Students should be secluded or physically restrained only in emergencies, and only as a last resort. That’s what the research says, and that’s the policy adopted in Maine in 2012, after it became clear there was little oversight on how schools were handling students exhibiting problem behaviors.

Five years later, the situation in Maine has greatly improved, if only for the uptick in accountability and increased attention on doing things right.

But ineffective techniques are still used too frequently, and disproportionately against students with disabilities. If the last five years have been about finding the scope of the problem, then the next five should be about getting help where it is needed.

The rules put in place in 2012, known as Chapter 33, were designed to reduce the use of physical restraint — such as a bear hug or forcing a child to lie on the floor — and seclusion — when a student is placed involuntarily in a safe room. School districts are now required to report to the state each use, and to review their use quarterly. Prior, the use of restraints and seclusion often were not even reported to parents.

Districts were slow to respond, particularly to the reporting requirements, according to a recent study by Disability Rights Maine. Just 54 percent of schools issued reports in the first year, 2013, though 88 percent did last year. And because of discrepancies in how things were reported by individual schools, the data is not wholly reliable.

But what is clear is that restraints and seclusion are used multiple times against a small percentage of students, and far more often against students with disabilities.

In 2016, for instance, emergency interventions were used in Maine more than 13,000 times, but with only 1,800 students. In the last four years, 86 percent of interventions were used against students with disabilities, who represent just 16 percent of the 166,000 or so students in Maine.

Not surprisingly, the interventions took place disproportionately in the special purpose private schools and school-based day programs that only educate students with disabilities.

That makes sense — that population includes students exhibiting the most difficult behaviors, and teachers and other staff often have to intervene in extreme ways to keep students and adults safe.

However, the use of those interventions is usually a sign that something has gone wrong — in most cases, there is a way to respond, before the situation gets too bad, to keep a student’s behavior from escalating. Besides, those interventions are rarely effective in the long term.

“There continues to be no evidence that using restraint or seclusion is effective in reducing the occurrence of the problem behaviors that frequently precipitate the use of such techniques,” a 2012 report from the U.S. Department of Education said, summing up the research.

It’s reasonable for an educator put in such a position to wonder how they can go from restraining someone more than 75 times in three months, as happened to a 10-year-old student with autism mentioned in the DRM study, to something close to zero. But they don’t have to do it alone.

The state should target help — better programming, more training — to the programs with continued high rates of interventions. They should be reserved for emergencies, and it’s hard to believe that emergencies have to happen 13,000 times a year.