With Medicaid Errors at $36 Billion, HHS Told to Improve Data Quality

With changes in Medicaid high on the Republican Congress’ agenda, the Government Accountability Office is calling for improved quality checks on state-provided data.  

“We determined that available Medicaid expenditure and utilization data do not provide [HHS’ Centers for Medicare and Medicaid Services] with sufficient information to consistently ensure that payments are proper or that beneficiaries have access to covered services, which is inconsistent with federal internal control standards,” GAO wrote in a report released on Monday. “Federal Medicaid administrators rely on state-reported data to inform oversight activities, but we found continuing concerns regarding such data's completeness, accuracy and timeliness.”

Erroneous payments in fiscal 2016 were estimated at $36 billion out of the roughly $550 billion federal-state program for low-income patients.

To oversee Medicaid spending, CMS relies on spending and health care utilization data that states submit from their Medicaid Management Information Systems, which have been inconsistently implemented nationwide, delaying data sometimes for years, GAO noted. Only 18 states are currently providing complete data, though HHS officials expect improvements soon.

GAO made technical recommendations to improve data oversight. A transformed system, auditors estimated, could replace a system with 200 variable data points spread over five files and checked quarterly for quality with a system that provides 550 data variables over eight files, with 3,500 quality checks performed monthly.

HHS officials agreed.

The same day the report was released, the Justice Department announced it had reached a $60 million settlement under the False Claims Act with a major hospital service provider, TeamHealth Holdings. It had allegedly billed Medicaid (along with Medicare, the Defense Health Agency and the Federal Employees Health Benefits Program) “for higher and more expensive levels of medical service than were actually performed (a practice known as “up-coding’).”