The National Rifle Association (NRA) is not at its most popular at the moment, as American support for gun control rises. But perhaps people have been misjudging them, for it seems the NRA is cutting gun injuries, and probably deaths, albeit for just a few days a year.

NRA conventions are huge events; 81,000 people attended the 2017 annual meeting. Those who come are often there to buy guns, although some have other agendas, but are less likely to shoot them on site. Dr Anupam Jena of Harvard University and Columbia graduate student Andrew Olenski wondered whether having all these people in a controlled environment, rather than home with the guns they already own, would have an impact on the rate of gun injuries.

NRA rhetoric would suggest not. The organization and its supporters blame the deliberate use of guns on a limited pool of bad actors, and accidental injuries on inexperienced owners. In their view, neither are likely to be the sort to attend conventions. An alternative perspective holds that even law-abiding citizens sometimes snap in the wrong circumstances, and anyone can trigger an accident with a dangerous weapon if they handle it enough.

In the New England Journal of Medicine, Jena and Olenski report visits to emergency departments and hospitalizations for firearm injuries drop during NRA conventions compared to equivalent days three weeks before and after. The data was limited to people with private health insurance. Injuries dropped from 1.5 per 100,000 people to 1.2, a 20 percent decrease.

The finding was robust when controlled for age, sex, and state. Moreover, the drop was biggest among the demographics most likely to attend NRA conventions; men living in high gun-owning states. The effect was largest in regions close to the convention location.

The reduction in gun injuries was a result of fewer accidents – crime rates didn't change to any statistically significant extent. It's possible those who attend conventions are much less competent with guns than they think, and account for a large proportion of the accidental injuries – both to themselves and others caused by guns. Alternatively, the authors admit, there may be wider effects. A group of buddies may postpone a hunting trip, for example, if one is attending the convention, but it might be another of the friends who shoots someone when they do go.

The study didn't consider deaths. Thankfully, gun deaths are much less common than injuries, making it harder to get a statistically significant signal from such an analysis. However, it seems implausible that so many additional injuries can occur without sometimes resulting in fatalities.

So it seems the claims have been right all along: Guns don't kill people. People who really like guns kill people, even if they often do it accidentally.