Abstract
The effect of diet on blood cholesterol concentrations has become controversial. We assessed whether industry-funded studies were more likely than non–industry-funded studies to report conclusions that were not supported by their objective findings. PubMed and Cochrane Central Register of Controlled Trials searches through March 8, 2019, yielded 211 relevant articles. The percentage of industry-funded studies increased from 0% in the 1950s to 60% for 2010 tp 2019 (P < .001). Of 94 non–industry-funded intervention studies for which the effect of egg ingestion on cholesterol concentrations could be determined, net cholesterol increases were reported in 88 (93%) studies (51% statistically significant, 21% not significant, 21% significance not reported). Among 59 industry-funded intervention studies, net cholesterol increases were reported in 51 (86%) studies (34% statistically significant, 39% not significant, and 14% significance not reported). No studies reported significant cholesterol decreases. Nonsignificant net cholesterol decreases were reported by 6 (6%) non–industry-funded and 8 (14%) industry-funded studies. However, 49% of industry-funded intervention studies reported conclusions that were discordant with study results (ie, net cholesterol increases were described as favorable in the articles’ stated conclusions), compared with 13% of non–industry-funded studies. Readers, editors, and the public should remain alert to funding sources in interpreting study findings and conclusions.
‘. . . among published studies reporting relationships between egg consumption and blood cholesterol concentrations, the number and proportion of industry-funded studies has increased over time.’
Circulating cholesterol plays a major role in cardiovascular disease, a leading cause of mortality and morbidity. 1 Although dietary cholesterol is less potent than saturated fat in its effect on blood cholesterol concentrations, its ability to elevate total and low-density lipoprotein (LDL) cholesterol was well established by 2002, when the Institute of Medicine summarized the accumulated evidence in its publication on Dietary Reference Intakes, concluding that “serum cholesterol concentrations increase with increased dietary cholesterol . . . and the relationship of serum cholesterol concentration to CHD risk or mortality increases progressively . . .” 2
Eggs are highly concentrated in cholesterol, raising concerns about effects on blood cholesterol concentrations. In recent years, the egg industry, working especially through US federally administered programs, has funded studies investigating the effects of eggs on blood cholesterol concentrations. In a 2013 review of prior intervention studies on the effects of dietary cholesterol and plasma lipoprotein profiles, 10 of the 12 included studies were funded by egg industry programs. 3 In 2015, one of the authors of that review served on the Dietary Guidelines Advisory Committee, which reported that “available evidence shows no appreciable relationship between consumption of dietary cholesterol and serum cholesterol . . .” 4 Although that statement was not carried forward in the final Guidelines, which called for eating “as little dietary cholesterol as possible . . . ,” 5 the potential for industry funding for research to influence nutrition policy decisions has become an important concern.
The current study tested the hypothesis that, among published studies reporting relationships between egg consumption and blood cholesterol concentrations, the number and proportion of industry-funded studies has increased over time. It also assessed whether study conclusions reflected their objective findings. It did not attempt to provide a summary figure for the effect of eggs or other cholesterol sources on circulating cholesterol through meta-analysis; that has been the subject of previous reviews.2,6-11
Materials and Methods
The study was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, where relevant; we note that our study’s focus on funding sources rather than on patient outcomes made some PRISMA guidelines inapplicable.
Data Sources and Search Strategy
The electronic databases PubMed and Cochrane Central Register of Controlled Trials were searched for articles published through August 2015 (the year of the Dietary Guidelines Advisory Committee report on dietary cholesterol) containing the key words “egg” and “cholesterol,” limited to studies on humans. A second PubMed search used the same strategy for articles published from September 1, 2015, through March 8, 2019. For the PubMed searches, the terms [(“egg”) AND (“cholesterol”)] were entered into the main search field, and the search was restricted to studies on humans. Analogous terms were used for the Cochrane search. From the reference lists of reviewed articles, additional articles were identified.
Study Selection
Two reviewers (BL and JMF for the first search and NB and RF for the second search) independently scanned the retrieved abstracts to identify articles, using the following inclusion criteria:
Articles reported results of observational studies, intervention studies, case reports, systematic reviews, or meta-analyses involving adult humans.
Articles reported on studies assessing intake of eggs or products containing egg yolk.
Articles reported on studies describing total or LDL cholesterol (blood, serum, or plasma) concentrations.
Articles limited to abstracts for which no full article had been published were included if they provided data sufficient for review. Publication status and publication date were not grounds for exclusion. Articles were excluded if they were limited to assessing the effects of egg white, crystalline cholesterol, or products other than eggs or if they reported data that were published in another, more comprehensive article.
For the selected studies, full-text articles were retrieved. Articles identified in the computer search were translated into English when necessary; non-English-language articles identified through reference lists were rejected.
Data Extraction and Quality Assessment
For each study, data regarding funding sources were extracted, when reported. For articles citing no direct industry funding, information on other industry associations was also noted, including financial relationships, employment, consultancies, or previous funding from the egg or poultry industries. Any disagreements were resolved by consensus. Studies were also evaluated to note whether observing changes in blood cholesterol concentrations was a key study objective.
Studies were considered “industry-funded” if they reported financial support from egg or poultry industries, companies promoting egg-derived nutritional products, or industry-promotional programs. Funding from government agencies was not considered “industry funding,” unless it was from an industry-funded program specifically intended for egg promotion. Funding from commercial sponsors that were ostensibly unaffiliated with the egg and poultry industries was not considered “industry funding,” and provision of egg products alone was not considered “industry funding.”
Studies were deemed to be intervention studies if a dietary intervention was administered to human participants and blood concentrations of total or LDL cholesterol were subsequently measured. For intervention studies, sample size, study duration, and the use of design elements to minimize bias (defined as the use of a control group, using participants as their own controls, or a crossover design) were noted.
For articles reporting intervention studies of greater than 1 day’s duration, eggs’ reported effect on blood cholesterol concentrations was noted. Study results were classified based on changes in LDL cholesterol or, if LDL cholesterol was not reported, total cholesterol, compared with a control dietary condition, or with baseline if no control condition was available.
These results were classified as “significant net decrease,” “nonsignificant net decrease or no change,” “significant net increase,” or “nonsignificant net increase.” A “net decrease” was a reduction in LDL (or total) cholesterol concentration, compared with a control dietary condition (or with baseline if no control condition was available), or a blunted rise in these concentrations, compared with a control condition. A “net increase” was an increase in LDL (or total) cholesterol concentration, compared with a control dietary condition (or with baseline if no control condition was available) or had a blunted decrease, compared with a control dietary condition.
Statistical significance was based on P values <.05 for changes in LDL cholesterol in relation to a control condition (or, in the absence of a statistical comparison to a control condition, to baseline values) or, if LDL was not reported, total cholesterol. If changes in LDL did not reach statistical significance but changes in total cholesterol did so, the results were deemed “not statistically significant.” Classification as “nonsignificant net decrease” or “nonsignificant net increase” was based solely on the direction of change compared with a control condition or with baseline values (without regard for the magnitude of change), with a P value that failed to reach the above-mentioned criterion for significance.
The reported effect (net increase or net decrease, typically based on the article’s results section and/or tables) was then compared with that described in the article’s conclusions (typically in the article abstract, discussion, or conclusion sections) to ascertain whether the author’s conclusions agreed with the objective results; these were then rated as either concordant or discordant. Author’s conclusions that eggs led to a net decrease in total or LDL cholesterol elevations, that these concentrations did not increase, or that increases were effectively and adequately counteracted by overall lipid changes (eg, reduced TC–HDL ratio) so as to pose no cardiovascular risk were deemed “favorable.” Authors’ acknowledgements that egg consumption raised LDL or total cholesterol, without the above-mentioned qualifications, were deemed “unfavorable.” Articles concluding that eggs led to a net increase in blood lipid concentrations incidentally as part of investigations on other effects of egg consumption (eg, studies of eggs’ effects on plasma lutein) were deemed “unfavorable,” provided they did not dispute or minimize the effects of eggs on lipids in their abstracts or discussion sections. Articles were excluded from the concordance analysis if the study duration was no more than 24 hours (eg, postprandial studies), if effects of eggs could not be separated from those of other interventions (eg, interventions included eggs plus fatty foods, or eggs plus an exercise regimen), if reported data were insufficient to determine the effect of egg consumption on total or LDL cholesterol, or if results diverged in study subgroups.
Statistics
Studies were grouped according to decade of publication. The Cochran-Armitage test assessed whether there was a significant trend in the proportions of industry-funded studies across successive time blocks, which were treated as ordinal. All statistical procedures were performed using SAS software, version 9.4. An α of <.05 was considered significant.
Results
The PubMed and Cochrane Central Register of Controlled Trials searches yielded 794 and 142 citations, respectively, for a total of 936 articles (Figure 1). Removal of duplicates left 843 unique articles. An additional 134 articles were identified through reference lists or other sources. Of the 977 total articles, 211 met the inclusion criteria. These included 27 observational studies, 172 intervention studies, 3 case reports, and 9 systematic reviews.
Flow diagram.
Articles’ funding status is summarized in Table 1 and Figure 2 and described in detail in eTable 1. Industry funding increased progressively from zero in the 1950s and 1960s to 60% in the years between 2010 and 2019. When the analysis was limited to studies for which an examination of effects on blood cholesterol concentrations was a key study objective, 58% of studies published between 2010 and 2019 were industry-funded. In both analyses, the increasing trends in the proportions of articles that were industry-funded over decades were highly statistically significant (P < .001).
Industry Funding of Studies on Diet and Blood Cholesterol Concentrations.
Studies were considered “industry-funded” if they received funds from egg or poultry industries, companies promoting egg-derived nutritional products, or industry-promotional programs.
Industry funding over time.
Among intervention studies, the median size of their egg-intervention participant groups was 25 participants in industry-funded studies and 18 in non–industry-funded studies. In recent decades, industry-funded studies have used somewhat smaller participant samples (median 29 participants in both the 2000s and the 2010s, compared with 31 in the 2000s and 48 in the 2010s in non–industry-funded studies). Studies have not differed greatly with regard to median study duration (6 weeks in industry-funded studies, 4 weeks in non–industry-funded studies; eTable 1). The use of appropriate controls in study design (defined as the use of a control group, using participants as their own controls, or a crossover design) was reported in 85% of industry-funded studies and 66% of non–industry-funded studies.
Of the 172 intervention studies, 19 did not permit evaluation of the effects of eggs on blood cholesterol concentrations (studies were limited to immediate postprandial effects [8], effects of eggs could not be separated from those of other interventions [5], data were insufficient to determine the effect of egg consumption on lipids [3], or results diverged in study subgroups [3]), leaving 153 (94 non–industry-funded and 59 industry-funded) studies, for which the effects of egg consumption were reported. Of 94 non–industry-funded studies, net increases in cholesterol concentrations were reported in 88 (93%) studies. Of these increases, 48 (51%) were statistically significant, 20 (21%) did not reach statistically significance, and 20 (21%) did not describe significance. Among 59 industry-funded studies, net increases in cholesterol concentrations in response to egg ingestion were reported in 51 (86%) studies. Of these, 20 (34%) were statistically significant, 23 (39%) did not reach statistically significance, and 8 (14%) did not report significance of the increase. No studies reported significant net decreases in cholesterol concentrations. Nonsignificant net cholesterol decreases were reported by 6 (6%) non–industry-funded and 8 (14%) industry-funded studies (Figure 3).
Research results.
In contrast to the similarity between industry-funded and non–industry funded studies with regard to size, duration, use of appropriate controls, and the frequency of reporting significant and nonsignificant net changes in lipid concentrations, there was a noteworthy dissimilarity between industry-funded and non–industry-funded intervention studies in their stated conclusions (listed in eTable 1) (See supplemental files). Overall, in 14 studies, results showing net decreases in cholesterol concentrations, as defined above, were described as “favorable” in the study’s conclusions (ie, they were concordant for favorable effects of eggs on cholesterol concentrations). In 98 studies, net increases in cholesterol concentrations, as defined above, were described as “unfavorable” in the study’s conclusions (ie, they were concordant for unfavorable effects). However, in 41 reports, net increases (as defined above) were described as favorable in the study’s conclusions (discordance. Table 2). Among industry-funded intervention studies, discordance was found in 49% of studies, compared with 13% of non–industry-funded studies.
Concordance Between Reported Results and Conclusions in Intervention Studies on the Effect of Eggs on Blood Cholesterol Concentrations.
Abbreviations: LDL, low-density lipoprotein; TC, total cholesterol; HDL, high-density lipoprotein.
The reported effects of eggs on blood cholesterol levels (significant net decrease, nonsignificant net decrease, significant net increase, or nonsignificant net increase) were compared with authors’ conclusions. Authors’ conclusions that eggs led to a net decrease in total or LDL cholesterol elevations, that these concentrations did not increase, or that increases were effectively and adequately counteracted by overall lipid changes (eg, reduced TC–HDL ratio) so as to pose no cardiovascular risk were deemed “favorable.” Authors’ acknowledgements that eggs elevated LDL or total cholesterol, without the above qualifications, were deemed “unfavorable.” The reported effects were then compared with each article’s conclusions and were rated as either concordant or discordant.
In most cases, discordance related to differences in interpretation of net increases cholesterol concentrations that failed to reach statistical significance; some studies reported these findings objectively as nonsignificant changes; others reported them as an absence of changes. An example comes from a 2014 study of college freshman. The addition of 2 eggs at breakfast, 5 days a week over 14 weeks, was associated with a mean LDL cholesterol increase of 15 mg/dL (0.38 mmol/L)—a change that did not reach statistical significance. Rather than concluding that the 15 mg/dL increase did not reach statistical significance and therefore a chance finding could not be ruled out, the investigators concluded that the “additional 400 mg/day of dietary cholesterol did not negatively impact blood lipids.” 12 Similarly, in a 2014 study in the Netherlands, a 1-year trial testing a beverage containing 1.5 egg yolks daily found that mean LDL cholesterol concentration increased by 9.3 mg/dL (0.24 mmol/L), compared with a rise of 3.5 mg/dL (0.09 mmol/L) in a control group consuming an egg-free beverage; the result did not reach statistical significance. Rather than concluding that the difference in the increases in LDL cholesterol did not reach statistical significance, the investigators concluded that changes in LDL cholesterol “were not different between the 2 groups.” 13
There were no dramatic changes over time in either the degree of discordance or the difference in percentages between industry-funded and non–industry-funded studies (Table 2).
Discussion
The present study has yielded 2 primary findings.
First, industry involvement in research on the effect of eggs on blood cholesterol concentrations increased greatly from 0% of studies in the 1950s and 1960s to 60% for the years 2010-2019.
Second, although the vast majority of studies reported either significant or nonsignificant net increases in blood cholesterol as a result of egg-feeding, industry-funded articles reporting the results of intervention studies were more likely than non–industry-funded articles to portray the effects of egg consumption in a positive or neutral light that did not fully reflect study results.
The effect of eggs and of dietary cholesterol on blood cholesterol concentrations has been examined in several reviews and meta-analyses.2,6-11 In a 2018 meta-analysis based on 28 randomized controlled intervention studies, egg consumption overall elevated both total cholesterol and LDL cholesterol concentrations by 5.6 mg/dL and HDL cholesterol by 2.1 mg/dL (all P < .0001), compared with control groups. 10 The dose-response relationship between dietary and circulating cholesterol was examined in a 2019 meta-analysis of intervention studies, finding that a 100 mg/day increase in dietary cholesterol was associated with increases in circulating LDL cholesterol concentrations ranging from 1.90 to 4.58 mg/dL, depending on the choice of model. The association differed depending on baseline cholesterol concentration. At baseline LDL cholesterol concentrations of 100, 125, 150, and 175 mg/dL, a 100 mg/day increase in dietary cholesterol was predicted to increase LDL cholesterol concentration by 2.7, 3.6, 4.6, and 5.5 mg/dL, respectively. 11
Interventions that influence LDL cholesterol have been shown to change the risk of major vascular events. In a meta-analysis of 49 intervention studies including 312 175 participants with 39 645 vascular events, the relative risk for major vascular events per 38.7 mg/dL reduction in LDL cholesterol was 0.77 (P < .001) for statins and 0.75 for dietary changes, bile acid sequestrants, ileal bypass, and ezetimibe (P = .002). 14
In recent years, a major funder of research on eggs is the American Egg Board, a federally authorized, industry-funded program whose mission is “to increase demand for eggs and egg products through research, education and promotion.” 15 The American Egg Board’s research funding is administered by the Egg Nutrition Center. Other countries, notably the United Kingdom and Australia, have similar entities that fund research for the purpose of promoting egg consumption.
The massive increase in industry funding in recent years has not necessarily altered the objective findings of studies on the effects of dietary cholesterol. However, the interpretation of those changes may have been affected by funding source. We found discordance between research results and study conclusions in 49% of industry-funded studies, compared with 13% of non–industry-funded studies. In many cases the issue relates to conclusions regarding changes that did not reach statistical significance. It is important to differentiate between the absence of a statistically significant difference and the absence of a difference.
As industry funding has increased, the conclusions of research reviews may have been influenced. In a 1992 meta-analysis, 29% of the 41 cited studies on dietary cholesterol had been paid for by industry. It reported that serum cholesterol concentration “is clearly increased by added dietary cholesterol but the magnitude of predicted change is modulated by baseline dietary cholesterol.” 7
Two decades later, a 2013 review based its findings solely on studies published since 2003. 3 Of the 12 included studies, 10 were funded by the egg industry. In contrast to the 1992 review, the 2013 review’s conclusions were much more muted, concluding that the effect of dietary cholesterol is “modest and appears to be limited to population subgroups.”
Similarly, a 2013 review attributed to the American Heart Association and American College of Cardiology excluded studies published prior to 1998. In addressing the issue of the effect of dietary cholesterol on blood cholesterol concentrations, the review noted the absence of adequate studies during this frame and concluded, “There are insufficient data to make a statement.” 16 By disregarding studies published prior to 2000, authors of reviews may have relied to a large degree on industry-funded research, and to the extent they rely on study conclusions, rather than study data, they may neglect the effects of eggs and other sources of dietary cholesterol.
Beyond the potential effect of industry funding on the integrity of individual research studies, the publications that flow from them, and later reviews, the possibility that industry funding may also influence the judgment of funded scientists and their colleagues at funded institutions on related matters merits consideration. In that regard, Tufts University researchers were funded by the Egg Nutrition Center to conduct a meta-analysis on dietary cholesterol and cardiovascular disease, in which 15 of 18 included studies had been industry-funded, at the same time that 2 individuals from the same university were actively serving on the Dietary Guidelines Advisory Committee and tasked to objectively comment this scientific issue.4,8
The effect of industry funding on health research has been the subject of prior studies. A 2007 study found a significant relationship between industry funding sources and conclusions in nutrition-related intervention studies. 17 Unfavorable conclusions were found in 37% of studies with no industry funding, compared with 0% for industry-funded studies (P = .009). In contrast, a 2012 study examining articles on dairy products and obesity detected no biases based on funding source. 18 An investigation published in JAMA Internal Medicine quoted correspondence and other documentation demonstrating the sugar industry’s influence on reported scientific findings. 19 An accompanying commentary provided a historic overview of the efforts of other food industry entities to skew research in their favor. 20
The present study has limitations. Its scope was limited to assessing the change in frequency of industry funding over time and the degree to which study results were reflected in reported conclusions. It did not aim to provide a summary figure for the effects of eggs on blood cholesterol concentrations via meta-analysis. Heterogeneity in study designs, sample sizes, baseline characteristics, choice of comparator, and other variables make meta-analyses in this area perilous unless they are restricted to a small number of homogeneous studies. Nonetheless, several prior meta-analyses have been conducted and, regardless of funding source, have reported that eggs and other sources of dietary cholesterol increase blood cholesterol concentrations.2,8-13
We note that, if eggs had no effect on plasma lipid concentrations, results might be expected to be roughly evenly split between those showing lipid elevations and reductions. Specifically, if an infinite number of studies with identical study design and study size were conducted and results were normally distributed, they would be expected to fall evenly as mean elevations and reductions. However, the vast majority of intervention studies reported elevations of plasma lipids associated with egg consumption, many of which reached statistical significance, while no articles reported statistically significant reductions in plasma lipids. However, for the reasons given above, we did not believe that a new meta-analysis was justified.
Our determinations regarding which studies received industry funding were conservative. Studies by researchers who had received industry funding for prior studies but not specifically for the study in question or who did not declare industry funding (eg, in an abstract for which no fuller report was available) were classified as “not industry funded.”
In conclusion, industry has played an increasing role in funding research on the effects of eggs on blood cholesterol concentrations and has become the dominant funding source in this area. While the vast majority of studies reported that egg consumption increased blood cholesterol concentrations to either a statistically significant or nonsignificant degree, industry-funded intervention studies were more likely than non–industry-funded studies to report conclusions that did not fully reflect study results. Discordance was found in 49% of industry-funded intervention studies, compared with 13% of non–industry-funded intervention studies. Readers, editors, and the public should remain alert to funding sources in interpreting study conclusions.
Supplemental Material
eTable_1_Egg_funding_data_2019_09_20 – Supplemental material for Industry Funding and Cholesterol Research: A Systematic Review
Supplemental material, eTable_1_Egg_funding_data_2019_09_20 for Industry Funding and Cholesterol Research: A Systematic Review by Neal D. Barnard, M. Blaire Long, Jennifer M. Ferguson, Rosendo Flores and Hana Kahleova in American Journal of Lifestyle Medicine
Footnotes
Acknowledgements
We thank Richard Holubkov for handling statistical operations.
Author Contributions
Dr Barnard had full access to all study data and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Dr Barnard
Acquisition, analysis, or interpretation of data: All authors
Drafting of the manuscript: Dr Barnard
Critical revision of the manuscript for important intellectual content: All authors
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr Barnard is the uncompensated President of the Physicians Committee for Responsible Medicine and the Barnard Medical Center, organizations that promote the use of low-fat, plant-based diets and discourage the use of animal-derived, fatty, and sugary foods. He also writes books and articles and gives lectures on nutrition, and has received royalties and honoraria from these sources. M. Blaire Long is employed by Integrated Rehabilitation Consultants and Jennifer M. Ferguson is employed by Trident Medical Center, Charleston, SC. Mr Flores and Dr Kahleova are employees of the Physicians Committee for Responsible Medicine. Dr Long, Dr Ferguson, Dr Kahleova, and Mr Flores received compensation from the Physicians Committee for Responsible Medicine for their work on this research study. There are no patents, products in development, or marketed products to declare.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: There was no external funding for this project. However, Dr Long, Dr Ferguson, Dr Kahleova, and Mr Flores received compensation from the Physicians Committee for Responsible Medicine for their work on this research study. M. Blaire Long is employed by Integrated Rehabilitation Consultants and Jennifer M. Ferguson is employed by Trident Medical Center, but neither of these organizations provided support for work done on this project, nor did these organizations have any role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Ethical Approval
Not applicable, because this article does not contain any studies with human or animal subjects.
Informed Consent
Not applicable, because this article does not contain any studies with human or animal subjects.
Trial Registration
Not applicable, because this article does not contain any clinical trials.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
