What Does Payment Integrity Mean on a Self-Funded Plan?

What does payment integrity cover on a self-funded plan?

Payment integrity covers every check between a claim arriving and the plan's money leaving, alongside every review that looks back at money already gone. On a self-funded plan, the distinction from fully insured coverage matters immensely. Each overpayment comes directly out of the employer's account, and each recovery goes back into it.

The sponsor, acting as the plan's fiduciary under the Employee Retirement Income Security Act (ERISA), is ultimately accountable for both (Source: Department of Labor, 2026). A TPA's payment integrity program is the set of these checks, the order they run in, and the specific record of who acted on each result.

The checks generally group into four distinct families:

• Administrative edits: Testing a claim against eligibility, benefits, pricing, and coding rules.
• Payer sequencing: Duplicate claim detection and coordination of benefits checks to ensure the right payer makes the first payment.
• Pattern detection: Fraud, waste and abuse screening that identifies issues no single claim reveals.
• Clinical review: Assessing whether the billed service was medically necessary.

Key Takeaways:

• Payment integrity protects the sponsor's direct funds.
• Fiduciary responsibility under ERISA falls on the plan sponsor.
• Checks span administrative, sequential, pattern, and clinical reviews.

What is the difference between prepayment claim review and a postpayment audit?

Prepayment claim review stops a claim before it pays to verify accuracy, while a postpayment audit examines a claim after funds are disbursed to recover errors. Prepayment review is significantly cheaper per error. An unpaid claim is corrected via a simple adjustment, whereas a paid claim requires a recoupment letter, provider disputes, and potential offsets against future payments.

However, postpayment audits remain necessary because some errors only appear in aggregate over time. Examples include a provider billing an impossible number of daily hours, unbundled procedures across a quarter, or other coverage discovered after the fact. A highly effective payment integrity program balances both methods carefully.

Key Takeaways:

• Prepayment review prevents errors and is cheaper to administer.
• Postpayment audits catch aggregate fraud and retroactive changes.
• Moving checks to prepayment delays claims; moving them to postpayment increases recovery costs.

What do claim edits check on a self-funded plan?

Claim edits on a self-funded plan automatically verify variables against the data the TPA already holds. These edits check if the member was eligible on the date of service, if the provider is in-network, and if the service is a covered benefit within the plan's limits. They also verify if a required authorization is on file and if the billed amount matches the contracted rate.

For coding rules, TPAs typically rely on standardized, published sets. For example, the National Correct Coding Initiative (NCCI) publishes procedure-to-procedure edits to prevent improper code pairings and limits units per service (Source: Centers for Medicare & Medicaid Services, 2026). Plans adopt these edits by choice through their specific plan document. Importantly, an automated edit that fires is not a final determination; it is simply a mandated reason for a human to look closer.

Key Takeaways:

• Edits verify eligibility, authorizations, and pricing against TPA records.
• TPAs utilize standard sets like CMS NCCI edits.
• An edit flag is a trigger for human review, not a final denial.

How do duplicate claim detection and coordination of benefits checks work?

Duplicate claim detection compares an incoming claim with historical files for the same member, provider, date, and service. The check must distinguish a true resubmission from a corrected claim or a second genuine service delivered on the same day. Claim frequency codes and line-item details usually settle exact matches, but near-matches require human intervention.

A coordination of benefits check determines if another health plan should pay first. This starts with the member's other-coverage record and the plan's order-of-benefit rules. For example, Medicare Secondary Payer rules can make a group plan primary for an active employee, forcing the plan to apply specific rules rather than assuming Medicare pays first (Source: Centers for Medicare & Medicaid Services, 2026). Deciding that another plan is primary and reducing payment accordingly constitutes a formal benefit determination.

Key Takeaways:

• Duplicates require distinguishing between corrections and distinct same-day services.
• Coordination of benefits relies on accurate, up-to-date member coverage files.
• Determining primary vs. secondary payer status is a formal benefit determination.

What does fraud, waste and abuse screening look for?

Fraud, waste and abuse screening looks for billing anomalies that a standard examiner cannot spot on a single, isolated claim. Prepayment screening flags immediate outliers, such as a modifier applied to every single line item or a member receiving services at two distant geographic locations on the same day.

Postpayment data mining finds broader patterns across time. Definitions matter here (Source: Centers for Medicare & Medicaid Services, 2026):

• Fraud: Deliberate misrepresentation to obtain improper payment.
• Waste: Overuse or misuse of services without malicious intent.
• Abuse: Actions inconsistent with sound medical or business practices.

The screening output is a referral to a special investigations unit or the plan sponsor. The TPA's specific authority to hold claims, request records, or report findings is strictly defined by its administrative services agreement.

Key Takeaways:

• Screening identifies patterns invisible on a single claim.
• Fraud requires intent; waste and abuse do not.
• Flags result in referrals for investigation, governed by the TPA agreement.

Who handles claim hold governance on a self-funded plan?

A specifically named person must decide when a claim is held, following written criteria approved by the plan sponsor. The plan document grants benefit determination discretion to the plan administrator (the sponsor), who delegates daily execution to the TPA's examiners. A hold delays payment and starts the compliance clock under the ERISA claims procedure regulation (Source: Department of Labor, 2026).

Clear claim hold governance must answer four critical questions: what triggers a hold, who places it, who releases it, and how long it can stand before escalation. This timeline is vital because a large claim held past the stop-loss contract's paid-claims window can fall entirely outside reimbursement eligibility, leaving the sponsor to absorb the total loss.

Key Takeaways:

• Claim holds must be executed by named, authorized personnel.
• Holds trigger ERISA compliance timelines for claims decisions.
• Poor governance can push claims outside of stop-loss reimbursement windows.

Why must software not make final claim decisions?

Software must not make final decisions because a flag is a probability and a denial is a determination with rights attached. Payment integrity often tempts TPAs with complete automation, promising found savings and high processing speeds. However, auto-denying claims based purely on a software score threshold creates compliance risks, angers providers, and generates massive appeal queues.

Structurally, ERISA requires that a formal determination be made by a person and accompanied by a specific, plan-based reason. An auditor cannot accept "a score exceeded a threshold" as a valid denial reason. Furthermore, third-party postpayment vendors paid via shared savings have financial incentives to find errors. Because of this conflict, the decision to send a recoupment letter is the plan's decision, not the software's or the vendor's.

Key Takeaways:

• Software provides probabilities; humans make determinations.
• Automated denials violate ERISA requirements for specific reasoning.
• Financial incentives for recovery vendors necessitate independent human oversight.

How can agents support payment integrity without making decisions?

AI agents can securely handle the vast administrative burden of payment integrity, provided they do not cross into adjudication. To accelerate workflows, agents run every check and assemble the evidence and then stop. For a duplicate flag, agents place claims side-by-side; for an itemized bill, they reconcile individual charges to claim lines; for a recovery, they draft the recoupment letter for a human examiner to review, sign, or discard.

This exact methodology is how Daybook Health integrates with a TPA's core system. Daybook agents perform the high-volume assembly and leave a dated audit trail, ensuring 100% of denial and medical-necessity decisions remain with the TPA's designated staff. Applying this agentic workflow to checks like duplicates is a direct extension of standard pend management. All work is handled domestically; absolutely nothing is offshored.

Key Takeaways:

• Agents assemble evidence, reconcile data, and draft documents.
• 100% of final determinations must remain with human TPA examiners.
• Every agent touchpoint requires a transparent, dated audit trail.

What should a plan sponsor ask its TPA about overpayment recovery?

A plan sponsor evaluating an overpayment recovery program on a self-funded plan should ask its TPA four direct questions. First, which checks run prepayment versus postpayment? Second, which specific coding edit sets apply, and who selected them? Third, who has the authority to place and release a claim hold? Finally, does any check result in a denial or recovery without a named human examiner's signature?

A TPA that can answer all four clearly from its internal records has a robust, defensible payment integrity program. A TPA that answers the fourth question with a software vendor's name has improperly delegated the sponsor's fiduciary determinations.

Key Takeaways:

• Demand a clear split of prepayment vs. postpayment checks.
• Ensure naming accountability for every claim hold and release.
• Verify that software vendors are not making final, automated determinations.