Domestic violence on children: development and validation of an instrument to evaluate knowledge of health professionals 1

ABSTRACT Objective: to develop and validate an instrument to evaluate the knowledge of health professionals about domestic violence on children. Method: this was a study conducted with 194 physicians, nurses and dentists. A literature review was performed for preparation of the items and identification of the dimensions. Apparent and content validation was performed using analysis of three experts and 27 professors of the pediatric health discipline. For construct validation, Cronbach's alpha was used, and the Kappa test was applied to verify reproducibility. The criterion validation was conducted using the Student's t-test. Results: the final instrument included 56 items; the Cronbach alpha was 0.734, the Kappa test showed a correlation greater than 0.6 for most items, and the Student t-test showed a statistically significant value to the level of 5% for the two selected variables: years of education and using the Family Health Strategy. Conclusion: the instrument is valid and can be used as a promising tool to develop or direct actions in public health and evaluate knowledge about domestic violence on children.

The expansion of primary care services that recently occurred in Brazil, by increasing the number of Family Health Strategy (FHS) teams, is an opportunity for greater surveillance of domestic violence. In theory, these teams should be able to deal with situations of violence, but little is known about the knowledge on this subject, or the ability of team members to approach the problem. Expanding the capacity to detect violent situations is necessary to allow the construction of social support networks and confrontation of these situations (8) .
Given the complexity of this problem, its approach cannot be competence a single area of or professional category. This complexity involves a multidisciplinary view and intersectoral action, as part of collective actions. Within the FHS framework, information can be produced and exchanged with those of other sectors.
Thus, professionals must employ means such as social networking, reflecting on the vulnerable conditions of life, aiming to ensure rights, in the same way as developing potential actions against violence upon children (9) .
The knowledge of health professionals on this issue must be evaluated. However, effective tools to assess this knowledge are not available. Although tools for evaluation of the victims of psychological, physical and sexual violence are available, instruments aiming to evaluate health professionals were not found (10)(11) .
Further studies are necessary to verify the knowledge of professionals that guide public health policies, and to combat domestic violence in children. This study aimed to develop and validate an instrument that evaluates the knowledge of health professionals about domestic violence in children.

Method
The study consisted of the development and  Step 1 Step 2 Step 3 Step 4 Step 1-Identification of the dimensions related to domestic violence on children The identification of the instrument's items related to "Knowledge about domestic violence in children in the practice of healthcare professionals" was based on education materials/articles and on national and state guidelines on the theme (12)(13) . An integrative literature review was conducted using the descriptors child health, domestic violence, health personnel and family health.  Step 2-Apparent and content validation The instrument was subjected to content and semantic structure analysis by three experts in the area: a physician with a doctorate in pediatrics and two nurses with a doctorate, experts in pediatric health, who evaluated the presence or absence of the comprehensiveness, objectivity and relevance criteria.
The instrument was also administered to 27 professors of undergraduate and graduate programs in the health area that taught disciplines related to child health. After the analysis of experts and professors, the instrument was redesigned, according to directions and suggestions, and six items were excluded. Next, a pilot study was performed with six health professionals from a private health system the municipality, in order to verify the adequacy and clarity regarding the interpretation of the instrument.

Step 3-Construct validation
The preliminary version of the instrument, "Knowledge about domestic violence on children in the practice of health professionals" with 79 items, was administered to 194 professionals (physicians, nurses and dentists) registered at the FHS of Montes Claros,

MG.
Those items correctly answered by more than 90% (very easy) and less than 10% (considered too difficult level) of the FHS professionals were excluded (14) . The correlation of each item with the mean score of the full questionnaire was performed, using the minimum cutoff point of 0.2 for the correlation coefficient between the total mean score and each item of the instrument, excluding those items with lower values (15) . After expert and researcher analysis, seven items considered important with respect to content were maintained.
As the internal coherence decreased when dimensions classified the instrument, it was considered as a single scale; thus, the Chronbach α value was calculated for the full instrument with 56 items, reflecting a satisfactory level of internal consistency (16) . After this analysis, the  (17) .

Step 4-Criteria validation
Initially, the scores were calculated using the sum  (19) .

Results
Among the professionals who participated in the validation process, a predominance of women (74.7%) was verified; nurses (42.8%), with less than five years since graduation (55.6%). The title that was most often identified was that of family health resident (37.6%).
Most of them had one to five years working in the FHS analyses by 194 professionals, 11 more items were excluded that were considered too easy or too hard, that is, as the agreement and errors by more than 90% of the professionals. Among that, 56 items remained, whose Cronbach's α analysis was 0.734, reflecting a satisfactory level of internal consistency. Table 1 presents the result of the Kappa statistic in the reproducibility of the instrument. More than 80% of the items showed a moderate to very good agreement (12) . The results of criteria validation, based on the time working in FHS and time since completing education are shown in Table 2. Both the criteria used showed a statistically significant association.
Rev. Latino-Am. Enfermagem 2016;24:e2772  (10) . Authors translated, adapted and validated the content of the Childhood Trauma Questionnaire, which is administered to adolescents and adults, with the aim to investigate the history of abuse and neglect during childhood (11) .
The process of preparation followed the recommended methodological and statistical aspects (20)(21)(22) , with satisfactory internal consistency Semantic analysis was performed with a more sophisticated sample (higher skill) of the studied population, ensuring that the items in the instrument were intelligible to the lower stratum. Thus, the instrument was referred for an analysis of experts on the subject, suggesting addition or modification of items.
This same process was conducted in the study of Stelkoo-Pereira and colleagues, whose goal was the validity and internal consistency of the School Violence Prevalence Investigation Questionnaire -student version, a situation similar to that which was found in the study of Hermida and Araújo that aimed for the development and validation of the nursing instrument (24)(25) .
In construct validation, the dimensions of the items of the instrument of this study, must be simultaneously