NutriStack
Most supplement combinations have barely been studied.
We ran a relevance-filtered PubMed search for every one of the 2,517 supplement and supplement-drug pairs in the NutriStack database. For 795 pairs (31.6%), PubMed holds no record that names both substances in an interaction context. 1,399 (55.6%) have fewer than five. The full table, query and method are below, free to reuse.
Findings
The headline numbers.
Counts are PubMed records that name both members of a pair, found with the query described in the method. A record is evidence that someone studied or discussed the two together, not that they interact.
Summary
| Measure | Value |
|---|---|
| Pairs searched | 2,517 |
| Supplement + prescription pairs | 1,445 |
| Supplement + supplement pairs | 1,072 |
| Pairs with zero qualifying PubMed records | 795 (31.6%) |
| Pairs with fewer than two records | 1,036 (41.2%) |
| Pairs with fewer than five records | 1,399 (55.6%) |
| Pairs with at least one clinical trial listed | 730 (29.0%) |
Where the complete record lists show it
- For the 1,901 pairs with 20 or fewer records, we reviewed every record's PubMed publication type and MeSH tags. 1,263 of them (66.4%) have no human study naming both.
- 1,527 (80.3%) have no clinical trial of any kind.
- Pairs with more than 20 records were not counted here, because the harvest stored only the first 20 records for each pair.
Distribution
Supplement and drug pairs are the least studied.
The gap is widest exactly where the stakes are highest: a supplement taken alongside a prescription medicine.
PubMed records per pair
| Records | Supplement + prescription | Supplement + supplement |
|---|---|---|
| 0 records | 493 (34.1%) | 302 (28.2%) |
| 1 record | 151 (10.4%) | 90 (8.4%) |
| 2-4 records | 217 (15.0%) | 146 (13.6%) |
| 5-19 records | 284 (19.7%) | 194 (18.1%) |
| 20+ records | 300 (20.8%) | 340 (31.7%) |
The form gap
Research names the mineral. Shops sell the form.
The substances with the largest share of unstudied pairs are mostly specific salt forms, such as magnesium glycinate, taurate, malate and threonate. These are the forms most heavily marketed online, and PubMed research almost never names them in a combination.
Largest share of pairs with zero records (substances in 15 or more pairs)
| Substance | Pairs | Pairs with zero records | Share |
|---|---|---|---|
| Magnesium Glycinate | 98 | 93 | 94.9% |
| Magnesium Taurate | 17 | 16 | 94.1% |
| Magnesium Malate | 16 | 15 | 93.8% |
| Magnesium L-Threonate | 15 | 14 | 93.3% |
| Berberine HCl | 16 | 14 | 87.5% |
| Iron Bisglycinate | 17 | 12 | 70.6% |
| Magnesium Citrate | 27 | 19 | 70.4% |
| L-Theanine | 42 | 29 | 69.0% |
| L-Citrulline | 32 | 22 | 68.8% |
| Psyllium Husk | 35 | 24 | 68.6% |
| Nattokinase | 22 | 15 | 68.2% |
| Valerian Root | 16 | 9 | 56.2% |
| Evening Primrose Oil | 17 | 9 | 52.9% |
| Vitamin K1 | 19 | 10 | 52.6% |
| Ashwagandha | 51 | 26 | 51.0% |
A form-specific search is narrower than a search for the parent mineral, and that is the point: what a label names is often not what the research tested. The forms guide explains what differs between salt forms.
For contrast: the most consistently studied
| Substance | Pairs | Pairs with zero records | Share |
|---|---|---|---|
| SAMe | 17 | 0 | 0.0% |
| Alcohol | 66 | 1 | 1.5% |
| Turmeric/Curcumin | 35 | 2 | 5.7% |
| Selenium | 16 | 1 | 6.2% |
| Vitamin E | 31 | 2 | 6.5% |
| Calcium | 80 | 6 | 7.5% |
| Potassium | 75 | 6 | 8.0% |
| Warfarin | 36 | 3 | 8.3% |
The other end
The most studied pairs.
At the other extreme, a few mineral and vitamin pairs have thousands of records, mostly because they share absorption pathways that nutrition science has studied for decades. Linked pairs have an evidence page.
Most PubMed records
| Pair | Records |
|---|---|
| Copper and Zinc | 11,384 |
| Iron and Zinc | 6,680 |
| Beta-Carotene and Vitamin A | 6,503 |
| Copper and Iron | 6,380 |
| EGCG (Epigallocatechin Gallate) and Green Tea Extract | 4,755 |
| Iron and Manganese | 3,491 |
| Calcium and Iron | 3,332 |
| Calcium and Zinc | 3,325 |
| L-Arginine and L-Lysine | 3,262 |
| Alcohol and Cannabis (THC-Dominant) | 3,172 |
| Manganese and Zinc | 2,851 |
| Vitamin B12 and Vitamin B9 | 2,522 |
| Elderberry Zinc Lozenges and Selenium | 2,069 |
| Selenium and Zinc | 2,068 |
| L-Glutathione and Selenium | 2,027 |
Method
How the search worked.
Every number on this page can be reproduced from the CSV, which includes the exact PubMed query used for each pair.
Search
- Source: PubMed, via the NCBI E-utilities API, harvested on 2026-08-01.
- Each substance was searched under its name and common synonyms (for example 5-HTP and 5-hydroxytryptophan).
- A record counts for a pair when both names appear in the title, or when both appear in the title or abstract together with an interaction, co-administration or pharmacokinetic term. Bare co-mention was rejected in testing: it matched hundreds of papers that were about something else.
- Study design comes from PubMed publication types and MeSH tags (Humans, Animals). No design was inferred by NutriStack.
- The pair list is the NutriStack database: 2,517 pairs of supplements, and of supplements with prescription medicines, that users commonly combine.
Limitations
- PubMed is not the whole literature. Work that is not indexed there, or that phrases the combination differently, is missed.
- A record naming both substances is not evidence of an interaction, or of its absence. Zero records means the combination is under-studied, not that it is safe.
- Combination products (for example zinc lozenges with elderberry) were searched under their ingredients' names, so they can share a count with the base pair.
- Design counts are only complete for pairs with 20 or fewer records.
- This report makes no claim about any specific combination. Questions about your own medicines belong with your pharmacist or prescriber.
Reuse and citation
The data is free to reuse under CC BY 4.0. Please cite: Williams B, Gassen D. The supplement evidence gap, 2026. NutriStack. https://nutristackapp.com/reports/evidence-gap
Why this matters
Few studies means you have to keep your own records.
When the literature is thin, the best information about how a routine is going is your own. NutriStack is a free iPhone app for logging what you take and when.