Assessment of Testamentary Capacity in Older Adults: Description and Initial Validation of a Standardized Interview Instrument

Abstract

Objective

Testamentary capacity (TC) is a legal construct, which concerns a person’s mental capacity to make or amend a will. Although expert clinicians are frequently asked to assess TC in forensic settings, there are few instruments and little empirical research to inform and guide their assessments. The present study describes the development and psychometric properties of a standardized assessment measure of TC (Testamentary Capacity Instrument-TCI), and investigates its reliability and validity.

Methods

The TCI is an interview-based, psychometric measure, which assesses a testator’s knowledge of four conceptual elements, which together comprise the legal basis for TC in the Anglo-American legal system: (1) what a will is, (2) nature and extent of assets/property, (3) possible heirs/claimants to property, and (4) plan to distribute assets to heirs after death. Cronbach’s alpha and percentage exact agreement were used to examine TCI element reliabilities. Using independent samples t-tests, MANOVA and MANCOVA, we investigated validity by comparing TCI element performance of cognitively intact older adults (n = 22) and older adults with ad dementia (n = 20).

Results

The TCI elements showed good internal consistency and good inter-rater reliability. The ad group performed significantly below the control group on all four TCI elements, with effect sizes exceeding 1.2, suggesting that the TCI has content and construct validity.

Conclusions

Relative to cognitively intact older adults, older adults with ad dementia showed significant impairment on all four TCI conceptual elements. The TCI has promise as a standardized quantitative measure of TC to support clinical assessment of TC in forensic settings.

Introduction

The freedom to choose how one’s property and other possessions will be disposed of following death—known as the right of testation—is a fundamental right under Anglo-American law (Frolik, 2001; Marson, 2020; Marson & Hebert, 2006). A key requirement of the law of testation is that a testator (person making the will) have testamentary capacity or competency (TC): “that measure of mental ability recognized in law as sufficient for the making of a will”(Frolik, 2001). If TC is lacking at the time of execution of the will, the will is invalid and void in effect, and a court will distribute the testator’s assets in accordance with either a preexisting will or statutory guidelines if no will exists. In the latter situation, assets may be distributed among the decedent’s survivors with little or no consideration of the decedent’s values or preferences (Frolik, 2001; Walsh, Brown, Kaye, & Grigsby, 2019). The legal requirement of TC exists across all state jurisdictions in the USA (Marson, 2020; Marson, Huthwaite, & Hebert, 2004).

TC is an important medical-legal construct in our aging society, and disputed matters of TC are increasingly common in the courts and legal system of the USA (Brenkel et al., 2018). We live in an unprecedented aging society where cognitive decline and dementia in later life are highly prevalent and readily diminish TC and other legal and clinical capacities in older adults (Shulman et al., 2015). In addition, the USA is an individualistic and litigious society where issues of conflicting family relationships and distribution of wealth are very often addressed through capacity litigation and the legal system rather than through private and informal family discourse. Rates of divorce and remarriage have increased dramatically in recent decades resulting in “blended” families, heightened family tensions, and conflict about inheritance decisions (Brenkel et al., 2018; Nedd, 1998). Current economic trends also reflect that substantial wealth is now being transferred from Baby Boomers to successive generations, providing additional impetus for conflict with regard to distribution of wealth among potential heirs (Brenkel et al., 2018; Nedd, 1998). Thus, TC has significant important legal, economic, and public policy implications.

TC is also increasingly an area of clinical forensic assessment. The proliferation of capacity and undue influence litigation in probate and other civil courts has led to increasing involvement of clinicians experienced in forensic assessment (Mart, 2016; Peisah & Shulman, 2012). Clinical neuropsychologists, psychiatrists, geriatricians, and other mental health professionals are increasingly being asked by attorneys, courts, and sometimes family members, to conduct either a contemporaneous (prospective) evaluation of the TC of a living testator (Brenkel et al., 2018), or a retrospective evaluation of the TC of a now deceased testator at the time the will was executed (Voskou, Douzenis, Economou, & Papageorgiou, 2017).

Contemporaneous assessments of TC represent evaluations of a testator conducted close in time to a will execution and are intended to provide potential supporting clinical evidence to the testator’s attorneys charged with determining whether a client legally has TC. Contemporaneous TC evaluations are typically requested when concerns exist regarding a known mental illness or neurological disorder that may affect the testator’s decisional competency, or when ongoing or anticipated family conflict foreshadow a subsequent will challenge (American-Bar-Association, 2005; Haldipur & Ward, 1996; Spar & Garb, 1992).

Although the demand for contemporaneous TC assessments is increasing, assessment approaches vary widely in quality and approach (Brenkel et al., 2018; Marson et al., 2004). This is due to numerous factors, including the absence of an accepted conceptual and theoretical framework of TC validated by empirical research (Marson et al., 2004), the lack of standardized direct assessment instruments of TC (Marson & Hebert, 2006), limited to nonexistent formal training for conducting TC assessments among health care professionals (Grisso, 2003; Marson, Schmitt, Ingram, & Harrell, 1994), and difficulties inherent in forensic assessments that depend on corroborating information from family members and other informants who may also be in a position personally to benefit from the execution of the will (Marson & Hebert, 2006).

For similar reasons, very few empirical studies exist to inform and guide clinical assessment of TC (Heinik, Werner, & Lin, 1999; Papageorgiou, Voskou, Economou, Beratis, & Douzenis, 2018; Roked & Patel, 2008). To date we have identified only four such empirical studies. Heinik et al. (Heinik et al., 1999) asked 31 community dwelling elderly persons seen in a geropsychiatric outpatient service to define “testament” as part of their clinical interview (The term “testament” in this study appears to be equivalent to the meaning of a “will” as defined by Anglo-American law.) Participant responses were rated according to a six-item Testament Definition Scale (TDS) using the following criteria: (1) a testament is a document, (2) a testament is made by a person, (3) a testament is made during a person’s lifetime, (4) a testament involves property, (5) a testament involves a receiver or receivers [heirs], and (6) a testament comes to fruition after death. In the study, the most common elements of a “testament” identified by participants included that it involved a person (58.1%) and that it took effect after a person’s death (45.2%). Less than 40% of the sample identified the other key four criteria of testament.

In another early study (Roked & Patel, 2008), TC was assessed in patients with mild (n = 27), moderate (n = 27), and severe (n = 20) Alzheimer’s disease (ad) [dementia] who were age 55 or older. Assessment of TC was based on guidelines outlined by the British Geriatric Society, British Medical Association and The Law Society (Association & Society, 2004). The guidelines operationalized TC criteria set forth in the British legal case of Banks v Goodfellow (1870), which consist of a testator’s (1) understanding of the testamentary act or its effect, (2) understanding of the extent of the property being disposed, and (3) appreciation of the claims to which the testator ought to give effect. The study stated that an “independent investigator” was trained to assess TC using the above criteria via interviews of study participants, and these interviews were then independently assessed by a second rater who was a geriatric psychiatrist. It was not clear in what form the interview information was provided to the second rater, and it was also not stated whether a participant’s capacity status was based on one or both rater assessments, or how capacity status was determined in cases where the TC outcomes of the two raters differed.

The study found that dementia severity was associated with loss of TC, with 6% of mild ad patients, 38% of moderate ad patients, and 56% of severe ad patients found to lack TC (Roked & Patel, 2008). In addition, using logistic regression, two global cognitive screening measures (the MMSE and CAMDEX-R) both equivalently predicted participants’ TC status in 87% of cases. At the cognitive domain level, language measures predicted TC status in 84% of the cases, whereas short-term memory and attention were not good predictors of TC status.

As part of a broader financial capacity study, our own group assessed participants’ knowledge of their financial assets, a financial ability highly relevant to TC (Earnst et al., 2001) We assessed “knowledge of personal assets and estate arrangements” in a sample of 23 cognitively intact older adults and 20 persons with mild to moderate ad type dementia. Results indicated that the ad group performed significantly below controls concerning knowledge of their financial holdings and estate arrangements. The findings thus suggested that cognitive decline associated with ad would impair an individual’s recall and understanding of the nature and extent of his/her property and belongings—a core TC element for valid execution of a will.

A major limitation of these early empirical TC studies is that they did not involve a psychometric measure of the full legal construct of TC, but instead addressed smaller components. In an important paper, Shulman and his group recently made a strong case for the field’s need of a conceptually grounded standardized psychometric measure of TC that could serve as a contemporaneous assessment instrument (CAI) in forensic settings (Brenkel et al., 2018). This is a need that our own research group also made less formally in an earlier paper in which we conceptually outlined a prototype testamentary capacity assessment instrument (TCI) [Marson et al., 2004; Brenkel et al., 2018].

The field appears ready now to embark on this task. Dr. Papageorgiou and colleagues recently developed the Testamentary Capacity Assessment Tool (TCAT), a brief specialized instrument for TC screening in patients with dementia (Papageorgiou et al., 2018). The TCAT has four subtests that reflect the authors’ clinical conceptual model of TC for a testator: (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. Using the judgments of an expert psychiatrist as the gold standard, the authors found that the TCAT and an assigned cut score showed good reliability, and sensitivity and specificity in predicting the expert clinician outcomes. They concluded that the TCAT appeared to be a reliable measure for screening TC in dementia, which could be used by both expert and non-experts alike (Papageorgiou et al., 2018).

While the TCAT represents a creative and novel assessment tool, it has some conceptual limitations, insofar as its model of TC is not theoretically grounded and appears idiosyncratic, and in particular is not linked to established legal elements and models of TC found in Anglo-American and other international legal systems. As articulated by Appelbaum and Grisso (Appelbaum & Grisso, 1995) and as noted by Shulman and his group (Brenkel et al., 2018), a CAI measure of TC should use measurement tasks “derived from the legal standards that a court would find most relevant.” (Brenkel, p. 28). Thus currently the legal relevance of the TCAT conceptual model and test items to TC and to probate litigation remains unclear.

As the above handful of studies makes apparent, only modest scientific research to date has been conducted in the area of TC. There continues to be a need for a conceptually well-grounded psychometric measure of TC—linked to established legal elements or standards of TC—that can support contemporaneous clinician assessments of TC in forensic settings. There also continues to be an equivalent need for related empirical studies of TC as a forensic construct.

In this study we follow up on our earlier conceptual work in TC (Marson et al., 2004), and present an interview-based psychometric assessment measure of TC called the Testamentary Capacity Instrument (TCI). Our purposes are twofold: (1) to describe the TCI’s development and psychometric properties (Haldipur & Ward, 1996; Walsh et al., 2019); and (2) to present initial evidence of validity by investigating the TCI’s ability to discriminate the performance of cognitively-intact older adults and persons with ad type dementia.

Methods

Participants

Twenty persons with mild-to-moderate Alzheimer’s type dementia (ad) and 22 cognitively intact older adults (controls) 60 years or older participated in the study. Control and ad participants were recruited from a longitudinal research study conducted through the Alzheimer’s Disease Research Center (adRC) at the University of Alabama at Birmingham (UAB).

Control or ad diagnosis was established prior to study recruitment. A diagnosis of Alzheimer’s type dementia was based upon guidelines set forth by the National Institute for Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Dementias Association (McKhann et al., 1984). Participants with ad and older controls underwent a thorough medical and neurological history and examination, laboratory evaluation, neuroimaging, and a comprehensive neuropsychological evaluation. Controls and ad participants were excluded from the study if at the time of participation they suffered from any of the following conditions, which have been found to interfere with cognitive function: neurodegenerative diseases 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 and materials: testamentary capacity instrument (TCI)

The testamentary capacity instrument (TCI) is a standardized interview-based instrument designed to help expert clinicians assess TC in forensic settings. The TCI is designed to clinically assess four conceptual elements that together comprise the legal construct of TC in Anglo-American law and found across state legal jurisdictions in the USA:

• 1) Testator’s knowledge of the meaning and purpose of a will (E1-Purpose);

• 2) Testator’s knowledge of the nature and extent of his/her property (E2-Assets);

• 3) Testator’s knowledge of his/her heirs or “objects of bounty” (E3-Heirs); and

• 4) Testator’s knowledge of basic plan for disposition of assets to heirs (E4-Plan).

These four TC elements were selected based on a review of the Anglo-American legal literature (Brenkel et al., 2018; Walsh et al., 2019) and in particular derive from the previously referenced seminal legal case of Banks v. Goodfellow (Banks v. Goodfellow, 1870). As discussed in more detail below, for each TCI element, the senior author developed a range of individual test questions and scoring to quantitatively assess clinically different aspects relevant to the specific TC legal element/construct.

Description, structure, scoring of element 1 [purpose of will]

The first TCI element concerns a testator’s knowledge of the meaning and purpose of a will (Spar & Garb, 1992; Walsh et al., 2019). E1 (Purpose) consists of 13 test items and has a maximum score of 28 points. Question types include a definitional question (direct production), a declarative knowledge question (direct production), 9 true-false questions, and 2 multiple choice questions. The testator is first asked to indicate whether or not he/she has a will (item is not scored). The testator is next asked to state in his or her own words what a will is and what it does. This response is scored using defined criteria. The testator then answers true/false questions concerning the nature and purpose of a will. Other direct production and recognition questions tap the steps involved in making a will, and the disposition of assets.

Description, structure and scoring of element 2 [assets]

The second TCI element concerns a testator’s knowledge of the nature and extent of his/her assets that would pass by will to his/her designated heirs (Spar & Garb, 1992; Walsh et al., 2019). This is a more challenging element than Element 1 [Purpose], as it requires a testator to recall the different types of assets he or she owns, and to have knowledge of their relative current value. A testator’s knowledge of his or her current property will depend on both historical and short-term memory for assets acquired, sold, gifted, or otherwise relinquished both recently and in the past (American-Bar-Association, 2005; Marson et al., 2004).

Element 2 first tests for asset knowledge by having the testator review 20 test questions, which each reference a different common form of property (e.g., home, automobiles, savings accounts, works of art, etc.). For each question, the participant answers whether he/she owns that type of property (yes/no), and also provides a financial value for the type of property owned. Because independent and accurate estimation of an asset’s value may be difficult, for example due to appreciation or depreciation changes over time, participants provide answers within value ranges.

Scoring is based on a comparison of the testator participant’s ownership and value responses to those of a trusted informant (collateral source) who possesses accurate knowledge of the testator’s property holdings. The scoring metric for each question is 2, 1, 0, with 2 points awarded if the participant correctly responds concerning both ownership (or not) of the type of property, and its relative value. Participants get 1 point if they respond correctly concerning ownership of property type (yes/no), but misjudge its relative value range. 0 points are awarded if the participant answers incorrectly concerning his/her ownership of the type of property.

Question 21 of Element 2 asks participants to generate spontaneously a list of assets that they own which could be included in a will. The participant’s responses are then compared to the property ownership responses provided by the collateral source for the preceding original 20 property questions.

Because some states emphasize the role of memory function in TC (Walsh et al., 2019), Question 22 asks participants to name three personal assets that they intend to include in their will. Actual short-term recall for these three items is subsequently assessed as part of Element 4 (Plan) (see below).

Scoring of element 2 [assets]

TCI Element 2 consists of 22 questions and has a maximum score of 46 points. As discussed, Questions 1–20 use a recognition memory format to test knowledge of types of property ownership and respective property values, and together have a score range of 0–40 points. Question 21 concerns property the testator owns and wishes to pass by will. Scoring for this question is first calculated as a percentage agreement figure (0–100%) reflecting the percent agreement of property items generated by the participant that were also previously endorsed by the informant (collateral source) as being owned by the participant. The percent agreement score is then converted to a centile score with 0 points = 0–24% agreement, a 1 point score = 25–49% agreement, a 2 point score = 50–74% agreement, and a 3 point score = 75–100% agreement.

Question 22 is a production item that asks the participant to name three items he/she wishes to pass by will and has a score range of 0–3. The participant’s recall of these three items is separately tested later in the TCI.

Description, structure and scoring of element 3 [heirs]

Element 3 of the TCI addresses a testator’s knowledge of his or her heirs, also known as the “objects of one’s bounty” (Spar & Garb, 1992; Walsh et al., 2019). The testator is asked to recall relatives who may have a legal claim to his/her property or possessions after death, as well as friends, associates, beloved pets, and charities the testator may wish to include in the will (Spar & Garb, 1992; Walsh et al., 2019). This recollection of potential heirs is independent of their actual final status within the testator’s will.

Element 3 consists of four questions designed to assess a testator’s knowledge of potential and actual heirs. Question 1 asks whether the testator is or has been married, and requests information regarding spouse’s name, age, and if he/she is living or deceased. Question 2 asks about the testator’s children, and requests information regarding name, residence, age, and if the child is living or deceased. Question 3 asks about close relatives other than spouse and children, including grandchildren and siblings, and requests information regarding name, residence, age, and if the close relative is living or deceased. Question 4 asks the testator independently to name three heirs that the testator intends to include in his/her will, which can include groups and organizations as well as persons, and the nature and length of the relationship. These actual heirs may or may not overlap with the spouse, family members and close relatives identified in Questions 1–3.

Scoring of element 3 [heirs]

The total score range for Element 3 is 0–16 points. As was the case with Element 2, scoring for this domain is dependent on comparisons of participant responses to corroborating information obtained from a knowledgeable informant. Question 1 concerning a possible spouse has a range of 0–4 points (3 points for correct name, 1 point for age).

The testator can earn up to six points for each listed child or other relative. Three points are earned for correct name, an additional two points for correct location (city/state), and one point for correct age. Due to variability in the number of children, siblings, and grandchildren that would be potentially listed by each testator, Questions 2 and 3 reflect a percentage coefficient in which total points earned by the testator as described above and are then divided by total points possible as established by the knowledgeable informant. As with E2 Question 21, the percent coefficient score for Questions 2 and 3 are both then converted to a centile score with 0 points = 0–24% agreement, a 1 point score = 25–49% agreement, a 2 point score = 50–74% agreement, and a 3 point score = 75–100% agreement.

Question 4 of Element 3 has a maximum of 6 points. It examines a testator’s ability to independently identify three persons, groups, or organizations that he or she intends to include in a will, and to indicate the nature and length of his/her relationship to these heirs.

Description, structure, scoring of element 4 [plan]

Element 4 of the TCI concerns a testator’s ability to describe a basic plan for disposition of his or her assets to heirs under a will (Spar & Garb, 1992; Walsh et al., 2019). This element involves the participant’s integration of information from the three prior Elements, in particular Element 2 [Assets] and Element 3 [Heirs].

Element 4 is comprised of five questions. In Question 1, the participant is asked to describe his/her plan for distributing his/her property to chosen heirs after death. In Question 2, the testator is asked to recall the three assets that he/she previously named in Element 2 as property to be passed by will. For each of Questions 3–5, the testator is asked to indicate the heir to which each asset would pass by will, and the reasons for making such distribution to this heir.

Element 4 consists of five questions and has a total possible score range of 0–14 points. Question 1, which concerns the testator’s description of a plan for his/her will has a score range of 0–2. For Question 2, a participant receives a point for recalling each of the three assets previously identified, and thus carries a range of 0–3. Each of Questions 3–5 has a score range of 0–3, for a collective point range of 0–9.

In summary, TCI E4 (Plan) consists of 13 scored test items and a maximum score of 28 points. Question types include a definitional question (direct production), a property short-term recall item (direct production), and 3 property-heir distribution questions (direct production).

Cognitive and mood measures

Dementia rating scale, second edition (DRS-2)

The DRS-2 is a brief but comprehensive measure of cognitive status in older adults with neurocognitive conditions such as ad and related disorders (Jurica, Leitten, & Mattis, 2001) (29). Five subscales comprise the DRS-2, including attention, initiation and perseveration, construction, conceptualization, and memory.

Folstein mini mental state examination (MMSE)

The MMSE (Folstein, Folstein, & McGugh, 1975) is a popular screening measure of global cognitive functioning consisting of 11 tasks that assess orientation, memory, attention, language, ability to follow commands, and visual construction. The maximum score is 30 points.

Wide range achievement test-third edition—Reading subscale (WRAT-3 reading)

WRAT-3 Reading assesses reading achievement in children and adults (Wilkinson, 1993) and involves a person reading aloud a list of words. Current and prior versions of the Reading subtest (WRAT-R and WRAT-3) have been found to correlate highly with other measures of Verbal and Full Scale Intelligence Quotients (VIQ and FSIQ) derived from Wechsler Scales (Lezak, Howieson, & Loring, 2004; Wilkinson, 1993).

Geriatric depression scale (GDS)

The GDS is a widely-accepted, self-report screening measure of depression in older adults (Yesavage et al., 1982).

Design and procedures

Written informed consent meeting the requirements of The University of Alabama and UAB Institutional Review Boards was obtained from study participants. Consent to participate in this research study was obtained from a family member or other legally authorized representative for ad patients who lacked consent capacity.

Participation in the study was completed in one or two visits based on the participant’s schedule, preference, or observed level of fatigue toward the end of the initial visit. Due to difficulties associated with contacting and scheduling participants, the interim period between initial cognitive testing and subsequent administration of the TCI was greater than 1 month for 38% of participants but typically fell within 3 months or less.

The TCI was administered in clinic following completion of the cognitive testing on the first assessment day, or at the participant’s home on a second assessment day. The TCI was recorded to assist with scoring, ensure standardized administration, and allow for analysis of inter-rater reliability. Participants were paid $20 cash following completion of the TCI.

Informant data

All participants were asked to identify an informant knowledgeable about the participant’s assets and the participant’s potential heirs such as a spouse, other family member, or close friend (Spar & Garb, 1992). The informant completed a separate questionnaire providing information regarding the participant’s assets and family members, which was used in scoring Element 2 and Element 3 of the TCI. Informant questionnaires were completed either at the time of the visit at home and mailed back to the investigator in a pre-paid stamped envelope, or over the telephone. Informants were not compensated financially for completing the questionnaire. The present study obtained respondent interviews for all 42 participants (100% completion rate), allowing for proper assessment of memory for assets and potential heirs as well as suitability of scoring criteria.

Statistical analyses

Power analyses

With 80% power, an effect size of 1.2 was used to detect group differences at an alpha level of 0.01. Previous research involving similar samples produced an equivalently large effect size (d = 1.23) for detecting group differences in knowledge of personal assets and estate arrangements (Marson, 2001).

Demographic comparisons

Chi-square analyses or Fisher Exact tests (when the chi-square test was not appropriate) were performed to investigate differences in frequencies for demographic variables of sex and ethnicity. Between-group differences in demographic variables of age and education were examined using independent samples t-tests.

Group comparisons

Multiple analyses were performed to compute differences between participant groups on cognitive and TCI variables. First, multivariate analysis of variance (MANOVA) was performed to minimize Type I error resulting from multiple univariate analyses. Next, a series of independent samples t-tests were conducted with a Bonferroni-corrected alpha level (p ≤ 0.01) to adjust for multiple comparisons. The potential influence of covariates on group differences in TCI elements was examined using multivariate analysis of covariance (MANCOVA).

Assessing violations of assumptions

The assumptions of univariate and multivariate ANOVAs and Pearson product–moment correlation were checked using Levene’s test for equality of variance and Box’s test for equality of covariance as well as the Kolmogorov–Smirnov and Shapiro–Wilk tests for normality. For those TCI elements that violated parametric assumptions of normality and homogeneity of groups, nonparametric tests were performed (Mann Whitney U and Spearman rho correlation coefficient). Results of nonparametric tests also met pre-established alpha levels for between group comparisons (α ≤ 0.01) and level of association between variables (α ≤ 0.025). All scores were checked for outliers, using a 99% confidence interval, and analyses were recalculated excluding participants with moderate ad (n = 4) and outliers. Results from all of these secondary analyses were unchanged. All analyses were performed using SPSS for Windows Release 13.0.

Results

Participant characteristics

Concerning assessment location, of the 41 participants who completed both the TCI and the cognitive test measures, 7 controls (17%) and 3 ad participants (7%) elected to complete the TCI at the conclusion of their visit. The remaining 31 participants were administered the TCI on a separate day at their home.

The interim period between initial cognitive testing and subsequent capacity testing ranged from 0 to 5 months, with the majority of participants (n = 35, 85%) completing the TCI within less than 3 months of initial cognitive testing. The average interim period for the control group was 38.27 days (SD = 48.27) with a range of 0 to 137 days, whereas the average interim period for the ad group was 43.42 days (SD = 45.95) with a range of 0 to 158 days. An independent samples t test revealed no significant difference between participant groups based on the interim period (t (40) = −5.39, p = 0.593).

Table 1 compares demographic variables across groups. Control (n = 22) and ad (n = 20) groups did not differ with respect to age, education, sex and ethnicity. The average age of the ad group was 75.6 years with a range of 60 to 82 years versus the control group, which was 72.6 years with a range of 60 to 91 years. An independent samples t test revealed no significant difference between groups based on age, t (40) = −1.10, p = 0.277). Groups were also statistically equivalent with regard to years of education, t (40) = −0.17, p = 0.865, with each group closely approximating 15 years of education.

Table 1.

Participant demographic and clinical characteristics

	Controls (n = 22)	ad group (n = 20)	Effect	p
Gender (M/F)	9/13	10/10	.35a	0.55
Race (W/AA)	17/5	18/2	1.22a	0.27
Age	72.6 (7.0; 60–82)	75.6 (10.5; 60–91)	−1.10b	0.28
Education	14.9 (2.1; 12–20)	15.1 (3.2; 9–20)	−.17b	0.87
WRAT-3 Reading	107.5 (8.2; 87–117)	96.7 (13.8; 66–118)	3.1b	0.004
MMSE	29.2 (1.0; 26–30)	22.4 (4.6; 13–28)	6.9	<0.001
DRS-2c Total Score	139.9 (3.0; 132–144)	113.6 (14.6; 82–134)	8.2	<0.001
GDSd	3.9 (4.4; 0–17)	7.2 (6.0; 0–22)	−2.0b	0.053

^aChi-square

^bIndependent samples t-test

^cDementia Rating Scale-2

^dGeriatric Depression Scale; scores within () = mean/standard deviation and range.

Females constituted 50% (n = 10) and 59% (n = 13) of the ad and control groups respectively, with no significant group differences in gender composition, χ² (1) = 3.49, p = 0.554. Regarding ethnicity, 18 ad participants (90%) and 17 controls (77%) were Caucasian, while the remaining 2 participants with ad (10%) and 5 controls (23%) were African-American. There were no significant ethnicity differences across groups, χ² (1) = 1.22, p = 0.269.

Groups were found to differ with regard to reading level, with the control group performing significantly better than the ad group (WRAT-3 reading), t (38) = 3.07, p = 0.004. As the groups were generally equivalent with regard to ethnic composition, the differences in reading level are believed to reflect a degradation of reading ability resulting from cognitive decline associated with ad rather than differences in the quality of education received (Manly, Jacobs, Touradji, Small, & Stern, 2002; Manly, Touradji, Tang, & Stern, 2003).

Although differences in self-ratings of depression between groups approached significance, t (39) = −1.99, p = 0.053, with the ad group demonstrating a higher level of depressive symptoms, both group means fell below the clinical threshold for mild depression.

We performed a MANCOVA to address the potential influence of reading level and depressive symptoms on group differences in TCI scores. Results demonstrated that significant group differences on TCI elements remained after adjusting for both reading level and self-rated depression.

Participant performance on cognitive measures

Control and ad groups differed significantly on the global cognitive screens of the MMSE and the DRS-2, with the control group scoring substantially better (higher) than the ad group on both measures (see Table 1). Cognitive test results were unavailable for one moderate ad participant whose MMSE score was 13 out of 30 and fell below the cut score needed for administration of neuropsychological testing.

Participant performance on the TCI

A series of independent sample t-tests applying a Bonferroni correction demonstrated significant group differences on the four elements of the TCI (all ps < 0.001) (Table 2). For Element 1 (Purpose), the controls performed better (23.9, SD 2.1) than ad participants (mean 18.8, SD 3.8) with a Cohen’s d value of 1.7. Compared to controls, ad patients had an impaired ability to define the features of a will and explain steps to making a will.

Table 2.

Group performance on four TCI elements

Variable	Controls (n = 22)	ad (n = 20)	t/X2/Z	p	d
Element 1—purpose
Q2—what is will	6.3 (1.4), 4–8	4.6 (1.6), 0–7	3.8	<0.001	1.1
Q3–10—T/F re wills	7.2 (0.9), 5–8	5.8 (1.2), 4–8	4.5	<0.001	1.3
Q11—steps making will	6.5 (1.1), 4–8	4.9 (1.9), 1–8	3.2	0.002	1.0
Q12—not a step for will	1.0 (0.0), 1–1	0.9 (0.4), 0–1	3.6	0.099	n/a
Q13—effect of death	2.0 (0.0), 2–2	1.7 (0.7), 0–2	3.6	0.099	n/a
Q14—assets after death	1.0 (0.2), 0–1	1.0 (0.2), 0–1	0.005	1.0	n/a
Total score (max = 28)	23.9 (2.1), 19–28	18.8 (3.8), 8–25	5.6	<0.001	1.7
Element 2—Assets
Q1–20—asset type/value	34.1 (2.8), 27–38	29.3 (3.7), 24–40	4.8	<0.001	1.5
Q21—poss assets in will	1.9 (0.6), 1–3	0.8 (0.8), 0–2	4.1	<0.001	1.6
Q22—name 3 will assets	3.0 (0.0), 3–3	2.2 (1.2), 0–3	3.0	0.003	0.9
Total Score (max = 46)	39.0 (2.6), 33–43	31.8 (6.0), 14–45	5.2	<0.001	1.6
Element 3—heirs
Q1—marriage/status	3.95 (0.2), 3–4	3.0 (1.0), 1–4	4.0	<0.001	1.3
Q2—children/status	1.91 (0.61), 1–3	0.80 (0.77), 0–2	4.1	<0.0001	n/a
Q3—close relatives/status	3.0 (0.0), 3–3	2.6 (0.9), 0–3	2.5	0.014	0.6
Q4—name 3 heirs in will	6 (0.0), 6–6	4.6 (2.3), 0–6	2.8	0.007	0.9
Total Score (max = 16)	16.0 (0.2), 15–16	12.8 (4.3), 1–16	3.5	0.001	1.1
Element 4—plan
Q1—description of plan	2.0 (0.2), 1–2	1.2 (0.8), 0–2	3.8	<0.001	1.4
Q2—name 3 will assets	2.9 (0.3), 2–3	1.2 (1.2), 0–3	5.4	<0.001	1.9
Q3—heir for asset 1	2.7 (0.5), 2–3	2.3 (0.8), 0–3	2.0	0.050	0.6
Q4—heir for asset 2	2.6 (0.5), 2–3	2.4 (1.0), 0–3	0.6	0.588	0.3
Q5—heir for asset 3	2.7 (0.5), 2–3	2.2 (0.9), 0–3	2.0	0.048	0.7
Total Score (max = 14)	12.9 (1.1), 11–14	9.2 (2.5), 1–13	6.2	<0.001	1.9

Note. Values are mean (SD), range. t/Z = value for independent t test or Mann–Whitney U (questions 1–5). ad = Alzheimer’s disease, TCI = Testamentary Capacity Instrument, d = Cohen’s d, n/a = not applicable.

For Element 2 (Assets), ad patients in comparison with controls had lower agreement levels with their Respondents as to their assets owned and the value of those assets (Element 2, see Table 2). When examining score distributions (i.e., 0, 1, 2 points) between the groups, the ad group had higher frequencies of 0 and 1 point scores as exemplified on Questions 5 and 9 (see Table 3). ad participants also had consistently lower agreement levels with their Respondents regarding information as to the names, ages, locations, and living/deceased status of their family members (i.e., potential heirs) as compared controls (Element 3, see Table 2). Finally, relative to controls, ad patients were impaired in their ability to describe their plans for distributing personal assets to prior identified heirs (Element 4, see Table 2). For example, controls were better at naming those heirs, and providing specific details and a rationale for disbursement of each listed asset.

Table 3.

Score distribution for Questions 5 and 9 of TCI element 2 (Knowledge of Assets)

Score (points)	Q5—own cars?/value		Q9—own savings acct?/value
	C (n = 22)	AD (n = 20)	C (n = 22)	AD (n = 20)
0	0 (0.0)	2 (10.0)	2 (9.1)	7 (35.0)
1	11 (50.0)	13 (65.0)	8 (36.4)	9 (45.0)
2	11 (50.0)	5 (25.0)	12 (54.5)	4 (20.0)

Note. cells = frequency (%).

C = control, AD = Alzheimer’s disease.

For ad participants, age was found to be inversely associated with knowledge of assets (Element 2: r = −0.45, p = 0.045) and potential heirs (Element 3: r = −0.45, p = 0.047). No significant age-related associations emerged within the control group.

Across the four TCI elements, no significant correlations were found between either group and years of education, or between either group and reading level.

TCI reliabilities

Table 4 lists internal consistency data for TCI Elements 1 and 4. Cronbach’s alpha for Element 1 (α = 0.58) fell below the threshold level of ≥ 0.60 (DeVellis, 1991; Nunnally, 1978). However, Element 1 achieved adequate internal consistency (0.64) upon deletion of three forced-choice items with low corrected item-total correlations (see Table 4). These items were maintained within the scale based on prior clinical recommendations that multiple-choice and forced-choice items be incorporated in the assessment of TC to accommodate for ad related difficulties in verbal expression (Spar & Garb, 1992). Cronbach’s alpha for Element 4 reached acceptable levels and was noted to increase from 0.67 to 0.70 after a memory item (Item 2) (unrelated to developing a plan for disposition of assets) was eliminated.

Table 4.

Scale reliabilities for the TCI elements

	Coefficient α (n = 42)	% Exact interrater agreement (n = 10)
E1: Understanding/purpose of a will	0.58	98.5
Item 2: OEa—define a will		92.5
Item 3: T/Fb—a legal document		100.0
Item 4: T/F—names person to make medical decisions		100.0
Item 5: T/F—plan for dealing out property/belongings†		100.0
Item 6: T/F—takes effect only after a person’s death†		100.0
Item 7: T/F—legally certifies your death		100.0
Item 8: T/F—must be written down to be legally binding†		100.0
Item 9: T/F—can be created or changed over the phone		100.0
Item 10: T/F—can distribute property before death		100.0
Item 11: OE—describe steps in making a will		84.0
Item 12: MCc—item 11		100.0
Item 13: OE—describes what happens to property in a will		100.0
Item 14: MC—item 13		100.0
E2: Knowledge of assets	—	100.0
E3: Knowledge of potential heirs	—	100.0
E4: Plan for disposition	0.67	92.5
Item 1: OE—describe your plan for distribution of assets		90.0
Item 2: OE—recall of 3 items earlier selected for will‡		100.0
Item 3: OE—identify heir and provide reason for choice		100.0
Item 4: OE—identify heir and provide reason for choice		80.0
Item 5: OE—identify heir and provide reason for choice		70.0

^aOpen-ended items

^bTrue/false items

^cMultiple choice items

^†Cronbach’s alpha for Element 1 increases to 0.64 after deleting these items.

^‡Cronbach’s alpha for Element 4 increases to 0.70 after deleting this item

Internal consistency estimates for TCI Elements 2 and 3 were not calculated due to differences across participants in the maximum number of points that could be attained per item within these scales. For instance, within Element 3 participants were scored on their recall of the name, age, and general place of residence (i.e., city and state) for each child. Participants varied with regard to number of children, thus requiring this item be scored as a percentage. As such, scores for items within the same scale varied widely and prohibited calculation of Cronbach alpha.

Table 4 lists the initial interrater reliability data for TCI Elements 1 and 4. Interrater reliability was calculated for Elements 1 and 4 where subjective interpretation may affect application of scoring criteria. For each item in Elements 1 and 4, we used a conservative criterion of percentage of exact agreement between two raters across a subsample of controls and patients with ad (n = 10). The percentage of agreement per item was then averaged for each element. Adequate interrater reliabilities (>80% exact agreement) were found for both Element 1 and Element 4 as well as for the items comprising these scales. The exception was Item 5 of Element 4 [who would you leave an asset to by will?] which was marginal (70% exact agreement). However, Item 5 is identical in form and task to Items 3 and 4 of Element 4, which achieved 100% and 80% exact agreement respectively.

TCI Elements 2 and 3 involve simple yes/no questions and listings of information with specific scoring criteria resulting in exact interrater agreement (100%).

Consistency of respondent and participant ratings of asset ownership and knowledge of heirs

As discussed, participant scores on elements assessing knowledge of assets (Element 2) and potential heirs (Element 3) were derived from corroborative information provided by a knowledgeable respondent. As such, both participants and their respondents were asked to rate the respondent’s knowledge of these areas on a Likert scale ranging from 0 (no knowledge) to 10 (excellent knowledge). Independent samples t-tests were conducted to determine if differences in respondents’ level of knowledge regarding participants’ assets and family existed between the two participant groups.

Table 5 presents results of group comparisons of respondents’ level of knowledge pertaining to participants’ assets and heirs. ad and control participants did not differ significantly in their ratings of their associated respondents’ knowledge of assets and heirs. Likewise, no group differences emerged with regard to respondents’ self-ratings of their knowledge of the participants’ heirs. However, a group difference emerged with respondents’ self-ratings of the participants’ assets, with ad respondents indicating a higher level of knowledge regarding participants’ assets than control respondents. Overall, both participants and respondents rated respondents as having very good to excellent knowledge of a participant’s assets and family members, a finding supportive of the use of respondent ratings across groups in the study.

Table 5.

Group differences in ratings of respondents’ level of knowledge pertaining to assets and heirs

	Controls (n = 22) X (SD)	ad Group (n = 20) X (SD)	t	P
Participant ratings
E2: Knowledge of assets	8.89 (1.22)	9.15 (0.96)	−0.77	0.445
E3: Knowledge of heirs	9.27 (0.869)	9.30 (0.79)	−0.11	0.916
Respondent ratings
E2: Knowledge of assets	8.84 (1.08)	9.72 (0.64)	−3.18	0.003
E3: Knowledge of heirs	9.16 (1.27)	9.52 (0.88)	−1.08	0.288

Discussion

Importance of testamentary capacity as medico-legal construct

Testamentary capacity (TC) is an important legal construct that concerns a person’s mental capacity to make or amend a will. In our aging society, legal disputes concerning the TC of older adults continue to increase (American Bar Association/American Psychological Association Assessment of Capacity in Older Adults Project Working Group, 2008). Enormous intergenerational transfers of wealth are currently underway between the World War II, Baby Boomer and Millennial generations (Feiveson & Sabelhaus, 2018), and these transfers are often accompanied by legal disputes about an older individual’s testamentary capacity to pass assets by will or trust. Neuropsychologists and other clinicians, in turn, are increasingly asked by attorneys and the courts to participate in these forensic proceedings by clinically assessing and offering testimony concerning the TC of older adults (Peisah & Shulman, 2012). These clinical forensic assessments, while forming only a part of the evidentiary basis of a civil competency case, often have significant legal consequences for testators, family members, legal professionals, and the legal system. Yet unlike other important older adult clinical capacities such as treatment consent (Kan, 2020), financial capacity (Nowrangi, Sevine, & Kamath, 2019), driving (Wadley et al., 2009), and medication management (Adeola, Fernandez, & Sherer, 2020), which have received considerable empirical research attention over the past two decades (Triebel, Gerstenecker, & Marson, 2018), there has been a striking paucity of assessment instruments and corresponding empirical research concerning TC.

The present paper represents an initial effort to address these TC research gaps with respect to both measure development and empirical research. First, the paper introduces and describes development of the Testamentary Capacity Instrument (TCI), a prototype psychometric measure designed by the senior author to directly and quantitatively measure testator performance across four well-established legal elements (E) of TC generally found across state jurisdictions in the USA and in Anglo-American law generally. Second, this paper investigated the TCI measure by evaluating its performance in a sample of cognitively normal older adults (controls) and older adults with ad. ad is the most prevalent form of late life dementia and is most often the neurocognitive/neuropsychiatric disorder at the clinical center of forensic cases of TC in the probate court. Accordingly, empirical study of TC in persons with ad, using a direct assessment measure like the TCI, has high relevance to both clinical and legal professionals.

Development of the TCI—validity and reliability

The TCI was designed to address the need for a standardized assessment instrument to support and guide the contemporaneous assessment of TC by forensic clinicians and to assist attorneys, courts, and juries in making ultimate legal judgments. As noted above, a critical attribute of the TCI is its breadth and scope—it is constructed and operationalized to assess quantitatively each of four well-accepted conceptual elements or dimensions of the TC legal construct under Anglo-American law: (E1-Purpose of a Will); (E2-Knowledge of Assets); (E3-Knowledge of Heirs); (E4-Plan of Distribution) (Spar & Garb, 1992; Walsh et al., 2019). As outlined in the Methods section, for each TCI element, individual test questions and associated scoring were developed to quantitatively assess different aspects of that TC element/construct and to maximize content validity. Although the TCI elements were designed as constructs for clinical assessment, each maintained a close correspondence to the legal constructs ultimately at issue. The goal was to create an instrument that would provide clinicians with quantitative information relevant to each of the four TC legal constructs. At the same time, the four TCI elements are clinical constructs that are not coextensive with the four legal TC constructs. In this way, the TCI was designed to test functions considered by both healthcare and legal professionals to be integral to executing a valid will. As such, the TCI instrument and its elements were designed to have good face and content validity.

As discussed further below, the TCI and its elements also showed initial evidence of construct validity. On empirical testing, controls performed significantly better than ad participants across each of the four TCI elements (see Table 2). By discriminating ad participant performance from that of cognitively intact older adults, the TCI and its four elements demonstrate initial construct validity. As no other standardized TC measures exist that assess the four elements of TC, the criterion validity of the TCI could not be determined in the present study.

With regard to scale reliability, the TCI demonstrated acceptable levels of internal consistency for two of the four legal elements (Element 1—Understanding a Will and Element 4— Plan). Internal consistency estimates could not be obtained for Element 2 (Assets) and Element 3 (Heirs) as scores for the individual items within these scales inherently varied widely across participants. Good inter-rater scoring reliabilities were obtained across the four legal elements of the TCI.

Investigating testamentary capacity in patients with ad dementia using the TCI

As noted, we used the TCI to investigate TC in patients with ad dementia as compared to cognitively healthy older adults. As expected, ad patients performed below controls on all four of the TCI elements. The patterns of impairment are discussed below.

TCI element 1—Knowledge and purpose of a will

Relative to controls, ad patients had an impaired understanding of the nature and purpose of a will (see Table 2). For example, the ad group had greater difficulty than controls on an item asking participants to state what a will is (t = 3.8, p = 0.001). Similarly, on a set of true/false questions concerning basic attributes of a will, ad group performance fell below that of controls (t = 4.5, p < 0.001). The ad group also demonstrated difficulty understanding the steps needed to make a will (t = 3.4, p = 0.004), and in verbally describing what happens to property named in your will after your death (X² = 3.6, p = 0.06). These findings are consistent with neuropsychological research indicating that persons with ad experience declines in semantic and conceptual knowledge, and in verbal fluency, that impair their ability verbally to describe or define concepts or terms such as “will” (Murphy, Rich, & Troyer, 2006; Weingartner, Kawas, Rawlings, & Shapiro, 1993), correctly identify attributes of a will, or understand basic steps in making a will.

At the same time, the ad group retained some basic knowledge of a will relevant to Element 1. For example, the ad group performed equivalently with controls on simpler multiple choice items such as knowing that talking with a physician was not a necessary step in making a will (X² = 0.011, p = 0.99), and knowing what happens to property listed in a will after a person’s death (X² = 0.005, p = 0.95).

We found that including easier multiple or forced-choice test items (in contrast to more difficult free recall items) helped to limit the impact of dementia-related impairments in expressive language and conceptual knowledge on Element 1 performance in the ad group. As further examples, both ad participants and controls correctly indicated that a will is a legal document (100% for both groups), that a will is a plan for distributing property or belongings to others (90% of ad versus 100% of controls), that a will takes effect only after a person’s death (90% of ad versus 100% of controls), that it must be in writing to be legally binding (90% of ad versus 86% of controls), and that it cannot be created or changed over the telephone (95% versus 100% of controls).

In contrast, group differences emerged on more complex forced-choice items. For example, 70% of ad participants versus only 14% of controls indicated that a will names a person to make medical decisions for you. Similarly, 65% of ad participants versus 14% of controls endorsed a will as similar to a death certificate. About half of the ad group versus 14% of controls indicated that in some cases a will distributes property before a testator’s death. Thus, the level of conceptual complexity of the forced-choice item differentiated group performance on Element 1.

In summary, we found that in comparison to controls, participants with ad dementia had difficulty verbally describing the features and purposes of a will, and in understanding and expressing the more complex conceptual aspects of a will.

TCI element 2—knowledge of nature and extent of assets

Relative to controls, ad participants had an impaired knowledge of the “nature and extent of their assets” eligible for inclusion in a will (Element 2) (see Table 2 and Methods Section).

Control respondents demonstrated a greater knowledge of general asset ownership and associated asset values than did their counterparts with ad (Table 2, Element 2, Questions 1–20). This finding likely reflects that controls had better historical, as well as short-term recall of overall property ownership and property values than did persons with ad. Similarly, compared to ad participants, controls had better historical and short-term recall of types of property that could be included in their will. Controls were also better able than ad participants to recall and identify three specific assets they intended to include in their will. Thus, it appears that performance on TCI Element 2 [Assets] drew heavily on participants’ preserved historical memory for assets that have been owned over time, as well as their short-term recall for more recently acquired assets. Both short-term memory and historical memory loss are cognitive deficits characteristic of persons with mild-to-moderate ad dementia.

A review of asset ownership “errors” across groups (participant ownership responses discrepant with that of their respondents) showed that the ad group had a higher number of errors than the control group regarding property ownership status (42 errors versus 7 errors in controls). The greatest proportion of errors within the ad group involved more abstract and complex “monetary” assets, such as savings accounts, stocks, and bonds. This finding suggests that in ad dementia, diminished historical and short-term recall of ownership of abstract monetary assets may occur sooner than diminished recall of ownership of more concrete assets such as a home or car.

A small group difference emerged in connection to the accuracy of monetary values assigned by participants to specific asset ownership. For established ownership items (i.e., ownership endorsed by both a participant and his/her respondent), disagreement in estimating the value of those assets was noted on 109 occasions (29.8%) of control participant/respondent dyads, versus on 126 occasions (42.2%) for ad participant/caregiver dyads (i.e., 8.5% higher in the ad group) (X² = 11.2, p < 0.001). These observations suggest that, at least in the mild-to-moderate stage of ad dementia, value range estimates of owned assets between controls and ad participants may be relatively equivalent across groups, with ad participants prone to a slightly higher error rate in value estimation.

In summary, the study results indicate that cognitive decline in ad was associated with impairments in knowledge of personal assets eligible for inclusion in a will. Older adults with ad had diminished historical and short-term recall of the full range of owned assets eligible for inclusion in a will and of specific assets to be targeted for inclusion in a will. In particular, ad participant recall appeared to diminish sooner for abstract monetary and financial assets—that will in many cases comprise the primary component of their estate—than for more concrete forms of asset ownership.

TCI element 3—knowledge of possible heirs

Although a testator has freedom to include in his or her will any person, group, or organization, probate courts typically recognize the rights of “natural heirs”—persons with a close blood relationship with the testator, such as children and grandchildren—in probating wills and in will contests. Spouses are not natural heirs but are entitled to a distribution by will through marital or community property laws. Therefore, for purposes of Element 3 of the TCI, potential heirs pertain to immediate family members—a spouse, and children and other close blood relatives such as siblings and grandchildren, whom the law recognizes as being next of kin.

Overall, we found that, relative to controls, ad participants had an impaired knowledge of their potential heirs, specifically their spouse, and their children and other close relatives (see Table 2). Review of ad participant test performance for Element 3 reflected several types of errors that contributed to group differences in knowledge of potential heirs. ad participants made more errors than controls in recalling the ages of their spouse and their children, the maiden name of their spouse and/or surnames taken by daughters in marriage, and recent changes in place of residence for adult children or other extended family (siblings and grandchildren). However, all ad participants (n = 20) correctly recalled the first name of their spouse and children, and whether these family members were living or deceased. Only one moderately impaired ad participant (MMSE = 13) was unable to provide the name of her sister and also a niece who also served as her primary caregiver.

As noted, the final test item of Element 3 required participants to select up to three individuals, groups and/or organizations they intended to include in their will. Controls evidenced no difficulty on this task, whereas 25% of ad participants (5 out of 20) had some difficulty selecting heirs and/or describing their relationship to the chosen heirs. These difficulties were most apparent among participants with moderate ad dementia.

While the findings indicated that persons with ad dementia exhibit impaired knowledge of their potential heirs, the findings also reflected some variability in the knowledge impaired. For example, knowledge of longstanding and static autobiographical information (e.g., first name of spouse and of adult children) appeared to remain intact even in persons with moderate ad. In contrast, participants with ad demonstrated diminished recall for fluid and/or peripheral information concerning potential heirs, such as ages, surnames resulting from marriage or divorce, and places of residence. While such diminished recall of fluid or changing interpersonal information may not always be critical to a testator’s knowledge of his/her heirs, it may also generalize to other circumstances more relevant to prospective testamentary capacity, such as a testator’s recent estrangement from or reconciliation with potential heirs.

TCI element 4—Plan of disposition

We found that ad participants also performed below controls with respect to developing a plan for disposition of assets under a will (Element 4) (see Table 2 and Methods Section).

A review of Element 4 item performance showed that 45% of the ad participants required prompting or guidance in describing their plan for disposition of assets under a will, as opposed to only 9% of controls. Regarding the assignment of specific identified assets to selected heirs, ad participants (particularly those with moderate ad) had more difficulty with this task than controls. The ad group also had greater difficulty providing an explanation for selection of heirs that took into account the specific asset and heir versus other potential heirs.

In summary, persons with ad dementia were impaired in their ability to develop a plan for disposition of assets to heirs under a will. Persons with ad had difficulty in selecting heirs for specific assets and in providing reasons for their choice of heir in relation to an asset. Deficits in short-term memory, long-term episodic memory, and executive functioning characteristic of mild-to-moderate ad dementia likely contribute to these performance difficulties.

Implications and limitations of the study

The present study has described a new interview-based psychometric measure (the TCI) for prospectively evaluating TC, and has investigated the TCI’s validity in discriminating the performance of cognitively healthy older controls and persons with ad dementia on four core conceptual components of TC.

The TCI was developed by the senior author to address the need for a standardized assessment instrument to support and guide the contemporaneous assessment of TC by forensic clinicians. We believe that the TCI demonstrated face and content validity by assessing four core components of TC found in Anglo-American jurisprudence, as well as initial construct validity by discriminating the performance of healthy older controls and ad dementia participants on these four components. The TCI also showed good inter-rater reliabilities and internal reliability for Element 1 (Purpose of a Will) and Element 4 (Plan for a Will). Due to the nature of their items and scaling, internal reliabilities could not be calculated for TCI Element 2 (Knowledge of Assets) and Element 3 (Knowledge of Heirs).

The initial validity study of the TCI provided strong evidence for impaired TC in persons with mild-to-moderate ad dementia. On the TCI, cognitively healthy older adults performed significantly better than their ad counterparts on all four core TCI components, and on many if not most of each component’s underlying test items. Thus, the new measure performed well discriminating groups in a clinical sample where one would anticipate strong TC differences. This finding augurs well for the development of future forensic psychometric measures of TC.

It should be noted that the TCI is intended to serve as a tool to assist clinicians assessing TC in forensic settings. While there may ultimately be TCI cut scores developed for evaluating impairment on the four constructs, the authors do not envisage TCI cut scores for intact TC versus lack of TC in particular cases, as TC is a legal and not a clinical judgment. While the TCI can provide useful information on core TC legal constructs, ultimately a clinician’s opinions regarding a testator’s TC in a forensic matter must reflect his or her own clinical judgments, based on the evidence as a whole and not just the TCI scores or other relevant test scores. In addition, a clinician’s judgments using the TCI are solely clinical judgments — intended to assist attorneys, juries, and courts in arriving at dispositive legal judgments of TC in legal cases.

The study has other limitations. First, this was an initial study with relatively small samples. However, significant group differences were found, and the effect sizes were large and robust. Studies that include larger and more heterogeneous samples will be helpful, particularly for the development of meaningful psychometric scores indicative of impairment for each legal element.

Second, the authors recognize that the TCI’s conceptualization of the four core components of TC likely do not represent the sole conceptual sources or frames of reference with regard to TC under Anglo-American law. As such, other investigators may not fully agree with either the items selected for inclusion in the TCI or the way in which some items seek to operationalize legal elements of TC. The current study’s findings reflect core TC constructs as operationalized by the TCI.

Third, future research should seek to expand normative data on the TCI and to develop cut scores for identifying impairment for each legal element. As previously mentioned, cut scores are not intended to serve as a criterion for capacity determinations but simply to provide clinicians with additional information by which to formulate their clinical judgments. Future research should also seek to confirm newly established neurocognitive predictors of TC with additional well-characterized clinical samples. Finally, future research should seek to establish the criterion validity of the TCI using external TC judgments of expert forensic clinicians knowledgeable about testamentary capacity in older adults.

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 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.