RESEARCH ARTICLE

Cognitive predictors of medical decision-making capacity in mild cognitive impairment and Alzheimer’s disease Sara Stormoen1,2, Ove Almkvist3,4, Maria Eriksdotter5,6, Erik Sundström7,8 and Ing-Mari Tallberg1,2 1

Department of Clinical Sciences, Intervention and Technology (CLINTEC), Karolinska Institutet, Stockholm, Sweden Department of Speech and Language Pathology, Karolinska University Hospital, Stockholm, Sweden 3 Department of Psychology, Stockholm University, Stockholm, Sweden 4 Division of Alzheimer Neurobiology Center - Department of Neurobiology, Care Sciences and Society, Stockholm, Sweden 5 Department of Geriatric Medicine, Karolinska University Hospital, Stockholm, Sweden 6 Division of Clinical Geriatrics - Department of Neurobiology, Care Sciences and Society, Stockholm, Sweden 7 Division of Neurodegeneration - Department of Neurobiology, Care Sciences and Society, Stockholm, Sweden 8 Stiftelsen Stockholms sjukhem, Stockholm, Sweden Correspondence to: S. Stormoen, E-mail: [email protected] 2

Objective: Impaired capacity to make decisions in everyday life and situations of medical treatment

is an inevitable consequence of the cognitive decline in Alzheimer’s disease (AD). The objective of this study was to identify the most powerful cognitive component(s) that best predicted medical decision-making capacity (MDMC) in patients with AD and mild cognitive impairment. Method: Three groups of subjects participated in the study: patients with AD (n = 20), mild cognitive impairment (n = 21), and healthy control subjects (n = 33). MDMC was assessed by the linguistic instrument for medical decision-making (LIMD) and related to demographics and 27 cognitive test measures. Results: The cognitive tests were found to aggregate into four components using a principle component analysis. The four components, which correspond to verbal knowledge, episodic memory, cognitive speed, and working memory, accounted for 73% of the variance in LIMD according to a stepwise regression analysis. Verbal knowledge was the most powerful predictor of LIMD (beta = 0.66) followed by episodic memory (beta = 0.43), cognitive speed (beta = 0.32), and working memory (beta = 0.23). The best single test as shown by the highest correlation with LIMD was Reading speed (R = 0.77). Conclusion: Multiple factors are involved in MDMC in subjects with cognitive impairment. The component of verbal knowledge was the best predictor of MDMC and Reading speed was the most important single cognitive test measurement, which assessed both rapid Reading and understanding of text. Copyright # 2014 John Wiley & Sons, Ltd. Key words: medical decision-making capacity; Alzheimer’s disease; cognitive predictors; verbal knowledge; text reading History: Received 10 October 2013; Accepted 7 March 2014; Published online 16 April 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/gps.4114

Introduction A debated issue in dementia research is the ethical dilemma of obtaining informed consent from patients with impaired cognitive function (Peterson and Wallin, 2003; Slaughter et al., 2007). A challenge is to secure the patients’ autonomy as well as the medical benefit as far as possible. Studies infer that reduced medical decision-making capacity (MDMC) is a common consequence among patients with cognitive Copyright # 2014 John Wiley & Sons, Ltd.

decline (Marson et al., 1995a; Marson et al., 1996; Gurrera et al., 2006; Jefferson et al., 2008; Okonkwo et al., 2008). Research on subjects with impaired cognitive function brings about important issues of how to understand the relation between MDMC and cognitive abilities. Decisional capacity in patients with cognitive impairment has in previous studies been evaluated with different instruments and methods showing good psychometric features (Marson et al., 1995b; Gurrera Int J Geriatr Psychiatry 2014; 29: 1304–1311

Medical decision making in Alzheimer’s disease

et al., 2006; Jefferson et al., 2008). Consent abilities, referred to as legal standards (LS) have been suggested and are well-established in some societies. Five LS are commonly used: LS1, expressing treatment choice; LS2, making a reasonable choice; LS3, appreciating consequences of choice; LS4, providing rational reasons for choice; and LS5, understanding treatment situation and choices (Marson et al., 1995b). Understanding, LS5 is described as a relevant standard for evaluating competency to consent and also very likely to be impaired in dementia (Marson et al., 1996). Often, so-called vignette methods are applied when assessing capacity to consent. In these methods, the subject is presented with a hypothetical trial, usually describing diseases and treatments (routine or clinical trials). After being informed, the subject is asked certain questions about the content (Schmand et al., 1999; Vellinga et al., 2005). We have recently developed a Swedish quantitative research instrument based on the vignette methods denoted as the Linguistic Instrument of Medical Decision-making (LIMD) that have good psychometric properties (test–retest reliability was r = 0.94, and validity defined as the total variance between groups was 80.3%; Tallberg et al., 2013). LIMD consists of three vignettes about hypothetical clinical trials, an interview, and a scoring protocol. Three so-called criteria of decision-making capacity are evaluated in LIMD: (1) comprehension (understanding the content); (2) evaluation (ability of evaluating risk and benefit); and (3) intelligibility (ability to express a decision). The LIMD criteria are similar to previous well-known standards of consent abilities (Marson et al., 1995b). However, the standard LS2 (make a reasonable choice) was excluded, because the interpretation of the concept “reasonable” is based on personal evaluations and therefore difficult to operationalize. Also, in LIMD, LS3 and 4 were merged into one criterion (“evaluation”) for the same reason. Instead, we examined utterances considering risks and benefits occurring in the transcriptions of the interviews. Each of the three LIMD criteria was defined, and each criterion received a score on a continuous scale, and the summed score was used as a measure of LIMD. The LIMD score is based on predefined linguistic features such as independence of the utterances, accuracy, inferences, and coherence for each established criterion (see Tallberg et al., 2013 for more details). Several previous studies have investigated to what extent cognitive performance can predict decisional capacity in patients with cognitive impairment (Marson et al., 1995a; Marson et al., 1996; Gurrera et al., 2006; Jefferson et al., 2008; Okonkwo et al., Copyright # 2014 John Wiley & Sons, Ltd.

1305

2008). Gurrera and collaborators (2006) investigated the impact of cognitive performance in mild to moderate dementia on four LS of capacity of treatment decisions: expression of choice, appreciation, reasoning, and understanding. Their research showed that neuropsychological performance significantly could predict all standards and indicated that verbal retrieval was the strongest predictor of several decisional abilities. Marson and collaborators (1995a) did not find verbal reasoning and memory highly associated to the capacity to provide rational reasons for a treatment choice. Moreover, the authors indicated that basic language functions such as verbal fluency had stronger impact on rational reasoning in MDMC compared with higher language functions such as verbal abstractions and reasoning. However, many MDMC instruments require a time-consuming assessment procedure, and the matter of how to develop an instrument for assessment of MDMC appropriate for use in clinical settings is under discussion. The objective of this study was to identify the most powerful cognitive component(s) that best predicted MDMC in patients with Alzheimer’s disease (AD) and mild cognitive impairment (MCI), using LIMD to assess MDMC and a comprehensive battery of cognitive tests as possible predictors, and taking demographic factors into account. Our hypothesis was that there are one or several specific cognitive components responsible for MDMC. Methods Subjects

Three groups participated in the study: patients with AD (n = 20; mini mental state examination [MMSE]: 24.1 ± 3.6), MCI (n = 21; MMSE: 26.6 ± 2.4) and a group of healthy control subjects (HC, n = 33; MMSE: 29.1 ± 1.0). All subjects had Swedish as native language (inclusion criterion). Groups differed significantly (F = 31.63, df = 2/72, p < 0.001) in dementia severity by the MMSE (Folstein et al., 1975) because of disease. Groups were matched in demographic characteristics as demonstrated by non-significant group effects (all p > 0.10) for gender distribution, age, and years of education, as presented in Table 1. The two groups of patients were examined and diagnosed at the Memory Clinic, Karolinska University Hospital, Huddinge, as part of the clinical routine; see succeeding text. Patients with an MMSE 9 were excluded. Int J Geriatr Psychiatry 2014; 29: 1304–1311

1306

S. Stormoen et al.

Table 1 Demographic characteristics (gender, age, and years of education) for the three groups of subjects: Alzheimer’s disease and mild cognitive impairment patients and healthy control subjects Demographics N (women/men) Age (M ± SD), years Education (M ± SD), years

AD

MCI

HC

p

20 (9/11) 72.5 ± 7.6 10.3 ± 3.6

21 (14/7) 69.1 ± 8.8 12.3 ± 2.9

33 (23/10) 69.2 ± 6.5 12.4 ± 3.3

ns ns ns

AD, Alzheimer’s disease; MCI, mild cognitive impairment; HC, healthy control subjects; SD, standard deviation; ns, non-significant.

Patient records were screened by two clinicians (ME and SS), reviewing diagnosis, MMSE scores, and exclusion criteria. If considered eligible for participation, a letter of invitation describing the study was sent to the patient. The group of HC was recruited among spouses to the participating patients and by advertising for healthy, age-matched volunteers (aged 55–80 years). An in-house questionnaire regarding age, education, health problems, vision, hearing, dyslexia, and cognitive symptoms was used for the HC before inclusion. Only those with no health remarks were included.

Hospital, Huddinge. The duration of the total assessment session was approximately 3–4 h per subject, and pauses were offered when needed. Three clinicians were engaged in the test procedure. One person (SS) performed the total data collection. Scoring of cognitive assessments was done in consultation between two persons (SS and OA), blinded to identity and diagnosis of the subjects. A third person (IMT), blinded to identity and diagnosis of the subjects, scored the LIMD protocols twice after an interval of 3 months and after giving each subject a new code. Measures Medical decision-making capacity. Linguistic instrument for medical decision-making consists of three parts: (1) three vignettes; (2) a standardized interview; and (3) a standardized scoring protocol. As the test– retest reliability of LIMD has been shown to be very good (R = 0.94, p < 0.001; Tallberg et al., 2013), the present study refers to the LIMD score as a mean value between two evaluations made by the same experienced rater (IMT). Cognitive function. All patients and healthy controls

Diagnosis

The diagnosis of AD patients followed the criteria set by the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA; McKhann et al., 1984). The diagnosis of MCI was made according to modified clinical criteria (Petersen et al., 1999; Winblad et al., 2004). All patients were diagnosed by a multidisciplinary team using all information from a standard comprehensive clinical examination including medical history; somatic, neurologic psychiatric status, and cognitive screening; magnetic resonance imaging; cerebrospinal fluid biomarkers (total tau, phospho-tau and Abeta42); routine laboratory examination of blood and urine; and functional status on cognition, language, and daily life (Wahlund et al., 2003). Information from a close informant war reported using the Informant Questionnaire on Cognitive Decline in the Elderly (Jorms, 2004). Symptoms of depression were assessed by the Cornell Depression Index (Alexopoulos et al., 1988). Procedures

The assessments of decision-making capacity and cognitive function were conducted at Karolinska University Copyright # 2014 John Wiley & Sons, Ltd.

were assessed regarding global cognition, aspects of linguistic competence, visuospatial function working, episodic memory, executive function, and attentional function by means of 27 test measures. The procedure regarding instruction and scoring followed international standards (Folstein et al., 1975; Wechsler, 1981; Kaplan et al., 1983; Golden and Freshwater, 2002; Järpsten, 2002; Johansson, 2002; Lezak et al., 2004; Tallberg, 2005; Östberg et al., 2008; Tallberg et al., 2008). The psychometric properties (reliability and validity) of all test measures (MMSE, Boston Naming Test, Word sequence production, Sentence repetition, Inference, Logico-grammatical sentences, Reading speed, Reading aloud, Word fluency, Information, Similarities, Block design, Digit span, Digit symbol, Corsi block tapping, Rey auditory verbal learning and retention, Rey-Osterrieth copying and retention, Trail making A and B, and Stroop word and word color) were in good agreement with standard requirements. Statistics

Descriptive statistics (mean ± standard deviation) were used to present sample characteristics as well as cognitive test results for the three groups of participants. Univariate analysis of variance (ANOVA) was used to analyze possible group differences in age, education, Int J Geriatr Psychiatry 2014; 29: 1304–1311

Medical decision making in Alzheimer’s disease

1307

LIMD, MMSE, and all cognitive test measures. Tukey’s post hoc t-test was used to analyze possible pairwise group differences. The effect size is reported as degree of explained variance in ANOVA’s (eta, η2). A principle component analysis was used to reduce the number of test measures into independent components covering as much as possible of the total test variance. Eigen values >1 were used to decide the number of components, and all components should have high loadings (>0.70) in at least two test measures. Varimax rotation was used to identify a parsimonious structure. A stepwise linear regression analysis was used to examine the relationship (common variance, r 2) between the LIMD score and possible predictors (component scores). The degree of importance for each component regarding LIMD was presented as β weights. The Pearson correlation coefficient was used to describe the association between pairwise variables (test–test and test–component). Statistical significance was indicated using three levels of probability (p < 0.05,

p < 0.01, and p < 0.001). The statistical software used was SPSS, version 20. Ethical approval The Regional Ethics Committee in Stockholm approved the study, 2008/1276-31/2, 2009/1764-32. All participating subjects gave both oral and written consents. Results Medical decision-making capacity and cognitive function

The performances in groups of subjects (AD, MCI, and HC) of medical decision-making capacity (LIMD) and cognitive functions are presented in Table 2.

Table 2 Test results in raw scores or time in seconds of medical decision-making capacity, dementia severity, and cognitive function for the three groups of subjects: Alzheimer’s disease and mild cognitive impairment patients and healthy control subjects Test LIMD MMSE BNT Word sequence, forward (s) Word sequence, backward (s) Inference Sentence repetition Logico-grammatical Word fluency, FAS Word fluency, noun Word fluency, verb Information Similarity Block design Rey-Osterrieth, copying Digit span, forward Digit span, backward Digit symbol Reading speed (s) Read aloud, word (s) Read aloud, non-word (s) Corsi blocks RAVLT; total RAVLT; retention Rey-Osterrieth, retention TMT, A (s) TMT, B (s) Stroop, reading color words Stroop, naming word colors

2

AD

MCI

HC

F

p

η

8.0 ± 7.1 24.1 ± 3.6 40.7 ± 13.7 6.5 ± 2.8 29.3 ± 15.7 12.7 ± 6.8 13.3 ± 7.6 16.8 ± 7.0 29.7 ± 10.7 12.6 ± 6.5 9.3 ± 5.5 19.5 ± 4.7 13.5 ± 5.2 17.8 ± 8.0 31.6 ± 5.0 5.2 ± 1.0 3.3 ± 0.7 28.3 ± 9.4 11.5 ± 6.8 61.2 ± 17.2 68.2 ± 20.5 4.9 ± 0.7 21.5 ± 9.9 1.9 ± 2.2 6.3 ± 6.2 56.9 ± 18.0 173.0 ± 60.7 70.4 ± 14.4 18.5 ± 9.4

17.1 ± 6.2 26.6 ± 2.4 50.0 ± 6.7 5.7 ± 2.4 19.2 ± 12.8 22.9 ± 6.7 21.4 ± 5.6 22.9 ± 5.6 35.0 ± 9.9 17.9 ± 6.2 16.4 ± 6.3 20.2 ± 3.5 19.3 ± 3.8 21.8 ± 9.0 32.5 ± 4.0 5.3 ± 1.1 3.9 ± 0.9 30.8 ± 8.2 17.5 ± 8.0 53.9 ± 14.9 61.8 ± 19.1 4.9 ± 0.9 37.0 ± 10.8 4.9 ± 4.1 10.6 ± 7.1 65.9 ± 29.1 170.3 ± 72.1 74.7 ± 19.8 22.8 ± 7.1

21.9 ± 4.8 29.1 ± 1.0 54.9 ± 3.9 5.4 ± 0.9 10.6 ± 4.8 25.4 ± 4.1 24.6 ± 5.3 25.5 ± 4.8 47.8 ± 11.3 22.0 ± 6.0 19.2 ± 5.7 21.2 ± 3.6 20.6 ± 3.0 28.2 ± 6.9 34.5 ± 1.9 6.0 ± 1.0 4.7 ± 1.1 42.6 ± 10.0 22.9 ± 6.4 44.5 ± 9.5 44.6 ± 9.8 5.4 ± 0.8 47.0 ± 10.0 10.1 ± 3.4 16.0 ± 5.2 49.2 ± 15.5 108.9 ± 39.4 92.3 ± 9.5 35.6 ± 9.6

F(2, 71) = 35.05. F(2, 70) = 31.63 F(2, 71) = 17.96 F(2, 71) = 3.42 F(2, 71) = 17.89 F(2, 71) = 32.34 F(2, 71) = 15.38 F(2, 71) = 15. 68 F(2, 71) = 19.73 F(2, 71) = 14.40 F(2, 71) = 18.59 F(2, 71) = 10.64 F(2, 71) = 21-20 F(2, 70) = 11.30 F(2, 69) = 4.27 F(2, 70) = 5.64 F(2, 70) = 13.62 F(2, 71) = 18.05 F(2, 71) = 17.04 F(2, 66) = 8.93 F(2, 66) = 13.75 F(2, 71) = 3.84 F(2, 70) = 41.07 F(2, 70) = 41.16 F(2, 67) = 16.04 F(2, 69) = 3.96 F(2, 69) = 11.29 F(2, 69) = 17.49 F(2,67) = 25.09

*** *** *** * *** *** *** *** *** *** *** *** *** *** * * *** *** *** *** *** * *** *** *** * *** *** ***

0.50 0.48 0.53 0.09 0.33 0.48 0.30 0.31 0.36 0.28 0.34 0.23 0.37 0.24 0.11 0.14 0.28 0.34 0.32 0.21 0.29 0.10 0.53 0.54 0.32 0.10 0.25 0.34 0.43

AD, Alzheimer’s disease; MCI, mild cognitive impairment; HC, healthy control subjects; LIMD, linguistic instrument for medical decisionmaking; MMSE, mini mental state examination; BNT, Boston Naming Test; FAS, Phonemic F,A,S; RAVLT, Rey Auditory Verbal Learning Test; TMT, Trail Making Test. *p < 0.05; **p < 0.01; ***p < 0.001

Copyright # 2014 John Wiley & Sons, Ltd.

Int J Geriatr Psychiatry 2014; 29: 1304–1311

1308

S. Stormoen et al.

Mean values in test performances (LIMD, MMSE, and cognitive test measures) showed that AD patients performed worse than MCI patients and that MCI patients performed worse than HC subjects. The groups differed significantly in LIMD and cognitive function (p < 0.05). AD patients and HC differed significantly in all tests (p < 0.05). AD and MCI patients differed significantly on LIMD and several cognitive test measures assessing language production and comprehension, reading capacity, and episodic memory (p < 0.05). Further on, MCI patients and HC differed significantly on LIMD and several cognitive tests assessing language production, reading capacity, spatial ability, short-term memory, episodic memory, executive function, attention, and speed (p < 0.05).

Similarities. This component was interpreted as “verbal knowledge”. The second component, interpreted as “episodic memory”, accounted for 15% of the variance, and it had high loadings (>0.70) in three tests: Rey-Osterrieth retention, Rey Auditory Verbal Learning Test (RAVLT) learning and retention. The third component, interpreted as “cognitive speed”, accounted for 15% of the variance, and it had high loadings (>0.70) in three tests: forward Word sequence, Trail making A and B. The fourth component accounted for 14% of the variance, and it had high loadings (>0.70) in three tests: forward and backward Digit span and Stroop color words. This component was interpreted as indicating “working memory”. In Table 3, the test–component loadings are presented (only loadings >0.30 are presented).

Structure of all neuropsychological tests LIMD in relation to cognitive function and test measures

By using principle component analysis on 27 cognitive test measures, four components were obtained taking 68% of the total variance into account. The first component accounted for 23% of the variance, and it had high loadings in many verbal tests and high loadings (>0.70) in three tests: Inference, Information, and

A regression analysis on LIMD score as criterion and the four component scores as possible predictors showed that LIMD was significantly related to all four components (multiple R = 0.86, adjusted R2 = 0.73, F = 40.23, df = 4/55, p < 0.001). LIMD was most

Table 3 Component loadings (>0.30) for cognitive tests after varimax rotation Cognitive test

Verbal knowledge

Similarities Information Inference Word fluency, verb Logico-grammatical BNT Sentence repetition Reading speed (s) Word fluency, Noun Word fluency, FAS Read aloud, non-word (s) Block design Rey-Osterrieth, copying RAVLT, retention RAVLT, total Rey-Osterrieth, retention Digit symbol Stroop, reading word colors Word sequence, backward (s) TMT, A (s) Word sequence, forward (s) TMT, B (s) Read aloud, word (s) Corsi blocks Digit span, forward Digit span, backward Stroop, naming color words

0.83 0.79 0.71 0.70 0.69 0.68 0.66 0.65 0.57 0.56 0.51 0.42 — — 0.34 — — 0.34 0.32 — — — 0.59 — — — 0.31

Episodic memory — — 0.42 — — 0.41 — 0.36 0.53 0.39 — 0.31 — 0.85 0.78 0.78 0.53 0.51 0.39 — — 0.37 — — — — —

Cognitive speed — — — — — — — 0.44 — 0.48 0.32 — — — — 0.51 0.30 — 0.73 0.72 0.65 0.61 0.40 — — 0.47

Working memory — — — 0.39 0.32 — 0.46 — — 0.46 0.36 — — — — — — 0.44 0.32 — — — 0.34 — 0.85 0.72 0.68

BNT, Boston Naming Test; FAS, Phonemic F,A,S; RAVLT, Rey Auditory Verbal Learning Test; TMT, Trail Making Test. Loadings >0.70 are in bold.

Copyright # 2014 John Wiley & Sons, Ltd.

Int J Geriatr Psychiatry 2014; 29: 1304–1311

Medical decision making in Alzheimer’s disease

strongly associated with the “verbal knowledge” component (beta = 0.66, p < 0.001), secondly with the “episodic memory” component (beta = 0.43, p < 0.001), thirdly with the “cognitive speed” component (beta = 0.32, p < 0.001), and fourthly with the “working memory” component (beta = 0.23, p < 0.01). This pattern of results remained also after taking demographic characteristics into account. Turning to single tests, it was found that LIMD was most strongly related to the Reading speed test (included in the verbal component) as demonstrated by a significant correlation (R = 0.77, p < 0.001, Figure 1). If combinations of tests are preferred to predict LIMD, the second strongest correlation was found for the total score on the Rey Auditory Verbal Learning Test (RAVLT) as shown by a stepwise regression analysis, when the Reading speed test was first entered into the analysis (beta = 0.38, p < 0.001), and RAVLT (total learning score) was second (beta = 0.26, p < 0.001). Inclusion of still more tests led to significant contribution by the Inference (beta = 0.23, p < 0.001) and Sentence repetition (beta = 0.22, p < 0.001) tests. However, the additive power in predicting LIMD was only marginally increased by adding two or more tests. It was interesting to note that the MMSE test was not included into the regression model as a significant predictor of LIMD, neither as a single test nor as a test in combination with other tests. Discussion The relationship between MDMC and cognition has been the focus in many studies. In the assessment of

Figure 1 Reading speed score in relation to linguistic instrument for medical decision-making (LIMD) score for three groups of individuals: Alzheimer’s disease (AD) and mild cognitive impairment (MCI) patients and healthy control subjects (HC).

Copyright # 2014 John Wiley & Sons, Ltd.

1309

MDMC by LIMD, the subject is expected to read, listen, and decode information about different hypothetical clinical trials, and the performance on this task can be assumed to be related to cognitive functioning. Our result showed that four cognitive components (verbal knowledge, episodic memory, cognitive speed, and working memory) are important for MDMC. The components are all associated in various specific and basic cognitive tests, all of which are impaired in AD and MCI patients. When single tests were investigated, it was shown that the most powerful test for predicting MDMC was the Reading speed test (Järpsten, 2002). The most powerful combination of two tests was Reading speed combined with RAVLT (Lezak et al., 2004). It is possible to increase the predicted power further by adding more tests, for instance, Inference and Sentence repetition (Laakso et al., 2000). These results emphasize the strong influence of verbal functions on MDMC. The most powerful cognitive component for prediction of MDMC was verbal knowledge. This is in agreement with task demands when a patient has to consider participation in a clinical trial. Information about the trial is presented orally by the health-care personal and also in written format. Furthermore, verbal interaction between the patient and the health-care personal takes place. In this situation, the patient has to apprehend, encode, recall, reason about pros and cons for participation, and finally present the decision verbally to the health-care personal. This process is to a high extent related to various verbal functions (Marson et al., 1995a; Gurrera et al., 2006). Our results also showed a strong association between MDMC and episodic memory, as the summed score of LIMD was highly associated to measurement of verbal learning (RAVLT; Lezak et al., 2004). In the standardized interview performed as part of the LIMD procedure, the subject did not need to remember the whole content of the vignette texts as they were allowed to keep the written information in front of them during the entire interview. However, a certain degree of preserved memory function would possibly be afforded for obtaining a high LIMD score, as the subjects at least needed to keep the current question as well as some information of the vignette in mind long enough to be able to answer it and also for memorizing some parts of the vignette text. The importance of information processing speed and working memory has been reported in previous research on MDMC. Jefferson and collaborators (2008) indicated that executive function and information processing were strongly correlated to the ability to understand information relevant for making a Int J Geriatr Psychiatry 2014; 29: 1304–1311

1310

choice. It is a general finding that cognitive speed is a crucial factor in diseases associated with cognitive impairment and aging (Salthouse, 2010) an outcome that could affect the MDMC. According to the stepwise regression analysis in the present study, cognitive speed added power to the prediction of LIMD performance when the contribution of the other components had been included. The most powerful single test was the Swedish reading speed test (Järpsten, 2002). Notably, the Reading speed test examined not exclusively reading speed but also includes different aspects of reading capacity such as sustained reading comprehension and the ability to draw conclusions. The test is standardized and described in detail in the Swedish manual (Järpsten, 2002). The test consists of a text inserted with 36 parentheses with three words of which one fits in the context. The task is to read the text as fast as possible and to mark which word in each parenthesis is correct. The time limit is 4 min. The ability to draw conclusions from implicit information was assessed by the test Inference. Previous research has found a strong relation between inferencemaking and comprehension ability. When making an inference, presented information and existing knowledge need to be recalled and integrated for the correct solution (Cain et al., 2001). Our result showed that the test performance on Inference was a strong predictor of the LIMD summed score, which underlines the importance of a preserved ability to process language on an abstract level. Finally, the test Sentence repetition showed to be another strong predictor of the LIMD summed score. The subject is asked to repeat long sentences, an ability that involves different cognitive components beside the ability to repeat and communicate sentences, such as encode prior knowledge and remember a certain amount of new information. The present study has some limitations. The definition of decision competence varies among studies of MDMC and calls for clearly expressed criteria and a future consensus on MDMC. Methods used to study MDMC vary considerably, and a comparison between the present study and previous research has to be made with caution because of differences in the methodology. The limited number of subjects and the sample characteristics may affect the statistical power and the possibility to generalize results to a population of patients with mild dementia (MMSE ≥ 20). An advantage of the present study was the use of a comprehensive cognitive test battery, also including tests that not previously have been used in research on cognitive predictors of MDMC. The LIMD as Copyright # 2014 John Wiley & Sons, Ltd.

S. Stormoen et al.

measurement of MDMC has good psychometric features. It was designed in order to provide a systematic and detailed analysis of interview responses assumed to be particularly important in MDMC. Also, possible floor and ceiling effects were avoided by including both patients with mild AD and MCI as well as HC as participants. To conclude, verbal knowledge assessed by a text reading test was the cognitive component and single test measurement, which best could predict MDMC in groups of subjects with impaired cognitive function and healthy controls. As many MDMC instruments (e.g., LIMD) are time-consuming and clinically less suitable, the development of a more easily administered test with good psychometric properties predicting MDMC is an important future matter. A future goal is to develop a short and easily administered cognitive test with good psychometric properties predicting MDMC. A test similar to the Reading speed test, which assessed the ability to read text with sustained comprehension, could be a potential and powerful alternative to other MDMC instruments. Further research is needed to apply theoretical knowledge of MDCM into clinical praxis for the benefit of different group of patients. Conflict of interest None declared. Key points MDMC is associated to several cognitive components interpreted as verbal knowledge, episodic memory, cognitive speed, and working memory. The cognitive component that best could predict MDMC was verbal knowledge. The cognitive test measure best associated to MDMC was Reading speed, a tool that assessed both rapid reading and understanding of text.



• •

Acknowledgements The Swedish Brain Power Consortium supported this study. Financial support was also provided through the regional agreement on medical training and clinical research (ALF) between Stockholm County Council and the Karolinska Institutet. The authors thank all patients and subjects that participated in the study. Int J Geriatr Psychiatry 2014; 29: 1304–1311

Medical decision making in Alzheimer’s disease

References Alexopoulos GS, Abrams RC, Young RC, Shanoian CA. 1988. Cornell Scale for Depression in Dementia. Biol Psychiatry 23: 271–284. Cain K, Oakhill JV, Barnes MA, Bryant PE. 2001. Comprehension skill, inferencemaking ability, and their relation to knowledge. Mem Cognit 29(6): 850–859. Folstein MF, Folstein SE, McHugh PR. 1975. “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12: 189–198. Golden CJ, Freshwater SM. 2002. Stroop Color and Word Test. A Manual for Clinical and Experimental Uses. Stoelting Company: Chicago. Gurrera RJ, Moye J, Karel MJ, Azar AR, Armesto JC. 2006. Cognitive performance predicts treatment decisional abilities in mild to moderate dementia. Neurology 66: 1367–1372. Järpsten B. 2002. DLS™ för skolår 7–9 och år 1 i gymnasiet. Handledning: Stockholm: Psykologiförlaget. Jefferson A, Lambe S, Moser D, Byerly L, Ozonoff A, Karlawitch J. 2008. Decisional capacity for research participation in individuals with mild cognitive impairment. J Am Geriatr Soc 56: 1236–1243. Johansson MG. 2002. LS™ Reviderad. Klassdiagnoser i läsning och skrivning för högstadiet och gymnasiet. Handledning. Psykologiförlaget: Stockholm. Jorms AF. 2004. The informant questionnaire on cognitive decline in the elderly (IQCODE): a review. Int J Clin Psych 16: 275–293. Kaplan E, Goodglass H, Weintraub S. 1983. Boston Naming Test. Lea & Febiger: Philadelphia. Laakso K, Brunnegård K, Hartelius L, Ahlsén E. 2000. Assessing high level language in individuals with multiple sclerosis: a pilot study. Clin Ling Phon 14(5): 329–349. Lezak MD, Howieson DW, Loring HJ, Fisher JS. 2004. Neuropsychological Assessment. Oxford University Press: New York. Marson DC, Cody H, Ingram K, Harrell LE. 1995a. Neurological predictors of competency in Alzheimer’s disease using a rational reasons legal standard. Arch Neurol 52: 955–959. Marson DC, Ingram K, Cody H, Harrell LE. 1995b. Assessing competency of patients with Alzheimer’s disease under different legal standards. Arch Neurol 52: 949–954. Marson DC, Chatterjee A, Ingram KK, Harrell LE. 1996. Toward a neurologic model of competency: cognitive predictors of capacity to consent in Alzheimer’s disease using three different legal standards. Neurology 46(3): 666–672.

Copyright # 2014 John Wiley & Sons, Ltd.

1311 McKhann G, Drachman D, Folstein M, et al. 1984. Clinical diagnosis of Alzheimer’s disease; report of the NINCDS-ADRDA work group under the auspices of the department of health and human service task force on Alzheimer’s disease. Neurology 34: 934–939. Okonkwo OC, Griffith HR, Belue K, et al. 2008. Cognitive models of medical decision-making capacity in patients with mild cognitive impairment. J Int Neuropsychol Soc 14: 297–308 Östberg P, Fernaeus SE, Bogdanovic N, Wahlund LO. 2008. Word sequence production in cognitive decline: forward ever, backward never. Log Phon Voc 33(3): 126–135. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. 1999. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 56: 303–308. Peterson G, Wallin A. 2003. Alzheimer disease ethics-informed consent and related issues in clinical trials: results of a survey among the members of the research ethics committees in Sweden. Int Psychoger 15(2): 157–170. Salthouse T. 2010. Selective review of cognitive aging. J Int Neuropsychol Soc 16: 754–760. Schmand B, Gouwenberg B, Smit JH, Jonker C. 1999. Assessment of mental competency in community-dwelling elderly. Alzheimer Dis Assoc Disord 13: 80–87. Slaughter S, Cole D, Jennings E, Reimer MA. 2007. Consent and assent to participate in research from people with dementia. Nurs Ethics 14(1): 27–40. Tallberg IM. 2005. The Boston Naming Test in Swedish: normative data. Brain Lang 94(1): 19–31. Tallberg IM, Ivachova E, Jones Tinghag K, Östberg P. 2008. Swedish norms for word fluency tests: FAS, animals and verbs. Scand J Psychol 49: 479–485. Tallberg IM, Almkvist O, Stormoen S, Eriksdotter M, Sundström E. 2013. Medical decision-making capacity using a protocol based on linguistic features. Scand J Psychol 54: 386–392. Vellinga A, Smit JH, van Leeuwen E, van Tilburg W, Jonker C. 2005. Decision making capacity of elderly patients assessed through the vignette method: imagination or reality? Aging Ment Health 9: 40–48. Wahlund LO, Pihlstrand E, Jönhagen ME. 2003. Mild cognitive impairment: experience from a memory clinic. Acta Neurol Scand Suppl 179: 21–24. Wechsler D. 1981. Manual for the Wechsler Adult Intelligence Scale-Revised (WAISR). The Psychological Corporation: New York. Winblad B, Palmer K, Kivipelto M, et al. 2004. Mild cognitive impairment – beyond controversies, towards a consensus: report of the international working group of mild cognitive impairment. J Int Med 256: 240–246.

Int J Geriatr Psychiatry 2014; 29: 1304–1311

Cognitive predictors of medical decision-making capacity in mild cognitive impairment and Alzheimer's disease.

Impaired capacity to make decisions in everyday life and situations of medical treatment is an inevitable consequence of the cognitive decline in Alzh...
151KB Sizes 0 Downloads 5 Views