Another day, another scam defrauding insurers and governments. For purposes of full disclosure, the case I'm highlighting today comes from Washington's Labor and Industries (L&I), the agency where I formerly worked and headed up fraud prevention efforts, and the investigation dates back to my time there. During my time there, I saw so much fraud and abuse from interpreter agencies that I once said I would give a reward for anyone that could find an honest one. That's hyperbole, but it reflects the exposure that workers' compensation insurers, Medicaid and Medicare face from providers, particularly those with lower credentials and huge risk for fraud.
The latest case from L&I comes in at $600,000 in value. The owners of Hispanic Voices, the firm providing services (sometimes) and billing (all the time!), used a mix of schemes. They ranged from phantom billing for services never rendered, billing two or three times as long as real services took, and billing for services provided by non-certified interpreters (aka grab your cousin and start interpreting).
Other cases I've seen showed all sorts of bad behavior. Billing for providing interpretation services from the same interpreter in multiple locations at the same moment (very impressive!). Charges for significant travel time to provide the service which never occurred. Some of the more interesting schemes involved recruitment of non-English speaking workers in smaller communities to file false workers' compensation claims, with the same primary care providers and interpreters billing and everyone splitting the ill gotten gains.
Okay, enough bashing of interpreters. There are plenty of other high risk provider categories and treatments out there - durable medical equipment, physical therapy, chiropractors, spinal cord stimulators and the like. While lower in some specialties, known fraud cases range the gamut of virtually every type of procedure and provider type. So, what to do about it?
Beyond increasing requirements for licensing, background checks and the like, the real opportunity is to ensure that data sets are analyzed appropriately. Bouncing billing time for services rendered by the interpreters off the the addresses of the patient and doctor, as well as the services provided by the doctor, begins to show a range of what may be likely on an individual billing. Going further to analyze the number of licensed interpreters versus billing hours, which can regularly show more than 24 hours worked in a day helps as well. Good analysis of patient populations, need and risk continue to expand analysis. For government agencies, matching against other data sets - like revenue/tax base, ownership and employee reporting quickly begin to show the gaps that reflect the fraud.
These approaches work regardless of the type of provider. However, beginning with the areas of biggest risk can help significantly. One state, working with SAS, began to segment Medicaid billings, and started with high risk areas. As an example, non-emergency medical transportation was a significant cost. When properly analyzed for risk, outliers and with predictive models based on past cases utilized, they were able to save tens of millions of dollars annually. From just one type of service!
The best approach is to use multiple data sets, multiple approaches and ensure that fraud detection is continuous monitoring, not just ad hoc runs. The latter produce successes, but fail to guard against future risks, serving more like fraud Whack-a-Mole than a true plan.