Research methodology

Sources

Every factual statement cites a primary source from one of these categories:

• FDA labeling and records: DailyMed structured product labels, Drugs@FDA application records and FDA approval letters.
• Trial registries: ClinicalTrials.gov records, read through its public API.
• Peer-reviewed literature: PubMed records, identified by PMID and DOI.
• Reference databases: PubChem for chemical identity and UniProt for protein and receptor data.

Secondary sources such as news articles, vendor pages and forums are not used as evidence. Each citation shows the date it was last retrieved.

Which trials are included

The trial database includes a registered trial when the tesamorelin arm has published or posted results. Excluded:

• Trials without any published or posted results (for example, trials still recruiting or terminated early with few participants).
• Registry entries that describe themselves as mock or test records.

How data are extracted

• Results are copied as the source reports them, with the same units, comparator, confidence interval and p-value.
• The source of each number is shown: the publication (PMID), the registry posting or the FDA label.
• When a registry and its publication disagree (for example on enrollment or masking), both versions are shown. Neither is silently picked.
• Dose arms, dosing schedules and administration details are not extracted. The label is linked instead.

No pooled statistics

This site does not combine results from different studies into a single estimate. The studies differ in population, duration, imaging method and analysis. Published meta-analyses are cited as their authors reported them, with their heterogeneity statistics where available.

Evidence grades

Each trial gets a design-based grade, adapted from the GRADE approach. Randomized, placebo-controlled trials start high and are downgraded for small size, single-centre design, missing publication, open-label design or inconsistency. The reasons for each downgrade appear on the trial's record.

• High: Multicentre, randomized, double-blind, placebo-controlled, adequately sized; no serious limitations identified.
• Moderate: Randomized and placebo-controlled, but small, single-centre, short, or with registry/publication discrepancies.
• Low: No placebo arm, open-label, pilot-sized, or results available only as an unpublished registry posting.
• Very low: Uncontrolled before/after design; useful for physiology, not for outcome conclusions.

A grade describes how much confidence a study's design supports. It says nothing about whether a result was favourable.

How results are described

• The study, not the compound, is the subject of a sentence: “the trial measured…”, “the authors reported…”.
• The population is always named. A result in one population is never presented as applying to another.
• Approved uses are kept separate from research findings. Approved medicine is kept separate from research-grade material.

Verification

An automated check runs before publishing. It confirms that every registry ID and PubMed ID resolves, that every label link returns its document, and that page text contains no dosing or administration language. Corrections are handled as described in the editorial policy.