Does Consumer Credit Precede or Follow Changes in Cognitive Impairment Among Older Adults? An Investigation in the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) Trial

Abstract

Objectives

We assessed the relationships between pre- and post-morbid consumer credit history (credit scores, debts unpaid, or in collections) and classification of mild (or greater) cognitive impairment (MCI).

Methods

Generalized Estimating Equation models assessed pre-and post-morbid credit history and MCI risk among 1740 participants aged 65+ in the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) study, linked to TransUnion consumer credit data.

Results

Each 50-point increase in credit score was associated with up to 8% lower odds of MCI in the next 3 years. In contrast, new unpaid collections over doubled the odds of having MCI in the next 3 years. MCI was associated with subsequent credit score declines and a 47%–71% greater risk of having a new unpaid collection in the next 4 years.

Discussion

Credit declines may signal risk for future MCI. MCI may lead to financial challenges that warrant credit monitoring interventions for older adults.

Keywords

cognitive dysfunction aging financial activities

Introduction

Emerging evidence suggests that consumer credit has relevance for understanding cognition (Li et al., 2015), an especially important outcome for older adults in the United States (US). In the US, an individual’s consumer credit history represents a cumulative series of financial decisions, based on a person’s potential to manage the timely payment schedules of loans, bills, credit cards, and debts (Israel et al., 2014; Li et al., 2015; Traub, 2013). Older adults comprise a population for whom rising bankruptcy and default rates due to untimely payments of bills is a growing concern due to the erosion of social safety nets for health (e.g., Medicare) (Thorne et al., 2018), and due to greater individual financial responsibility to cover health needs after decreased retirement benefit supports.

Changes in credit may occur before or after cognitive decline. Changes in credit may be an early marker for older adults who are at elevated risk for changes in cognition but who have yet to display overt declines. For example, studies in financial decision-making have suggested that poorer decision-making predicts incidence of later Alzheimer’s Disease and related dementias (ADRD) and cognitive decline (Kapasi et al., 2021; Stewart et al., 2019). Changes in credit after cognitive decline may suggest a lessened ability to manage financial affairs. For example, studies have shown that the ability to manage finances is lower in adults with MCI compared with cognitively normal adults of similar age and that, even among cognitively normal adults, those with more Alzheimer’s Disease pathology (beta-amyloid and tau neurofibrillary tangles) show reduced financial abilities (Gonzalez et al., 2021; Tolbert et al., 2019). These findings suggest that financial decision-making abilities wanes in the very early stages of cognitive decline. While studies in financial decision-making can help us understand mechanisms between credit and cognitive decline, financial decision-making is just one component of credit. Consumer credit goes beyond financial decision-making by reflecting both decision-making and contextual factors that are embedded in credit estimation (length of credit, type of debit, neighborhood factors) (Dean et al., 2018b). While the literature on health decision-making and cognitive decline is broad, the literature on consumer credit and cognition is more limited.

One study that focused on credit scores and cognitive decline found relationships between consumer credit both before and after ADRD diagnosis. In that study of 54,062, Medicare beneficiaries, those diagnosed with ADRD as compared to those who did not develop ADRD were more likely to miss payments on credit accounts up to 6 years prior to and 3 months after an ADRD diagnosis and have subprime credit 2.5 years prior to ADRD diagnosis (Nicholas et al., 2020). this study makes an important advancement and suggests the value of using consumer credit to understand cognition; nonetheless, it left gaps in understanding associations between consumer credit and early indicators of subsequent ADRD diagnosis, such as prodromal phases of ADRD known as mild cognitive impairment (MCI) (Petersen et al., 1999; Winblad et al., 2004).

The Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) randomized controlled trial provides a unique opportunity to assess the temporality of changes in credit and health, across a range of physical and mental health outcomes over a 20-year period. To the extent that we are aware, and after exhaustive review of both domestic and international studies, this study will be the first opportunity to examine the timing between credit score changes and early cognitive decline. This analysis aims to assess the temporal ordering of credit and health by addressing the questions: (1) Can credit or changes in credit predict MCI? and (2) How quickly might MCI influence subsequent credit? We hypothesized that poor credit history is associated with future MCI, and that MCI is associated with subsequent poor credit score. Additionally, we explored the degree to which changes in credit predict upcoming MCI, and whether MCI predicts changes in future credit characteristics. The results of this analysis will provide evidence for whether consumer credit, or changes in credit may be an effective indicator of MCI.

Methods

Sample

The Advanced Cognitive Training for Independent and Vital Elderly was a 4-arm, multi-site, single-blind, randomized controlled trial (N = 2802) with recruitment from March 1998–October 1999 in six metropolitan areas, designed to study the effectiveness of three cognitive training interventions compared to a no-contact control arm (Ball et al., 2002; Jobe et al., 2001). Participants had to be >65 years of age, which would make the vast majority eligible for health insurance under Medicare, the US’s health insurance program for older adults. Exclusion criteria included significant cognitive dysfunction (Mini-Mental State Examination [MMSE] score <23 (Folstein et al., 1975)); functional impairment (dependency or regular assistance in activities of daily living (ADL) as measured by the Minimum Dataset (MDS) Home Care (Morris & Morris, 1997)); self-reported diagnoses of Alzheimer’s disease; stroke within the last 12 months; certain cancers; current chemotherapy or radiation therapy; or poor vision, hearing, or communicative ability that would have interfered with the interventions or outcome assessments. Each treatment arm received ten 60–75-min sessions of training in small groups over 5–6 weeks for memory, speed of processing, and reasoning. Intervention details have been previously described (Ball et al., 2002; Jobe et al., 2001).

The ACTIVE study was approved by institutional review boards at the University of Alabama at Birmingham; Wayne State University, Detroit, Mich; the Hebrew Rehabilitation Center for the Aged, Roslindale, Mass; the Johns Hopkins University School of Medicine, Baltimore, Md; Indiana University, Bloomington; Purdue University, Indianapolis, Ind; Pennsylvania State University, University Park; the University of Florida, Gainesville; and the New England Research Institutes, Watertown, Mass. For this project, we linked the survey data to credit data provided by TransUnion, a global information solutions company. The ACTIVE team sent TransUnion a dataset with minimally necessary identifiable information that enabled TransUnion to append credit history data to the ACTIVE dataset. After appending the data, TransUnion removed the original identifiers and sent the ACTIVE team a de-identified dataset with new unique study IDs, so that the original dataset could not be re-identified.

The analysis sample consisted of ACTIVE participants who had matched credit data in 2001. ACTIVE follow-up interviews were conducted 1, 2, 3, and 5 years after baseline. Credit scores were matched to each of these follow-up visits using the closest year before or after the visit (January 2001, 2002, 2003, 2004, 2005, 2007, 2009, 2010, 2011, 2014, 2017), based on name, address, and Social Security Number (SSN) provided at the ACTIVE baseline interview. (See Figure 1 for a timeline and sources of data collected). Credit history may be missing due to lack of available identifying data (e.g., address) that allowed for matching to ACTIVE data.

Figure 1.

Timeline of ACTIVE and Credit History Data Collection.

Measures

Mild (or greater) cognitive impairment classification was previously derived in the ACTIVE study populations (Thomas et al., 2020), and was based on a comprehensive neuropsychological approach (Bondi et al., 2014; Jak et al., 2009). Participants were classified as MCI if they performed greater than one SD below (i.e., <16^th percentile) the expected mean on at least two cognitive measures in the same cognitive domain. There were seven cognitive scores used to determine MCI classification, including three memory measures: Hopkins Verbal Learning Test immediate free recall (sum of three learning trials), Rey Auditory Verbal Learning Test (AVLT) immediate free recall (sum of five learning trials) and AVLT delayed recognition (score = hits - false positives +35); two reasoning measures: Word Series and Letter Sets (total correct); and two speed of processing measures: Digit Symbol Substitution (total correct) and Useful Field of View (UFOV) Task 2. The z-scores for each test were based on the discrepancy between demographically adjusted (age, education, sex, race) expected scores of a robust normal control group within ACTIVE and each participant’s actual performance [z-scores = (Observed Test Score – Predicted Test Score)/Standard Error of the Estimate]. Participants were classified as MCI or cognitively normal at each ACTIVE visit. While dementia was an exclusion at the ACTIVE baseline visit, it is possible that some participants classified as MCI at follow-up visits would be better characterized as possible/probable dementia. The label of MCI is used throughout, but for participants classified as MCI at a visit after baseline, MCI should be interpreted as “MCI or greater impairment” since there was no clinical dementia workup as part of ACTIVE. Once a participant was classified as MCI, they retained that status for each subsequent visit. For missing cognitive status classifications (unimpaired or MCI), we used a last value carried forward approach. If the participant died, we used all available data up until death.

We separately assessed four consumer credit history variables: continuous credit score, prevalence of major derogatory accounts, and prevalence of any unpaid accounts in collections, with and without medical debt. Consumer credit score was represented by the VantageScore^® 2.0, a proprietary method that the three major credit bureaus in the US use as a summary measure of the credit risk of an individual based on past credit events. Scores ranged from 501 to 990, with higher numbers reflecting better creditworthiness. Consistent with previous studies (Dean et al., 2018a; Knapp & Dean, 2018), we rescaled the credit score into 10 categories of 50-point increments for interpretability of results. We also assessed scores in categories of subprime (≤700), near prime (701–800), and prime (VantageScore >800). To assess short-term defaults, we assessed major derogatory accounts within each time period (none vs. any), reflecting accounts of any amount that are not paid within 120–180 days following a due date; this measure suggests a likelihood of default (collections, charge-offs, foreclosures, bankruptcies and repossessions) and would have negative implications on a credit score. We assessed any collections accounts (none vs. any), which are accounts that go unpaid for 180 days beyond their due debt and whose debt is sold to a third party (charge-off). Paying off a collection might not improve credit since even paid collections are still considered a derogatory mark. We assessed both any unpaid collections accounts and collections accounts that did not include medical debt. For each of the defaulted accounts we assessed both the history (prevalent) of defaults and new (incident) additional defaults for each time period. We excluded delinquent accounts, which would be accounts that are unpaid for as little as 1 day past the due date, and might instead reflect administrative challenges with payment (payment posting too late despite attempts to pay near the due date), rather than actual negligence. Credit history variables captured a look-back period that summarized credit behavior for up to 36 months. We used a last value carried forward for missing credit data, due to an inability to match the participant’s address at that time point.

We included covariates from both the ACTIVE surveys and the credit reports. From the survey, we included ACTIVE intervention group (dummy codes for each of the four groups), baseline age, sex, self-reported race/ethnicity, and education. From the credit reports, we included current home or auto loan status (may reflect use of credit due to ongoing loans as opposed to new uses of credit). In the models regressing credit score/history on MCI classification, we additionally controlled for past credit score/history.

Analysis

We calculated descriptive statistics of demographics, cognitive impairment, and credit variables across study visits. Means and standard deviations were reported for continuous variables. Frequencies and percentages were used to summarize categorical variables. Chi-square tests were used to assess the difference in demographic characteristics and cognitive outcomes at baseline between the matched and unmatched sample. To illustrate the fluctuation of consumer credit before and after onset of MCI, we plotted annual credit scores each year for ACTIVE participants, with separate trend lines for participants with or without MCI onset.

We used a series of time-lagged regression models to assess the temporal associations between consumer credit and MCI. To account for the correlation between multiple visits for an individual, we used Generalized Estimating Equations (GEE) for all models (Liang & Zeger, 1986). For our first hypothesis, the credit variable at the previous visit was the exposure and the subsequent heath variable—MCI status—was the dichotomized outcome in each time-lagged model. Odds ratios (OR) and their 95% confidence intervals (CI) from GEEs with logit link were reported as the effect of credit history on MCI at the follow-up visits. For assessing whether changes in credit preceded MCI, we calculated how much change in credit there had been since the previous visit (e.g., previous visit credit score minus current visit credit score). For the second hypothesis, MCI was the exposure and consumer credit was the outcome variable at subsequent visits. For assessing whether changes in consumer credit followed onset of MCI, we calculated the degree to which credit variables changed from one visit to the next upcoming 4 years of visit (e.g., credit score at current visit minus credit score at next visit) and whether the participant had existing or new defaults (derogatory accounts or unpaid accounts in collections), adjusting for baseline credit score. We also examined whether the type of derogatory collection mattered: medical versus non-medical delinquencies. GEE models with identity link were used to assess the relationship between and the credit score at follow-up visits. The coefficients and their corresponding 95% confidence intervals were reported. For dichotomized outcomes, ORs (95% CIs) were reported. All analyses were performed using SAS (v.9.4; SAS Institute Inc, Cary, North Carolina, USA).

Results

Of the 2802 ACTIVE study participants, 1740 were included in our final analytic sample (Figure 2) after excluding participants due to a lack of data enabling us to match to TransUnion data or due to unique characteristics of study participants that would have enabled re-identification after being de-identified. There was no difference in MCI between the matched and unmatched groups (Supplemental Table 1).

Figure 2.

ACTIVE-TransUnion Match Process Diagram.

In any given year, roughly one-third of our sample were classified as having MCI. Average credits scores were high, with 85% or more having a near prime or high prime rate. Fewer than one-fourth of participants had a major derogatory, unpaid or collections account in any given year (Table 1). Like the full ACTIVE sample, approximately two-thirds of the sample was under age 75, most were White (71%) and female (76%). Most did not have a currently reported auto or home loan (62%).

Table 1.

Demographic Characteristics of ACTIVE Study Participants Matched to TransUnion Credit Score Data (N = 1740).

N (%) for categorical variables and mean(sd) for continuous variables	Year 1 (Visits 2 & 3) N = 1740	Year 2 (Visit 4) n = 1733	Year 3 (Visit 5) n = 1727	Year 5 (Visit 6) n = 1720
Health outcome
Mild cognitive impairment	474 (28%)	423 (29%)	410 (31%)	358(33%)
Credit variables
VantageScore, SD (same year)	818 (93)	820 (97)	823 (100)	824 (99)
Subprime (≤700)	229 (13%)	246 (14%)	247 (15%)	235 (14%)
Near prime (701–800)	340 (20%)	308 (18%)	281 (17%)	288 (17%)
Prime (>800)	1157 (67%)	1154 (68%)	1164 (69%)	1150 (69%)
Any major derogatory versus none	389 (22%)	413 (24%)	441 (26%)	445 (26%)
Any unpaid collection versus none	167 (10%)	186 (11%)	205 (12%)	214 (12%)
Any collections, excluding medical	148 (9%)	161 (9%)	175 (10%)	186 (11%)
Demographic characteristics
Age group
65–69	570 (33%)
70–74	572 (33%)
75–79	366 (21%)
80+	232 (13%)
Race
Black, including Hispanic	507 (29%)
White, including Hispanic	1233 (71%)
Sex
Female	1331 (76%)
Male	409 (24%)
Education
≤12 years of education	697 (40%)
13–16 years of education	577 (33%)
>16 years of education	466 (27%)
Current home or auto loan	666 (38%)	661 (38%)	654(38%)	636 (37%)

The plots in Figure 3 show the trajectories of credit scores for those with (2b) and without (2a) MCI. For those without MCI, we show the trajectory of credit scores since 2001. For those with MCI, we show the event-study graph, plotting credit scores before and after the MCI classification (time 0). Credit scores begin to decline nearly 5 years prior to MCI onset, and remain at a lower level than those without MCI for the remainder of the time period. Overall, there is a significant decrease over time (beta = −.59; p = .02) for those with MCI (60% of participants with MCI had a negative slope) but no significant change in slope (beta = .06; p = .76) for the non-MCI group.

Figure 3.

(a) Plot of mean consumer credit score over time for participants who were not classified as MCI or greater cognitive impairment during 2001–2017. (b) Plot of mean consumer credit score over time for participants with MCI or greater cognitive impairment.

Associations between credit history in years prior to MCI classification can be found in Table 2. Each 50-point increase in credit score was associated with 5% lower odds of MCI in the next 1 and 2 years, and 8% lower odds 3 years later, though the amount of change in score was not associated with increased risk of MCI. Having near or high prime credit was associated with up to 29% lower odds of having MCI within the next 2 years. Having a history of derogatory accounts was associated with 30% greater risk of MCI in the next year, but a history of derogatory accounts was not significantly associated with MCI risk beyond the next year (e.g., 2 or 3 years later). Having a new additional derogatory account was not associated with future MCI risk. Having a history of non-medical debt in collections was not associated with future MCI risk; however, new additional non-medical debt in collections predicted greater MCI risk of up to at least 28% up to 2 years later. Both prevalent unpaid collections and new unpaid collections in the previous year were associated with 22% and 25% greater risk of MCI, respectively. New unpaid collection in three prior years was associated with over double the odds of MCI risk.

Table 2.

Odds of Mild Cognitive Impairment (MCI) or Greater Cognitive Impairment Based on Past Credit History: Results From ACTIVE Study Participants Matched to TransUnion Credit Score Data.

	Odds of MCI
	GEE or Logistic regression(#)
Credit scores
Score as continuous 1 year prior	.95 (.92, 0.98)*
Score as continuous 2 years prior	.95 (.91, 0.99)*
Score as continuous 3 years prior#	.92 (.86, 0.99)*
Credit quality (ref: subprime)
Near prime credit 1 year prior	.72 (.61, 0.86)*
Prime credit 1 year prior	.71 (.58, 0.86)*
Near prime credit 2 years prior	.80 (.61, 1.04)
Prime credit 2 years prior	.71 (.54, 0.94)*
Near prime credit 3 years prior#	.84 (.54, 1.32)
Prime credit 3 years prior#	.73 (.48, 1.09)
Change in credit score since previous visit
Score change from 1 year prior	1.02 (.99, 1.06)
Score change from 2 years prior	1.03 (.99, 1.07)
Score change from 3 years prior#	1.09 (.97, 1.22)
Presence of derogatory account
Major derogatory 1 year prior	1.30 (1.09, 1.55)*
Major derogatory 2 years prior	1.19 (1.00, 1.42)
Major derogatory 3 years prior#	1.38 (1.00, 1.92)
Report of new derogatory account since previous visit
Major derogatory 1 year prior	1.18 (.94, 1.48)
Major derogatory 2 years prior	1.20 (.97, 1.49)
Major derogatory 3 years after#	1.50 (.92, 2.45)
Presence of account in collections, excluding medical debt
Any collection excluding medical debt 1 year prior	1.17 (.96, 1.42)
Any collection excluding medical debt 2 years prior	1.07 (.90, 1.26)
Any collection excluding medical debt 3 years prior#	1.14 (.71, 1.83)
Report of new accounts in collections, excluding medical debt, since previous visit
Any collection excluding medical debt 1 year prior	1.28 (1.01, 1.63)*
Any collection excluding medical debt 2 years prior	1.28 (1.00, 1.65)
Any collection excluding medical debt 3 years prior#	1.45 (.79, 2.65)
Presence of unpaid collections accounts
Any unpaid collections 1 year prior	1.22 (1.05, 1.41)*
Any unpaid collections 2 years prior	1.08 (.87, 1.32)
Any unpaid collections 3 years prior#	.74 (.46, 1.20)
Report of new unpaid collections account since previous visit
Any unpaid collection 1 year prior	1.25 (1.01, 1.54)*
Any unpaid collection 2 years prior	1.27 (.97, 1.66)
Any unpaid collection 3 years prior#	2.01 (1.18, 3.41)*

*p < .05. Credit scores are treated as continuous in 50-point increments. Covariates included are ACTIVE group, baseline age, sex, race/ethnicity, education and current asset ownership status; models where change in credit history was the predictor included previous time point’s credit as a covariate. Major derogatory accounts are those unpaid up to 180 days; collections accounts are those unpaid for longer than 180 days and whose debt is sold to a third party (charge-off).

Associations between the changes in credit scores after MCI classification can be found in Table 3. Credit scores continued to see significant decrements up to the 4 years after MCI classification, with larger decrements as time increased. While often not showing significant associations in the first year after MCI classification, credit score decreases in subsequent visits were significant at 3 and 4 years after MCI classification. MCI classification was associated with a lower likelihood of having near or high prime credit for up to 3 years. Odds of derogatory accounts and non-medical debt accounts in collections increased in the subsequent years after MCI classification. MCI classification was associated with a 40%–66% greater risk of having any unpaid collections each year, for up to 4 years, and a 47%–71% greater risk of having a new unpaid collection each year for up to 4 years.

Table 3.

Regression Results for Past Mild Cognitive Impairment (MCI) or Greater Cognitive Impairment on Future Consumer Credit (β/OR, 95%CI): Results From ACTIVE Study Participants Matched to TransUnion Credit Score Data.

	MCI
	GEE model
	β (95% CI)
Credit scores
Score as continuous 1 year after	−.05 (−.10, −.01)*
Score as continuous 2 years after	−.13 (−.22, −.04)*
Score as continuous 3 years after	−.18 (−.28, −.08)*
Score as continuous 4 years after	−.26 (−.38, −.14)*
Credit quality (ref: subprime)
Near prime credit 1 year after	.72 (.57, 0.91)*
Prime credit 1 year after	.63 (.49, 0.82)*
Near prime credit 2 years after	.69 (.51, 0.93)*
Prime credit 2 years after	.63 (.45, 0.88)*
Near prime credit 3 years after	.74 (.55, 0.99)*
Prime credit 3 years after	.59 (.43, 0.81)*
Near prime credit 4 years after	.77 (.54, 1.10)
Prime credit 4 years after	.52 (.36, 0.76)*
Credit score loss at upcoming visits
Score change to 1 year after	.06 (.01, 0.10)*
Score change to 2 years after	.13 (.04, 0.22)*
Score change to 3 years after	.18 (.09, 0.28)*
Score change to 4 years after	.16 (.04, 0.28)*
	OR (95% CI)
Presence of derogatory accounts
Major derogatory 1 year after	1.24 (.95, 1.61)
Major derogatory 2 years after	1.53 (1.24, 1.89)*
Major derogatory 3 years after	1.33 (1.10, 1.61)*
Major derogatory 4 years after	1.31 (1.05, 1.63)*
Report of new derogatory account at upcoming visit
Major derogatory 1 year after	1.14 (.83, 1.56)
Major derogatory 2 years after	1.33 (.95, 1.85)
Major derogatory 3 years after	1.13 (.84, 1.50)
Major derogatory 4 years after	1.21 (.91, 1.61)
Presence of account in collections, excluding medical debt
Any collection excluding medical debt 1 year after	1.22 (.90, 1.67)
Any collection excluding medical debt 2 years after	1.35 (1.03, 1.76)*
Any collection excluding medical debt 3 years after	1.22 (.93, 1.62)
Any collection excluding medical debt 4 years after	1.40 (1.04, 1.90)*
Report of new accounts in collections, excluding medical debt, at upcoming visit
Any collection excluding medical debt 1 year after	1.39 (.95, 2.04)
Any collection excluding medical debt 2 years after	1.53 (1.03, 2.28)*
Any collection excluding medical debt 3 years after	1.29 (.89, 1.87)
Any collection excluding medical debt 4 years after	1.58 (1.10, 2.27)*
Presence of any unpaid collections
Any unpaid collections 1 year after	1.40 (1.06, 1.86)*
Any unpaid collections 2 years after	1.57 (1.20, 2.06)*
Any unpaid collections 3 years after	1.41 (1.10, 1.82)*
Any unpaid collections 4 years after	1.66 (1.27, 2.17)*
Report of new unpaid collections at upcoming visit
Any unpaid collections 1 year after	1.47 (1.05, 2.07)*
Any unpaid collections 2 years after	1.79 (1.25, 2.55)*
Any unpaid collections 3 years after	1.47 (1.07, 2.02)*
Any unpaid collections 4 years after	1.71 (1.24, 2.35)*

*p < .05. Credit scores are treated as continuous, in 50-point increments. Covariates included are ACTIVE group, past credit score, baseline age, sex, race/ethnicity, education and current asset ownership status. Major derogatory accounts are those unpaid up to 180 days; collections accounts are those unpaid for longer than 180 days and whose debt is sold to a third party (charge-off).

Discussion

Our goal was to examine the associations between consumer credit in relationship to MCI among older adults in the ACTIVE study. Our results suggest that credit scores start modestly declining up to 5 years before an algorithmically defined MCI classification, with derogatory accounts and accounts in collections showing strong associations in the 3 years prior to MCI. Our findings in MCI align with previous recent findings about credit and ADRD (Nicholas et al., 2020), and previous work showing poorer financial abilities in preclinical stages of AD (Gonzalez et al., 2021; Tolbert et al., 2019), while newly suggesting that there are implications for consumer credit in even the prodromal stages of ADRD, namely, MCI. Our results suggest that credit scores may be an alert of future MCI within the upcoming 3 years; however, the amount of change in credit itself was not suggestive of future MCI risk. This counter-intuitive result may be because of the built-in lagged time between when a default occurs and when it appears on a credit account, which is supported by our finding of the strong impact of credit in the previous 1 year, which would have had little time to show change.

Most consistently, unpaid debt in the prior year was most strongly associated with future MCI, though new major derogatory accounts were not associated with MCI in the next year. This may be because there was not enough time for it to become a major derogatory account prior to MCI onset. We found limited evidence that non-medical debt collections predicted future MCI, suggesting that the relationships found could be due to increased debt from medical costs as cognition worsens in the lead-up year to MCI.

After MCI, lower credit scores, negative credit scores, and poorer credit outcomes linger for up to 3–4 years later, with greater likelihood of increased loss of credit in later years as well as increases in new unpaid collections. For those not classified as having MCI during the entire observation period, credit was stable; in contrast, those with MCI experienced steep declines in credit scores, which then remained low. MCI had strongest associations with unpaid collections up to 4 years later, suggesting that changes in credit after MCI have sustained impact, and can cause some of the most severe types of defaults that have implications on long-term creditworthiness.

These results highlight two overarching pathways by which credit and cognitive decline are linked: (a) credit may be predictive of who is more likely to develop, treat, or manage MCI; and (b) credit may be a consequence of the financial impacts of healthcare treatments and long-term care needs (Dean & Nicholas, 2018). The first pathway may operate through the development of MCI and its impacts on financial decision-making. Previous research suggested that poor financial decision-making is associated with a 2.8 times greater likelihood of MCI among older adults (Stewart et al., 2019), independent of starting level of cognition. Another study found that poorer decision-making and greater susceptibility to scams reflected early signs of changes in brain pathology that signaled future cognitive impairment (Kapasi et al., 2021).

While financial decision-making may be one pathway by which consumer credit foreshadows future cognitive decline, consumer credit does not only reflect financial decision-making. Credit history, and credit scores particularly as a summary measure, are based on both current decision-making that is proximate to the MCI designation, but also historical decision-making of finances over the entirety of one’s credit history. It is also based on resource levels, the quality of credit products available, and financial literacy (Li et al., 2015). Furthermore, there is evidence that credit scores themselves are based on a combination of individual decision-making behavior and one’s neighborhood context, such that people with identical credit activities, but who live in different neighborhoods, can have very different credit scores (Avery et al., 2004; Cohen-Cole, 2011). These distinctions suggest that credit captures something beyond individual-level decision-making, and may provide additional context for understanding the range of individual and contextual factors that influence cognitive decline.

Post-morbid credit may be due to changes brought on by the cognitive impairment itself. For example, previous studies have found that financial decision-making is poorer among people with MCI, which may limit their ability to rectify or respond to changes in their credit (Han et al., 2015; James et al., 2015). In our sample, approximately 60% of those with MCI experienced a subsequent decline in credit. The remaining 40% of participants who developed MCI and also maintained good credit may have had supports that helped them maintain good credit, (e.g., another family member may have stepped in to help manage affairs) though our study does not have data to assess that. It could also be that those with MCI who did not experience a decline in credit had very little credit to manage, but that is unsupported by our data: participants in the MCI group and who had a decreasing slope in credit were only using an average of 17% of their credit compared with those in the MCI group with no decreasing slope who were using on average 33% of available credit. It is more likely that, as a result of MCI, family members or others may have stepped in to shift some of the financial management away from the person with MCI, which may have kept their credit from declining. Other possible explanations are that those who maintained good credit despite MCI were able to ensure management of types of credit that are less likely to result in credit default (for example, prioritizing mortgage payments over credit card payments) or that the impact of MCI does not extend to financial behaviors for all people. Additional data beyond that which is available for our study would be needed to confirm which of these explanations is at work.

Limitations

Limitations of our analysis suggest that our findings may overestimate or underestimate the associations between credit and MCI. Nearly 1/3 of the ACTIVE sample was unable to be matched to credit data, mostly due to challenges with identifying an address for the participant. Inconsistent address information may be a marker of residential instability that may be related to consumer credit, meaning we may be missing a sample of people who are the most impaired and underestimate our associations; however, we did not find differences in MCI when we compared the matched to unmatched groups. As described in the methods, our MCI classification after baseline likely includes participants with dementia as well, as there was no clinical dementia diagnostic evaluation within ACTIVE at follow-up visits making it uncertain whether participants had MCI or would be more appropriately classified as possible/probable dementia. This limitation may have combined groups with more severe cognitive impairments with those with less severe cognitive difficulties, and biased our estimates away from a null finding. Due to small sample sizes and lack of MCI assessment prior to entering the ACTIVE study, we were unable to assess incident MCI. We mitigated this limitation by including credit score values pre-baseline and assessing prevalent MCI at each time point. The MCI criteria in ACTIVE are also limited by the lack of delayed free recall trials available on the memory measures. Despite this limitation, these MCI criteria have previously demonstrated that those classified as MCI, particularly those with a learning/memory impairment, had faster declines in everyday functioning relative to those without MCI (Thomas et al., 2020). It is possible that credit data reflected activities of the financial manager of the household, who may not have been the ACTIVE participant for whom MCI was assessed. The study does not have information on the financial manager; however, this likely means our study underestimates the relationship between MCI and credit, as some financial challenges may have been avoided due to intervention by a financial manager.

Conclusion

Poorer credit is evident both before and after MCI; however, the role of credit before and after MCI operates differently: pre-morbid consumer credit may indicate risk of cognitive impairment (or the lack thereof), while post-morbid consumer credit may reflect the ability to quickly respond and manage the economic impact of cognitive impairment. Monitoring of credit and credit changes, especially among those with already poor credit, among older adults may help identify potential future MCI. After indication of MCI, older adults may benefit from additional credit monitoring and assistance managing changes in credit or other financial challenges that arise.

Supplemental Material

Supplemental Material - Does Consumer Credit Precede or Follow Changes in Cognitive Impairment Among Older Adults? An Investigation in the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) Trial

Supplemental Material for Does Consumer Credit Precede or Follow Changes in Cognitive Impairment Among Older Adults? An Investigation in the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) Trial by Lorraine T. Dean, Shang-En Chung, Alden L. Gross, Olivio J. Clay, Sherry L. Willis, Michael Crowe, Ian M. McDonough, Kelsey R. Thomas, Michael Marsiske, Jaya Aysola, Roland J. Thorpe, Jr., Cynthia Felix, Melissa Berkowitz, and Norma B. Coe in Journal of Aging and Health

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the NIA grant P30AG059298. Effort for Dr. Clay is also supported by University of Alabama at Birmingham Alzheimer’s Disease Research Center [P20AG068024]. Effort for Dr. Thomas was supported by the U.S. Department of Veterans Affairs Clinical Sciences Research and Development Service [IK2CX001865] and NIH/NIA grants [R03 AG070435].

ORCID iDs

Lorraine T. Dean

Olivio J. Clay

Kelsey R. Thomas

Roland J. Thorpe, Jr.

Supplemental Material

Supplemental material for this article is available online.

References

Avery

R. B.

Calem

P. S.

Canner

G. B.

(2004). Consumer credit scoring: Do situational circumstances matter? Journal of Banking & Finance, 28(4), 835–856. https://doi.org/10.1016/s0378-4266(03)00202-4

Ball

Berch

D. B.

Helmers

K. F.

Jobe

J. B.

Leveck

M. D.

Marsiske

Morris

J. N.

Rebok

G. W.

Smith

D. M.

Tennstedt

S. L.

Unverzagt

F. W.

Willis

S. L.

for the ACTIVE Study Group (2002). Effects of cognitive training interventions with older adults: A randomized controlled trial. JAMA, 288(18), 2271–2281. https://doi.org/10.1001/jama.288.18.2271

Bondi

M. W.

Edmonds

E. C.

Jak

A. J.

Clark

L. R.

Delano-Wood

McDonald

C. R.

Nation

D. A.

Libon

D. J.

Galasko

Salmon

D. P.

for the Alzheimer's Disease Neuroimaging Initiative . (2014). Neuropsychological criteria for mild cognitive impairment improves diagnostic precision, biomarker associations, and progression rates. Journal of Alzheimer's Disease, 42(1), 275–289. https://doi.org/10.3233/JAD-140276

Cohen-Cole

(2011). Credit card redlining. Review of Economics and Statistics, 93(2), 700–713. https://doi.org/10.1162/rest_a_00052

Dean

L. T.

Knapp

E. A.

Snguon

Ransome

Qato

D. M.

Visvanathan

(2018a). Consumer credit, chronic disease and risk behaviours. Journal of Epidemiology and Community Health, 73(1), 73–78. https://doi.org/10.1136/jech-2018-211160

Dean

L. T.

Nicholas

L. H.

(2018). Getting credit for your health? Using credit scores to understand predictors and consequences of disease. American Journal of Public Health, 108(11), 1503–1505. https://doi.org/10.2105/ajph.2018.304705

Dean

L. T.

Schmitz

K. H.

Frick

K. D.

Nicholas

L. H.

Zhang

Subramanian

Visvanathan

(2018b). Consumer credit as a novel marker for economic burden and health after cancer in a diverse population of breast cancer survivors in the USA. Journal of Cancer Survivorship, 12(3), 306–315. https://doi.org/10.1007/s11764-017-0669-1

Folstein

M. F.

Folstein

S. E.

McHugh

P. R.

(1975). “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198. https://doi.org/10.1016/0022-3956(75)90026-6

Gonzalez

Tommasi

N. S.

Briggs

Properzi

M. J.

Amariglio

R. E.

Marshall

G. A.

Alzheimer’s Disease Neuroimaging Initiative . (2021). Financial capacity and regional cerebral tau in cognitively normal older adults, mild cognitive impairment, and Alzheimer’s disease dementia. Journal of Alzheimer's Disease, 79(3), 1133–1142. https://doi.org/10.3233/jad-201122

10.

Han

S. D.

Boyle

P. A.

James

B. D.

Bennett

D. A.

(2015). Mild cognitive impairment is associated with poorer decision-making in community-based older persons. Journal of the American Geriatrics Society, 63(4), 676–683. https://doi.org/10.1111/jgs.13346

11.

Israel

Caspi

Belsky

D. W.

Harrington

Hogan

Houts

Ramrakha

Sanders

Poulton

Moffitt

T. E.

(2014). Credit scores, cardiovascular disease risk, and human capital. Proceedings of the National Academy Of Sciences, 111(48), 17087–17092. https://doi.org/10.1073/pnas.1409794111

12.

Jak

A. J.

Bondi

M. W.

Delano-Wood

Wierenga

Corey-Bloom

Salmon

D. P.

Delis

(2009). Quantification of five neuropsychological approaches to defining mild cognitive impairment. The American Journal of Geriatric Psychiatry, 17(5), 368–375. https://doi.org/10.1097/JGP.0b013e31819431d5

13.

James

B. D.

Boyle

P. A.

Han

S. D.

Bennett

D. A.

(2015). Cognitive decline is associated with risk aversion and temporal discounting in older adults without dementia. Plos One, 10(4), 1–12. https://doi.org/10.1371/journal.pone.0121900

14.

Jobe

J. B.

Smith

D. M.

Ball

Tennstedt

S. L.

Marsiske

Willis

S. L.

Rebok

G. W.

Morris

J. N.

Helmers

K. F.

Leveck

M. D.

Kleinman

(2001). Active: A cognitive intervention trial to promote independence in older adults. Controlled Clinical Trials, 22(4), 453–479. https://doi.org/10.1016/s0197-2456(01)00139-8

15.

Kapasi

Stewart

Schneider

J. A.

Bennett

D. A.

Boyle

P. A.

(2021). Association of amyloid-β pathology with decision making and scam susceptibility. Journal of Alzheimer's Disease, 83(2), 879–887. https://doi.org/10.3233/jad-210356

16.

Knapp

E. A.

Dean

L. T.

(2018). Consumer credit scores as a novel tool for identifying health in urban US neighborhoods. Annals of Epidemiology, 28(10), 724–729. https://doi.org/10.1016/j.annepidem.2018.07.013

17.

Gao

Enkavi

A. Z.

Zaval

Weber

E. U.

Johnson

E. J.

(2015). Sound credit scores and financial decisions despite cognitive aging. Proceedings of the National Academy Of Sciences, 112(1), 65–69. https://doi.org/10.1073/pnas.1413570112

18.

Liang

K.-Y.

Zeger

S. L.

(1986). Longitudinal data analysis using generalized linear models. Biometrika, 73(1), 13–22. https://doi.org/10.1093/biomet/73.1.13

19.

Morris

J. N.

Morris

S. A.

(1997). ADL assessment measures for use with frail elders. Journal of Mental Health and Aging, 3(1), 19–46.

20.

Nicholas

L. H.

Langa

K. M.

Bynum

J. P. W.

Hsu

J. W.

(2020). Financial presentation of Alzheimer disease and related dementias. JAMA Internal Medicine, 181(2), 220-227. https://doi.org/10.1001/jamainternmed.2020.6432

21.

Petersen

R. C.

Smith

G. E.

Waring

S. C.

Ivnik

R. J.

Tangalos

E. G.

Kokmen

(1999). Mild cognitive impairment: Clinical characterization and outcome. Archives of Neurology, 56(3), 303–308. https://doi.org/10.1001/archneur.56.3.303

22.

Stewart

C. C.

Wilson

R. S.

Bennett

D. A.

Boyle

P. A.

(2019). Healthcare and financial decision making and incident adverse cognitive outcomes among older adults. Journal of the American Geriatrics Society, 67(8), 1590–1595. https://doi.org/10.1111/jgs.15880

23.

Thomas

K. R.

Cook

S. E.

Bondi

M. W.

Unverzagt

F. W.

Gross

A. L.

Willis

S. L.

Marsiske

(2020). Application of neuropsychological criteria to classify mild cognitive impairment in the ACTIVE study. Neuropsychology, 34(8), 862–873. https://doi.org/10.1037/neu0000694

24.

Thorne

Foohey

Lawless

R. M.

Porter

(2018). Graying of US bankruptcy: Fallout from life in a risk society. Sociological Inquiry, 90(4), 681–704. https://doi.org/10.1111/soin.12323

25.

Tolbert

Liu

Hellegers

Petrella

Weiner

Wong

Doraiswamy

P. M.

(2019). Financial management skills in aging, MCI and dementia: Cross sectional relationship to 18F-florbetapir PET cortical β-amyloid deposition. The Journal of Prevention of Alzheimer's Disease, 6(4), 274–282. https://doi.org/10.14283/jpad.2019.26

26.

Traub

(2013). Credit reports and employment: findings from the 2012 national survey on Credit Card Debt of low-and Middle-Income Households. Suffolk UL Rev., 46, 983.

27.

Winblad

Palmer

Kivipelto

Jelic

Fratiglioni

Wahlund

L. O.

Nordberg

Bäckman

Albert

Almkvist

Arai

Basun

Blennow

de Leon

DeCarli

Erkinjuntti

Giacobini

Graff

Hardy

Petersen

(2004). Mild cognitive impairment–beyond controversies, towards a consensus: Report of the international working group on mild cognitive impairment. Journal of Internal Medicine, 256(3), 240–246. https://doi.org/10.1111/j.1365-2796.2004.01380.x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB