On the morning of August 20 1986, U.S. postal worker Patrick H. Sherrill, 44, walked into the post office where he worked in sleepy Edmond, OK and started shooting.

First, he killed his supervisor who had given him a poor performance review the day before. From there, he systematically found and shot 19 other coworkers with three semiautomatic pistols. When the shooting stopped, Sherrill had killed 14 coworkers and injured six more before taking his own life. It was, at the time, the third largest mass shooting in American history and it would be the eventual origin of “going postal.” It would also make the phrase “workplace violence” a part of the American psyche.

The shooting prompted a federal investigation, a one-day Congressional hearing and a 7,000 page report.

30 years later, a cracked and tarnished monument stands out front of that Post Office and these incidents have become common. Usually, they only produce fleeting headlines (think San Bernardino or the beheadings in Moore, OK) despite the devastation they leave behind.

So where are we now?

The most recent numbers from the National Bureau of Labor Statistics show 14,770 workplace homicides took place between 1992 and 2012, averaging more than 700 per year. FBI data from 2013 shows homicides at work made up 9 percent of all workplace deaths.

Workplace homicides also trend along gender lines, meaning it matters who the victims are. They are the second leading cause of death for women at work. And although the sample sizes are vastly different, in 2013, more than 90 percent of workplace homicide victims were men.

There’s also a unique detail to workplace deaths in the U.S. Most are committed with guns. In 2010, 78 percent of workplace homicides were shootings, according to the Census of Fatal Occupational Injuries (CFOI).

It’s no secret the United States has a unique problem with gun violence. There are roughly 89 guns for every 100 people in the country. To put it another way, the U.S. population is 4.43 percent of the global population, but Americans possess 42 percent of all the guns in the world.

Research shows more guns equal more gun deaths, plain and simple. America has six times the amount of gun deaths as Canada and 16 times more than Germany, according to the Human Development Index report from 2012. That means the ubiquity of firearms and the number of deaths can translate or at least add to the number of workplace homicides.

But workplace violence is a broader issue than just homicides or mass incidents alone. According to the Occupational Safety and Health Administration, two million people are the victims of workplace violence each year. Those include incidents of harassment, assault and threats. However, these numbers do appear to be going down.

In 2013, the FBI reported fewer than 400 workplace homicides, a decrease of more than 12 percent. But it’s not clear how reliable reports about workplace violence even are. There are serious problems with how workplace violence data is collected.

Why is it so hard to track?

Workplace violence statistics in the U.S. are difficult to pin down for a few reasons. The first reason, the definition of workplace violence varies depending on the incident. Take the shootings in San Bernardino and the two on the U.S. Army post Ft. Hood, for example. All three had elements of both terrorism and workplace violence, but only one was considered the latter (The first Ft. Hood shooting was workplace violence).

Even incidents like the beheading attack at an Oklahoma food processing plant tread a blurred line between the two. That kind of ambiguity can lead to misreporting and possibly a misunderstanding of motives behind certain incidents.

The definition also varies depending on the study. The United Nations Office on Drugs and Crime points out several problems in its 2009 “Manual on Victimization.” Sample size poses a threat to research depending on the industry being studied. There’s also a big problem determining when an act of workplace violence starts and ends. This matters because most incidents are often considered on-going and may lead to under or over-reporting in certain circumstances.

When it comes to deaths in the U.S., things get even more tricky. Because of a law passed in 1996, the CDC hasn’t tracked gun violence for 20 years out of fear of having its funding stripped by members of Congress, backed by the National Rifle Association. That ban has since been lifted, but there still hasn’t been any movement on collecting new numbers.

Things get even worse when it comes to tracking mass shootings in work places. One major issue is the lack of a concrete definition on what exactly is a mass shooting. Many shooting and gun death trackers have landed on four or more people shot, following the model of the FBI. For instance, the tracking site GunArchive.org has tracked 245 mass shootings so far in 2016, 78 of them in the workplace. However, some use more or less deaths while others differentiate between people shot and people killed, making it difficult to get a broader sense of mass shooting and workplace violence deaths.

Possibly the largest difficulty is the number of unreported incidents of workplace violence. Experts differ on the actual number of unreported incidents and frankly, it’s most likely impossible to tell. But what we do know about it is that most people, at least anecdotally, say they wouldn’t report workplace violence out of fear of retaliation or escalation.

What do we know about how to prevent workplace violence?

Workplace violence is being reported with greater frequency and accuracy than 30 years ago. OSHA has specially designed handbooks for how to deal with workplace violence in industries where employees are at greater risk and suggest implementing a zero-tolerance policy.

The FBI also has a set of trigger environments or situations including things like whether or not money is exchanged, whether someone works in an isolated office or small group and whether or not there are unstable people around on a daily basis. Companies now routinely have new employees train on what to do about violence and how to report it.

All of those are big improvements, but there is still a lot to be done. The better we can understand how, where and why workplace violence happens, the better the U.S., the safer the work place can be for employees and customers.