National Collaborative on Gun Violence Research (NCGVR) Firearm Injury Study
We’ve set out to study firearm injury in a way that has not been done before…
…to unlock new insights on how to prevent firearm deaths.
(Above) The five-year age-standardized firearm mortality rates are shown on the map of the United States. Counties with lower mortality rates are shown in blue and green, and higher mortality rates are shaded in yellow, orange, and red. Nationally, firearm deaths and suicides increase from 1989-1993 to 2015-2019 while firearm homicides decreased; however, these national trends varied across counties and geographical regions. The West and Midwest had increases in firearm suicide rates and the Southeast had increases in firearm homicide rates. Figures originally published in JAMA Network Open.
Between 1989 and 2019, there were well over 1 million firearm deaths in the United States
In that time, suicide was the leading cause of firearm deaths (589,285 lives) followed by homicide (412,231 lives)
To learn why some counties did not follow the national trends, we started by identifying counties that were especially “low” or “high” in firearm mortality rates…
Click through the tabs below to see which counties are low and high outliers for all firearm deaths, as well as deaths by homicide and suicide. The low outliers are shown in yellow and the high outliers are shown in red. States are shaded green or blue depending on their number of low or high outlier counties, respectively, with darker shading meaning more outliers.
Low and High County Outliers for Firearm Mortality Rates in 2015-2019
These counties had the greatest percent increases in firearm deaths
from the 1989-1993 period to the 2015-2019 period
Very rural counties (populations of less than 10,000 people) experienced the biggest percent increases in all firearm deaths and firearm suicides.
The greatest percent increases in firearm homicides happened in slightly larger rural counties (populations between 10,000 and 50,000 people).
Though they are not graphed here, it is noteworthy that St. Louis City County (Missouri) and Baltimore City County (Maryland) experienced 48% and 60% increases in firearm homicides, respectively, between the two time periods. Out of all counties that experienced large increases, these two counties had the highest firearm homicide rates in 2015-2019 (66.69 deaths per 100,000 people and 47.43 deaths per 100,000 people, respectively).
Note: counties with firearm mortality rates equal to zero were not included as a percent cannot be calculated
These counties had the greatest percent decreases in firearm deaths
from the 1989-1993 period to the 2015-2019 period
Very rural counties (populations of less than 10,000 people) also experienced the biggest percent decreases in all firearm deaths and firearm suicides.
The District of Columbia demonstrated a large decrease in all firearm deaths, mainly as a result of the significant decrease in firearm homicides.
Several urban counties in New York (Kings, Bronx, and New York counties; populations greater than 1 million people) had very large decreases in firearm homicides.
Another noteworthy observation (not graphed) is the 24.36% decrease in firearm homicides in Orleans Parish County (Louisiana). Despite that, the 2015-2019 firearm homicide rate remains high at 39.43 deaths per 100,000 people.
Note: counties with firearm mortality rates equal to zero were not included as a percent cannot be calculated
Looking for more data?
Use the CDC WONDER data query system to find public health data
Once we knew which counties had firearm mortality rates significantly lower or higher than expected, we asked…
“How are these counties similar or different?”
…in hopes we can start to explain why some counties are more affected by firearm deaths while other counties are less affected.
Comparisons of County Characteristics by Outlier Status
Firearm Homicide
| Characteristics | Non-Outlier Counties (n=3034) Mean (SD) | Low Outlier Counties (n=14) Mean (SD) | High Outlier Counties (n=63) Mean (SD) | Significance (p value) |
|---|---|---|---|---|
| Geography | ||||
| Rurality-urbanicity (1 to 9, where 1 = most urban and 9 = most rural) | 5.10 (2.67) | 6.71 (3.02) | 5.11 (3.02) | .079 |
| Land area (sq miles) | 957.17 (1314.71) | 1002.42 (1019.95) | 751.95 (639.92) | .7 |
| Sociodemographic | ||||
| Sex ratio (M/F) | 0.99 (0.09) | 1.01 (0.11) | 0.94 (0.07) | <.001 |
| Aged 15-65 (%) | 66.05 (3.30) | 66.44 (2.30) | 64.79 (2.59) | .023 |
| Black (%) | 8.44 (13.57) | 23.88 (25.54) | 35.48 (27.47) | <.001 |
| Hispanic (%) | 7.12 (12.55) | 11.74 (20.83) | 5.64 (10.20) | .5 |
| Education | ||||
| High school graduate (%) | 83.18 (7.30) | 75.96 (7.49) | 79.45 (7.45) | <.001 |
| Economic | ||||
| Median household income ($) | 39,258.12 (10,039.63) | 31,921.36 (7,169.45) | 31,278.94 (5,987.54) | <.001 |
| Unemployment rate (%) | 5.34 (1.72) | 6.56 (2.82) | 6.86 (2.44) | <.001 |
| Poverty (%) | 15.20 (6.38) | 21.64 (8.80) | 22.83 (8.08) | <.001 |
| Politics | ||||
| Republican voters (%) | 60.48 (12.31) | 55.82 (20.80) | 50.08 (17.63) | <.001 |
| Health | ||||
| Heavy drinkers (%) | 13.17 (2.34) | 14.25 (1.72) | 13.21 (2.17) | .079 |
| Trauma care access (# of level I trauma centers within 60mi) | 1.45 (2.87) | 0.64 (1.65) | 0.94 (1.27) | .092 |
| Firearm dealers | ||||
| Firearm licenses (firearm dealer or pawnbroker) per capita | 26.05 (43.39) | 4.57 (3.55) | 26.54 (40.04) | <.001 |
Firearm Suicide
| Characteristics | Non-Outlier Counties (n=3046) Mean (SD) | Low Outlier Counties (n=12) Mean (SD) | High Outlier Counties (n=53) Mean (SD) | Significance (p value) |
|---|---|---|---|---|
| Geography | ||||
| Rurality-urbanicity (1 to 9, where 1 = most urban and 9 = most rural) | 5.04 (2.66) | 8.83 (0.39) | 8.13 (1.53) | <.001 |
| Land area (sq miles) | 940.55 (1292.12) | 1101.69 (555.57) | 1647.50 (1810.42) | <.001 |
| Sociodemographic | ||||
| Sex ratio (M/F) | 0.99 (0.09) | 1.03 (0.11) | 1.01 (0.05) | <.001 |
| Aged 15-65 (%) | 66.06 (3.25) | 64.44 (4.91) | 64.25 (4.14) | <.001 |
| Black (%) | 9.23 (14.69) | 1.22 (3.38) | 0.64 (1.33) | <.001 |
| Hispanic (%) | 7.07 (12.56) | 4.67 (7.04) | 9.73 (13.11) | .3 |
| Education | ||||
| High school graduate (%) | 83.01 (7.34) | 85.75 (5.16) | 86.26 (7.13) | .002 |
| Economic | ||||
| Median household income ($) | 39,168.94 (10,069.29) | 33,035.00 (7,003.28) | 34,369.85 (6,701.12) | <.001 |
| Unemployment rate (%) | 5.38 (1.74) | 5.29 (2.34) | 5.12 (2.42) | .06 |
| Poverty (%) | 15.36 (6.51) | 18.58 (10.82) | 15.83 (6.11) | .4 |
| Politics | ||||
| Republican voters (%) | 60.04 (12.46) | 66.68 (15.35) | 70.55 (13.48) | <.001 |
| Health | ||||
| Heavy drinkers (%) | 13.15 (2.26) | 16.62 (5.99) | 14.35 (4.07) | .045 |
| Trauma care access (# of level I trauma centers within 60mi) | 1.46 (2.87) | 0.00 (0.00) | 0.25 (0.73) | <.001 |
| Firearm dealers | ||||
| Firearm licenses (firearm dealer or pawnbroker) per capita | 26.42 (43.59) | 3.17 (3.30) | 4.98 (4.18) | <.001 |
What does it mean?
From the tables above and other analyses, there are a number of key differences between counties based on their “outlier” status…
1. More federal firearm licenses (i.e., the number of places where you can buy a gun) were located in high county outliers for all firearm deaths and firearm homicides.
2. High firearm homicide outlier counties were clustered in the Southeast, including Alabama, Mississippi, and Georgia.
3. Counties with unexpected increases in firearm homicide over time had higher unemployment rates, higher poverty rates, and lower median household incomes. High county outliers for homicide also had a larger Black population (35%) compared to non-outlier counties (8%). Taken together, these findings point to the ongoing public health consequences of racist policies and practices related to housing, development, banking, and more.
4. Counties with unexpected increases in firearm suicide over time were largely White, rural, and with less access to level I trauma care centers. These findings highlight the public health issue of firearm suicide in rural counties and the lack of access to health services in rural America.
Putting it all together…
This study identified national trends in firearm mortality rates over time, and the counties that deviated from those trends in positive or negative ways.
In doing so, we’ve identified counties that can be examined further to understand what exactly is causing firearm deaths to get better or worse. Examining this at the county level (as opposed to national or state) provides new opportunities for more accurately understanding and addressing firearm deaths.
We now have more information to address the public health issue of losing lives to firearms. Researchers and policymakers should look to these county outliers to learn what has worked in reducing firearm deaths and to develop targeted local interventions for communities that are most affected.
Want to hear more about what we found?
Let us break it down another way…
A “tweetorial” from one of the study authors that breaks down the study in a Twitter thread that you can share
Everyone wants to reduce #GunViolence. But how? Let the data show us. Our new @JAMANetworkOpen study led by @Michelle_Degli "hotspots" U.S. counties in which gun deaths have ⬆️ or ⬇️ much more than expected over the past 20 years What did we learn?https://t.co/KmgkXrbNXh
— M. Kit Delgado, MD, MS (@kit_delgadoMD) June 6, 2022
A blog post written by study authors, published by the Leonard Davis Institute (LDI) of Health Economics
Read the full article
The open access article is published in JAMA Network Open and provides additional information on the analyses and takeaways.
Stay tuned for more...
Thank you to the research sponsors
This work is funded by The Arnold Foundation through the National Collaborative on Gun Violence Research and administered by the RAND Corporation. It is one of 25 projects funded by The Arnold Foundation over the last 3 years.
