When a dental clinic began quietly changing procedure codes on previously rejected claims, altering dates of service and billing for treatments patients never received, it was data — not a tip — that caught it.
Sun Life’s in-house analytics platform flagged the suspicious billing pattern. Investigators moved in, reviewing patient charts maintained by the dentist. The clinic initially issued a refund when confronted with the concerns, but the billing irregularities persisted. Sun Life ultimately filed a formal complaint with the dentist’s regulatory body and delisted the provider.
Cases like that one illustrate how profoundly fraud detection has changed in the group benefits industry over the past decade, said Shelley Frohlich, AVP of fraud risk management at Sun Life.
From rules to machine learning
Historically, fraud detection in the group benefits industry was largely rules-based and relied on tips and routine claim audits. These fraud detection tools when used on their own, have become static, and limited.
“Over the years, schemes have become increasingly sophisticated and widespread. As schemes evolve, so must technology. Advanced data analysis techniques and pattern recognition algorithms allow us to quickly analyze vast amounts of data to detect unusual claiming patterns that can be indicative of fraud or abuse,” Frohlich said. “And we’re using artificial intelligence to spot concerning patterns in large amounts of data.”
Sun Life now uses a combination of AI, machine learning predictive models, complex risk rules, behavioural pattern analysis, and human intelligence to screen claims. According to the company, all claims, no matter the submission channel – electronic and paper – go through fraud detection, using a risk-based approach.
Red flags
The detection system considers multiple data points, tracking both individual claims and the full claims history tied to a certificate.
“Our data trending and profiling tools, which include provider data capture, can identify irregular claiming patterns and behaviours — plan members or providers testing coverage limits either through electronic or paper claims, unusual or suspicious provider or pharmacy claiming patterns, and potential collusion between service providers and plan members,” Frohlich said.
When suspicious patterns emerge, investigators take over — contacting employees, medical and dental providers, or pharmacies to validate the claim, or conducting site visits when warranted, among other types of investigative techniques.
Built in-house for speed
Sun Life’s fraud detection tools are built internally. said Frohlich. This means they can be updated quickly as new fraud schemes emerge.
“Our fraud detection tools are built in-house, which ensures we continuously and quickly evolve and adapt to the ever-changing fraud schemes landscape,” she said.
The company’s Fraud Risk Management team is supported by more than 100 experts with backgrounds in data analysis, law enforcement, investigations, healthcare, and insurance. Data scientists develop machine learning algorithms to detect anomalies in health and dental benefits claims. These systems leverage years of accumulated data to identify irregular claiming patterns and flag anomalous claims.
The pharmacy line of defense
For pharmacy claims specifically, the pharmacy benefit manager — an entity that administers direct drug submissions from pharmacies on behalf of health insurance plans — uses its claims processing system as a first line of defense. Every electronic claim from a pharmacy goes through several verifications confirming benefit coverage, plan rules, and cost before being processed.
”We leverage data trending and profiling tools to uncover pharmacies exhibiting red flags and claiming irregularities that require investigation,” Frohlich said. Investigators work alongside the pharmacy benefit manager to examine flagged leads. Investigations may include direct contact with employees, or pharmacies and removal of the pharmacy’s eligibility to submit claims, when warranted.
Looking ahead
” With growing recognition of the power of AI in fraud detection and the increasing evidence of its effectiveness, I anticipate continued growth in this area,” Frohlich said.
For employers, the evolution of fraud detection means the tools now in play are capable of finding schemes that were likely to have been missed a decade ago. But technology alone cannot do it — human investigators remain essential for interpreting what the data reveals and acting on it. Plan sponsors and members also play a critical role in fraud prevention by recognizing fraud and abuse and taking steps to report it.
