How many DMEPOS suppliers actually serve each state?

Every territory plan in durable medical equipment starts with a list. How many suppliers sell this product in the state, which ones matter, and how many can one rep cover? The number usually comes from a purchased database, a trade association directory, or a regional manager’s memory.

There is a public file that answers the first part directly. CMS publishes, for every product code and every state, how many suppliers filled Medicare orders and for how many beneficiaries. Read naively, it gives a supplier count per state. Read carefully, it shows that the count means something different from what a territory plan assumes.

How many DMEPOS suppliers serve each state?

It depends on the product, and the range is wide. For CPAP device rentals in 2024, the median state had 275 suppliers filling Medicare orders. Florida had 1086 and Hawaii had 66, a 16.5x spread. For blood glucose test strips the median state had 1166 suppliers; for lumbar-sacral orthoses it had 37.0.

Those are counts of suppliers that filled at least one order written by a clinician in that state. They are not counts of suppliers located there. The distinction sounds pedantic until the state counts are added up. For CPAP rentals they sum to 18006, against 4454 distinct suppliers nationally. The same supplier is being counted in several states at once.

That overlap is the finding. A state supplier count in this market is mostly a measure of reach: how many suppliers, from anywhere, take orders from that state’s referring clinicians. For some products it is closer to a local count. For others it is almost entirely national suppliers reaching in.

Where does this data come from?

The Medicare Durable Medical Equipment, Devices & Supplies by Geography and Service file, published by CMS on data.cms.gov. It reports usage, supplier counts and referring-provider counts by product code and state. The release used here covers the 2024 data year and Medicare fee-for-service only.

Twelve product codes were pulled, chosen to span the categories a device or DME manufacturer would recognise as its own market: five rental devices (CPAP, bi-level respiratory assist, oxygen concentrator, semi-electric hospital bed, standard wheelchair), two positive airway pressure resupply items, a folding walker, two orthoses, and two diabetes supply codes. Drug codes were left out, because the pharmacy channel behind them has different intermediaries.

For each code, every state row and the national row were retrieved. The state table covers the 50 states and the District of Columbia, 51 geographies in all, and holds 607 rows. A second table holds one row per product code with the national supplier count and the sum of the state counts.

Why do the state counts add up to more than the national total?

Because the file assigns each order to the referring clinician’s state, not the supplier’s. A supplier filling orders from four states appears once in the national row and four times across the state rows. Dividing the summed state counts by the national count gives the average number of states each supplier draws orders from.

At the top, the average oxygen concentrator supplier drew orders from 7.65 states: 4192 suppliers nationally, appearing 32052 times across the state rows. At the bottom, the average lumbar-sacral orthosis supplier drew orders from 1.35 states: 1580 suppliers, 2129 state appearances.

A value near one describes a local channel. The supplier sits near the clinician, the product is fitted or delivered in person, and a state count is close to a count of businesses operating there. A value well above one describes a channel where suppliers routinely fill orders across state lines, which usually means central fulfilment, mail order, or a multi-branch operator filing under shared identifiers.

The ratio is a floor. CMS withholds any state figure built from very few beneficiaries, so a supplier with a handful of orders in a small state drops out of that state’s count but stays in the national total. The true footprint of every code in the chart is at least as wide as shown.

Which product channels are local, and which are national?

Resupply is the most national channel in the basket, and fitted orthoses the most local. For the same therapy, the channel changes shape between device and consumables: CPAP rental suppliers averaged 4.04 states, while suppliers of the full face mask averaged 5.37 and heated tubing 5.95.

That split inside one therapy matters because a manufacturer selling both the device and its consumables is selling into two differently shaped channels. The device rental starts with a set-up, often in person, near the patient. The replacement mask and tubing are shipped on a cadence, and the supplier shipping them does not need to be anywhere near the patient or the clinician who wrote the original order.

The mobility and bed rentals sit in the middle. Standard wheelchair rental averaged 2.65 states, semi-electric hospital beds 2.53 and the folding walker 2.9. These are bulky items delivered to a home, and the ratios are consistent with regional operators serving a state and its neighbours rather than with national shippers.

The two orthoses are the most local items in the file. The adjustable knee orthosis averaged 1.83 states and the lumbar-sacral orthosis 1.35. Both are typically fitted, and fitting happens where the patient is. A state count for these codes is the closest thing in the file to a count of local businesses.

The diabetes codes are a useful caution against reading too much into any single ratio. Test strips averaged 2.02 states and the glucose monitor supply allowance 1.73, lower than the positive airway pressure resupply items. Yet test strips have by far the largest supplier base, 39872 nationally. A large share of that base is retail pharmacy, and a pharmacy fills local prescriptions. The ratio measures geographic reach. Channel size moves independently of it.

Why do small states have more suppliers per beneficiary?

Because out-of-state suppliers set a floor under every state’s count. For CPAP rentals, the ten states with the fewest beneficiaries had a median of 51.7 suppliers per 1,000 beneficiaries; the ten largest had 21.4. That is a 2.4x gap, and it runs in the direction national reach would produce.

The file carries its own denominator (the number of beneficiaries on each product code in each state), so a per-beneficiary rate is available without importing a population figure from another source. That matters, because a per-capita rate would compare suppliers against the whole population instead of the Medicare fee-for-service population they actually serve.

The pattern holds across the basket. Across all 51 geographies, the log of a state’s CPAP beneficiary count and its suppliers per 1,000 beneficiaries have a correlation of -0.81. For heated tubing it is -0.82, for oxygen concentrators -0.78, and for the knee orthosis -0.74. The smaller the state, the more suppliers per beneficiary, on every code checked.

At the extremes the gap is large. The District of Columbia had 69.5 CPAP rental suppliers per 1,000 beneficiaries and Massachusetts had 15.4, a 4.5x difference between two dense, well-served jurisdictions. Semi-electric hospital beds show the widest small-to-large gap in the basket: 186.45 suppliers per 1,000 beneficiaries in the ten smallest states against 41.1 in the ten largest, a 4.5x difference.

These move together. The data does not show why. National reach is the explanation consistent with the footprint ratios above: a supplier active across many states shows up in every state’s count, and in a small state it is a larger share of a smaller base. A second explanation is that small states have genuinely fragmented local markets. The file cannot separate the two, because it never records where a supplier is.

What does a large state’s count actually contain?

A significant fraction of the national supplier base. Florida’s CPAP referrals reached 1086 suppliers, 24.4% of all CPAP rental suppliers in the country. For oxygen concentrators, Florida’s referrals reached 1710 suppliers, 40.8% of the national total. A supplier list for one large state is a large slice of the national one.

The same holds for resupply and diabetes supplies. Florida’s test strip referrals reached 5632 suppliers and New York’s 5625, each 14.1% of the national base. California, with more strip beneficiaries than either, reached 5432.

This is where state-based planning quietly goes wrong. Three large states’ supplier lists, merged, do not produce three times as many accounts. They produce a list with heavy overlap, and the overlap is largest precisely in the product lines where footprints are widest: oxygen and positive airway pressure resupply. In these categories, deduplicate by supplier identifier before counting accounts.

How was this calculated?

One request per product code against the CMS data API, filtered on product code and rental indicator, returning the national row and every geography row. State rows were kept for the 50 states and DC. States per supplier is the sum of state supplier counts divided by the national count. Data retrieved 2026-09-30.

The request, for any one code, is:

https://data.cms.gov/data-api/v1/dataset/27c150fd-8578-43b1-bba5-6388987e32af/data?size=5000&filter[HCPCS_Cd]=E0601&filter[Suplr_Rentl_Ind]=Y

Replace the code and rental indicator to reproduce any row. The response includes the national row, each state, the territories, and the Armed Forces, foreign and unknown geographies. Only the 51 state-level geographies are kept, because none of the others is a place a sales territory could be drawn around.

The rental indicator matters. The same product code can appear twice, once for suppliers renting the item and once for suppliers selling it, with very different counts. CPAP devices, oxygen concentrators, hospital beds, wheelchairs and bi-level devices were taken on the rental side, which is where their volume sits; the resupply items, walker, orthoses and diabetes codes on the purchase side.

Suppliers per 1,000 beneficiaries is suppliers divided by beneficiaries on the same code in the same state, multiplied by 1,000. Beneficiaries here are those with at least one order for that code; the rate is not comparable across codes. The small-state and large-state medians rank the 51 geographies by beneficiaries on each code and take the median rate in the bottom ten and top ten. The correlations use the natural log of beneficiaries, because state sizes span two orders of magnitude.

The full query string, every derived figure with its formula, and the dataset notes are in the sources file for this article.

What can this data not tell you?

Where any supplier is based. The file places each order in the referring clinician’s state and counts suppliers by identifier, never by address. Every statement here about local versus national channels is inferred from the footprint ratio. None of it is observed. A supplier appearing in eight states could be one central warehouse or eight branches.

That second possibility matters. Suppliers are counted by national provider identifier. A chain that registers each branch separately looks local: many suppliers, each in few states. A chain that files centrally looks national. The ratio in the chart mixes both behaviours, and the file gives no way to tell them apart.

Suppression biases every count downwards. State cells built from very few beneficiaries are withheld, which is why the lumbar-sacral orthosis appears in only 46 of the 51 geographies and why every footprint ratio is a lower bound. The smaller the state and the rarer the product, the larger the undercount.

The population is Medicare fee-for-service only. Medicare Advantage, commercial, Medicaid, cash and hospital-outpatient channels are absent. A supplier that does most of its volume outside fee-for-service Medicare can be invisible here or badly under-represented. For products where Medicare is a small share of volume, the counts describe only a slice of the channel.

Counts are per product code. A supplier serving all twelve codes in the basket is counted once in each, so the figures cannot be added across codes into a single “suppliers in this state” total. Twelve codes also sample the market; a manufacturer should pull its own.

Nothing here measures quality, share or size of any supplier. A supplier with one order and a supplier with ten thousand each count as one.

What should commercial ops do with a state supplier count?

Treat it as a count of who can reach that state’s clinicians, not a count of local accounts, and check the footprint ratio for the product before drawing territories around it. Where suppliers draw orders from four or more states, a territory built on state lines will split one account across several reps.

The practical sequence is short. Pull the national row and the state rows for the product codes that describe the portfolio. Compute the ratio. For codes near one, a state-based territory and a state supplier list are reasonable approximations of the market. For codes well above one, build the account list nationally, deduplicated by supplier identifier, and assign accounts to owners before assigning geography.

The ratio also predicts where attribution will break. A supplier filling orders across many states will generate demand signals in regions where no rep has touched it, and a lead-source field that records the region of the rep, or the state of the order, will misassign it. The lead-source schema for distributor channels separates the party that created demand from the party that booked the order for exactly this reason, and the footprint ratio is a quick way to size how often those two will differ.

Rep capacity planning changes too. In a small state, a raw supplier count overstates the local market by the most, because national suppliers are a large share of it. Dividing reps by supplier count will over-staff small states and under-staff large ones, unless the count is first reduced to suppliers with meaningful local volume. That requires the supplier-level file rather than this one.

For marketing, the resupply finding is the one to act on. A manufacturer whose consumables move through suppliers averaging five or six states per supplier is marketing into a channel where regional campaigns and regional rep relationships map poorly onto how the product actually moves. National programmes aimed at the supplier accounts themselves fit that shape better than state-by-state field coverage.

None of this depends on a purchased list. The file is public, the request is one line, and the whole analysis for a product line runs in minutes. It sits alongside the clearance-register approach in where new device competitors are entering: public regulatory and claims data describe the structure of a market well, and say very little about demand. More on channel attribution sits in the attribution pillar.

Is a larger supplier count better for a manufacturer?

Not necessarily. A large count means many parties can fill the product, which widens distribution and fragments attribution at the same time. More suppliers means more partner relationships to manage and more places for demand to be recorded under someone else’s name. The count describes channel structure only.

What a manufacturer can reasonably infer is narrower. A product with a wide footprint and a large supplier base is one where the manufacturer’s own visibility into who created demand will be weakest, and where investment in demand capture before the order reaches a supplier returns the most. A product with a narrow footprint is one where relationships with local suppliers carry more of the weight, and a state-based field model is a closer fit.