How to Set a Defensible Reference-Based Pricing Benchmark

If your reference-based pricing (RBP) plan sets every hospital claim at a flat percentage of Medicare — 150%, 170%, whatever your TPA recommended — you're not really pricing anything. You're applying a national rule of thumb to a market where negotiated rates for the same procedure routinely differ by 3x from one hospital to the next.

It's also worth being honest about what "percentage of Medicare" even is: Medicare's rate isn't a market rate. It's an administratively set price, adjusted for wage index and a handful of other factors, that was never designed to answer the question an RBP plan actually needs answered — "what does a real payer actually pay this specific hospital for this specific procedure?" Using it as the anchor and then applying one multiplier to it nationally compounds an approximation on top of an approximation.

That gap is where three problems live: you can't tell if your TPA's "savings" number means anything, your plan has no documented basis for the number when a provider disputes a balance bill, and you have no way to know whether your flat percentage is too aggressive in some markets and leaving money on the table in others.

The proof: a flat benchmark can't survive contact with real rate data

Look at a single, common, well-defined procedure — CPT 27447, total knee arthroplasty — across hospitals that have published their actual negotiated rates under CMS's price transparency rule. Pulling that data directly (324 hospitals reporting this code under one consistent description, filtered for a sane price range):

• Median negotiated rate: ~$21,700
• Middle 50% of hospitals: roughly $12,200 to $33,700
• Some hospitals sit well outside even that band in either direction.

That's the interquartile range for the identical procedure code. A single flat percentage — applied nationally, or even statewide — is definitionally wrong for most of the hospitals your plan members actually use. It's either overpaying at the low end of the range or setting your plan up for provider disputes at the high end, and you have no data-backed way to know which is happening where.

This is a pattern we see over and over when a self-funded plan finally pulls its own market's data (illustrative and composite — not a specific engagement): the plan is running a flat percentage of Medicare across its whole network, and it turns out to be quietly wrong at every facility, just in different directions. The community hospital handling most of the plan's routine volume clears comfortably below that flat number — meaning the plan is overpaying on every claim there, month after month, with no one ever noticing because the aggregate "% of Medicare paid" still looks fine. Meanwhile the regional health system twenty minutes away, where the complex cases and the joint replacements go, sits well above it — which is exactly where the balance bills and the provider disputes concentrate. Averaged across the network, the number looks defensible. Facility by facility, it's wrong almost everywhere. That's a worse failure mode than being uniformly too aggressive or too generous, because it survives every high-level review and only shows up when someone finally looks at the per-hospital data.

The path: calibrate the benchmark to real local data, not a rule of thumb

1. Pull the actual negotiated-rate distribution for your plan's real utilization, not a national average. Your members don't get care at a random hospital nationally — they get care at a specific set of facilities in your specific service area. The benchmark that matters is the rate distribution there, for the procedures your plan actually pays for, sourced from those hospitals' own published machine-readable files. One thing that trips up anyone building this by hand: CMS's rule lets hospitals report a negotiated rate as a dollar amount, a percentage of charges, a per diem, or an algorithm — not always as a clean number you can drop straight into a benchmark. A meaningful share of MRF line items come through as "percentage of charge," which means you can't treat every reported rate the same way; you have to resolve it against that hospital's own chargemaster before it means anything. Skip that step and your "real data" benchmark quietly inherits the same fuzziness as the flat-percentage approach it's supposed to replace.

2. Set the benchmark as a percentile, not a flat multiplier. "120% of the local market's 50th percentile negotiated rate" is a defensible, data-backed number. "150% of Medicare, applied everywhere" is not — it can't explain why it's the right number for any specific hospital, because it was never calculated from that hospital's data in the first place.

3. Build the documentation trail before you need it, not after a dispute. When a provider disputes a balance bill, "our TPA told us to use 150%" is not a defensible position in any venue that matters — not to the provider's billing office, not to a state regulator, not to a judge if it gets that far. A benchmark calculated from that specific facility's own published rates, with the calculation shown, is. Build that documentation at benchmark-setting time — a repeatable report per facility, not a one-off spreadsheet built during a dispute.

4. Audit TPA-claimed savings against the same real data — don't take their comparison on faith. A TPA reporting "you saved 40%" is comparing your paid rate to something — usually billed charges, which are the least meaningful number in healthcare pricing (list prices routinely run 3-5x negotiated rates). That comparison flatters everyone involved: the TPA gets a big number to put in the renewal deck, and nobody has to look at whether the underlying benchmark was ever calibrated to a real market. The number that actually matters is your paid rate versus the real negotiated-rate distribution at that hospital. If your TPA's savings claim doesn't hold up against real transparency data, that's worth knowing before renewal, not after.

What this actually requires

None of this requires new data collection — every hospital covered by CMS's price transparency rule is already required to publish exactly this data. What it requires is turning thousands of individual hospital machine-readable files — with all their inconsistent rate types, naming conventions, and formatting — into a usable, per-facility benchmark, and refreshing it as those files update. That's the part that's genuinely hard to do by hand across a national network, and it's the specific gap between "we have a rule of thumb" and "we have a defensible, auditable benchmark."

Medlyze's Benefits Analytics is built around exactly this: percentile-based rate benchmarking calculated directly from CMS machine-readable files, with a documentation trail per facility, so your RBP benchmark is a specific number you can defend — not a rule of thumb you inherited from a vendor. It's also worth pairing with a direct look at payer-level price transparency compliance and Medlyze's rate lookup tool if you're auditing a specific facility or payer relationship rather than building a plan-wide benchmark.

If your current RBP program is running on a flat percentage today, the fastest way to find out how exposed that leaves you isn't another vendor deck — it's seeing the real distribution for the hospitals your plan actually uses.