(CNN) - After 26 children and educators were gunned down at Sandy Hook Elementary School in 2012 and President Barack Obama talked about increasing gun control, Americans bought more guns. But instead of offering protection, at least in some cases, the sales increased the chances that people would be accidentally shot and killed, according to a study published Thursday in Science magazine.
An additional 20 children and 40 adults, beyond what would be expected, died in accidental gun violence in the five-month period after Sandy Hook, the study found. This is considered a statistically significant increase during a time in which, the authors estimate, 3 million additional guns were sold beyond the general rate at which people typically buy guns.
No other spike in accidental deaths "of that magnitude" matches what happened after Sandy Hook, the authors wrote, and the states that had the biggest increase in sales saw the biggest increase in accidental gun deaths. The states with the smallest increase in gun sales saw the smallest number of additional accidental deaths.
The authors suggest that these numbers again confirm the theory that increasing access to guns increases the risk of accidents.
Phillip Levine, a professor in the Department of Economics at Wellesley College, and Robin McKnight of the National Bureau of Economic Research looked at data from Google searches, information from the National Instant Criminal Background Check System, which is used when people buy guns, and National Vital Statistics System mortality data.
Licensed gun dealers check with the background check system when someone buys a gun, but the database does not account for people who buy guns at shows or from private dealers. The number of people run through the system increased beyond the standard number after Sandy Hook, they found.
Google searches using the terms "buy gun" -- including phrases like "where can I buy a gun?" -- and "clean gun," meaning "how often should I clean my gun?" also spiked after Sandy Hook, and that was especially true on the days Obama talked about gun control.
"What's nice about Google data is that you can see specific cause and effect like this," Levine said. "Right after President Obama made his gun speech, search terms like 'buy a gun' immediately went up."
Dr. Jonathan Fielding of the University of California, Los Angeles' Fielding School of Public Health agreed that the numbers confirm the theory about increased access to guns. He complimented the strength of the study and its approach.
"It is so interesting to me, this phenomenon that happens that our initial reaction to hearing about mass shootings like Sandy Hook is to protect ourselves. It would make common sense to some people to think 'what I need to do to protect myself is to have a gun,' but the unfortunate truth is that this normal reaction is wrong," said Fielding, who was not involved in the new study.
Fielding pointed out that studies have shown that having a gun increases your risk of being killed by firearms, when it comes to homicide and suicide. Other studies have found that is true even regardless of storage practice. The new study did not find that more sales equated to more suicides or homicides, only that gun accidents increased.
"If you have an estimated 3 million more guns in homes, in all likelihood, not all of those guns will be safely stored and may be loaded and accessible, and that results in unintended harm. Humans are imperfect, despite their best intentions," said Dr. Stephen Hargarten, a professor and chairman in the Department of Emergency Medicine and director of the Injury Research Center at the Medical College of Wisconsin. "We don't even know the whole story, though, because this doesn't get at the whole scope of the gun violence, since presumably there are a lot of nonfatal gun events that happen, too."
David Hemenway, a professor of health policy at the Harvard T.H. Chan School of Public Health, agreed that the study is solid and that it will be helpful in public health discussions. "It is nice to say 'yes, this is actually the case; here are the numbers,' " he said. "Again, keep in mind, it is only one study."
There are limitations spelled out in the study: Vital Statistics typically undercounts the actual number of people who die from gun violence, and this study counts only deaths. The data do not capture the number of people injured in gun accidents, and it's unclear what impact these sales have on long-term trends toward gun violence.
"Personally, I think it's pretty convincing data overall, and I hope others will as well," Levine said. "We see a causal effect here is a pretty strong one, and when it happens at the same time like this, it's probably not an accident."