The Health Integrity Project

Research Publication Classification

Our workflow helps classify research publications based on study quality, evidence strength, population type, and sample size to provide clear evidence ratings.

About Our Classification System

Our seven-step workflow ensures consistent and transparent evaluation of research publications:

  1. Validation: Screens for fundamental issues — conflict of interest and narrative reviews
  2. Quality Checks: Evaluates study design, control groups, bias management, and statistical methods
  3. Evidence Strength: Determines if the result directly settles the claim (Strong → continue; Weak → Inconclusive)
  4. Human Study Verification: Confirms findings are applicable to human populations
  5. Study Type: Distinguishes observational studies from clinical trials
  6. Sample Size Classification: Assesses robustness and generalizability based on participant numbers
  7. Paper Stance: Reviewer concludes whether the paper supports or contradicts the claim

Full Review Process

Each health and wellness claim submitted to Evidence Decoded is reviewed by one or more experts using this standardized workflow to ensure rigorous, transparent evaluation.

1 Validation

Objective: Identify fundamental issues that may invalidate the credibility of the research before deeper review.

Note: If any validation issue is found, the publication is marked as Invalid and further evaluation is not conducted.

Conflict of Interest:
  • Was the study funded by a company that would benefit from the results?
  • Do authors receive personal fees or own shares in such companies?
Narrative Review:
  • Is this a narrative review, commentary, or editorial summarizing results across studies?
  • Not a primary research article — conclusions based on qualitative interpretation

2 Quality Assessment

Objective: Evaluate the methodological rigor of the study.

Note: If any quality criterion is not met, the publication is classified as Inconclusive.

Study Design:
  • Is the study design appropriate for the research question?
  • Are outcomes clearly defined and consistently measured?
Control Group:
  • Are control or comparison groups present and properly selected?
  • May include wildtype, baseline, placebo, standard of care, or matched cohort
Bias Addressed:
  • Were confounding variables identified and tracked?
  • Important factors like age, sex, comorbidities, and socioeconomic factors considered
  • Mitigation methods include randomization, blinding, balanced cohorts
Statistics:
  • Were statistical tests appropriate for the study design and data type?
  • Are assumptions stated and checked?
  • Was multiple test correction applied?

3 Evidence Strength

Objective: Determine whether the paper's findings directly settle the claim.

Strong: Statistically significant result that directly speaks to the claim — clear support or clear refutation of the specific assertion.
Weak: Non-significant result, or gaps (wrong endpoint/population, underpowered, marginal effect) that prevent the paper from settling the claim either way.

Weak evidence forces the paper to Inconclusive. Strong evidence continues to the Human Study step.

4 Human Study Verification

Objective: Determine whether findings are directly applicable to human populations.

Cell Culture: Studies using cell lines
Animal Models: Studies using animal subjects
Human Studies: Research conducted in human participants

Cell culture and animal model studies are categorized as Not Tested in Humans and require further human research.

5 Study Type Classification

Objective: Categorize based on specific study design to guide interpretation of evidence strength.

Study Tags:
  • Observational Study: Researchers observe participants without intervention (cohort, case-control, cross-sectional studies)
  • Clinical Trial: Experimental study with active intervention testing effectiveness and safety of treatments
  • Women Not Included: Tracks gender representation when women/females were excluded
Population Representation:
  • Ethnicity of study participants
  • Age ranges of participants

6 Sample Size Classification

Objective: Assess robustness and generalizability based on participant numbers.

Less than 100 participants: Limited Tested in Humans
100 to 500,000 participants: Tested in Humans
More than 500,000 participants: Widely Tested in Humans

7 Paper Stance

Objective: Conclude whether the paper's evidence supports or contradicts the claim being reviewed.

Supporting: The paper's evidence is consistent with the claim.
Contradicting: The paper's evidence runs against the claim.

Stance is determined by the reviewer at the end, based on all prior steps. It is not assigned at intake.

Classification Outcomes

Depending on how a publication moves through the workflow, it is assigned one of the following classifications:

Invalid
The study has fundamental issues (conflict of interest, is a review/meta-analysis, etc.)
Inconclusive
The study has quality issues in design, control group, bias handling, or statistics
Not Tested in Humans
The study was conducted on cells, animals, primates or other non-human systems
Limited Tested in Humans
The study has fewer than 100 human participants
Tested in Humans
The study has 100-500,000 human participants
Widely Tested in Humans
The study has more than 500,000 human participants

Expert Review Comments

After completing all seven steps, experts provide a comprehensive summary including: