What 18 Months of Claims Data Taught
One Manufacturer About Their Health Plan

This is a case study in what happens when a mid-market employer actually looks at their health plan data — and what they find. We've changed certain details to protect client confidentiality, but the financial outcomes, the analytical findings, and the interventions described are drawn from real client experience. The lessons apply broadly.

The starting point: a plan that seemed fine

The client was a Midwest manufacturer with 385 employees, fully insured with a major regional carrier for six years. Their renewal increases had been manageable — 6–9% annually, roughly in line with what their broker told them was "market." They had decent benefits by industry standards, competitive with peers in their labor market. They had no particular reason to think their plan was broken.

What they didn't have was data. Their annual benefits review consisted of a renewal presentation from their broker, a premium comparison between their current carrier and one or two alternatives, and a recommendation to stay put or switch carriers. No claims analysis. No cost driver identification. No pharmacy breakdown. No utilization benchmarking. Just premium versus premium.

When they engaged Benefits Collective, the first thing we did was request their claims data from the carrier. The carrier, per ERISA, was obligated to provide it. It took three requests and six weeks, but we got it.

What the data showed

Eighteen months of claims data, analyzed against industry benchmarks for manufacturers of similar size and demographics, produced several findings that surprised the client — and explained why their premium trajectory had been what it was.

Finding #1: Pharmacy was 34% of total claims — and almost entirely brand

Industry benchmark for pharmacy as a percentage of total claims: 26–29%. Their plan was running 34%, driven almost entirely by brand-name prescriptions on conditions where generic equivalents were available and clinically appropriate. Analysis of the formulary revealed that the carrier's PBM had a preferential rebate arrangement with several brand manufacturers, creating financial incentive to steer utilization toward branded products even when generics were less expensive.

Specific example: one drug class with 22 members on therapy, all on branded medications at an average of $380/fill. The generic equivalent: $18/fill. Annual savings opportunity from generic conversion in that single drug class alone: $96,000.

Finding #2: ER utilization was 40% above benchmark

The plan's emergency room utilization rate was 380 visits per 1,000 members annually, against a benchmark of 270 for a manufacturing workforce of similar demographics. When we mapped ER visits by member ZIP code and cross-referenced against primary care claim patterns, the picture became clear: 68% of ER visits were from members who had no primary care claim in the prior 12 months. The plan's primary care access was poor — high co-pays, limited network depth in the plant's geography, and no telemedicine option — and employees were using the ER as their default care setting.

Average ER visit cost on the plan: $2,100. Annualized excess ER cost above benchmark: $237,000. Most of these visits were for conditions amenable to primary care management: upper respiratory infections, musculoskeletal complaints, minor lacerations, urinary tract infections.

Finding #3: Three members accounted for 28% of total claims

Claims concentration is the variable that most significantly affects stop-loss structuring. In this case, three members — one with a chronic autoimmune condition requiring specialty biologic therapy, one who had experienced a cardiac event mid-year, and one with a complex pregnancy — had generated $1.1 million in claims against a $3.9 million total. This concentration was not unusual for a group this size, but it had implications for how the plan was structured and what stop-loss protection was appropriate.

Under the existing fully insured plan, the client had no visibility into this concentration. They had no way to know that three members represented 28% of the cost the carrier was covering on their behalf — cost that was directly driving their renewal pricing. Self-funding, with appropriate stop-loss structuring, would let them manage this exposure explicitly rather than absorbing it through opaque premium increases.

What changed — and what it produced

Based on the analysis, the client transitioned to a self-funded plan at their next renewal. Key structural changes: move to a transparent pass-through PBM with 100% rebate passthrough, generic step therapy protocol implemented for the identified drug classes, Direct Primary Care clinic contracted for the manufacturing site, and specific stop-loss structured at $125,000 to cap the exposure from high-cost claimants.

Eighteen months into the self-funded plan, the results were measurable. Pharmacy spend had declined 38% from the prior year — driven by generic conversion, biosimilar substitution on the autoimmune biologic, and elimination of PBM spread pricing. ER utilization had declined 29%, consistent with DPC adoption rates. Total plan cost per member per month was 11% below the equivalent fully insured premium.

The broader lesson

The findings in this case are not unusual. In our experience evaluating mid-market employer health plans, the pattern repeats with remarkable consistency: pharmacy economics significantly worse than benchmark, avoidable ER utilization driven by primary care access gaps, and claims concentration that explains renewal pricing but is never disclosed. What's unusual is that this employer actually looked.

The data is available. ERISA requires carriers to provide it. Analyzing it requires expertise and benchmarks, but it's not technically complicated. What it requires is an advisor who has an incentive to find the problems — because finding the problems is the first step toward fixing them, and fixing them is the entire point.

If you've been with the same carrier for more than two years and have never seen a detailed claims analysis, that absence of information is itself informative. The data exists. The question is whether anyone has looked at it on your behalf.