What stroke patients can benefit from computer-based cognitive rehabilitation? : identifying patient characteristics that lead to adherence in a computer-based cognitive rehabilitation program

Society is affected by stroke because of higher institutionalization rates and greater health care

costs, and stroke patients often suffer both cognitive and emotional consequences. As stroke is a

common disease, 41.000 Dutch citizens are diagnosed with stroke yearly, it is necessary to provide

personalized and cost-effective rehabilitation programs. Computer-based cognitive rehabilitation

(CBCR) is an inexpensive addition to current rehabilitation. However, the effectiveness of CBCR

is still controversial, partly due to a high rate of non-usage which complicates research. This

study aims to identify characteristics in stroke patients that facilitate CBCR adherence.

A group of 53 stroke patients took part in Lumosity’s brain gaming program.

Neuropsychological tests and psychological questionnaires were carried out before and after

intervention, and were correlated to program adherence.

The results suggest that patients with higher quality of life, but higher rates of subjective

cognitive failure, are most inclined to show adherence to a CBCR program. Rehabilitation

practitioners could use this insight in setting up personalized rehabilitation programs by using

CBCR as primary rehabilitation intervention in this specific patient group, while providing more

coaching and supervision for patients who do not fit this profile.

Identifying patient characteristics that lead to adherence in

a computer-based cognitive rehabilitation program.

E. Stevens

Sophia Rehabilitation, The Hague, The Netherlands

University of Amsterdam, The Netherlands

emmastevens1987@hotmail.com

Masterthesis Clinical Neuropsychology & Health Psychology

University of Amsterdam

Student number: 5815142

Supervisor: J. Buitenweg, MSc

Second supervisor: R. van de Ven, MSc

External supervisor: Drs. A. de Kloet, Sophia Revalidatie, Den Haag

Submitted: 01-09-2014

Chapter 1

Chapter 2

Chapter 3

Chapter 4

Chapter 5

INTRODUCTION

...

₃

METHODS & PROCEDURE...

5

SAMPLE CHARACTERISTICS...

OPERATIONALISATION ...

MATERIALS...

INTERVENTION...

DATA ANALYSIS...

RESULTS

...

₁₂

PATIENT CHARACTERISTICS...

OVERALL ADHERENCE...

Influence of demographic characteristics on program usage 13 Influence of psychological factors on program usage 13

DISCUSSION

...

₁₄

could negatively affect long-term cognitive recovery and neurological outcomes. The impact of cognitive impairment is immense, and reaches beyond these personal adverse consequences as diminished functioning in daily life (Paker et al., 2010), to higher rates of institutionalization for rehabilitation purposes (Pasquini et al., 2007), and greater health-care costs (Claesson et al., 2005). These negative effects of cognitive impairment accentuate the importance of cognitive rehabilitation. The effect of the compensatory approach in cognitive rehabilitation has been established quite firmly: patients learn about and practice to apply compensations for cognitive limitations (Cicerone, 2011). In the remedial approach, the aim is to train, practice and stimulate to directly improve the underlying cognitive limitations (Sohlberg & Mateer, 1989). Among these, computer-based cognitive rehabilitation (CBCR) is often used: patients use a computer as an intervention tool (Cha & Kim, 2013). Although the effects of CBCR are sometimes presented in very exciting and promising ways (e.g. “Based on extensive research, Lumosity improves memory, attention, processing speed, and problem-solving skills so you can feel more confident in your abilities“; http://lumosity.com/how-we-help), there is also a large body of research that doubt the effectiveness and generalizability of CBCR. In 1993, a review study by Carroll (1993) indicated very limited success in fundamentally and permanently Stroke is the leading cause of death worldwide, and

every year 41.000 Dutch citizens are diagnosed with stroke (Shim, 2014; Hersenstichting.nl, 2014). Among the survivors of stroke, many show cognitive impairment: estimated percentages ranging from 22% to 50% (respectively Douiri et al., 2013 and Paul et al., 2007). Up to 32% of patients show persistent cognitive impairment up to three years after their first stroke (Patel et al., 2003). In addition, emotional disturbances have been frequently reported: depression and anxiety are commonly observed in stroke patients (Lincoln et al., 2013), and depression has been estimated to affect 33% of all stroke patients (Hackett et al., 2005). Kravetz (1995) found the self-concept of people with brain damage to be more negative and less self-confident than that of individuals without brain injury.

The cognitive and emotional consequences of stroke seem to interact, as studies have reported that impairments in cognition can give rise to feelings of decreased competence in everyday life (van Zandvoort et al., 1998). Multiple studies also show that post-stroke emotionality disturbance is associated with significant cognitive impairment and poor rehabilitation outcomes (Starkstein et al., 2008; Akerlund et al., 2013; Nys et al., 2005). Furthermore, D’aniello et al. (2014) reported that this emotional disturbance following stroke

individualized treatment based on each patient’s neuropsychological patterns and repeatedly stimulate the impaired cognitive areas. An extra advantage is that patients can work at their personally preferred pace (Connor & Standen, 2012; Redick et al., 2013; Björkdahl et al., 2013). An individualized format could benefit patients that experience performance anxiety in a more traditional group-format intervention (Kueider et al, 2012). As a result of these obvious advantages of CBCR over the more expensive strategy training, a lot of current research focuses on the question whether CBCR can lead to any general improvement in cognitive functioning. However, most studies struggle to report reliable and consistent findings due to a high number of non-users (Connor & Standen, 2012; Bossen et al., 2013). Qualitative data suggest several factors that impede the usage of CBCR. Reduced participation has been associated with fatigue, a common problem in stroke survivors (Lundqvist et al., 2010; Johansson & Tornmalm, 2012). Additionally, the cognitive impairment itself often leads to reduced participation because the intervention demands are too high (Bossen et al., 2010). For example, participants have reported that they found the task demands too challenging, starting with the login process (Connor & Standen, 2012).

As CBCR studies encounter high rates of non-usage, it is important to identify the factors that are related to usage. A proposed important factor changing one’s general intellectual abilities

thus far. Also more recently, studies have been published showing that even though improvement occurs on the trained tasks, the beneficiary effects did not transfer to untrained tasks, not even those tasks that were cognitively closely related (Owen et al., 2010; Redick et al., 2013). Despite the doubts regarding transfer of training to everyday life, trained subjects do sometimes report subjective improvement in various aspects of cognition, even in absence of any objective evidence for change (Redick et al., 2013). Also, a recent study that focused on computerized working memory training after brain injury shows that less depressive and anxiety symptoms were reported by the participants after training (Åkerlund et al., 2013). However, these positive effects of training on experienced cognitive functioning and mood are not specific for CBCR and are also observed in compensatory approaches as cognitive strategy training programs (Cicerone et al., 2011). The most extensively described advantage offered by CBCR over traditional cognitive compensatory training programs is that it is a more cost-effective alternative. This offers the possibility of more widespread usage of CBCR. Additionally, the fact that CBCR can be done at home creates the possibility that also homebound patients can access cognitive rehabilitation (Kueider et al., 2012; Cha & Kim, 2013). Moreover, CBCR can be used to provide

The psychological factors chosen are based on previous research and consist of scales for self-efficacy, subjective cognitive impairment, fatigue and overall quality of life. Following earlier research outcomes, it was hypothesized that a high level of self-efficacy would enhance adherence to a CBCR program, whereas fatigue is expected to be an impeding factor. For the rest of the psychological factors, subjective cognitive impairment and overall quality of life, no clear hypotheses could be formed because of contradictory results of previous research. These factors will be also looked into in an explorative manner. By means of this study, which looks at possible predictors of adherence, we could better identify patients that will profit from CBCR and cultivate better patient specific rehabilitation.

SAMPLE CHARACTERISTICS

From two rehabilitation centres (Het Rijnlands Revalidatiecentrum and the Sophia Revalidatie) participants were selected which were diagnosed with stroke within 12-36 months from September 1, 2012. Patients were invited if they were 45-75 years old, had access to a computer with internet access, were able to visit the rehabilitation centre, had enough time for assessment, and were able to play videogames 5 times a week.

is the level of self-efficacy (Bossen et al., 2013). Self-efficacy is described as the belief in one’s overall competence to effect performances across a wide variety of situations, even when the tasks are difficult or novel (Bandura, 1997). Self-efficacy and health outcomes seem to be highly correlated (Luszczynska, 2005; Gramstad et al., 2001). Qualitative data from a recent study pointed out several patient characteristics that seem to play a key role in adherence. Patients that felt personally responsible for their progress outcome were most likely to finish. Lack of self-discipline, low mood and additional health problems were indicated as negative factors and seemed to result in non-usage (Bossen et al., 2013).

The main goal in this study was to further examine and possibly identify the patient characteristics that facilitate adherence to CBCR. Even though the effects of CBCR are already being investigated, there has been relatively little work done to investigate what patients actually show adherence to CBCR. In the present study, we examined the effect of several demographic patient characteristics on program usage, among them gender, age and level of education. As no clear expectations could be formed about demographic patient characteristics on the basis of previous research, they will be examined in an exploratory manner. Furthermore, we looked at psychological factors and their influence on adherence to the intensive CBCR program.

METHODS &

PROCEDURE

in the end 110 patients were present at the first assessment day. Of these 110 patients, 53 patients were eventually assigned to the intervention group.

OPERATIONALISATION

To answer our research questions, we will analyse data from the individuals who took place in the study and were assigned to the intervention group. Before the first meeting, all participants received multiple questionnaires in their e-mail. During the first meeting, all participants were subjected to a neuropsychological assessment. Hereafter the participants were randomized and either allocated to the intervention or the control group. For the intervention group (which is of interest for this study) the intervention followed, which consisted of an eight week long serious brain gaming program provided by Lumosity Inc. (www.lumosity.com, 2014). Participants were asked to play 5 days a week, for 15 to 20 minutes a day. The control intervention consisted of reading a fact about the brain every week for the same period of eight weeks. Immediately after these eight weeks, all participants were retested (all neuropsychological tests and questionnaires were repeated, parallel tests were used where possible) and the complete assessment was done again after 16 weeks. Adherence to the program was remotely monitored by inspecting the number of times logged in, and total play time in minutes was registered. We will look at how well the demographic characteristics and questionnaire outcomes predict total time Exclusion criteria were clinical depression,

aphasia and/or other sensory integration problems, lack of computer skills, no computer or internet access. Patients also could not participate if they were already involved in another cognitive rehabilitation or intervention program. According to the registries 889 patients were potentially eligible for the study and were invited (203 in Leiden and 587 in The Hague). Participants were asked to return the agreement form attached to the invitation when they met the inclusion criteria and were willing to participate. Eventually 146 patients (64 in Leiden and 82 in The Hague) returned the agreement form and gave permission to receive a telephone call to answer several additional questions regarding the exclusion criteria. As four patients did not answer the telephone calls, 142 out of the 146 patients were reached. Eventually 31 out of the 142 patients were not included in the study as they met at least one of the exclusion criteria: antidepressant use (n=5), aphasia and/or other sensory integration problems (n=5). Another 21 out of the 142 responded patients actually did not meet the inclusion criteria for one of the following reasons: not enough time to participate (n=6), illiteracy (n=2), no Internet access (n=5), CVA diagnosis more than 40 months ago (n=6) or not meeting the inclusion criteria for age (n=2). Finally, 115 patients were included and randomized. Directly after randomization,

Lumosity Inc.

Questionnaires

Cognitive Failure Questionnaire

The Cognitive Failure Questionnaire (CFQ), a self-rating scale, was used to rank subjective perceived cognitive failure. The questionnaire covers self– reported failure in different cognitive domains, i.e. attention, memory, perception, and motor function (Broadbent et al., 1982). Participants were asked to identify how often a specific cognitive mistake has occurred in their life in the last 6 months (e.g., “Do you fail to notice signposts on the road?”, “Do you forget appointments?” and “Do you bump into people?”). The CFQ is designed as a frequency scale and includes 25 items, each being a five-point scale item ranging from never (0) to very often (4). The overall score lies within the 0 – 100 range. Broadbent et al. (1982) reported that the questionnaire has high test-retest correlation and high internal consistency. This was confirmed in a 1995 Dutch study by Merckelbach, in which the test-retest reliability was high (0.83) and also internal validity was high (Cronbach’s α = 0.81). General Self-efficacy Scale

The scale assesses a general sense of perceived self-efficacy, which reflects an optimistic self-belief and the belief that one can both perform a novel or difficult task and cope with adversity (Sherer et al., 1982). This self-administered scale consists of 10 items that are all scored on a 4-point scale (e.g., I can can always manage to solve difficult problems

played and hereby hope to identify patient characteristics that facilitate adherence to computer-based cognitive rehabilitation. For this reason, all analyses were done using the first complete assessment.

MATERIALS

Assessments were carried out at three moments: immediately before the start of the intervention (t0), 8 weeks from baseline (t1) , and 16 weeks from baseline (t2). At all test occasions, the assessment consisted of both questionnaires and a neuropsychological test battery. Questionnaires were filled in at home via a digital link. All tests were conducted in the rehabilitation center.

Patient characteristics

Sociodemographic characteristics were recorded at baseline for every participant: participants were asked to inform about age, gender and educational level. The variable ‘educational level’ was divided into the categories ‘low’ (VBO, VMBO and MAVO), ‘moderate’ (HAVO, VWO and MBO) and ‘high’ (HBO and WO), by using the CBS classification (www.nationaalkompas.nl). Also, patients were asked about work participation and living situation (independent vs. dependent and together vs. alone). Additionally, type of stroke (infarction vs. haemorrhage) was registered.

Adherence

Adherence was employed as a continuous dependent variable, i.e. total playtime in minutes. Participants’ total playtime was provided by

response choices. The questionnaire is organized into eight multi-item scales: physical functioning, role limitations due to physical health problems, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional problems and general mental health (Jenkinson et al., 1994). All raw scale scores are linearly converted to a 0 to 100 scale, with higher score indicating higher levels of functioning or well-begin (Aaronson, 1998).

Neuropsychological Assessment

Computerized versions were used for all tests. The complete neuropsychological battery consisted of the Trail Making Test (TMT), the Corsi Block-tapping Test, Digit Span, Eriksen Flanker Task and the Raven Colored Progressive Matrices. The entire test battery was completed in approximately one hour. Tests were administered in a fixed order and every patient followed the same instructions. Assistance during testing was provided by one of the researchers. In this study, the main interest was to identify factors that play a role in adherence to a CBCR program. As intelligence was opted as a possible influencing factor, only the Raven Colored Progressive Matrices, a test used to administer fluid intelligence, was included in analyses. The other neuropsychological test are therefore described in a concise manner.

Raven Colored Progressive Matrices

This multiple-choice test has been used to administer fluid intelligence (Lezak et al., 2004). Each item contains a pattern problem with one

if I try hard enough). Overall scores thus range from 10 (low perceived self-efficacy) to 40 (high perceived self-efficacy). In samples from 23 nations, reliability was high with Chronbach’s α ranging from .76 to .90 (Schwarzer et al., 1995). An extensive 2001 review confirms that the instrument has high internal consistency reliability (ranging from .86 - .91). Test-retest coefficients showed that the scale was stable over time (r = .67, Chen et al., 2001).

Stroke Specific Quality of Life Scale

The Stroke Specific Quality of Life Scale (SS-QOL) is a self-administered scale which aims to measure quality of life after stroke. It consists of 49 items in 12 domains: energy, family roles, language, mobility, mood, personality, self-care, social roles, thinking, upper extremity function, vision, and work/productivity (Post et al., 2010). Each item is scored on a 5-point scale, with higher scores indicating better functioning (e.g., I felt I was a burden to my family). Domain scores are the unweighted averages of the item scores, averaging these unweighted domain scores gives the total score. Summary scores therefore also range from 1 to 5, respectively low to high quality of life. All 12 domains showed high internal reliability with Chronbach’s α of at least 0.73 (Williams et al., 1997) The SF-36 Health Survey (SF-36)

The SF-36 Health Survey is a self-report questionnaire to measure general health status and consists of 36 questions with standardized

et al., 2008). As also this test was computerized, the arrangement of blocks was presented on the computer screen for one second. After a 500-ms blank, the participant was asked to identify the block sequence in the correct order. Set sizes ranged from two to nine digits. After giving a correct answer, another block was added to the test. If a participant repeated the sequence incorrectly twice, the test stopped automatically. The score consists of the highest number of blocks a participant could correctly reproduce.

Digit Span

The Digit Span test was used for measuring span of immediate verbal memory (Shum et al., 1990). The test comprises two different subtests, Digits Forward and Digits Backward. Both tests consist of pairs of random number sequences that were presented on a computer screen, one second per each digit, with a 500-ms blank screen between each digit. In the Digit Forward subtest, the patient had to repeat these digits in the correct sequential order. In the Digit Backward subtest, participants were asked to repeat the digits in reversed order. Set sizes ranged from two to nine digits. After giving a correct answer, another digit was added to the sequence. If a participant repeated the sequence incorrectly twice, the test stopped automatically. Test scores represent the number of correct sequences given.

Eriksen Flanker Task

The Eriksen Flanker Task (Eriksen, 1995) was used part removed. The participant has to choose

which of the six pictures below the item contains the correct pattern. The original Raven’s Matrices consists of 60 items. In this study, three 20-item tests were created, to ensure 3 different sets of items for the three assessment moments. The 20 items consisted of incomplete figures; the missing part is depicted in one of the six alternatives given below the picture. The items get more complex and shift from pattern completion to reasoning by analogy, ranging from quite simple to increasingly difficult items. Internal consistency coefficients tend to cluster around .90 for adults (Llabre, 1984). Retest reliability correlations run in the range of .70 and .90 (Eichorn, 1975; Llabre, 1984), even when retesting is involves administrations six to 12 months apart. Trail Making Test

The Trail Making Test (TMT) was used for assessing divided attention (Strauss et al., 2006). The test is given in two parts, A and B. The subject must first draw lines to connect consecutively numbered circles on one work sheet (Part A), and then connect the same number of consecutively numbered and lettered circles on another work sheet by alternating between the two sequences (Part B) (Lezak et al., 2004). Test administration consisted of the amount of seconds it takes to complete the two separate parts and the number of correct items. Corsi Block-tapping Test

The Corsi Block-tapping Test was used to assess visuospatial short-term memory (Kessels

one game was finished, the next game could be started pressing the ‘next’ button. In each game,

participants began at the same level of difficulty. The difficulty level of each game was raised or lowered depending on the performance in the previous round of the respective game. The software provided feedback to participants as they completed each exercise and after each session. The participants completed the training at home using their own computers and Internet access. Participants were asked to complete five sessions of training per week. Adherence was monitored remotely by visual inspection of the number of times logged in and minutes played on each cognitive domain in on the web site of the study. To promote training adherence, all participants were followed up with weekly telephone calls during training by their personal coach. If a person did not complete the game session, they were asked by e-mail or telephone to complete their session. Participants who were unable to complete a session on one day were instructed to complete an extra session on another day. People received a reminder in their e-mail to improve compliance when they wanted to and could contact their personal coach for any problems or questions about the games. After 8 weeks of training participants were asked to temporarily stop their training for 8 weeks with the aim to measure whether training effects remain after a period of rest. After those 8 weeks non-training (t2) participants could restart their non-training.

for assessment of focused attention. In the flanker task, subjects are asked to identify a central target and ignore the flanking items. In the computer version that was used in this study, the central target stimulus was flanked by either four response-compatible (e.g., ←←←←←) or four response-incompatible (e.g., ←←→←←) stimuli. Participants were instructed to press the ‘C’ button if the central sign pointed to the left and asked to press the ‘M’ button if the sign pointed to the right. The tests consisted of five sets of 30 stimuli presented on the screen. Between each set the participant had a 15 second break. The score of the test is the reaction time in milliseconds. The difference in reaction time for congruent and incongruent items shows how well participant are able to inhibit irrelevant information and focus on the relevant information (focused attention).

INTERVENTION

The intervention of the study consisted of a serious gaming program provided by Lumosity Inc. (intervention group) and reading facts about the brain (control group). The training consisted of gaming during a period of 8 weeks, at least 5 days per week, approximately 15-20 minutes per day. The training software was supplied by Lumosity Inc. A number of sixteen games were selected and five broad cognitive domains targeted: attention, speed, memory, flexibility and problem solving (e.g. Figure 1). Per session three games were randomly assigned to the participants. After

(analysis of variance) was chosen for the independent variable education level, as this could not be taken into the regression analysis. The remaining variables were analyzed using multiple regression analysis. If one or more of the sociodemographic variables showed it had significant value in predicting adherence, these predictors were used in further analysis.

Second, a multiple regression analysis was conducted, to determine what psychological factors predict total playtime best. Logistic regression analysis with a stepwise backward selection procedure was used to build the eventual most parsimonious prediction model. All analyses were performed using SPSS Statistics 22.

DATA ANALYSIS

53 patients in the intervention group finished the complete protocol. On this group, descriptive analyses were performed to describe participant characteristics and adherence.

Adherence was employed as a continuous dependent variable, namely total playtime in minutes. A variety of sociodemographic and psychological variables were taken into account as independent variables. The sociodemographic variables consisted of age, gender, kind of stroke, education level, IQ, living situation and work participation. The psychological variables taken into account were self-efficacy, subjective cognitive impairment, fatigue and quality of life.

Statistical analyses were done in two phases. First, all independent variables were tested for association with playtime. An ANOVA

RESULTS

OVERALL ADHERENCE

Seven of the 53 participants (13 percent) in the intervention group failed to play the Lumosity games, five due to challenging task demands with their computer or Internet access, and two because of serious health problems. The number of days and time played are shown in Table 2. The median of the number of days played was 45 with a range of 4-63, and the total time played in minutes was respectively 562 with a range of 63 till 1264. Total minutes of play time were equally distributed in participants over the period of training, except for three participants

PATIENT CHARACTERISTICS

Of the 53 participants in the intervention group, 50 participants (94%) completed the entire study, including follow-up assessment. Reasons for withdrawal in the treatment group were: illness and experiencing too much stress because of participation in the study. At the first meeting, which is the meeting of interest in this study, all 53 participants were present. The demographic characteristics of the study participants are shown in Table 1. Participant characteristics were similar in intervention and control group at baseline.

stroke, haemorrhage or ischaemic, did not predict adherence either (β = .060, p=.724). Work participation (β = .162, p=.872), independent or dependent living (β = -.241, p=.811), living together or alone (β = .252, p=.234), had no predictive value for adherence.

Additionally to the multiple regression an ANOVA was conducted to identify whether education has an effect on adherence. The main effect of education was not significant F(2,43) = .038, p=0.963). Participants with low, medium and higher education did not differ on total playtime.

As no sociodemographic variable showed to be significant predictors of adherence, none of the variables are taken into account in the following analyses.

Influence of psychological factors on program usage

Multiple regression analysis was used to test if participants’ psychological characteristics

I

Multiple regression analysis was used to test if participants’ sociodemographic characteristics significantly predict adherence to the CBCR program. The results of the regression indicated that none of the sociodemographic characteristics significantly predict adherence (R²=.102, F (7, 33) =.535, p=.802). Data were independently and identically distributed and the assumptions of conditional normality, linearity of the regression relationship, multicollinearity and homoscedasticity were tenable.

Looking into the separate predictors, female participants had a higher play time (M=622.48, SE=63.88), than men did (M=526.22, SE=50.39). However, gender did not significantly predict adherence (β = .120, p=.676). Age was not predictive of adherence (β = -.085, p=.680), nor was IQ estimation (β = .046, p=.827). Kind of

significantly predicted adherence. The results of the backward regression indicated the two predic-tors that explained 18.8% of the variance (R²=.188, F(1,42)=5.56, p<.05). It was found that this model, including quality of life after stroke (SSQoL: β = .559, p<.05) and subjectively perceived cognitive failure (CFQ: β = .346, p=.058), significantly pre-dicted adherence to the training program. Data were independently and identically distributed and the assumptions of conditional normality, linearity of the regression relationship, multicollinearity and homoscedasticity were tenable.

Higher scores on the SSQoL, indicating higher subjective quality of life, lead to higher adherence. Higher scores on CFQ, indicating more subjective cognitive failure, lead to higher adherence. Self-efficacy and fatigue did not show significant predictive value for adherence.

In this study we tried to identify patient characteristics that facilitate adherence to CBCR. The primary finding of this study is that quality of life and subjective cognitive failure of post-stroke patients have a significant influence on adherence to the computer-based cognitive rehabilitation program Lumosity. People that report higher quality of life and more cognitive failure, are most inclined to show compliance and complete the brain training program. Simple demographic

characteristics such as age, gender and educational level are of no predictive value in program adherence, nor are the other psychological factors self-efficacy and fatigue. Quality of life after stroke was administered using a questionnaire that consists of multiple domains, including energy, mobility, mood, thinking and self-care. It could be argued that relatively good quality of life is a necessity to start and follow an intensive CBCR program, which requires self-motivation. When quality of life is low, the energy for self-care and adherence to such a program might just be lacking. The fact that a higher indicated level of cognitive failure leads to better program adherence could be explained by its motivation generating quality: if cognitive impairment is experienced, this could motivate adherence to a cognitive rehabilitation program. This interpretation is in line with previous research that show that motivation is an important patient variable in better treatment and educational outcomes, and also for cognitive rehabilitation (Medalia & Richardson, 2005). Although it was hypothesized that higher self-efficacy would lead to greater adherence to the program, no such result was found in the current study. This is unexpected, as self-efficacy and health outcomes have been shown to be highly correlated (Luszczynska, 2005). A possible explanation for this is that self-efficacy is a well-defined but limited concept.

It has been indicated that stroke patients

and possibly benefit from, a CBCR program like Lumosity. This knowledge is important for future research into the effectiveness of CBCR programmes. As noted earlier, one of the biggest problems in this research area is the high number of non-users which causes studies to be less reliable. To investigate whether CBCR can improve cognitive functioning, this amount of non-usage needs to drop. Moreover, it has recently been shown that treatment success depends strongly on treatment intensity: it even seems that irregular attendance is not much better than no attendance at all. This even further accentuates the need for studies to ensure intervention adherence if CBCR effectiveness wants to be investigated. Future studies into CBCR effectiveness could best focus on the group of post-stroke patients that have been now shown to comply with such an intervention.

As the findings in this study shed new light on the non-usage problem in CBCR, this could lead to better patient selection for CBCR programmes. The high rate of non-usage also shows that not all patients are suitable candidates for CBCR. Now, before submitting post-stroke patients to a CBCR program, a careful consideration should be made on the basis on patients’ subjective quality of life and experienced cognitive impairment. People that have relative good quality of life, but do experience a lot of cognitive impairment, will probably show the highest level of adherence to such a program. This new insight could help rehabilitation have significantly poorer levels of self-concept

across a number of domains (Ponsford et al., 2014) and are found to be more negative and less self-confident (Kravetz, 1995). This drop in self-confidence together with a low self-concept leads patients to perceive themselves as having performance difficulties (Ponsford et al., 2014). This could explain why the more overarching factor of quality of life did show to lead to greater adherence: self-efficacy is an important factor, but quality of life is multidimensional. The factor quality of life takes all inter-relationships into account between self-concept, self-efficacy, self-confidence and mood. It could be argued that not just self-efficacy, but an interacting cluster of factors in self-concept influence adherence to a CBCR program. Furthermore, the hypothesis that fatigue would cause intervention adherence to drop, could neither be supported. This could be explained by the fact that fatigue was only indirectly measured by means of a more overarching questionnaire on general health status.

By choosing a questionnaire that is more specifically designed to measure subjective fatigue and related behavioral patterns, for example the Checklist Individual Strength (CIS), future research could make more reliable claims regarding fatigue and its impact on adherence (Vercoulen et al., 1994; Bültmann et al., 2002).

This study indicates that a particular group

practitioners to choose treatment forms as CBCR for specific patients and hereby cultivate better and more individually suitable rehabilitation programmes.

This study could also inspire alternative usage of CBCR for post-stroke patients that do not show the specific characteristics that lead to adherence. As this study has shown, patients who experience cognitive impairment, yet rate their quality of life positively, are most capable and willing to complete a CBCR training. This leaves a large group of post-stroke patients that do not seem to be capable or motivated to take part in a CBCR program. Even though it seems plausible that this group could not benefit from CBCR, the fact remains that CBCR has clear advantages over more traditional interventions. The cost-effectiveness, accessibility for homebound patients and possibilities for individualized treatment could still show to be beneficial for other post-stroke patients. The absence of a trainer or ‘coach’ has been identified as a limiting factor in CBCR programmes (Connor & Standen, 2012).

Possibly, patients that do not primarily seem to show adherence to CBCR, can profit from extra guidance in form of a ‘coach’ during the training. The potential problems with a lack of supervision are increased in stroke patients given that they often require more structure than healthy adults. This may be particularly true for executive function

impairments that reduce behavioral regulation and rehabilitation participation (Skidmore et al., 2010;

Cumming et al., 2013

Using a coach, or neuropsychological therapy, in combination with training methods seems to be the preferred combination: the therapeutic support leads to better adherence as the therapy helps rebuild quality of life (Åkerlund et al., 2013). If however, a coach will indeed be included as support for the CBCR program, it is advisable to use a coach with the right level of training. It has been previously shown that not only treatment outcome but also treatment adherence is affected by the level of training of a therapist: patients benefit most when trained by doctoral-level staff with a sense of intrinsic motivation and commitment to the training task (Medalia & Richardson, 2005). Several limitations to this study must be pointed out, starting with the most salient one. In light of previous studies in other patient groups several predictors for program adherence and completion were identified. In the current study most of these potential predictors have been tested and studied. One important factor has however not been taken into account. This is the negative influence of comorbidity, additional health problems, on intervention adherence. Several studies show that an important part of non-usage can be explained by comorbidity (Bossen et al., 2013; Björkdahl et al., 2013).

It could be argued that comorbidity has an effect on quality of life, which was taken into account in the current study, but in future research it would be interesting to examine the direct effects

symptoms, improvement in functioning and better quality of life were observed (Cicerone et al., 2011; Sarajuuri, 2005). With this in mind, future research could make an attempt in comparing CBCR training, CBCR training combined with neuropsychological treatment in form of a coach and the more comprehensive-holistic approach in combination with CBCR. In doing so, a clearer understanding of what stroke patients benefit most from what rehabilitation approach, could be achieved.

Given the huge impact of stroke, both the adverse personal consequences as the overarching negative effects in society, the quest for effective and affordable treatment has a high priority. Further identifying predictors of adherence for different patient groups and more research into effectiveness of different rehabilitation approaches could lead to more individualized care and cost-effective treatment.

This study shows that despite all favorable aspects of CBCR, only a distinct group of stroke patients shows adherence to a computer-based cognitive rehabilitation program. As a consequence, it is not advisable to make CBCR the standard rehabilitation practice. Every patient should have access to a personalized rehabilitation program, which could include CBCR, if future research proves the effectiveness of CBCR.

of comorbidity on program adherence.

It would also have been interesting to look at general motivation to take part in a CBCR program and ask for personal expectations. It could be hypothesized that highly motivated participants show better adherence to an intervention program such as CBCR. Another addition to the current study could be an extensive exit questionnaire at the end of the study, including motivational questions and questions regarding the program itself. Participants that encounter problems with the program and playing the games, could become less motivated and thus less inclined to show adherence to the program. Future studies could add these aspects to get a better and more complete understanding of adherence to CBCR.

A question that arises from this study is how patient groups that at first sight seem less prone to complete such a program, can profit from CBCR. As has been noted, treatment adherence and outcome is positively affected by combining neuropsychological therapy with a training method as CBCR. A recent study on TBI patients compared the effectiveness of standard rehabilitation with a comprehensive-holistic neuropsychological rehabilitation, which included both individual and group therapies. This holistic approach emphasizes metacognitive and emotional regulation for cognitive, emotional and interpersonal difficulties. Patients were twice as likely to make clinically significant ains compared to conventional rehabilitation, a reduction in

Aaronson N.K., Muller M., Cohen, P.D.A., Essink-Bot M., Fekkes M., Sanderman R., Sprangers M.A.G., te Velde A., Verrips E. (1998). Translation, Validation, and Norming of the Dutch Language Version of the SF-36 Health Survey in Community and Chronic Disease Populations. J Clin Epidemiol, 51(11), 1055–1068.

Åkerlund, E., Esbjörnsson, E., Sunnerhagen, K. S., & Björkdahl, A. (2013). Can computerized working memory training improve impaired working memory, cognition and psychological health?. Brain Injury, 27(13-14), 1649-1657.

Bandura, A. (1997). Self-efficacy: The exercise of control. Macmillan.

Björkdahl, A., Åkerlund, E., Svensson, S., & Esbjörnsson, E. (2013). A randomized study of computerized working memory training and effects on functioning in everyday life for patients with brain injury. Brain Injury, 27(13-14), 1658-1665.

Bossen, D., Buskermolen, M., Veenhof, C., de Bakker, D., & Dekker, J. (2013). Adherence to a Web-Based Physical Activity Intervention for Patients With Knee and/or Hip Osteoarthritis: A Mixed Method Study. Journal of medical Internet research, 15(10).

Botvinick, M.M., Cohen, J.D., & Carter, C.S. (2004). Conflict monitoring and anterior cingulate cortex: an update. Trends in Cognitive Sciences, 8, 539-546.

Broadbent, D.E., Cooper, P.F., FitzGerald, P., & Parkes, K.R. (1982). The Cognitive Failures Questionnaire (CFQ) and its correlates. British Journal of Clinical Psychology, 21, 1-16.

Bültmann, U., Kant, I. J., Van Den Brandt, P. A., & Kasl, S. V. (2002). Psychosocial work characteristics as risk factors for the onset of fatigue and psychological distress: prospective results from the Maastricht Cohort Study.Psychological medicine, 32(02), 333-345.

Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytic studies. Cambridge University Press.

Cha, Y. J., & Kim, H. (2013). Effect of computer-based cognitive rehabilitation (CBCR) for people with stroke: A systematic review and meta-analysis. NeuroRehabilitation, 32(2), 359-368.

Chen, G., Gully, S.M., & Eden, D. (2001). Validation of a new general self-efficacy scale. Organizational Research Methods, 4, 62-83.

Cicerone, K. D., Langenbahn, D. M., Braden, C., Malec, J. F., Kalmar, K., Fraas, M., Felicetti, T., Laatsch, L., Harley, J. P., Bergquist, T., Azulay, J., & Ashman, T. (2011). Evidence-based cognitive rehabilitation: updated review of the literature from 2003 through 2008. Archives of physical medicine and rehabilitation, 92(4), 519-530.

Claesson, L., Lindén, T., Skoog, I., & Blomstrand, C. (2005). Cognitive impairment after stroke–Impact on activities of daily living and costs of care for elderly people. Cerebrovascular Diseases, 19(2), 102-109. Connor, B. B., & Standen, P. J. (2012). So much technology, so little time: factors affecting use of computer-based brain training games for cognitive rehabilitation following stroke.

Cumming, T. B., Marshall, R. S., & Lazar, R. M. (2013). Stroke, cognitive deficits, and rehabilitation: still an incomplete picture. International Journal of Stroke, 8(1), 38-45.

Jenkinson C, Wright L, Coulter A. (1994). Criterion validity and reliability of the SF-36 in a population sample. Qual Life Res., 3(1), 7-12.

Johansson, B., & Tornmalm, M. (2012). Working memory training for patients with acquired brain injury: Effects in daily life. Scandinavian journal of occupational therapy, 19(2), 176-183.

Kessels R.P., van den Berg E., Ruis C., Brands A.M. (2008). The Backward Span of the Corsi Block-Tapping Task and Its Association With the WAIS-III Digit Span. Assessment 15(4): 426-434.

Klingberg, T. (2010). Training and plasticity of working memory. Trends in cognitive sciences, 14(7), 317-324.

Kravetz, S., Gross, Y., Weiler, B., Ben-Yakar, M., Tadir, M., & Stern, M. J. (1995). Self-concept, marital vulnerability and brain damage. Brain Injury, 9(2), 131-139.

Kueider, A. M., Parisi, J. M., Gross, A. L., & Rebok, G. W. (2012). Computerized cognitive training with older adults: a systematic review. PloS one, 7(7), e40588.

Lezak, M.D., Howieson, D.B., & Loring D.W. (2004). Neuropsychological Assessment (Fourth Edition). OXFORD University Press.

Lincoln, N. B., Brinkmann, N., Cunningham, S., Dejaeger, E., De Weerdt, W., Jenni, W., ... & De Wit, L. (2013). Anxiety and depression after stroke: a 5 year follow-up. Disability and rehabilitation, 35(2), 140-145.

Llabre, M.M. (1984). Standard Progressive Matrices. In D.J. Keyser & R.C. Sweetland (Eds.), Test critiques (Vol. 1). Kansas City, MO: Test Corporation of America. D’Aniello, G. E., Scarpina, F., Mauro, A., Mori, I.,

Castelnuovo, G., Bigoni, M., ... & Molinari, E. (2014). Characteristics of anxiety and psychological well-being in chronic post-stroke patients. Journal of the neurological sciences.

DeRenzi, E., Faglioni, P., & Previdi, P. (1977). Spatial memory and hemispheric locus of lesion. Cortex, 13, 424-433.

Douiri, A., Rudd, A. G., & Wolfe, C. D. (2013). Prevalence of poststroke cognitive impairment South London stroke register 1995–2010. Stroke, 44(1), 138-145.

Eichorn, D. H. (1975). The Raven Progressive Matrices (review). In W.K. Frankenburg & B.W. Camps (Eds.), Pediatric screening tests. Springfield, IL; Thomas.

Eriksen, C.W. (1995). The flankers task and response competition: A useful tool for Investigating a variety of cognitive problems. Visual Cognition, 2, 101-118.

Gramstad, A., Iversen, E., & Engelsen, B. A. (2001). The impact of affectivity dispositions, self-efficacy and locus of control on psychosocial adjustment in patients with epilepsy. Epilepsy research, 46(1), 53-61.

Hackett, M. L., Yapa, C., Parag, V., & Anderson, C. S. (2005). Frequency of depression after stroke a systematic review of observational studies. Stroke, 36(6), 1330-1340.

Hersenstiching (2014). Cijfers over patienten. Hersenstichting.nl. Retrieved March 25, 2014, from https://www.hersenstichting.nl/alles-over-hersenen/ hersenaandoeningen.

Lundqvist, A., Grundström, K., Samuelsson, K., & Rönnberg, J. (2010). Computerized training of working memory in a group of patients suffering from acquired brain injury. Brain Injury, 24(10), 1173-1183.

Luszczynska, A., & Schwarzer, R. (2005). Multidimensional health locus of control: comments on the construct and its measurement. Journal of Health Psychology, 10(5), 633-642.

Medalia, A., & Richardson, R. (2005). What predicts a good response to cognitive remediation interventions? Shizophrenia Bulletin, 31(4), 942-953.

Merckelbach, H., Muris, P., Nijman, H., & de Jong, P.J. (1995). Self-reported cognitive failures and neurotic symptomatology. Personality and Individual Differences, 20, 715-724.

Nys, G. M. S., Van Zandvoort, M. J. E., Van der Worp, H. B., De Haan, E. H. F., De Kort, P. L. M., & Kappelle, L. J. (2005). Early depressive symptoms after stroke: neuropsychological correlates and lesion characteristics. Journal of the neurological sciences, 228(1), 27-33.

Owen, A. M., Hampshire, A., Grahn, J. A., Stenton, R., Dajani, S., Burns, A. S., ... & Ballard, C. G. (2010). Putting brain training to the test. Nature, 465(7299), 775-778.

Paker, N., Buğdaycı, D., Tekdöş, D., Kaya, B., & Dere, Ç. (2010). Impact of cognitive impairment on functional outcome in stroke. Stroke research and treatment, 2010.

Pasquini, M., Leys, D., Rousseaux, M., Pasquier, F., & Henon, H. (2007). Influence of cognitive impairment on the institutionalisation rate 3 years after a stroke.

Journal of Neurology, Neurosurgery & Psychiatry, 78(1), 56-59.

Patel, M., Coshall, C., Rudd, A. G., & Wolfe, C. D. (2003). Natural history of cognitive impairment after stroke and factors associated with its recovery. Clinical Rehabilitation, 17(2), 158-166.

Paul, S. L., Srikanth, V. K., & Thrift, A. G. (2007). The large and growing burden of stroke. Current drug targets, 8(7), 786-793.

Perrig, W. J., Hollenstein, M., & Oelhafen, S. (2009). Can we improve fluid intelligence with training on working memory in persons with intellectual disabilities? Journal of Cognitive Education and Psychology, 8(2), 148-164.

Ponsford, J., Kelly, A., & Couchman, G. (2014). Self-concept and self-esteem after acquired brain injury: A control group comparison. Brain Injury, 28(2), 146-154.

Post M.W.M., Boosman H., van Zandvoort M.M., Passier P.E.C.A., Rinkel G.J.E., Visser-Meily J.M.A. (2009). Development and validation of a short version of the Stroke-Specific Quality of Life Scale. Journal of Neurology, Neurosurgery and Psychiatry 82, 3 (2010) 283.

Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z., ... & Engle, R. W. (2013). No evidence of intelligence improvement after working memory training: A randomized, placebo-controlled study. Journal of Experimental Psychology: General, 142(2), 359.

47. Sarajuuri, J. M., Kaipio, M. L., Koskkinen, S. K., Niemelä, M. R., Servo, A. R., Vilkki, J. S., (2005). Outcome of a comprehensive neurorehabilitation program for

Van Zandvoort, M. J. E., Kappelle, L. J., Algra, A., & De Haan, E. H. F. (1998). Decreased capacity for mental effort after single supratentorial lacunar infarct may affect performance in everyday life. Journal of Neurology, Neurosurgery & Psychiatry, 65(5), 697-702.

Vercoulen, J. H., Swanink, C., Fennis, J. F., Galama, J., van der Meer, J. W., & Bleijenberg, G. (1994). Dimensional assessment of chronic fatigue syndrome. Journal of psychosomatic research, 38(5), 383-392.

Williams, L. S., Weinberger, M., Harris, L. E., Clark, D. O., & Biller, J. (1999). Development of a stroke-specific quality of life scale. Stroke, 30(7), 1362-1369.

patients with traumatic brain injury. Archives of physical medicine and rehabilitation, 86, 2296-2302.

Schwarzer, R., & Jerusalem, M. (1995). Generalized self-efficacy scale. Measures in health psychology: A user’s portfolio. Causal and control beliefs, 1, 35-37.

Sherer, M., Maddux, J. E., Mercandante, B., Prentice-Dunn, S., Jacobs, B., & Rogers, R. W. (1982). The self-efficacy scale: Construction and validation. Psychological reports, 51(2), 663-671.

Shim, H. (2014). Vascular Cognitive Impairment and Post-Stroke Cognitive Deficits. Current neurology and neuroscience reports, 14(1), 1-6.

Sholberg, M. M., & Mateer, C. A. (1998). Introduction to Cognitive Rehabilitation: Theory and practice. New York, NY, Guilford Press.

Shum, D.H.K., McFarland, K.A., & Bain, J.D. (1990). Construct validity of eight tests of attention: Comparison of normal and closed head injured samples. The Clinincal Neuropsychologist, 4, 151-162.

Skidmore, E. R., Whyte, E. M., Holm, M. B., Becker, J. T., Butters, M. A., Dew, M. A., ... & Lenze, E. J. (2010). Cognitive and affective predictors of rehabilitation participation after stroke. Archives of physical medicine and rehabilitation, 91(2), 203-207.

Starkstein, S. E., Mizrahi, R., & Power, B. D. (2008). Antidepressant therapy in post-stroke depression.

Strauss E, Sherman EMS, & Spreen O. A compendium of neuropsychological tests: Administration, norms, and commentary. New York: Oxford University Press; 2006.