EFSA OpenFoodTox Study Quality Assessment Guide

Toxicological study curation and study quality evaluation form the foundation of chemical safety dossiers for FMCG food, beverage, and packaging raw materials across the European Union. Procurement managers, toxicologists, and regulatory affairs leads face high stakes when auditing supplier ingredient dossiers against official benchmarks. Relying on superficial study summaries or unverified database entries creates severe supply chain vulnerabilities, regulatory delays, and potential product recall risks under EU food safety laws.

OpenFoodTox database discrepancies and scientific opinion priority

EFSA OpenFoodTox 3.0, published on Zenodo in April 2026 under DOI 10.5281/zenodo.19388272, serves as the machine-readable database for chemical hazards in food and feed, offering IUCLID 6 archives and flat spreadsheets covering 7,880 chemical substances. However, EFSA explicitly warns that OpenFoodTox is a compiled index and does not guarantee freedom from data entry or migration errors. In any case of discrepancy between database values and published scientific opinions, the original PDF output in the EFSA Journal takes absolute legal and regulatory priority.

Furthermore, traditional database queries depend on Klimisch reliability scores (Klimisch 1 or 2), which naturally favor Good Laboratory Practice (GLP) industry guidelines while devaluing non-GLP academic research. To overcome this limitation, modern evaluation tools like SciRAP and ToxRTool allow regulatory scientists to systematically evaluate non-GLP academic studies against design quality criteria such as blinding, randomization, and power calculations, enabling academic data inclusion in EFSA authorization dossiers.

SCHEER 2026 Weight of Evidence integration rules

When integrating multiple studies into a unified line of evidence, EFSA scientific panels apply the SCHEER Memorandum on Weight of Evidence approach (updated January 2026). Under Section 6.1, a line of evidence achieves Grade A (High Quality) only if at least 75 percent of studies (and a strict minimum of 3 studies) are high quality. Similarly, Consistency Grade I (High Consistency) requires at least 75 percent of studies (minimum 3) to reach concordant toxicological conclusions.

A critical edge case occurs when an assessment contains 2 or fewer supporting studies. Under SCHEER 2026 rules, consistency cannot be judged reliably on sample sizes of two or fewer studies, making matrix grading impossible. In such small sample cases, toxicologists must report evidence qualitatively rather than assigning a formal matrix category. When 3 or more studies exist, matrix integration applies: Quality A with Consistency I yields Very Strong evidence, while Quality A with Consistency III indicates Strong to Moderate evidence.

Predictive QSAR acceptance and TEST score exclusion dead zones

When empirical experimental toxicity data is absent, EFSA working groups allow predictive in silico Quantitative Structure-Activity Relationship (QSAR) models under strict applicability domain rules. QSAR predictions are accepted only if the target chemical falls inside the model prediction domain. For VEGA QSAR models, predictions require a consensus score of at least 0.5 or a reliability score of at least 0.7.

QSAR Platform or Tool	Reliability Score Threshold	Acceptance Decision	Operational Rule
VEGA QSAR	Consensus score >= 0.5 or Reliability >= 0.7	Accepted	Valid for in silico hazard screening
US EPA TEST QSAR	Reliability score > 0.65	Accepted	High confidence model output
US EPA TEST QSAR	Reliability score 0.35 to 0.65	Strictly Rejected	Excluded as unvalidated dead zone
Danish (Q)SAR Database	Inside prediction domain	Accepted	Retained for read-across support
Structural Read-Across	Similarity score >= 0.85 (0.90 complex)	Accepted	Resolves diverging model results

For the US EPA Toxicity Estimation Software Tool (TEST), EFSA rules enforce a strict exclusion zone. Reliability scores between 0.35 and 0.65 are classified in a forbidden middle zone and strictly rejected. Software tools that treat QSAR reliability as a linear scale violate EFSA guidance by accepting middle-zone predictions. When QSAR models yield conflicting conclusions, read-across requires chemical analogue structural similarity scores of at least 0.85 (or 0.90 for complex re-evaluations).

Qualichem multi-expert confidence levels and MIXTOX rules

Evaluating study quality across diverse expert panels often introduces scientific controversy. The Qualichem framework grades in vivo toxicological studies against 45 distinct quality criteria. A criterion is flagged as critical if expert score divergence is at least 2 points (or any score <= 3) and meets severe failure conditions, such as a 4-point expert difference, any single score of 1, or a median score of 4 or lower. Global study confidence requires fewer than one-third critical criteria (14 or fewer out of 45) for High Confidence.

For chemical mixture assessments (MIXTOX), EFSA guidelines establish clear prioritisation thresholds. A chemical contributing 10 percent or more to total mixture risk triggers detailed cumulative assessment. Low-priority pesticides are excluded from cumulative risk groups if their Hazard Quotient at the 99.9th exposure percentile HQ(P99.9) is 0.01 or less. For non-pesticide food contaminants, the exclusion cutoff defaults to HQ(P95) <= 0.01.

Automated study quality assessment with fmcg.network

FMCG quality assurance, procurement, and regulatory teams can automate toxicological study audits and SCHEER Weight of Evidence checks using fmcg.network Business Capabilities. The Toxicological Study Quality Assessment capability cross-references chemical parameters against EFSA OpenFoodTox 3.0 benchmarks, evaluates SCHEER 2026 WoE matrices, checks QSAR dead zones, and grades Qualichem confidence levels automatically.

To run an automated study quality evaluation, connect your AI assistant to fmcg.network and submit the query:

“Evaluate Weight of Evidence for Bisphenol A with 4 studies (3 high quality, 3 consistent) and test QSAR acceptance for Carbendazim with TEST score 0.52 against SCHEER 2026 and EFSA OpenFoodTox 3.0 rules.”

The assistant queries the network registry, calculates exact line-of-evidence matrix grades, identifies QSAR dead-zone rejections, and returns factual reference positions without issuing subjective legal advice or compliance verdicts.

Install fmcg.network in Claude, ChatGPT, Copilot, or Cursor, then explore the full Business Capability Directory.

Frequently Asked Questions

Does EFSA provide an API for OpenFoodTox 3.0 datasets? OpenFoodTox 3.0 datasets are hosted on Zenodo (DOI 10.5281/zenodo.19388272) as IUCLID 6 archives and flat Excel files. Programmatic access uses Zenodo REST APIs or direct spreadsheet parsing.

What procedure applies if an ADI in OpenFoodTox differs from an EFSA Journal PDF? In any case of discrepancy between compiled OpenFoodTox database files and published scientific opinions, EFSA rules dictate that the original PDF in the EFSA Journal takes absolute legal preference.

How can SciRAP or ToxRTool upgrade non-GLP academic study reliability? SciRAP and ToxRTool evaluate non-GLP academic studies against objective design criteria such as randomization, blinding, and statistical power, allowing high-quality academic data to achieve regulatory reliability recognition.

Why does EFSA reject TEST QSAR predictions with a reliability score of 0.52? EFSA QSAR rules strictly exclude TEST model reliability scores between 0.35 and 0.65 as an unvalidated middle zone. Only TEST predictions with reliability scores above 0.65 are accepted.

What is the minimum structural similarity score for read-across analogues? EFSA read-across guidelines require structural similarity scores of at least 0.85, increasing to 0.90 for highly complex chemical re-evaluations.

How is the SCHEER 2026 Weight of Evidence matrix applied to small sample sizes? If a line of evidence contains 2 or fewer studies, SCHEER 2026 rules state that consistency cannot be judged reliably and matrix grading cannot be assigned. Evidence must be evaluated qualitatively.

What does SCHEER mean by Grade A Quality with Grade III Consistency? Grade A Quality means at least 75 percent of studies (minimum 3) are high quality. Grade III Consistency means 50 percent or fewer studies agree on the conclusion, resulting in a Strong to Moderate evidence rating.

How are pesticides excluded from EFSA cumulative risk assessment groups? Pesticides are excluded from cumulative risk groups if their individual Hazard Quotient at the 99.9th exposure percentile HQ(P99.9) is 0.01 or less.