U.S. Department of Education data on how frequently restraint and seclusion are used in the nation’s schools is incomplete and inaccurate, government investigators say.

Numbers reported through the agency’s civil rights data collection drastically undercount schools’ use of the practices, which disproportionately affect students with disabilities, the Government Accountability Office said in a report released this week. GAO indicated that the Education Department needs to take “immediate action” to remedy the situation.

More than 17,000 school districts — or 70 percent — told the Education Department that they did not have a single instance of restraint or seclusion during the 2015-2016 school year. However, GAO found that nine of the largest districts reported zero incidents, even though that wasn’t accurate or they did so when in fact they didn’t have data at all.

The findings likely point to a larger issue, GAO said.

“While it is difficult to know the full extent of underreporting of restraint and seclusion in the (Civil Rights Data Collection), the fact that only one of the 10 largest districts that reported zeros actually affirmed that it had no incidents calls into question the data showing zero incidents for 70 percent of the nation’s public school districts,” the report found.

GAO said the Education Department should remind school districts when to report zero instances on the data collection versus leaving the line blank, which indicates that numbers are unavailable. In addition, investigators want the agency to follow up with any districts that report zero in the current data collection for the 2017-2018 school year and they want the agency to work to ensure that school plans for providing future figures are submitted and complete.

According to the report, the Education Department agreed to follow through on those recommendations, but pushed back on GAO’s idea that the agency should correct known errors in past data collections saying only that it would “amend the data notes to ensure the public is aware of potential changes” in the 2015-2016 report.

Federal data on restraint and seclusion in schools was first collected for the 2009-2010 school year and is now reported every two years.

“The GAO’s report makes clear that school districts across the country are not accurately reporting the use of restraint or seclusion — disciplinary practices that jeopardize student and staff health and safety,” Reps. Bobby Scott, D-Va., and Don Beyer, D-Va., said in a joint statement. “Today’s report is also further evidence that the state patchwork of restraint and seclusion standards is failing to protect students and educators.”

The lawmakers said they plan to reintroduce legislation later this year to establish federal standards for the use of restraint and seclusion in schools.

The Education Department said that its Office for Civil Rights was already working to improve the quality of its data on restraint and seclusion as part of a major initiative announced in January.

“OCR is pleased that GAO agrees that data quality is an important goal in this context. OCR is already in the midst of making improvements and therefore agrees with the spirit of GAO’s recommendations,” the agency said in a statement.