The GPO Question

In 1987, Congress rewrote the rules of hospital buying in a single stroke that hardly anyone outside the industry noticed. The Medicare and Medicaid Patient and Program Protection Act carved an exception into the federal anti-kickback statute: payments from a vendor to a group purchasing organization — a GPO — would be legal. The Department of Health and Human Services later wrote the safe harbor in regulation: as long as a GPO's fees were 3 percent or less of the purchase price, or were specified in a written agreement and disclosed annually to the hospitals involved, the money the manufacturers paid the negotiators was above board.

That decision built the modern hospital supply chain. Nearly forty years later, three such organizations — Vizient, Premier, and HealthTrust — serve as the primary GPO for 84.1 percent of American hospitals, according to a September 2026 JAMA Health Forum study, up from 61.7 percent a decade before. The question that has never been settled, and that the system's scale has made impossible to ignore, is the obvious one: does it actually work?

The case for the machine

Start with the strongest version of the defense, because it is not obviously wrong.

Buying in hospitals is ferociously complicated. A mid-sized health system manages tens of thousands of line items: sutures in twelve sizes, IV catheters in nine gauges, gloves in nitrile and latex, generics with rolling shortages, implants with surgeon-specific preferences, capital equipment with ten-year service contracts. The administrative burden of sourcing, vetting, contracting, and monitoring all of it is enormous, and the overwhelming majority of hospitals have neither the staff nor the volume to do it well.

Industry-commissioned research has repeatedly put the savings from group purchasing in the double digits as a share of product spend, and no serious critic disputes the underlying logic: a purchasing agent that aggregates thousands of buyers will extract better terms than any of them could alone. The GAO itself, in its 2014 review, noted that competition among GPOs can push them to negotiate the lowest possible prices, since hospitals are free to switch memberships.

For a rural or community hospital, this is the whole case. It cannot hire a procurement team. It cannot bench-test supplier claims. What the GPO sells it is not merely discounts — it is the administrative function of buying, outsourced to specialists and paid for by someone else.

The case against

The prosecution opens with arithmetic. The fee a GPO collects is a percentage of what its members buy. Higher prices mean larger fees. The negotiator's revenue rises with the number it is supposed to minimize. The GAO, in that same 2014 report (GAO-15-13), stated the conflict plainly: the vendor-fee funding structure creates incentives "for GPOs to negotiate higher prices for medical products in order to increase the amount of vendor fees that they receive." The office reviewed five large GPOs and found they had collected about $2.3 billion in administrative fees in 2012, passing nearly 70 percent back to members and owners and retaining the rest — a business, in other words, with billions of reasons to keep the prices it negotiates exactly where they are.

The contracting practices compound the concern. In 2003, GAO testimony before the Senate (GAO-03-998T) examined how GPOs awarded medical-surgical contracts and warned that sole-source awards, bundles of unrelated products, and contracts running five years or longer could, in the hands of a dominant GPO, lock out competing manufacturers and discourage new entrants. A small manufacturer with a better product but no GPO contract does not merely lose a bid; it can lose practical access to thousands of hospitals. In 2004, the Department of Justice issued subpoenas to more than a dozen device companies as it investigated whether the fee-and-rebate systems used by Novation, then one of the largest GPOs, were inflating the prices Medicare and Medicaid paid. A series of Senate Antitrust Subcommittee hearings beginning in 2002, under Senator Herb Kohl, pressed the same question for years.

The deepest problem is epistemological. GAO-15-13 noted there was little empirical evidence available to either support or refute the competing claims — because the prices are secret, the savings are computed against secret baselines, and no independent party can audit either. The industry's double-digit savings figures are real numbers derived from real comparisons, but they are the industry's numbers, using baselines the industry chose. Nobody can prove the machine costs money. Nobody can prove it saves it, either.

A fair verdict

Both stories are true at once, which is why the debate has run for four decades without resolution.

The GPO model plainly delivers administrative efficiency and real bargaining leverage to hospitals that have no alternative source of either. That is not a small thing. For the majority of American hospitals — community, rural, independent — the GPO is the procurement department. The savings claims may be unverifiable in detail, but the mechanism is sound and the alternative, unaggregated buying by thousands of under-resourced institutions against concentrated global suppliers, is not a serious proposal.

But the model is also, structurally, a tax on opacity that grows with the prices it is meant to restrain, governed by a safe harbor written in an era of fragmentation and never stress-tested for an era of three. The 3 percent rule assumed GPOs would be one option among many. The GAO's call for better reporting of the fee revenues that flow back to hospitals went largely unheeded — HHS had not reviewed hospital reporting of GPO revenue since 2005, by the office's own account. The conflict the GAO identified is real even if its effects are unmeasured, and it is most acute precisely where competition is thinnest: physician-preference items, where a single manufacturer dominates a category and the GPO's choice of a sole-source winner can determine market structure for years.

The fair verdict, then, is conditional. GPOs save hospitals money on the categories where the model is soundest: high-volume, standardized commodities bought by institutions that could never negotiate alone. They have never convincingly shown they save money on the categories that matter most — the high-value physician-preference items where savings would be largest and where the fee model's incentives cut against them. And the savings they do produce accrue to hospitals in the dark: no buyer can verify its prices against the market, because the market is invisible by design.

That invisibility is the real answer to the GPO question. The model works well enough that the industry built itself around it, and opaquely enough that no one can say how well it works. The next chapter of American hospital purchasing will not be written by whoever negotiates the best contracts. It will be written by whoever turns on the lights.