Virtual second opinions sound like a no-brainer. They help employees avoid unnecessary surgeries, catch misdiagnoses, and feel more confident about their care. Cleveland Clinic’s own program data shows 67% of its virtual second opinions recommend a change in diagnosis or treatment plan. Employers love the marketing. Brokers love the savings narrative. But after spending years inside benefits administration systems, I’ve noticed something that rarely gets discussed: these programs are quietly messing up your claims data.
It’s not intentional. But the way most VSOs are implemented creates what I call a ghost claim: a treatment that was avoided but never recorded in the systems that actually predict risk and cost. And that ghost comes back to haunt your underwriting, your stop-loss coverage, and even your pharmacy rebates.
The Data Void You Didn’t Know Existed
Your TPA’s analytics engine lives and breathes paid claims. It looks at diagnosis codes, procedure codes, and provider patterns to forecast what your population will need next year. When an employee with chronic back pain gets a VSO that says “skip the surgery and try physical therapy,” the system doesn’t learn that the surgery was avoided. It still sees that employee as high risk. The only cost that shows up in the medical ledger is the physical therapy, if they even go.
The VSO fee itself gets logged somewhere else entirely, often in a wellness or advocacy budget line. It’s completely invisible to the claims engine. Your risk models keep expecting a surgery that will never happen. You’re paying for a decision that improves care, but your data says nothing changed.
Three Ways This Breaks Down
1. The Non-Event Blind Spot
Stop-loss carriers underwrite from claims history, and they laser known high-cost members by assigning them a higher deductible based on diagnosis. A member with chronic back pain stays on the underwriter’s radar whether or not the surgery happens. But when a second opinion cancels the surgery, no record of that decision reaches the underwriter. The carrier prices next year’s policy as if the surgery could still occur, and you pay for risk the VSO already removed.
2. Conflicting Diagnoses Create Chaos
VSO platforms usually operate completely outside your primary care network’s electronic health record. When the VSO specialist says “this is a rare autoimmune condition,” but your local doctor says “it’s just stress,” the two systems never talk. Your population health vendor runs a gap analysis and sees a patient flagged for anxiety who isn’t filling their prescriptions. They start sending reminders for the wrong thing. Meanwhile, the correct diagnosis is sitting in a PDF, locked inside the VSO vendor’s portal, invisible to your care managers.
3. The Pharmacy Rebate Trap
This one is almost never caught. Suppose the VSO recommends switching from a branded biologic to a generic. The employee follows the advice. You skip the high-cost brand claim. Big savings, you think. But your PBM rebate contract is based on volume of brand-name drugs. By reducing that volume, you may fall short of your rebate guarantee. The medical savings evaporate when the PBM sends you a smaller rebate check. Because the TPA handles medical claims and the PBM handles pharmacy, no single system connects the dots.
What the Data Gap Costs Patients
Underwriting and rebate leakage are only part of the cost. When a second opinion flags a rare autoimmune condition and the local record still says anxiety, the care manager chases the wrong diagnosis. Reminders, screenings, and follow-ups target the condition the system thinks the patient has, while the real one goes unmanaged.
A 2023 analysis from Johns Hopkins and Harvard’s CRICO group, published in BMJ Quality & Safety, estimates that 795,000 Americans die or are permanently disabled by diagnostic error each year. A second opinion that catches a serious condition but can’t push it back into the care team’s workflow becomes a diagnostic error with extra steps.
A corrected diagnosis that reaches the claims engine and the care manager’s screen at the same time protects the patient and the employer’s stop-loss renewal.
How to Actually Fix This
Stop treating VSOs as a standalone vendor service. You need to integrate them into your data ecosystem. Ask your vendor and your TPA to do three things:
- Record the averted procedure with standard codes. When a VSO recommends against a specific procedure, the clinician can document it as a procedure not carried out because of the patient’s decision, using the ICD-10 Z53.2 code series. That places the decision in the medical record, where underwriting and risk models can see it. Don’t invent a private code for the non-event. A made-up code is its own data-integrity problem, and underwriters discount anything they can’t tie to a standard.
- Bridge the clinical records. Require the VSO vendor to provide a structured data feed, like a FHIR API, that pushes the second opinion diagnosis back into your population health data warehouse. Your care managers need to see both opinions so they can stop chasing the wrong condition.
- Run a rebate impact analysis. Before launching a VSO program, model what happens if 20% of patients shift from high-rebate brands to low-rebate generics. You might discover that the program increases net spend under a heavy rebate-driven PBM contract.
The Bottom Line
Virtual second opinions are a powerful tool. But right now, most employers are deploying them as a Band-Aid on a broken data pipeline. You’re paying for better clinical decisions, but those decisions aren’t reaching the systems that determine your future costs.
Treat VSOs as a master data management challenge. If the decision doesn’t enter your claims engine, your risk models, and your pharmacy rebate calculations, you haven’t saved money. You’ve just made your data less accurate.
Ask your vendor for their data integration plan. If they don’t have one, they’re selling a great clinical experience but a poor system outcome. And that’s a cost you’ll keep paying, quietly, every time you renew your stop-loss or audit your PBM rebates.