AUSTIN, Texas >> The massacre at the First Baptist Church in Sutherland Springs has focused national attention on the Air Force’s failure to report to the FBI a 2012 military court conviction that should have prevented the shooter from buying an assault-style weapon he used to kill 26 people and injure 20 more.

But gun experts say problems with the background check system for gun purchases go well beyond gaps in military conviction records and have factored into many of the mass shootings that have gripped the country in recent years.

“The military failure is just the issue we know about most recently,” said Garen Wintemute, an ER doctor and gun violence researcher who directs the University of California’s Violence Prevention Research Program. “Failure to report is a widespread problem that has been recognized at the federal level and by reporting agencies for decades.”

The National Instant Criminal Background Check System — a network of databases maintained by the FBI that federally licensed gun sellers check before making sales — is supposed to prohibit certain groups of people from buying guns, including drug abusers, felons, fugitives, unauthorized immigrants, people convicted on domestic violence charges and those involuntarily committed to a mental health facility.

But there are significant gaps in the records caused by a myriad of factors:

>> The federal government cannot force states to report conviction data, and many don’t submit all of the relevant records.

>> Congress has not spent more than $1 billion of authorized funding for a program meant to incentivize states to improve their reporting of court records.

>> Differences in how states and military courts classify certain crimes, such as misdemeanor domestic violence offenses, result in some convictions being left out.

>> Ambiguity in federal law about certain categories of people who shouldn’t be able to buy guns leads to inconsistent reporting.

Despite these issues, the background check system, commonly called NICS, has prevented about 2.5 million people from buying guns since it launched in 1998, which is about 1 percent of the 250 million transactions it processed through the end of last year.

The system has improved in many areas since Congress in 2007 approved a $1.3 billion program to encourage states to improve their reporting. The bill was prompted by the 2007 shooting at Virginia Tech University that left 32 dead and was carried out by a student who purchased weapons after having been determined by a court to be dangerously mentally ill.

But only $110 million of that money had been distributed by the federal government by the end of last year, as Congress and the Bush and Obama administrations in their annual budgets routinely sought and distributed less money for state reporting grants than the law authorized.

The man who attacked the Sutherland Springs church, Devin P. Kelley, had a military court domestic violence conviction that should have barred him from buying a gun. Yet he purchased one gun each year since 2014, according to officials with the federal Bureau of Alcohol, Tobacco, Firearms and Explosives.

U.S. Sen. John Cornyn, R-Texas, last week announced he will push legislation to strengthen the reliability of the background check system, and many members of Congress have called for better reporting of military convictions.

Previous efforts to bolster the NICS have faltered because of political and logistical hurdles. The National Rifle Association, which favors background checks over other forms of gun control, has played a “mixed role,” said Lindsay Nichols, federal policy director for the Giffords Law Center to Prevent Gun Violence.

While the influential NRA has publicly supported measures like the 2007 law, it has also pushed to exempt more groups from the prohibited gun buyers list, protect loopholes that allow buyers to circumvent the background check system, and defeat spending measures that could improve it, Nichols said.

“The background check system has been effective to the extent that the records are there,” Nichols said. “But as we have seen certainly in the Sutherland Springs shooting, every time a record does not exist, it has the potential to lead to great tragedy.”

Even if the database were complete, purchasers can evade background checks altogether through a provision known as the “gun show loophole,” which exempts small private sellers from having to use the FBI system.

“Getting the reporting improved is important, but it can’t be used for a smoke screen to hide the fact that this other huge loophole exists,” Nichols said.

BACKGROUND INVESTIGATIONS

The 1993 Brady Handgun Violence Prevention Act — named after James Brady, the press secretary who eventually died of injuries suffered in a 1981 assassination attempt against President Ronald Reagan — established the NICS, which launched four years later.

A gun background check includes searches of the Interstate Identification Index, which tracks state court records; the National Crime Information Center, which compiles many nonconviction records like domestic violence protection orders; and databases set up specifically for the NICS for records not captured by the other two.

In most states, including Texas, federally licensed gun sellers call the FBI to conduct the search. For “point of contact” states like California, which have stronger laws on gun purchases than the federal government, the sellers call a state law enforcement office that performs the NICS check.

A vast majority of the checks take only a couple minutes, but ambiguities in a buyer’s history can result in a lengthier investigation. If the probe goes beyond three days, federal law allows the store to sell the guns while the case is resolved. If a purchaser is later found to have been ineligible, the Bureau of Alcohol, Tobacco, Firearms and Explosives is responsible for seizing the weapons.

DOMESTIC VIOLENCE

For most types of criminal offenses, federal law only prohibits people from buying firearms if they have been convicted on felony charges. But for domestic violence, misdemeanor offenders are also barred.

Because states classify domestic violence charges differently — some do not even have a specific charge for family violence and instead charge abusers with assault and battery — the FBI faces significant challenges in determining which misdemeanor records should trigger the prohibition on buying guns, and it frequently has to follow up with state and local officials as cases come up.

The difficulty in interpreting domestic violence cases results in longer background check investigations, a 2016 report by the Government Accountability Office showed. For about 30 percent of the 59,000 misdemeanor domestic violence offenders who tried to buy guns between 2006 and 2015, the investigations took longer than 30 days.

Because gun sellers can — but are not required to — sell guns to purchasers whose cases take longer than three days to adjudicate, more than 6,200 prohibited domestic violence offenders were able to buy guns during that 2006-15 period, the report found.

Only 12 states, including Texas, proactively flag misdemeanor court records that likely prohibit convicts from buying guns in the future, the National Consortium for Justice Information and Statistics found in another report.

NONCRIMINAL RECORDS

Another major gap in the NICS records is for events that should disqualify people from buying guns but don’t involve criminal convictions.

For instance, federal law prohibits all abusers of illegal drugs, not just those who have been convicted on drug charges, from buying guns. That means state records showing someone has admitted to drug use or failed a drug test are supposed to be entered into the NICS indexes.

The number of those records in the database has increased dramatically since the 2007 law, from about 5,600 to almost 23,000. But as of the end of 2016, 26 states had never submitted information from a noncriminal drug use case for inclusion in the NICS database, according to a report by the National Consortium for Justice Information and Statistics.

People subject to protective orders for domestic abuse are also prohibited from buying guns. While states consistently report protective order data to the National Criminal Information Center, most do not consistently flag them with a “Brady indicator” that automatically triggers a prohibition on gun purchases.

It is difficult to determine whether orders without the indicator should prevent someone from buying guns because the prohibition only applies to protective orders in which the accused abuser is the spouse, ex-spouse, parent or cohabitant of the victim.

In 2015, only about 37 percent of protective orders contained the indicator, according to the Government Accountability Office report.