Psychometric properties of the Attitudes Scale facing Alcohol and Alcoholism in nursing students 1

ABSTRACT Objective: to verify the psychometric properties of the Attitudes Scale facing Alcohol and Alcoholism (EAFAA) and people with disorders related to the use of alcohol in nursing students. Method: a convenience sample (n=420) completed the EAFAA, the data were submitted to Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA). Results: the EFA resulted in an instrument composed of 48 items divided into four factors. The CFA has established the validity of the factorial structure. The internal consistency of the scale was considered adequate (α=0.85) presenting a sensitivity of 70% and specificity of 75%. Conclusion: the EAFAA constitutes a reliable instrument to identify the attitudes of nursing students towards alcohol, alcoholism and persons with disorders related to alcohol use.


ior but also
o everything that can be logically inferred from the external behavior.Also according to these theories (1) an attitude is an enduring organization of beliefs and general cognitions, endowed with emotional charge, for or against a defined social object, which predisposes to an action consistent with the cognitions and emotions related to this object, in the case of multidimensional constructs involving affective, cognitive and behavioral components.

Given the diversity of definitions of the term attitudes, at least three common elements can be found in most of them, namely: attitudes are conscious or unconsciou mental state, a value, a belief or sentiment facing a particular object, a predisposition to a certain behavior and or action (2) .

Regarding the attitudes of nurses towards alcohol and associated issues, according to scholars of the subject (3) , these are influenced by the values and moral per eptions internalized during childhood and also in daily life.These values lead to conceiving persons with disorders related to alcohol use as without character, and guilty of their health problems it also evokes fear in childhood, which is turned into rejection and avoidance in adulthood.Thus, the concepts and predispositions acquired in relation to those people in the social, emotional and intellectual development of the individual bring up other values, in the case f nurses, manifested through negative attitudes to people with disorders related to alcohol use (3) .


Studies investigating the issue of training nursing

students in the field of psychoactive substances have shown that most of them are often faced with people with disorders associated with the use of alcohol and other drugs, in various specialties during their training (4) .

Students ca reproduce the attitudes of practitioners facing this clientele, which have been described as negative and permeated by the moral connotations of the problem (3) .This phenomenon is not clear, given the scarcity of national scientific literature seeking to identify the perceptions and attitudes of nursing students towards the patient with disorders related to alcohol use (5) .

North American studies, mainly in the last three decades of the last century, provide evidence that interventions made with nursing students through training sessions are effective in increasing knowledge; however, changes in attitude occur less frequently, and when they occur, they linger for short periods of time (6) ; there is also evidence that after appropriate training sessions nursing students feel safer and better prepared to intervene with alcoholic patients with more positive attitudes (6) .A more recent study (7) found that after a theoretical training on the approach towards patients with disorders related to alcohol use, improvement was observed in attitudes relating the perception of legitimacy and capacity to work with alcohol-related problems patients among students, however, they manifested less motivation to work with this clientele at the end of training.

In Latin America and specifically in Brazil, studies on the ttitudes of nursing students regarding alcohol issues suggest positive that they have positive attitudes (4-5)   , even though they prefer not to deal with this kind of patient (6) .These difficulties have been attributed by the researchers of this phenomenon (4)(5)(6) to the sparse attention g ven to the teaching of this subject in the curricula of Brazilian nursing schools.

It must be emphasized, however, that despite their significant contribution to the knowledge of this phenomenon in Brazil, national studies have obtained their results using instruments mostly not validated for use among nursing students, a fact tha can contribute to biases in their findings.

Given the scarcity of Brazilian studies assessing the attitudes of nursing students to issues related to alcohol (6) , to the need to map the situa ion in geographic

areas not yet explored, and also the importance of identifying the attitudes of this population to propose strategies to characterize the attitudes of future nurses.

The availability of a robust and reliable instrument can be considered essential to systematically study the attitudes of nursing students when facing issues related to alcohol and alcoholism.

The scale of attitudes towards alcohol, alcoholism and pe ple with disorders related to the use of alcohol (EAFAA), was developed and tested psychometrically to measure this construct among nurses and health practitioners (3,8) , however, there are no studies of it psychometric properties among nursing students.

The EAFAA consis s of 50 items divided into four fully agree) showing reliability indices close or equal to 0.90 (3,8) .

The positively oriented EAFAA items are usually ordered in a way to avoid the acquiescence in the style of responses, i.e., the tendency to constantly endorse "agree" or "disagree" in response to the items.

The estimated time to complete the instrument is 15 minutes (8) .

Given that the estimates of the validity of instruments and its reliability depend mainly on the nature of the samples,

d the fact
that whenever an instrument is used in a new context or in different groups, it is necessary to re-establish its psychometric properties (9) .It also needs to be considered that the psychometric studies published on EAFAA result of its application in health practitioners (3,8) .This study aimed to study the psychometric qualities of EAFAA among nursing students.


Objective

Verify the psychometric properties of the Attitude S

le towa
d Alcohol, Alcoholism and people with disorders related to the use of alcohol, in nursing students.

Method 420 nursing students of the fourth and third year were recruited from three nursing scho ls (two private, one public) in São Paulo during the period of January 2011 to December 2012.Participants were randomly divided into two samples, one of them composed by 75% of participants (n=298) and the second consisting of the remaining 25% (n=122).In order to preserve the anonymity of the institutions, the schools are named as School A School B and School C.

e collec
ion instruments were constituted from the EAFAA composed of 50 items (8) and a sociodemographic questionnaire with information on age, gender and year (third or fourth) attended at the time of collection.

For data collection, the questionnaires were distributed in the classroom in envelopes to be deposited in an urn in the back.The participants who agreed to participate in the research deposited the answered instruments without identification.The Research Ethics Committee of the institutio of the main researchers approved the proposal prior to its beginning.Of the respondents, the majority, 61%, were from School A.

Table 1 shows the demograp ic data of the participants according to the Exploratory Facto Analysis (EFA) and


Confirmatory Factor Analysis (CFA).

Given that it was an instrument already validated in populations of nurses and other ealth professionals, The criteria adopted for the composition of the factors and for keeping items in th model were the same used in validating the original version of EAFAA (8) .This standard has also been followed to determine the reliability of the scale, using the Cronbach's alpha for the instrument in its entirety and for each of

he extracte
factors.The Pearson correlation coefficient was used to verify the correlation between the variables of EAFAA, and between the scores obtained by the sample in the instrument before and after the changes brought about by the AFE.

The cutoff of the EAFAA nursing students version, was identified from the analysis of the ROC technique (receiver operating characteristic), and defined as the one that maximized the Youden index (10) .


Results

The criteria of normality and sphericity checked by the Kaiser-Meyer-Olkin and Bartlett's sphericity tests were attended presenting respective values of (0.85) and significance of (<.0001).The model consists of 50 items distributed in 4 factors and it was the best solution for this version of the scale.

The model was submitted for consideration, with the exclusion of 2 items that made up the initial version of the instrument (8) (F2 33 -I consider the patient with Rev. Latino-Am.Enfermagem 2016;24:e2823 disorders related to alcohol use as a lost cause & F3 27 -Dissatisfied people abuse of alcohol) because they have a factor load below the determined cut-off point to stay in the instrument (0.40), and because they have shown significant factor load on more than one factor at the end of the rotation.

Excluding these two items from the EAFA for nursing students did not cause damage to the reliability coefficients of the respective factors and the EAFAA.The observed result was consistent with the results of the factor analysis of the original sca e (8) , keeping 48 items distributed in 4 factors (Table 1).


The reliability coefficients measured by

Cronbach's alpha for the scale in full (α=0.85)

an also for each of the 4 individual factors were considered satisfactory in both cases (Table 2), with a significant correlation between the factors and between these and the full version of the scale (Tabela 2).As expected, a significant correlation was found (r=0.90, p<000.1) between the scores obtained by the sample in EAFAA (8) and in the version modified by the EFA, supporting the validity criteria of the EAFAA for nursing students.

The analysis for the selection of the cut-off point of the EAFAA adopted from the ROC curves technique pointed out as optimal for nurs ng students the 3.29 score presenting a sensitivity of 70% and specificity of 75%.

The model with four factors resulting from the EFA, was assessed in the second sample of participants (n = 122) using the CFA for this purpose.

The results of this analysis showed reasonable fit of the model as suggested by the indexes (Table 2).

The reduced model is shown in Figure 1.All four factors presented inter-correlation among them (Table 2)..620.700

*Significant correlation to the level 0.01; † significant correlation to the level 0.05; ‡ degrees of freedom; §.Tucker-Levis Index; II Comparative Fit Index; ¶ Root Mean Square Error of Approximation; **Parsimony Normed Fit Index; † † Parsimony Comparative Fit Index.

Elements in bold in the main diagonal of the correlation matrix between the factors correspond o the reliability factor.


Discussion

The results of this study provide evidence to support the validity of the EAFAA for use among nursing students, the factorial structure of the resulting scale of the AFE is consistent with that obtained in the original version (8) .Although composed of 48 items, two less than in the previous version (8) , the EAFAA maintained observing a significant correlation coefficient with the scores observed in the health practitioners' version (8) .

The internal consistency of the EAFAA for nursing students, as established by Cronbach s Alpha, showed adequate levels of relia

lity for th
full version as well as for each one of its four factors (11) .The factors 1, 2 and 3 showed excellent, good and adequate indexes

Vargas D, Rocha FM.

respectively of reliability (11) , the factor 4 showed acceptable reliability index (α=0.60) (11), consistent with previous studies that evaluated the psychometric properties of the EAFAA in different populations (3,8) .It was also noted a low correlation of this fact with the EAFAA and the other factors of the scale, a result also observed in previous studies (8) Among the reasons for this result