A Pennsylvania-based Blue Cross Blue Shield affiliate deserves high marks for taking an $11.5 million-plus bite out of healthcare fraud that costs the U.S. as much as $170 billion annually. The company, Highmark, also earns an “A” in cross-functional collaboration, antifraud technology, and relationship management with providers and other stakeholders.
Three years ago, Highmark, which provides health insurance to more than 25 million individuals, developed a new antifraud analytics application it dubbed the “Financial Investigation Reporting System for Tomorrow” (FIRST). After one year of use, FIRST, along with the sleuthing savvy of the company's special investigations unit (SIU) personnel and the analytics mastery of Highmark's healthcare informatics group, helped the company to generate $11.5 million in claims recovery and cost avoidance.
Since then, the annual cost savings — produced by a combination of claims recovery and cost avoidance as well as process and systems improvements sparked by the antifraud investigative work — generated each year have been “more significant than the prior year,” SIU Director Tom Brennan reports. “I still need to demonstrate that monthly ROI to my boss, and we're still generating plenty of ROI.”
Although healthcare fraud may at first blush sound like an industry-specific challenge, the problem's effects (and remedies) can be felt by (and applied to) other industries and organizations (and risk management challenges).
Healthcare fraud qualifies as a massive problem inside and outside of the healthcare industry. It can weaken the quality of care patients receive. Fraud also creates significantly higher costs for insurers and, of course, the individuals and organizations that pay insurance premiums to cover these costs. The National Healthcare Anti-Fraud Association estimates that fraud adds $51 billion to $170 billion to U.S. healthcare costs each year.
This explains why healthcare insurers have SIU departments staffed with tough-talking, fair-minded investigators like Brennan, whose descriptions of recent cases, which sound as if they should be accompanied by Law and Order background music, include reports of fraudsters' prison sentences, multimillion-dollar restitution payments, and seizures of mansions, luxury cars, and, in the instance of one fraudulent chiropractor, 86 Rolex watches.
The details drive the home the fact that fraud, regardless of whether it is perpetrated against insurance companies or other corporations, is a crime with serious costs and penalties. And while few companies in other industries possess special investigations units or informatics functions, most organizations do house similar skills in their internal audit and finance or IT functions (think of the financial planning and analysis wizard everyone turns to when they need to unearth that compelling nugget of data).
The interplay between these two skill sets — investigative/risk management and data analytics — is central to Highmark's antifraud success; it is also applicable to a wide range of risk management capabilities in all industries, particularly within companies that seek to predict and prevent costly problems before they occur.
When Shawn McNelis, Highmark's vice president of health plan informatics, joined the company three years ago, the first thing he did was to begin meeting with his business partners. During those meetings, McNelis, whose background includes artificial intelligence and knowledge management work at IBM and PNC, would announce his group's intention of becoming more strategic in their partnering. His group, which provides analytical support to numerous internal and external clients (including the U.S. Centers for Disease Control and Prevention), then began executing this intention while providing its partners with information, analyses, predictive modeling, and other reports.
“Over time, you build up a reputation and you also have a series of showcase examples of what's possible when other groups partner with you,” he notes.
SIU's leadership espoused a similar approach to working with business partners, and it seemed destined that the two functions would create an extremely beneficial partnership.
This partnership initially was fueled by highly manual collaboration. SIU relied predominantly on homegrown applications to act on the fraud tips it received. “But we're so large, and our data is so massive … that running an extract could take hours, days, or even weeks, depending on the case,” says Brennan.
The extracts were performed on a data warehouse of claims data that is fed by data from Highmark's operational systems and care management systems. As the primary user of this warehouse, McNelis's group suggested to Brennan that they could provide some technical support to the investigations. An analyst would work with an SIU investigator who would describe a potential fraud under investigation. The analysts would then devise an algorithm, which was then used to plumb the data warehouse for data and patterns that would either support or refute the possibility of fraud.
Equipped with this information, the investigator would then continue the investigative process, which routinely includes requests for medical records, reviews of the records by a medical consultant, interviews of patients, and interviews of other witnesses, among other steps. Frequently, the SIU investigator would return to the analyst with requests for additional data drill-downs.
“It was a fairly iterative process,” McNelis recalls, “and it was labor-intensive.”
It was also a successful process. “We were doing well,” Brennan confirms, noting that the collaboration yielded numerous investigation cases, many of which were determined to be fraud. “We recognized that we didn't need more cases. We needed to work better, faster.”
Brennan and McNelis realized that addressing this need required automation. “Informatics and SIU wanted to expand this work,” recalls McNelis. “We weren't able to do that given the labor that an expanded workload would require, so we looked for ways to automate.”
Since the company already used SAS technology as its primary analytics platform, it made sense to consider SAS along with other vendors for a tool that could be tailored to produce ad hoc reports that SIU could generate on its own. The duo evaluated several applications. “We wanted something more advanced than just rules-based technology,” Brennan notes. “We know most of the fraud schemes that take place, so we hoped to find a tool that could help us to discover schemes that we don't already know about. This got us into predictive modeling.”
Highmark invested in SAS Enterprise Miner, which became the basis for the FIRST system. The application's front end, or user interface, is Web-based and therefore looks familiar to the SIU specialists who use it. After an initial training on the new tool from the informatics group, SIU conducts its own training of new employees.
Today, SIU specialists no longer need to consume hours of their informatics partners' time with requests for data warehouse queries; they use the tool to run the queries themselves.
“We now run reports and get extracts in minutes — sometimes seconds — where before it took us hours, days, or even longer,” Brennan reports. “And we're now able to identify issues much more quickly.”
The productivity gains that Brennan's SIU unit posted sound impressive. Shortly after the tool's introduction, previous caseload constraints faded away. The unit's caseload increased by 30 percent, which translates to productivity gains and staffing savings of roughly $200,000 per quarter, according to Brennan.
The informatics department has also benefited. “This application has allowed SIU to do some multiple of work without requiring that equivalent significant multiple of work from my staff,” says McNelis. “There is a huge financial impact. In other situations, there may be other layers of business intervention or involvement that sort of clouds the value of contributions you have made to that effort. In this case, I think that there is a clear and distinct line between the analytic work and the financial outcome.”
The impact extends beyond fraudulent claims, in large part because SIU is very careful — and sensitive of the company's relationships with providers, the insured, pharmacies, and third-party billers — to identify what Brennan describes as “weird but OK” situations.
“We want and strive to maintain good relationships with providers we contract with,” McNelis emphasizes. “So we don't just leap on every odd billing pattern and look into provider operations as potential fraud. There are many reasons why something might appear suspicious. The folks in SIU do a very good job of conducting exhaustive research into a potential for fraud.”
The primary method that SIU uses to identify fraud is a form of exception reporting in which investigators compare providers and other billers to their peers. When a provider's claims pattern falls outside of normal billing behavior, that's weird. Highmark's investigators then conduct additional research to determine why the provider qualifies as an outlier. For example, a pain specialist may be working primarily in the field of podiatry, so his claims activity more closely resembles a podiatrist's claims activity; that's weird, but OK.
“We also recognize that we're not identifying only fraud in these situations,” Brennan says. Odd claims patterns may also be the result of errors in billing, payment, or coding systems, or they may be the result of manual errors that sometimes occur or outdated policy or contract terms. Correcting these errors can also yield significant returns.
“We've identified issues with our medical policies that led to adjustments to medical polices that resulted in millions of dollars of savings,” Brennan says. “We've identified issues with claims processing where relatively minor edits to the system have resulted in millions of dollars of savings. The same holds true for contracts.”
These savings demonstrate that Brennan has been very generous with the “weird but OK” insights that his investigators uncover.
“If you develop these solutions and want to just hold them under your vest, you really are not doing the corporation the justice it deserves,” he says.
These highly visible cost savings and the growing use of the tool also help to pave the way to expand the tool's use. Other business units and functions within Highmark, including the company's internal audit department and its provider claims review department, now use the tool.
The analytics tool can also be used in more sophisticated — and more proactive — ways, as SIU and informatics continue to demonstrate. SIU recently developed a predictive medical model that (based on sophisticated data analysis) ranks 21,000 providers according to the likelihood that their billing patterns are either fraudulent or “weird but OK.” When the model first spit out a list of the top 200 providers (flagged because of outlying claims activity), SIU investigators let out a collective “Ah-ha!” It turns out that 44 of the 200 companies were already the subjects of investigations. The model also identifies the volume of potential fraudulent activity, which among the 44 providers under investigation was 20 percent higher than SIU had projected. In other words, the predictive modeling showed SIU exactly what it didn't know, as Brennan had hoped.
Again, SIU treats the rankings as information to be confirmed or refuted with additional investigation; many providers are on this list due to a variety of system and process errors. When these errors are discovered, a newly formed cross-functional triage team (consisting of SIU investigators, claims experts, and medical experts) evaluates them and then relays the information to the appropriate Highmark function or unit to resolve.
FIRST's success has Highmark thinking about how to make fraud prevention even more proactive. A new process would analyze individual claims before they are paid; questionable claims would be investigated and resolved before the insurer pays the bill, which would greatly increase cost recovery volumes.
“In a pay-and-chase environment, you're only going to recoup 20 percent to 30 percent once the money goes out the door,” Brennan explains. “If you can stop it before it goes out, that's where you make the most significant impact.”
And this would mean even higher marks on fraud prevention for the highly valuable collaboration between SIU and informatics at Highmark.
Three years ago, Highmark's antifraud activities generated $11.5 million in savings via claims recovery and cost avoidance. Since then, the annual savings from these activities as well as system process improvements — sparked by other information generated by antifraud explorations — have “increased significantly,” reports Tom Brennan, director, special investigations unit (SIU), for the Pennsylvania-based Blue Cross Blue Shield affiliate.
The success of Highmark's healthcare-fraud prevention activities hinges on the following factors:
Reputation. Highmark's SIU has earned a reputation as a model SIU in the healthcare industry, in large part because the department has long produced impressive fraud-detection results with relatively modest resources. This credibility has helped to open doors to cross-functional collaboration.
Collaboration. Collaboration is almost always a key component of GRC (governance, risk, and compliance) success; what's noteworthy at Highmark is how various functions and departments collaborate. The partnership between SIU and informatics, the group responsible for analytical support throughout the company, started as an informal “How can we help each other?” discussion. After they teamed up and implemented a new analytical application to, as Brennan notes, “work better faster,” SIU began sharing its reports with several other areas of the company. The information-sharing led to system and process improvements, along with major cost savings, in Highmark's claims-processing and contract management areas.
Human intervention. Highmark's new automated fraud-detection capabilities, fueled by an SAS application, lets SIU scour a massive warehouse packed with data from the company's claims and operational systems to generate information that helps to support or refute possible instances of fraud. However, SIU examines these query results carefully and sensitively — keeping a vigilant watch for “weird but OK” results that relate to people or process changes or errors, but not necessarily to fraud. Doing so helps Highmark to preserve a healthy relationship with its providers.