How US Hospitals Buy New Medical Devices: Who Decides, Through Whom and on What Evidence

1. Reports

By Saroj Jha · 11 min read

What public data shows about who buys, through whom, under what financial pressure and on what evidence, and what it does not show, from GAO, AHA, CMS, MedPAC, KFF and peer-reviewed studies.

Selling a device to a US hospital is rarely one decision. This report sets out what public data shows about who buys, through whom, under what financial pressure and on what evidence, from the Government Accountability Office (GAO), the American Hospital Association (AHA), the Centers for Medicare & Medicaid Services (CMS), the Medicare Payment Advisory Commission (MedPAC), KFF and peer-reviewed studies. It also lists what the public data does not show, because for a founder planning a sales cycle the gaps matter as much as the figures.

Three purchasing groups serve about 84% of US hospitals, but most of the rise since 2014 is one merger, and about 70% of community hospitals sit inside a health system. Hospital budgets are tight, Medicare pays a fixed rate per stay, and the best public evidence says buyers want credible clinical evidence and a case made in their own budget. No public source shows how long a purchase takes.

Who has to say yes

In AAJ’s working model, a new device needs three yeses: a clinician who asks for it, a committee that approves it, and a contract that lets the hospital buy it. The public data is strongest on the third, so most of this report is about contracts, money and evidence rather than about committees.

Purchasing groups reach nearly every hospital

A group purchasing organization (GPO) negotiates contracts with vendors on behalf of many hospitals. GAO’s 2014 review quotes a trade association, the Healthcare Supply Chain Association, estimating that “approximately 96 percent to 98 percent” of US hospitals buy through GPO contracts^[1]. That is an industry estimate that GAO reports rather than measures.

The most recent measurement of who belongs to which group is a 2026 research letter in JAMA Health Forum. Using hospital data from the American Hospital Association survey and the IQVIA database, it found that Vizient, HealthTrust and Premier together served 84.1% of 4,760 hospitals in 2024, up from 61.7% of 5,008 hospitals in 2014. Another 5.6% of hospitals had no GPO in either year^[2]. Two details change how the headline reads:

• Most of the rise is a merger. The letter notes that MedAssets, which held 20.3% of hospitals in 2014, was acquired by Vizient in 2016. Counting MedAssets as a large GPO in 2014, the large groups already reached about 82% of hospitals that year (AAJ’s sum of the letter’s counts, 4,113 of 5,008). Vizient’s own share went from 23.0% to 43.4%, almost exactly what Vizient and MedAssets held together in 2014 (43.3%).
• The shares describe each hospital’s primary pharmaceutical GPO. Because hospitals can join several groups, the authors used the primary pharmaceutical affiliation. They report that pharmaceutical and medical-surgical affiliations agree in 94% of cases, which is reassuring but means a device company should still ask a hospital which group covers its category.

Contract terms matter as much as membership, and the public numbers are more than a decade old. GAO surveyed the five largest GPOs about 2012; GAO’s earlier report covers 2008.

Vendors pay the fees, which is why a GPO contract is a commercial negotiation for a device company: the fee is usually a percentage of what hospitals buy. GAO found that administrative fees made up about 92% of the five GPOs’ revenue in 2012, and that nearly 70% of the roughly $2.3 billion in fees was passed on to customers or owners^[1]. The federal Anti-Kickback safe harbor lets GPOs collect these fees if, among other conditions, the written agreement with customers states that fees are 3% or less of the purchase price or specifies the amount^[1]. The contract figures are old; AAJ found no newer public source for today’s mix of terms and fees.

Most hospitals sit inside a system

The AHA counts 3,567 of 5,121 US community hospitals (70%) as part of a health system^[4]. KFF reports the share of hospitals affiliated with a system rising from 58% in 2010 to 69% in 2023^[5]. For a device company, a win at one hospital is a reference inside its system, and a loss at the system level can close several hospitals at once.

The money: margins and payment

Hospitals are buying under financial pressure, though the picture depends on the measure and on who owns the hospital.

MedPAC’s January 2026 slides give a preliminary fiscal 2024 all-payer operating margin of 6.5%, excluding investment and donation income, and a fee-for-service Medicare margin of −12.1%^[6]. KFF finds that for-profit hospitals earned much higher operating margins in 2023 (14.0%) than nonprofit (4.4%) and government hospitals (3.4%)^[5]. The most recent flash report from Kaufman Hall says roughly 43% of hospitals are operating at a loss, compared with roughly 33% before the pandemic^[7]. That report appears on the website of Vizient, which is also the largest purchasing group, so read it with that in mind.

Medicare’s payment system adds a second constraint. For inpatient stays, CMS pays “at predetermined, specific rates for each hospital discharge,” so a new device’s price comes out of a payment that does not change with it. A new-technology add-on payment can cover part of the extra cost, no more than 65% of the technology’s cost (or of the amount by which the case cost exceeds the standard payment, whichever is lower), and only for technology that shows “substantial clinical improvement” over what exists^[8].

Coverage is slower still. A 2023 study of 281 devices and diagnostics authorized through the premarket approval, De Novo or breakthrough 510(k) routes from 2016 to 2019 found that 64 (23%) were novel, meaning they lacked coverage decisions or relied on temporary billing codes. Only 28 of the 64 (44%) reached explicit or implicit Medicare coverage, and the median time from FDA authorization to at least nominal coverage was 5.7 years^[9].

The evidence hospital buyers weigh

AAJ found two directly relevant studies, and both have limits.

Read together, they support leading with credible clinical evidence. The review also points to earlier work finding that budget impact, the cost to the hospital’s own budget at its own volume, matters more to investment decisions than a cost-effectiveness ratio, although the executives in the survey ranked cost effectiveness among the most important factors. Neither study shows how much weight any one factor carries at a particular hospital.

What the public data does not show

AAJ looked for each of the following and found no source with a stated method and a representative sample:

• How long a value analysis committee takes from request to decision, or the full purchasing cycle for a new device.
• What share of hospitals run a formal value analysis committee today, and who sits on it.
• Which kinds of evidence committees require, such as randomized trials or economic models.
• The share of hospital spending that goes through GPO contracts today, and the current mix of contract terms.
• When hospitals’ fiscal years end, in numbers, and when budgets are set.

Two figures that circulate widely are left out. One is the claim that 73% of hospitals’ non-labor purchases go through GPOs, which GAO quotes from a 2009 study by an industry consultancy rather than finding itself. The other is value analysis survey results from a supply-chain software vendor and a professional association, whose percentages do not fit the stated sample sizes. The practical consequence: ask each target hospital for its committee’s form, meeting dates and submission deadline, and record your own cycle times.

What this means for a device startup

1. Find out the contract route first. With about 84% of hospitals served by three groups, ask early customers which group and contract they buy through. If your category is already on contract with a competitor, plan for that.
2. Sell to the system, not just the hospital. Seven in ten community hospitals sit in a system; aim the first site’s result at the system’s committee.
3. Make the case in this year’s budget. Lead with the evidence, then show cost per patient and budget impact at the hospital’s own volume; a long-term cost-effectiveness ratio can follow but should not carry the case alone.
4. Do not count on new payment. Build the case on the hospital’s existing payment unless your evidence can meet the add-on payment’s “substantial clinical improvement” test.
5. Measure your own cycle. Since no public data shows how long committees take, record request, submission, decision, contract and first order dates for every hospital.

Method and limits

AAJ used primary sources: government reports (GAO, CMS, MedPAC), the AHA’s statistics, KFF, the Kaufman Hall flash report and peer-reviewed studies. Each source was opened on October 22, 2026, and every figure was checked against it. The limits are real. The group shares come from a one-page research letter and describe pharmaceutical affiliations. GAO’s contract data is from 2012 and 2008. The MedPAC figures are from preliminary slides and may change. The executive survey had 34 respondents and covered orthopedic implants, and the systematic review covers studies from 2000 to 2012. The AHA is an advocacy organization, though its hospital counts are standard reference data. This report is general information, not legal or financial advice.