Supplement Evidence Density Map 2026

• This map matched no published RCT for 22% of cleaned supplement-goal pairings (475 of 2,156). That figure is an upper bound on the true evidence gap because it includes known matching failures.
• 38% had at least 3 mapped RCTs, the threshold HAA treats as a more meaningful evidence base.
• Only 3.7% reached the 10+ RCT bucket.
• 52.3% had at least one mapped meta-analysis or systematic-review signal.
• 122 no-match pairings still had at least 1,000 estimated monthly searches. Treat this as a mapping-review queue, not evidence that trials are absent.
• The map measures evidence density, not product quality, dose adequacy, individual suitability, or medical benefit.

The matrix is not evenly distributed. A large no-match tail sits beside a smaller set of heavily mapped pairings. Because the no-match bucket includes known pipeline gaps, it cannot be used to estimate how often published trials are absent.

Some categories have large bodies of human research; others are mostly market demand plus thin clinical support. Cluster-level views prevent one strong ingredient from making the whole supplement landscape look stronger than it is.

High search demand is not the same as strong human evidence. This is why HAA separates demand, evidence, commercial availability, and YMYL risk before deciding which pages deserve full editorial treatment.

These pairings had the highest mapped RCT counts in the cleaned matrix. Counts are evidence-density signals, not automatic recommendations.

Read this table as a review queue for HAA’s mapping, not as a finding that published trials are absent.

For these high-demand pairings, the matrix matched no RCT. `rctCount` was produced by keyword-matching a capped per-supplement PubMed sample rather than querying each supplement-goal pairing directly, so a no-match can mean that relevant trials were outside the sample or missed by the keywords. The source data also contains unmerged aliases that can assign different counts to the same ingredient and goal. For example, Citrulline / exercise performance is mapped to 0 while L-Citrulline / exercise performance is mapped to 9.

Every row below also has at least one mapped meta-analysis or systematic-review signal, now shown in the table. Do not cite this table as evidence that a supplement is unsupported for a goal.

Suggested citation: Healthy Aging Atlas Research Team. (2026). Supplement Evidence Density Map 2026 (Version 2026.1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.21412552

Download the CSV or review the methods page before reusing the headline number. The denominator is the cleaned matrix after removing editorial rejects, a confirmed category-as-goal artifact, and duplicate supplement-goal identities created by legacy aliases or formulation slugs.

• This is an evidence-density map, not a clinical recommendation engine.
• RCT count does not measure trial quality, sample size, duration, risk of bias, or whether the dose matches products on the market.
• The matrix uses search and evidence proxies to identify marketed supplement-goal pairings; it is not a legal audit of every brand label or advertisement.
• The PubMed snapshot reflects the HAA matrix available at generation time. New studies can change the evidence density over time.
• A no-match result means this pipeline did not match an RCT in its source sample. It does not establish that no published trial exists.
• `rctCount` is a matching result, not a PubMed census. It came from keyword matching against a capped per-supplement sample rather than a direct query for every supplement-goal pairing, so the no-match bucket includes an unquantified share of pipeline gaps.
• Alias normalization is incomplete. Confirmed examples such as Citrulline / L-Citrulline and Carnitine / L-Carnitine can carry different mapped counts for the same goal. A corrected remap and alias canon are tracked separately.