How to Tell a Good Clinical Study from a Bad One

You find a supplement with a clinical trial attached. You buy it. Six months later, someone tells you the study had 22 people in it, ran for eight weeks, and was funded by the company that makes the product.

That is not a fringe case. That is Tuesday in the longevity supplement industry.

The problem is not that you were not paying attention. It is that nobody told you which parts of a study actually matter and which parts are noise dressed up as science. A journal name does not tell you that. A p-value does not tell you that. Neither does "double-blind, placebo-controlled" on a product page, which can mean almost anything depending on what you do not read (1).

This article gives you eight specific things to check. Not a methodology course. Eight checks, in plain language, that take less than five minutes per study and will immediately change how you read any health claim.

Check 1: What Kind of Study Is It?

This is the most important question and the one most product pages skip entirely. Studies are not interchangeable. A mouse study and a human clinical trial are not different points on the same scale. They are answers to completely different questions.

The randomised controlled trial (RCT) is the strongest study design for assessing causal effects in humans. Participants are randomly assigned to the intervention or a control, which distributes confounding variables equally across both groups and makes causal inference possible (2). High-quality systematic reviews and meta-analyses of RCTs sit at the top of the evidence hierarchy (3).

Animal and cell-based (in vitro) studies provide early, preliminary evidence but have limited direct applicability to humans. Rapamycin, resveratrol, and dozens of other compounds extended lifespan in mice and showed far weaker effects in human trials. Animal data is a reason to keep watching, not a reason to act (4). For a detailed breakdown of how this plays out across popular longevity compounds, see our overview of supplements claimed for anti-aging.

Check 2: How Many People Were in It?

Small studies are structurally unreliable. A study's ability to detect a real effect depends on its statistical power, which is driven by participant numbers. Too few people and the study misses genuine effects, or produces impressive-looking results by chance that collapse when someone tests them in a larger population (4).

Many supplement trials involve 20 to 60 people over 8 to 12 weeks. That is enough to generate a press release. As a working rule, treat small RCTs with limited participant numbers cautiously, especially if they have not been independently replicated. Also watch what "statistically significant" actually means: a a p-value below 0.05 indicates that the observed result would be unlikely under the assumption that there is no true effect.. It says nothing about whether the effect is large enough to matter. Confidence intervals are more useful — a wide interval or one that nearly touches zero should make you cautious regardless of the p-value (5).

Check 3: Did Anyone Know Who Was Taking What?

In a double-blind trial, neither the participants nor the researchers know who received the active compound and who received the placebo. This matters more than most people realise, because the placebo effect is real and biologically measurable. People who believe they are taking an active compound report genuine improvements in energy, focus, mood, and sleep regardless of what is actually in the capsule (5).

Open-label trials, where everyone knows the allocation, produce systematically inflated results for subjective outcomes. When a trial reports improved energy or cognitive clarity without double-blinding, the finding is difficult to interpret, and that limitation is worth looking for on any product page citing the research.

Check 4: Could Something Else Explain the Result?

Confounding is why observational research is so hard to act on. A confounder is a third variable that is associated with both the thing being studied and the outcome, making two things look connected when they are not directly causing each other (6).

In longevity research this is pervasive. People who supplement with NMN or NR also tend to exercise regularly, prioritise sleep, eat well, and have higher incomes. Each of those factors independently predicts almost every health outcome researchers measure. A study that finds NMN users have better metabolic markers without rigorously accounting for all of that is probably detecting something about the kind of person who takes NMN, not what NMN does.

Check 5: Are You Seeing All the Studies, or Just the Ones That Worked?

Journals publish positive results far more readily than null results (7). Studies that find no effect tend to stay unpublished, so the literature you can access systematically overstates how well interventions work. If ten groups test a compound and two find a result by chance, but only those two get published, the public record looks like solid backing.

The corrective is pre-registration: researchers declare their hypothesis, primary outcomes, and analysis plan on a public registry like ClinicalTrials.gov before collecting any data. Pre-registered trials produce substantially fewer inflated positive findings (8). It takes 90 seconds to check whether a trial was pre-registered and whether the published outcomes match what was originally specified. That check alone filters out a significant proportion of unreliable findings.

Check 6: Did They Measure Something That Actually Matters?

Most longevity supplement trials do not measure lifespan or disease incidence. They measure surrogate endpoints: biomarkers that are assumed to proxy for outcomes that take years to appear. NAD+ levels, telomere length, inflammatory markers, insulin sensitivity. These are informative signals. They are not the same as the outcome.

Improving a surrogate does not guarantee improvement in the underlying outcome. Several drugs with strong surrogate data have failed or caused harm at the clinical outcome level (9). A supplement that shifts a biomarker and one shown to reduce disease risk are different claims, even when both appear on the same product page. Our clinician's perspective on NMN walks through exactly this distinction using one of the most studied longevity compounds as a worked example.

Check 7: Who Paid for It?

Industry-funded nutrition and supplement studiesare significantly more likely to report results favourable to the sponsor (10). Commercially sponsored research is not automatically wrong. It is a signal that warrants additional scrutiny, particularly when a finding has not been replicated by a group with no financial stake in the outcome.

Funding disclosures appear at the end of most peer-reviewed papers. Check whether the sponsor had a role in data analysis or manuscript preparation, and if the only evidence for a compound comes from trials the manufacturer paid for, treat it as a reason to wait for independent replication before acting. We applied this lens directly when reviewing the NMN evidence base — including the conflicts of interest disclosed in several of the earliest human trials.

Check 8: Has Anyone Else Found the Same Thing?

A single well-designed RCT is a starting point, not a conclusion. The findings that hold up across independent research groups, different populations, and different study designs carry a qualitatively different kind of weight (4). Omega-3 fatty acids, creatine, and vitamin D in deficient populations are among the most extensively studied compounds, with findings replicated across multiple research groups.. The process of finding a well-replicated omega-3 supplement that meets this bar is a useful illustration of how replication shapes real purchasing decisions.

When a supplement is promoted on one trial alone, the absence of replication is itself data. The finding may be recent. Independent attempts may also have produced null results that attracted less attention than the original.

The Eight Checks at a Glance

• What type of study is this: RCT, observational, animal, or in vitro?
• How many participants, and how long did it run?
• Were participants and researchers blinded to the allocation?
• Which confounders were measured and adjusted for?
• Was the trial pre-registered, and do published outcomes match the original plan?
• Did the study measure a clinical outcome or a surrogate biomarker?
• Who funded the research, and did the sponsor have a role in analysis or reporting?
• Has the finding been independently replicated?

How Augment Life Uses This Framework

Every ingredient in Augment Life formulations is evaluated against these eight criteria. The threshold for inclusion is human clinical evidence from double-blind, placebo-controlled trials at doses matching those used in the research. Where the evidence is strong and consistently replicated across independent groups, that is stated. Where it rests on animal data, small trials, or a single unreplicated finding, that is stated too. The goal is an accurate account of what the research shows, including its limits.

A Final Note

Longevity research is moving faster than the quality controls that govern more established fields. These eight checks are not a guarantee against being misled. They are a significant upgrade over reading the abstract and scanning for the words "clinical trial." The gap between those two approaches is where most bad supplement decisions happen.

Read More From Augment Life

• Supplements Claimed for Anti-Aging: Results, Claims and Potential Risks
• Nicotinamide Mononucleotide (NMN): A Clinician's Perspective
• The Longevity Debate: Can Supplements Replace Exercise?
• The Role of Autophagy in Health and Longevity
• Your Cells Can Build New Mitochondria
• What Is an Optimal Longevity Diet?

Literature Sources

1. Sacks G et al. Conflicts of interest and the quality of recommendations in clinical guidelines. PLOS Medicine. 2016. doi: 10.1371/journal.pmed.1001798
2. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010. doi: 10.1136/bmj.c332
3. Higgins JPT, Thomas J (eds). Cochrane Handbook for Systematic Reviews of Interventions. Version 6.4. Cochrane, 2023.
4. Ioannidis JPA. Why most published research findings are false. PLOS Medicine. 2005. doi: 10.1371/journal.pmed.0020124
5. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology. 3rd ed. Lippincott Williams & Wilkins; 2008.
6. Rothman KJ. Epidemiology: An Introduction. 2nd ed. Oxford University Press; 2012.
7. Turner EH et al. Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med. 2008. doi: 10.1056/NEJMsa065779
8. Chan AW et al. Empirical evidence for selective reporting of outcomes in randomised trials. JAMA. 2004. doi: 10.1001/jama.291.20.2457
9. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann Intern Med. 1996. doi: 10.7326/0003-4819-125-7-199610010-00011
10. Lundh A et al. Industry sponsorship and research outcome: a Cochrane review. Cochrane Database Syst Rev. 2017. doi: 10.1002/14651858.MR000033.pub3