Medicaid managed care plans have been touted for keeping spending down, and a recent government report finds they are successful in taming improper payments.

However, the findings might be far from accurate.

The Government Accountability Office, in a May 7 report (PDF), said that about 0.3% of the $171 billion that Medicaid paid to managed care organizations last year, or about $500 million, was improper.

That's compared to about 10%, or about $37 billion, under Medicaid as a whole.

The drastic difference between two rates made the GAO question the managed care rate. The watchdog said the Center for Medicare & Medicaid Services' estimate for managed care has limitations and might "minimize the appearance of program risks."

GAO recommended that the agency take steps to mitigate the program risk, which could include revising estimation methodology and focusing additional audit resources on managed care. HHS agreed with the recommendations.

RELATED: GAO recommends CMS continue prior authorizations

"Without addressing key risks, such as the extent of overpayments and unallowable costs, CMS cannot be certain that its estimated improper payment rate for managed care accurately reflects program risks," GAO said.

Lawmakers also weighed in on the findings.

“These are important programs with limited resources, and it is imperative they are maximizing every dollar,” Rep. Greg Walden, R-Ore., chairman of the House Energy and Commerce Committee, said in a statement. “Updating their error measurement tracking will be an important tool in reducing the amount of improper payments in the Medicaid program, allowing CMS to focus on their beneficiaries and the care they need.”

Additionally, in a separate report (PDF), the GAO found that the Department of Health and Human Services spent about $90 billion on improper payments in 2017.

On top of the $37 billion under Medicaid, improper spending for fee-for-service Medicare was $36.2 billion last year, while Medicare Advantage and Part D hit $14.4 billion and $1.3 billion, respectively.