clinicians.build · interactive · september 2, 2026

The Integration Curve

Brendan Keeler on Scan.com's $220M: networks are the most durable business in healthcare, and the label does not come for free. So here is the arithmetic that decides whether one is buildable. Integrate the ten biggest billers in Medicare lab and you hold 22.3% of the volume. The ten biggest in imaging get you 1.08%.

Primary source: Health API Guy (Brendan Keeler), “Scanning for a Network,” September 1, 2026
Data via MIMI Labs: CMS Medicare Physician & Other Practitioners, by Provider and Service
Performance year 2024 · US providers · 174,484 imaging NPIs · 163,177 lab NPIs

Read “Scanning for a Network” → CMS source data → Or read today's newsletter →

Pharmacy got Surescripts. Imaging and labs did not. The usual explanation is regulatory — e-prescribing had a mandate and a DEA hook, planned care never did. That is true and it is not the whole answer.

The other half is a counting problem. A network is only buildable if a small number of endpoints holds a large share of the volume, because that is what lets you launch with coverage worth paying for. So: how many organizations do you actually have to wire up?

the endpoint budget — drag it

endpoints you can integrate 10

lab & pathology

—

of Medicare lab services

imaging

—

of Medicare imaging services

office visits (E&M)

—

the floor case: no network

lab ÷ imaging

—

coverage advantage

10 endpoints 100 1,000 10,000 what imaging needs for 80%

Two companies are a fifth of Medicare lab volume. The largest single imaging biller in the country is half of one percent.

The five largest lab billers in Medicare Part B are LabCorp and Quest entities — one NPI alone is 4.4% of all lab services. The five largest imaging billers are companies most clinicians have never heard of: Symphony Diagnostic Services, Mobile Images Acquisition, Berger and Burrow Enterprises. Mobile X-ray vendors, mostly. Number one is 0.51%.

who is actually at the top

rank	lab & pathology	share	imaging	share

Now every state, and the thing that looks like a finding

If concentration is what makes a network buildable, some state should be the easy beachhead. Each dot below is one state or territory. Horizontal is how many distinct NPIs bill Medicare imaging there (log scale, 1 to ~15,000). Vertical is what the ten biggest of them hold. A dot high on the left is a market you could corner with ten phone calls.

the critical lens — a volume floor

minimum imaging services in state no floor

states shown

—

max top-10 share

—

median top-10 share

—

across shown states

volume covered

—

80/20 lens

With no floor, the chart has states where the ten biggest imaging billers hold 100% of the volume. Those are the Army, Air Force and Pacific military post offices and the Northern Mariana Islands — each with ten or fewer billing NPIs. Just below them sit the Virgin Islands at 92% (20 NPIs) and Guam at 82% (37). Drag the floor to 70,000 services and every one of them is gone, Puerto Rico included. What remains is a wall: no state above 37% (Wyoming, 306 NPIs), and the five largest imaging states — California, Florida, Texas, New York, Illinois — all under 5.3%.

The one real outlier is Maryland at 23%, and it is one company: Symphony Diagnostic Services, the mobile x-ray firm that is also the largest single imaging biller in the country. A state-level “market” you could corner turns out to be a vendor with a van.

The builder read: if your imaging product depends on retrieval, there is no beachhead market where ten integrations gets you coverage. Price the fax, the phone call and the CD-ROM into the plan from day one — then ask honestly whether retrieval is the product rather than a cost you absorb.

What this can't tell you

An NPI is not an integration endpoint. This is the biggest caveat and it cuts both ways. Under Part B carrier claims, most imaging is billed by the individual radiologist who read it, not the facility that acquired it. Flip the toggle above to organizations only: just 2,289 organizational NPIs bill imaging, and they account for only 4.3% of the volume — which tells you the org-level view is not a cleaner census of imaging centers, it is a different, tiny slice. The real endpoint count sits somewhere between the two, and no public dataset publishes it.
Medicare Part B fee-for-service only. No Medicare Advantage, no commercial, no self-pay, no cash-pay imaging — and Scan.com's actual US wedge, per Keeler, is personal injury litigation, which is none of the above. Nothing here measures that market.
Hospital outpatient imaging is largely absent. These are physician/supplier claims. A large share of US imaging happens under hospital outpatient billing, which lands in a different CMS file. Read the imaging curve as the independent and professional slice, not the whole modality.
CMS suppresses cells under 11 beneficiaries. The smallest billers are already filtered out before you see them, so every count here is a floor, and the true tail is longer than the chart.
Concentration is necessary, not sufficient. Labs are concentrated and still never produced a Surescripts. Counting endpoints tells you whether a network is possible. It does not tell you whether anyone will build it, or who pays for the transaction when they do.