Journalist, investor or curious observer

Start with these three chapters

Separate anecdotes from comparative evidence, read the study critically, then check the original records.

1. Read a claim without losing its limits
2. Read a study as a complete record
3. Primary-source toolkit

Then use the reading path and books for more depth, and complete the learning worksheet to record what you found.

Read a claim without losing its limits

Begin with one answerable question. “Does it improve wellbeing?” covers many different things. Pain, sleep, muscle strength, a blood measurement and hospital admission are separate outcomes. Write the exact substance, the people studied, the comparison and the outcome before deciding whether a finding answers your question.

An online experience is a report of what someone noticed. It can identify questions worth investigating, including possible harms. It usually cannot separate the substance from changes in training, food, other treatments, expectations or the natural course of a problem. Many similar posts still do not create a controlled comparison. Popularity and scientific certainty are different measurements.

Laboratory and animal experiments can investigate mechanisms and generate hypotheses. Human observational studies examine associations in real settings, with potential differences between groups. Randomized comparisons are designed to reduce those differences, but their usefulness still depends on execution, size, follow-up and the outcomes measured. The study design must fit the question; a small trial is not automatically a complete safety record.

Keep benefit and harm in the same notes. Record how many people were observed, how long they were followed, what events occurred and whether investigators considered them related to treatment. An event after treatment does not by itself establish causation. Conversely, an uncertain causal connection is not a reason to erase a serious event.

When reading numbers, retain the denominator. A hypothetical change from 2 in 100 to 1 in 100 is a one-percentage-point absolute reduction and a 50% relative reduction. Both descriptions concern the same comparison. Ask for uncertainty around estimates and whether the result was planned or selected from many analyses. A headline alone rarely supplies all of this.

Read a study as a complete record

Begin with the primary paper and registry where available. A registry describes a planned or reported study; its presence does not prove benefit. Compare the registered population, outcomes and timing with the published account. Preserve publication date, study period and later update date as different facts.

Write the research question using population, intervention, comparator and outcome. Include species, formulation, route and duration. A comparison against placebo answers a different question from a comparison against an active medicine. A study in one population may leave other groups unrepresented.

Look at how participants entered the analysis. An estimate based on everyone assigned may differ from one describing participants who remained on treatment. Missing observations and treatment discontinuations can affect interpretation. Record the paper’s stated analysis and assumptions instead of merging estimates from different methods.

Check whether an outcome was primary, secondary or exploratory. Multiple measurements increase the opportunity for chance findings; the analysis plan and adjustments matter. A confidence interval describes uncertainty under the analysis assumptions. Statistical significance does not by itself establish a meaningful improvement in daily life.

Read withdrawals, adverse events and serious events alongside benefits. Retain reported causality assessments and comparator counts. Distinguish no event observed from evidence that an event cannot occur. Small studies and short follow-up may have limited ability to characterize uncommon or delayed harms.

Finally, separate the result from its interpretation. Note sponsor involvement, available methods and missing details. A sponsor release can report a development before full results are public; attribute it clearly and leave unresolved methods visible until the paper or registry supplies them.

Know which product the evidence describes

A molecule name is a starting point, not a complete product description. Record the active ingredient, chemical form, formulation, route, manufacturer or compounder and the document date. Products using similar names may differ in ways that matter. A result for one preparation does not automatically establish the performance of another.

FDA approval concerns a particular drug product and its authorized labeling. A facility registration, product listing or National Drug Code does not mean FDA has approved the product. A company can have legitimate paperwork for one activity without every product or claim being authorized.

Compounded drugs are prepared under a different framework and are not FDA-approved. They can serve particular patient needs, but FDA does not review each compounded product for safety, effectiveness and quality before marketing in the way it reviews approved medicines.

Investigational means a product or use is being studied; it does not mean success is established. Research-use labeling concerns a claimed purpose and cannot establish suitability for human use. Keep regulatory status, analytical quality and clinical evidence as three separate entries. None can substitute for the other two.

Build a simple comparison: what the paper studied, what the official label describes, and what the current document or service names. Mark missing information rather than filling it with a brand assumption. This comparison is useful for a patient preparing questions and for a professional reviewing a brochure.

Pharmacy types

Section 503A covers compounding for an identified individual patient, including by a pharmacist in a state-licensed pharmacy, subject to its conditions. Section 503B covers registered outsourcing facilities that remain subject to current good manufacturing practice (CGMP), a system of manufacturing quality requirements. These facilities can supply office stock for healthcare providers without an individual prescription, subject to applicable conditions. Neither pathway makes a compounded product FDA-approved. Registration is not approval of the facility’s products. The linked Pharmacist guide explains the differences in more depth.

Sources for this addition checked October 31, 2026.

What “gray market” leaves unexplained

“Gray market” is a loose conversational label, not a quality grade or a legal conclusion. It can blur several different issues: an unapproved product promoted for human use, an authentic product outside a documented supply chain, a counterfeit identity, misleading research labeling, or a preparation that does not meet a compounding pathway’s conditions. Name the actual question before deciding what a document or testimonial proves.

A research-use statement describes a claimed purpose; it does not automatically settle intended use. In a March 31, 2026 warning letter to Gram Peptides, FDA explained that website evidence established intended human use despite research-only and not-for-human-consumption disclaimers. This is an example of FDA’s stated enforcement findings, not a final court judgment or a claim about the company’s later status. The broader lesson is to read the surrounding claims and presentation, not a disclaimer in isolation.

Counterfeit and unapproved are also different concepts. Counterfeit identity concerns a product falsely represented as authentic. Unapproved status concerns the absence of applicable approval. Lawfully compounded medicines are themselves not FDA-approved, so “unapproved” alone does not describe every applicable legal condition. Conversely, calling something compounded does not establish that it meets those conditions. Classification requires the exact product, source, claims and regulatory context.

Chemical names require precision. On its current GLP-1 concerns page, FDA states that retatrutide and cagrilintide cannot be used in compounding under federal law. The same page distinguishes semaglutide sodium and acetate from the active ingredient in approved medicines and says the agency is unaware of a lawful basis for their compounding use. These are product-specific statements, reviewed October 31, 2026; do not generalize them to every peptide, salt or jurisdiction.

A separate FDA clarification dated April 1, 2026 discusses restrictions on preparations that are essentially copies of commercially available drugs. It explains that adding another active ingredient does not automatically avoid those restrictions. That dated statement should remain dated: it is not an independently checked live shortage-database result. An April 30, 2026 announcement proposing exclusion of three GLP-1 substances from the 503B bulks list was a proposal, not a final decision.

For an educational review, record exactly what is known: the named item, original source, date, claimed use and unresolved identity or quality questions. Distinguish a reported adverse event from proven causation and a compelling anecdote from a controlled study. Neither a disclaimer nor a dramatic personal story turns an uncertain product into a verified clinical option.

Explain a claim with its evidence attached

A reader receives a message from the complete presentation: headline, image, testimonial, spoken words and qualifications. Write down the actual impression before judging the evidence. A document titled “education” can still communicate a product benefit or sales claim.

Keep a claim ledger with the exact wording, intended audience, primary source and relevant limits. Record the population, product, outcome and date. Attach a decision: supported as written, needs narrowing, or unanswered. This makes review repeatable when a sentence moves from a website to a printed handout.

Use plain language without removing uncertainty. Explain an unfamiliar term when it first appears. Put a consequential limitation beside the claim it qualifies. A small disclaimer elsewhere cannot repair an unsupported headline. Separate community reports from controlled evidence and company statements from independently established findings.

Give each document a content owner, version and review date. Preserve sources and consequential limitations when adapting it for another audience. Keep a change record so later readers can identify which claim changed and why.

Changed claims about presentation checked against FTC guidance October 31, 2026.

Primary-source toolkit

ClinicalTrials.gov: find a study’s registered design, eligibility, outcomes, status and posted results when available. Compare record versions with the paper. A registration entry alone does not establish a favorable result or clinical benefit.

Drugs@FDA: locate approval records, reviews and approved labeling for covered drug products. Match the application, product and date. It does not validate every similarly named preparation.

DailyMed: read company-submitted labeling currently in use. It includes approved and unapproved products, and its labeling may differ from the most recent FDA-approved version. Presence in DailyMed is not proof of approval; check Drugs@FDA for that question.

PubMed: locate biomedical citations and follow them to the actual study. An abstract is a starting point, not a replacement for methods, full results and limitations. Being indexed is not an endorsement of a conclusion.

FDA warning-letter database: find the agency’s stated concerns, letter date and any linked responses or closeouts. The database warns that later interactions may change the status of the issues. A missing letter is not a certificate of compliance.

Sources for this addition checked October 31, 2026.

Reading path and books

Continue in the Church Library

Choose the depth that fits your background. Library entries include reading options and affiliate disclosures where applicable.

Browse all Library rooms →

Examples in practice

These examples identify published resources and their uses; they are not evaluated services.

Questions and explanations

Glossary

Comparator

The reference group or intervention against which an outcome is compared, such as placebo or another treatment.

Endpoint

An outcome defined for a study. A biological measurement and a patient-important outcome are not necessarily the same thing.

Formulation

The composition and presentation of a product, including ingredients beyond the active substance.

Adverse event

An unwanted medical occurrence during observation; occurrence does not itself establish a causal connection to treatment.

Investigational

Under investigation. This describes research status, not proof of safety or benefit.

Gray market

An informal description of uncertain or unauthorized distribution channels, not a regulatory approval category.

Confidence interval

A range expressing statistical uncertainty under an analysis’s assumptions; it does not remove bias or establish applicability.