VALIDATION STUDIES IN NURSING: INTEGRATIVE REVIEW ESTUDOS DE VALIDAÇÃO NA ENFERMAGEM: REVISÃO INTEGRATIVA ESTUDIOS DE VALIDACIÓN DE ENFERMERÍA REVISION INTEGRADORA

The aim of this study was to analyze the methods used for validation studies in nursing research considering the clinical investigation as phenomenon. We carried out literature at BIREME, CINAHL, PUBMED which contain the terms 'validation studies', 'nursing' and 'clinical' and 21 articles were included in the review. The majority of the studies were conducted in 2008; in North America (USA) and European Community (62%) in the area of adult health. Most of the phenomena investigated were related to nursing care involving physical and emotional aspects. The content validity has been cited in 71.4% of the articles, criterion validity in 28.5% and construct validity in 23.8%. The reliability by means of Cronbach's alpha was used in the majority of the studies. It was found a knowledge gap concerning the validation estudies in the area of public health, child health and the social phenomena related to nursing care.

A current concern of nursing researchers is related to the evaluation of its phenomena, as research carried out with new tools or instruments used by other researchers must be validated (1) . If the validity and reliability of evaluation tools do not reflect the theoretical concepts being tested, the conclusions drawn from the empiric phase shall be invalid, thus failing to promote development (1) .
The term validity is also defined as the degree to which it is adequate to measure the true value of what it is intended to evaluate, enabling to infer the extent to which results obtained with the applied instrument represent the true or are opposed to it (2) . There are three main validity principles that change according to the information offered and the goal of the researcher: content validity, construct validity and criterion validity (1) .
The content validity represents the content universe or the control of constructed data that provides the structure and the base for the formulation of questions that shall adequately represent the contents.
In this type of validity, the researcher must define the concept and identify the dimensions of the concept components (1,2) .
The construct validity is based on the extent to which a test measures a feature or theoretical construct and seeks to validate a theoretical body subjacent to the testing of hypothetical relations (1) .
However, validity related to a criterion demonstrates the level at which the performance of the researched subject and its real behavior are related when using the evaluation tool. Therefore, criterion is the second measure that evaluates the same studied concept (1) .
For example, when comparing the oxygen saturation level in newborns by applying fetal hemoglobin and the oxygen saturation level obtained through a pulse oxymeter device, both results should be equivalent (3) . This study serves as an example of criterion validity.
The correlation level of the evaluation with a criterion that is external to the measured phenomenon seeks to evaluate the degree to which the instrument discriminates among persons who differ in a certain feature in agreement with a standard criterion (4) .
Validation studies are widely used in several research areas both in national and international literature. In nursing, in particular, these types of studies are not recent. Examples are the patient classification system created in 1960 (5) or nursing diagnosis validation studies, which began in 1979 (6) .
Translation and adaptation of scales such as pain (7) , pressure ulcer prevention (8) , infant anxiety (9) and a questionnaire of drugs use (10) are other examples of validation studies.
International literature includes many nursing diagnosis validation studies (6) such as Gordon and Sweeny, the Fehring and Hoskin models, being the Fehring one the most accepted in Brazil (11) .
Considering this context, the goal of this study is to analyze validation methods used in clinical nursing research, as we consider that there is a scarcity of works of this nature in literature. Due to their importance for nursing practices, it is necessary to evaluate the advances and gaps of this knowledge.
The integrative review study is performed in seven stages: hypothesis selection or questions for review; selection of the research that shall be included in the review sample; definition of primary research features that compose the review sample; analysis of findings from articles included in the review; interpretation of results; review narration and findings critical analysis (12) . In this review we opted for following these stages. At BIREME, all available databases in this library were included by entering Health Science keywords (DeCS/MeSH), the keywords "validations studies", "nursing" and "clinic and the Boolean Operator AND.
The search strategy was built in agreement with the specificities of each database. In order to grant database search uniformity, a question and the inclusion criteria were used as guidelines. Two researches performed the investigation independently, being initially selected those articles that had abstracts chosen by both of them in accordance with the set criteria.
In relation to the kind of validation described in the analyzed articles, content validation was mentioned in 71.4% of cases, followed by criterion validation (28.5%) and construct validation (23.8%). Content validation only was used in 52.4% of studies, criterion validation in 14% and construct validation in just 9.5% of cases.
In studies that used more than one validation type, content and criterion validation were used together in 9.5% of studies, content and construction in 9.    that the matter has attracted, confirming that nursing as a science has taken into account the evaluation and measurement of its results (1) .
Therefore, scientific production in the nursing area has increased in the last decade both in the national and international scenarios, including those DISCUSSION studies that use integrative reviews as a research methodology (29) .
In Brazil, the growth in the number of scientific research in the area consolidates post-graduate teaching, masters and doctorate courses, dramatically contributing to the development of Brazilian nursing (30) , even if in this review there were only 9.5% of Brazilian studies.
A remarkable aspect of this research is the validation type: 71.4% of studies used content validity (9,10,(14)(15)18,20,(25)(26)(27)(28) . The explanation lies in the need for nursing to develop measurement instruments that contemplate all aspects of the researched phenomenon, once this validation is applied to cases in which a universal behavior can be clearly defined (31) .
Although most studies are content validation ones, in this review, only two nursing diagnosis validation studies were identified (27)(28) , even knowing that nowadays this kind of research is considerably large. This is possibly due to the fact that these studies do not involve phenomenon analysis in clinical practice.
As for the phenomena studied, results enabled to identify that most of them were linked to assistance aspects such as care quality (19) , brain trauma (20) , newborns with respiratory discomfort saturation levels (3) , nursing diagnosis related to mechanical ventilation patients (28) among others, confirming that today"s nursing is seeking to qualify the assistance.
Nursing requires conceptualization of the phenomena that it treats and/or it takes care/of.
Consequently, validation studies are still fundamental for the practice to be scientifically based and to overcome the elaboration of inductive/deductive nursing care or diagnosis, seeking to enable a quality improvement and better visibility of professional practices and as a consequence, achieve the long desired autonomy (32) .
In this integrative review, it was observed that more than half (66.7%) of instruments started from the measure scale, confirming the nursing concern about using an instrument that can offer more precision to identify the phenomenon under study.
It was also verified that a large majority (66.7%) of populations studied was composed of inpatients. They were in critical areas such as intensive care units (21.4%) (17,20,28) and the emergency care sector (7.1%) (16) ; which demonstrates that this kind of study has been limited to the hospital environment and to critical patients who require qualified and specialized assistance.
Another interesting aspect is the statistical analysis. In more than half (52.4%) of studies, the Cronbach"s, alpha coefficient was applied, which is used when the item may have more than two alternatives.
The result of this coefficient suggests an internal consistence indicator related to the researched scale and consequently, to the items that compose it. The advantage of its use lies in its capacity to obtain a very valid and reliable measure employing the smaller number of items possible. In other words, selecting items that contribute to achieve maximum reliability and validity (33) .
Through the analysis of this study, it can be affirmed that there is a certain similarity in the ways content, criterion and construct validation studies are carried out. However, one of the limitations is the fact that validation studies that may bring some contribution might have been excluded during the first stage, when researchers often read only titles and abstracts, as only well-structured summaries that met inclusion criteria were selected. Another limitation is that articles written in other languages other than Portuguese, Spanish and English were not included.
Despite the significant number of articles found, the selected sample size was reasonably sufficient.
However, the number of publications on the topic has grown in recent years, demonstrating the interest of nurses on the subject.
Most studies in this review aimed at content validation with phenomena analysis in clinical practice.
It was observed that the most studied phenomenon was related to the assistance to patients/clients with regards to physical and emotional aspects. The most frequently used instrument was the measurement scale. In more than half of articles, Cronbach"s alpha was used for statistical analysis and the most studied area was adult health in the hospital environment.
There is a clear knowledge gap related to the validation of instruments that aim at evaluating public health, child health and social phenomena related to nursing care.