|
Introduction Epistemonikos is an open access, multilingual, health decision-making tool that is freely available to anyone with internet access. The website is maintained by the Epistemonikos Foundation, a nonprofit organization based in Santiago, Chile. Originally called the Epistemonikos Database, it was founded in 2012 with the admirable goal of creating an easy to use and accessible global repository containing the best available evidence in healthcare. It purports to be a “living project” with collaborators from all over the world helping users to find, appraise, and disseminate health-related knowledge. Content Epistemonikos is intended for use by health professionals, researchers, and health decision-makers that need rapid and reliable access to quality scientific information. Their website states that Epistemonikos is the “largest source of systematic reviews” and attempts to provide the best available scientific evidence for real-life medical questions. Content is found by regularly screening multiple databases, including but not limited to the Cochrane Database of Systematic Reviews, PubMed, EMBASE, CINAHL, and PsycINFO to locate systematic reviews on health topics. At present, Epistemonikos indexes over 100,000 systematic reviews and hundreds of thousands of individual studies for health decision-making. In addition to systematic reviews, it also provides the studies within those reviews, and an overview of the reviews to interlink all relevant evidence in a convenient fashion. Available content largely includes journal articles and pre-prints, which fall into color-coded content icons including broad synthesis (green), primary studies (red), and systemic reviews (blue). Epistemonikos shares an explicit process and PRISMA chart for the database’s inclusion criteria. The resource offers transparency on how eligible systematic reviews are located, revealing that both humans and “robots,” or machine learning tools, select reviews following the inclusion and exclusion criteria. The “robots” also help the humans along with assessing the full-text of reviews to identify their included studies. Then, they sort the relevant content that answers the same PICO question before synthesizing relevant material into a matrix of evidence table. Features/Functionality The layout interface is refreshingly uncluttered and clean with a straightforward, no-frills Google-esque landing search page. There is a single free text search box on the home page for composing simple searches. An advanced search link appears below the main search box for more control to create traditional Boolean logic searches and to search by title, abstract, title/abstract, or authors. 
Image 1: Advanced search landing page example (Credit: Epistemonikos) While the user interface is clear and offers basic filters like publication year and publication type, less control is offered in terms of how results appear in the list. There is no way to sort the order in which results appear. According to the How To page, “Results will be sorted by relevance to your query (i.e. how well represented in the article are the terms you entered), which is calculated by an algorithm developed by Epistemonikos team.” Search results can be exported as an RIS file for citation management tools like EndNote or Zotero. When reading a systematic review entry, users are provided with basic abstract information, including links to the article in PubMed or its DOI. What is unique to Epistemonikos is that they include resources with evidence related to the article broken down into primary studies, systematic reviews, and broad syntheses. Other languages are also provided when possible. 
Image 2: Systematic review entry example (Credit: Epistemonikos) The entry for the primary studies linked to a systematic are clear to read and sort through for further context. 
Image 3: Example of primary study list from the systematic review (Credit: Epistemonikos) It is more challenging to produce a search that resulted in a matrix of evidence than expected. When located, it can help provide context about how frequently an article is used when researching a topic. The intent with the interconnected entries is that all of the available evidence is connected so as to click into all available evidence related to the same question or topic. 
Image 4: A matrix of evidence table (Credit: Epistemonikos) The Help features are useful to explain things in a simplistic, clear fashion. The simple search quick guide is easy to follow with small moving GIFs to demonstrate functionalities. The transparent nature of the entire website is an asset. For example, it lists the date when a search was completed as well as how papers are selected. Articles are clearly color-coded with icons to make the type of paper clear to users. An orange lightning icon appears on a paper when the study was reviewed by a machine and indicates human confirmation is “pending.” Lastly, it is helpful and straightforward to export up to 100 articles as an RIS file for importing to citation management tools. While there are many excellent features, there is room for improvement. For example, when searching more simplistically, results were somewhat scattered and many not relevant in scope to the search topic. The inability to sort the order of results and rely on a vague internal algorithm is somewhat frustrating at times when producing large amounts of results. It would be beneficial to offer the option to exclude results for pre-prints or those that have not gone through a human review process. It appears as if most of the robotically-crafted reviews do not contain any of linked or related primary studies. At present, the advanced search feature is only accessible in English and evidence matrixes are found sparingly across various topics. Expanding available searching languages and evidence matrix coverage will be helpful. According to Epistemonikos, developing new versions of their databases is in the works, including a database of randomized trials and an updated version of the database of systematic reviews and other types of evidence synthesis. These new databases are purported to “offer a modern and enhanced advanced search experience, addressing known issues and incorporating numerous new features, including Epistemonikos’ Evidence Terms, the controlled vocabulary developed by the Epistemonikos Foundation.” So, it seems like they are aware of some of the existing shortcomings and are looking to address them. The timeline for these changes is unclear. Business Model This resource is open access, making it freely available to anyone with an internet connection. Anyone can search without logging in, though there is an option to create an account. When signing up for a personal account there is an ample terms and conditions agreement along with a privacy policy statement. Not requiring a login is a benefit to some who may not want to give up privacy rights. Breakthrough One of the more unique aspects and greatest value of Epistemonikos to the health information community is that it offers multilingual text when available. At present, they hold nine languages, including Arabic, Chinese, Dutch, English, French, German, Italian, Spanish, and Portuguese. Because it is open access and focused primarily on answering health-related questions with higher levels of evidence like systematic reviews, it could be an alternative resource for smaller libraries or labs that may not be able purchase expensive platforms like Cochrane. However, any full-text article would have to be interlibrary loaned. The easy to access links present evidence and display a cross-tabulation of reviews and studies when available. Overall, Epistemonikos offers a fine open access resource to locate and connect health-related questions to bigger questions.
|
