Mobility Atlas research: walking-cane-guide
Data and method for the article at mobilitatlas.com/blog/walking-cane-guide/
License: CC BY 4.0 (credit "Mobility Atlas")
Prepared: 17 October 2026

WHAT WAS READ
Every FDA MAUDE medical device report with text filed under product code IPS (Cane; 21 CFR 890.3075), KHY (Cane, Safety Walk; 21 CFR 890.3075) or IPR (Crutch; 21 CFR 890.3150), and every report with text filed under INP (Tips And Pads, Cane, Crutch And Walker; 21 CFR 890.3790) whose text names a cane or a crutch, received at any date up to the latest openFDA update read on 17 October 2026 (data last updated 7 October 2026): 1,214 reports, all coded with one codebook for both devices. 597 are about canes, 609 about crutches and 8 about another device (coded O). This data folder is shared by the Mobility Atlas cane and crutch guides.
All 19 FDA recall records for product codes IPS, IPR, KHY and INP, downloaded through openFDA on 17 October 2026 (recall_facts.py).
Makers: grouped from the manufacturer and brand names in each report (Medline, Maxhealth, Sunrise/Guardian, Drive, Invacare, HurryCane, other).

HOW IT WAS CODED
1. A codebook (codebook.txt) was written before any report was coded. Its SHA-256 fingerprint, recorded at 2026-10-17T12:05:55Z, is e38637d9df80225c1dd5a4fdb013371a4600c0adeb6c96f1360a2293813c981e. 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 96.6% (kappa 0.957); part named 93.3% (kappa 0.923); mode 94.4% (kappa 0.924); injury 99.5% (kappa 0.989); fall 99.3% (kappa 0.986); arrival 99.7% (kappa 0.979); what was found 98.6% (kappa 0.948).
4. Every report where the coders differed was decided by the lead analyst under the written rules below (200 reports marked adjudicated). A consistency pass then re-applied the rules across the whole table and changed codes in 207 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 (kind of crutch or cane not stated): use the text first, then the header; a cane with no other detail is S and a crutch with no other detail is K. A needs an underarm sign in the text or header (underarm or arm pad, redness under the arms, axillary, top or crossbar, wing tube or wing nut) or the codebook words standard, aluminium or wooden; L needs forearm, elbow, Lofstrand, Canadian, a cuff or an arm band; size words (adult, youth, tall, junior, adolescent), push-button and a brand alone do not decide. A report that names no device at all is coded by the product code it is filed under (IPS S, KHY Q, IPR K). Invacare model numbers whose kind another report in the pool states are coded by that kind: 3920-2 and 7827 quad canes (Q), 8153 forearm crutch (L), 8115 axillary crutch (A).
C2: DEVICE (quad, self-standing, folding, other): Q when the report gives the cane a base, base plate, welded base, several legs or several feet; T when the cane stands on its own because of a pivoting, triangular, tripod or quad foot at the end of a single shaft (every Hurrycane, a cane with self-standing tips, a folding cane with a pivoting quad base), even when it also folds, because the codebook names that feature under T; F only for a folding cane without such a foot; a cam-lock toggle is a height lock, not a fold; O for a walker or another device, with PART N and MODE O.
C3: PART (whole device): when the report says the cane or crutch itself bent, buckled, snapped, or broke in half, in two or into pieces (also near the base or on the lower half), the failed part is the shaft S, the only part that can do that, with MODE D for bent or buckled and K for snapped or broke; it is Q only when the report says no more than that it broke, collapsed, fell apart, wobbled or was unstable, and with PART Q the MODE is O as the codebook says.
C4: PART (height adjustment): J for the push, snap or detent button, pin, clip, cam lock, plastic lock piece and locking collar (a collar on a forearm crutch is the locking collar, not the cuff), and for a tube that broke, cracked or bent at an adjustment hole (height holes or hand-grip holes), because the codebook puts wear or cracking at the adjustment holes under J; a break at the extension or near another hole (a structural hole, a rivet) with no adjustment hole named is S; a red clip is coded by what it holds (the hand grip H, the leg or crutch height J).
C5: PART (pieces and hardware): the plastic brace, bracket, clamp or piece at the bottom of an axillary crutch that joins the uprights to the lower leg, the foot piece or extension leg, the bow and the fold joint of a folding cane are S; the cuff post, cuff rivet, cuff screw, cuff bolt, cuff mounting sleeve and the locking nut of the cuff are C; a wing nut, bolt or rivet that holds or adjusts the hand grip is H; a screw that holds a foot or tip is T; W only for hardware the report ties to no part, including a frame rivet the maker places on the frame.
C6: PART (welds): a weld on a cane with no stated location is B, because every located cane weld in this pool is at the base, the base plate or where the legs attach; a weld at the handle is H; a weld that separated is MODE K.
C7: PART and MODE (tips): rubber tips, caps, glides and feet at the end of a shaft or a quad leg, and the pivoting or triangular foot of a self-standing cane, are T; the metal base, plate and legs are B; a tube that cut, punched, pierced or wore through the tip, or a tip that is falling apart, busting or tearing, is MODE W; a tip that split or cracked is K, came off L, slipped S.
C8: PART and MODE (several parts, order, maker evidence): when several failed parts are named with no order of failure the first in the codebook list is coded (T, B, J, S, H, P, C, W, Y), separate complaints in one report included; when the reporter or the maker gives a sequence (a cracked armpad that led to the tube break, a hand-bar piece that collapsed in the event) the first in that sequence is coded; when the maker's inspection or photo review confirms the failure and states the part or how it failed more precisely than the reporter (clip broken off, washer fractured, foot piece bent, bolt missing), that description is coded; when the maker finds no fault or says the fall caused the break, the allegation is kept, as the codebook says.
C9: PART X vs Q vs a named part: X when nothing is alleged to have failed (an intact part that is too large or does not fit, a fall or injury while using the device, a cut from an intact part when the report states no malfunction); Q when the report blames the device for the harm or instability but names no part or problem (an attorney claim, unstable, an unidentified bracket or pad); a named part alleged to harm by its material or shape (a latex tip, a sharp corner where the hands rub) is coded as that part with MODE O.
C10: MODE (general): the first mode the report states is coded; broke off and snapped off are K; peeling of a cover is W; for PART J: K when the part broke, L when the button, pin, clip or piece popped out, fell out, slid or disengaged, D when the cane or crutch collapsed, buckled or telescoped and that is stated first or the button movement is not described, O when the lock merely did not hold or was missing with no movement described; a missing bolt is O.
C11: INJURY: y when the text states harm to the body (pain, bruise, scrape, cut, fracture) or care; n when it states only that the user hit the head or hit something, that a cast broke, or that injuries were claimed by a third party and not confirmed by the user.
C12: FALL: y when the user fell, went down, dropped down, tried to use the device to catch a fall, or an injury is stated that only going down could cause (ribs broken after the crutch broke, severe injury when the crutches collapsed on stairs); n when the user almost fell without hitting anything or only slipped with no fall stated.
C13: ARRIVAL: y when the problem was found new, out of the box, in the package or at first handling or use (including new crutches that left a metal taste at first touch); n when the maker only suggests possible shipping damage noticed later, or when a fall at first use has no problem found.
C14: FINDING: D when the maker's own inspection, photo review or test confirms the reported failure, even with the root cause unknown; N when the maker examined the sample or a photo and states it shows no fault (intact, works as designed, part not fractured), or names an outside cause, judged for the part that is coded; X when the device was not evaluated or the maker draws no conclusion (not returned, hedged text such as most likely user instability but unknown if a malfunction occurred, device history, lot or inspection record review only, a photo or video that cannot show the issue, warranty or goodwill replacement, a lot pulled by a facility) and when the opinion is that of a dealer, provider, facility or user.
C15: PAIRS: reports that repeat the same event in the same words get identical codes.

KEYWORD COUNTS (Python regular expressions, matched against the upper-case report text without its header line)
KW = {'fracture': r'FRACTUR', 'head': r'\bHEAD\b', 'hip': r'\bHIP\b', 'wrist': r'WRIST', 'stairs': r'STAIR', 'surgery': r'SURGER', 'emergency': r'EMERGENCY|\bER\b|HOSPITAL',
      'weight': r'\d{3} ?(LB|POUND)', 'button': r'BUTTON', 'weld': r'WELD', 'rivet': r'RIVET', 'tip': r'\bTIPS?\b|FERRULE', 'new_box': r'OUT OF THE BOX|BRAND NEW|NEW OUT|FIRST USE|FIRST TIME', 'ice': r'\bICE\b|\bICY\b|\bSNOW|\bWET\b|\bRAIN\b'}

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 walking-cane-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."
- Most 2026 reports with text are one maker's filings (Medline); a few makers filed most of the 2011-2014 reports. Both reflect filing, not product changes.
- 356 crutch reports do not say which kind of crutch.
- Reports go back to 1992; old reports describe products that may no longer be sold.
- After adjudication, a consistency pass re-applied the written rules to rows both coders had agreed on and changed 207 of them (marked consistency).
- 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
The full report texts are public: search MAUDE or openFDA by report key.
walking-cane-guide-documents.csv  the documents and pages read, with links