DEX Score
The DEX score combines effectiveness and evidence evaluation scores to represent supplement value and performance data. It's how SupplementDEX combines Effect Size, Evidence, and Trust into a single 1–5 score.
The DEX Score is SupplementDEX’s single 1–5 readout for how strong the case is that an ingredient helps for a given condition. Generally, a DEX Score of 4–5 stars means a supplement is worth the money. You’ll see it as stars next to the overall verdict, and as a sortable column on the Effect heatmap.
How it’s built
Each condition is graded on three metrics. All three use a five-level scale (shown as colored dots):
- Effect Size: Very High → High → Medium → Low → Very Low / None
- Evidence and Trust: Very High → High → Medium → Low → Very Low
| Metric | What it answers |
|---|---|
| Effect Size | How large / clinically meaningful is the benefit when it works? (Very Low / None = negligible or no meaningful effect) |
| Evidence | How much research exists (volume and coverage)? |
| Trust (Confidence) | How much can that research be trusted (design, consistency, quality)? |
The DEX Score weights them as:
Effect Size ×3 · Evidence ×1 · Trust ×1
Effect Size carries more weight because a large, clinically meaningful benefit matters more than simply having many studies.
How we grade Evidence
Evidence stays on the same five-level scale. It is a structured judgment of research volume and coverage for that specific use (not “this ingredient in general”).
| Level | Typical pattern |
|---|---|
| Very High | ≥10 human studies for this use, including at least one systematic review or meta-analysis; or a dense set of large RCTs spanning doses/populations that indicates a larger evidence base and providing generalisability |
| High | ~5–9 relevant human studies (trials and/or reviews) that cover the main clinical question |
| Medium | ~3–4 human studies; partial coverage (e.g. one population, one dose, or short duration only) |
| Low | 1–2 human studies, or a thin mix of small trials / secondary analyses |
| Very Low | No adequate human trials for this use — abstracts, case reports, animal/in vitro data, or mechanistic speculation only |
A relevant systematic review or meta-analysis floors Evidence at High, even when the local study list is short — because it already synthesizes a broader evidence body.
Human review
There are edge cases where we use human review — when large study populations can necessitate human intervention, and vice versa.
How we grade Trust
Trust stays on the same five-level scale. It is a structured judgment of study design, consistency of findings, and whether the research applies to real-world use (dose, form, population).
| Level | Typical pattern |
|---|---|
| Very High | Findings driven by well-conducted RCTs and/or meta-analyses; results agree in direction; studied dose/form/population match typical use; outcomes are clinically relevant |
| High | Mostly RCTs with adequate controls; direction largely consistent; only minor gaps (one underpowered trial, short follow-up, or limited population) |
| Medium | Mix of RCTs and weaker designs, or mixed results across similar trials; applicability issues (wrong dose, surrogate endpoints, narrow sample) |
| Low | Dominated by small, open-label, or uncontrolled studies; serious inconsistency; or industry-heavy single-site evidence that doesn’t replicate |
| Very Low | Cannot trust a positive or a null finding — speculative claims, surrogate-only signals, or too little quality data to judge reliability |
Very Low Trust means the finding isn’t reliable enough to recommend — the DEX Score caps guidance at Reconsider or below, even if Effect Size looks large. Proven purchase requires High or Very High Trust.
When the evidence body is mostly multi-ingredient / fixed herbal combinations, we treat it as combination evidence: Effect Size is capped at Low and Trust at Medium, so a combo product is not scored as if it were the single ingredient alone.
Mixed study directions soften Trust only mildly. Large RCT/meta bodies are not collapsed by raw +/- tallies; disagreement among meta-analyses matters more.
Human review
There are edge cases where we use human review — when study design, population, dose, conflicting results, or combination products don’t map cleanly to the rubric and require intervention.
Combined DEX Score
The DEX score combines effectiveness and evidence evaluation scores to represent supplement value and performance data.
The DEX score and its subscores are updated regularly and use the following mapping for the score values:
| Guidance | Score | Stars |
|---|---|---|
| Do not purchase | ≤39 | ★★★★★ |
| Reconsider purchase | 40–59 | ★★★★★ |
| Consider purchase | 60–79 | ★★★★★ |
| Proven purchase | 80–100 | ★★★★★ |
How to read it
- A condition can have lots of Evidence and still score poorly if the effect is small or the data are shaky.
- A high DEX Score usually means a meaningful effect plus research you can lean on.
- A low DEX Score is a caution flag — not always “does nothing,” but not a strong case yet.
1–2 stars (Do not purchase) — Ineffective or insufficient evidence. Weak case; skip for this use.
3 stars (Reconsider purchase) — No meaningful benefit. Mixed or modest effect.
4 stars (Consider purchase) — Possibly effective. Almost all good-quality studies find a benefit, and the evidence is strong enough to be worth the money for many people.
5 stars (Proven purchase) — Clear evidence for benefit and effect — sometimes approved by official guidelines.
On the Effect heatmap, sort by DEX Score when you want the strongest signals first instead of alphabetical order.
Where you’ll see it
- Purchase guidance + stars beside each condition in the SupplementDEX Database
- The DEX Score column when you expand metrics on the Effect heatmap
- Cross-condition comparisons on the Effect graph and on supplement pages with many studied uses
For Simple vs Practitioner Evidence Review and the rest of the page layout, see How to read evidence.