Which recovery equipment Medicare pays for — and which has been recalled

Medicare publishes what it pays for durable medical equipment. The CPSC and FDA publish what has been recalled. Nobody publishes them together, so nobody can answer the question a patient actually has the week before surgery: is the thing I am about to buy both covered and safe? This page is that join, for the 68 equipment codes that matter during recovery at home. The whole dataset is free to download and reuse.

The fact

Medicare paid an estimated $179,520,882 across 4,218,036 services in DY2024 for the 68 HCPCS codes covering surgery-recovery equipment. 28 of those 68 codes have at least one CPSC or FDA recall record attached to the equipment they describe.

Utilization is DY2024, published with roughly an 18-month lag — the most recent year CMS has released. 96 of 102 recall records match a code in this scope; the remaining 6 are listed in the dataset rather than dropped.

Source: CMS DMEPOS Fee Schedule Jul2026 (DME26-C) + CMS DMEPOS utilization DY2024 + Medicare PDAC DMECS + CPSC SaferProducts + openFDA

Download CSV — 68 rows Download JSON — full records + provenance

CC BY 4.0. Cite as: VerifiedCareData, “Which recovery equipment Medicare pays for — and which has been recalled,” 2026-08-29. No affiliate links, tracking parameters, or account required.

Three things the join shows that neither source shows alone

1. The one item here that has killed people is one Medicare pays for

Bed side rails are the only equipment in this dataset whose recall narratives name deaths: 13 across six CPSC recalls, every one an adult portable bed rail, with entrapment and asphyxiation as the stated hazard. They carry 35 recall records in total, and the pace is current rather than historical — five of them landed in 2026 alone. Medicare covers them. It billed 20,821 services for the two codes in DY2024, at a purchase fee of $12.31–$22.32 for half-length (E0305) and $129.76–$203.16 for full-length (E0310).

Coverage is not a safety signal. It is a payment category, and this is the clearest case of the two coming apart. If you are borrowing or buying a bed rail secondhand, check the model against CPSC before it goes on the bed.

2. Bath and transfer safety equipment is excluded by statute — and it has been recalled

All 9 bath and transfer codes (E0240–E0248: shower chairs, tub rails, raised toilet seats, transfer benches) show zero national claims and no fee-schedule entry. That is not an oversight and not a quality judgement: bath-safety equipment is excluded from the DME benefit category by statute — Social Security Act §1861(s)(6), the “primarily medical in nature” test. There is no appeal that turns a shower chair into a covered item.

Meanwhile 7 distinct recall records in this dataset cover exactly that equipment — shower and shower/commode chairs from Aquatec, Invacare, ActiveAid and others, all through FDA device recall channels. So the category you must pay for entirely out of pocket is also one where the recall history is worth checking yourself, because no supplier is doing it for you as a condition of billing Medicare.

3. Recall volume does not track how many products exist

It would be reasonable to assume the most-recalled category is simply the one with the most models on the market. It is the opposite here, and the gap is wide.

The folding wheeled walker (E0143) is the busiest item in the dataset — 508,460 services and $24,064,635 in DY2024, across 1,002 distinct products from 103 manufacturers. It carries 17 recall records and no deaths. Bed side rails carry 35 records and 13 deaths from a catalogue of 47 products — a catalogue roughly one twentieth the size.

So the concentration is real and it is not a reporting artifact: a product category about twenty times smaller than walkers produces twice the recalls and all of the deaths. If you take one operational thing from this dataset, it is that adult portable bed rails deserve a model-number check before use in a way a walker does not.

What Medicare spends the most on

Top 10 of 68 codes by estimated DY2024 Medicare payment.
CodeEquipmentDY2024 servicesDY2024 paymentProductsRecalls
K0005 Ultralightweight wheelchair 21,620 $40,854,859 298
E0260 Hospital bed, semi-electric (head and foot adjustment), with any type side rails, with mattress 758,660 $37,429,249 169
K0001 Standard wheelchair 1,305,390 $24,581,790 991
E0143 Walker, folding, wheeled, adjustable or fixed height 508,460 $24,064,635 1,002 17
K0003 Lightweight wheelchair 471,617 $13,079,796 636
E0163 Commode chair, mobile or stationary, with fixed arms 139,220 $7,078,214 202 5
E0261 Hospital bed, semi-electric (head and foot adjustment), with any type side rails, without mattress 122,881 $5,939,865 52
K0006 Heavy duty wheelchair 66,273 $3,364,492 310
K0007 Extra heavy duty wheelchair 42,671 $3,146,791 699
K0002 Standard hemi (low seat) wheelchair 69,565 $2,147,958 549

What Medicare does not pay for at all

12 of the 68 codes appear in no Medicare fee schedule and recorded no national claims in DY2024. Nine of them are the statutory bath-safety exclusion above. The others are listed here so the absence is legible rather than inferred.

CodeEquipmentProducts on marketRecalls
E0118 Crutch substitute, lower leg platform, with or without wheels, each 47
E0172 Seat lift mechanism placed over or on top of toilet, any type 4
E0240 Bath/shower chair, with or without wheels, any size 1 2
E0241 Bath tub wall rail, each 0
E0242 Bath tub rail, floor base 0
E0243 Toilet rail, each 0
E0244 Raised toilet seat 0
E0245 Tub stool or bench 0
E0246 Transfer tub rail attachment 0
E0247 Transfer bench for tub or toilet with or without commode opening 1 5
E0248 Transfer bench, heavy duty, for tub or toilet with or without commode opening 0 5
K0008 Custom manual wheelchair/base 0

How the join works, and where it is imperfect

A recall attaches to an HCPCS code when one of the search terms the recall was retrieved under appears in that code's CMS long_description as adjacent words, tolerating a plural and at most one intervening word (so the term 'bed rail' reaches CMS's 'Bed side rails'). One intervening word and no more: two would also pull 'bed rail' into 'Hospital bed, ... with any type side rails', and a hospital bed's integrated rails are not what those recalls are about. Derived from the two source files, with no editorial mapping table.

The honest limits, all of which are also stated inside the dataset:

One consequence is worth stating plainly: do not sum the recall column. A recall that matches several codes is counted under each, so the 35 bed-rail records appear under both E0305 and E0310. Distinct totals live in the dataset's totals object, and every figure on this page uses those.

Correction, 29 August 2026

The first version of this page, published earlier today, overstated every recall count. It reported 191 recall records; the correct in-scope figure is 102. Three faults, all in how the source file was assembled rather than in the agencies' data:

  • Duplicates. openFDA's device/recall and device/enforcement endpoints return the same recall, and the original dedupe compared dates the two endpoints format differently (2024-08-26 against 20240826), so overlapping records were kept twice.
  • Out-of-scope matches. The agency searches run on product-name text, so “walker” returned infant walkers, activity push toys, a Classic Walker Wagon, Skywalker Via Ferrata climbing sets and Sky Walker tree stands; “bed rail” returned children's bed rails, a different product under a different mandatory standard.
  • Wrong device. “Rebound Air Walker” and “Rebound Diabetic Walker” are orthopedic walking boots, not mobility walkers.

The correction changed a conclusion, not just a number. The earlier version argued that recall volume tracks how many models exist. On the cleaned data that is false and backwards: bed side rails produce more recalls and all of the deaths from a catalogue roughly one twentieth the size of walkers'. Section 3 above is rewritten accordingly.

The filter is deliberately conservative — a record that cannot be confidently placed in scope is excluded — so these counts are a floor. The rules, with a reason attached to each, are in data/pipeline/clean_recalls.py and echoed in the dataset's own metadata. Some openFDA recalls are still listed once per colour or SKU; the counts are therefore recall records, not recall events.

Use it

The dataset is CC BY 4.0. Republish it, chart it, load it into a model, check our arithmetic. The only ask is attribution back to this page so a reader can trace a number to CMS, CPSC and FDA rather than to a summary of a summary. If you find an error, tell us and it gets fixed and noted, not quietly overwritten — see how we verify.

Sources

  1. CMS DMEPOS Fee Schedule, July 2026 (DME26-C) — purchase (NU) fee ranges, min/max across state non-rural columns
  2. CMS Medicare DMEPOS utilization by Geography and Service, DY2024 — national services, beneficiaries and estimated payment; ~18-month publication lag
  3. CMS HCPCS Level II quarterly file, July 2026 release — code existence and official long descriptions
  4. Medicare PDAC Product Classification List (DMECS) — 9,270 products classified to these codes
  5. CPSC SaferProducts.gov Recall API — consumer recalls; the only source for the deaths-mentioned field
  6. openFDA Device Recall API — device recalls, phrase-quoted product searches
  7. openFDA Device Enforcement API — enforcement reports for the same equipment terms
  8. Social Security Act §1861(s)(6) — the DME benefit definition behind the bath-safety exclusion