How to Read a Clinical Trial: A Primer for Non-Scientists

This content is educational and explains research methodology. It is not medical advice. Decisions about your own health belong with a qualified healthcare professional.

Why this matters

Consumer health content is saturated with study references. Almost every supplement page, podcast episode, and wellness video cites "research" — usually one or two studies, almost always interpreted in the most favourable possible light. Most readers have no way to evaluate whether the study cited actually supports the claim being made.

This page is a primer. It does not require a science degree. By the end you will be able to look at a clinical trial summary and ask the right questions: who was studied, what was measured, what was found, and whether the finding actually supports the claim someone is making.

Proco's editorial position is that the gap between what published research describes and how that research is then communicated is the single biggest problem in consumer health information. Closing that gap requires both us doing the translation work carefully and readers being able to push back when they see translation done badly.

The hierarchy of evidence

Not all evidence is equal. Researchers describe a rough hierarchy:

The hierarchy is rough — a well-conducted cohort study can be more informative than a poorly conducted RCT, and meta-analyses are only as good as the trials they include. But the hierarchy is a useful starting frame.

When you see a health claim, the first question is: what type of evidence does it rest on? "A trial" usually means an RCT. "Studies show" can mean almost anything. "Research suggests" often means a single observational paper.

The anatomy of a clinical trial: PICO

Researchers use the acronym PICO to structure trial questions:

• Population: Who was studied?
• Intervention: What did they receive?
• Comparator: What was the control?
• Outcome: What was measured?

Reading any trial means asking these four questions and getting clear answers.

Population

Who was studied? "Healthy adults aged 18-45" is a different population from "postmenopausal women with osteoporosis." A finding in one population doesn't automatically generalise to the other.

Common population traps in popular coverage:

• Trial run in mice, reported as if applicable to humans
• Trial run in athletes, reported as applicable to sedentary readers
• Trial run in one age group, reported without age qualification
• Trial run in patients with a specific condition, reported as applicable to people without it
• Trial run in one country, where dietary and environmental baselines differ from other populations

Intervention

What did participants receive? Dose, form, frequency, duration. A trial of 5,000 IU vitamin D for 6 months in a deficient population tells you something different from a trial of 1,000 IU for 4 weeks in a replete population.

Comparator

What were participants compared with? A placebo? An existing treatment? Nothing? "Improved sleep by 20%" means little without knowing what the comparison was.

Common comparator traps:

• "No control group" — observational, not causal
• "Active control" — comparison against an existing intervention, not placebo
• "Historical control" — comparison against past data, not a contemporaneous group
• "Self-comparison" — before-and-after in the same person, vulnerable to regression to the mean

Outcome

What was measured? "Improved cognitive function" can mean a 2% improvement on one specific reaction-time test. "Reduced inflammation" might mean a statistically detectable change in a single biomarker.

Surrogate outcomes (biomarkers, lab values, scan findings) are often easier to measure than the outcomes that matter to people (disease, death, quality of life). A trial that shows a supplement lowers a biomarker is suggestive; a trial that shows it reduces disease incidence is decisive. Most consumer health claims rest on surrogate outcomes.

Sample size and statistical power

How many people were studied? A trial with 20 participants is doing something different from a trial with 20,000.

Small trials can detect very large effects but routinely miss medium and small ones. Small trials are also more vulnerable to chance findings — random variation can produce statistically "significant" results that don't replicate.

When you see a positive finding from a trial with fewer than ~100 participants, the right response is interest, not conviction. Replication in larger trials is what moves a finding from suggestive to established.

P-values: what they mean (and don't)

The p-value is the most misunderstood number in research.

A p-value is the probability of observing the data you have (or more extreme data) if there were actually no effect. A p-value of 0.05 means "if there were no real effect, we'd see results this extreme 5% of the time by chance alone."

The conventional threshold for "statistically significant" is p < 0.05. This is a convention, not a law.

A p-value does NOT tell you:

• The probability that the finding is real
• The probability that the finding would replicate
• The size of the effect
• The practical importance of the finding
• The probability that the alternative hypothesis is true

A small p-value with a tiny effect size on a surrogate marker is consistent with "we detected a real but trivial change in a number that may or may not matter." Communicate that as "significant improvement" at your peril.

Effect size matters more than significance

This is the most important section on this page.

A statistically significant finding can be trivially small in real-world terms. A non-significant finding can still describe a meaningful trend. The two are different questions.

When reading a study, look for the effect size — the actual magnitude of the change reported. Sleep onset shortened by 17 minutes is a different finding from sleep onset shortened by 3 minutes, even if both are "statistically significant."

Researchers report effect sizes in several standardised ways: Cohen's d (for continuous outcomes), hazard ratios (for time-to-event outcomes), odds ratios and relative risk (for binary outcomes), mean differences. Each has interpretive conventions.

Rough rule of thumb: when a popular article says "a study found X improves Y," the next question is "by how much."

Conflicts of interest

Who funded the trial? Who designed it? Who employs the authors?

Industry-funded studies of industry products consistently report more favourable findings than independently funded studies of the same products. This is one of the most replicated findings in research methodology [Lundh et al. 2017 Cochrane review].

This doesn't mean industry-funded studies are wrong. It means they require extra scrutiny. Look for:

• Funding disclosure
• Author affiliations
• Whether the funder had a role in study design, data analysis, or publication decision
• Independent replication

A finding that's been replicated by independent groups in different settings is much stronger than a single industry-funded trial, regardless of statistical significance.

Pre-registration and selective reporting

A trial that was registered before it began — typically on ClinicalTrials.gov or a similar registry — has committed in advance to its primary outcome, sample size, and analysis plan. This makes it harder to fish for positive findings after the data is collected.

Trials that were not pre-registered, or whose published primary outcome differs from the registered one, deserve more scrutiny. Selective reporting is a known problem: when researchers measure many outcomes, some will appear "significant" by chance alone, and there is pressure (career, financial, publication) to report those rather than the registered ones.

This is why systematic reviews and meta-analyses focus on pre-registered trials when possible.

Red flags in popular coverage

When a popular article describes a study, watch for:

• No link to the source. If you can't find the paper, you can't check the claim
• No reporting of effect size. "Significantly improved" without numbers
• No reporting of sample size. "Studies show" without N
• Animal or in-vitro studies described as human findings
• Conflation of correlation with causation ("People who do X live longer" ≠ "X makes people live longer")
• Single-study coverage as if definitive. Almost no field is settled by one trial
• No mention of replication or systematic reviews
• Author has a commercial interest in the conclusion
• The headline is much stronger than the abstract

These are not absolute rules. Quality reporting exists in popular outlets, and lousy reporting exists in academic ones. But the pattern is reliable.

Where to find studies

Anyone can read primary research. The main free sources:

• PubMed (ncbi.nlm.nih.gov/pubmed) — Index of biomedical literature, free abstracts, often free full text
• Cochrane Library (cochranelibrary.com) — Systematic reviews, often free
• ClinicalTrials.gov — Registry of registered trials, including ongoing ones
• Google Scholar — Broader search, mixed quality, useful for citation tracing
• Sci-Hub — Note: paywall workaround of disputed legal status in most jurisdictions; we mention it for completeness but do not endorse use

Reading abstracts is a good entry point. Reading methods sections is where the real understanding lives.

What this primer does not cover

This page is a starting frame. It does not cover statistical methods in depth, study design subtleties beyond PICO, the politics of academic publishing, or the specific quirks of nutrition and supplement research (which has its own well-documented methodological problems).

For deeper coverage, see future Proco pages on:

• How to evaluate a meta-analysis (coming soon)
• The problem with nutrition research methodology (coming soon)
• What "evidence-based" actually means (coming soon)
• How surrogate outcomes mislead consumers (coming soon)

Sources and further reading

Proco provides educational, research-based information. This page describes how research is conducted and reported. Decisions about your own health belong with a qualified healthcare professional.

Schema (for implementation)

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "How to Read a Clinical Trial: A Primer for Non-Scientists",
  "description": "A reader's guide to evaluating clinical research: the hierarchy of evidence, PICO framework, p-values, effect size, conflicts of interest, and red flags in popular coverage.",
  "datePublished": "2026-06-01",
  "dateModified": "2026-05-31",
  "author": {"@type": "Organization", "name": "Proco"},
  "publisher": {"@type": "Organization", "name": "Proco", "url": "https://procohq.com"},
  "about": {"@type": "Thing", "name": "Clinical research methodology"}
}