Mobility Atlas research: walker-rollator-guide
Data and method for the article at mobilitatlas.com/blog/walker-rollator-guide/
License: CC BY 4.0 (credit "Mobility Atlas")
Prepared: 16 October 2026

WHAT WAS READ
FDA MAUDE medical device reports filed under product code ITJ (Walker, Mechanical; 21 CFR 890.3825) or NXE (Walker, Mechanical, Poly Vinyl Chloride), read in openFDA on 16 October 2026, that carry text: (a) every such report received 1 January 2019 to 31 December 2025 (786); (b) every such report received in 2026 from a maker other than Medline (77); (c) a systematic sample of 300 of the 4,124 reports with text that Medline filed in 2026, taken as 50 consecutive reports at six evenly spaced positions of the openFDA result list (skips 0, 687, 1375, 2062, 2749 and 3437). Pool: 1,163 reports, all coded; 6 describe no walker event (coded N), leaving 1,157. The stratum of each report is in strata.csv.
All 29 FDA recall records for product codes ITJ and NXE, downloaded through openFDA on 16 October 2026 (recall_facts.py).
Makers: grouped from the manufacturer and brand names in each report (Drive, Medline, Compass Health/Carex, Roscoe, Invacare, Nova, other).

HOW IT WAS CODED
1. A codebook (codebook.txt) was written before any report was coded. Its SHA-256 fingerprint, recorded at 2026-10-16T04:17:56Z, is 81d766a73c58a26d623505b9b7a663a714e601542f236eb7cb393c9ebf05eb03. 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 94.9% (kappa 0.894); category 94.8% (kappa 0.937); injury 99.0% (kappa 0.977); fall 99.6% (kappa 0.991); what was found 97.5% (kappa 0.931).
4. Every report where the coders differed was decided by the lead analyst under the written rules below (155 reports marked adjudicated). A consistency pass then re-applied the rules across the whole table and changed codes in 0 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
W1: INJURY: y only when the text names harm to the body, pain, or medical care, or the maker states the event caused or contributed to serious injury; a head strike, hitting a body part, or a fall with no harm named is n, and the event type Injury alone is not enough (Death is y).
W2: FALL: y only when the text says the user fell or was thrown to the ground (including a mention of "the fall" as having happened); a device that folded, collapsed or slid away with no fall stated, or a leg slamming down while the user stayed on the device, is n.
W3: DEVICE: R when the text or brand line says rollator, or the text names hand brakes or wheel locks, a seat or backrest, the user sitting on the device, four wheels or rear wheels; K for a knee walker or scooter, a steering limiter or tiller, or a user whose leg rests on the device; P for a gait trainer, child walker, bariatric or heavy-duty walker; W only when a standard, folding, front-wheeled (wheels plus back caps or glides) or hemi walker is described.
W4: DEVICE: "walker" (or "the device") with no kind named and none of the W3 cues is U, even when a part (leg pin, fold button, wheel, weld) could belong to several kinds and even when the brand line only says walker; a "mobility scooter" with a swivel seat is U.
W5: CATEGORY: by precedence S wins over wheel or frame whenever the user fell while sitting on or getting onto the seat (or was dumped from it in transport or wheelchair mode), and when the report says the seat or seat padding, seat bar or backrest (including its holder) broke, bowed or collapsed; a part that broke while the user sat with no fall is coded by that part.
W6: CATEGORY: damage found on arrival, out of the box, at assembly before use, or a missing part with no use event, is A, because categories 1-8 describe failures in use; when the maker's evaluation refutes the alleged problem and finds a different failed part on a used unit, the found part is coded.
W7: CATEGORY: a fall with no failure from 1-7 named is L, including a walker that tipped, moved unexpectedly, felt unstable or stopped at a threshold while the user walked, and a part bent or broken by or after the fall; N when no walker-family device is named or described in the event; X when an injury from use is alleged with no device problem or mechanism, or a device tipped with no fall and no problem named.
W8: CATEGORY: a whole walker that "collapsed", "folded" or "fractured" with no part named is F (frame or folding); "it broke" or "the top twisted" with no part named, an unstable design complaint with no event or fall, a device that "locks" when pushed (brake or wheel unclear), an unidentified knob or side arm, and a wheel off the ground or a crooked unit with no broken part named are U.
W9: CATEGORY: parts that carry the wheel itself (fork, caster, axle, stem bolt, wheel bracket, caster insert, bearing) are H; frame tubing, legs, welds, the tube a wheel inserts into, height or wheel adjustment knobs and their threads, fold buttons and pins are F; a handle or knee-walker tiller that broke, bent or spun is G.
W10: CATEGORY: when the report names a brake problem (brakes did not hold, needed adjusting so a wheel stuck, or brake issues reported with the event) B wins over wheel or unclear by precedence.
W11: FINDING: D when the maker states it confirmed the reported failure on the unit or from photos, found a broken or loose part, or recorded a cause code of supplier or manufacturing error, even if the root cause is unknown, shipping damage, or the investigation is otherwise marked not confirmed; unless W12 applies.
W12: FINDING: N when the maker attributes the problem to a cause outside the device (misuse, self-propelling, inadequate maintenance, overtightening, extreme external force or impact, overuse beyond rating, stairs) even if it confirmed a broken part, or when its review of the unit or photos found the components normal or no fault.
W13: FINDING: X when the device was not returned or evaluated, the investigation is pending, the maker only quotes warnings or the manual, cites tests of other units or the batch, or reports a customer's or family member's view, or evaluated the wrong unit, or says the issue is unconfirmed and a sample is required.

KEYWORD COUNTS (Python regular expressions, matched against the upper-case report text without its header line)
KW = {
 'fracture': r'FRACTUR|BROKE (HER|HIS|THEIR|A|AN)? ?(LEFT |RIGHT )?(HIP|ARM|LEG|WRIST|RIBS?|ANKLE|FOOT|SHOULDER|FEMUR|ELBOW|PELVIS|BACK|NECK|COLLAR ?BONE)',
 'hip': r'\bHIP\b',
 'head': r'\bHEAD\b',
 'surgery_or_hospital': r'SURGERY|HOSPITAL|EMERGENCY ROOM|\bER\b',
 'stairs_curb': r'STAIR|STEP|CURB',
 'pushed_while_seated': r'PUSH\w* (HER|HIM|THEM|THE USER|THE END.USER|THE PATIENT)|TRANSPORT|WHEELCHAIR MODE|WHILE (BEING )?PUSHED',
 'sitting': r'\bSAT\b|SITTING|SIT DOWN|SEATED',
 'weld': r'WELD',
 'brake_cable': r'CABLE',
 'adjust': r'ADJUST',
 'new_out_of_box': r'OUT OF (THE )?BOX|NEW |ARRIV|ASSEMBL',
 'weight_limit': r'WEIGHT (LIMIT|CAPACITY)|OVER ?WEIGHT',
}
PK = {'brake_part': r'BRAKE', 'finger': r'FINGER|THUMB|HAND', 'fold': r'FOLD', 'cut_or_laceration': r'\bCUT|LACERAT|STITCH', 'leg': r'\bLEG|SHIN|CALF|ANKLE|KNEE'}

QUALITY NOTES (disclosed in full)
- The article went through 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 walker-rollator-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."
- The Medline 2026 reports are a sample of one maker's retrospective batch; they are counted separately and not added to the other strata.
- One maker files most of the 2019-2025 reports, which reflects its size and filing, not its safety.
- The codebook puts any fall while sitting on a rollator seat under seat, even when a wheel or leg broke at the same time.
- The same event is sometimes filed twice; we count reports, not events.

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
strata.csv          which part of the pool (stratum) each report came from
The full report texts are public: search MAUDE or openFDA by report key.
walker-rollator-guide-documents.csv  the documents and pages read, with links