Validation of the Finnish version of the Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS) and evaluation of the applicability of the Multiple Sclerosis Neuropsychological Questionnaire (MSNQ) and the Fatigue Scale for Motor and Cognitive Functions (FSMC)

Abstract Objectives Cognitive impairment is frequent in multiple sclerosis (MS) as approximately half of the patients manifest some degree of cognitive impairment. The Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS) has been designed for brief cognitive evaluation. The purpose of the study was to validate the BICAMS along with the Finnish versions of one self‐rating questionnaire each for cognition and fatigue. Methods A total of 65 MS patients and 45 healthy controls (HC) were assessed with the BICAMS, the Multiple Sclerosis Neuropsychological Questionnaire (MSNQ), and the Fatigue Scale for Motor and Cognitive Functions (FSMC) twice, approximately within nine days. Results MS patients scored markedly lower than the HCs on each of the three tests of the BICAMS. Of the patients, 60% scored at least 1.5 SD below the mean of the HCs on at least one test; 49% on the SDMT, 26% on the CVLT‐II, and 28% on the BVMT‐R. Correlation coefficients for the repeated measurement were between 0.75 and 0.89 for the three tests in the whole study sample. MS patients reported more cognitive symptoms and more fatigue than the HCs. Cronbach's alpha was 0.94 for the MSNQ and 0.98 for the FSMC. Correlation coefficient for the repeated measurement was 0.91 for the MSNQ and between 0.92 and 0.94 for the FSMC scores for the whole study sample. Conclusions The present study supports the validity of the Finnish version of the BICAMS. The SDMT was the most sensitive of the three BICAMS tests and showed cognitive impairment in half of the patients. The Finnish versions of the MSNQ and the FSMC proved useful tools in approaching concerns related to cognition and fatigue.


| INTRODUC TI ON
Cognitive deficits are a common manifestation in multiple sclerosis (MS) occurring in about 50%-60% of patients (Sumowski et al., (2018)). Slowed information processing as well as memory and learning dysfunction are regarded as the most frequent cognitive deficits (Benedict et al., 2020;Sumowski et al., 2018). Slowed information processing speed in particular is thought to be the core feature of cognitive decline in MS. The functional consequences of MS-related cognitive impairment can be striking (Hämäläinen & Rosti-Otajärvi, 2014). Cognitive deficits may have effects on physical independence, quality of life, employment, social and recreational activities, driving skills, and rehabilitation outcome, as well as on caregiver strain (Benedict et al., 2020). Since cognitive deficits can have a multidimensional impact on patients' activities of daily living, these symptoms should be considered in the diagnostics and treatment.
Despite the high frequency and obvious negative impact on functioning, cognitive impairment often remains undiagnosed; invisible symptoms, especially mild cognitive impairments, are not observed during routine neurological examinations. To improve the detection of cognitive impairments and to make follow-up easier, brief assessment tools have been suggested for routine use. An international expert committee agreed on a short battery, the Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS), which is considered a valid and reliable measure of cognitive functioning in MS when comprehensive neuropsychological assessment is not available . The BICAMS includes the Symbol Digit Modalities Test (SDMT) (Smith, 1982) evaluating information processing speed, the California Verbal Learning Test (CVLT) (Delis et al., 2000) evaluating verbal memory and learning, and the Brief Visual Memory Test-Revised (BVMT-R) (Benedict et al., 1997) evaluating visual memory and learning.
In clinical practice, self-reports provide an important source of information on subjective symptoms. The Multiple Sclerosis Neuropsychological Questionnaire (MSNQ) (Benedict et al., 2013) has been used to assess cognition-related concerns. Furthermore, the Fatigue Scale for Cognitive and Motor Functions (FSMC) (Penner et al., 2009) offers a possibility to not only evaluate subjective overall fatigue but also the cognitive and motor components of the symptom. Self-reports are valuable especially in cases where objective assessment is not available, and they serve as a way to approach a delicate topic. However, self-reports are vulnerable to different sources of errors and require validation before use in new populations and as new translations. Whereas self-perceived cognitive and fatigue symptoms have been found to be associated with depression scores, controlling for mood state is necessary.
The primary objective of the present study was to evaluate whether the Finnish BICAMS is a valid measure of cognitive status in MS by employing the validation procedure suggested by Benedict et al. (2012). The secondary aim was to evaluate the applicability of the Finnish translation of the MSNQ and the FSMC in patients with MS.  (Polman et al., 2011). The other inclusion criteria were age between 18 and 65 years, Finnish as a native language, adequate visual acuity, and audition based on interview, no reported alcohol or substance abuse, no other neurological illness except MS, no severe psychiatric illness, no primary learning disability, and no relapse during one month prior to the study. The medical records of patients with MS were prescreened for inclusion criteria. After this prescreening, 73 patients with MS were informed of the study and eight of them refused to participate, mainly due to unwillingness to be assessed. HCs were recruited from the personnel of the rehabilitation center as well as their relatives and friends following the inclusion and exclusion criteria described except for those related to multiple sclerosis. A total of 65 HCs were given the study information; five of them did not fulfill the inclusion criteria, 14 refused to participate mainly due to unwillingness to be assessed, and one was not willing to continue after the baseline assessment. Age, gender, educational degree, | 3 of 9 HÄMÄLÄINEN Et aL.
All the participants performed the BICAMS tests and completed the questionnaires on cognition, fatigue, and mood twice, at baseline and after approximately nine days later.

| SDMT
The Symbol Digit Modalities Test (SDMT; 5) measures the speed of information processing. The test consists of a sheet with nine symbols presented in pseudo-randomized lines. Each symbol is paired with a digit 1-9 in a key at the top of the sheet. The participant is asked to pair, in order, as many of the symbols to the corresponding digits as they can in 90 s. The existing Finnish version of the test and the instructions was employed, and same version of the test was used in both assessments. The number of orally given correct answers during 90 s served as the dependent variable.

| CVLT-II
The California Verbal Learning Test II (CVLT-II; 6) measures verbal learning. The immediate recall consists of five learning trials of a word list of four words each in four semantic categories. The examiner reads the words aloud at a steady pace during 20 s. The participant listens to the complete list and is asked to recall as many words as possible in any order. The list of 16 words had previously been adapted and standardized into Finnish (Vuorivirta, 2006). The same test version was used in both assessments. An alternate version of the CVLT-II is not available in Finnish. The dependent variable was the total number of words recalled during the five trials.

| BVMT-R
The Revised version of the Brief Visuospatial Memory Test (BVMT-R; 7) measures visual learning. The test consists of six abstract symbols on a sheet of paper. Participants are given 10 s to look at the symbols and are then asked to draw as many symbols as they can recall in the right order on an empty sheet of paper. Performance is scored on accuracy and location with 0-2 points per symbol. The task is repeated three times. For the present study, the existing Finnish version of the test and the instructions were employed. There are several alternative forms of the test; version 1 was used during baseline and version 2 during retest. The sum score on the three trials served as the dependent variable.

| 2.3. Self-rating questionnaires
Subjective cognitive complaints were assessed by using the Finnish version of the MSNQ (Benedict et al., 2013), which consists of 15 questions assessing cognitive restrictions with the scale ranging from 0 (never) to 4 (frequently). The total score served as the dependent variable. Subjective feelings of fatigue were evaluated with the FSMC (Delis et al., 2000). The questionnaire consists of 20 statements related to motor and cognitive aspects of fatigue with the scale ranging from 1 (totally disagree) to 5 (totally agree). The total score as well as the sub-scores for motor and cognitive fatigue served as the study variables. Mood was assessed CES-D questionnaire (Radloff, 1977).

| Statistical analyses
Groups were compared with the Mann-Whitney U test and the Wilcoxon test for continuous and ordered variables and the chisquare test for binary variables. Results were considered statistically significant when p <.05, without correction for multiple testing.
Group differences were quantified using the Common Language Effect Size statistic (CLES; McGraw & Wong, 1992) and Cohen's d (Cohen, 1988). Relationships between the study variables and testretest reliability were evaluated with Spearman rank order correlations. The test-retest reliability was considered acceptable when the correlation coefficient was greater than 0.70. Performance on individual tests was considered impaired if at or below the −1.5 SD level of the HC distribution (Sumowski et al., 2018). Overall cognitive performance was defined as impaired if performance at least on one test of the BICAMS was impaired. The internal consistency of the MSNQ and the FSMC questionnaires was evaluated with Cronbach's alpha, with 0.70 considered acceptable. Statistical analyses were performed with IBM SPSS 24.0.

| Results
The background variables of the study groups are reported in Table 1. The mean interval of the baseline and the retest was 9.0 (SD 3.4) days. The study groups were statistically similar with respect to gender, age, and years of education, but differed in employment status and self-rated mood state. Based on the PREDSS, 28% of the patients had mild to moderate disability (EDSS 0-3), 61% severe disability (4-6.5), and 11% were restricted to a wheelchair (Benedict et al., 1997;Benedict et al., 2013;Penner et al., 2009). A majority of the patients (62%) had a relapsing-remitting and a minority (38%) a progressive form of the disease. There were no missing values in the data.
The MS patients scored significantly lower than the HCs on each single test of the BICAMS both at the baseline as well as at the retest ( Table 2). The between-groups Cohen's ds were from 0.69 to 1.20 showing medium to very large effect sizes (Cohen, 1988). Both groups showed practice effects as the performance at the retest exceeded that observed at the baseline.
At baseline, 60% (39/65) of the patients were impaired on at least one of the three BICAMS tests. Of the patients, 29% (19/65) showed impaired performance on one test, 19% (12/65) on two tests, and 12% (8/65) on all three tests. The SDMT was the most sensitive test of the BICAMS as almost half of the patients had impaired performance (Table 3).

MS patients mean (SD)
Healthy controls mean ( Correlations between the study variables, the BICAMS, the MSNQ, the FSMC, and the CES-D are presented in showed impairment on at least one of the tests of the BICAMS battery and 50% (20/40) specifically on the SDMT. Correlation between the FSMC scores and the CES-D total score is presented in Table 5.
Test-retest reliability results are reported in

| D ISCUSS I ON
The aim of the present study was to validate the BICAMS in a Finnish population with MS by employing the validation procedure suggested by Benedict et al. (2012). The secondary aim of the study was to evaluate the applicability of the Finnish version of the MSNQ (Benedict et al., 2013) and the FSMC (Penner et al., 2009)  Both groups showed practice effects on the tests of the BICAMS. The performances at the retest exceeded those observed at the baseline in both groups. The differences in practice effects

TA B L E 4
The results of MS patients and HCs on the MSNQ and the FSMC during baseline and retest (SD = standard deviation) between the groups were small. The same versions of the SDMT and the CVLT-II tests were used for the repeated measurements.
Instead, parallel versions were used for the BVMT-R. The practice effects can be suggested to be more evident when same test version is repeated than when parallel versions are applied. This was also the case in the present study the difference on the SDMT being 3.9 points for the MS group and 4.9 for the HCs, on the CVLT-II 8.7 and 9.5 points, and on the BVMT-R 1.6 and 0.6 points, respectively. In an Italian study by Goretti et al. (Goretti et al., 2014) with a sample of 243 HCs tested twice, the baseline performance was slightly better than in the HCs of the present study. The finding is probably due to the fact that their patients were 11 years younger and slightly more showed the impairment rate of 58% (Dusankova et al., 2012;Walker et al., 2016), Irish 57% (O'Connell et al., 2015), and Hungarian 52% (Sandi et al., 2015). The Finnish version of the BICAMS seems to tap MS-related cognitive impairment at a satisfactory level and, thus, can be considered as a useful and valid measure to identify MS patients who may have cognitive impairments.
From the three single tests of the Finnish version of the BICAMS, the SDMT was the most sensitive followed by the BVMT-R and the CVLT-II, showing impairment rates of 49%, 28%, and 26%, respectively. O'Connell and colleagues (O'Connell et al., 2015) reported an impairment rate of 37% for the SDMT, 10% for the BVMT-R, and 40% for the CVLT-II using the same criteria as used in the present study.
Polycroniadou and colleagues (Polychroniadou et al., 2016) reported an impairment rate of 43% for the SDMT, 22% for the BVMT-R, and 20% for the CVLT-II using the 5th percentile as a cut-off score. Our results corroborate the earlier findings on the sensitivity of the SDMT.
The SDMT has been suggested as the most sensitive single task to tap MS-related cognitive deficits, especially those related to processing slowness López-Góngora et al., 2015).
The test-retest reliability of the BICAMS was evaluated with the correlation coefficients. For the SDMT as well as the CVLT-II, the correlations for the whole study sample as well as for the MS group were > 0.80 indicating good test-retest reliability. For the BVMT-R, the correlation was > 0.70 showing adequate test-retest reliability. These results are in line with the findings from the other BICAMS validation studies in which the correlations for the SDMT and the CVLT-II have been higher than those for the BVMT-R (Filser et al., 2018;Goretti et al., 2014;Niino et al., 2017;Walker et al., 2016). Our results also show that the translation and adaptation of the California Verbal Learning Test into Finnish is appropriate and has a good test-retest reliability.  Unsurprisingly, MS patients reported significantly more cognitive complaints than the HCs on the MSNQ. Altogether 46% of the patients reported subjective cognitive complaints with a total score equal to or over 27 points (Benedict et al., 2008). To compare, 63% of them showed impairment on at least one of the tests of the BICAMS battery and 57% specifically on the SDMT. A third of the patients who reported cognitive complaints did not show impairment on any of the BICAMS tests. The MSNQ showed high internal consistency. The correlation between the total score of the MSNQ and the SDMT was negative and statistically significant, whereas the correlation between the MSNQ and the CVLT-II, and the BVMT-R were statistically non-significant. Instead, correlations between the total score of the MSNQ and the total score as well as sub-scores of the FSMC, and the CES-D were all statistically significant, supporting the earlier findings that low mood state and other symptoms, like fatigue may explain patients' cognitive complaints (Benedict et al., 2008;O'Brien et al., 2007).
The test-retest reliability of the MSNQ was good, the correlations for the whole study sample as well as for the MS group and HCs separately being > 0.80, as observed also previously (Benedict et al., 2008;Morrow et al., 2010). The results of the present study confirm the earlier findings that the MSNQ score is related to the elevated scores in depression questionnaires (Benedict et al., 2008;O'Brien et al., 2007) and, thus, should be used together with an evaluation of mood state. The MSNQ might better serve as a tool to approach this delicate topic than as a screening instrument for cognition per se.
MS patients reported significantly more fatigue than HCs on the FSMC. Altogether, 69% of the patients reported at least mild overall fatigue (Penner et al., 2009) (Vuorivirta, 2006) with good test-retest reliability.
In the validation, we used the same versions of the SDMT and the CVLT-II during baseline and retest as suggested in the original validation procedures . To evaluate how similar the parallel forms of the BVMT-R are, two different versions were used.

ACK N OWLED G M ENTS
We thank the voluntary patients with MS and the healthy controls for participation in the study. We also thank statistician Matias Viitala for his contribution to statistical analyses. The study was financially supported by Novartis AG, Finland which is gratefully acknowledged. The funder had no role in or influence on any other aspect of the study, including data collection, data analysis, writing of the manuscript, and decision to publish.

CO N FLI C T O F I NTE R E S T
None of the authors have any conflict of interest related to this study. P Hämäläinen belonged to the International Committee which originally suggested the BICAMS to be used in cognitive evaluation of patients with MS.

AUTH O R CO NTR I B UTI O N
Hämäläinen involved in planning the study, acquiring the funding, administration of the project, permission for the validation from the test publishers and authors, supervision, analyzing the data, and writing and editing the manuscript. Leo involved in data collection, data curation, statistical analysis, formatting the tables, and editing the manuscript. Therman involved in statistical analysis, formatting the tables, and editing the manuscript. Ruutiainen involved in planning the study, neurological expertise, and editing the manuscript.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.2087.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data collected in the present study can be shared only if permitted by the local ethics committee.