If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author at: Global Organization for EPA and DHA Omega-3s (GOED). 222 South Main Street, Suite 500, Salt Lake City, UT 84101, United States
EPA/DHA are an active area of research. Each month, 200–250 new articles on EPA/DHA are indexed by PubMed.
•
Despite a vast body of existing research, there remain significant gaps in our understanding of EPA/DHA health effects.
•
The volume of existing research makes it challenging to conduct systematic investigations to identify or resolve those gaps.
•
This article introduces the GOED Clinical Study Database (CSD), a comprehensive, manually curated database that catalogs this research. It also outlines the search strategies describes the methods for gathering data describes the database capabilities and current and potential applications.
Abstract
Due to their multiple mechanisms of biological action, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) have been the focus of ongoing active research for decades. In spite of the resulting body of knowledge, there remain significant gaps in our understanding of EPA/DHA health effects. Further, the volume of existing research makes it challenging to conduct systematic investigations to identify or resolve those gaps.
The purpose of this article is to introduce the GOED Clinical Study Database (CSD), a comprehensive, manually-curated relational database that catalogs this research.
Eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) are the two main omega-3 long-chain polyunsaturated fatty acids of marine origin. Their intake and high levels in blood have been linked to multiple positive health effects. The outcomes for which a positive effect has been reported include but are not limited to: reductions in all-cause mortality [
The cardiovascular benefits alone are the basis of ISSFAL's intake recommendation that healthy adults consume a minimum of 500 mg/d of EPA and DHA combined [
International Society for the Study of Fatty Acids and Lipids (June 2004). Report of the sub-committee on recommendations for intake of polyunsaturated fatty acids in healthy adults. [online] Available at: http://www.issfal.org/news- links/resources/publications/PUFAIntakeReccomdFinalReport.pdf [Accessed January 2022].
], but few adults globally reach the recommended intake. In the United States, for example, the mean intake of EPA+DHA among adults is estimated to be 113 mg/day [
U.S. adults are not meeting recommended levels for fish and omega-3 fatty acid intake: results of an analysis using observational data from NHANES 2003-2008 [published correction appears.
Nutr. J.2014; 13 (64]. Nutr J. 2014;13:31. Published 2014 Apr 2. doi:10.1186/1475-2891-13-31)
]. This includes both seafood and dietary supplements.
The underlying mechanisms responsible for these effects are diverse. EPA and DHA omega-3 fatty acids are important components of cell membranes and can regulate membrane properties [
]. As a result of these diverse mechanisms of action and health effects, EPA and DHA are among the most researched compounds in the scientific literature (Fig. 1).
Fig. 1Number of articles and articles describing randomized controlled trials indexed by PubMed. The count represents the number of articles annotated with the most used MeSH terms in the Chemicals and Drugs Category. eicosapentaenoic OR docosahexaenoic OR omega-3 OR "fish oil" OR "cod liver oil" OR "krill oil" OR "n-3PUFA" OR "n-3 PUFA".
A PubMed search including these fatty acids, and the most common marine-origin oils that contain them as ingredients for human nutrition (Fig. 2), yields in excess of 46,000 articles, with approximately 240 new articles being added each month.
Fig. 2Basic PubMed query used by the CSD to identify articles that are potentially about EPA and/or DHA.
Such a wealth of scientific research represents a great opportunity: it is possible to examine the existing literature to write systematic reviews and meta-analyses, to identify gaps in knowledge and to design new trials. At the same time, it represents a significant challenge; the number of published articles makes it difficult to keep up to date with the research and to conduct the comprehensive literature reviews and data extraction needed to take advantage of this body of knowledge.
In order to alleviate these problems and to foster further research, the Global Organization for EPA & DHA Omega-3 s (GOED) developed the Omega-3 Clinical Study Database (CSD). The dual objectives of this relational database and its accompanying graphical user interface are to catalog the existing literature on EPA and DHA, and to facilitate searches of relevant literature.
It is important to mention that while GOED derives its funding from commercial entities involved in the omega-3 industry, the CSD is a scientific and not a commercial project. All of the data stored in the database has been extracted and reviewed by independent research assistants contracted by GOED but not affiliated with the organization or its member companies. This is to ensure the database's scientific integrity: the data is extracted according to established best practices and free of potential biases derived from conflicts of interest.
The use of relational databases for sharing and disseminating research information is a well-established practice in biomedical research. The volume of current research and the development of high-throughput molecular methods and the rise of omics disciplines makes it necessary to manage ever-increasing amounts of data. The journal Nucleic Acid Research maintains a partial list of relevant databases, spanning a wide range of research fields and data types, and publishes a yearly issue dedicated to documenting the contents of and changes to this list. For the latest issue, Rigden and Fernández [
] revised the database list, adding 90 new resources and removing 86 obsolete databases, which brought the revised total to 1641 databases.
But while there is a large number of databases available, few of them are dedicated to exploring the effects of individual compounds on human health. To the best of our knowledge, the CSD is the first fully comprehensive database dedicated to the health effects of specific nutrients. A similar database exists, dedicated to dietary fibers and human health [
], but the data collection is centered on 10 pre-specified outcomes – the CSD collects data on all outcomes recorded by authors of articles on interventional trials.
A second relevant database was developed by Hooper et al. [
PUFAH group. Creation of a database to assess effects of omega-3, omega-6 and total polyunsaturated fats on health: methodology for a set of systematic reviews.
]. The authors were commissioned by the World Health Organization (WHO) Nutrition Guidance Expert Advisory Group (NUGAG) to conduct a series of meta-analyses to assess the health effects of total polyunsaturated fatty acids, omega-3 and omega-6 supplementation. The necessary literature search focused on identifying long-term (> 24 weeks) randomized controlled trials, and identified 311 completed trials, of which 221 modified the intake of long-chain omega-3 s. Data extracted from this corpus of research was the basis for the creation of their database.
One unique and remarkable aspect of this database derives from the fact that the team responsible for its creation contacted the authors of individual studies asking for information on any assessed outcomes that were not included in the final articles, as a way of reducing the risk of publication bias in systematic reviews.
One important difference between this database and the CSD derives from its authors’ deliberate choice to focus only on long-term RCTs, which is only a fraction of the existing research based on interventional trials (the CSD has identified over 3500 interventional studies in human subjects). Many important areas of research, including COVID-19 (SARS-CoV-2) treatment, peri‑operative and pregnancy outcomes or the effect of EPA/DHA on the side effects of chemotherapy require or can only be conducted using shorter interventions. Additionally, a large number of trials addressing the effect of EPA/DHA intake on biomarkers of disease risk, including blood triglycerides and blood pressure, are shorter.
A fully comprehensive database like the CSD has several potential applications, the most important of which is the ability to conduct efficient literature searches, either as a first step in conducting a systematic review, or as an aid in determining whether a particular research idea has already been investigated, or to determine the ideal target population and intervention dosage and composition for a proposed RCT. Identifying relevant literature, an important initial step for both systematic and non-systematic reviews and in interventional study design, normally requires a lengthy literature search. Since the CSD already catalogs existing articles by study type and by whether the articles are about the effects of EPA or DHA, this step is shortened considerably.
Not only does the CSD have multiple applications to omega-3 research, but importantly the relational database and user interface developed could be used, with minor modifications, for other nutrients or substances.
The CSD is available through a fee-based subscription model at https://www.goedomega3.com/omega-3-science/goed-clinical-study-database.
2. Data collection and processing
The database was populated in three stages. The first stage is automated and the last two depend on manual data curation by a group of research assistants. Details of each stage is presented in the following sections.
2.1 Stage one - Article search
The CSD executes a PubMed search using the query shown in Fig. 2. This search is conducted weekly, to keep the database updated with newly published research. Once new articles are identified, they are added to the CSD. It is important to note that the query used in this step is insufficiently specific, and that many of the articles returned are not about EPA or DHA. This is to be expected. PubMed's great strength is in the breadth of its coverage of the biomedical literature, but such coverage comes at the cost of inaccuracies in the artificial intelligence-based decisions of what each article is about. The choice to identify articles that are potentially relevant to the CSD by conducting too wide a search is deliberate and made to ensure that the CSD is comprehensive and no potentially relevant article is missed. The responsibility to decide whether an article is about EPA/DHA is determined by the research assistants according to criteria defined below.
2.2 Stage two - Initial screening
The purpose of the initial screening is two-fold. The first objective is to identify all articles that studied EPA/DHA and to classify them by study type. The second is to identify all interventional articles (including RCTs) in human subjects for which the intervention included EPA/DHA, and to flag them for subsequent extensive data extraction. In this stage, research assistants examined the title and abstract of each article to answer three questions:
1
What type of study is covered in the article?
2
Is the article about EPA/DHA?
3
In the Abstract, did the authors report a positive, negative or neutral effect of supplementation on clinical outcomes?
All articles in the database were classified by study type, according to a prespecified list that includes, among others: systematic and non-systematic reviews, meta-analyses, interventional studies, prospective or retrospective cohort and case-control studies, in vitro or animal model studies, etc. The choices are not exclusive: if an article could be classified as two or more study types, it was classified as belonging to multiple types.
The second step was to verify whether the article studied the health effects or biological action of EPA/DHA. All sources of EPA/DHA (fish oil, krill oil, algal oils, etc.) were considered. The CSD excludes studies where the only agent studied was alpha-linolenic acid (ALA). Ultimately, the goal of the CSD is to catalog what is known about the health effects of EPA/DHA in people. Although ALA can be converted to EPA and DHA in humans, the conversion is limited [
] and ALA may have health effects beyond those attributable to this conversion.
Finally, articles involving human participants were classified according to whether the authors declared in the Abstract that EPA/DHA intake or status was associated with positive, neutral or negative health effects. Only one conclusion was recorded per study. Articles where the authors concluded there were deleterious effects, or with evidence of damage or negative side effects for any outcome, were automatically labelled as negative. The remaining articles needed either a statement from the authors claiming a benefit or one positive statistically significant outcome to be labelled as positive, otherwise they were classified as neutral.
Each abstract was screened by two independent research assistants, and in case of a disagreement on any of the three criteria, a third research assistant made the final decision. This procedure, conducted using a dedicated data entry management system and graphical user interface, follows the standard best practices for literature reviews used for systematic reviews [
Higgins J.P.T., Thomas J., Chandler J., Cumpston M., Li T., Page M.J., Welch V.A. (editors). Cochrane handbook for systematic reviews of interventions version 6.2 (updated February 2021). Cochrane, 2021. Available from www.training.cochrane.org/handbook [Accessed January 2022].
Interventional articles in human subjects for which at least one of the interventions included EPA/DHA underwent detailed data extraction. The decision of which studies needed this data extraction was solely based on study type and the nature of the intervention, and was independent of other considerations including outcomes measured, author conclusions, study size and duration, EPA/DHA source or form (dietary supplements, pharmaceuticals, enriched foods, etc.).
This stage consists of two phases: data extraction and review. The data extraction phase consisted of manually entering all relevant information from each article into the database. This includes:
1
A short summary of the article based on the abstract
2
Group data including number of participants, genders and ages, their baseline characteristics and static measurements taken throughout the study
3
Study timepoints
4
The form and dose of EPA/DHA administered to participants, placebos used, and additional interventions
5
Any result or outcome measured with a quantifiable datapoint.
All data was entered to reflect the authors’ explicit statements; no assumptions were made at any stage. Static measurements were entered from tables and text. Any measurements presented only in figures were excluded so as not to estimate incorrectly. Results with ambiguous wording or unclear group/time comparisons were excluded. Only results comparing two groups at the same timepoint, or one group at two separate timepoints, were included; any comparisons involving three or more groups were excluded.
Baseline characteristics, static measurements and results were entered using a hierarchically organized controlled vocabulary. This vocabulary is an extension of the Medical Subject Heading (MeSH) thesaurus [
National Library of Medicine. Medical subject heading – home page. 2020, Available from https://www.nlm.nih.gov/mesh/meshhome.html [Accessed January 2022].
]. Additional terms were created and placed in the hierarchy as needed to accurately capture data from the articles.
After a full review was completed by the first researcher, all entered data was confirmed by a second researcher. Efforts were focused on ensuring all numbers, dosages, terms, and comparisons were correctly entered and accurately reflected the information presented in the article.
3. Graphical user interface
The relational database and accompanying graphical user interface were designed by Dayspring Partners of San Francisco, California, a digital consultancy company. The database was implemented using MySQL [
ECMA International. ECMAScript 2021 language specification. 12th Edition/June 2021. Available at https://www.ecma-international.org/wp-content/uploads/ECMA-262_12th_edition_june_2021.pdf [Accessed January 2022].
The graphical user interface is divided into three sections. The first implements the data entry logic and is used to manually populate the database with information from the relevant articles (Fig. 3).
Fig. 3Data entry form used to describe subgroups of participants within an interventional clinical trial.
The other two sections of the graphical user interface are used to mine and retrieve the information entered as part of the initial screening and full review, respectively.
The initial screening records the type of study covered by each article, and, for studies involving human participants, verifies whether the article is indeed about the health effects of EPA/DHA. It is possible to mine this data to conduct literature searches and to create curated lists of articles of certain types, and to display publicly available information about them. For example, the graphical user interface can be used to create and export lists of articles of interventional clinical trials about pregnancy, to display information about what MeSH terms are most commonly associated with them, a network of co-authors (Fig. 4), and information about the journals publishing this research.
Fig. 4Part of the co-author network for interventional articles about Pregnancy. Node size indicates number of articles and color represents the proportion of the articles that report positive/neutral/negative results. Influencers were identified using the PageRank link analysis algorithm
Brin S., Page L. The anatomy of a large-scale hypertextual Web search engine. 2020, Available at http://infolab.stanford.edu/∼backrub/google.html/[Accessed January 2022].
Full review data can be mined to find those articles that measured any desired health condition, either as part of the participant characteristics, or as a study outcome (Fig. 5). The search can be restricted to studies of certain size, or studies that use certain placebos or that use dosages in a pre-specified range. The search can also be restricted to only include studies that included participants with pre-specified characteristics (age, gender, baseline conditions, etc.).
Fig. 5Part of the user interface showing the articles on interventional trials using EPA/DHA as intervention that report measuring Myocardial Infarction and its subtypes. Node size represents the number of articles found. Elements in this result page can be used to refine the base search or to navigate to study lists, article details or graphical representations of the results found.
Because the data stored by the CSD is comprehensive and the outcomes recorded are indexed by health condition, its use can provide a faster and more complete picture of the effect of EPA/DHA supplementation on any outcome of interest. For example, searching PubMed for the effect of more than 1 g/day of EPA+DHA on systolic blood pressure requires a lengthy literature search and data extraction and, depending on the screening search parameters, some relevant articles may be missed. Because detailed data extraction for the CSD includes all articles of interventional clinical trials for which EPA and DHA are part of an intervention, the database can be used to effectively identify all relevant articles and visualize these results.
The ability to query the CSD in real time has several important scientific and commercial uses, including, but not limited to:
1
Identification of gaps in research
2
Literature searches for systematic reviews and meta-analyses, as well as for regulatory applications
3
Information gathering for scientific outreach
4
Study design
The most important scientific application of the CSD is the ability to conduct efficient literature searches. Researchers interested in investigating the effect of EPA/DHA on blood pressure can easily search for all articles that PubMed has tagged with the MeSH terms blood pressure or hypertension (or any of their descendants in the MeSH ontology) to identify the 250+ interventional studies to date (Fig. 6) and to export a list of these studies. Alternatively, they can query the database for all interventional articles that report a measurement of any of the MeSH terms Blood Pressure, Systolic; Blood Pressure, Diastolic, or Hypertension as outcomes, to find a larger list of studies (Fig. 7) – many non-cardiovascular studies report measurements of blood pressure. They can then examine the list of studies to learn about their size and duration and about the EPA/DHA dosages used (Fig. 8), and use this information to identify gaps in research or to refine the search, if needed.
Fig. 6Results of a possible search for all articles about EPA/DHA that have been tagged by PubMed as being about blood pressure, classified by study type. Only studies involving human participants are included in the results. This example search was executed on May 2022.
Fig. 8Partial list of interventional studies that measured any of a short list of terms related to blood pressure as outcomes. The study list displays the study duration and size, as well as the dosage used.
In addition to saving time in the process of conducting a literature review, another distinct use for the CSD involves the evaluation of adverse effects in a clinical trial. The COASPIRE trial [
ClinicalTrials.gov, Bethesda (MD): National Library of Medicine (US). EPA-FFA to treat hospitalised patients with COVID-19 (SARS-CoV-2) [Identifier: NCT04335032], 2020. Available at https://clinicaltrials.gov/ct2/show/NCT04335032 [Accessed January 2022].
] is a multi-site, interventional, triple-blinded, placebo-controlled study to determine whether the administration of 2 g/day of EPA results in improved outcomes for patients (n = 284) hospitalized with COVID-19 (SARS-CoV-2). During intervention, a subject experienced clinically significant changes in Alanine Aminotransferase (ALT), Aspartate Aminotransferase (AST) and hemoglobin (Hgb) levels that the site investigators classified as adverse events. The investigators asked the medical monitoring committee whether the study medication should be modified or ceased. The CSD was utilized to identify trials administering dosages > 1 g/day of EPA or total long-chain omega-3 s in which changes in ALT, AST or Hgb were analyzed. The data extracts were used by the medical monitoring committee as one input to determine the actions the trial site needed to take regarding the study medication. Importantly, the analysis was completed and data interpreted in less than eight hours, allowing for a quick decision by the medical monitoring committee.
5. Strengths and limitations
One current limitation of the CSD derives from the fact that the initial article search is performed using only PubMed. Other relevant articles may be present in other databases, like EMBASE [
Cochrane central register of controlled trials (CENTRAL), 2020. Available at https://www.cochranelibrary.com/central/about-central [Accessed January 2022].
]. While this may result in missed articles when conducting literature searches, it is unlikely that this will introduce systematic biases in the interpretation of the results for any one topic.
A second limitation is that all data contained in the database has been manually curated, and while the database contains multiple best-practice-based mechanisms to ensure the integrity and correctness of the data, errors are inevitable. The user interface contains methods to correct these errors when they are identified, and to maintain a log of these corrections.
The database will be “living,” and will be updated as new studies are published.
We believe this database is the first fully comprehensive database dedicated to the health effects of any nutrient - similar projects only cover a limited number of prespecified outcomes or restrict data extraction to a small subset of the existing interventional trials. Because the CSD identifies all articles about EPA/DHA, catalogs them according to study type, and collects data on all outcomes measured in interventional trials we are confident it will become an important tool in identifying gaps in existing knowledge and furthering omega-3 research. Additionally, not only does the CSD have multiple applications to omega-3 research, but the relational database and user interface developed can be used, with only minor modifications, for compounds, including nutrients or drugs.
The database is populated in three stages.
1
The CSD executes a (weekly) PubMed search, using a broad query string to identify any articles potentially about EPA/DHA.
2
Research Assistants (RAs) manually review the abstracts to determine which articles are indeed about EPA/DHA, and to classify them by article type.
3
For interventional articles on human subjects, RAs extract detailed information from the full text of the article. The data extracted includes the size and duration of the study, the type of agents and placebos and EPA/DHA dosage used, the baseline characteristics of participants, the outcomes measured and results reported.
To guarantee data integrity, all manual extraction steps have built-in verification and correction procedures. Each abstract is reviewed independently by two RAs (and a third in case of disagreement), and full data extraction from interventional studies is performed by a single RA, then reviewed and approved by a second.
As of May 2022, the CSD contains information about more than 46,000 articles, of which 414 are meta-analyses, 1116 systematic reviews, 5326 epidemiological studies, 3513 human interventional studies, and 15,963 studies in animal models.
The relational database is accompanied by a graphical interface to query its contents in a user-friendly manner. The ability to query the CSD in real time has several important scientific and commercial uses, including, but not limited to:
1
Identification of gaps in research
2
Literature searches for systematic reviews and meta-analyses, as well as for regulatory applications
3
Information gathering for scientific outreach
4
Study design
5
Product design and marketing
The clinical study database is available through a fee-based subscription model at https://www.goedomega3.com/omega-3-science/goed-clinical-study-database.
Declaration of Competing Interest
Aldo Bernasconi, Ph.D. and Kaitlin Roke, Ph.D. are employees of the Global Organization for EPA and DHA Omega-3 s (GOED), a 501(c)6 not-for profit trade association. GOED's goals are to increase consumption of omega-3 s to adequate levels around the world and to ensure that the industry is producing quality omega-3 products that consumers can trust. Allison M Wilkin, Ph.D. is contracted by GOED as the project manager overseeing the research assistants and full review process and has no other competing interests to declare. Adam Ismail is an employee of the KD Pharma Group SA, which is a manufacturer of active pharmaceutical ingredients and nutraceuticals containing omega-3 fatty acids. At the initiation of the project to develop the CSD, he was employed by GOED and has continued involvement in GOED as a board member. The funding for the creation of the Clinical Study Database was provided by GOED.
Acknowledgments
The authors are grateful for the thousands of hours of hard work for Abstract Review and Data Entry contributed by an incredible team of Research Assistants, who are listed on the GOED website (www.goedomega3.com). The authors would also like to thank the team at Dayspring Partners for helping this database come to life.
References
Harris W.S.
Tintle N.L.
Imamura F.
et al.
Blood n-3 fatty acid levels and total and cause-specific mortality from 17 prospective studies.
International Society for the Study of Fatty Acids and Lipids (June 2004). Report of the sub-committee on recommendations for intake of polyunsaturated fatty acids in healthy adults. [online] Available at: http://www.issfal.org/news- links/resources/publications/PUFAIntakeReccomdFinalReport.pdf [Accessed January 2022].
U.S. adults are not meeting recommended levels for fish and omega-3 fatty acid intake: results of an analysis using observational data from NHANES 2003-2008 [published correction appears.
Nutr. J.2014; 13 (64]. Nutr J. 2014;13:31. Published 2014 Apr 2. doi:10.1186/1475-2891-13-31)
PUFAH group. Creation of a database to assess effects of omega-3, omega-6 and total polyunsaturated fats on health: methodology for a set of systematic reviews.
Higgins J.P.T., Thomas J., Chandler J., Cumpston M., Li T., Page M.J., Welch V.A. (editors). Cochrane handbook for systematic reviews of interventions version 6.2 (updated February 2021). Cochrane, 2021. Available from www.training.cochrane.org/handbook [Accessed January 2022].
National Library of Medicine. Medical subject heading – home page. 2020, Available from https://www.nlm.nih.gov/mesh/meshhome.html [Accessed January 2022].
ECMA International. ECMAScript 2021 language specification. 12th Edition/June 2021. Available at https://www.ecma-international.org/wp-content/uploads/ECMA-262_12th_edition_june_2021.pdf [Accessed January 2022].
Brin S., Page L. The anatomy of a large-scale hypertextual Web search engine. 2020, Available at http://infolab.stanford.edu/∼backrub/google.html/[Accessed January 2022].
ClinicalTrials.gov, Bethesda (MD): National Library of Medicine (US). EPA-FFA to treat hospitalised patients with COVID-19 (SARS-CoV-2) [Identifier: NCT04335032], 2020. Available at https://clinicaltrials.gov/ct2/show/NCT04335032 [Accessed January 2022].
Cochrane central register of controlled trials (CENTRAL), 2020. Available at https://www.cochranelibrary.com/central/about-central [Accessed January 2022].
00070-9&id=gr1.jpg){kind=link}
00070-9&id=gr2.jpg){kind=link}
00070-9&id=gr3.jpg){kind=link}
00070-9&id=gr4.jpg){kind=link}
00070-9&id=gr5.jpg){kind=link}
00070-9&id=gr6.jpg){kind=link}
00070-9&id=gr7.jpg){kind=link}
00070-9&id=gr8.jpg){kind=link}