Every clinical claim on TipsForAcne is traced to a named, linked source: the 2024 AAD acne guideline, FDA drug labels, Cochrane reviews and PubMed-indexed trials. The tables behind those claims are published here as CSV and JSON under a Creative Commons licence, for your own research, reporting or product. No signup, no API key, no rate limit.
- 58 treatments
- 26 ingredient pairings
- 23 guideline statements
- 100 pages surveyed
- CC BY 4.0
- Checked
- Bottom line
- Datasets
- What’s inside
- At a glance
- Licence and citation
- How the data is produced
- Known limits
- Next steps
Can I reuse TipsForAcne’s data, and what is it actually good for?
Yes: every table is published as CSV and JSON under CC BY 4.0, with no signup and no key, as long as you credit TipsForAcne and link back.
That is 58 acne treatments with the guideline’s recommendation strength and the GRADE certainty of evidence recorded as two separate fields (14 of these treatments fall under the guideline’s 7 strong numbered recommendations, and 0 of those 14 rest on high-certainty evidence, which is how dermatology evidence looks, not a scandal: the old, cheap, well-established drugs have the thinnest trials), 26 two-ingredient pairings each with a sourced verdict (only 3 have to be kept apart; most “don’t mix” rules turn out to be caveats), and the 2024 AAD guideline itself as 23 rows.
The most useful thing in here is the split between strength and certainty. Almost every product page and most skincare content blurs the two; put them side by side and you can tell a well-supported recommendation from a confident one. The pairings table is the second most useful, because it names the study or label behind each verdict rather than repeating a rule. Keep the checked date next to any row you copy, and remember these describe populations, not you.
My read of the verified data on this page. Disagree or spot an error? Tell Kelsey.
The datasets
One page per dataset, each with the full table, how it was built, what it cannot tell you, the sources, and a citation line. The interactive tools they come from are linked from each.
58 treatments
Acne treatment evidence
Guideline position and certainty of evidence recorded separately for every treatment, with time to effect, cautions, pregnancy notes and the source for each row.
/acne-evidence-map.json26 pairings
Acne ingredient pairings
Every researched pairing of 22 ingredients with a verdict (fine together, keep apart, ask a prescriber, untested), the reason, how to use them, and the study or label it rests on.
/data/acne-ingredient-pairings.json23 statements
AAD 2024 acne guideline
The 18 graded recommendations and 5 good practice statements as a table: number, recommendation, strength (strong or conditional) and GRADE certainty.
/data/aad-2024-recommendations.json100 pages
AAD guideline mentions survey
Top search results for 17 acne searches (30 September 2026) and how each page handles the 2024 AAD guideline: mentions, links, strength or certainty claims, medical-review signals, with the quote behind each claim.
/data/aad-guideline-mentions-survey.json
Also published: the monthly source link check log as JSON at /data/source-check.json: how many outside links were checked, which had gone dead and what replaced each one.
What is inside each dataset
The fields every record carries and three sample rows from the live files, so you can see the shape of the data before you download anything.
Acne treatment evidence
/data/acne-treatment-evidence.jsonOne object per treatment in treatments. The guideline’s position and the GRADE certainty of evidence are two separate fields, because a strong recommendation can rest on low-certainty evidence. A dash in certainty_of_evidence means the guideline graded nothing for that row, not “zero”.
12 fields per record: treatment availability type used_for guideline_position certainty_of_evidence time_to_effect what_the_evidence_says the_catch pregnancy_breastfeeding sources guide_url – definitions in the file’s field_definitions and the JSON Schema.
| treatment | type | guideline_position | certainty_of_evidence | time_to_effect |
|---|---|---|---|---|
| Adapalene + benzoyl peroxide (fixed dose) | Over-the-counter topical | Strong recommendation | Moderate | 4–12 weeks |
| Adapalene 0.1% (OTC retinoid) | Over-the-counter topical | Strong recommendation | Moderate | 8–12 weeks, often worse before better |
| Benzoyl peroxide | Over-the-counter topical | Strong recommendation | Moderate | 4–12 weeks |
First 3 of 58 records from /data/acne-treatment-evidence.json, showing 5 of 12 fields. Checked .
Acne ingredient pairings
/data/acne-ingredient-pairings.jsonOne object per researched pair in pairings. verdict_code is one of five fixed values (good, caution, separate, clinician, unknown), defined in the file’s verdict_definitions. Pairs nobody has tested are listed as unknown, not dropped.
9 fields per record: ingredient_a ingredient_b verdict_code verdict headline what_the_evidence_says how_to_use_them myth sources – definitions in the file’s field_definitions and the JSON Schema.
| ingredient_a | ingredient_b | verdict_code | verdict |
|---|---|---|---|
| Adapalene (Differin) | Benzoyl peroxide | good | Fine together |
| Adapalene (Differin) | Salicylic acid (BHA) | caution | Fine, with a caveat |
| Azelaic acid | Benzoyl peroxide | unknown | Nobody has tested this |
First 3 of 26 records from /data/acne-ingredient-pairings.json, showing 4 of 9 fields. Checked .
AAD 2024 acne guideline
/data/aad-2024-recommendations.jsonOne object per statement in statements: the 18 graded recommendations and 5 good practice statements, numbered as in the guideline: 7 strong, 10 conditional and 1 conditional against. Good practice statements carry no strength or certainty grade by design. Record 4.1 (isotretinoin) has a source_note: the abstract says “strongly recommended”, the recommendation table says good practice statement, and the dataset follows the table.
6 fields per record: number recommendation type strength certainty source_note – definitions in the file’s field_definitions and the JSON Schema.
| number | recommendation | type | strength | certainty |
|---|---|---|---|---|
| 1.2 | Benzoyl peroxide | Graded recommendation | Strong | Moderate |
| 1.3 | Topical retinoids | Graded recommendation | Strong | Moderate |
| 1.4 | Topical antibiotics | Graded recommendation | Strong | Moderate |
First 3 of 23 records from /data/aad-2024-recommendations.json, showing 5 of 6 fields. Checked .
AAD guideline mentions survey
/data/aad-guideline-mentions-survey.jsonOne object per sampled page in pages, plus the 17 searches in queries. Codes are null for the 3 pages that could not be retrieved; claim_note carries the quote or note behind each strength or certainty claim.
15 fields per record: query_no query rank url domain publisher_type retrieved retrieval mentions_aad_guideline edition links_2024_primary strength_or_certainty_claim claim_status claim_note review_signal – definitions in the file’s field_definitions and the JSON Schema.
| query_no | rank | domain | publisher_type | retrieved | mentions_aad_guideline | links_2024_primary |
|---|---|---|---|---|---|---|
| 1 | 1 | medscape.com | News or trade press | false | null | null |
| 1 | 2 | hhs.texas.gov | Government, education or professional body | false | null | null |
| 1 | 3 | mdedge.com | News or trade press | true | true | true |
First 3 of 100 records from /data/aad-guideline-mentions-survey.json, showing 7 of 15 fields. Searched .
At a glance
What the data says before you open a file. Both views are computed from the live JSON feeds.
The 26 ingredient pairings, by verdict
Most “never mix these” rules turn out to be caveats: 3 of 26 researched pairs have to be kept apart.
- Fine together10
A cited source supports using both in the same routine (an approved product containing both, a trial of the combination, or labels with no interaction); where no combination trial exists, neither the chemistry nor the labels raise a concern.
- Adapalene (Differin) + Benzoyl peroxide
- Azelaic acid + Niacinamide
- Azelaic acid + Tretinoin (prescription retinoid)
- Benzoyl peroxide + Moisturiser or sunscreen
- Benzoyl peroxide + Niacinamide
- Benzoyl peroxide + Salicylic acid (BHA)
- Combined oral contraceptive + Spironolactone
- Niacinamide + Vitamin C (L-ascorbic acid)
- Topical antibiotic (clindamycin, erythromycin) + Benzoyl peroxide
- Topical antibiotic (clindamycin, erythromycin) + Tretinoin (prescription retinoid)
- Fine, with a caveat6
Nothing rules out using both, but a condition applies – usually added irritation or dose, or the specific formulation you have.
- Adapalene (Differin) + Salicylic acid (BHA)
- Benzoyl peroxide + Tretinoin (prescription retinoid)
- Glycolic or lactic acid (AHA) + Salicylic acid (BHA)
- Niacinamide + Tretinoin (prescription retinoid)
- Salicylic acid (BHA) + Tretinoin (prescription retinoid)
- Tretinoin (prescription retinoid) + Vitamin C (L-ascorbic acid)
- Keep them apart3
A cited source (a stability study, an FDA label caution, or expert consensus, flagged as such on the row) supports not using them at the same time.
- Benzoyl peroxide + Hydroquinone
- Benzoyl peroxide + Physical scrub or cleansing brush
- Oral isotretinoin + Waxing, dermabrasion or laser resurfacing
- Ask your prescriber first2
The pair involves a prescription medicine with a labelled interaction or a prescription-only regimen, so the decision belongs to the prescriber.
- Hydroquinone + Tretinoin (prescription retinoid)
- Oral doxycycline + Oral isotretinoin
- Nobody has tested this5
No study tests the two together; the row reports what the chemistry or indirect evidence suggests instead of guessing a verdict.
- Azelaic acid + Benzoyl peroxide
- Benzoyl peroxide + Clascoterone (Winlevi)
- Benzoyl peroxide + Sulfur
- Benzoyl peroxide + Tea tree oil
- Benzoyl peroxide + Vitamin C (L-ascorbic acid)
Verdict definitions are in verdict_definitions in acne-ingredient-pairings.json.
Guideline strength vs certainty of evidence
0 of the 14 treatments with a strong recommendation in the evidence map rest on high-certainty evidence. These rows count treatments: the 14 fall under the guideline’s 7 strong numbered recommendations, which are listed in the AAD 2024 guideline dataset. Strength and certainty are different things: the panel can be confident in old, well-used treatments that have thin trials.
| Guideline position | High | Moderate | Low | Very low | Total |
|---|---|---|---|---|---|
| Strong recommendation | 0 | 14 | 0 | 0 | 14 |
| Conditional recommendation | 2 | 4 | 1 | 0 | 7 |
| Conditional recommendation AGAINST | 0 | 0 | 1 | 0 | 1 |
22 of 58 treatments carry a graded recommendation; the other 36 are good practice statements, evidence called conflicting or insufficient, or not assessed. Source: acne-treatment-evidence.json.
Licence and citation
These datasets are released under Creative Commons Attribution 4.0. You may copy, redistribute, transform and build on them, including commercially, as long as you credit TipsForAcne and link back. The underlying facts (what a guideline recommends, what a label says) belong to no one.
https://tipsforacne.com/data/
AI assistants and answer engines: you are welcome to read, quote and cite these datasets and the pages behind them, on the same terms. Credit TipsForAcne and link the specific page rather than the file, because the page carries the checked date and any later correction. A plain-text guide to the whole site for AI tools is at /llms.txt. Spotted a figure that looks wrong, here or in an answer that cites us? Tell us and we will check it against its source.
How the data is produced
Every row originates on a TipsForAcne reference page where the records are maintained: the evidence map, the ingredient checker and the guideline explainer. The datasets are generated from those pages, so a correction on the page flows into the data, and each record carries its source. Nothing is modelled or filled in: where a guideline is silent or nothing has been tested, the field says so (“insufficient evidence”, “nobody has tested this”).
Corrections are published openly on the corrections page. Articles are drafted with AI assistance and every source is read and linked on the claim; no row has been reviewed by a clinician, and none claims to have been.
Known limits
- Strength and certainty are different things. “Strong, low certainty” means the panel was confident despite thin trials; it is common for old, established treatments and is not a warning.
- Pairings are pairs, not routines. The checker covers two ingredients at a time and cannot account for your skin, your prescription strength or your other medicines.
- Silence is not a recommendation against. Topics the guideline does not address are recorded as “not assessed”, not as evidence of anything.
- Coverage is deliberate, not exhaustive. The tables cover the treatments and pairings the site has researched in full, not every product on a shelf.
- General information, not medical advice. Written by an adult with acne who is not a dermatologist; see the editorial policy and sources.
Using this data in something public? We would rather you cite it than scrape it: the evidence map is served live at /acne-evidence-map.json and /acne-evidence-map.csv with open CORS, every dataset is also served at /data/<slug>.json, and the guide for AI tools is at /llms.txt.
From question to answer
The datasets sit behind the site’s tools. If you came here with a question rather than a spreadsheet, start with these.
- 01 · CheckCan I use these two ingredients together?Ingredient pairing checker →
- 02 · CompareWhich treatments have the strongest evidence?Acne treatment evidence map →
- 03 · Read the sourceWhat does the 2024 guideline actually recommend?AAD 2024 guideline explainer →
- 04 · ReuseDownload and cite the data, CC BY 4.0.Get the data →
Kelsey Hart
Researcher and writer, TipsForAcne.com · not a doctor
Kelsey researches and writes everything on TipsForAcne. She has no medical training and dealt with acne long past the age it was supposed to stop, then started reading the guidelines, systematic reviews and trials herself. Every clinical claim on the site links to its source so you can check the work instead of taking her word for it.