ASSESSING MOTOR NEUROLOGICAL SOFT SIGNS IN SCHIZOPHRENIA: CONFIRMATORY FACTOR ANALYSIS OF PORTUGUESE VERSION OF BMS-P

Motor Neurological Soft Signs (MNSS) has been increasingly studied in people with schizophrenia spectrum, contributing to the knowledge of psychomotor disorders. However, there is still scarce the availability of valid and reliable instruments for this assessment worldwide, and in Portugal there is none. Objective: The aim of this study was to analyze the factor structure model of the Portuguese version of the Brief Motor Scale. Method: The confirmatory factor analysis was conducted in a sample of 103 participants with schizophrenia, 82 males and 21 females, mean age 44.21 (+11.25). A one-factor model was compared with three two-factor models based on the literature and a previous exploratory factor analysis. Results: The results of this study showed good fit indices and provide evidence to support a two-factor model for the assessment of motor neurological soft signs in persons with schizophrenia. Conclusions: It was proposed a new construct for MNSS assessment consists of 2 domains and 9 items distributed among them.


INTRODUCTION
Motor Neurological Soft Signs (MNSS) are the motor components of Neurological Soft Signs (NSS) and they have been increasingly studied in persons with Schizophrenia Spectrum, contributing to the knowledge of the psychomotor disorders (Afonso et al., 2021;Guimarães et al., 2017Guimarães et al., , 2018Guimarães et al., , 2020;;Janssens et al., 2017;Zakaria et al., 2013) that seems to precede and predict the onset of the disorder.
MNSS have been described as having two components: motor coordination and motor sequentialization, indicators of neurological abnormalities in persons with schizophrenia (Bachmann et al., 2014).These both domains are considered valid indicators of neurological deficits (Jahn et al., 2006a, b, Zakaria et al., 2013), referring the existence of minor brain lesions, namely in the prefrontal areas of the cortex (Bombin et al., 2005;Gil, 2014;) and sensory-motor areas (Chan et al., 2006).Chan et al. (2006) found that sensorimotor cortex, thalamus, and right-side cerebellum have more activity in more complex motor tasks.Neuroimaging studies suggest that the frontal lobe is associated with motor coordination tasks, and the prefrontal lobe with motor sequencing tasks (Bombin et al, 2005).The screening of MNSS may be useful in psychomotor planning, monitoring effects and tracking outcomes.
Historically MNSS have been studied within the larger evaluation of neurological soft signs (NSS) corresponding only to a one or two factors in those scales.All the NSS include motor tasks to measure the MNSS and distribute them in a nonconsensual way in several domains.Various motor tasks are assessed, but the motor domain allocated to the tasks is frequently different (Table 1).In the Neurological Evaluation Scale -NES (Buchanan;Heinrichs, 1989) only Pearson's correlation was performed and findings pointed out that all domains were related to the total and none were correlated with each other.The Heidelberg NSS Scale (Schröder et al., 1992) revealed a Cronbach's alpha of .85,interrater reliability of .88(p<.005) and a six-factor principal component factor analysis (rotated by VARIMAX), that explained 63% of the total variance (Schröder et al., 1992).The reliability and validity of the Cambridge Neurological Inventory -CNI (Chen et al., 1995) were addressed by interrater reliability with Kendall's w between .82 and 1.00 (p<.05) and Pearson's correlation between groups with significant correlation between almost groups (Chen et al., 1995).The Standardized Examination Assessing Neurological Soft Signs -SEA (Krebs et al., 2000) demonstrated a Croanbach's Alpha of .85,and a five-factors principal component factor analysis (rotated by VARIMAX) which explained 55.15% of total variance (Krebs et al., 2000).4 longer versions of soft neurological signs assessment, but restricting to specific items, making this a quick and easy evaluation.BMS proposed 10 motor tasks (items) distribute by two motor domains (Guimarães et al., 2018;Jahn et al., 2006a): motor coordination (Diadochokinesia, Oseretzky, foot taping, bilateral rhythm tapping, gaze impersistent) and motor sequentialization (pronation-supination, finger-thumb-opposition, fist-ring, fist-edge-palm, rhythm production).Motor coordination assessment requires motor planning and generalization, which allows the person to choose the best position and speed for the movement (Fonseca, 2010).In motor sequencing, movements evaluated are those require planning and praxis learning (Guimarães et al., 2018;Jahn et al., 2006a,b;Zakaria et al., 2013).In the original BMS, Cronbach's alpha was .77,Cohen's kappa for interrater reliability was greater than .70,and the two subscales were confirmed by principal component analysis and explain 53.1% of the total variance (Jahn et al., 2006a).
Validation of assessment instruments using statistical measures that are considered robust from an international scientific perspective is essential and crucial to its use for correct, valid, and scientifically based therapeutic purposes.In Portugal, the BMS-P was translated, adapted, and validated to Portuguese persons with schizophrenia, with content validity values >.75, reliability between .82 and .87,and Pearson correlations between scale items between .93 and .99,for p<.01 (Guimarães et al. 2017(Guimarães et al. , 2020)), indicating that can be a reliable instrument to be used.However, a precise validation of the instrument and its theoretical construct is required.
Therefore, the aim of this study was to analyze the factor structure model of the Portuguese version of the Brief Motor, obtained through CFA.The BMS-P assumed the same model and organization (items and subscales) than the original version (Guimarães et al, 2017(Guimarães et al, , 2020)).Based on previous EFA which pointed out two principal components that explain 58.91% of the scale variance, a model with two principal components but with different observed variables for each factor was hypothesized (Guimarães et al., 2020).This study aims to contribute, nationally and internationally, to the theoretical aspects of the assessment of MNSS in people with schizophrenia, and to bring evidence in the areas of psychomotricity that will contribute to a better and more effective psychomotor assessment and, consequently, to the validation of psychomotor intervention programs.

PARTICIPANTS
A convenience sample included 103 participants with a psychiatric diagnosis of schizophrenia according to the DSM 5 (APA, 2013), 82 males and 21 females, aged between 19 and 69 years (44.21+11.25).More than half of participants studied at least 9 years (n=63; 61.2%), 26 (25.2%) studied between 9 and 12 years and 14 (13.6%) had a university degree, and only 29.13% had a professional or academic activity.All participants had a diagnosis of schizophrenia from at least one year ago, being (26.21%) of the participants attended both types of structures (socio-occupational and residential).Only 9 (8.74%) had occasional support in this type of community structure.All of participants were under medical care from a psychiatrist.The exclusion criteria were the presence of a condition of substance abuse or any motor impairment.

BMS-P (PORTUGUESE VERSION OF BMS)
The Portuguese version of Brief Motor Scale (BMS-P) is the translated version of instrument developed originally in Germany, which evaluates exclusively the MNSS of adults with schizophrenia spectrum disorders, in two dimensions: motor coordination (MOCO) and motor sequencing (MOSE) (Guimarães et al., 2018;Jahn et al, 2006a;Jahn et al, 2006b).Both versions, the original and the Portuguese, have been demonstrated to be a reliable instrument with identical psychometrics characteristics to other extensive versions (Guimarães et al., 2020;Jahn et al, 2006a;Jahn et al, 2006b, Zakaria et al, 2013).Each subscale consists of five items, with a maximal summary score for each subscale of 10 points (Guimarães et al., 2020).Each item is classified according to the quality of performance: 0 = no disturbance or difficulty in movement, 1 = some difficulties or disturbances in movement, 2 = greater difficulty or disturbance of the movement.The MOCO and MOSE value are obtained by the arithmetic mean from the correspond items, and the total BMS-P score is the arithmetic mean from the MOCO and MOSE factors (Guimarães et al., 2018).

PROCEDURES
The study was carried out in compliance with the Helsinki Declaration, and the Ethic Committee of Faculdade de Motricidade Humana of Universidade de Lisboa approved all design and materials for Investigation (Ethic approved no.32/2017).All participants signed a written informed consent statement, explaining the study goals and procedures, and ensuring the data confidentiality and participant's anonymity.The BMS-P was applied by qualified psychomotor therapists, who had been previously trained for the scale administration.Training sessions were performed to practitioners who agreed to contribute as applicants.The BMS-P should be administered individually in a quiet room without distractions.Materials needed are a table, two chairs, a pen or pencil, and a sheet of paper.Each session takes 20 minutes.

DATA ANALYSIS
Confirmatory factor analysis (CFA) is the preferred approach to formally test a scale's dimensionality when existing theory and empirical evidence support a particular latent structure of the data, as well as the interrelationships between dimensions (Collier, 2020).Thus, CFA, as model basedapproach, were used to test the latent structure of BMS-P and to confirm if data support the scale's The first phase of research involved examining the first-order measurement model of BMS-P (Model 1).On the second phase, three second-order models were proposed: Model 2 for the original BMS (Jahn et al., 2006a); Model 3 based on previous EFA (Guimarães et al, 2020) that hypothesized BMS-P had two main factors (latent variables), like the original BMS, but different items (observed variables) in each one; and Model 4 based to on previous EFA but without Gaze Impersistent item, because in previous statistic tests it was showing only a weak correlation with one item and no correlation with other items, domains or total score (Guimarães et al., 2020).Models' qualities were evaluated using the Chi Squares Test (ꭕ 2 /df), Comparative Fit Index (CFI), Goodness-of-Fit (GFI), Adjusted Goodness-of-Fit (AGFI), Normed Fit Index (NFI), Root Mean Residual (RMR), Root Mean Square Error of Approximation (RMSEA).The model was considered to have acceptable fit if Chi Squares Test was under 5, RMR and RMSEA were less than .05,CFI and NFI higher than .90(Collier, 2020) and GFI equal or higher than .93when N>100 (Cho et al., 2020).

RESULTS
Concerning to the comparison the models it was listed the fit indices for each model (Table 2).
Model 1, Model 3, and Model 4 were the most acceptable, nevertheless, and due the knowledge of the literature, Models 3 and 4 were the first choice, with better adjustment of the Model 4.  EFA it did not correlate with almost items and it proved to be a weak measure (Guimarães et al., 2020).
In addition, Average Variance Extracted (AVE) and Composite Reliability (CR) were calculated and found to be AVE=.44 and CR=.82 for motor coordination and AVE=.48 and CR=.78 for motor sequencing.According to Forrell and Lacker (1981), an AVE below .50 could be considered if CR is above .70.
The most recent scale -Evaluation Psychomotrice des Signes Doux -EPSiD (Marionneau; Servant; Albaret, 2016) do not present any statistical data.Despite of the importance of MNSS, that tend to reflect a vulnerability for schizophrenia, all the above scales are besides not focused exclusively on motor skillsmaking use of a more broader assessment methodology, are extensive tests, that presents methodological limitations offering itemclusters/subscales not confirmed, since they are mainly based on correlation measures and Cronbach's alpha.Only two of them used Principal Component Analysis Factory for an exploratory analysis.The hierarchical structure of such scales was not analyzed yet.Having a robust and sound psychometric instrument is therefore emergent.In 2006, Jahn et al. (2006a) developed a specific instrument for assessing exclusively the motor neurological soft signsthe Brief Motor Scale (BMS).Originally, this instrument result from previous RECIMA21 -REVISTA CIENTÍFICA MULTIDISCIPLINAR ISSN 2675-6218 ASSESSING MOTOR NEUROLOGICAL SOFT SIGNS IN SCHIZOPHRENIA: CONFIRMATORY FACTOR ANALYSIS OF PORTUGUESE VERSION OF BMS-P Susana Guimarães, Ana Sofia Santos, Ana Paula Lebre dos Santos Branco Melo RECIMA21 -Ciências Exatas e da Terra, Sociais, da Saúde, Humanas e Engenharia/Tecnologia v.4, n.9, 2023

5
medicated with atypical antipsychotics (53), typical and atypical antipsychotics (35), or receiving monthly electroconvulsive therapy treatment (3).At overage, the participants had 2.75 (+1.89) hospitalizations, and the mean age of onset was 20.21 (+5.78).Participants were recruited from rehabilitation community centers, according to the following inclusion criteria: adults (over 18 years) with a diagnosis of schizophrenia.62 (60.19%) of the participants frequented a socio-occupational structure, 5 (4.85%) a residential structure, and 27 . In particular, the research question is whether the data support the theoretical 2-factor second-order structure, that is, whether MNSS can be theoretically explained by two dimensions: motor coordination and motor sequencing.Statistical analysis was conducted using the IBM Statistical Package for the Social Sciences (SPSS) Statistics 28 to presented descriptive statistics, means and standard deviation to characterize the sample, and the IBM SPSS Amos 28 to run Confirmatory Factorial Analysis (CFA).A p-value less than .05was considered statistically significant.

Figure 1
Figure 1 is referred to as a path diagram and provides a visual representation of the theoretical relationships between the observed and latent variables.The criteria for keeping the items were the values above .40and a statistically significant (p<.01).Item 10 was deleted because in the previous

Table 1 .
BMS-PT motor tasks and their distribution by domains in the Neurological Soft Signs scales (authors' summary)

Table 2 -
Fit Indices from Confirmatory Factor Analysis Models