USE OF THE BIBLIOGRAPHIC REFERENCE MANAGER IN THE SELECTION OF PRIMARY STUDIES IN INTEGRATIVE REVIEWS

Objective: present the use of the bibliographic reference manager EndNote ® as a support tool in the construction of a database and the selection of primary studies to develop integrative reviews. Method: a reflection study was undertaken, based on the international and the Brazilian literature. Results: the elaboration of an integrative review permits summarizing the knowledge about the theme of interest delimited in the health area, which can contribute by providing recommendations based on research results for clinical practice, as well as by identifying knowledge gaps, driving future research development. The use of bibliographic reference management tools is important for the selection of primary studies, permitting the location of repeated references, identification of languages, selection of publication periods and the creation of citations in the elaboration of the scientific text, among other resources. Conclusion: the advance in technologies, particularly the software EndNote ® , has provided resources for the practicability of knowledge synthesis methods, in a transparent and replicable manner, consequently entailing a better quality of this type of study.


3/13 INTRODUCTION
The synthesis of knowledge is the scientific method for summarizing evidence from various studies on a specific theme, identifying gaps in research and suggesting new studies, as well as for providing the best evidence available for making decisions in health.The knowledge synthesis methods include the systematic review, meta-analysis, qualitative synthesis, and integrative review.] The quantity and complexity of new knowledge production in the health area, added to the professionals' time constraints, has led to an increase in the development of knowledge synthesis methods among health researchers, especially in nursing. 3he integrative review (IR) is a method that permits summarizing knowledge through a systematic and rigorous process.The development of the IR needs to be based on the same principles of methodological rigor in the development of research. Figure 1 -Steps of the integrative review, Ribeirão Preto, São Paulo, 2017 2,[4][5] The importance of methodological rigor in IR and the need to use tools to manage the selection of primary studies were the aspects considered to construct this reflection for health researchers.Thus, in view of the authors' experience in the practice of review methods using the reference manager, in addition to the productions in Brazilian nursing that used IR as a method, the objective of this reflection is to present the use of the EndNote ® bibliographic reference manager as a support tool in the construction of a database and in the selection of primary studies for integrative reviews.

REVIEW QUESTION AND PRIMARY STUDY SEARCH STRATEGY
The researcher should delimit the topic of interest of the review, usually a theme that motivates the development of the study and is related to clinical practice.The definition of the guiding question is essential and permits the determination of the research problem, the target population, as well as the key variables. 2In addition, the review question will be the framework to determine the selection criteria of the primary studies.
The question is also relevant for the identification of the descriptors for the search of the primary studies, and, therefore, for the construction of the search strategy.The proper definition of the question is fundamental to avoid identifying irrelevant studies in view of the review purpose.To develop the IR, the suggested format for the guiding question involves the definition of the population, the intervention or condition or problem or phenomenon and the results of interest.It may also include the type of study (research design) or other criteria such as language, geographical location or time period. 6hus, the use of the PICO format (an acronym for patient, intervention, comparison, outcomes) can help in the proper construction of the review question, 7 however, it is emphasized that in IR, it is not always possible to delimit all the elements of the PICO strategy.This strategy can also be used in the PICOT format, where T stands for time, or PICOS, where S is the type of study.
Currently, most databases have interfaces to include the elements of the PICO strategy, enabling the reviewer to implement practically the same format in all selected databases. 8The elements of the strategy mentioned to formulate the review question, the definition of each element and the possible questions are described below: P) Problem, patient, population -What will be the group of interest?I) Intervention, exposition or topic of interest -Which intervention or topic will be studied?C) Comparison -Is there any comparison of the intervention?O) Outcome or results -What will be the effect of the intervention?;] After constructing the question, the reviewer will start delimiting the controlled descriptors according to each selected database (baseline vocabulary) and keywords.The controlled descriptors and keywords should be combined according to each element of the elaborated PICO strategy, based on a logical structure, which will be the basis of the primary study search strategy.
The difference between the use of controlled descriptors and keywords is highlighted.The keywords are random and extracted from free language texts, that is, they are not structured.Controlled descriptors are organized into hierarchical structures, facilitating the search and identification of studies of interest to the reviewer. 11Structured vocabularies (controlled descriptors) allow the researcher to identify the information with the exact term, providing objective results. 12For the database (PubMed) National Library of Medicine National Institutes of Health, the terms are identified in the Medical Subject Headings (MeSH), in the Cumulative Index to Nursing and Allied Health Literature (CINAHL), in the Headings and in the Latin American and Caribbean Health Sciences Literature (LILACS) through the Descriptors in Health Sciences (DeCS), among other vocabularies.
In the identification of the controlled descriptors in the databases selected to conduct the review, the researcher may also find synonyms relevant to the term of interest.As an example, when searching for the descriptor "Cuidados de Enfermagem", the synonyms "Assistência de Enfermagem" and "Atendimento de Enfermagem" can be identified within DeCS.In this example, the search strategy for primary studies in the LILACS database could be constructed as follows: Cuidados de Enfermagem OR Assistência de Enfermagem OR Atendimento de Enfermagem.
Among the sources of access to the scientific literature, the databases and electronic libraries are highlighted.The choice of the databases will depend on the review question. 13In Brazilian nursing, the most used databases to search for primary studies are PubMed, CINAHL,

5/13
Institute for Scientific Information Web of Science and LILACS.Among the electronic libraries, the following stand out: Scientific Electronic Library Online (SciELO), Capes / MEC Journals Portal and Science Direct.] The search strategy is defined as the technique or set of rules that makes it possible to identify the publications stored in a given database, which can answer the review question.In order to answer the guiding question, it is necessary to perform "tactical operations" to identify valid and relevant information for the purpose of the review.] A priori, there are three steps in the process of planning the search strategy: 1) decision regarding the database; 2) selection of search terms and their suitability in each database; 3) logical formulation of the strategy.The search strategy should be sensible and replicable. 13,15n IR, the search for primary studies should be extensive enough to ensure the identification of relevant and necessary surveys to answer the guiding question.There needs to be a balance between the search for comprehensiveness and the maintenance of relevance in its development though, in order to enable its management.
The structure of the search strategy should include a range of terms (controlled descriptors and keywords) for each element of the PICO strategy.This construct implies the use of database vocabulary terms, text words, synonyms and related terms for each concept, joining each term within each concept using Boolean operators. 14The terms employed can be combined using (AND, OR or NOT) to compose the search strategy. 13The "NOT" operator, as far as possible, should be avoided to reduce the risk of inadvertently removing relevant studies from the search. 14mong the different terms, "AND" is usually used to identify studies on two themes (intersection).Among the synonyms, "OR" is used in order to locate indexed searches with terms different from the base vocabulary.The "NOT" operator is used to exclude a specific subject from the search.In IR studies, adopting limits or restricting characteristics such as language, date or country of publication should be avoided in the search strategy, as it may represent the loss of potentially relevant studies. 13ther search techniques include the use of parentheses (used to establish order in the search process and separate the sets of terms), truncation (a method that uses the dollar sign ($) at the end of a word), employed in the search for word roots; and quotes, used in case of compound terms.Professional librarian assistance is recommended if the reviewer does not have the necessary resources to implement the search for primary studies in the literature. 13,15

MANAGEMENT OF PRIMARY STUDY SEARCH RESULTS
] Many reviewers use Microsoft Excel spreadsheets, while others select studies manually, with the printing of titles and abstracts, which makes the process complex, not ecological and time-consuming.Currently, the workflow in the review process is more streamlined with the use of computer programs. 19n a study published in 2013, the authors pointed out that more than half of the systematic reviews used the software EndNote ® . 20As the SR are similar to the IR, the use of this technology is proposed in order to facilitate the selection process of the primary studies.

6/13
The use of bibliographic reference management programs (e.g.EndNote ® , Mendeley ® and Zotero ® ) is important to operationalize the selection of primary studies in IR. 13 These programs make it possible to locate repeated references, identify languages, select publication periods, create groups according to the selection criteria (inclusion and exclusion), compare the selected references among the reviewers, as well as to create quotations while writing the scientific text, making it easier to format the references at the end of each manuscript preparation process.The choice of the software is influenced by its availability (free or paid) and by the reviewer's adaptation to the available options.2] In general, the reference managers perform basic functions of locating publications, storing the results and citing references when writing the manuscript.
2] For example, the easy use of Mendeley ® is one of its strengths.On the other hand, this manager has the weakness of not allowing searches in databases within its own structure, nor does it permit dynamic changes in the formatting of citations in the texts. 22Thus, it is necessary to know each software to identify the one that best adapts to each reviewer, which is why it is worth mentioning the use of EndNote ® , which the authors of this article have used for more than 15 years.
In this reflection, we will discuss the use of the software EndNote ® , version XV.It should be noted that there is the free online version (EndNote ® Basic) available on the Thomson Reuters website (www.endnote.com),which does not have all the features available in the paid version.
When creating the file in EndNote ® , one can also outline different groups in order to organize the search.After the search of the primary studies in the databases, in which the reviewer crossed the controlled descriptors and keywords from the PICO strategy, one can easily export the search in a file and import through the use of filters for EndNote ® .For example, searching the PubMed database requires the selection of the "PubMed (NLM)" filter; in the LILACS database, the filter of choice is the "Reference Manager (RIS)".Other databases, such as CINAHL and Web of Science, permit the direct import of references by double-clicking the file exported from the database.EndNote ® gives its users a YouTube channel with different videos that teach how to use this software (https://www.youtube.com/user/EndNoteTraining).In addition, free, user-friendly guides are available to the users.
After importing the search in EndNote ® and directing the records referring to the group that indicates each database, it is recommended to save the back-up file with the original search, without managing the content, for the purpose of retrieving publications that may have been inadvertently excluded by the user, besides having the record of the number of references identified in each database.
Another feature is to change (or customize) the available fields within the EndNote ® library to identify the source of each search.The user needs to select "Tools", click "Change and Move Fields", select a field, for example "URL", select "Replace whole field with", in the right frame, enter the name of the database that group of exported references refers to and click "OK".This action changes the entire contents of the references column for the group.When executing this change in each database, the reviewer will have the exact record of the source of each reference.This should be one of the first steps taken by the reviewer.The step-by-step for the described feature and other important commands for performing review methods is available on the YouTube channel "EndNote for Systematic Reviews" * .Figure 2 presents an example of the creation of groups within the reference database in the development of review methods.To do this, just right-click within "My Groups", where you can create group sets ("Create Group Set"), with new groups within each larger group ("Create Group").These * Available from: <https://www.youtube.com/watch?v=QHXWaAZZBqE&list=PLF2KCgTC6mbTl5ngY2sxcKkeJ8iRCaJh7>

7/13
actions permit the creation of different filtrations from the set of identified primary studies.To add references in each group, simply select the publication and drag it to the required group.The use of EndNote ® also permits changing fields within the reference base, in order to manage the inclusion and exclusion criteria independent reviewers performed.In this case, each reference receives a label, with one or more codes, justifying each reviewer's inclusion or exclusion of the primary study.In the literature, in a published study, the authors elaborated a guide on how to use EndNote ® functions to enable the selection of primary studies in a rigorous and organized way. 23n the selection of the primary studies, references should be repeated (duplication of studies) in the searches performed in the different databases.Before removing duplicate studies, one needs to rename the "URL" column, indicating the database from the reference, as mentioned earlier.As shown in Figure 3, you can create a group called "Duplicate Studies" and direct all repeated references to this folder.
To select duplicate studies, the reviewer should click on the top tab named "References" and select "Find Duplicates".Through this action, the program highlights the duplicate references.The next step is to delete these references, an action that sends the duplicate texts to the "Unfiled" folder.If the program sends the duplicate references to the software's trash, you need to select these records again, right-click and select "Restore to Library".This action directs the recovered references to the "Unfiled" folder, where it can be added to the folder of interest as already mentioned.Relevant step is to select all the texts forwarded to the "Unfield" folder, right-click, and add these references ("Add References To") to the folder that will contain the duplicate studies.It is possible that, if this action is taken, some duplicate study still remains, as the reference has been extracted from different bases and is not always indexed identically.
For the selection of the year of publication, for example, the last five or ten years, the reviewer should click on the "Year" column, which appears as the heading immediately before the references.By clicking on this column, the program sorts the references according to the year of publication in ascending and descending order.The reviewer can select the references of the years of interest and do the same action previously described, that is, delete the references that are not part of the delimited period, sending them to the folder "Unfiled".Next, the reviewer should select all the texts forwarded to this folder, right-click, and add these references ("Add References To") to the folder that will contain the studies published in a period not selected to compose the sample of the review.
Two ways exist for the purpose of language identification.The first is to search within the reference database.The reviewer should note that, when clicking any folder of the groups created, different fields can be searched using the "Search" tool.When clicking on the existing options for the search, the reviewer should select the "Language" field.After this action, enter the languages the reviewer does not master or is not interested in.Depending on the database the study was extracted from, the language may be described in detail or identified by the initial three letters, for example "eng" to identify English.It may also be blank, that is, it does not contain information about the language of the reference.
The second way is to use the same "Search" tool, select the "Title" field in the search options and type the bracket symbol "[" because, most of the time, references that are in a language other than English have their titles in English in brackets.After this action, the reviewer should delete those references by submitting the articles to the "Unfiled" folder.Then select all the texts forwarded to this folder, right-click, and add these references ("Add References To") to the folder that will contain references published in other languages that will not be part of the review sample.
For the identification of texts that are not scientific articles, the reviewer should click on the "Ref Type" column, which appears as a heading immediately before the studies listed in the reference database.By clicking on this column, the program orders the references according to their type, making it easy, for example, to identify theses and book chapters.
Another useful feature of the program is the publication summary, an important aspect for the selection of the primary studies that will be included in the IR, by reading the title and summary of each reference.Thus, when clicking on the folder of interest, in the same place where the search takes place within the references themselves, there is an icon called "Preview", which permits the complete visualization of the reference and the abstract.Note that, for this view, in the upper tab, where it is possible to select the reference style, "Annotated" should be clicked, otherwise the user will see the last reference style used for formatting references.
To complete the advantages of using EndNote ® , the program also contains a full text search tool.To do this, the reviewer should select the references, right-click, and select "Find Full Text".It should be noted that the program searches for groups of 250 references, so that the reviewer should pay attention to this when searching for texts in Portable Document Format (PDF).This functionality of the program is also conditioned to the computer's IP, retrieving a larger number of texts in the institutions affiliated with the research portals, or when remote access is possible beyond the academic environment through the Virtual Private Network (VPN) service.

EXAMPLE QUESTION AND SEARCH STRATEGY
In the attempt to provide the reader with an understanding of the content addressed in the previous items, an example was elaborated, which considers the review question and the search strategies of the primary studies.
The guiding question outlined for the elaboration of IR was: "What evidence is available in the literature about the publication of integrative review studies by nurses in the Brazilian context?"Thus, the elements of the PICO strategy for the review question would be as follows: P) Nurses; I) Integrative review studies; C) Does not apply; and O) Number of studies published in Brazil.
The literature search of the primary studies was carried out via the Internet, in the following databases: PubMed, CINAHL and LILACS.To ensure a broad and careful search, the controlled descriptors and keywords were delimited in the Thesaurus according to each database, i.e.Medical Subject Headings, CINAHL Headings and Descriptors in Health Sciences.Next, the search strategies applied in each database are presented (chart 1).
For each search in the databases, an export file to EndNote ® was produced, containing all the references identified after crossing the elements Problem/Participants (P) AND Intervention/Aspect of interest (I) AND Outcome/Results (O) (figures 3 and 4).

Figure 2 -
Figure 2 -Examples of groups that can be created in the selection of primary studies for integrative review, Ribeirão Preto, São Paulo, 2017