DAY 64

Health & Longevity: Who Paid For It
Tracing the Money Behind Health Research

2026-07-30 · BigCat's Vitality Protocol

Day 61 was about how a study is done. Today is about who is pushing it. Funding is not fraud — but it is a measurable, traceable source of bias.

SUB · Funding Structure / Quantified Bias

Funding ≠ Supplying the Product: Ask Which Link They Hold

Money Enters the Evidence Chain at Four Points, Not One

Bottom Line

Industry-sponsored drug and device trials are significantly more likely to reach sponsor-favorable conclusions than independently funded ones. But a trial where "the company donated product only, while design and analysis stayed independent" carries far less risk. Don't veto a paper on sight of a company name — first work out which link the company holds.

Evidence Grade

Systematic review (methodological meta-research): Lundh et al., Industry sponsorship and research outcome (Cochrane, 2017 edition), pooling hundreds of studies, found risk ratios of roughly 1.27–1.34 linking industry sponsorship to favorable results and favorable conclusions.

Science & Mechanism

Bias almost never runs on fabricated data. Money enters the evidence chain at four links, and the risk is wildly uneven:

LINK 1

Paying

Low risk
decides if it happens

LINK 2

Supplying product

Moderate
sets dose and formula

LINK 3

Design + statistics

High
picks endpoints, comparators

LINK 4

Writing + submitting

Highest
how it reads, whether it runs

What actually bends the conclusion is ③ and ④, not ①

The techniques are selective design and selective publication: pick a surrogate endpoint your product wins on, choose an under-dosed active comparator, file the negative trial in a drawer. Cochrane's sharper finding is that sponsored studies show worse agreement between results and conclusions — same data, sunnier abstract.

Actionable Protocol

Sort every funding statement into one of three tiers:

Tier	What the statement looks like	Verdict
Independent funding + donated product	"Funded by the NIH; Company X supplied study product and placebo and had no role in design, analysis, or writing"	Low risk. Example: the VITAL vitamin D/omega-3 trial
Industry-funded, independent team	"Sponsored by Company X; data analyzed by an independent statistical center"	Moderate. Check whether the data custodian is truly independent
Industry across the whole chain	Company affiliations among the authors; acknowledgments thank "writing support provided by …"	High. Read it as marketing material

For Women + Common Myths

Women's angle: Menopause supplements, probiotics, collagen, cycle-based nutrition — these categories run mostly on small manufacturer-funded trials. The problem isn't that "research on women can't be trusted"; it's that independent confirmation of the same claim is thinner. The counter-example is worth remembering: the WHI hormone therapy trials were NIH-funded with the company supplying only the drug, which is precisely why their field-overturning negative results could be published at all.

Myth: "Industry-funded = fabricated." Most bias lives at the design and publication layers; the raw data usually survives scrutiny. The reverse myth is just as dangerous: assuming "government or university funded = no interests." Academia has its own stakes — promotion, faction, grant renewal.

Try This Week + A Question

TRY THIS WEEK

Take one supplement you're currently using, find the "clinical study" its website cites, read only the funding statement at the end, and decide which tier it belongs to.
Question: if every piece of evidence for a product comes from its manufacturer, how large an effect size would it take before you'd pay?

SUB · Conflict Disclosure / Self-Reporting

"No Conflicts of Interest" — and the Contradiction on the Same Page

Disclosure Is a Self-Report, Not a Switch

Bottom Line

A COI statement isn't an on/off switch; it's a self-report. The most common red flag isn't a missing disclosure — it's a contradiction on the same page: the paper declares "no conflicts of interest" while the funding line names a company and the acknowledgments thank that company's writer.

Evidence Grade

Expert consensus (ICMJE uniform disclosure requirements; CONSORT requires a separate statement of the funder's role) plus cross-sectional meta-research: multiple studies matching paper disclosures against the US CMS Open Payments database (in oncology, orthopedics and other fields) found a substantial share of industry payments went undeclared in the corresponding papers.

Science & Mechanism

ICMJE asks authors to declare honoraria, consulting fees, equity, patents, and writing support from the past three years — but enforcement runs on the honor system. Most under-reporting isn't concealment, it's a gap in the rules: the three-year window expires and the payment "resets to zero"; a grant paid to the department is judged not to be a personal interest; consulting fees arrive via a third-party medical-education firm. So the informative signal isn't the declaration itself — it's whether the three statements agree with each other.

Actionable Protocol

Read three sections at the end of the paper and ask four questions (2 minutes):

Section	What to ask
Funding	Who paid? Is the funder's role spelled out?
Conflicts of Interest	Does it contradict the funding line above it?
Acknowledgments	Any "medical writing / editorial assistance provided by" — i.e. possible ghostwriting?
Author contributions	Who holds formal analysis and writing – original draft? A company employee holding both is the strongest red flag there is

Rule: when the three sections conflict, weight the evidence by the heaviest of them, not by the declaration.

For Women + Common Myths

Women's angle: Menopausal hormone therapy, contraception, and fertility sit under commercial and ideological pressure at once, and the same dataset routinely gets half-quoted by both camps. On these topics, look one step further: who is citing it.

Myth: treating "disclosed" as "neutralized." Disclosure only informs you; the tilt at the design layer is already baked into the data. A second myth: assuming a longer disclosure is more suspicious. A long list usually just means an active author reporting honestly — the dangerous one is suspiciously short.

Try This Week + A Question

TRY THIS WEEK

Pick one health study you've shared and do exactly one thing: read the Funding section and the COI section side by side and check whether they're consistent.
Question: an author discloses 20 relationships — do you trust them more or less? Why might both instincts be wrong?

SUB · Independent Replication / Double Counting

Apparent Replication: Five Papers Are Really One Piece of Evidence

Shared Authors, Shared Cohort, Shared Upstream Source

Bottom Line

"Five studies support this" is often an illusion: those five may share the same authors, the same cohort, the same upstream dataset. Genuine replication requires an independent sample + an independent team + independent funding — miss one and it doesn't count.

Evidence Grade

Meta-research (quantified case study): the classic 1997 BMJ analysis by Tramèr et al. showed that covertly duplicated data in ondansetron antiemetic trials led meta-analysis to overestimate efficacy by about 23%.

Science & Mechanism

Evidence converges on a single source three ways: shared authors (one lab publishing serially, with consistent method preferences — and consistently propagated errors); shared cohort (one cohort spawning dozens of papers is normal practice, but counting those as dozens of independent findings in a review is double counting); and shared upstream source (many papers citing one early primary study, forming a citation cascade).

Paper A

Paper B

Paper C

Paper D

Paper E

↘ ↘ ↓ ↙ ↙

One n=42 cohort · one funder

Looks like five findings; it's one finding echoed five times

Supplements are the worst offender — the raw data usually comes from one small cohort funded by one company.

Actionable Protocol

The independence triple-check (5 minutes):
① Check author overlap: line up the author lists of 3–5 papers; an overlap of ≥ 2 people means not independent.
② Check registration and cohort: the same NCT number, or the same cohort/biobank name, means the same data; suspiciously identical sample sizes are a signal too.
③ Check the citation source: follow the references upstream two hops and see whether they converge on one original paper.
Rule: in a real replication, the authors, the sample, and the funder have all changed.

For Women + Common Myths

Women's angle: Women-specific claims (an ingredient easing premenstrual symptoms, a formula reducing hot flashes) often come from repeated analyses of one small cohort. Look for replication from a different country with a different funder rather than counting papers.

Myth: treating meta-analysis as automatically top-tier. Its ceiling is the quality of what it includes — pooling 12 double-counted industry trials just amplifies the bias and gives it the halo of the pyramid's peak.

Try This Week + A Question

TRY THIS WEEK

Pick a health claim you believe, pull the three papers behind it, and compare nothing but the author lists.
Question: when all the evidence in a field comes from two or three labs, is that "high barrier to entry" or "evidence that isn't independent"?

SUB · Working Procedure / 12-Minute Traceback

A Three-Step Funding Traceback: 12 Minutes to Filter Out Marketing-Grade Evidence

End Matter → Registry Cross-Check → Independence Test

Bottom Line

You don't need to be a methodologist. For any supplement or device study, run "end matter → registry cross-check → independence test" once and 12 minutes will filter out the overwhelming majority of marketing-grade evidence.

Evidence Grade

Expert consensus (ICMJE disclosure requirements, CONSORT reporting standards, trial pre-registration). The three steps map onto the three doors bias uses to enter the evidence chain: the self-report layer, the prior-commitment layer, and the independence layer.

Science & Mechanism

The hardest of the three is the registry layer. ClinicalTrials.gov preserves the pre-registered primary outcome and sample size, and every edit leaves a trace. Compare the primary outcome reported in the paper against the one registered up front: outcome switching can't be papered over after the fact, which makes it the single most decisive red flag available to you.

Actionable Protocol

Step	What to do	Red flags
1. End matter 2 min	Read Funding / COI / Acknowledgments and Author contributions	Who paid, who supplied, who ran the stats, who wrote it — two of the four pointing to the same company
2. Registry cross-check 5 min	Use the NCT number to check Sponsor / Collaborators / Primary Outcome on ClinicalTrials.gov; for US authors, check CMS Open Payments	Primary outcome doesn't match the registration; published years after completion; no registration at all
3. Independence test 5 min	Run the previous card's triple-check on the 3–5 papers behind the claim	Authors / cohort / funder converge on one source

End with one sentence: is this claim "n independent findings" or "one finding echoed n times"?

For Women + Common Myths

Women's angle: Pregnancy and lactation products (infant formula, prenatal supplements) are almost never tested in large independent RCTs — ethics and recruitment make it impractical — so the industry-funded share is higher. Price the thinness of the evidence itself into these decisions: "no evidence against" is not the same as safe.

Myth: "It's in a top journal, no need to check." Top journals publish industry-funded trials too (which is normal); they screen for methodological quality, not for commercial motive. A second myth: treating traceback as fault-finding. Its output isn't "believe / don't believe" — it's assigning the evidence a weight.

Try This Week + A Question

TRY THIS WEEK

Pick a supplement you're considering buying, run the full three steps, and save the verdict as one sentence in your notes — reuse it on the next decision of the same kind.
Question: if only manufacturers will fund research on a given question, should we lower the evidence bar, or accept that the question is unanswerable for now?

Going Deeper

Without industry money, half of all clinical trials wouldn't exist — where do you draw the line between dependence and independence?

The answer isn't cutting the money off, it's cutting the path of influence: the funding can come from industry, but data custody, the statistical analysis plan, and the decision to submit must sit with an independent party, with the analysis plan pre-registered. Banning commercial funding outright would simply make a great many trials disappear — that isn't cleaner, it's more ignorant.

Does mandatory disclosure create a form of "moral licensing"?

That risk is real. Behavioral economics describes a "disclosure paradox": after disclosing, advisers skew further toward their own interest because they've said it out loud, while recipients under-discount. Which is why disclosure only works alongside structural separation — independent statistics, pre-registration, data sharing.

Which funding bias is harder to spot: supplements and food, or pharma?

Usually the former. Drugs face heavy regulation: trials must be pre-registered and primary outcomes are publicly checkable. Dietary supplements are regulated as food in most jurisdictions, need no pre-market efficacy approval, and their studies are often unregistered, small, and built on biomarkers standing in for hard endpoints. The looser the regulation, the more the traceback falls to the reader.

In the age of AI search, does the citation cascade get amplified or broken?

Both forces are in play. Amplifying: models answer by corpus frequency, so a much-repeated error carries more weight, and the citation chain is erased — you can't see that five sources are one. Breaking: matching authors, cohorts, and registration numbers is exactly what machines are good at. Use AI as a traceback tool, not as the thing that reaches the conclusion for you.

← Back to Index