (BC CI) using 5,000 samples for indirect effects. Mediation was considered to be significant if the BC CI did not include zero. Mediation was considered to be partial if the direct effect of depression on everyday functioning remained significant after cognition was held constant.
Results
Participants were largely White, highly educated, majority female, and moderately depressed based on MADRS scores. Full participant characteristics can be found in Table 1. The executive functioning composite score was derived through confirmatory factor analysis and was composed of D-KEFS TMT condition 4, D-KEFS VF conditions 1-3, and D-KEFS CWI conditions 3 and 4. RBANS Index scores were used as the composite scores for the remaining cognitive domains.
Descriptive Statistics
Note. Cognitive test mean scores are z-scores. WRAT-4 = Wide Range Achievement Test – Fourth Edition; PASS = Performance Based Assessment of Self-Care Skills; ECog = Everyday Cognition Scale - 12 item
Correlation analyses as well as descriptive statistics can be found in Table 2 for cognitive tests, everyday functioning, and depression scores. In general, everyday functioning measures (PASS, ECog-12 Self) were negatively related with each cognitive domain score, with the exception of the ECog-12 Informant which was only weakly related to immediate and delayed memory. The PASS demonstrated marginally stronger relationships (r’s > .3) with cognitive domains including immediate memory, executive function, and visuospatial ability compared to the ECog-12 Self. MADRS scores were weakly positively correlated with the PASS and the ECog-12 Self, but not ECog-12 Informant, and weakly negatively associated with attention and immediate memory. There were several cognitive domains with moderate amounts of overlap (r’s > .5) including executive functioning and attention, executive functioning and language, and immediate and delayed memory. This overlap is likely due in part to tests in each domain measuring similar constructs (e.g., verbal fluency in both language and executive functioning and aspects of processing speed in both executive functioning and attention).
Correlation Matrix – Demographics, Everyday Functioning, Depression, and Cognitive Variables
*= p < .05, ** = p < .001, *** = p < .0001. MADRS = Montgomery-Åsberg Depression Rating Scale; PASS = Performance Based Assessment of Self-Care Skills; ECog = Everyday Cognition Scale - 12 item
Linear regressions revealed that MADRS scores were significant predictors of all three everyday functioning measures (PASS, ECog-12 Self, and ECog-12 Informant) as well as the domains of attention, executive functioning, and immediate memory after correcting for covariates (age, education, sex). Depression did not significantly predict any other cognitive domains. Full regression coefficients can be seen in Table 3.
Linear Regression Models for Depression Scores Predicting Everyday Functioning and Cognitive Domains
*All models include age, sex, and education as covariates. df = degrees of freedom.*= p < .05, ** = p < .001, *** = p < .0001. PASS = Performance Based Assessment of Self-Care Skills; ECog = Everyday Cognition Scale - 12 item
Given the significant predictive ability of attention, immediate memory, and executive functioning on everyday functioning, these cognitive domains were assessed as mediators. Immediate memory was found to partially mediate the relationship between depression and everyday functioning as measured by the PASS while attention and executive functioning were found to partially mediate the relationship between depression and everyday functioning as measured by the ECog-12 Self. No other mediation models were significant due to BC CI (bias-corrected confidence intervals) including zero. Full mediation model characteristics can be seen in Table 4 and Figure 1.
Mediation Models for Cognitive Domains and Everyday Functioning
*All models include age, gender, and education as covariates. df = degrees of freedom.* = p < .05, ** = p < .001, *** = p < .0001. MADRS = Montgomery-Åsberg Depression Rating Scale; PASS = Performance Based Assessment of Self-Care Skills; ECog = Everyday Cognition Scale - 12 item
Mediation models for signicant partial mediators. Note. *= p < .05, **= p < .001, ***= p < .0001.
Discussion
The current study aimed to examine whether the relationship between depression and everyday functioning is mediated by cognition in older adults with TR-LLD. We found that higher levels of depressive symptoms predicted worse everyday functioning, with the relationship being stronger using our performance-based measure as compared to the self- and informant-reports, as well as worse performance on measures of attention, executive functioning, and immediate memory. Although these relationships were significant, the overall effect sizes were small, ranging from β = .16 – .26 for the everyday functioning measures and from β = .11 – .13 for cognition. Attention and executive functioning were found to partially mediate the relationship between depression and self-reported everyday functioning, while immediate memory was found to partially mediate the relationship between depression and performance-based everyday functioning.
Our results are in line with those from prior studies demonstrating a direct effect of depression on everyday functioning using both self- and informant-rated (Burton et al., Reference Burton, O’Connell and Morgan2018; Hughes et al., Reference Hughes, Chang, Bilt, Snitz and Ganguli2012; Numbers et al., Reference Numbers, Jang, Brodaty, Sachdev, Draper and Reppermund2022) as well as performance-based measures of everyday functioning (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Gross et al., Reference Gross, Rebok, Unverzagt, Willis and Brandt2011; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011). However, our finding that greater depression severity predicted worse attention, executive functioning, and immediate memory scores, but not delayed memory scores, was surprising given that prior work has shown a robust relationship between depression severity and delayed memory (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Butters et al., Reference Butters, Whyte, Nebes, Begley, Dew, Mulsant, Zmuda, Bhalla, Meltzer, Pollock, Reynolds and Becker2004; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011). One prior study utilizing the RBANS in a demographically similar sample found that depression severity was associated with only the immediate memory index score, and more specifically with the list learning subtest (Hook et al., Reference Hook, Han and Smith2010). It is possible the lack of an association with delayed memory in the current study may be a result of reduced sensitivity related to fewer words to recall in the RBANS to detect this relationship as studies using more sensitive measures of delayed free recall with more words to recall such as the Hopkins Verbal Learning Test and the Rey Auditory-Verbal Learning Test have shown strong associations between depression severity and delayed memory (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011).
Our finding that attention and executive functioning partially mediate the relationship between depression severity and self-reported everyday functioning suggest that greater depression severity may be associated with worse performance on these cognitive processes, in turn influencing an individual’s subjective assessment of their ability, but not their capacity, to carry out everyday activities. This differs from past studies which report that executive functioning and attention mediate the relationship between depression and scores on performance-based assessments of everyday functioning (Brewster et al., Reference Brewster, Peterson, Roker, Ellis and Edwards2017; Gallo et al., Reference Gallo, Rebok, Tennsted, Wadley and Horgas2003; Yen et al., Reference Yen, Rebok, Gallo, Jones and Tennstedt2011). These studies utilized the Everyday Problems Test (EPT), Observed Tasks of Daily Living (OTDL), and Timed Instrumental Activities of Daily Living Test (TIADLT) which differ from the PASS in key ways that may explain the differences in our results. For example, although the EPT assesses similar activities to the PASS (e.g., medication management, bill payment), it requires participants to read materials and provide written responses while the PASS requires participants to carry the requested tasks out. Additionally, TIADLT assesses similar constructs but imposes a time limit on task completion. The measures of executive functioning used in the current studies, as well as those used in many previous studies, are sensitive to the impact of processing speed which may have also influenced results. Future studies examining this relationship using multiple performance-based functioning measures within the same sample as well as methods to factor out variance associated with processing speed (Karr et al., Reference Karr, Hofer, Iverson and Garcia-Barrera2019) from executive functioning scores are required to better understand this variability in findings.
Immediate memory was the only cognitive domain to partially mediate the relationship between depression and the PASS, suggesting that worse immediate memory due to greater depression severity may impact capacity to carry out activities, but not necessarily one’s subjective assessments of their ability. Interventions targeting immediate memory performance may therefore be particularly important for preserving everyday functional capacity in those with TR-LLD. A recent study examining the feasibility, accessibility, and tolerability of one such group-based intervention (“Club Connect”) in older adults has shown promising results (Woolf et al., Reference Woolf, Kaplan, Norrie, Burke, Cunich, Mowszowski and Naismith2024). While the study was not powered to evaluate efficacy, preliminary results found moderate improvements on a verbal list learning task assessing immediate memory and clinician-rated IADL functioning. Additionally, given that we found only partial mediation for all significant models, future work should focus on the role non-cognitive factors such as pain (Connolly et al., Reference Connolly, Garvey and McKee2017; Lin et al., Reference Lin, Yen, Chen and Chen2014) and physical disability (Lenze et al., Reference Lenze, Rogers, Martire, Mulsant, Rollman, Dew, Schulz and Reynolds2001; Seidel et al., Reference Seidel, Brayne and Jagger2011) play in influencing the relationship between LLD and everyday functioning.
There are several notable limitations to consider within the current study, including the use of mediation analyses on cross-sectional data. The primary goal of mediation is to better understand causal mechanisms of change over time, which requires establishment of temporal precedence between variables of interest. While there is growing evidence to suggest depression precedes cognitive impairment in some older adults (Butters et al., Reference Butters, Young, Lopez, Aizenstein, Mulsant, Reynolds III, DeKosky and Becker2008; Zahodne et al., Reference Zahodne, Stern and Manly2014), a consensus has yet to be reached, and future studies utilizing longitudinal data are required in order to provide more robust evidence for or against depression leading to cognitive impairment. The current study lacked a never depressed comparison condition, preventing us from examining whether cognitive performance in our sample was lower than expected compared to age and education matched peers.
Additionally, our sample was predominantly White, female, and well educated, preventing us from examining whether the strength of the relationships differed based on demographic characteristics. For example, prior studies have shown that the relationship between cognition and everyday functioning differs across ethno-racial groups and is generally less robust among Black older adults than White (Jackson et al., Reference Jackson, Rentz, Aghjayan, Buckley, Meneide, Sperling and Amariglio2017), making it difficult to know the extent to which the current results generalize outside of this sample. The measures used to evaluate executive functioning (D-KEFS subtests) have higher sensitivity than those used to measure the remaining cognitive domains (RBANS subtests) when detecting more subtle cognitive impairments. It is possible that the strength of our findings for the executive functioning domain is a result of this differential sensitivity. Future studies should take care to select tests with similar levels of sensitivity to avoid this issue. Lastly, the tasks measured by the PASS may be outdated given increased reliance on technological advances, and future work in this area should consider measures of functioning that capture daily technology use.
Despite these limitations, the present study had several strengths. To our knowledge, this was the first study to examine the relationship between depression, cognition, and everyday functioning in a sample of participants where all participants had a well-characterized history of clinically elevated depressive symptoms, expanding our understanding of how these variables interact in more severely depressed individuals. We also evaluated this relationship using multiple methods of measurement for everyday functioning, allowing for direct comparison of results within the same sample and a more nuanced understanding of their relationships with cognition. Our results also move us incrementally closer to understanding whether TR-LLD poses an increased risk for cognitive impairment and dementia as they show that depression severity within this critically ill sample is significantly associated with both worse cognition and everyday functioning cross-sectionally. Future studies using longitudinal data from the OPTIMUM and OPTIMUM-Neuro trials will help to further advance our understanding of the role TR-LLD plays in dementia risk.
In sum, we found that depression severity in older adults with TR-LLD was related to worse performance in immediate memory, attention, and executive functioning as well as worse everyday functioning. Attention and executive functioning partially mediated the relationship between depression severity and self-reported everyday functioning, while immediate memory partially mediated the relationship between depression severity and performance-based everyday functioning.
Funding statement
This work was supported by the National Institutes of Health R01 MH114969 (MAB), R01 MH114970 (AV), R01 MH114981 (HL), R01 MH114966 (JSS), and R01 MH114980 (PB).
Competing interests
The authors have no conflicts of interest to report.
