A CRITICAL ANALYSIS OF AN INTERPROFESSIONALISM ASSESSMENT SCALE APPLIED TO HEALTH SCIENCES UNDERGRADUATE STUDENTS

| Introduction: The search for initiatives that stimulate interprofessionalism should incorporate activities developed by educational institutions. The BAHIANA School of Medicine and Public Health (EBMSP) runs a project, approved by the Government Program of Education for Work in Health PET-Saúde, which promotes interprofessional practice. Objective: Evaluate the psychometric properties of the IEPS when applied to Brazilian students. Method: We conducted an Exploratory Factorial Analyses and took into account the eigenvalues and the theoretical model in order to estimate the number of factors. The Cronbach alpha coefficient was also calculated. Results: A three-dimensional factorial solution was tested. These three factors jointly explain 66.5% of variance in interprofessionalism. Three items were excluded, due to empirical weaknesses. The Cronbach alphas obtained for each dimension were 0.84, 0.70 and 0.16. Conclusion: The study attained its objective, gathering initial evidence regarding the validity of this measure in a Brazilian context.

Technological and scientific advances have led to greater specialization in theoretical and practical knowledge in health sciences and, as a consequence, to the fragmentation of its disciplines.Rather than contributing to a better understanding of phenomena, however, due to its complexity, an excess of fragmentation has begun to hamper comprehension.This has created the need for a reorganization of knowledge, through a search for common denominators between the several fields of knowledge: interdisciplinarity arises from just this scenario 1,2 .
Since the relevance of interdisciplinarity has been recognized, it has become essential to reform processes for the construction and socialization of knowledge.Such practices are essentially aimed at maximizing the benefits provided to patients through cooperation between different professionals 3 .
The search for initiatives that stimulate interprofessionalism should incorporate activities developed by educational institutions.By promoting the integration of health courses into the teaching and service process, the Government Program of Education for Work in Health -PET-Saúde, promotes interprofessionalism.The BAHIANA -School of Medicine and Public Health (EBMSP) runs a project approved by this government program 4 .
The IEPS has been widely adopted in research on interdisciplinarity 5,6,7 .In 2007, a study was published analyzing its psychometric properties when applied to a new participant sample 8 .In this study, the authors proposed a reformulation of the instrument, which involved withdrawing one of its dimensions -

METHOD Participants
The sample consisted of 84 students, all candidates for admission to the PET-Health / Networks program at EBMSP.Of the 84 participants, 76.2% were female, 58.3% were in their fifth semester at the time of the research, while the remaining 41.7% were in their sixth.The courses included in the study were medicine (54.8%), nursing (31%) and psychology (14.2%).
The study was approved by the local ethics committee (CAAE 57164216.1.0000.5544).Informed consent was obtained from all participants.understanding the value of other professionals -and excluded certain items from the other dimensions.The reformulated version of the IEPS was composed of 12 items distributed across three dimensions: 1. Competence and autonomy (5 items, α = 0, 79); 2. Perceived need for cooperation (2 items, α = 0.40); 3. Perception of actual cooperation (5 items, α = 0.83).Subsequently, Leitch (2013) 9 conducted a comparative study between the structures proposed by Luecht et al. (1990)  3 and McFadyen et al. (2007) 8 , asserting that the solution suggested by the latter authors it a better fit for application to undergraduate students.
In Brazil, the IEPS scale was translated by a group of researchers for exclusively didactic purposes.The translated version has been used in training workshops for health professionals.However, no studies have been conducted regarding the psychometric adequacy of the scale in its Portuguese translation.
This article aims to evaluate the psychometric properties of the IEPS when applied to Brazilian students.We applied the 12-item version, since it represents an improvement over the original scale.

Instrument:
The Interdisciplinary Education Perception Scale (IEPS) was used in its Portuguese translation.In response to the scale, the participant was invited to express their degree of agreement with each of the items on a 6-point Likert scale (1 = totally disagree to 6 = strongly agree).The original items in the English language (3, 8) and their respective translations may be found in Table 1.

Data analysis:
Initially, the Kaiser-Meyer-Olkin Measurement of Sampling Adequacy (KMO) and Bartlett's sphericity test were analyzed for factor matrix checkability.The factorial structure of the scale was then evaluated through Exploratory Factorial Analysis (EFA), using the Principal Axis Factoring (PAF) extraction method and Direct Oblimin rotation.In order to estimate the number of factors, the eigenvalues were observed and, to complement this, the theoretical model adopted to construct the scale was considered as a criterion for factor definition.In order to verify internal reliability, the Cronbach alpha coefficient was calculated for each dimension, with values at or above 0.70 considered acceptable, and indices starting at 0.80 as highly reliable 10 .

RESULTS
In the EFA, the KMO was 0.718 and Bartlett's sphericity test was significant (p <.001), ensuring matrix factorability.In order to estimate the factors, we initially considered the eigenvalue criterion, which suggested the existence of up to four factors, as seen in Table 2.
From the combination of the eigenvalue criterion and the theoretical model, a three-dimensional factorial solution was tested, in line with predictions in the international literature 8,9 .Table 3 presents the item distribution by factor and respective factorial loads.

Table 3. Factorial loads of the IEPS Tri-factor Solution composed of 12 items
Looking at the EFA results (Table 3), we can observe that four items presented empirical behavior divergent from expectations, according to the theoretical model we adopted: item 7 did not present a factorial load for any of the factors, items 8 and 11 presented high factor loads for factors not congruent with expected semantic representations, and finally, item 3 presented factorial ambiguity.
In view of these results, we decided to withdraw items 7, 8 and 11 and conduct a new EFA on the remaining In Table 4, we can see that these three factors jointly explain 66.5% of the explained variance of the phenomenon in question -interprofessionalism.From the factorial loads obtained by these items (Table 5), we can observe that the first factor grouped four items related to the perception of actual cooperation; the second involved three items regarding competence and autonomy, while, for its part, the third factor grouped the two items regarding the perceived need for cooperation.The Cronbach alphas obtained for each of the dimensions were 0.84, 0.70 and 0.16, respectively.
items.The decision to maintain item 3 for the next phase was because it had a factorial load within the correct dimension, and its factorial ambiguity may have been derived from empirical confusion caused by items 8 and 11 (which obtained factor loads not congruent with theory).
In the new EFA, the obtained KMO was 0.712 and Bartlett's sphericity test was significant (p <.001), attesting, once again, to data matrix adequacy.According to Table 4, the eigenvalue criterion predicts the existence of up to three different factors.

DISCUSSION
When examining the EFA results, we can confirm that the best solution for the Portuguese version of the IEPS is a scale composed of nine items, distributed across three factors: the first factor -"Perception of actual cooperation" -was the most representative of the phenomenon of interprofessionalism, accounting for 36.7% of variance.The second factor, called "Competence and autonomy", was able to explain 18.2% of variance, while the third and last factor, "Perceived need for cooperation", demonstrated the least explanatory power and was responsible for 11.6% of phenomenon variance.
In order for the factorial solution to be interpretable and compatible with those suggested by international validation studies 3,8,9 , it was necessary to exclude three items.Instrument translation should be considered as a potential reason for the empirical inadequacy of these items.Item 11 is a clear example of this weakness, since the original and translated versions differ significantly in content (Individuals in my profession think highly of other related professions/ Indivíduos na minha profissão pensam fortemente na minha profissão) -the focus in the original version is on other professions, while the translated version directs participant's attention to their own profession.Item 7, for its part, provided a rigorous translation of the original item ("Individuals in my profession believe in the judgment of other professions").However, it seems pertinent to adapt this to express a more suitable meaning for the Brazilian context, since the term "judgment" has a different connotation from the original item (which, in Portuguese, may represent a personal and evaluative view).
In addition to translation inconsistencies, we note other weaknesses in the translated version of the IEPS, which consist of aspects "inherited" from the original version.

CONCLUSIONS
This study aimed to evaluate the psychometric properties of the Portuguese version of the IEPS when applied to a sample of Brazilian students.We recognize, however, that the study contained important limitations.Among these are weaknesses regarding the quality of item translation and the low number of participants.Despite these limitations, the work attained its objective, which was to gather initial evidence of the validity of the measure in the Brazilian context.We recommend further studies for this research agenda, which should include a careful review of the measure (preferably proposing improvements in terms of the quantity and quality of items), followed by a new application, with a larger sample, with greater variability and a re-evaluation of the psychometric properties obtained.

Table 1 .
Original and Portuguese translated versions of the items in the Interdisciplinary Education Perception Scale (IEPS) a = competence and autonomy; b = Perception of need for collaboration; c = Perception of real cooperation.

Table 4 .
Number of factors suggested in the IEPS by Eigenvalue criterion (9 items) For example, in the first version, proposed by Luecht et al. (1990) 3 , the Perceived Need for Cooperation dimension is composed of only two items, which almost inevitably leads to low factor reliability.The results obtained in our study (α = 0.16) were lower than those obtained in the studies of Luecht et al. (1990) 3 and McFadyen et al. (2007)8, however, these surveys had already obtained unsatisfactory reliability values (α = 0.56 and 0.38, respectively).

Table 2 .
Number of factors suggested in the IEPS by Eigenvalue criterion (12 items)

Table 5 .
Factorial loads of the IEPS Tri-factor Solution composed of nine items