Scale of attitudes toward alcohol - Spanish version: evidences of validity and reliability 1

ABSTRACT Objective: validate the Scale of attitudes toward alcohol, alcoholism and individuals with alcohol use disorders in its Spanish version. Method: methodological study, involving 300 Colombian nurses. Adopting the classical theory, confirmatory factor analysis was applied without prior examination, based on the strong historical evidence of the factorial structure of the original scale to determine the construct validity of this Spanish version. To assess the reliability, Cronbach’s Alpha and Mc Donalid’s Omega coefficients were used. Results: the confirmatory factor analysis indicated the good fit of the scale model in a four-factor distribution, with a cut-off point at 3.2, demonstrating 66.7% of sensitivity. Conclusions: the Scale of attitudes toward alcohol, alcoholism and individuals with alcohol use disorders in Spanish presented robust psychometric qualities, affirming that the instrument possesses a solid factorial structure and reliability and is capable of precisely measuring the nurses’ atittudes towards the phenomenon proposed.


Introduction
The percentage of alcohol use in Latin America surpasses the global average (40%) (1)(2) . These percentages entail countless health problems, putting this problem high on the regional Public Health agenda. This demands training and knowledge from the health professionals, who are increasingly confronted with this population in the different care contexts (3) . As the professionals' knowledge on the problem is directly associated with their attitudes during care delivery to the users, it is important to identify their attitudes towards alcohol and aspects related to its use (4) .
Although attitudes have been largely defined since the 1960's, some characteristics currently prevail as structural components, including the values, beliefs and feelings; and their importance as predictors of behaviors or actions towards a specific situation (5) .
These components can be characterized as apparently immeasurable latent traits, but can be assessed by means of psychological measures, which permit evaluating and understanding their nature, causes and consequences in the different contexts, theoretical advancing in social psychology and its application to important behavioral changes in the study population (6) .
An analysis of the tools available to measure the nurses' attitudes towards individuals with alcohol use disorders reveals that most are available in English (7)(8) , and were therefore developed in the Anglo-American social and cultural context. Only two of the instruments available to measure the nurses' attitudes in that language have been translated to Spanish in Latin America and applied to populations in the same region (9)(10)(11) . Nevertheless, the validation processes of these tools are not available in the literature (9) , while other sources discuss these processes superficially (10) , which permits questioning the method adopted to verify the validity of the scales used thus far in Spanish. in Spanish, which complies with the international guidelines for these processes, aiming to guarantee the reliability of its use in that language.
Among the existing instruments in Latin America, the Scale of attitudes toward alcohol, alcoholism and individuals with alcohol use disorders, elaborated in Brazil in 2008 (12) , stands out because it was elaborated in a country of Latin American culture, which puts it closer to the context of the target population for the instrument validation. On the other hand, as opposed to the scales used in the studies executed in the region, evidences have been published on its elaboration and refinement process, demonstrating robust psychometric qualities in its language or origin (12)(13)(14) . En este sentido, el presente estudio tiene por objetivo buscar evidencias de estructura factorial y confiabilidad de la Escala de actitudes frente al alcohol, al alcoholismo y a la persona con trastornos relacionados al uso de alcohol (EAFAA) en enfermeros hispanohablantes en el contexto colombiano.

Method
A methodological study was undertaken, involving

Sample
To guarantee the quality of the psychometric analysis of the scale, in the calculation of the sample, at least five participants were guaranteed for each instrument item (15) . Initially, the sample consisted of 650 nurses from three hospitals (60%) and two (40%) primary health care services, 303 (46%) of whom returned the completed instrument. The following inclusion criteria were considered: being a nurse, professionally active at the time of the data collection.
The sample predominantly included female (85%), single participants (49.5%), who informed that they had received some type of training to work in the field of alcohol and other drugs (57%).

Instruments
To collect the data, a sociodemographic questionnaire attitudes toward alcohol, alcoholism and individuals with alcohol use disorders, translated and validated to the Spanish language (16) . This scale was elaborated in Brazil in Portuguese, conceived based on the classical test theory (CTT) and the latent trait theory, consisting of 49 assertions grouped in four factors, which could be answered using a Likert-type scale with five alternative answers, ranging between (1) Strongly disagree and (5) Strongly agree.

Data collection procedure
To collect the data, the nurses were contacted individually at work, in all shifts (morning, afternoon and night). After describing the research objective and guaranteeing that their identity would be preserved, they received a sealed envelope, which contained the  type of studies (17) .
To analyze the internal consistency of the scale and each of its four component factors, Cronbach's alpha and Mc Donald's Omega were used (18) . The sensitivity and specificity of the EAFAA were verified using the ROC curve technique (19) . For all statistical tests, a 5% confidence level was adopted.

Ethical aspects
This study received approval from the research

Results
The data matrix showed to be fit for the factorial analysis (FA), indicating a MO coefficient of 0.86.
The adequacy rates of the factorial model appointed satisfactory coefficients (Table 1)

Discussion
The objective in this study was to validate the Spanish version of the EAFAA when applied to Spanish speaking nurses in the Colombian context, demonstrating that 75% of the four component factors and the factor loading of the 49 items resulting from the content validation process (16) presented adequate correlation coefficients (98% of the correlations between the factors ranged between moderate and strong), indicating a satisfactory representation of the construct measured, in line with the studies developed in Brazil, whose results indicated significant correlations between 75% and 80% of the factors, suggesting that this distribution indicates that evidence exists of a consistent factorial structure.
After concluding the analysis of the factor loadings resulting from the confirmatory factor analysis, one item was excluded: 45: "Individuals with alcohol use disorders cooperate with their treatment", because it did not reach the cut-off point permitted (0.30) (17) . This fact can be associated with the cultural differences and the approach used in care for this population in Colombia, being centered on internment, which presupposes the user's unconditional cooperation in the treatment.
The exclusion of this item was analyzed in detail to guarantee that it would not affect the reliability of the (α=0.90) (12) , and in later studies intended to improve the scale in Brazil, with coefficients of α = 0.90 for an 83-item version (14) and α = 0.89 for the current 50-item version (13) . These data indicate evidence to sustain the reliability of the EAFAA and its stability when reproduced in other languages and contexts, turning the scale increasingly robust to measure the attitudes towards alcohol, alcoholism and individuals with alcohol use disorders, suggesting the validation of its structural and metric equivalence (20) . Therefore, studies are needed to validate the use of the EAFAA in those scenarios.

Study limitations
These study results indicate that the EAFAA is a

Conclusions
The validation process of the EAFAA -Spanish version suggested that the scale is a valid and reliable instrument with robust psychometric properties, capable of measuring nurses' attitudes in the Colombian culture towards alcohol, alcoholism and individuals with alcohol use disorders.