Cognitive Correlates of Impaired Testamentary Capacity in Alzheimer’s Dementia

Abstract

Background

Testamentary capacity (TC) is a legal construct about the ability to make or change a will. Although studies of financial and medical capacities have noted a strong association between capacity and cognition, no study has examined the cognitive correlates of TC in Alzheimer’s disease (ad).

Methods

Study participants were 22 cognitively healthy controls and 20 persons with mild to moderate ad who were administered a neuropsychological battery and the Testamentary Capacity Instrument (TCI), a new performance-based measure of TC. Both TCI Element and TCI Total scores were calculated. TCI Total scores were calculated for the purposes of expanding correlational analyses (i.e., Pearson’s product–moment) and are not intended for forensic TC evaluations. Final predictors were identified using linear and logistic regression modeling.

Results

All ad participants but one obtained TCI Totals that fell >2.5 standard deviations below the control group mean. Initial cognitive correlates of TCI performance were measures of general cognition, language, verbal memory, executive function, and processing speed. The four cognitive variables showing the highest t values and correlating with TCI Total score > 0.7 were entered into logistic and linear regression models. Final predictor models consisted of measures of verbal memory, executive function, and semantic knowledge and demonstrated shared variance of 0.71 (linear) and 0.82 (logistic).

Conclusions

A diagnosis of ad dementia is associated with clinical impairment in TC and associated cognitive abilities and indicates that a legal assessment of TC is warranted. Second, the results offer insight into the cognitive basis of TC in persons with Alzheimer’s dementia.

Introduction

Testamentary capacity (TC) is a medical-legal construct that refers to an individual’s ability to execute a will (Frolik, 2001; Marson, Huthwaite, Hebert, 2004; Spar & Garb, 1992; Walsh, Brown, Kaye, & Grigsby, 1997). TC is becoming increasingly important due to our aging society, and disputes about TC are becoming more prevalent in United States courts (Brenkel et al., 2018). The act of testation lessens the need for judicial intervention to settle disputes amongst potential inheritors by providing a legal mechanism for the orderly transfer of goods and possessions from the decedent to designated heirs. In the absence of a will, probate court is left to determine how assets will be distributed among a decedent’s survivors and may give little or no consideration to the preferences of the deceased (Frolik, 2001; Marson, Huthwaite, Hebert, 2004; Walsh, Brown, Kaye, & Grigsby, 1997).

TC is composed of a set of elements, or specific abilities, determined by the legal system for the execution of a valid will (Frolik, 2001; Haldipur & Ward, 1996; Walsh, Brown, Kaye, & Grigsby, 1997): (1) understand the nature and purpose of a will, (2) understand property and belongings, (3) express knowledge of persons who are potential heirs, and (4) exhibit rational reasoning in developing a plan for disposing of assets following death. The way in which these elements are weighed by courts in determining the validity of a will varies from state to state (Frolik, 2001;Spar & Garb, 1992 ; Walsh, Brown, Kaye, & Grigsby, 1997). Although some states require that the testator meet only one of the criteria for a will to be considered valid, other states require a testator to meet most or all criteria (Spar & Garb, 1992; Walsh, Brown, Kaye, & Grigsby, 1997). For example, the State of Alabama requires that a testator understand that a will is being executed, understand his/her property and belongings, and be able to express knowledge about potential heirs (Walsh, Brown, Kaye, & Grigsby, 1997).

In previous empirical research, our group has investigated the relationship between cognition and financial and medical capacities in a number of neurological and neurodegenerative disorders (Gerstenecker et al., 2015a; Gerstenecker et al., 2018; Gerstenecker, Hoagey, Marson, & Kennedy, 2017; Gerstenecker et al., 2017a; Gerstenecker et al., 2017b; Gerstenecker, Triebel, Eakin, Martin, & Marson, 2018; Gerstenecker, Triebel, Martin, Snyder, & Marson, 2016; Martin, Gerstenecker, Nabors, Marson, & Triebel, 2015; Niccolai et al., 2017; Triebel et al., 2015). These studies have highlighted that financial capacity and medical capacity are heavily mediated by cognition and that declines in these capacities occur in correspondence with declines in cognition (Triebel, Gerstenecker, & Marson, 2014). However, to our knowledge, only a handful of studies have been conducted in which aspects of TC have been investigated in older adults with cognitive impairment.

In one study conducted by Heinik, Werner, and Lin (1999), 31 older adults seen at a geropsychiatric outpatient service in Tel Aviv, Israel were asked to define “testament” as part of a clinical interview (Heinik, Werner, & Lin, 1999) (The term “testament” in this study appears to be equivalent to the meaning of a “will” as defined by Anglo-American law.). Participants’ responses were recorded verbatim and rated according to a six-item scale. The most common elements of a testament reported were that it involves a person (58.1%) and takes effect after a person’s death (45.2%). Limitations of the study included a high proportion of the sample being diagnosed with dementia (24/31) and lack of a control group.

In another study, TC was evaluated in groups of persons with mild, moderate, or severe Alzheimer’s disease (ad) (Roked & Patel, 2008). Determinations of TC were derived from a clinical interview based on guidelines outlined by the British Geriatric Society as well as the British Medical Association and The Law Society. Overall, 6% of persons with mild ad, 38% of persons with moderate ad, and 56% of persons with severe ad were judged to lack TC. In addition, performance on two cognitive screening measures was found to be highly predictive of TC status. However, similar to the aforementioned study by Heinik and colleagues (Heinik, Werner, & Lin, 1999), a control group was not included.

Our group investigated “knowledge of financial assets and estate arrangements” in a sample of 23 cognitively intact older adults and 20 persons with mild to moderate ad type dementia (Earnst et al., 2001). Although “knowledge of financial assets and estate arrangements” is distinctly relevant to TC, the aforementioned study was primarily focused on financial capacity. Results showed that the sample of persons diagnosed with ad had significantly poorer knowledge of their financial assets and estate arrangements. Thus, our findings indicated that memory deficits associated with ad likely affect an individual’s recall and understanding of the nature and extent of his/her property and belongings.

In an article by Papageorgiou and colleagues (Papageorgiou, Voskou, Economou, Beratis, & Douzenis, 2018) a brief measure designed to screen persons with dementia for TC was introduced. The Testamentary Capacity Assessment Tool (TCAT) contains four subtests: (1) memory, both autobiographical and other, (2) absence of psychopathology, (3) knowledge of financial parameters (including assets, also bills, and other financial matters), and (4) intention (including vignettes and theory of mind). The authors compared performance on the TCAT to the judgments of an expert psychiatrist and reported that a derived TCAT cut score showed good reliability, sensitivity, and specificity in predicting the expert’s ratings. However, although the introduction of the TCAT is novel and creative, it nevertheless contains conceptual limitations including an idiosyncratic and not theoretically based model of TC that is not linked to established legal elements and models of TC found in Anglo-American and other international legal systems (Martin, Gerstenecker, Hebert, Triebel, & Marson, in submission).

In a conceptually driven article previously published by our group, we outlined how various cognitive abilities may associate with different aspects of TC (Marson, Huthwaite, Hebert, 2004). For understanding the nature and purpose of a will, the testator should be asked to define in his or her own words what a will is and what it does, be able to acknowledge the existence of a prior will, and to understand the purpose and consequences of an amendment to a will. Thus, these requirements were theorized to rely heavily on several cognitive abilities including semantic memory, verbal abstraction and conceptualization, and expressive language (Marson, Huthwaite, Hebert, 2004). For understanding property and belongings, a testator’s knowledge of the nature and extent of his or her assets is assessed. Thus, a testator’s knowledge of assets is likely dependent not only on his or her memory for the meaning of terms such as “assets” or “ownership,” but also on short-term and historical memory for items purchased, sold, or relinquished both recently and in the past (Marson, Huthwaite, Hebert, 2004). For expressing knowledge of persons who are potential heirs, the testator should be asked to name his or her potential heirs and to describe their relationship to him or her. Particular interest should also be given to those individuals the testator has specifically chosen to exclude from the will. As such, this element appears to be strongly linked to autobiographical memory (Marson, Huthwaite, Hebert, 2004). For exhibiting rational reasoning in developing a plan for disposing of assets following death, a testator’s reasoning ability in forming a plan for disposition of assets is assessed. The testator may be asked to describe why he or she chose to bequeath a specific property to a given heir. In addition, the testator should be asked to project the will’s future consequences for potential heirs. As such, this element appears to rely not only on skills of semantic memory and historical, biographical, and short-term memory abilities, but is also highly dependent on executive control skills involving reasoning, planning, and organization.

Although we have previously highlighted potential cognitive abilities underlying TC, there is a dearth of empirical research pertaining to the cognitive correlates of TC. This dearth of research is troubling given that physicians’ judgments have been considered the historical gold standard for determining capacity in medical and legal practice (Appelbaum & Gutheil, 1991), despite it having been shown that subjective ratings of capacity by physicians have poor reliability (D. C. Marson, McInturff, Hawkins, Bartolucci, & Harrell, 1997). To address this gap in the research literature, we administered a neuropsychological test battery and performance-based measure of TC to a group of healthy older adults and a group of persons clinically diagnosed with ad. The following hypotheses were made: (1) TC and cognition will be highly correlated and (2) impairments in cognitive abilities common in persons with ad will contribute to TC impairment in the ad group in comparison to healthy controls.

Methods

Participants

Following institutional review board approval and the obtainment of informed consent, 42 participants (20 persons clinically diagnosed with ad and 22 cognitively intact older adults) were recruited to participate in a study of legal decision-making in late life. Diagnoses of control or ad were made by a consensus conference that was composed of neurologists, neuropsychologists, and nurses with experience in ad and other dementias.

Control and ad participants were recruited from a longitudinal research study conducted through the Alzheimer’s Disease Research Center at the University of Alabama at Birmingham (UAB) between July 2007 and October 2008. Control or ad diagnosis was established before study recruitment.

Diagnoses of ad were made using National Institute for Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Dementias Association criteria (McKhann et al., 1984). Persons determined to have severe ad (defined as scored ≤10 on the Mini Mental State Examination [MMSE]) were excluded.

Controls received a Clinical Dementia Rating (Morris, 1993) staging score of 0 and scored in the intact range on standardized tests of mental status (i.e., MMSE (Folstein, Folstein, & McGugh, 1975)) and global cognitive status (i.e., Dementia Rating Scale – Second Edition [DRS-2] (Jurica, Leitten, & Mattis, 2001)).

Both ad and control participants underwent thorough medical and neurological history and examination. Controls and ad participants were excluded from the study if at the time of participation they suffered from any of the following conditions that have been found to interfere with cognitive function: neurodegenerative disease other than AD, another neurological illness, cancer, severe pulmonary, renal, and/or liver disease, cardiac disease, autoimmune disease, alcoholism, untreated major depression or any other severe psychiatric disorder, and/or severe behavioral problems.

Measures

TC

The Testamentary Capacity Instrument (TCI) is an experimental instrument specifically designed to assess TC based on four legal elements: (1) understanding of the nature and purpose of a will (purpose), (2) recollection of the nature and extent of one’s property (assets); memory for potential heirs or objects of one’s bounty (heirs); and rational plan for the disposition of assets (plan). Elements were selected based on a review of current Anglo-American legal requirements for TC across the 50 United States as documented by Walsh and colleagues (Walsh, Brown, Kaye, & Grigsby, 1997) and the senior author’s forensic knowledge and experience. In addition, individual items were designed and reviewed in collaboration with psychologists and attorneys knowledgeable of TC and/or dementia populations. Items were then grouped according to the construct or legal element that they were believed to measure.

To score the TCI, items within each conceptual legal element are summed to create a respective Element score. For the purpose of this paper, the four TCI Element scores were then summed statistically to create a TCI Total score. The TCI Total score was created solely to expand the correlational analyses of the paper and is not intended or suited to serve as a global forensic measure of TC. In the TCI, TC is conceptualized strictly according to the four TCI Elements, which represent the four legal constructs commonly accepted in Anglo-American law as the basis of TC. For all TCI variables, higher scores indicate better performance. A detailed description of administration and scoring of the TCI can be found in the companion paper to the current study (Martin et al., in submission).

Cognitive measures

The DRS-2 (Jurica, Leitten, & Mattis, 2001) is a measure of general cognitive functioning that is widely used in geriatric clinical and research assessments. It yields age- and race-corrected scores on five subscales (Attention, Initiation/Perseveration, Construction, Conceptualization, Memory), and an age-, education-, and race-corrected total score.

The Digit Span subtest from the Wechsler Adult Intelligence Scale - Third Edition (WAIS-III) (Wechsler, 1997) is a measure of attention and concentration, in which subjects repeat orally presented digit strings, forwards, and backwards.

To evaluate semantic verbal fluency, participants were given 1 min to name as many animals—beginning with any letter—as possible (Reitan & Wolfson, 1993).

The Boston Naming Test (BNT) (Kaplan, Goodlass, & Weintraub, 1983) is a measure of confrontation naming and semantic memory in which subjects are required to name line drawings. For the current study, the 30-item version of the BNT was used.

The California Verbal Learning Test – Second Edition (CVLT-II) (Delis, Kramer, Kaplan, & Ober, 2000) is a measure of verbal learning and memory in which subjects learn 16 words over five learning trials. After a 20-min delay, free recall of the list is queried.

Trail Making Test Parts A and B from the Halstead-Reitan Neuropsychological Battery (Reitan & Wolfson, 1993) are measures of visuomotor knowledge speed and set-shifting, respectively. For Trails A, subjects asked to draw a line connecting 25 numbers in numerical order. For Trails B, subjects are asked to draw a line connecting 25 numbers and letters in alternating sequential order (i.e., 1 to A, A to 2, 2 to B). For both tasks, raw score is equal to the number of seconds to completion.

The Digit Symbol Coding subtest from the WAIS-III (Wechsler, 1997) is a measure of knowledge speed and divided attention in which subjects have 120 s to match and transcribe specific numbers with their paired symbols using a key at the top of the page.

Mood measure

Symptoms of depression were assessed with the Geriatric Depression Scale (GDS) (Yesavage et al., 1982).

Data analyses

Independent t-tests were used to determine whether the control or ad groups differed according to age or education. Chi-square tests were used to determine whether the two study groups differed according to gender or race. Independent t-tests or their non-parametric alternatives were used to determine whether the control or ad groups differed on TCI, cognitive, or mood variables.

Persons with ad were classified as exhibiting intact or impaired scores on TCI variables using psychometric cutoff scores derived from control performance. This method is useful in categorizing presence or absence of decisional impairment and has been utilized in earlier capacity studies (Gerstenecker et al., 2015a; Gerstenecker et al., 2015b; Triebel et al., 2015). A series of independent t-tests were then conducted to determine if neurocognitive performance, age, education, and GDS scores varied according to testamentary status. Pearson’s Chi-square tests were then conducted to determine whether gender or race varied according to testamentary status.

Pearson’s product–moment correlations were calculated to determine which cognitive variables significantly associated with TCI Elements and Total score. Independent t-tests or their non-parametric alternatives were calculated to investigate if cognitive variables varied according to testamentary status. As the cognitive correlates were highly similar across TCI Elements, a TCI Total score was calculated for the purpose of simplifying our expanded correlational analyses. However, as noted earlier, the TCI Total score was derived solely for model generation and is not intended or suited for clinical forensic evaluations (please see Measures – TC for a more detailed discussion).

The five variables most associated with TCI Total score, or that varied according TC impairment status using the TCI Total score were then used to construct two regression models. Because the same five cognitive variables were found to have the strongest association with each TCI Element, those five cognitive variables were used to construct regression models using TCI Total score. For the first model, the five designated cognitive variables were placed into a forced-entry linear regression to predict TCI Total score. For the second model, the five cognitive variables showing the largest t statistic were placed in a forced-entry binary logistic regression to identify persons with intact and impaired TCI Total score (using >1.5 standard deviation [SD] below the control group mean as indicative of impairment). For both models, relative predictive power was obtained through either R² or the coefficient of determination of Nagelkerke or pseudo-R².

To help balance the probability of committing a type I or type II error, a two-tailed alpha of 0.01 was used for all analyses.

Results

Sample Characteristics

Table 1 shows the sample characteristics.

Table 1.

Demographics and cognitive performance of Control and ad groups

Measure	Control (n = 22)	ad (n = 20)	t/χ 2	p	d
Demographics
Age	72.6 (6.9), 58–82	75.6 (10.5), 57–91	-1.1	0.277	0.34
Education	14.9 (2.1), 12–20	15.1 (3.2), 9–20	-0.2	0.865	0.07
Female/male	13/9	10/10	0.3	0.757	n/a
Caucasian/AA	17/5	18/2	1.2	0.414	n/a
General cognition
DRS-2	139.9 (3.0), 132–144	113.6 (14.6), 82–134	8.2	<0.001	2.5
MMSE	29.2 (1.0), 26–30	22.4 (4.6), 13–28	6.0	<0.001	2.0
Attention
Digit span	16.6 (3.1), 12–24	12.4 (6.4), 5–29	2.7	0.010	0.84
Expressive language
Animal naming	20.9 (6.2), 10–31	10.9 (3.9), 5–20	6.1	<0.001	1.9
BNT	28.0 (1.6), 24–30	20.6 (5.4), 6–28	6.1	<0.001	1.8
Memory
CVLT-II immediate	48.4 (8.6), 38–69	20.2 (8.1), 4–36	10.8	<0.001	2.6
CVLT-II long delay	10.8 (2.9), 5–15	1.2 (2.5), 0–9	11.4	<0.001	3.5
Executive function
Trails B	77.1 (20.7), 46–141	241.5 (88.6), 100–300	8.5	<0.001	2.6
Digit symbol coding	59.2 (8.9), 46–80	30.4 (14.8), 2–60	7.6	<0.001	2.4
Processing speed
Trails A	29.6 (8.7), 16–47	75.0 (47.7), 30–180	4.4	<0.001	1.3
Depression
GDS	3.9 (4.4), 0–17	7.2 (6.0), 0–22	2.0	0.053	0.63

Note: Except for female/male and Caucasian/AA, values are mean (SD), range. For male and female and Caucasian/AA, values are n/n. p value for t-test or non-parametric alternative analyzing group differences or Pearson’s Chi-square test (gender and race).

ad = Alzheimer’s disease; AA = African American; DRS-2 = Dementia Rating Scale – Second Edition; BNT = Boston Naming Test; CVLT-II Immediate = California Verbal Learning Scale – Second Edition Immediate Recall; CVLT-II Long Delay = California Verbal Learning Scale – Second Edition Long Delay Free Recall; GDS = Geriatric Depression Scale; n/a = not available, d = Cohen’s d.

The sample was mainly composed of Caucasian, well-educated, older adults. Both genders were fairly well distributed. Mean GDS scores did not meet threshold for clinical significance in either group.

TCI and Cognitive Performance

Tables 1 and 2 show group performances on the cognitive measures and TCI, respectively.

Table 2.

TCI performance of Control and ad groups

Measure	Control (n = 22)	AD (n = 20)	t/χ 2	p	d
TCI Variables
Purpose	23.9 (2.0), 19–28	18.7 (3.8), 8–25	5.6	<0.001	1.7
Assets	39.0 (2.6), 33–43	31.8 (6.0), 14–45	5.2	<0.001	1.6
Heirs	16.0 (0.2), 15–16	12.8 (4.3), 1–16	3.5	0.001	1.1
Plan	12.9 (1.1), 11–14	9.2 (2.5), 1–13	6.2	<0.001	1.9
Total	91.8 (2.7), 87–97	72.5 (14.0), 24–90	6.3	<0.001	1.9

Note: Values are mean (SD), range. AD = Alzheimer’s disease; TCI = testamentary capacity inventory; n/a = not available; d = Cohen’s d.

The ad group performed significantly below the control group on all cognitive measures except for WAIS-III Digit Span.

The ad group performed significantly below the control group for both TCI Total score and all TCI elements (i.e., purpose, assets, heirs, plan).

TCI Impairment in Relation to Control Group Performance

Table 3 contains impairment ratings for the ad group. All but one participant in the ad group performed >2.5 SD below the control group mean.

Table 3.

TCI impairment ratings for the ad group

TCI element	No impairment	Mild impairment	Moderate impairment	Severe impairment	Total impaired
Purpose	7/20 (35%)	1/20 (5%)	3/20 (15%)	9/20 (45%)	13/20 (65%)
Assets	3/20 (15%)	3/20 (15%)	3/20 (15%)	11/20 (55%)	17/20 (85%)
Heirs	7/20 (35%)	0/20 (0%)	0/20 (0%)	13/20 (65%)	13/20 (65%)
Plan	2/20 (10%)	5/20 (25%)	0/20 (0%)	13/20 (65%)	18/20 (90%)
Total	1/20 (5%)	0/20 (0%)	0/20 (0%)	19/20 (95%)	19/20 (95%)

Note: Impairment ratings are derived from control group performance and are not meant to indicate incapacity.

Mild impairment= > 1.5 SD below control group mean. Moderate impairment= > 2.0 SD below control group mean. Severe impairment= > 2.5 SD below control group mean.

AD = Alzheimer’s disease; TCI = testamentary capacity instrument.

Associations between TCI and Cognitive Variables

Table 4 contains associations between TCI variables and the four cognitive variables showing the strongest relationships.

Table 4.

Whole sample (N = 42) correlations between cognitive variables and TCI elements and total score

TCI element	CVLT-II immediate	CVLT-II delay	Digit symbol coding	TMT B	BNT
Purpose	0.72, <0.001	0.71, <0.001	0.66, <0.001	−0.69, <0.001	0.65, <0.001
Assets	0.55, <0.001	0.59, <0.001	0.54, <0.001	−0.62, <0.001	0.42, 0.007
Heirs	0.57, <0.001	0.53, <0.001	0.64, <0.001	−0.64, <0.001	0.58, <0.001
Plan	0.67, <0.001	0.67, <0.001	0.61, <0.001	−0.56, <0.001	0.57, <0.001
TCI Total	0.77, <0.001	0.78, <0.001	0.75, <0.001	−0.79, <0.001	0.67, <0.001

Note: TCI Total represents an artificial construct used for the purpose of correlational analyses. The five cognitive variables showing the strongest associations with TCI Elements are listed. These five cognitive variables, in varying order, were observed to have the strongest relationship with each individual TCI Element.

TCI = testamentary capacity instrument; CVLT-II Immediate = California Verbal Learning Test – Second Edition Immediate Recall; CVLT-II Delay = California Verbal Learning Test Long Delay Free Recall; BNT = Boston Naming Test.

Performances on the following neuropsychological tasks were significantly associated with TCI Total scores: Trails B (r = −0.787, p < 0.001), CVLT-II Long Delay Free Recall (r = 0.777, p < 0.001), CVLT-II Immediate Recall (r = 0.771, p < 0.001), WAIS-III Digit Symbol Coding (r = 0.746, p < 0.001), DRS-2 Total (r = 0.744, p < 0.001), BNT (r = 0.670, p < 0.001), Animal Naming (r = 0.572, p < 0.001), and Trails A (r = −0.426, p = 0.005). WAIS-III Digit Span was not significantly associated with TCI Total (r = 0.332, p = 0.034).

Performances on the following variables were significantly poorer for participants with impaired TCI Total score: CVLT-II Long Delay Free Recall (t = 9.4, p < 0.001), CVLT-II Immediate Recall (t = 8.5, p < 0.001), Trails B (t = 7.6, p < 0.001), WAIS-III Digit Symbol Coding (t = 6.8, p < 0.001), BNT (t = 5.8, p < 0.001), Animal Naming (t = 5.1, p < 0.001), and Trails A (t = 4.1, p < 0.001). Scores on WAIS-III Digit Span did not differ significantly between the impaired or intact groups (t = 2.5, p < 0.017).

Using Cognitive Performance to Predict TCI Total Score and Impairment Rating

The final regression models can be found in Table 5.

Table 5.

Results of the linear regression and logistic regression used to predict TCI Total scores and testamentary capacity impairment, respectively

Model	F/χ2; df; p	R 2	SEE	β, SE
Linear regression  CVLT-II immediate  CVLT-II long delay  Digit symbol coding  Trails B  BNT  Constant	17.0; 38; <0.001	0.72	5.97	0.11, 0.18 0.21, 0.53 0.08, 0.10 −0.41, 0.02 0.26, 0.29 74.77, 9.0
Logistic regression  CVLT-II immediate  CVLT-II long delay  Digit symbol coding  Trails B  BNT  Constant	38.0; 5; <0.001	0.84	n/a	−0.86, 0.14 0.03, 0.40 −0.04, 0.08 0.01, 0.02 −0.034, 0.34 11.6, 8.43

Note: SEE = standard error of the estimate of the regression model; β = unstandardized beta weights; SE = standard error of coefficient; TCI = testamentary capacity instrument; CVLT-II Immediate = California Verbal Learning Test – Second Edition Immediate Recall; CVLT-II Delay = California Verbal Learning Test Long Delay Free Recall; BNT = Boston Naming Test; n/a = not applicable.

For the linear regression model used to predict TCI Total score, R² was 0.71. The resulting model met all assumptions for regression, including normality of residuals (Shapiro–Wilk[39] = 0.10, p = 0.200).

Using a definition of impairment as scores falling >1.5 SD below the control group mean, we next constructed a logistic regression model to predict impaired or intact TC. Nagelkerke-R² was 0.82 and the receiver operating characteristic (ROC) value based on this model was 0.98.

Discussion

TC is a legal construct that refers to an individual’s ability to execute or make changes to a will (Spar & Garb, 1992; Walsh, Brown, Kaye, & Grigsby, 1997). Core legal components of TC include a testator’s ability to understand the nature and purpose of a will, understand the nature and extent of his/her assets and property, demonstrate knowledge of his/her natural heirs, and have a basic plan for distribution of assets to heirs following death (Frolik, 2001; Haldipur & Ward, 1996; Walsh, Brown, Kaye, & Grigsby, 1997). Identifying the specific cognitive functions that underlie these four components of TC can lead to better assessments of people at risk for impaired TC. In the current study, we administered a performance-based measure of TC (i.e., TCI) and neuropsychological battery to a sample of healthy controls and persons clinically diagnosed with ad. Results showed that TC, as conceptualized earlier, is associated with a wide range of cognitive variables—in particular verbal short-term memory, executive function, and semantic knowledge—and is strongly associated with impaired cognition. Results also indicated that cognitive decline has advanced to the point of affecting TC—in comparison to demographically-similar controls—by the time a diagnosis of clinical ad is applied. However, it is important to note that clinical impairment in TC does not necessarily indicate a lack of TC as defined by legal constructs and that a person with ad who wish to make or change a will should undergo a legal assessment of TC. It is also important to note that although clinical TC impairment was noted in a majority of our ad sample, one person with ad exhibited intact TC.

Our results indicate that the act of testation is highly affected by cognitive ability and that deficits in many cognitive domains may affect a person’s capacity to make or change a will. This observation was supported by analyses that investigated the cognitive correlates of TC as well as differences in cognitive performances of participants with either intact or impaired TC. Using the TCI as a proxy for TC, TCI Elements and Total score were found to be associated with measures of general cognition, language, memory, executive function, and knowledge speed. Performance on measures from the aforementioned cognitive domains were also found to vary according to TC status (i.e., intact versus impaired).

Although this information adds to the research literature and is useful in a clinical context, expanding on this information would be helpful for clinicians and attorneys trying to determine the relative risk of TC impairment exhibited by a patient or client. Thus, we next constructed two regression models using the five cognitive variables that were most associated with TCI Elements in our sample (each of the four cognitive variables belonged to the cognitive domains of executive function, verbal memory, or semantic knowledge). For one model, we used the aforementioned cognitive variables to predict TCI Total scores. For the other model, measures of executive function, verbal memory, and semantic knowledge were used to identify participants with either intact or impaired TC. As expected, although both models accounted for over 70% of shared variance, the resulting models did not account for the full variance on TCI performance. This means other variables (e.g., sensory processes, fatigue, anxiety, disease factors, etc.) may be contributing. Nevertheless, the resulting models indicated that TC decline and impairment is largely driven by declines in executive function and verbal memory. These results provide clinicians useful information when questions about TC arise. For instance, if a patient with memory deficits is expressing interest in changing a will, then clinicians on the patient’s treatment team should recommend that a legal examination of TC take place before any will changes being enacted.

The pattern of scores on the DRS-2 indicates that loss of ability in specific cognitive domains represents a better indicator of TC impairment than does general cognitive function. For example, two participants in the current study obtained scores of 134 and 132 on the DRS-2. However, the participant scoring higher on the DRS-2 (i.e., 134) was observed to have impaired TC in relation to healthy controls, whereas the participant scoring lower on the DRS-2 (i.e., 132) was observed to exhibit intact TC in relation to healthy controls. However, the participant scoring higher on the DRS-2 (i.e., 134) exhibited impairments on measures of executive functioning and verbal memory, whereas the participant scoring lower on the DRS-2 (i.e., 132) did not. Thus, scores on measures of executive function and verbal memory were better predictors of TC status than were scores on the DRS-2, which is one of the most widely used measure of general cognitive function administered in the older adult population. Thus, clinicians should not rely on DRS-2 scores or dementia cutoffs to make clinical decisions about TC. Instead, verbal memory and executive function should be given preference over measures of general cognitive function.

The finding that TC impairment is strongly associated with executive function, verbal memory, and semantic knowledge make sense from a conceptual perspective and are supported by a conceptually based article published previously by our group (Marson, Huthwaite, Hebert, 2004). Different aspects of TC include the ability to understand the purpose of a will, knowledge of personal property and belongings, express knowledge of potential heirs, and have a plan for asset distribution to heirs following death (Frolik, 2001; Haldipur & Ward, 1996; Walsh, Brown, Kaye, & Grigsby, 1997). For someone with verbal memory impairment or difficulties with semantic knowledge, knowledge about his/her property and potential heirs and understanding of a will may be diminished. Executive functioning is an umbrella term used to describe regulatory cognitive processes that both involve and coordinate working memory, inhibition, sequencing, sustained and selective attention, multitasking, and cognitive flexibility (Chan, Shum, Toulopoulou, & Chen, 2008). For someone with deficits in executive function, the ability to develop a plan for asset allocation may be compromised.

As discussed in our companion paper (Martin et al., in review), TC is highly dependent on autobiographical/personal historical memory. For that reason it may be initially surprising that our measures of episodic memory and executive function proved to be significant predictors of autobiographical TCI components (i.e., knowledge of heirs, knowledge of assets). However, the scientific literature in ad shows that performance on measures of complex working memory, inhibitory processing, episodic recall, and generative fluency are actually strongly associated with autobiographical memory (Bahk & Choi, 2018). Studies of neuroimaging also support the connection between episodic memory and autobiographical memory. For example, strong and relatively consistent network associations involving aspects of the default mode network frontal lobe regions (i.e., ventral medial prefrontal cortex) and parahippocampal/entorhinal/hippocampus regions to autobiographical memory have also been found (El Haj, Antoine, Nandrino, & Kapogiannis, 2015). Thus, measures of episodic memory used in this study may be able to serve as forms of proxies for autobiographical memory such as knowledge of heirs and knowledge of assets.

Finally, at this juncture, it is interesting to compare these empirical results to our prior conceptualization of the cognitive correlates of TC hypothesized in an earlier article (Marson, Huthwaite, Hebert, 2004). In the prior conceptual paper, we theorized that autobiographical memory, semantic memory, verbal abstraction, and expressive language would cognitively underpin TCI Element 1 (plan), verbal memory would be strongly associated with TCI Elements 2 and 3 (assets and heirs), and verbal memory, semantic memory, and executive function would underlie TCI Element 4 (plan). Given the dearth of empirical studies in this area, in the current paper, we employed an exploratory approach in cognitively modeling the TCI elements, by including in the regression models the four cognitive variables that showed the highest associations with each TCI element. The results showed that, with varying orders, verbal memory, executive function, and semantic memory correlated with each of the four TCI elements. Thus, the current results provided some preliminary support for the prior 2004 conceptualization, but also extended beyond it, by suggesting that a similar multidimensional cognitive basis informs each TCI element. These initial empirical findings will need further investigation and elaboration by the field.

There are some limitations and future directions to be considered. First, our sample was relatively small, largely Caucasian, and well educated. Larger and more diverse samples should be examined in future studies. Second, study groups included healthy controls and persons with ad, which represents people at opposite ends of cognitive functioning. Future studies should examine the utility of the study models in people with less severe cognitive impairment (e.g., mild cognitive impairment). Third, anxiety may affect performance on tests with a cognitive component. Future studies should consider including a measure of anxiety when cognitive testing is conducted. Fourth, although we have previously published a conceptually driven article that detailed potential cognitive correlates of different aspects or elements of TC, the models included in the current study were exploratory in nature due to the lack of empirical studies on the topic. Thus, the methods employed may have led to overfitting of study models and future studies are needed to verify our findings. Finally, and likely most importantly, a rating of impairment on our measure of TC does not represent legal determinations of having or lacking TC. As stated previously, determinations of TC vary from state to state and often a low threshold of ability can lead to a person being determined able to make or change a will according to legal standards. In the current study, we took a clinical approach and analyzed the performance of persons with ad in relation to healthy controls and made determinations of impairment based upon the mean performance of the control group.

Regardless of limitations, however, the current study contains several important points. First, TC is a cognitively mediated legal construct that is highly affected by verbal memory, executive function, and semantic knowledge. Second, when making determinations about the need for a legal evaluation of a person’s TC, clinicians should more heavily weigh scores on measures of verbal memory, executive function, and semantic knowledge than performance on measures of general cognitive function. Finally, by the time a diagnosis of ad is applied, virtually all older adults wishing to make or amend a will should undergo a legal evaluation of TC.

Funding

Clinical data collection in this study was supported in part by grants from the National Institute on Aging (Alzheimer’s Disease Research Center (P50 AG16582) and (R01 AG021927).

Disclosures

The Testamentary Capacity Instrument (TCI) measure was created by senior author Dr Marson as an unfunded project and is owned by the UAB Research Foundation (UABRF). Dr Marson and the UABRF have not received licensing fees or royalties for the TCI.