Psychometric evidence of the transcultural adaptation of the Vulnerability Abuse Screening Scale ( VASS ) for the detection of violence against the elderly

Objetivo: Apresentar evidencias psicometricas preliminares da adaptacao transcultural da Vulnerability Abuse Screening Scale (VASS). Metodo: Estudo transversal e analitico. Os passos para adaptacao transcultural verificaram a equivalencia conceitual e de itens, a semântica e a de mensuracao. Para a verificacao de equivalencia de mensuracao e dos dados psicometricos do estudo participou uma amostra de 30 e 66 idosos, respectivamente. Para as analises dos resultados foi utilizada estatistica descritiva e inferencial (KR-20, Teste t de Student, correlacao de Pearson, ANOVA univariada e o indice Kappa de Fleiss). Resultados: Verificou-se que a idade dos participantes variou entre 60 e 84 anos, prevalecendo respondentes idosas (n=38). O valor do KR-20 para o escore geral do instrumento foi de 0,688. Os valores encontrados para as quatro dimensoes propostas pelos autores do estudo foram 0,528, 0,289, 0,552 e 0,303, respectivamente. Apenas os valores das subescalas Vulnerabilidade e Coercao mostraram-se aproximados aos do estudo original (0,550 e 0,390). Conclusao: Ressalta-se que os valores de consistencia interna encontrados no presente estudo variaram de moderados a bons, indicando que os resultados sao satisfatorios ainda que iniciais.


INTRODUCTION
Population aging is a global phenomenon.Simultaneously, there is a need to guarantee opportunities for health, social participation and social security in a continuous manner throughout life, with the development of public policies aimed at comprehensive care for the elderly.These should provide assistance in the fields of social security, pensions, health, education and all the other fundamental rights established by relevant legislation 1 .However, comprehensive protection for the elderly is not currently guaranteed, especially in the context of ill-treatment and violence against this population.
Violence is growing worldwide, permeating, in an expressive manner, interpersonal relationships and relations between groups.As a social and systemic phenomenon which is subjective and triggers social contradictions, violence needs to be understood both in its expression and in terms of the impact on those affected by it.The World Health Organization (WHO) promulgated a classification that has been widely used by elderly care services 2,3 , according to which violence against the elderly can be classified as physical, psychological or emotional -which includes verbal violence -sexual, economic or financial, and as neglect and self-neglect 4 .
In addition, despite being a growing phenomenon with increasing social and media visibility, difficulties are still encountered in screening, identifying and preventing violence.Among the reasons for the difficulty in screening, arising from a shortage of information and problems in registering complaints, and which are listed in literature as the main agents that create an underreporting of violence, are a degree of proximity and/or kinship between the aggressor and the victim, or the presence of an affective-emotional, caregiving, or financial type dependent relationship 3,5 .
It is worth noting that violence against the elderly is usually only verified when a complaint is made and verified by means of examinations that assess, in an exaggerated manner, signs and indications of physical ill-treatment only.Factors that suggest the evidence of violence of other kinds, such as psychological or financial aspects and/or neglect and self-neglect, are not considered.
As there is a clear lack of instruments available in Portuguese for the detection of domestic violence against the elderly, it seems opportune and relevant to make available and adapt Lusophone versions of tools derived from consistent research programs 6 .One of the programs recognized for developing a tool for violence tracking 7-9 produced the Vulnerability Abuse Screening Scale (VASS), based on studies by Schofield et al. 10 and Schofield and Mishra 11 , in Australia.This scale is the object of the proposal of the current study.
The VASS instrument was created based on modifications of the Hwalek-Sengstock Elder Abuse Screening Test.The VASS contains twelve dichotomous items that, through the self-reports of the elderly, establish the risk of domestic violence.Data collection is based on the perception that the elderly person has regarding everyday situations that can indicate if they are a victim of violence.The point score of the scale is obtained through the sum of the values assigned to each affirmative answer, except for items 4, 5 and 6, which score in the event of a negative answer.
The items seek to identify the phenomenon through four domains, namely Vulnerability (items Nº 01 to Nº 03), Dependence (Nº 04 to Nº 06), Dejection (Nº 07 to Nº 09) and Coercion (item Nº 10 to Nº 12).These, in turn, try to explain four of the six WHO classifications: physical, psychological and financial violence and neglect.VASS is therefore a short screening tool for abuse of the elderly, which can be used in a clinical and home context, and aims to establish the risk of violence.
The psychometric evidence studies of validity and precision were examined in a longitudinal research process 11 .The psychometric findings of the instrument in question were found to be satisfactory and adequate for the adaptability of the scale to other cultural contexts 6 .
In this context, the present study aims to present preliminary psychometric evidence of the crosscultural adaptation of VASS, which investigates the risk of domestic violence against the elderly.Specific objectives were applied to evaluate the equivalence measurement of the Brazilian version, including: 1) Evidence of validity based on the internal structure of the Brazilian version of the instrument, through analysis of the internal consistency of this version compared with the original instrument; 2) Evidence of apparent or face validity, by measuring the statistical agreement between the evaluation of the judges; 3) An initial evaluation of factors related to evidences of validity based on the correlations (convergence patterns) and differentiations (discriminant patterns) with the variables studied .

METHOD
A study with a methodological, transversal and analytical design was performed.The steps for cross-cultural adaptation followed the proposals of Herdman, Fox-Rushby and Badia 12,13 and used the methodological propositions of other authors about adaptation, such as Beaton, Bombadier, Guillemin and Ferraz, Guillemin, Bombadier and Beaton 14,15 .This model is based on the evaluation of conceptual, item, semantic, operational and measurement equivalence, and has been widely disseminated throughout Brazil, especially in research involving the elderly 16 .Data about the substages of conceptual, item and semantic equivalence can be seen in a recent publication 17 .

Operational equivalence
With respect to the operational equivalence substage, which involves performing a pre-test of the proposed synthesized version, we sought to evaluate the degree of comprehension and acceptability of the Brazilian version of the instrument among a group of specialist judges who had participated in the previous stages.This group was formed by postgraduate students in Psychology, who are familiar with the cross-cultural adaptation method, as well as being knowledgeable in gerontology, human development and instrument validation studies, and a group of elderly people.A population of elderly people with different social, economic and educational characteristics was chosen, so that a final or revised synthesis-version could be confirmed and thus the final formatting of the adapted instrument confirmed.

Measurement equivalence
Next, the measurement equivalence, although partial, of the instrument was undertaken.At this stage, the preliminary psychometric evidence of the instrument was investigated with a larger population of elderly persons, statistically verifying its internal consistency and preliminary validity evidence.

Sample
The samples of the elderly persons participating in the two stages were independent.In other words, the elderly participants of the operational equivalence stage were different from those of the measurement stage.These individuals were selected for convenience from services of health, social care and organized civil society.All were located in the metropolitan area of the municipality of Natal, Rio Grande do Norte.To define the sample participants the suggestion of the authors of the Transcultural Adaptation proposal was considered in the pre-test stage.This recommends the use of around 30 to 40 participants 14 , to delineate the participant sample from the operational equivalence stage of the study.In view of this, 30 elderly people participated in this study.
For the measurement equivalence step, the proposals of five respondents per item, on average, was followed 18 .In the case of the present study, the sample group of this stage should total 60 subjects.In practical terms, however, a total of 66 elderly persons were evaluated, adding 10% to the value suggested above.In both above steps, the sample comprised those aged 60 years or older who consented, in a clear and free manner, to participate in this research, and who had retained their mental functions.Elderly persons who did not meet the previously mentioned criteria and which interrupted the interview because they were emotionally affected by the subject matter were excluded from the study.

Data collection, protocols and instruments
Data was collected through interviews, carried out by two previously trained interviewers with a background in psychology.The data of the operational equivalence stage was collected between November and December 2014.The data for the measurement equivalence stage were obtained in January and February 2015.The study was carried out in health and social care units and organized civil society centers located in the metropolitan region of the city of Natal, Rio Grande do Norte.
The following instruments were used as protocols: a) A sociodemographic questionnaire, which investigated data such as age, gender, place of birth, years of schooling/educational level, marital status, religion, as well as clinical health characteristics and the self-evaluation of perception of interpersonal relationships; b) The Mini Mental State Exam (MMSE), for screening of cognitive impairment, as this instrument is designed for a rapid and practical clinical evaluation of the cognitive state of geriatric patients, for the application of the inclusion criteria of the study; and c) The Vulnerability Abuse Screening Scale (VASS).

Data analysis
Regarding data analysis, the Fleiss' Kappa (k) measurement was used to evaluate the judges of the panel of experts, as well as the elderly persons, regarding the comprehension and acceptability of the items, beyond consensus.This evaluates the reliability of the agreement in the evaluations of the judges and measures the apparent or face validity of the instrument.The interpretation of the Kappa coefficient can be classified as: a) Almost perfect agreement, where values are between 0.81 and 1.00; b) Substantial agreement, where they are between 0.61 and 0.80; c) Moderate agreement, between 0.41 and 0.60; d) Fair, between 0.21 and 0.40; e) Slight agreement, between 0.0 and 0.20 and no correlation, where values are less than zero (0), indicating an absence of agreement 19 .
The data obtained in the interviews was categorized and analyzed through descriptive statistics, including frequency, percentages, central tendency and dispersion, with the aim of characterizing the participant sample and the other quantitative results.In the case of dichotomous items, internal consistency analysis was performed using the Kuder-Richardson reliability coefficient (KR-20) to evaluate the internal consistency of the instrument.Inferential statistics were applied to verify discriminant and convergent patterns of the studied data using Student's t-Test, univariate ANOVA and Pearson's correlation, respectively, which were chosen through the verification of the normality of the data, using the Shapiro-Wilk test.Statistics software was used for the analysis of data.The value of statistical significance adopted was equal to or less than 0.05 (p<0.05).

Ethical aspects
The study was performed following the approval of the Ethics Research Committee of the Universidade Federal do Rio Grande do Norte (Rio Grande do Norte Federal University) (Nº 97.186/2012; CAAE: 05563712.8.0000.5537),and all the participants signed a Free and Clear Consent Form.

RESULTS
Operational equivalence was satisfactory, with an acceptability of 92.5% among the elderly population studied, which was a total of 30 elderly people, aged between 60 and 81 [Mean (M): 68.07,Standard Deviation (sd): 5.687] years.The reliability of agreement between the judges of the stages was 0.625 (non-elderly expert judges) and 0.585 (elderly judges), respectively.In general, the results presented were acceptable, with values varying from moderate to high, suggesting the surface or apparent validity of the instrument.
Regarding the measurement equivalence stage, the age of the participants ranged from 60 to 84 years old, with elderly respondents (n=38), representing 57.6% of the sample studied.The number of people sharing a residence with the elderly ranged from one to seven people (M: 4.32; sd: 1.511).The households that the elderly persons lived in have, on average, four to nine rooms (M: 5.80; sd: 0.980).The other sociodemographic data can be seen in Table 1.
Health data, regarding the hospitalizations, falls and disabilities of the elderly population studied, can be seen in Table 2.
Data of the evaluation of the perception of the relationship of the individual with different people in their social environment can be visualized in Table 3.A minimum variance of zero and a maximum variance of 10 points (M: 3.71; sd: 2.404) were observed for total VASS score.If the scores of international studies are taken as a reference, there is a prevalence of 13.6% of elderly people with scores that may indicate a risk of violence.
The KR-20 value for the overall instrument score was 0.688 (Table 4).The KR-20 values for the four subscales proposed by the authors were 0.528, 0.289, 0.552 and 0.303, respectively.Only the internal consistency values of the Vulnerability and Coercion subscales, 0.550 and 0.390 respectively, were approximate to those found in the original study.The removal of items Nº 04, Nº 06 and Nº 10 resulted in an increase in the internal consistency index of the total scale.In terms of the internal consistency of the subscales, only the removal of items no.09, from the Dejection scale, and no.12, from the Coercion subscale, led to an increase in these values.
Regarding the verification of psychometric properties, the discriminant patterns of the items and external variables were verified in relation to the total VASS score.The Student's t-test was used to verify if there is difference in the total VASS score between certain demographic and social variables, including gender, being alone for most of the day, and hospital admission.Variables such as gender, being alone for most of the day, receiving a pension and/ or benefit, hospitalizations in the last 12 months, having been bedridden in the last 15 days, as with items Nº 01, Nº 03, Nº 06, Nº 07 and Nº 09, did not exhibit significant differences in the total score of the scale.The other variables were significant (Table 5).
There was also evidence of the convergent validity of items and external variables in relation to the total score of the scale, which can be seen in table 5. Pearson correlations were used to verify the covariance of the data of each of the items of the instrument with the total VASS score.As expected, to the extent the items on the scale had a positive score, so there was a tendency towards a significant increase in the total score (p<0.05),except for item no.06, which did not exhibit this behavior (Table 6). to be continued

DISCUSSION
The agreement of the evaluations carried out during the semantic equivalence stage, as verified by the Fleiss kappa measure, was 0.625 and 0.585, which indicates a substantial and moderate agreement, respectively, among the judges 19 .The results of the cross-cultural adaptation of the VASS were satisfactory, indicating the apparent validity of the instrument, which suggests that it is apparently satisfactory for the taking of measurements 20 .
In the measurement equivalence step, a minimum variance of zero and a maximum variance of ten points (mean=3.71;sd=2.404) were observed in the total VASS score.If the score values of international studies are taken as a reference, there is a prevalence of 13.6% of elderly persons with scores that may indicate a risk of violence.The prevalence found here is low compared to the international studies cited, which range from 15% to 35% [21][22][23][24][25] .This prevalence is within the averages found in Brazilian studies, which vary from 0% to 21% [26][27][28] .
As Streiner and Norman point out, internal consistency values on the scale of those found in this study, which showed moderate to good internal consistency, are commonplace for brief screening instruments.Values higher than those of the present study can only be obtained with the addition of items to the instrument 29 .Urbina points out that estimates below 0.70 may suggest that the score derived from the test may not be reliable 30 .Therefore, it can be inferred that the increase of internal consistency indices due to the withdrawal of an item may indicate that the test may not be precise for the measurement of the desired construct.
The previously described result (Table 3) seems to indicate that the items with significant results (Nº 02, Nº 04 Nº 05, Nº 08, Nº 10, Nº 11 and Nº 12) are more sensitive for the detection of the phenomenon in question.On the other hand, the non-significance of the difference of the scores of the five other items (Nº 01, Nº 03, Nº 06, Nº 07 and Nº 09) may indicate the reverse (Table 1).
For Cohen, these results emphasize the care the interviewer must take regarding false negative or positive results which can exist among respondents of the screening scale in question, especially due to the number of items on the scale.The author concludes that this type of tool should be used with parsimony and only as a preliminary step in the screening and identification of cases that merit indepth investigation 7,8 .
As expected, to the extent the items on the scale had a positive score, so there was a tendency towards a significant increase in the total score, except for item no.06, which did not exhibit this behavior.The same behavior was observed for the subscales and the total score.This covariance behavior is It should be noted that falls correlated negatively with the score, although in a very weak manner.Such behavior diverges from the findings in literature, as falls have been related to the phenomenon of violence and/or mistreatment of the elderly 31 .
The main limitation of the study is the lack of a gold standard instrument to test the reliability, sensitivity and specificity of the instrument in question.Despite this limitation, the cross-cultural adaptation and verification of the preliminary psychometric properties of the instrument as a selfreporting measure that addresses the indication of domestic violence against the elderly is relevant, and the results were satisfactory.

CONCLUSION
It should be noted that these are preliminary results.Following the verification of suitability and the initial psychometric patterns regarding the use of the instrument for the elderly population, there remains a need to continue with the verification of its robust psychometric properties 32,33 .This can be carried out, for example, through analysis of the evidence of reliability in a test-retest situation, and the validity of the construct and criterion.It should be noted that the internal consistency values found in the present study ranged from moderate to good, indicating that the results are satisfactory, even if initial.Finally, it should be emphasized that when the instrument in question becomes available for practical use, it should not be used in isolation in the investigation of violence against the elderly.Instead it should be contextualized and comprise part of an evaluation protocol, which brings together various procedures, instruments and technologies to assist in the detection of violence against the elderly.

Table 1 .
Sociodemographic data of the population of elderly persons studied, Natal, Rio Grande do Norte, 2015.

Table 2 .
Aspects related to the hospitalization, falls and disabilities of the population of elderly persons studied, Natal, Rio Grande do Norte, 2015.
Created by authors.

Table 3 .
Evaluation of perception of relationships with different people in the social environment of the elderly person, Natal, Rio Grande do Norte, 2015.
Created by authors.

Table 4 .
Reliability: internal consistency of Brazilian version of Vulnerability Abuse Screening Scale (VASS) instrument.Natal, Rio Grande do Norte, 2015.

Table 6 .
Result of significant correlations between sociodemographic variables, subscales and items of instrument with the total score of the Vulnerability Abuse Screening Scale (VASS).Natal, Rio Grande do Norte, 2015.

Table 5 .
Mean of Vulnerability Abuse Screening Scale (VASS) and significance of difference with variables studied.Natal-Rio Grande do Norte, 2015.
26 mai 2015];57(3):243-49.Disponível em: http://www.sciencedirect.com/science/article/pii/S0167494313001064 logical and expected.However, the low values, or the nonexistence of a correlation, other than between items Nº 01, Nº 03 and Nº 09, highlights the care that must be taken with the results of the scale, regarding the possibility of false negative and/or positive results 7,8 .