Industry Funding Bias in Nutrition Research — Evidence Vault.
Three peer-reviewed findings, three different decades, the same structural pattern. Lesser et al. (2007) PLOS Medicine: industry-funded nutrition studies favoured sponsor outcomes at odds ratio 7.61 (95% CI 1.27–45.73) compared with no-industry-funded studies. Sacks et al. (2020) PLOS ONE: 55.6% of articles with food-industry involvement reported findings favourable to industry, versus 9.7% of articles without industry involvement — a 5.7× relative rate. National Academy of Sciences review (2023, cited via Harvard The Nutrition Source): industry-funded studies approximately 30× more likely to report statistically significant findings in favour of the sponsor. The point estimates differ; the direction does not. This brief is the peer-reviewed evidence base for the structural critique: the manufacturer sources the ingredients, employs the scientists, and funds the research that biases the science. The findings inform the SCANSMART no-sponsorship and no-advertising position, applied across CheckIT and the I500. Bias is documented as structural rather than fraud-requiring — the Harvard Nutrition Source frames the mechanism precisely: "Funding bias exists even as scientists insist that funding does not influence their analysis and conclusions."
Stale-date reminder: re-check after the next NAS review of sponsorship influence; after the next Cochrane methodological review on funding-source-and-conclusion relationships; and after substantial Coca-Cola / industry email-disclosure releases via US Right to Know and equivalent transparency organisations. Verification method: primary peer-reviewed sources fetched and read verbatim across two passes; specific claims traced to originating papers; corrections to third-party reports flagged where source-trace produced different findings.
Industry-funded studies favour sponsor outcomes at odds ratio 7.61.
Three peer-reviewed findings, in three different decades, point at the same structural pattern. The point estimates differ; the direction does not.
| Finding | Source | Year |
|---|---|---|
| Odds ratio 7.61 (favourable vs unfavourable conclusion, all-industry vs no-industry funding, 95% CI 1.27–45.73) | Lesser et al. PLOS Medicine | 2007 |
| 5.7× more likely to report findings favourable to industry (55.6% vs 9.7%) | Sacks et al. PLOS ONE | 2020 |
| 30× more likely to report statistically significant findings in favour of the sponsor | National Academy of Sciences review (cited via Harvard Nutrition Source) | 2023 |
Odds ratio 7.61, with the precision constraint cited honestly.
Primary source. Lesser LI, Ebbeling CB, Goozner M, Wypij D, Ludwig DS. "Relationship between Funding Source and Conclusion among Nutrition-Related Scientific Articles." PLOS Medicine 2007; 4(1): e5. DOI 10.1371/journal.pmed.0040005.
Verbatim from the abstract: "The odds ratio of a favorable versus unfavorable conclusion was 7.61 (95% confidence interval 1.27 to 45.73), comparing articles with all industry funding to no industry funding."
Precision caveat. The wide confidence interval (1.27 to 45.73) reflects a small interventional-study sub-sample. The point estimate is robust enough to publish; the precision is not. When citing, name both the point estimate and the CI to preserve honesty about the precision constraint.
The 0% vs 37% split. Same paper: "For interventional studies, the proportion with unfavorable conclusions was 0% for all industry funding versus 37% for no industry funding (p = 0.009)."
The paper's own headline framing: "In 111 scientific articles on nonalcoholic beverages, articles with all industry funding were more than 7 times more likely to have favorable conclusions compared with articles with no industry funding."
5.7× more likely to report favourable findings — corrected from the third-party report's wrong "7.3%" framing.
Primary source. Sacks G, Riesenberg D, Mialon M, Dean S, Cameron AJ. "The characteristics and extent of food industry involvement in peer-reviewed research articles from 10 leading nutrition-related journals in 2018." PLOS ONE 2020; 15(12): e0243144. DOI 10.1371/journal.pone.0243144.
Verbatim from the abstract: "Of articles with food industry involvement, 55.6% reported findings favourable to relevant food industry interests, compared to 9.7% of articles without food industry involvement."
The actual relationship is 5.7× more likely (55.6 ÷ 9.7 = 5.73). A widely-circulated third-party framing of "7.3% more likely to show positive results overall" is wrong; the correct statement is "more than five times more likely" (the paper's own framing).
Additional verified figures from the same paper: 13.4% of articles in the top 10 nutrition journals (2018) had food industry involvement. Journal of Nutrition specifically: 28.3% of articles had food-industry involvement.
Bias is structural, not requiring fraud.
Primary source. Harvard T.H. Chan School of Public Health, The Nutrition Source, "Navigating Industry Funding of Research." Published 29 July 2025; modified 4 September 2025.
Verbatim: "Funding bias exists even as scientists insist that funding does not influence their analysis and conclusions."
And: "A scientist may not be aware that they are making choices that tend to slant the outcomes of a study in a particular direction."
This is the load-bearing framing for the canonical voice rule. Bias is structural, not requiring fraud; the scientist's belief that they are uninfluenced does not change the statistical pattern.
Worked structural-critique example, primary-source-verified.
Primary sources. Two separate primaries cover this:
- New York Times exposure, 9 August 2015 (Anahad O'Connor): "Coca-Cola Funds Scientists Who Shift Blame for Obesity Away From Bad Diets" — original investigative reporting.
- Serodio P, Ruskin G, McKee M, Stuckler D. "Evaluating Coca-Cola's attempts to influence public health 'in their own words': analysis of Coca-Cola emails with public health academics leading the Global Energy Balance Network." Public Health Nutrition (Cambridge), PMC10200649 — subsequent academic analysis using the email archive.
Verbatim from PMC10200649 introduction: "In 2015, the New York Times revealed that Coca-Cola funded a global network of scientists, the Global Energy Balance Network (GEBN), ostensibly to divert attention from the contribution of sugar-sweetened beverages to obesity epidemic, instead blaming inadequate exercise."
Verified specifics: Coca-Cola contributed approximately $1.5 million in start-up funding, plus ~$4 million to two GEBN co-founders. Network disbanded within months of the August 2015 NYT exposure.
Citation correction: Milbank Quarterly 2019, not PMC10200649.
Important citation correction to widely-circulated third-party reports. The actual primary source for the WHO claim is:
Maani Hessari N, Ruskin G, McKee M, Stuckler D. "Public Meets Private: Conversations Between Coca-Cola and the CDC." The Milbank Quarterly 2019; 97(1): 74–90. DOI 10.1111/1468-0009.12368. PMC6422605.
Verified specifics. A senior US Centers for Disease Control and Prevention official communicated with a former Coca-Cola executive about strategising to convince WHO to collaborate with the food industry on the same physical-activity-focused obesity messaging. The actor in the documented exchange is the CDC official, not direct Coca-Cola staff.
BMJ 2020 — the analysed correspondence.
Primary source. Griffin S. "Coca-Cola sought to shift blame for obesity by funding public health conferences, study reports." BMJ 2020; 371: m4718. DOI 10.1136/bmj.m4718. PMID 33272915.
Verbatim: "Academics in Australia and the US worked with US Right to Know, which lobbies for transparency in the food industry, to obtain and analyse emails between Coke and public health figures about events run by the International Society for Physical Activity and Health (ISPAH). They analysed 36,931 pages of documents to identify exchanges referencing Coke's sponsorship of the International Congresses on Physical Activity and Public Health (ICPAPH) held in Sydney in 2012 and Rio de Janeiro in 2014."
Top-line thesis verbatim from BMJ 2020: "The Coca-Cola Company worked with its sponsored researchers on topics to present at major international public health conferences in order to shift blame for rising obesity and diet related diseases away from its products onto physical activity and individual choice."
Important framing: the conferences (ICPAPH) are run by ISPAH. Coca-Cola sponsored the events, not owned them. Precise framing matters legally and journalistically.
What this evidence base supports.
The structural critique, peer-reviewed at citation level.
The structural reading: the manufacturer sources the ingredients, employs the scientists, and funds the research that biases the science. The verified findings provide peer-reviewed evidence for the funding-bias clause specifically: industry-funded studies favour sponsor outcomes at odds ratio 7.61 (Lesser et al. 2007); industry-involved studies report findings favourable to industry at 5.7× the rate of independent studies (Sacks et al. 2020); 30× more likely to report statistically significant findings in favour of the sponsor (NAS 2023). The structural critique is defensible at peer-reviewed citation level.
SCANSMART editorial position — no sponsorship, no advertising.
The SCANSMART editorial position prohibits sponsorship and advertising arrangements that would create the funding-bias structural risk. The peer-reviewed defence: industry funding biases nutrition research at odds-ratio 7.61 and increases favourable-finding rates approximately 5.7-fold. The position is anchored in evidence rather than first-principles preference, and is applied across CheckIT, the I500, and the Knowledge Library editorial framework.
Institutional engagement — the manufacturer-tier constraint.
SCANSMART’s institutional engagement framework constrains manufacturer-side product placements specifically because of the structural funding-bias risk documented above. The peer-reviewed evidence on funding-source-and-conclusion bias supports an editorial-integrity governance protocol distinct from a general anti-commercial posture. Manufacturer-tier engagements (where they occur) are structured to preserve editorial independence verifiable to the peer-reviewed evidence base.
UK 2026: conflict-of-interest disclosure, research-funding transparency, and editorial-integrity standards.
Industry funding bias intersects with multiple UK and international regulatory and quasi-regulatory frameworks.
| Surface | Mechanism | UK status 2026 | Upstream actor | International parallel |
|---|---|---|---|---|
| Journal disclosure-of-interest standards | ICMJE Uniform Requirements; journal-specific conflict-of-interest disclosure rules | Industry-led standards. International Committee of Medical Journal Editors (ICMJE) uniform requirements; nutrition journals' editorial policies; enforcement uneven. | Journal editors; peer-review system. | International framework via ICMJE; equivalent journal-policy regimes globally. |
| Research-grant transparency | UKRI / NIHR funding disclosure; charity-funder disclosure | Required. UK Research and Innovation, National Institute for Health and Care Research, and major UK charity funders require disclosure of co-funding sources and of conflicts of interest at grant-application and publication stages. | Funder; principal investigator. | NIH (US); ANR (France); DFG (Germany); equivalent national funding bodies globally. |
| UK political transparency | APPG (All-Party Parliamentary Group) financial-interest disclosure; lobbying register | Required. APPG funding disclosure under House of Commons Register of All-Party Parliamentary Groups; lobbying activity tracked under the Transparency of Lobbying Act 2014. | Industry trade associations; lobbying firms; political-engagement professionals. | EU Transparency Register; US lobbying disclosure (Lobbying Disclosure Act). |
| NICE Conflict-of-Interest policy | Declaration of interest framework for guideline development committee members | Required. NICE Code of Practice for Declaring and Managing Conflicts of Interest applies to all NICE committees and advisory groups. | NICE; guideline development committee members. | USPSTF (US); equivalent international guideline-development frameworks. |
| Charity and CIC governance | Charity Commission conflict-of-interest rules; CIC Regulator framework | Required. Charities and Community Interest Companies must publish conflict-of-interest policies; the framework applies to public-health charities and food-system non-profits. | Trustees; CIC directors. | Equivalent charity-regulator frameworks globally. |
| Industry pledge frameworks | International Life Sciences Institute disclosure; industry-association codes of conduct | Voluntary. ILSI and industry-association codes operate without regulatory backstop; civil-society monitoring (US Right to Know, Sustain, Action on Sugar) supplements gaps. | Industry associations; corporate sponsorship arrangements. | Similar voluntary structures globally with varying transparency. |
How to read the map. The UK has multiple required-disclosure frameworks for research funding and political engagement. The mechanisms are real but enforcement is uneven and depends substantially on civil-society monitoring (US Right to Know via FOIA-and-equivalent requests; academic transparency-focused researchers; investigative journalism). The peer-reviewed evidence on funding-source-and-conclusion bias persists even within disclosed-funding frameworks — disclosure mitigates but does not eliminate the structural bias.
Six categories of research where the funding-bias risk is most pronounced.
Three live tensions in the literature and the policy frame.
1. Disclosure does not eliminate bias.
The journal-disclosure-of-interest mechanism (ICMJE; journal-specific policies) is the standard mitigation for funding bias. Multiple peer-reviewed analyses (including Lesser 2007 and Sacks 2020) document that disclosed-funding studies continue to show the funding-source-and-conclusion bias. Disclosure is necessary but not sufficient. The honest reading: disclosure shifts the bias from invisible to visible while the underlying statistical pattern persists.
2. The "research-question selection" mechanism.
The Lesser 2007 / Sacks 2020 evidence base focuses on the conclusion-bias mechanism: industry-funded studies report sponsor-favourable conclusions at elevated rates. A separate (and possibly larger) mechanism is research-question selection: which questions get asked, which hypotheses get tested, which methodologies get funded. The Coca-Cola GEBN case is a worked example: the structural effect was not just biased conclusions in individual studies but the strategic positioning of the physical-activity hypothesis at international conferences over multiple years (the BMJ 2020 / 36,931-emails analysis). The "what questions get asked" effect is harder to quantify than the "what conclusions get reached" effect but may be the larger structural problem.
3. The scientist's perception of independence.
The Harvard Nutrition Source framing is precise: "Funding bias exists even as scientists insist that funding does not influence their analysis and conclusions." Individual scientists may genuinely believe they are uninfluenced; the statistical pattern across the literature persists regardless. This is a structural-versus-individual-attribution distinction: the bias is real at the population-of-studies level even when no individual study can be identified as fraud-affected. The peer-reviewed mitigation strategies (replication; pre-registration; independent verification; funding-source-sensitivity analysis in meta-reviews) operate at the structural level rather than at the individual-study level.
Copy-paste-ready primary sources.
- Lesser LI, Ebbeling CB, Goozner M, Wypij D, Ludwig DS. "Relationship between Funding Source and Conclusion among Nutrition-Related Scientific Articles." PLOS Medicine 2007; 4(1): e5. DOI 10.1371/journal.pmed.0040005.
- Sacks G, Riesenberg D, Mialon M, Dean S, Cameron AJ. "The characteristics and extent of food industry involvement in peer-reviewed research articles from 10 leading nutrition-related journals in 2018." PLOS ONE 2020; 15(12): e0243144. DOI 10.1371/journal.pone.0243144.
- National Academy of Sciences (NAS), 2023 review: Sponsor Influences on the Quality and Independence of Health Research — cited via Harvard Nutrition Source.
- Harvard T.H. Chan School of Public Health, The Nutrition Source, "Navigating Industry Funding of Research." Published 29 July 2025; modified 4 September 2025.
- Maani Hessari N, Ruskin G, McKee M, Stuckler D. "Public Meets Private: Conversations Between Coca-Cola and the CDC." The Milbank Quarterly 2019; 97(1): 74–90. DOI 10.1111/1468-0009.12368. PMC6422605.
- Serodio P, Ruskin G, McKee M, Stuckler D. "Evaluating Coca-Cola's attempts to influence public health 'in their own words'." Public Health Nutrition (Cambridge). PMC10200649.
- Griffin S. "Coca-Cola sought to shift blame for obesity by funding public health conferences, study reports." BMJ 2020; 371: m4718. DOI 10.1136/bmj.m4718. PMID 33272915.
- New York Times, 9 August 2015 (Anahad O'Connor): "Coca-Cola Funds Scientists Who Shift Blame for Obesity Away From Bad Diets."
- Scientific American, 16 December 2020: "Food-Industry-Backed Research Gives Results Funders Want, New Analysis Shows" (Teresa Davis interview).
What this brief does not claim.
This evidence vault contains no allegation of unlawful conduct against any specific named scientist, manufacturer, or research institution. The structural critique operates at the level of statistical patterns across the peer-reviewed literature, documented by Lesser et al. 2007, Sacks et al. 2020, and the NAS 2023 review, supported by the Harvard Nutrition Source synthesis. Where specific named cases are referenced (the Coca-Cola Global Energy Balance Network case 2015; the CDC / Coca-Cola correspondence in Maani Hessari et al. 2019 Milbank Quarterly; the ISPAH / ICPAPH email-disclosure analysis in Griffin 2020 BMJ and Serodio et al. 2023 Public Health Nutrition), references are limited to public-record peer-reviewed academic analysis and contemporaneous investigative journalism (notably the New York Times 9 August 2015 exposure by Anahad O'Connor) covering events that have been publicly documented and acknowledged by the named parties. No factual claim is made about any specific scientist's individual conduct beyond what has been peer-reviewed or publicly documented. The "scientist's perception of independence" finding (Harvard Nutrition Source: bias exists even as scientists insist that funding does not influence them) is explicitly framed as a structural-pattern observation, not as an allegation against specific named individuals.
Where to go next.
The full Knowledge Library carries five streams. The structural critique meets the frozen aisle in Frozen Food in the UK (sodium enhancement, MFA fortification gap, sulforaphane loss). The cognitive-health evidence is documented in UPF Brain & Cognitive Claims. The behaviour-change defensibility argument is in Behaviour Change & Decision-Point Capture. The FSMA gold-standard evidence-vault companions to this brief are Brand vs Manufacturer (structural transparency gap at the manufacturer end), Reformulation Tracking (time-axis manufacturer-engineering), Impulse Buying Triggers and Food Marketing to Kids (environment-side critiques), Cultural Food Myths (diaspora-community application), Canned Goods, Alcohol Labelling (SAPRO ecosystem), Bottled Water, Protein Claims (diaspora-community application), Global Staple Foods, Dietary Patterns, Carbohydrate Types, and Caffeine and Health. The label-reading mechanics are in The SCANSMART Method and Ingredient Rules; the marketing-claim register is in Nutrition Claims, Decoded; the certification ecosystem is in Symbols & Certification Marks.
Industry Funding Bias Evidence Base v1.0 (gold-standard depth) · Compiled 1 May 2026; gold-standard upgrade 11 May 2026 · Stale-date reminder: re-check after next NAS sponsorship-influence review, Cochrane methodological review on funding-source-and-conclusion bias, and major Coca-Cola / industry email-disclosure releases · Verification method: primary peer-reviewed sources fetched and read verbatim across two verification passes; specific claims traced to originating papers; corrections to third-party reports flagged where source-trace produced different findings · Citation, language, and defamation-safety discipline applied · Educational register; structural-pattern framing rather than individual-scientist allegation.