Measuring the impact of Mild Cognitive Impairment on IADL in Parkinson's disease

(1)

Measuring the impact of Mild Cognitive Impairment on IADL in Parkinson’s

Disease

Author: Rachel Brouwer

Studentnumber: 6164692

University of Amsterdam – Brain & Cognition

Master thesis Clinical Neuropsychology

Mentor University of Amsterdam: Anne Geeke Lever

Mentor Academic Medical Center: Gert Geurtsen

Date 04-‐01-‐2015

(2)

ABSTRACT

Background: It has been increasingly recognized that Parkinson’s Disease (PD) is not only associated with motor deficits, but also with cognitive deficits. In PD there is a spectrum of cognitive functioning, ranging from a normal cognition (PD-‐NC) to an intermediate condition: ‘mild cognitive impairment (PD-‐MCI)’, to PDD. The presence of PD-‐MCI and PDD increases functional disability, caregiver burden and has a major impact on quality of life. Measurements of activities of daily living (ADL) have been widely investigated as a possible screening tool to detect and monitor the functional changes related to dementia and to evaluate the quality of life. However, little is known about Instrumental Activities of Daily Living (IADL) measurements and the association with PD-‐MCI. The primary objective of this study is to examine the relation between IADL and PD-‐MCI. Our second study objective is to validate two IADL questionnaires as PD-‐MCI screening and monitoring tools.

Method: Ten patients with PD-‐MCI and 13 PD-‐NC patients underwent neuropsychological tests, 23 healthy informants were administered two IADL questionnaires that evaluated IADL

functioning of the PD patients. The informants of the PD patients were also administered an ADL questionnaire. To determine PD-‐MCI we used the level 2 criteria formed by the MDS-‐task Force. Result: Our results showed no significant differences on the IADL questionnaires (PDCFRS and the AIADL) and the ADL questionnaire (ALDS) between PD-‐NC and PD-‐MCI. Moreover, none of the two questionnaires could distinguish better between the PD-‐NC and PD-‐MCI group. An optimal cut-‐off score for both questionnaires is not determined due to the abnormal ROC curves and related statistics. In addition our results show that PD patients with a NC and a MCI do not show significantly different scores on the ALDS ADL questionnaire. Finally we found that there is a significant negative correlation between the AIADL and the PDCFRS total scores.

Conclusion: The present study results do not confirm an association between cognitive

impairment and IADL difficulties. Our study results imply that the AIADL and PDCFRS cannot be used as a screening and monitoring instrument for PD-‐MCI. Due to a small sample size and unrepresentative sample of PD patients this should be further investigated in future research.

(3)

INTRODUCTION

Parkinson’s disease (PD) is a progressive neurological disease, which is mostly associated with motor deficits. PD motor impairments include amongst others, akinesia, tremor, and rigidity (Schoenberg, 1987). However, it has been increasingly recognized that PD is not only associated with motor deficits but also with cognitive and other non-‐motor deficits. The decrease in cognitive functions in PD patients include in particular a decline in memory and executive functions (Muslimovic et al., 2005). Difficulties in executive functioning include impairments in working memory, attentional processes, response inhibition, planning and visuospatial mental rotation (Bradley et al., 1989; Bokura et al., 2005; Gauggel et al., 2004; Monchi et al., 2004; Owen et al., 1997; Ransmayr et al., 1987; Rowe et al., 2002;). Non-‐motor deficits include: autonomic dysfunction, psychiatric changes, sensory symptoms and sleep disturbances (Sami et al., 2004).

In PD there is a spectrum of cognitive functioning, ranging from normal cognition (PD-‐NC) to an intermediate condition, to PDD. The intermediate condition is characterized by cognitive decline, without the severity of PDD. This ‘predemential level’ in PD patients is also called ‘mild cognitive impairment (PD-‐MCI)’ and is considered to be a harbinger of PDD (Litvan et al., 2012). PD-‐MCI occurs even in early stages of PD and is mainly characterized by impairments of executive functioning and attention. Aarsland & Kurz (2009) found that 75% of the Parkinson patients with cognitive deficits eventually develop PDD.

To distinguish between PD-‐NC, PD-‐MCI and PDD the Movement Disorder Society task force (MDS) developed guidelines, which are appropriate for cognition functioning in PD patients (Litvan et al., 2012). The proposed criteria for PD-‐MCI are combined in two levels of assessment. Level 1 is an abbreviate cognitive assessment and lead to ‘possible PD-‐MCI’ diagnosis. Level 2 is based on a comprehensive cognitive assessment and lead to ‘probable PD-‐MCI’ diagnosis. Level 1 is a quick and easy way to diagnose PD-‐MCI and provides less diagnostic certainty than level 2. Level 2 is a reliable method and allows for full cognitive subtyping of PD-‐MCI (Leroi et al., 2012; Litvan et al., 2012). In this study we use MDS level 2 guidelines to distinguish PD-‐NC and PD-‐MCI patients. The exact guidelines will be discussed more in detail in the ‘Method’ section.

Decline of physical functioning seen in PD patients decreases in tandem with cognitive decline and is associated with several factors: the severity of the motor symptoms, the overall decreasing cognitive functioning, depression, and an older age at time of diagnosis (Leroi, McDonald et al., 2012). This overall functional decline in cognition and physical functioning contributes to the inability to live independently. Moreover the presence of PD-‐MCI and PDD increases the number of nursing home placement and, caregiver burden and has a major impact on quality of life (Aarsland et al., 2000; Aarsland et al., 2003; Bouwens et al., 2009; Hely et al., 2008; Leroi et al., 2012; Marras et al., 2002; Post et al., 2007; Sabbagh et al., 2005; Starkstein et al., 1992; Willis et al., 2012).

(5)

Measurements of Activities of Daily Living have been widely investigated as a possible screening tool to detect and monitor the overall functional changes related to dementia and to evaluate quality of life. To determine quality of life and independence level of an individual, a distinction is made between Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL). ADL are basic self-‐care skills, which are essential for the direct care of an individual's self, for example dressing and showering. IADL are somewhat more complex activities and could be described as activities that are performed in the environment without being dependent on others. IADL include for example, doing groceries and cooking (Gobbens & Assen, 2014).

Studies that examined the overall functional decline in patients with PD often used questionnaires covering only ADL. However, in association with cognition overall functional decline in IADL is more interesting to explore than ADL. IADL includes higher order activities and are therefore more vulnerable to early effects of cognitive decline than ADL-‐activities. For example in Alzheimer Disease it is found that a decline in ADL-‐activities due to cognitive impairment often occurs in later, more severe stages (Barberger-‐Gateau et al., 2000; Peres et al., 2006; Stern et al., 1990; cited in Sikkes et al., 2011; Wilms et al., 2000). Due to the vulnerability to cognitive decline, IADL may be useful for diagnosing dementia conditions such as PD-‐MCI and PDD (Lawton & Brody, 1969). Moreover, problems in complex everyday activities are suggested to be one of the first indications of dementia noticed by the patient or family members, which contributes to recognizing cognitive problems that need to be evaluated (Desai et al, 2004). Performance decline in ADL and IADL due to cognitive impairment is already seen in Alzheimer Disease (Sikkes et al., 2011). An IADL questionnaire, the Amsterdam Instrumental Activities of Daily Living (AIADL), has been validated for patients with Alzheimer’s dementia, but not yet for PD patients (Sikkes et al., 2012). In PD, although there is an association between cognitive functioning in PD patients and difficulties in IADL (Kulivesky et al., 2013; Pirogovsky et al., 2013), little is known about the impact of cognitive impairment on IADL in PD-‐MCI. Up till now, judgments about IADL functioning are often derived from neuropsychological examination and/or with descriptions by caregivers (Troster, 2011; Goetz et al., 2008). Both methods may underestimate the patient’s disability and dysfunction (Kulivesky et al., 2013). In addition, judgments about IADL functioning in association with cognition are also derived from scales/questionnaires, which are intended for assessment in patients with other dementias and do not account for the motor impact of the PD (Feldman et al., 2001).

Kulivesky et al (2013) validated an IADL questionnaire; the Parkinson Disease Cognitive Functional Rating Scale (PDCFRS), as a valid and reliable instrument to examine the effects of Parkinson’s cognitive symptoms on IADL. However, he didn’t use the MDS Task Force level 2 criteria to distinguish between PD-‐NC and PD-‐MCI. The screening tool Parkinson Disease-‐ Cognitive Rating Scale (PDCRS) was used to assess cognition of PD patients and to distinguish between PD-‐NC and PD-‐MCI. This PD specific cognitive brief screening tool is not capable to provide a very accurate and reliable MCI diagnosis and is only suitable for the MDS level 1 criteria. Other studies also tried to validate ADL and IADL questionnaires as suitable questionnaires to distinguish between PD-‐MCI and PD-‐NC patients, however again according to level 1 MDS criteria (Brennan et al., 2015; Pirogovsky et al., 2013, Pirogovsky et al., 2014). To our knowledge the PDCFRS is still not validated according to MDS level 2 criteria as a screener and monitoring tool for functional changes related to PD cognitive impairment. Additionally,

(6)

there are also no other IADL questionnaires validated based on the MDS Task Force level 2 criteria in PD-‐MCI patients.

The lack of an instrument that is quick, valid, and reliable to screen and monitor the functional changes related to PD cognitive impairment, negatively influence clinical management and up-‐ coming research trials (Kulivesky et al., 2013; Leroi et al., 2013). Moreover, an intervention for PD-‐MCI is an unmet need, which is crucial for the overall care of PD patients. PD patients suffering from MCI should be treated as early as possible, to minimize and postpone further cognitive decline and its effect on the quality of life. More insight on the impact of PD-‐MCI on IADL functioning may contribute to a better understanding of PD-‐MCI overall functioning. This could be useful for clinical and research settings. In addition, a reliable identification of IADL deficits is needed, as patients with IADL deficits suffering from MCI have a higher risk of converting to dementia than MCI patients without IADL deficits (Jekel et al., 2015). The use of screening and monitoring tools specifically designed and validated for patients with PD-‐MCI is therefore strongly recommended.

Since cognitive decline has a tremendous influence on PD patients and their families, and since the current lack of knowledge on the association between IADL-‐functioning and cognitive decline in PD-‐MCI, the primary objective of this study is to examine the relation between IADL functioning and PD-‐MCI. Our second study objective is to validate the PDCFRS and AIADL as suitable screening and monitoring tools for PD-‐MCI. For this aim we try to establish their optimal cut-‐off scores for PD-‐NC and PD-‐MCI distinction. Additionally, we will examine which IADL questionnaire is able to make the best distinction between MCI and NC. Finally, we investigate the correlation between the total scores on the AIADL, PDCFRS and ALDS.

We recruit PD patients and informants (spouse) of the PD patients at the Academic Medical Center. The spouses conducted three questionnaires during neuropsychological assessment of the PD patient. There are three established methods to assess questionnaires, each of them has their own strengths and weaknesses: self-‐report, performance-‐based and informant-‐based report (Loewenstein et al., 2010; Desai et al., 2004; Pearson VI., 2000). However, due to the advantages for our study of the informant-‐based method, we used the informant-‐based reports. Advantages of this method include the ease of administration. While the informant is waiting at the AMC hospital for the patient to finish the assessment, the informant could complete the questionnaires. Therefore, in our study informant based reporting is time efficient. Moreover, the ratings are based on real-‐world functional performance of IADLs, thus the patient does not have to perform functional skills in a clinical setting. Also, the patient is not burdened by the assessment.

In the current study IADL activities are measured using two questionnaires; the PDCFRS (Kulivesky et al., 2013) and the AIADL (Sikkes et al., 2013). To determine PD-‐MCI we used level 2 criteria formed by the MDS-‐task Force (Litvan et al., 2012) on ten neuropsychological tests. In addition to the IADL questionnaires, also an ADL questionnaire was assessed. PD-‐MCI patients often perform ADL activities without any problems. ADL activities are less complex than IADL activities. Even by definition there are no ADL problems in patients who only suffer from MCI (Reppermund et al, 2011). We assessed the Amsterdam Linear Disability Scale (ALDS) to verify if PD-‐MCI patients experienced ADL problems. Given previous research results (Kulivesky et al., 2013; Sikkes et al., 2013) it is hypothesized that the PDCFRS and the AIADL are suitable

(7)

questionnaires to distinguish between PD-‐MCI and PD-‐NC. Previous research has shown that MCI patients performed worse than PD-‐NC patients on IADL-‐questionnaires and performed equally on ADL questionnaires. Therefore, it was hypothesized that PD-‐MCI patients will perform significantly worse on the AIADL and PDCFRS but not on the ALDS compared to PD-‐NC patients. It is also expected that the AIADL questionnaire is best suited to make this distinction; this questionnaire is a more extensive questionnaire than the PDCFRS. Finally we expect that the correlation between the PDCFRS and AIADL is larger than the correlation with the ALDS questionnaire, because both PDCFRS and AIADL measure IADL functioning.

METHODS

SUBJECTS

All included PD patients experienced motoric difficulties to such an extent that Deep Brain Stimulation (DBS) operation was considered an optional solution. Prior to this operation all patients underwent DBS screening to determine if they were suitable for the operation. We recruited PD patients and their informants at the AMC hospital visiting this hospital for neuropsychological assessment, which is a mandatory aspect of the DBS-‐screening. This neuropsychological assessment is done to determine cognitive functioning of the PD patient. After neuropsychological assessment and other neurological examinations, the suitability for DBS operation of the PD patient can be determined.

We could safely assume that the PD patients were accurately diagnosed with PD. Prior to the DBS-‐screening at the AMC-‐hospital, the PD patients had already been diagnosed with PD in other medical institutions and were also already treated for PD by one or more neurologists. At the AMC-‐hospital a neurologist reassessed the PD diagnosis. Patients were excluded if neuropsychological assessment leaded to Parkinson Disease Dementia (PDD). Patients were also excluded when diagnosed with a mental or neurological disorder (other than PD), major depressive disorder, psychosis, already underwent deep brain stimulation surgery previously, took medication prior to neuropsychological assessment known to influence cognitive abilities, and other known causes potentially interfering with cognitive abilities. Spouses were included as informants when they had a good insight into the performance on daily activities of the PD patient.

This study included 23 PD patients (17 male, 6 female) and 23 informants. Informants completed the AIADL and PDCFRS. Additionally ALDS data was gained from 7 informants. See for more descriptive statistics on subjects table 3 in section ‘Results’.

(8)

MATERIALS

Diagnostic Process Materials

Neuropsychological tests

Several neuropsychological tests were administered during the DBS screening (see Table 1). The tests were analyzed according to the MDS taskforce Level II criteria to categorize patients as PD-‐ NC or PD-‐MCI. The criteria included that (1) a gradual cognitive decline was reported by the clinician, caregiver or the patient him-‐/ her self, and (2) cognitive deficits must be present on at least 2 out of 10 tests of the neuropsychological battery administered during the DBS screening. If the patient scored at least 1 standard deviation below the mean scores for an age-‐ and gender-‐ matched healthy control group, this is defined as a PD-‐MCI test result. More information on how we applied the MDS taskforce Level II criteria in our study will be discussed in the section ‘Procedure-‐ Dementia screening: PD-‐MCI or PD-‐NC’.

Table 1. Included neuropsychological tests categorized by cognitive domains.

Cognitive domain Neuropsychological Test Test Performance Corrected for

Executive functioning GIT Letterfluency

TMT part B Education Age, education, sex and TMT A

Language BNT

WAIS IV similarities Age and education Age

Memory RBMT IR

Dutch version Rey AVLT DR

Age, education and sex Age, education and sex Speed and Attention TMT part A

Stroop test II (color)

Age and education Age, education and sex Visuospatial functioning GIT visuospatial reasoning

JOLO Age Age and sex

Abbreviations: GIT, Groninger Intelligence Test; TMT, Trail Making Test; BNT, Boston Naming Test; RBMT, Rivermead Behavioral Memory Test; AVLT, Auditory Verbal Learning Test; JOLO, Judgement Of Line Orientation.

Questionnaires

In addition to the neuropsychological tests, we administered three questionnaires to explore the association of functional impairment with continuous measures of cognitive functioning: the AIADL, PDCFRS and the ALDS (Sikkes et al., 2013; Kulivesky et al., 2013, Weisscher et al., 2007) .

(9)

PDCFRS (Kulivesky et al., 2013)

Based on the lack of specific instruments of functional assessment, which control for PD motor aspects, the PDCFRS is designed. The PDCFRS is an IADL-‐based questionnaire and is designed to detect a wide range of functional aspects suspected to be sensible for cognitive impairment in PD. The questionnaire consists of items that take the influence of motor impact on daily activities of PD into account. The questionnaire is already validated for PD patients with MDS level 1 criteria (Kulivesky et al., 2013). The spectrum of instrumental cognitive changes seen in PD is covered by a total of 12 questions (Kulivesky et al., 2012). All instrumental cognitive changes are considered in a period of two weeks before evaluation. All questions explore whether or not the patient has had any trouble in performing an activity with minimal motor involvement. Example questions relate to: handling money or personal mail, controlling drug schedule, and organizing daily activities. Each question has a 4-‐point scale response option (none; some of the time; most of the time; the subject has never done the activity in the past, scored respectively: 0, 1, 2, 8). If a question is scored with an 8 (which indicates the activity is never done in the past), the mean of all the items answered with 0 or 1 or 2 is calculated. This mean value (either 0, 1, 2) is replaced for all the items scored with an 8 (Kulivesky et al., 2013). The score 0 means no dysfunction. The maximum score of the questionnaire is obtained by the sum of the ratings as stated above and is 24. A total score of 24 means cognitive dysfunction. The mean time needed for completion of the questionnaire was approximately 5 minutes. This questionnaire is the only questionnaire we used in our study, which is not based on the IRT method. Consequently, all items are presented to all patients irrespective of their disability level.

AIADL (Sikkes et al., 2013)

The AIADL is a computerized and informant based IADL questionnaire aimed at measuring problems in IADL. The questionnaire is also intended to detect early-‐onset dementia and dementia in an early stage, and is already validated in Alzheimer patients. The questionnaire is based on the item-‐response theory (IRT) to improve scoring accuracy, and by only using the discriminative items, the test administration time is efficient. In this way, items are tailored and adaptive to the individual response. Activities that are non applicable to the patient are skipped. For example, if the patient did not use a dvd-‐recorder, there were no further detailed questions about the use of a dvd-‐recorder. Moreover, this method eliminates floor and ceiling effects. As a result of the IRT-‐method the AIADL is a branched questionnaire, which contains a minimum of 47 and a maximum of 70 activities. Each item contains a 5-‐point scale response option (scored 0-‐4, Sikkes et al., 2013). All activities are divided in seven categories, see Table 2. It took approximately 23 minutes to conduct the questionnaire. The scoring of the questionnaire is based on the item response theory as stated above. To make the results easier to interpret, the logistic scores are transformed. The mean of the score is transformed linearly to 50 and a SD of 10, which results in a scoring range from 20 to 80 (SD ± 3). Lower scores indicate poorer functioning (Sikkes et al., 2013). It is important to note that this questionnaire contains questions about activities from the last four weeks.

(10)

Table 2. AIADL questionnaire categories.

Category Examples from activities

1. Household Groceries -‐ Cooking

2. Home appliances Microwave -‐ Dishwasher

3. Administration Paying the bills-‐ Electronic banking

4. Work All sorts of work

5. Computer Using the internet

6. Appliances Handling the remote control

7 Free time Driving – playing games

Abbreviations: Amsterdam-‐ Instrumental Activities of Daily Living (AIADL)

Short summary differences: PDCFRS versus AIADL

The AIADL questionnaire is an extensive computerized questionnaire based on the IRT method. The PDCFRS questionnaire is a brief and paper based questionnaire and is not based on the IRT method. The PDCFRS is designed for Parkinson patients and is designed to detect a wide range of functional aspects suspected to be sensible for cognitive impairment in PD. The AIADL questionnaire emphasizes more physical IADL items than the PDCFRS and does not control for PD related motor impairments.

ALDS (Weischer et al., 2007)

The ALDS-‐questionnaire was developed to quantify the functional status in terms of the ability to perform Activities of Daily Living (ADL). The items cover a large number of activities and are suitable for many types of chronic conditions and for a very wide range of functioning (Holman et al., 2003; Holman et al., 2005). It is also a tool suitable to determine the level of disability in diagnosed PD-‐patients (Weisscher et al., 2007). The informants are asked whether the PD patient is able to carry out a given activity in the present time. The current version of the ALDS item bank uses (as well as the AIADL) an IRT framework and consists of 77 items ranging from relatively easy to difficult. The items are tailored to the ADL level of the patient. Even by using a small number of items a sufficiently and detailed clinical picture is obtained. In our study each informant was assessed with 15 randomly selected items to determine a total score. Due to the expired license of our ALDS test-‐account some of the informants had to answer the questions in a recently developed smartphone application (App). Moreover, due to this expired license problem, unfortunately we lost data from our test account and we only had access to the data collected by the app. This app gives us no insight in the conducted items and the answers, only a total sum score. To make the results easier to interpret, these logistic scores are transformed in the ALDS-‐app. Both the patient’s ability to carry out an activity and the difficulty of the items are arranged on a single hierarchical linear scale. After the completion of the questionnaire in the ALDS-‐app, a linear transformed ALDS-‐ score of 0(dead)-‐90 could be obtained. The value 10 represents the lowest possible level of functional status and the value 90 represents the highest possible functioning level and means that there is no dysfunction (see Figure 1). Higher scores on the questionnaire indicate a higher level of self-‐sufficiency. It took approximately 5 minutes to complete the questionnaire.

(11)

.

Figure 1. Measurement range of AMC Linear Disability Score

PROCEDURE

Participants – Data collection

The University of Amsterdam and the Ethical commission of the Academic Medical Center in Amsterdam approved the study. Written informed consent was obtained both from the patient as the spouse prior to the neuropsychological assessment of the PD patient. If the spouse was not present at the neuropsychological screening the informed consent was digitally obtained by email. The results of the neuropsychological assessment were only used after receiving informed consent both from the PD patient and the spouse. All questionnaires were conducted within one month after receiving the informed consent.

Data collection was scheduled from April 16th_{till
September
23
in
2015
at
the
AMC
hospital.}

In total 23 patients with idiopathic PD and their informants participated in this study. All subjects were Dutch except for one patient (Spanish, although could speak moderately Dutch). The patients underwent a comprehensive neuropsychological assessment, which we used to diagnose PD-‐MCI.

Data collected at screening included age, age at disease onset and educational level. Using the Dutch classification according to Verhage (Verhage, 1964) the level of education was categorized, ranging from 1 to 7 (low to high); 1= did not finish primary school, 2= finished primary school, 3= did not finish secondary school, 4= finished secondary school, low level, 5= finished secondary school, medium level, 6= finished secondary school, highest level, and/or college degree, 7= university degree.

(12)

For each PD patient, a knowledgeable informant completed the three questionnaires during the research visit. The questionnaires were self-‐administered on a computer at the AMC hospital in a quiet room. We conducted the questionnaires using two web surveys: ‘Qualtrics’ for the A-‐IADL and PDCFRS and ‘Questmanager’ for the ALDS. Some informants conducted the questionnaires at their own home, they installed the ALDS-‐App themselves on their own devices. The informants could submit the questionnaire on this smartphone.

In most cases, informants completed the questionnaires prior to the outcome of the neuropsychological tests. Therefore, they had no knowledge of the cognitive status of the patient according to the neuropsychological assessment. However, this was not always possible because the informant was not always present at the screening. Some informants received the questionnaires by email after informed consent was obtained. Informants were given as much time as needed to complete the questionnaires. It took approximately 30-‐45 minutes in total to administer the three questionnaires. All informants received the questionnaires in the same order: first two IADL questionnaires (the AIADL followed by the PDCFRS) and subsequently the ALDS ADL questionnaire.

PD-‐MCI or PD-‐NC

To establish whether the PD patient has PD-‐MCI or not (PD-‐NC), we used the PD-‐MCI diagnostic level 2 (a comprehensive assessment) criteria formed by the MDS Task Force (Litvan et al., 2012). The MDS Task force criteria stated that for full subtyping PD-‐MCI subtyping at least ten tests covering five cognitive domains must be included. We included two tests of each cognitive domain as previously listed (see table 1 and see Lindeboom et al., 2003 for more information about the tests), hereby we addressed all of the five cognitive domains equally. The PD patients were diagnosed with MCI and assigned to the PD-‐MCI group if they scored at least 1 standard deviation below the mean on two tests within one cognitive domain. This is called single domain PD-‐MCI. PD patients were diagnosed with MCI and assigned to the PD-‐MCI group if they scored at least 1 standard deviation below the mean on one task in two or more separate cognitive domains, which is called multiple domain MCI (Litvan et al., 2012). Moreover, for diagnosing PD-‐ MCI, a subjective cognitive decline had to be reported by either the patient or the informant (Litvan et al., 2012). All together, based on the level 2 diagnostic criteria (comprehensive assessment) we assigned the PD patients to one of two groups; the PD-‐MCI group or the PD-‐NC group (Litvan et al., 2012).

Patients were excluded from analysis when they met the Parkinson Disease Dementia criteria (PDD criteria; Emre et al., 2007).

(13)

STATISTICAL ANALYSIS

Data analysis methods

We filled in missing values if no more than 4 tests values, out of the 10 tests, were absent. This was done for 5 patients (with 4 task values missing for 1 patient, 3 task values missing for 1 patient and 1 task value missing for 3 patients). The missing data was due to insufficient time to complete the comprehensive neuropsychological assessment, colorblindness of one of the patients, fatigue, hyperkinesia, and Dutch not being the native language. In all cases missing values were replaced by the average performance of all the PD patients except for the fatigue condition. For this condition, missing values were replaced by the lowest mean performance on the tasks, as fatigue represented inability of the patient.

First, we investigated whether there were any outliers for the scores on the three questionnaires. We looked at box plots of the scores on the total PDCFRS, AIADL and the ALDS. A subject was defined as an outlier and was excluded from our analysis if he/she scored at least 2 standard deviations below the average performance on two questionnaires. We decided to use this outlier exclusion rule because we only gained ALDS data of 7 participants. Thereafter, we determined whether our data was normally distributed. The Shapiro-‐Wilk test is an appropriate normality test for small sample sizes (N<50). Owing to our small sample size (N=23), we used the Shapiro-‐Wilks test to check for normality. To determine whether the variances were equal, we executed the Levene’s Test. Next, we calculated the descriptive statistics for demographic and clinical variables, which included: means, percentages and standard deviations. The data are expressed as means ± the standard deviations. Due to the logistic transformed data of the questionnaires and because we want to determine the percentage correct classification of the questionnaires we performed logistic regression analyses. We performed two logistic regression analyses to determine whether the scores on the PDCFRS, AIADL and ALDS significantly differ for PD-‐NC and PD-‐MCI group. We were interested if the two questionnaires PDCFRS and AIADL measuring IADL are correlated. Therefore, we performed a bivariate correlation to determine if the total scores of the IADL questionaires PDCFRS and the AIADL, significantly correlate with each other for the PD patients. The correlation coefficient is commonly used to measure the size of an effect. Values of ± 0.1 indicate a small effect, ± 0.3 indicate a medium effect and ± 0.5

indicate a large effect (Field, 2009). To identify the discriminative power and the accuracy of the

PDCFRS and the AIADL, Receiver Operator Charasteristic (ROC) curves were generated. We looked at the Area Under the Curves (AUC) (< 0.6 = worthless; 0.6 -‐ 0.7 = poor; 0.7 -‐ 0.8 = fair; 0.8 -‐ 0.9 = good; 0.9 -‐ 1.0 = excellent) (Field, 2009). To determine whether the PDCFRS or AIADL is a better cognitive screening tool to identify PD-‐MCI, we analyzed the confidence intervals (CI) of the AUC. If the CI’s show overlap, none of the two questionnaires is significantly better than the other to distinguish between PD-‐MCI and PD-‐NC patients. The scores on the PDCFRS and AIADL were used as predictor variables and PD-‐MCI group as the state variable. SPSS version 22.0 for IOS was used for statistical analyses; p values < 0.05 was considered statistically significant. Significance values are displayed for each analysis.

(14)

RESULTS

POWER

As described above, based on a power analysis (G * Power 3.0.10) it was reasoned that for a power of 0.80 with a significance level of α = 0.05, and an effect size of d = 0.4, the sample size should be N = 34. However due to circumstances this criteria is not met, e.g. some patients refused to participate, and limited DBS screenings were conducted due to summer holiday. We included 23 participants. This sample size gives the statistical outcomes a power of 0.6.

Descriptive Statistics

We excluded one PDD patient from our analyses. After checking for outliers, none of the other included patients had to be excluded. Mean age of PD patients at onset of the diagnosis was 51.21 (SD= 7.84). Mean age of the PD patients at neuropsychological assessment was 61.00 (SD= 7.67). Mean education level was 5.39 (SD=1.12) (Verhage, 1964). See table 3 for a short overview of demographic and clinical variables for al PD patients, PD-‐NC, PD-‐MCI group and the comparison of these variables between the PD-‐NC and PD-‐MCI patients.

.

Table 3. Descriptive statistics.

Abbreviations: AIADL, Amsterdam-‐ Instrumental Activities of Daily Living , PDCFRS, Parkinson Disease-‐ Cognitive Functioning Rating Scale. ALDS, AMC Linear Disability Scale. PD-‐NC, Parkinson Disease patients with normal cognition, PD-‐MCI, Parkinson Disease patients with Mild Cognitive Impairment.

Note: Data expressed in means ± Standard Deviations.

¹ Education according to Verhage classification, ranging from 1 (low) to 7 (high). ² Derived from an independent T-‐test, not significant at the p<0.05 level.

All PD patients

(N=23) PD-‐NC (N=13, 57%) PD-‐MCI (N=10, 43%) P-‐value and F-‐value PD-‐NC and PD-‐MCI

group² Age 61.00 ± 7.67 59.69 ± 8.55 62.50 ± 6.43 p(0.40) F (1.80) Age at onset PD 51.21 ± 7.84 49.85 ± 9.39 53.00 ± 5.16 p(0.35) F(3.41) Education¹ 5.39 ± 1.12 5.69 ± 1.03 5.00 ± 1.15 p(0.15) F(0.02) A-‐IADL 58.40 ± 6.46 60.68 ± 6.79 55.43 ± 4.83 p(0.05) F(0.69) PDCFRS 3.30 ± 3.87 2.31 ± 3.38 4.60 ± 4.27 p(0.17) F(1.50) ALDS 82.86 ± 15.50 (N=7) 80.60 ± 18.35 (N=5) 88.50 ± 2.12 (N=2) p(0.60) F(1.91)

(15)

PD-‐NC versus PD-‐MCI

23 PD patients completed neuropsychological assessment. There were missing values for the GIT visuospatial reasoning, RBMT, Stroop and AVLT. In total 4,3% of the data was missing. According to the level 2 diagnostic criteria, 10 patients were diagnosed with PD-‐MCI (43%). All PD-‐MCI patients were considered multiple domain PD-‐MCI. The remaining 13 patients had a normal cognition (57%). The total scores for both the PD-‐NC and PD-‐MCI group on the PDCFRS, AIADL and ALDS were not normally distributed. The PD-‐NC and the PD-‐MCI groups did not show a significantly different evaluation on the ALDS, AIADL or PDCFRS questionnaire (see Table 4). The correct classifications determined with logistic regressions were: 69.6% for the AIADL, 71.4% for the ALDS and 60.9% for the PDCFRS.

Table 4. Outcome of the logistic regression analyses1_.

Statistics AIADL ALDS PDCFRS

Lower C.I. ¹ 0.74 0.78 0.93 Upper C.I. ¹ 1.01 1.52 1.51 Odds Ratio 0.87 1.08 1.18 B (SE) p-‐value² -‐0.15 (0.08) p(0.06) 0.08 (0.17) p(0.64) 0.17 (0.12) p(0.17) Model X (1) p-‐value² 11.55 p(0.17) 2.90 p(0.23) 6.35 p(0.17) %Correct Classification 69.60% 71.40% 60.90%

Abbreviations: AIADL, Amsterdam-‐ Instrumental Activities of Daily Living , PDCFRS, Parkinson Disease-‐ Cognitive Functioning Rating Scale. ALDS, AMC Linear Disability Scale.

Note:

¹ For Lower and Upper Confidence Interval, 95% C.I for EXP (B). ² Statistic signification level p<0.05.

(16)

Correlation between the AIADL and the PDCFRS

To determine whether PDCFRS and AIADL scores significantly correlate we executed a bivariate correlation. The total PDCFRS score was negatively related to the total AIADL score, r = -‐ 0.657 (large effect), p < .001 (see Table 5). Higher AIADL scores were associated with lower PDCFRS scores (see Figure 2, Scatterplot). The ALDS did not show a significant correlation with the PDCFRS or the AIADL (see Table 5).

Table 5. Bivariate correlations between PDCFRS, AIADL and the ALDS.

Total score

PDCFRS Total score AIADL Total score ALDS Total score PDCFRS Spearman Correlation P-‐value N 1 -‐ 23 0.657¹ 0.002 23 0.169 0.717 7 Total score AIADL Spearman Correlation P-‐value N -‐0.657¹ 0.002 23 1 -‐ 23 0.315 0.491 7 Total score ALDS Spearman Correlation P-‐value N 0.169 0.717 7 0.315 0.491 7 1 -‐ 7

Abbreviations: AIADL, Amsterdam-‐ Instrumental Activities of Daily Living , PDCFRS, Parkinson Disease-‐ Cognitive Functioning Rating Scale. ALDS, AMC Linear Disability Scale.

Note: ¹ Statistic signification P<0.01

(17)

F

igure. 2. Scatter plot of the PDCFRS and theAIADL. Total scores for PD patients with normal

cognition (NC) and mild cognitive impairment (MCI).

Abbreviations: NC, cognitive intact Parkinson's disease patients; MCI, mild cognitive impaired Parkinson's disease patients; PDCFRS, Parkinson's Disease-‐Cognitive Functioning Rating Scale; AIADL, Amsterdam Instrumental Activities of Daily Living Scale.

(18)

To identify the discriminative power and accuracy of the PDCFRS and the AIADL, Receiver Operator Characteristic (ROC) curves and other associated statistics were generated. The Areas Under the Curves (AUC) are <0.6 and overlapped for the PDCFRS: total AUC 0.265 (0.060-‐0.470; CI 95%) and the AIADL: total AUC 0.277 (0.053-‐0.501; CI 95%). Both AUC showed worthless values (<0.6). The ROC curves as seen in figure 3 are abnormal. It is not reliable to determine an optimal cutoff point. We decided to scope out other ROC curve related data, due to the abnormal ROC curves.

AIADL PDCFRS 1

Figure. 3 ROC curves

Abbreviations: AIADL: Amsterdam Instrumental Activities of Daily Living Scale; PDCFRS: Parkinson's Disease-‐ Cognitive Functioning Rating Scale.

Note: 1. Diagonal segments are produced by ties.

Measuring the impact of Mild Cognitive Impairment on IADL in Parkinson's disease