Unintended shootings in a large metropolitan area: An incident-based analysis☆☆☆★

Introduction 

In 1999, firearms were implicated in 28,874 fatalities in the United States.1 Although the vast majority of these deaths were suicides (16,599) or homicides (10,828), approximately 3% were unintentional (824), comprising 30,467 potential years of life lost. One hundred fifty-eight of these unintended deaths were pediatric (<18 years old).1 A disproportionate number of unintentional firearm injuries involve children.2, 3, 4

The case/fatality ratio for unintentional shootings is estimated to be between 13:1 and 100:1.2, 3, 5 Although some of these injuries are relatively minor, others are severe.3, 4, 5 The costs of acute care, rehabilitation, and long-term disability caused by firearm injury are substantial, as much as $2.3 billion in 1994.5, 6 Total costs, including lost productivity resulting from injury-related death and disability, might be 15-fold greater.7

Despite the magnitude of the problem, little is known about the factors that contribute to unintentional shootings.5, 8, 9 To determine the proportion that might be prevented by various strategies, we identified unintentional shootings in a major metropolitan area and classified them by causation.

Methods 

An electronic firearm injury surveillance system was used to identify fatal and nonfatal shootings in a 5-county area of metropolitan Atlanta, GA. The strengths and limitations of this system have been described in detail in an earlier report.10 Five medical examiner's offices, 22 area emergency departments, and 33 law enforcement agencies submitted case reports to the system. Records were linked to generate as complete a picture as possible of each event. This study was exempted by our institutional review board.

A case was defined as an injury caused by the unintentional discharge of a projectile from a powder firearm. Powder firearms included all kinds of pistols, rifles, and shotguns but excluded air rifles and BB guns. Incidents of unintended discharge that did not result in injury and incidents of injury resulting from blunt trauma (eg, gun dropped on a victim's foot) were excluded. Bystanders inadvertently shot during an attempted assault or drive-by shooting were considered victims of intentional injury and were excluded as well. The initial classification was made by the reporting agency. Our study interval was limited to shootings that occurred between May 1, 1996, and June 30, 2000.

Four members of our team (RBI, AR, RA, and ALK) independently studied each report to determine the circumstances of the shooting. Both the ED data collection form and, if available, the written law enforcement narrative were reviewed. In most instances, the police report was the only source that contained information about the circumstances of the shooting. In no case did the ED report or medical examiner's report conflict with a police report regarding injury circumstances. Cases were classified into 1 or more of 3 predefined categories: Preventable by safe storage, preventable by safer handling, and preventable by safer design. When the shooter was a minor who gained access to an adult's gun without the adult's permission, we considered the incident preventable by safer storage. Preventable by safer handling was selected when the narrative indicated that the firearm discharged (1) during cleaning, (2) while clearing a jammed round or attempting to unload, (3) while playing with or showing off the firearm, or (4) while moving, handling, or catching the firearm. Preventable by safer design was selected when the narrative specifically noted any of the following: (1) the shooter did not realize the firearm was loaded; (2) the magazine was out of the firearm or removed by the shooter before the trigger was pulled; or (3) the firearm discharged when dropped or placed on a hard surface. Some cases fell into more than one of these categories and were coded as such. After initial coding, our team met to review cases that were coded differently and to seek a consensus.

Confidence intervals (CIs) were estimated by using a z statistic for 1-sample proportions. Interrater reliability was measured by using the proportion of cases in which investigators agreed, with cases weighted by the percentage of investigators who agreed. For example, if 3 of the 4 raters judged a case to be preventable by safer design, that case would receive a weight of 0.75. A sensitivity analysis was performed by assuming that all cases with missing information would not have been preventable by any of the methods described. Statistical analyses were done with SAS for Windows statistical software package (version 8.02, SAS Institute, Inc., Chicago, IL).

Results 

During the 49-month study period, 5,735 cases of firearm injury were reported in our 5-county metropolitan area. A total of 247 (4.3%) of these were initially classified as unintentional. After a review of these case reports, 31 of these shootings were reclassified as either intentional or of uncertain intent, leaving 216 cases (3.8% of all shootings during the study period) in our analysis. Six (2.8%) of these unintentional shootings resulted in a fatality.

A matching police offense report was linked to the ED report for 145 (67%) cases. For most of the remaining cases, the only documentation available was the 1-page reporting form that lists victim age, sex, and race, as well as the manner and severity of injury. The form does not provide enough information to classify the specific circumstances of the event.10

Eighty-seven percent (187) of the 216 victims were male, 65% (141) were black, 24% (51) were white/non-Hispanic, and 8% (18) were members of other racial or ethnic groups. The age distribution of victims is similar to the national mortality profile for unintentional firearm-related deaths; one fourth involved victims younger than 18 years (Table 1).

Table 1. Unintentional firearm injuries in metropolitan Atlanta, GA, May 1996 through June 2000 (N=247).

	Variable	Included in Analysis, No. (%)	Insufficient Information, No. (%)	Excluded (Not Unintentional), No. (%)
	Age distribution, y
	0–4	1 (1)	2 (2)	0 (0)
	5–14	17 (14)	6 (6)	2 (6)
	15–24	51 (42)	41 (44)	13 (42)
	25–34	19 (16)	23 (24)	7 (23)
	35–44	18 (15)	9 (10)	5 (16)
	45–54	7 (6)	4 (4)	2 (6)
	55–64	5 (4)	2 (2)	1 (3)
	≥65	0 (0)	1 (1)	0 (0)
	Not specified	4 (3)	6 (6)	1 (3)
	Race distribution
	Black	79 (65)	62 (66)	17 (55)
	White	33 (27)	18 (19)	12 (39)
	Other	9 (7)	9 (10)	1 (3)
	Not specified	1 (1)	5 (5)	1 (3)
	Sex distribution
	Female	14 (11)	14 (15)	9 (29)
	Male	107 (88)	80 (85)	21 (68)
	Not specified	1 (1)	0 (0)	1 (3)

Nearly 40% (85) of the victims shot themselves. Sixteen percent (34) were shot by a friend or acquaintance and 5% (10) by a family member. One percent (3) were unintentionally shot by their spouse or intimate partner. Thirty-eight percent of unintended shootings (n=81; nearly two thirds of those in which the incident location was specified) occurred in a home. Seven percent (16) occurred in an automobile. Only 3 cases involved hunting. The vast majority of unintentional shootings (188 [87%]) involved a handgun (Table 2).

Table 2. Unintentional firearm injuries in metropolitan Atlanta, GA: Incident details.

	Variable	No.	%
	Type of firearm
	Handgun	188	87
	Shotgun	6	3
	Rifle	11	5
	Unknown	11	5
	Victim-shooter relationship
	Self	85	39
	Family	10	5
	Friend	34	16
	Spouse/intimate partner	3	1
	Stranger	7	3
	Not specified	77	36
	Incident location
	Residence	81	38
	Street or sidewalk	20	9
	Woods or field	4	2
	Bar or tavern	3	1
	Retail establishment	3	1
	Automobile	16	7
	Parking lot	9	4
	Other	3	1
	Not specified	77	36

Among the 204 shootings in which information about the nature of the wounds was available, 17 were to the head or face, 21 to the chest or abdomen, 5 to the back, 53 to an arm, and 104 to a leg. Most victims had only a single wound. Of those for whom the ED disposition was recorded, 48 were admitted, 1 was transferred to another hospital, and 52 were discharged home. Three were pronounced dead in the ED.

Enough information was available to characterize incident circumstances in 122 cases (57% of the total). All but 6 of these fell into 1 or more of 3 broad categories of causation: Preventable by safe storage, preventable by safer handling, or preventable by safer design (Table 3).

Table 3. Unintentional firearm injuries in metropolitan Atlanta, GA: Analysis of incidents and their causes.*

	Category	No.	%	95% CI, %	Lower Limit (Sensitivity Analysis),† %
	Preventable by safer storage	17	14	8–20	4
	Preventable by safer handling	90	74	66–82	35
	Preventable by safer design	39	32	24–40	13
	*The total number of cases is 122. Categories are not mutually exclusive, and therefore, the total exceeds 100%. †The sensitivity analysis assumed that the lower CI was correct and that none of the cases for which there was insufficient information were preventable in any way.

Ninety shootings (74% of those analyzed) were attributed to mishandling of the firearm. Eighteen occurred while the owner was cleaning the gun or clearing a jammed round, 42 occurred while the user was playing with or showing off the gun, and 30 occurred when the user moved, fumbled, or dropped (but then caught) the gun.

Thirty-nine shootings (32% of those analyzed) were attributed to potential deficiencies in the firearm's design. In 17 cases, the investigating officer specifically noted that the shooter was unaware that the weapon was loaded. A loaded chamber indicator is designed to alert the user that a round is in the chamber.4 Six shootings occurred while the pistol's magazine was removed, possibly giving the handler the mistaken impression that the weapon was unloaded. A magazine safety is designed to block the trigger when the magazine is removed to prevent discharge of a round retained in the chamber.4 Nineteen shootings occurred when the gun was dropped or struck a hard surface. A firing pin block, also known as a “drop safety device,” is designed to prevent a gun's firing pin from contacting the cartridge if the weapon is dropped or struck against a hard surface.4

Reviewers independently agreed on 98.2% of cases regarding exclusion because of lack of sufficient information. Complete agreement was reached on 94% of cases that were judged to be potentially preventable by safer storage, 78% of cases that were judged potentially preventable by safer handling, and 85% of cases that were judged to be potentially preventable by basic mechanical safety features. When disagreement occurred, there was a clear majority view in all but a handful of cases (2%, 6%, and 2%, respectively). Almost all of these initial disagreements were the result of a reviewer inadvertently overlooking a relevant fact or term in the police narrative. Once these were identified, complete agreement was reached on better than 99% of cases in each category.

Discussion 

Unintentional firearm injuries accounted for a small fraction of firearm-related fatalities in our community during the study interval but caused a somewhat larger percentage of nonfatal injuries (0.2% and 3.9%, respectively). Most of the victims were male patients between 15 and 34 years of age. An incident-based analysis revealed several options for prevention, including promoting safer storage of guns in the home, teaching safe handling of firearms, and incorporating basic safety features into new gun designs.4, 5, 11, 12, 13

Approximately 40% of US households contain 1 or more firearms.14 The average gun-owning household contains 4.15 Between one fourth and one third of gun-owning households contain at least 1 handgun.14, 15, 16 Rates of firearm ownership are greater in rural areas and small towns, but households in urban areas are more likely to contain handguns.16 The rate of gun ownership in metropolitan Atlanta is similar to that noted in other metropolitan areas of the south and west.17

Persons who keep a firearm for protection are much more likely to store the weapon loaded and readily available than people who own firearms for other reasons.8, 14, 16, 18, 19 However, keeping an unlocked and loaded gun in the home violates a central tenet of firearm safety. The National Rifle Association's “A Parent's Guide to Gun Safety” advises owners to “always keep the gun unloaded until ready to use” and to “store guns so that they are inaccessible to children and other unauthorized users.”11 The Clinician's Handbook of Preventive Services echoes this admonition.12 There is evidence that few gun dealers share this advice with customers, even those with young children.20 In any case, many gun-owning parents store their firearms in an unsafe manner.19, 21, 22, 23, 24

When children find a gun, they often play with it. Jackman et al25 placed pairs and trios of 8- to 12-year-old boys in a room with a 1-way mirror and observed them for 15 minutes. An actual .38 caliber pistol, altered so it could not be fired, was concealed in a drawer. Instead of a magazine of bullets, the pistol contained a radio transmitter that activated a light whenever the trigger was pressed with enough force to discharge the weapon. Of the 29 groups tested, 21 discovered the gun within 15 minutes of being placed in the room. Members of 16 (76%) groups handled the gun, and 1 or more members of 10 (48%) groups pulled the trigger. During subsequent questioning, nearly half of the boys said that they were unsure whether the gun was real or a toy. More than 90% reported having prior gun safety instruction. Hardy et al26 observed a group of young children and noted that when they were left unsupervised around guns, they touched and played with them, despite clear instructions not to do so.

When children gain unsupervised access to a gun, the consequences can be tragic. Wintemute et al27 studied fatal shootings of children by children in California and noted that more than half occurred when children played with a loaded gun they found in the home. Grossman et al28 studied unintentional and self-inflicted firearm injuries of children in Seattle and noted that many involved a gun found in the victim's home or the home of a relative or friend.

In hopes of teaching young children to avoid touching a gun if they find one, the National Rifle Association developed the Eddie Eagle program.29 Although the curriculum has reportedly been taught to more than 12 million children in 10,000 schools, it has not been objectively evaluated to confirm that it is effective.30

Adult training programs are not very effective at encouraging safe storage of guns in the home. Weil and Hemenway16 surveyed 605 adult gun owners and found that those who had received firearms training were no more likely to store their guns safely than those who did not. Cook and Ludwig14 analyzed responses to an independent survey of more than 1,600 gun owners and obtained similar findings. They did note, however, that gun owners trained by the National Safety Council were somewhat more likely to report storing their gun safely than those trained by other organizations.

Several states enacted laws that hold the owner responsible if a child gains access to the gun and is injured to promote safe storage of guns. An evaluation of these child access prevention laws concluded that enactment was associated with a 23% decrease in the rate of unintentional firearm-related deaths of children younger than 15 years of age.31 However, another group studied the effect of child access prevention laws in the 15 states in which they were in effect and found evidence of effect in only one, Florida.32

Safe handling might have prevented 66% to 82% of the unintentional shootings in our study. Firearm safety training programs emphasize several worthwhile concepts, including “always keep the gun pointed in a safe direction” and “always keep your finger off the trigger until ready to shoot.”33 Trainees are taught to assume that every firearm is loaded unless they can personally verify that it is unloaded by inspecting the chamber.

Despite the intuitive appeal of firearm safety training, researchers have found that education alone is often ineffective at promoting safe behavior, particularly when it involves a complex series of actions.34, 35 Not surprisingly, the demographic group at greatest risk of unintentional injury (ie, young men) is also the group that is least receptive to safety training.35 First-time applicants for a driver's license are required to demonstrate their ability to safely operate a motor vehicle, but first-time purchasers of a firearm are not required to learn or demonstrate safe handling skills.36

Little thought has been given to the idea of making safer guns.37 In 1988, the US General Accounting Office studied the extent to which child-proof safety devices or a loaded chamber indicator could prevent firearm-related deaths.5 Researchers randomly selected 107 fatal unintentional shootings from urban and rural jurisdictions across the United States and reviewed the case files to characterize each event. On the basis of the incident narratives, they concluded that a loaded chamber indicator might have prevented 23% of the deaths and that a child-proof safety device might have prevented another 8%. The remaining deaths were either judged to be nonpreventable by these 2 strategies, or the data were insufficient to make a determination. Other safety devices were not considered.

We found evidence that loaded chamber indicators, magazine safeties, and firing pin blocks might have prevented as many as one third of the unintended shootings in our series. Widespread adoption of these safety features, plus routine use of devices that prevent unauthorized child access, might have prevented as many as 46% of the unintended shootings in our series.

It is possible that other safety devices might have prevented additional shootings. For example, a grip safety device automatically locks the pistol's trigger mechanism unless the weapon's grip is properly grasped. A positive safety device is designed to prevent the firearm from being discharged unless it is purposefully disengaged.4 These devices might have prevented some of the shootings attributed to mishandling.

Emerging technologies could produce even safer guns.37 It should be possible, for example, to design a firearm that can be easily fired by adults but not by young children. Widespread adoption of child-resistant aspirin bottles prevented many unintentional poisonings; widespread adoption of child-resistant firearms might prevent many unintentional shootings.37, 38

The technology exists to manufacture personalized handguns that can only be fired by their owners.37, 38, 39 Personalized guns would be particularly useful for law enforcement because a surprising number of officers are shot and killed each year by an offender who grabs their service weapons.40

Despite the potential benefits of safer gun designs, there is no impetus for the gun industry to adopt them. Federal law specifically exempts domestic handgun manufacturers from consumer product safety regulations. The public is largely unaware of this fact. Half of the respondents in 2 recent polls by the National Opinion Research Center expressed the belief that all or some guns are regulated for safety.18 Sixty-eight percent of respondents, including 64% of gun-owning respondents, supported the idea of “government safety regulations for the design of guns.” Remarkably, 94% of respondents (including fully 93% of those who own guns) agreed that “handguns made in the United States should be required to meet the same federal safety and quality standards as imported handguns.” Eighty-eight percent of respondents to the National Opinion Research Center surveys, including 80% of gun-owning respondents, endorsed the idea that new handguns should be legally required to be child-resistant. Seventy-one percent of respondents (including 59% of gun owners) agreed with the statement, “All new handguns should be personalized.”18

There are several limitations in our study. First, the system we used to identify cases relied on voluntary reporting. Three sources of data were used to maximize rates of detection: A 1-page incident report faxed from area hospital EDs, county medical examiner's case files, and police offense reports from local law enforcement agencies.10 Between May 1996 and December 1998, an audit revealed that 13% of ED reports submitted to our system could not be matched with a corresponding police report.10 This strategy minimized the potential for missed cases, but it could not eliminate it entirely.41 Individuals who did not disclose their injury to the police or seek care in an ED, as well as those who were treated in an ED but not reported, would not be detected by using our system.

Misclassification can occur in any retrospective study. Our data on causation are only as good as the police reports on which they are based.3, 41 To minimize misclassification, we manually reviewed each case before including it in our series. Twenty-one shootings initially classified as unintentional were subsequently reclassified as intentional or indeterminate and excluded from our analysis.

Missing data complicated our efforts to categorize the circumstances involved in these shootings. No police report was available for 95 of the cases reported by ED personnel. Even when a police report was secured, the documentation was sometimes too sketchy to accurately characterize the event. At the outset of the study, we had hoped to identify the specific make and model of firearm involved in these shootings to confirm whether it lacked specific safety features. Unfortunately, investigating officers rarely documented the specific make and model in their offense reports.

Finally, our findings are limited to a single 5-county metropolitan area of a major southern city and might not be applicable to major cities in other parts of the country, much less to smaller communities or rural areas. For example, only 3 of our incidents were linked to hunting. Nationwide, as many of one fifth of all unintentional shootings are related to hunting.

Nonetheless, our results support 2 important conclusions. First, almost all of the unintended shootings we studied could be attributed to 1 or more of 3 factors: Child access, mishandling of the firearm, and deficiencies in design. All 3 causes might be amenable to intervention. Nearly half the shootings might have been prevented if the gun involved had incorporated 3 simple mechanical safety features—a loaded chamber indicator, a grip safety, and a firing pin block—and had been kept inaccessible to children. Most of the remaining incidents might have been prevented if the handler had followed the most basic safety rules. Identifying which strategy or which combination of strategies is most effective will require further research.

This study would not have been possible without access to data from a regional firearm injury reporting system. A nationwide system of this sort, with attention paid to prospective collection of high-quality data, could generate invaluable information about the circumstances, location, and contributing factors involved in thousands of shootings each year.42 The reports produced would not only be useful to local and state law enforcement, they would also be useful to firearm safety instructors, gun enthusiasts, gun industry executives, and others interested in preventing deaths and injuries caused by firearms.