Evaluating voting competence in persons with Alzheimer disease.

Voting by persons with dementia raises questions about their decision-making capacity. Methods specifically addressing voting capacity of demented people have been proposed in the US, but never tested elsewhere. We translated and adapted the US Competence Assessment Tool for Voting (CAT-V) to the Italian context, using it before 2006 elections for Prime Minister. Consisting of a brief questionnaire, this tool evaluates the following decision-making abilities: understanding nature and effect of voting, expressing a choice, and reasoning about voting choices. Subjects' performance was examined in relation to dementia severity. Of 38 subjects with Alzheimer's disease (AD) enrolled in the study, only three scored the maximum on all CAT-V items. MMSE and CAT-V scores correlated only moderately (r = 0.59; P < 0.0001) with one another, reflecting the variability of subjects' performance at any disease stage. Most participants (90%), although performing poorly on understanding and reasoning items, scored the maximum on the choice measure. Our results imply that voting capacity in AD is only roughly predicted by MMSE scores and may more accurately be measured by a structured questionnaire, such as the CAT-V. Among the decision-making abilities evaluated by the CAT-V, expressing a choice was by far the least affected by the dementing process.

1. Introduction

Mental disorders, including dementia, can impair competence, but a diagnosis of dementia does not imply a complete loss of competence [1]. There is a wide consensus on considering competence as the capacity of a person to make a specific decision [2]. Voting is a decision of particular interest since a consensus does not exist on which abilities the patient with dementia should retain to express a reliable choice [3]. Voting is among the fundamental rights of citizens in democratic countries. Thus, identifying patients who, despite the presence of dementia, maintain the capacity to vote and increasing their chance to take part in a ballot (e.g., allowing their caregivers to have a role in facilitating this) would be of crucial importance. Participation in the electoral process by citizens with dementia has become especially important in recent years, both for the growing number of individuals suffering from Alzheimer disease (AD) or other progressive cognitive disorders, and in light of the fact that in at least two cases (2000 US presidential elections and 2006 Italian elections for the Prime Minister designation), a small number of votes had a decisive effect on the results. It is especially in long-term facilities that inappropriate assumptions about the absence of voting capacity may deprive still capable and willing residents of the right to vote [4, 5].

Recently, a novel test to assess the capacity to vote has been proposed: the Competence Assessment Tool for Voting (CAT-V) [6], which evaluates an individual's performance on four decision-making abilities: understanding the nature and effect of voting, appreciating the reality of voting situation, making a choice, and reasoning about voting choices. In this paper, we report the results of a study that applied a modified version of the CAT-V to individuals with mild-moderate AD who were temporarily residents in a long-term care facility before 2006 Italian elections for designating the Prime minister. Our primary hypothesis was that although voting capacity would be inversely associated with dementia severity, the single decision-making abilities evaluated by the CAT-V would be affected unequally by the dementing process.

2. Methods

2.1. Subjects

The subjects included in the present study (n = 38) represent all the patients with mild-moderate dementia (Mini-Mental State Examination [7] (MMSE) ≥ 11) and a clinical diagnosis of probable AD (according to the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer Disease and Related Disorders Association (ADRDA) criteria) [8] who were admitted into the Alzheimer Centre of the Ospedale Gazzaniga (Bergamo, Italy) from sixty to thirty days before 2006 Italian general elections. Although, in some respects, our centre has several characteristics of a long-term care facility, no patient is a permanent resident. The primary requirement for a patient's admission into our centre is the presence of behavioural abnormalities or psychopathologic symptoms in the context of a dementing syndrome but, once these features are significantly relieved, the patient is discharged.

2.2. Assessment Tool and Scoring

The instrument we used to evaluate the capacity to vote was a modified version of the CAT-V, an instrument that measures a person's ability to understand the nature and effect of voting, make a choice, appreciate, and reason through a voting decision. These criteria were operationalized into five questions preceded by an introduction reminding each person that soon he/she would have the opportunity to take part in a ballot for the election of the Prime Minister. Thus, as opposed to the original US version of the CAT-V, in which subjects are asked to imagine that two candidates are running for Governor and that the day of the interview is the Election Day, the scenario we proposed was real rather than hypothetical. Furthermore, in order to shorten the time of interview, unlike the original, our version of the CAT-V did not include a question evaluating subjects' appreciation of the significance of voting. For each CAT-V item, the scores assigned to each person ranged from 2 (correct response reflecting adequate performance) to 0 (inadequate performance). The instrument and criteria for scoring are shown in the appendix.

Every participant was enrolled after an initial contact with his/her principal caregiver. Once informed about the characteristics of the study and made sure that its results would be used exclusively for research purposes, each participant (or his/her caregiver) provided a written informed consent. The study was approved by our local institutional review board.

2.3. Data Analyses

All of the 38 participants were interviewed and rated by one investigator (M. Sala), who was blinded to their MMSE score. Thirty of them were also interviewed and rated by another investigator (E. Chitò). Weighted kappa and Kendall tau-B were used to determine the interrater reliability. Twenty-nine subjects were again interviewed by M. Sala two weeks later, to evaluate the test-retest reliability. The scores included in the main data analysis are those assigned to all participants at baseline by M. Sala The Spearman correlation coefficient was used to examine the association of the capacity to vote (as expressed by the score on each of the CAT-V items) with severity of both cognitive impairment (as expressed by the MMSE score) and behavioural and psychopathological symptoms (as expressed by the Neuropsychiatric Inventory (NPI) [9] score). Each participant was administered the CAT-V, the MMSE, and the NPI during the same session.

3. Results

3.1. Subjects' Characteristics

All the subjects who were asked to participate in the project (n = 38) did complete the interview. Their demographic characteristics are reported in Table 1. There was a clear female preponderance. Severity of cognitive impairment was, on average, relatively mild. Behavioural and psychopathological symptoms were moderately severe at baseline but were significantly relieved prior to discharge.

Table 1

Demographics of the subjects (n = 38).

	Mean	Standard deviation	Range
Age (years)	81.1	5.1	68–93
Gender (M/F)	9/29	—	—
Education (years)	5.6	2.3	3–13
Disease duration (years)	3.3	1.6	1–8
MMSE score	18.2	4.2	11–27
NPI score at baseline	43.0	17.5	11–72
NPI score prior to discharge	21.2	11.7	0–50

MMSE, Mini-Mental State Examination, NPI, Neuropsychiatric Inventory.

3.2. Performance on CAT-V

Subjects' performance on CAT-V is shown in Table 2. Over a half of the subjects appeared to fully understand the nature of the vote, but only approximately a third was entirely able to understand its effect. However, the great majority of participants (~90%) was deemed to be completely able to make a choice. Conversely, subjects' ability to reason about voting by comparing the choices at disposal and, above all, by evaluating the possible consequences of the preference for a candidate on their life was considerably more impaired. In fact, only about 16% of the participants had a completely adequate performance on the latter measure.

Table 2

Subjects' score distribution on each item of the Competence Assessment Tool for Voting (CAT-V).

CAT-V item and score	N	%
Understanding the nature of voting
0	7	18,4
1	10	26,3
2	21	55,3
Understanding the effect of voting
0	18	47,4
1	6	15,8
2	14	36,8
Choice
0	4	10,5
2	34	89,5
Comparative reasoning
0	13	34,2
1	9	23,7
2	16	42,1
Generating consequences
0	21	55,3
1	11	28,9
2	6	15,8

As shown in Table 3, which relates subjects' combined performance on understanding and choice to their performance on reasoning, only three of the 38 participants (8%) scored the maximum on all items. As emerges from Table 4, there were better test-retest and interrater reliabilities for scores on understanding and choice than for scores on reasoning.

Table 3

Cross-tabulation relating scores on understanding and making a choice to scores on reasoning.

	Score on questions assessing understanding and choice
Score on questions assessing reasoning	Number of subjects scoring 0–5	Number of subjects scoring 6	Total
Number of participants scoring 0–3	25	9	34
Number of participants scoring 4	1	3	4
Total	26	12	38

Table 4

Competence Assessment Tool for Voting (CAT-V) scoring criteria, interrater, and test-retest reliabilities.

		N	Cohen Kappa	Kendall tau-b	P value
Interrater reliability*	Understanding and choosing	30	0.65	0.7	0.0001
	Reasoning	30	0.41	0.65	0.0001
Test-retest reliability#	Understanding and choosing	29	0.42	0.65	0.0001
	Reasoning	29	0.22	0.57	0.0001

*Determined in the first 30 of the 38 patients enrolled in the study.

#Retest was administered only to patients not yet discharged from our center after two weeks (29/38).

3.3. Relation of CAT-V Performance to Severity of Dementia

There was no relation of CAT-V scores to severity of behavioural and psychopathological symptoms (r = −0.14, P = 0.41). Conversely, as expected, lower CAT-V scores were associated with lower MMSE scores (Figures 1, 2, and 3). However, a great variability in subjects' performance was noted at any stage of disease. On questions evaluating understanding and choice (Figure 2), for example, only 58% of subjects with mild AD (MMSE ≥ 20) obtained the maximum score but, remarkably, over one-third of those who scored the maximum was beyond mild-stage disease (MMSE < 20).

Figure 1

Relation of scores on the Mini-Mental Exam to combined scores on all questions of the Competence Assessment Tool for Voting (r = 0.59, P < 0.0001).

Figure 2

Relation of scores on the Mini-Mental State Exam to scores on questions assessing understanding and choice (r = 0.61, P < 0.0001).

Figure 3

Relation of scores on the Mini-Mental State Exam to scores on questions assessing reasoning (r = 0.41, P = 0.01).

Lower CAT-V scores were also associated with fewer years of education but, as opposed to disease severity, poor education had an impact exclusively on measures of understanding and choice (r = 0.32, P = 0.049).

No relationship was found between education and MMSE scores (r = 0.05, P = 0.76).

4. Discussion

Among persons with mild to moderate AD, global measures of cognitive functioning, such as the MMSE, cannot adequately substitute for an assessment of voting capacity. The present study has examined the capacity to vote in a sample of 38 mild-moderate AD patients using a modified version of the CAT-V. Originally designed in the US, this instrument consists of a brief questionnaire, which has been translated and adapted to the Italian context and by which we have explored the following functional abilities: understanding the nature and effect of voting, expressing a choice, reasoning about the choices at disposal, and reasoning about the consequences of voting. We have also calculated the reliability of the test and its relation to dementia severity.

On the basis of our results, a full capacity to vote (as expressed by integrity of all above mentioned decision-making abilities) appears to be retained by a small minority of AD subjects (3/38), and exclusively at mild-stage disease (MMSE ≥ 20).

However, when we applied less restrictive criteria for determining voting competence, as those identified in the Doe standard (a legal standard based on a 2001 federal district court decision in Maine (US), which solely requires an intact ability to understand and make a choice) [10], subjects' voting capacity was not completely predicted by MMSE scores. On understanding and choice measures, in fact, only 58% of our less deteriorated subjects (MMSE ≥ 20) obtained the maximum score but, remarkably, over one-third of those who scored the maximum were beyond mild-stage disease (MMSE 13–19). As a result, the relation of these measures to severity of cognitive impairment was only moderate (r = 0.61).

Of note, over two-thirds of our AD patients, although still able to express a choice, did not appear to be entirely able to understand the nature and, especially, the effect of the vote, thereby failing to meet the Doe standard. A similar pattern (choice considerably less impaired than understanding) has previously been reported by US investigators [5]. In their study, however, the percentage of AD patients who failed to fulfil the Doe standard was lower than that seen in our study (55% versus 68%), and there was a much more substantial link between declining voting capacity and increasing dementia severity (r = 0.87 versus 0.61).

There are several possible reasons for the discrepancy between our results and those previously reported by the US investigators. For example, compared to these authors, we examined a sample characterized by more advanced age (81.1 versus 77.7 years), greater female preponderance (76% versus 52%), less severe impairment (mean MMSE 18.2 versus 16.4), and considerably lower education (5.6 versus 14 years). Furthermore, since we excluded patients with severe AD from analyses, the range of cognitive impairment was more compressed in our sample (MMSE 11–27) than in the US sample (MMSE 2–28). The exclusion of patients with severe AD may also explain the only low-moderate test-retest reliability values of our study (understanding and choice, k = 0.42; reasoning, k = 0.22), as well as the less satisfying agreement between our raters than between the raters of the US study (understanding and choice, k = 0.65 versus 0.91; reasoning, k = 0.41 versus 0.74). Subjects with severe AD have in fact a greater likelihood than those with mild-moderate disease to invariably provide completely inadequate performances, so as to be assigned the minimum score unambiguously and consistently over time.

Not requiring a particular expertise, the CAT-V is easily administrable, since no more than five minutes are needed for its administration. This time might further be shortened if one decides to skip the reasoning questions because, at least in our experience, these questions were too demanding even for mildly deteriorated patients and generated performances characterized by insufficient test-retest and inter-rater reliabilities. Interpreting subjects' performance remains, however, problematic even when analyses are restricted to the questions inherent in the Doe standard (understanding and choice). Clearly, performances at the extremes of the spectrum are not controversial, so that a performance generating the minimum score unequivocally indicates absence of voting capacity and, by contrast, a performance generating the maximum score indicates a full compliance with the Doe standard. However, intermediate scores need a judgement to be made, the basis of which is not obvious and deserves further comments. For example, in both the US and our study, almost all of the participants with intermediate scores appeared to be invariably capable to make a choice, while what varied was their ability to understand the nature and effect of voting. Consequently, if we had applied more liberal criteria than those identified in the Doe standard for determining voting competence and, for example, we had deemed a patient to retain the capacity to vote if he/she was able to express a choice regardless of understanding, the great majority of AD subjects (90% in our study and 88% in the US study) would have been categorized as such.

The limitations of this study relate to lack of data from nondemented persons, whose availability would have been extremely helpful in interpreting intermediate scores by the identification of appropriate cutoffs and of more detailed neuropsychological information than that provided by the MMSE. Other issues are the relatively small sample size, the relatively narrow range of cognitive impairment, and the presence of significant behavioural and psychopathological symptoms (although their severity did not appear to influence the performance of our patients on the CAT-V). Since this study was restricted to persons with mild-moderate disease, and most of them had significant behavioural disturbances, its results may be not entirely representative of all patients with AD. Despite these flaws, the primary hypothesis of the present study has been tested and verified. There was an inverse relationship between voting capacity and dementia severity but, somewhat unexpectedly, the strength of this association was not substantial. However, as predicted, the single decision-making abilities evaluated by the CAT-V were unequally impaired by the dementing process (reasoning > understanding > choice). On the basis of these data, the use of a structured interview, such as the CAT-V, may offer advantages over unstructured or clinical assessments, especially in light of the fact that global measures of cognitive functioning, such as the MMSE, do not appear to be strong predictors of the capacity to vote. Further studies are needed to refine the clinicians' approaches to identifying demented people who are still capable to vote from those who are no longer capable. Nevertheless, a tool like the CAT-V can adequately assist in this distinction (Table 5).

Table 5

What already known on this topic is
People with dementia are underrepresented at the polls. Many of them are denied the opportunity to vote even when retaining the mental capacity to do so.
Methods that address voting capacity of demented people, such as the Competence Assessment Tool for Voting (CAT-V), have been proposed and tested in the US, but never elsewhere.
Using the CAT-V in patients with Alzheimer Disease (AD), US investigators have shown a robust association between declining voting capacity and increasing dementia severity.
What this study adds
Using a modified version of the CAT-V, we found only a moderate association between declining voting capacity and increasing dementia severity in AD.
The capacity to express a choice is largely preserved even in moderate-stage AD.
Many patients with AD, although no longer capable of understanding the nature and importance of voting, are still able to express a choice. Their right to vote should therefore be respected.