Mobility Atlas research: wheelchair-cushion-guide
Data and method for the article at mobilitatlas.com/blog/wheelchair-cushion-guide/
License: CC BY 4.0 (credit "Mobility Atlas")
Prepared: 12 October 2026

WHAT WAS READ
Every FDA MAUDE medical device report filed under product code IMP (Cushion, Wheelchair, 21 CFR 890.3920) or KIC (Cushion, Flotation, 21 CFR 890.3175), as downloaded through openFDA on 11 October 2026: 313 reports, 308 with text, all coded. 248 concern a cushion used on a wheelchair or a seat; the rest concern other products filed under these codes.
All FDA recall records for product codes IMP and KIC, downloaded through openFDA the same day (two records).

HOW IT WAS CODED
1. A codebook (codebook.txt) was written before any report was coded. Its SHA-256 fingerprint, recorded at 2026-10-11T11:19:03Z, is 14bfcb9d56dfd8ccbe8ba6caeef2e4d7460b2991ab223c7b0d0a09df1bb3b493. Any later change to the codebook would change this fingerprint.
2. The reports were split into chunks. Two coders (AI language models working separately, each in a private scratch folder, each without access to the other's answers, told to read every report and not to use keyword scripts) coded every report in every chunk against the codebook.
3. Agreement between the two coders: device 86.4% (kappa 0.775); category 96.4% (kappa 0.952); injury 99.4% (kappa 0.987); skin damage 99.7% (kappa 0.993); fall 99.7% (kappa 0.968); what was found 97.7% (kappa 0.956).
4. Every report where the coders differed was decided by the lead analyst under the written rules below (56 reports marked adjudicated). A consistency pass then re-applied the rules across the whole table and changed codes in 24 more reports (marked consistency).
5. Counts in the article were produced by script from the final table, not by hand. Some sub-counts (parts, places or words named in reports) are keyword counts over the report text; the exact word patterns are listed under KEYWORD COUNTS. Different wording would give somewhat different counts.

ADJUDICATION RULES
C1: DEVICE: a cushion or back named by a wheelchair seating brand or model (ROHO, Jay, ContourU, Matrx, Varilite, Vicair, Air Care, an Invacare or Posey wheelchair seat) or a flotation cushion sliding off a wheelchair seat is W unless the text puts it on a bed, sofa or other chair.
C2: Hardware of a wheelchair back or positioning-cushion system (back shell, back pin, laterals, adductors, mounting brackets) is DEVICE W, CATEGORY M.
C3: Air bubbles in a fluid or gel bladder, or a fluid pad that went hard, is M (material), not A.
C4: A pressure injury with a set-up error only suspected, or a cushion said not to shape or work with no failed part found, is P.
C5: When a worn part and slipping are both named, the precedence list gives M.
C6: FINDING: a cause the maker calls only likely, with the cushion not returned or not examined, is X; a result line or evaluation naming a defect is D; malfunction not confirmed is N.
C7: SKIN is n for a blister on the back (not in the codebook list); INJURY is n when harm is not stated.
C8: A cushion on a sofa or an ordinary chair is DEVICE B.
C9: A vague complaint ("tailbone problem") with no injury or failure named is U.

KEYWORD COUNTS (Python regular expressions, matched against the upper-case report text without its header line)
KW_A = {
 'valve': r'VALVE|STEM',
 'not_noticed': r"NOT NOTICE|DIDN'T NOTICE|DID NOT REALIZE|NO FEELING|NO SENSATION|MINIMAL FEELING|NO LOWER SENSATION|WITHOUT NOTICING|DID NOT NOTICE",
 'no_medical_records': r'NOT SEEN MEDICAL RECORDS|NOT RECEIVED MEDICAL RECORDS|UNABLE TO ACCESS MEDICAL RECORDS',
 'one_section': r'QUADRANT|QUAD\b|QUADS\b|QUADRAT|COMPARTMENT NEAREST|SECTIONS WERE DEFLATED',
 'surgery_or_hospital': r'SURGERY|HOSPITAL',
}
KW_M = {
 'back_hardware': r'\bBACK(REST)?\b.{0,60}(HARDWARE|SHELL|BRACKET|CLAMP|PIN|HOOK|CLIP|MOUNT|CRACK|SPLIT|STRIPPED|LOCK)|(HARDWARE|SHELL|BRACKET|CLAMP|PINS?|HOOK|CLIPS?)\b.{0,60}\bBACK',
 'foam_breakdown': r'FOAM|BOTTOM(ED|ING)? OUT|FLATTEN|COMPRESSED|SQUISHED|SINKING|LOP SIDED|DECOMPRESS',
 'gel_fluid': r'\bGEL\b|FLUID|LIQUID',
}

QUALITY NOTES (disclosed in full)
- The article went through three fact-check passes by checkers who had not written it (a full pass with a reader review, then a control pass over the changes), each recomputing numbers from the files here and reading quotes against their sources; every error found was corrected before publication.

AMOUNTS
Every report text was searched for money figures; none gives one. Prices in the article are prices shown on makers', dealers' and retailers' pages and in the Mobility Atlas catalog, listed in wheelchair-cushion-guide-documents.csv and the article sources.

LIMITS
- Reports are not rates. Nobody publishes how many devices of each model are in use.
- A report is an allegation or a description; the FDA says the submission of a report "itself does not necessarily demonstrate that the device caused or contributed to the adverse outcome or event."
- Few makers file most of the reports, which reflects their size and filing, not their safety.
- Many reports were never evaluated by the maker; a finding of no defect is not proof that nothing went wrong.

FILES
codebook.txt        the codebook as fingerprinted
coded_reports.csv   one row per coded report: both coders' answers, the final code, and how a disagreement was resolved (resolution, rule_applied)
report_pool.csv     the reports in the pool with product code, brand, date received and event type
The full report texts are public: search MAUDE or openFDA by report key.
wheelchair-cushion-guide-documents.csv  the documents and pages read, with links