Methodology & Data Sources

Editorial Workflow

Content on PlainSafety is compiled by our editorial process. Raw data from CPSC NEISS, SaferProducts.gov, and CPSC recall announcements is ingested programmatically; narrative framing, guide text, rankings commentary, and methodology writeups are drafted and reviewed by our editorial team; the data pages themselves are generated from the source records rather than written individually. We follow rigorous editorial standards: source data is loaded directly from official agencies, never invented or interpolated. The templates, methodology and written guides are human-reviewed before publication; the generated data pages are not reviewed one by one, and we would rather say so than imply a level of individual review that a corpus this size does not receive. We do not accept payment for coverage, placement, or rankings, danger scores and all "most dangerous" lists are computed directly from CPSC public data.

Data Sources

PlainSafety combines three US Consumer Product Safety Commission (CPSC) public datasets. Their update schedules differ, so each source's coverage is stated below rather than presented as a live feed.

• NEISS (National Electronic Injury Surveillance System): A stratified probability sample of approximately 100 US hospital emergency departments, collecting product-related injury cases continuously since 1970. From sample cases, CPSC projects nationally representative estimates of product-related ER visits. Our database covers 2005-2024 (~7.3 million case records projecting to ~310 million national estimates).
• SaferProducts.gov: Consumer-submitted incident reports filed directly with CPSC. The portal currently uses a historical snapshot of more than 65,000 public reports; its former incident-report API endpoint is retired, so this snapshot is not presented as a live or continuously refreshed feed.
• CPSC Recalls: Official recall announcements including hazard descriptions, remedies, affected units, and product categories. PlainSafety refreshes this separate CPSC recall feed in batches; always confirm the current recall status with CPSC before acting.

Cross-referenced where available with the FDA Recalls, Market Withdrawals & Safety Alerts and the NHTSA National Center for Statistics & Analysis for context on cross-agency product-safety enforcement actions.

Danger Score Calculation

PlainSafety computes a danger score (1-100) for each of the 838 product categories using a composite of four factors:

• Injury volume (40% weight): Annual average estimated ER visits from NEISS, normalized to a 0-100 scale within the full product category population
• Hospitalization rate (25% weight): Percentage of NEISS cases requiring hospital admission, indicating injury severity
• Fatality rate (25% weight): Deaths associated with the product category from NEISS death module and case narratives
• 5-year trend (10% weight): Whether ER visit volume is increasing, decreasing, or stable compared to the prior 5-year average

A score of 50 means the product is more dangerous than half of tracked categories. Scores are relative rankings, not absolute risk probabilities.

Composite Safety Score

Every product page also shows a Composite Safety Score (0-100 + A-F grade) distinct from the danger score above: where the danger score is built entirely from NEISS injury-surveillance sub-factors, the composite blends that surveillance data with a second, independent CPSC data system: the public SaferProducts.gov consumer incident report count. Each dimension is benchmarked against the full 838-category population by percentile rank, not raw-value scaling, since a linear scale would let a huge-volume category (avg_annual_injuries ranges from 18 to over 1.5 million) dominate the composite.

For each dimension, a category's raw value is located against the population's p10/p25/p50/p75/p90 percentile breakpoints and linearly interpolated to a 0-100 percentile rank (0 = safest end of the population, 100 = most severe). The composite is the weighted average of the dimensions that have data. 5-year trend has no recorded value for 41 of 838 categories (insufficient prior-period NEISS sample); for those categories its 15% weight is redistributed proportionally across the other three dimensions rather than counted as zero risk or dropped silently, so a category is never penalized or favored for a dimension we simply lack data on. The letter grade follows: A (0-19, safest) · B (20-39) · C (40-59) · D (60-79) · F (80-100, most severe), the same bands the danger score's Low/Moderate/High/Very High labels use, extended to five letter grades.

Corpus placement (#N of M)

Product and category detail pages show an analytical #N of M placement inside the CPSC NEISS-derived PlainSafety corpus. These ranks describe relative standing in our published inventory:

• Products - danger score: national rank by composite danger score among scored categories (highest score = #1; ties share the best rank).
• Products - injury volume: national rank by total_injuries_20yr (highest estimated ER injuries = #1). Tie-break: product slug A-Z.
• Categories - ER volume: rank by summed average annual ER visits across scored products in the group (highest = #1), excluding uncoded/special groups. Tie-break: group name A-Z.

Corpus placement answers "where does this entity sit among the scored CPSC categories we publish?" It is not a per-use risk probability. Always read the score methodology above and the primary CPSC sources before acting.

NEISS Sampling Methodology

NEISS uses a stratified probability sample of ~100 hospitals selected to represent the ~6,000 US hospital emergency departments. Each case is assigned a statistical weight based on the sampling probability, allowing projection to national estimates. Sample hospitals are rotated periodically to maintain representativeness.

Statistical note: For product categories with few cases in the sample, national estimate projections carry higher coefficients of variation (statistical uncertainty). We follow CPSC guidelines and note when estimates are based on small sample sizes.

Product Category Linking

NEISS, SaferProducts.gov, and CPSC recalls each use their own product category codes. We link these three datasets by:

• NEISS product codes (2-digit and 5-digit) mapped to product category names
• SaferProducts.gov product types mapped to matching NEISS category names
• CPSC recall product categories matched to NEISS codes using CPSC's published crosswalk

Update Schedule

NEISS data is collected continuously and published annually by CPSC. CPSC recalls are issued as hazards are identified and are refreshed here in batches. The historical SaferProducts report snapshot has no working replacement export at present, so new incident reports may not be reflected. Between updates, newly issued recalls may also be absent. Danger scores are recalculated with each annual NEISS update to reflect the most current injury patterns.

Limitations

• Danger scores are based on emergency department injuries, not all injuries. Many product-related injuries are treated at home or by primary care providers and are not captured by NEISS.
• High absolute injury volumes may reflect widespread product use, not inherent product dangerousness. Per-user risk requires knowing how many households use each product.
• SaferProducts.gov reports are voluntary and not nationally representative. Dramatic incidents generate more reports than minor but more common injuries.
• CPSC recalls reflect official government action and may lag consumer reports by months or years.
• Product category crosswalks between NEISS, SaferProducts.gov, and CPSC recalls are approximate, some products may not link perfectly across all three data sources.

Not Affiliated

PlainSafety is not affiliated with the Consumer Product Safety Commission or any government agency. This site is for informational purposes only and does not provide safety certifications or product endorsements.

Frequently Asked Questions