Payment integrity in the age of AI and value-based care

| Article

The payment integrity (PI) industry plays a critical role in US healthcare, ensuring that claims adjudication is accurate and that physicians and health systems are reimbursed correctly for care delivery. We estimate the $9 billion industry has grown at a 7 percent CAGR in recent years, reflecting growth in US healthcare spending and persistent complexity in billing processes.

This growth has coincided with technological change, such as the rise of generative AI (gen AI), and payment model innovation, including the transition of about 60 percent of total care delivery reimbursement to value-based care (VBC) payments1—shifts that have the potential to transform the PI industry. This article discusses these trends and the implications for payers, care delivery organizations, PI companies, and investors.

What is payment integrity?

The PI value chain includes a series of capabilities that span the healthcare financial-transactions ecosystem (Exhibit 1).2Administrative simplification: How to save a quarter-trillion dollars in US healthcare,” McKinsey, October 20, 2021. These capabilities are typically grouped into two main categories: prepayment capabilities, which seek to proactively identify and correct billing or adjudication errors before a health plan disburses payment to a care delivery organization, and postpayment capabilities, which identify and resolve historical payment errors after disbursement of initial payment to the care delivery organization.

The payment integrity value chain is typically grouped into prepayment and postpayment capabilities.

This value chain includes a mix of software and analytics solutions, which typically identify and resolve relatively simple errors, and teams of coding experts and clinicians, who review claims and medical records in more complex cases. Services are provided through a combination of internal health-plan functions and an ecosystem of specialist PI companies that augment payers’ internal capabilities.

The goal of PI programs is to ensure accurate claims adjudication and payment, taking into account factors such as payment responsibility, including identifying potential errors related to eligibility, benefits coordination, and similar considerations; billing accuracy, including identifying potential coding errors and mistakes; and cases of fraud, waste, and abuse. They also look to ensure that claims are paid in accordance with the contract between a health plan and care delivery organization—that is, that each service is paid at the contractually indicated price.

Payment integrity is critical given the complexity of the US healthcare payments workflow, which results in a natural rate of error even in cases where all stakeholders strive to submit and pay claims accurately and in good faith (see sidebar, “Complexities of US healthcare payments”). Payers typically strive to identify cases of overpayment, and care delivery organizations (via a revenue cycle function typically called revenue integrity) often seek to identify cases of underpayment. This system of checks and balances helps ensure accurate payments in US healthcare.

Trends shaping the future of payment integrity

The PI industry is poised for transformation in the coming years. Contributing factors include growth in US healthcare spending, which we estimate will grow roughly 2.5 percent faster than GDP from 2022 to 2027,3The gathering storm in US healthcare: How leaders can respond and thrive,” McKinsey, September 8, 2022. increasing the addressable market for PI companies and attracting continued investment.

Persistent complexity in billing and claims processing is another factor. For example, the COVID-19 pandemic accelerated a shift in care to alternate settings, such as home and virtual care.4From facility to home: How healthcare could shift by 2025,” McKinsey, February 1, 2022; Oleg Bestsennyy, Greg Gilbert, Alex Harris, and Jennifer Rost, “Telehealth: A quarter-trillion-dollar post-COVID-19 reality?,” McKinsey, July 9, 2021. The subsequent emergence of new care delivery models (for example, hospital at home) and increase in the use of virtual-care have made it necessary for payers, health systems, and policy makers to design new billing and reimbursement guidelines, which have created new sources of potential claims errors.5

Market dynamics in the PI services ecosystem are also shaping the future of PI: consolidation has resulted in the formation of organizations with the scale and specialization to harness new tools, such as gen AI. And more recently, a new wave of PI start-ups has been launched, with the incentive to disrupt the industry by launching innovative, AI-powered solutions.

But two trends stand out as having particularly complex repercussions for the PI industry. Technological innovation, such as gen AI, could result in the launch of PI solutions that are more accurate and efficient than ever before. And increasing adoption of VBC creates new sources of reimbursement complexity, both for payers and care delivery organizations. Below, we explore the relevance of these trends for PI in further detail.

AI and innovation in payment integrity

Analytical AI, machine learning (ML), and gen AI have the potential to transform many industries, including healthcare.6Tackling healthcare’s biggest burdens with generative AI,” McKinsey, July 10, 2023. The effects could be particularly profound in payment integrity because PI capabilities depend on the rapid review and synthesis of a variety of data sources. Furthermore, even relatively small increases in accuracy and efficiency of claims adjudication can represent substantial financial impact for a range of healthcare stakeholders because of the complexity and scale of the US healthcare payments ecosystem.7

As Exhibit 2 shows, analytical and gen AI could augment existing capabilities across the PI value chain. For example, during the review or audit of complex claims, gen AI could review and synthesize a complex combination of structured and unstructured data—including encounter data, medical record details, and health plan reimbursement policies—and support the human reviewer in making decisions, improving the accuracy and efficiency of the review process.

AI can be applied across the payment integrity value chain.

As we have previously described,8Setting the revenue cycle up for success in automation and AI,” McKinsey, July 25, 2023. it will take time for healthcare organizations to fully understand and harness the potential of new AI and ML capabilities. But a few potential effects of the PI technology seem probable. First, AI and ML could substantially increase claims accuracy and administrative efficiency of PI programs, creating potential shifts in value pools. For example, predictive analytics could further shift value from postpay PI programs to prepay claims edits. In addition, early adopters may have an opportunity to create a strategic distance between themselves and competitors and meaningfully improve PI programs’ ROI (for payers) and internal profitability (for PI services companies). Finally, PI services companies could transform portions of the PI value chain, launching innovative new solutions made possible by AI and ML—such as accelerated, lower-cost, and more-accurate reviews or audits for complex claims.

VBC and the future of payment integrity

Similarly, growth in VBC adoption creates new incentives for innovation in PI. Broadly defined, the shift to VBC entails tying care delivery organization reimbursement to health outcomes or value for patients.9 Adoption of VBC has increased markedly in recent years: the Health Care Payment Learning & Action Network (HCPLAN) estimates that VBC adoption nearly doubled from 2015 to 2021.10 As of 2021,11 almost 60 percent of healthcare reimbursement has some tie to quality or value, ranging from pay-for-performance models to fully capitated reimbursement models (Exhibit 3).

About 60 percent of healthcare reimbursement is tied to quality or value.

This shift to VBC has the potential to improve care coordination and quality for patients.12 It also creates complexity that can result in payment errors.

For example, traditional sources of payment complexity and error associated with fee-for-service (FFS) billing persist in VBC scenarios. Indeed, these challenges exist in about 97 percent of US healthcare spending. Two factors contribute to this. First, in HCPLAN models 1 to 3, which account for about 93 percent of all US healthcare payments and include both FFS and VBC, care delivery organizations and payers submit and adjudicate claims throughout the year. They then must conduct a financial reconciliation process at contractually indicated periods (annually, semiannually, or quarterly).13 Thus, all challenges with traditional claims are still relevant with these models. Second, population-based payment models (that is, HCPLAN model 4B or capitation model) require continued FFS billing and payment for all care delivered by physicians and health systems other than the risk-bearing care delivery organization. In these scenarios, claims adjudication is performed either by the health plan or risk-bearing care delivery organization and can vary by contract, resulting in the potential for human error in IT system configurations for both care delivery organizations and payers.

VBC models also create new sources of complexity and error, including the following:

The need to “attribute” specific patients to a risk-bearing care delivery organization. This is often done algorithmically—for example, in cases when a patient does not formally select a primary-care physician (PCP). Algorithmic attribution methodologies can vary across organizations, and results often lag behind care delivery touchpoints. For example, a common attribution method assigns financial responsibility for a patient’s care to the PCP the patient saw most frequently in the prior plan year, creating inherent time gaps between patient–physician interactions and corresponding reporting. As a result, risk-bearing care delivery organizations do not always have transparency into the patients for which they are financially accountable until several months into a performance period.

Quality and financial-performance tracking. Such tracking is a core component of VBC models, is complex, and varies by contract. As a result, when coordinating care for a given patient, risk-bearing physicians sometimes do not know the metrics for which they will be held accountable.

Contract configuration. Configurations can vary by contract and model. For example, of the roughly 150,000 diagnosis and procedure codes involved in billing, the exact codes covered by a capitated arrangement vary by payer and care delivery organization. As a result, human error in system configuration can lead payers to incorrectly pay (or not pay) for a claim that was (or was not) covered in a capitated contract.

Reconciliation and settlement. This often involves complex and custom financial analysis of terms that vary by VBC contract. For example, to reconcile performance at the end of a VBC contract period, payers and care delivery organizations must agree on the patients attributed to the care delivery organization; performance against a variety of quality and financial metrics, which vary by contract; and settlement of any outstanding issues related to claims adjudication throughout the contract period. While VBC programs are meant to align the incentives of payers and care delivery organizations and foster increased partnership in the coordination of patient care, the complexity of this reconciliation process can cause unintended, negative impacts on relationships between payers and risk-bearing care delivery organizations.

As a result of these existing and new complexities, the shift to VBC may increase the importance of payment integrity functions for payers and care delivery organizations. While PI organizations don’t necessarily have to help payers and care delivery organizations navigate the complexity described above, there may be an opportunity for organizations that choose to. So far, a range of services companies—including VBC analytics and software solution providers, actuarial consulting firms, and traditional PI and revenue cycle firms—have supported health plans and care delivery organizations during the transition to VBC. However, complexity and pain points persist, creating opportunities for firms that innovate in response.

What it means for payment integrity stakeholders

In response to these trends, stakeholders across the value chain could take steps to make US healthcare reimbursement more efficient and accurate.

For payers and care delivery organizations, being first out of the gate to adopt AI-powered solutions could have a material impact on future performance because of the potential to increase reimbursement accuracy and efficiency relative to competitors. Likewise, organizations that adopt or build new PI solutions for VBC first could have a material advantage as the shift to VBC continues. Payers and care delivery organizations could act today to build the operating model, talent, technology ecosystem, and data infrastructure needed to use AI-powered solutions as they become commercially available.

For PI services companies, being first to market with new solutions could create a distinct opportunity to outpace competitors. Similarly, those that fall behind could risk erosion of their value proposition and market position. Services companies could invest now in solutions powered by AI or tailored to address emerging pain points in VBC models to seize this opportunity.

For investors, these trends bring the potential to create value by identifying and partnering with emerging providers of AI-powered or VBC-tailored PI solutions, helping to accelerate and scale solutions with the potential to increase accuracy and efficiency across the US healthcare payments workflow.

Change is coming for the PI industry. By understanding the ways AI and VBC could reshape the industry, stakeholders could not only brace for impact but also harness the potential, ushering in the next era of payment integrity.

Explore a career with us