How Artificial Intelligence Can Transform Payment Integrity
By Krithika Srivats, Vice President, Sagility Practice Office
This article was published in Becker’s Hospital Review.
In 2013, the actress Angelina Jolie published an opinion column in the New York Times that described her decision to undergo a preventive mastectomy after testing positive for the BRCA gene, a mutation that can lead to breast cancer.
Within two weeks of the column’s publication, more than 4,500 BRCA tests were ordered across the nation, a 64 percent rise from the same time period the previous year. Interestingly, researchers found no increase in the number of breast cancer diagnoses, and mastectomy rates actually declined. This suggested that many of the tests were given to populations at low-risk of carrying the mutation. In the end, a few hundred words written by a celebrity cost the healthcare system nearly $14 million for tests that were largely unnecessary.
Though this was a clear example of overutilization, the payment integrity programs of most health plans mounted responses that were either ineffective or non-existent. A closer look at most payment integrity programs reveal why they were ill-suited to the task.
While specific components of a payment integrity program can vary among health plans, most include a combination of human-based review and rules-based automation software to identify anomalies among thousands of claims. However, manually-driven payment integrity programs—even those augmented with automation and analytics software—identify less than half of all claims anomalies, which may or may not prove to be improper claims after further investigation. In addition, this resource-intensive and costly process was largely built on a pay-and-chase model, whereby outliers are targeted and recovery of previous paid claims are initiated.
A 2018 Gartner report, “U.S. Healthcare Payer CIOs Must Adopt Prospective Payment Integrity to Thwart Improper Claims Payment and Fraud,” concluded that health plan executives can significantly improve the financial returns from payment integrity programs” through prospective processes and “exponentially increase financial performance and improve provider relationships.”
Fortunately, technological advances have given health plans an opportunity to build a better payment integrity system—one that doesn’t just anticipate emerging trends from, say, a high-profile endorsement of an expensive medical test, but most other forms of overutilization.
The solution? Artificial intelligence (AI).
AI has been successfully applied in many areas of healthcare with proven results—from the creation of better care pathways to advances in disease research and reducing medical errors. AI also has the potential to be a significantly more reliable, efficient, and cost-effective solution to mitigate improper claims payments that cost the healthcare industry hundreds of billions of dollars every year.
However, before diving too deeply into AI’s potential in payment integrity, let’s take a closer look at the size and scope of overutilization.
Dimensions of the Problem
Fraud is often cited as the primary culprit behind improper claims payments, but the reality is that overutilization drives a much greater share of overall waste—as much as 40 percent, according to Truven Health Analytics.
There are many causes of overutilization—most of which have nothing to do with Angelina Jolie.
For example, some healthcare services are overutilized because of provider practice patterns and the availability of resources. Physicians in a particular region of the country prescribe more pain medications than the national average because that is the local standard of care. Similarly, more back operations are performed in some regions than in others due to the relative availability of surgeons and hospital beds.
Overutilization also can result from disparities in provider contracts with complex riders, a shortage of staff resources to oversee accuracy of billed vs treated charges and a lack of financial investments in quality auditing and technological solutions to catch mistakes in claims before payments are made.
A more fundamental and hard-to-detect issue stems from billing behavior. Providers frequently contract with third-party vendors to handle claims coding and billing. While outsourcing can generate efficiencies and cost-savings, lack of clarity and consistency in procedures can lead to significant upcoding and unbundling of claims or the unnecessary use of modifiers.
How AI Makes a Difference
Unlike the resource-intensive and largely static nature of traditional payment integrity processes, intelligent algorithms continually learn and evolve with each claim. As a result, health plans now have an opportunity to reduce unwarranted utilization by far more than they have been able to do up to now.
However, identifying anomalies only scratches the surface of what is possible
Payment integrity covers a broad spectrum of issues starting with outlier identification all the way to behavior-led organized crimes. Payment integrity solutions have moved from basic rules related to payment integrity such as policy and pricing edits, diagnostic vs. procedure mapping, and other similar linear dimensions to a multi-dimensional approach working across the enterprise
Healthcare technology advancements have focused on increasing the effectiveness of multi-dimensional analytics—aggregating data from medical records, contact centers, claims, authorizations, and provider networks to capture trends (and errors) across the claims value stream.
This multi-dimensional analytic capability can also be applied to quickly emerging overutilization trends.
The sheer volume of claims also makes this new type of analytics valuable. For example, overutilization trends in use of transcutaneous electrical nerve stimulation (TENS) for chronic pain management could be mitigated by using AI-driven analytics to leverage social posts, peer-reviewed journal outcomes, and claims-related insights to significantly increase the effectiveness of audits, target higher propensity to over-pay claims, and eliminate dead-end searches.
APIs, Interoperability, and NLP Drive Efficiency
Program integrity innovations now focus on the increasing ability of health plans to obtain medical records directly from the EHRs of providers with whom they have value-based contracts. That interoperability enables the automation of common tasks such as prior authorization and records requests to determine the medical necessity of services. This kind of automation can save many hours of work for reviewers and coders at health plans.
The time savings can be increased by another 40 percent when AI-led natural language processing (NLP) is applied to unstructured data in the initial stages of review. So, while increasing the hit rate for identification of anomalies, a system that incorporates AI and NLP can also increase staff efficiency and reduce the use of high-cost resources such as nurses.
Augmenting AI Capabilities with Multi-Disciplinary Engagement
It’s important to note that an investment in new or upgraded technologies or analytics alone is insufficient to address the task at hand.
Health plans seeking real financial improvement and administrative efficiencies need to seamlessly couple technology with service capabilities and workflow improvements across the claims processes. Health plans will have to integrate payment integrity as an enterprise-wide initiative, starting with provider credentialing, special investigation units, customer complaints teams, and utilization management—all the way to post-service claims administration services.
Most important is engaging service providers that have the global delivery capacity to deploy the clinicians and high-skilled examiners to intervene in high-impact claims and behavior management. The clinical reviewers look at site of service, level of care, and medical necessity protocols and help providers get on a path toward a collaborative approach. They engage in benchmarking and ranking providers on a longitudinal scorecard as well as engaging in peer review analysis.
In conclusion, AI-based analytics and a multi-dimensional solution can help health insurers significantly cut the payment of improper claims by enabling more of those claims to be identified and acted on earlier. By denying claims for unwarranted utilization, health plans can help root out the waste that plagues the U.S. healthcare system.