Joanna Briggs Institute Reviews Overview

The JBI is an international organization based at the University of Adelaide (Australia); it comprises a network of health scientists, professionals, researchers, and students committed to health practices based on reliable scientific evidence ^12,13. This organization aims to improve health outcomes through scientific production, training, and dissemination ¹⁴. The JBI developed the Evidence-Based Medical Care Model to bring this approach closer to clinical practice, a proposal applicable to all health workers that considers evidence synthesis as an essential component for EBP ^15,16.

The JBI network brings together about 80 centers of excellence and affiliated groups of more than 70 universities and hospitals worldwide, which continually develop methods and tools for rigorous SRs ^13,14. The JBI SRs are designed to answer questions that meet clinical needs and the requirements of healthcare service providers and health policy-makers ¹⁷.

The JBI has made methodological recommendations to conduct eleven types of reviews: effectiveness, qualitative studies, economic/cost assessment, prevalence or incidence, diagnostic test accuracy, etiology and risk, textual synthesis, mixed methods, umbrella reviews, psychometric reviews, and scoping reviews ¹². For all types of SRs, there is an open access manual to guide them ¹⁴.

Regardless of their classification, these reviews are characterized by the systematization of the literature selection process according to the objectives, research question, and inclusion/exclusion criteria stipulated a priori in a protocol. These criteria determine the eligibility of studies and provide the guidelines for conducting an exhaustive search of the literature in different databases and informational sources. Subsequently, a screening process is carried out to select the studies to be included, that is, those undergoing quality assessment and data analysis to create an objective synthesis of the results, evaluating the certainty of the evidence and the implications for practice through a transparent and detailed report on the methods used for the review ¹⁴.

An SR requires a methodological and technical approach; therefore, the JBI has devised a Comprehensive Systematic Review Training Program (CSRTP) ¹⁸ to train researchers, health workers, scholars, and people interested in making SRs using JBI methods. These methods include training in the systematic review software System for the Unified Management of the Assessment and Review of Information (JBI-SUMARI) ¹⁰.

The CSRT program has a workload of 40 hours distributed in three modules: Module 1: Introduction to evidence-based healthcare and systematic review of the evidence; Module 2: Making SRs of quantitative evidence; Module 3: SR of evidence generated by qualitative, narrative, and textual research. Course certification is awarded by the JBI ¹⁸.

In Latin America, the JBI Brazilian Center for Evidence-Based Medical Attention has continuously trained health professionals interested in using the CSRT program since 2009. This center has trained and certified more than 500 health workers, primarily nurses, in the South American region. In addition, it has launched an Evidence Implementation Training Program (EITP) aimed at health personnel to transform health service providers' care practices ¹⁹.

JBI initiatives have improved the training of researchers, systematic and scoping reviewers, consumers, and implementers to strengthen EBN. Currently, the JBI focuses its efforts on creating new affiliated groups in Latin America; the expansion of these groups will undoubtedly improve the results of health practices in this region ¹⁷.

Cochrane Living Systematic Review

Cochrane Living Systematic Reviews (LSRs) are a relatively new proposal, released in 2017 by the Cochrane Collaboration in a provisional guide for LSR pilots, which was subsequently validated and published in 2019 ^20). LSRs constitute an approach specially designed to cater to the need for continuous SR updating and publication ^21). This approach aims to answer a priority question when the level of evidence is uncertain due to scarce primary studies or a high probability that new evidence will modify current knowledge ^21,22.

LSRs usually predominate in medicine ²³, and their application in nursing represents a challenge because this type of approach involves a continuous update that, according to Cochrane, must be monthly ²⁰. Keeping SRs up-to-date bridges knowledge and time gaps in the dissemination of knowledge, especially when the level of certainty is low and the evidence is inaccurate ²³.

An LSR proposal can originate from a research question or a pre-existing systematic review. This review will identify the uncertainties of knowledge in priority areas for nursing.

The Cochrane Collaboration has a guide that describes complementary processes to the Cochrane manual ²⁴ designed to make and publish an LSR ²⁰. In general terms, the Cochrane LSR guide describes each stage of the LSR process and recommends tools called 'enabling,' which offer resources to supporting the work team in searching and screening primary studies, summarizing, and writing until publication ²⁵. These tools have been designed to offer potential gains in saving time and resources allocated to keep the SR permanently updated ²⁶. Researchers and health workers can find a list of Cochrane enabling tools in the Systematic Review Toolbox, available at http://systematicreviewtools.com/.

For example, the Task Exchange²⁸ and Cochrane Crowd³⁷ tools are platforms that support the working group because they help organize and distribute tasks in large teams involving researchers and citizen science collaborators ²⁷, which implies active participation of citizens ²⁹. Moreover, machine learning tools such as RCT Classifier help collaboratively assess and select the available evidence ³⁰, while text mining tools classify information according to importance, considering the most recent updates in the scientific literature ^31,32. Text mining automates information search and retrieval by identifying patterns or correlations between the terms used in databases ³³ through tools such as MeSH on Demand³⁴, designed to generate automatic searches by identifying MeSH (Medical Subject Heading) descriptors in the bibliography published on PubMed. Other similar tools such as Polyglot Search³⁵ and Medline Transpose³⁶ can be used for the same purpose.

For the eligibility and screening stage, machine learning algorithm tools such as Clasificador de ensayos clínicos aleatorizados²⁴ and Cochrane Crowd³⁷ help identify randomized clinical trials ²⁰. Additionally, full-text report retrieval can be automated and accelerated using tools such as CrossRef³⁸ and SRA-Helper³⁹. Another helpful tool for LSEs is Rayyan QCRI⁴⁰, designed to speed up the initial selection of abstracts and titles through a semi-automation process, reducing the time used to filter and preselect search results by 40 % ⁴⁰.

For data extraction, tools such as RobotReviewer⁴¹, The Content-Mine⁴², and Machine-Learning²⁰ allow extracting structured data according to the components of a PICO (Population, Intervention, Comparison, Outcome) question and integrating the risk of bias ²⁰. The facilitating tools for evidence synthesis and automatic text generation include RevMan Replicant⁴³ and RevMan HAL ⁴⁴, designed to automatically generate a text for the meta-analysis results section ²⁰.

Finally, to publish an LSR, the reviewers and the editorial team carry out an exhaustive review of the full text when it is the first version. Reviewers check modified sections for subsequent update versions to identify changes and optimize time ^20,45. In this case, three possible situations may occur: i) no new studies are found to include in the review and, in this case, the LSRs will not require a new peer review and may only be submitted for editorial review; ii) there is new evidence, but it is not possible to incorporate it into the review for reasons described in the protocol, or the new evidence will probably not change the results and conclusions of the previous version of the LSR, in which case the update will go through editorial review, while peer review will be optional; iii) new evidence was incorporated, and the update will require a peer review and a new editorial review ⁴⁶.

Systematic Reviews of Measurement Instruments

Self-reports are instruments to assess behaviors, attitudes, and values in individuals ⁴⁷. Health sciences tend to use them more frequently for patient assessment; however, these instruments must be valid and reliable to measure a subjective construct.

An SR of Measurement Instruments aims to evaluate the psychometric properties and validity evidence of instrument validation studies. These reviews facilitate selecting the most appropriate instruments to be employed in academic, clinical, and research contexts based on a given instrument's validity, reliability, and sensitivity evidence. Having validated and reliable measurement instruments allows progress in disciplinary knowledge and its appropriate application to nursing practice. Between 2015 and 2021, more than 2,900 nursing articles related to the validation of instruments derived from con struction, translation, and cross-cultural adaptation have been published, showing a constant increase in this type of publication and highlighting the need for methods that systematize and optimize the obtaining of results of an SR of measurement instruments. An SR in psychometrics seeks to characterize the articles published on a phenomenon of interest and identify the results on the different measurement instruments used to compare their reliability, validity, and other relevant statistics.

A tool can be used in this type of SR to evaluate article quality: the COSMIN (Consensus-based Standards for the Selection of Health Status Measurement Instruments) checklist assesses the methodological quality of studies on psychometric properties and measurement instruments ⁴⁸. This tool includes the assessment of three domains: validity, reliability, and sensitivity. Validity considers three measurement properties: content, construct (internal structure), and sensitivity. In turn, reliability includes three other measurement properties: internal consistency, reliability (based on inter- and intra-observer test-retests), and measurement error. Finally, sensitivity considers the ability of an instrument to correctly classify a sick individual based on a positive result ⁴⁸. Specificity is the probability of classifying an individual as healthy through a negative result. This checklist has been used to critically evaluate measurement instruments, strengthen EBP, and encourage the use of measurement instruments, with the consequent standardization of patient assessment for the health and nursing professionals in multiple contexts.

Nonetheless, the critical evaluation of an instrument entails techniques and procedures that ensure their accuracy according to international recommendations. For this, analyzing the validity of the content, internal structure, response processes, measure testing, relationship with other variables, reliability, and invariance has been suggested ⁴⁹.

For content validity, study reports should incorporate standards for cross-cultural adaptation. In health, a quasi-experimental study proposed five stages for the cultural adaptation of instruments ⁵⁰; besides, an integrative review showed that 49 °% of the studies had used these five stages for the cultural adaptation of instruments in nursing ⁵¹. Content validity should include the initial translation, the synthesis of translations, a back translation, the committee of specialists' assessment, and the pretest ⁵¹. Content assessment can be qualitative ⁵² or quantitative ⁵³, and authors should use it to adjust the content of the original instrument to the adapted version. Authors must analyze items' semantic, idiomatic, experiential/cultural, and conceptual equivalences in the qualitative assessment, as it shows the significance of dimensions, the relevance of items, and the meaning equivalence of those items. The quantitative assessments advise implementing a content analysis through the Content Validity Ratio (CVR) because it is one of the most rigorous indexes for content evaluation ⁵³.

Concerning the construct validity analysis, currently called 'internal structure' validity, authors should report data inspection techniques, analysis method, structure, extraction, factor retention, and factor loading cut-off points ^54,55. For data inspection, the application of two techniques is suggested, Kaiser-Meyer-Olkin (KMO) and the Sphericity Test, since both tools determine the adequacy of data for factor analysis ⁵⁶.

For selecting a factor analysis method, authors should have conducted an exploratory factor analysis (EFA) to determine the dimensionality of the instrument. Furthermore, authors should report necessary analysis parameters such as factor loadings of all items, total variance explained, and communalities ⁴⁹.

It is also advisable that the study reports present the confirmatory factor analysis (CFA) since it is a valuable method to confirm the theoretical model that underlies a measurement instrument. The factorial model should be evaluated using at least three adjustment coefficients ^57).

Another relevant aspect is the report of the factorial extraction method, and for this, the Optimal Implementation of Parallel Analysis (PA) is recommended, as it is the most accurate method of factor retention ⁵⁸. One study suggested resorting to more than one factor retention method since oblique rotation has been applied to psychosocial variables ⁵⁹.

The report of factor loading cut-off points is relevant since it allows evaluating the factor loadings and total variance explained of a theoretical model using factor loadings > 0.40, with a total variance explained > 60 °% and communalities > 0.40 ⁴⁹.

Study reports must describe the coefficient used to estimate reliability. Although Cronbach's alpha (α) ⁶⁰ is frequently adopted, several studies have shown limitations and assumptions of uncorrelated errors, tau equivalence (λ), and normality ⁶¹.

Other recommended analyses to measure reliability are test-retest, equivalent forms, split halves, inter-rater, and KR-20 (Kuder-Richardson 20, for instruments with dichotomous variables); similarly, other authors have suggested other techniques such as Greatest Lower Bound (GLB) or McDonald's Omega (ω) ⁶². When the total scores of the tests are normally distributed, the coefficient w should be the first option since it prevents the overestimation problems of GLB ⁶². However, when there is a low or moderate test skew, GLB should be used ⁶³.

The evaluation of invariance is becoming increasingly relevant. Therefore, studies should report its use through ΔCFI, ΔRMSEA, or other valuable indices to determine the difference between tested models so that authors can guarantee the invariance of the instrument for a given culture ⁶⁴.

Final comments

SRs are essential for health and nursing professionals because they allow the implementation of evidence-based nursing practices through rigorous knowledge synthesis methods.