Summary of Firearms Mortality in the U.S., 1999-2017

In the current issue of Health Affairs, Jason E. Goldstick and co-authors present a Datawatch article, “US Firearm-Related Mortality:  National, State, and Population Trends, 1999-2017” (available here).

Here’s the authors’ summary of the article:

“Nationwide firearm-related mortality rates increased in 2015–17 after remaining relatively stable in 1999–2014. Recent increases are reflected across most states and demographics to varying degrees, which suggests a worsening epidemic of firearm mortality that is geographically and demographically broad. In both time periods the fractions of firearm deaths due to suicide and homicide remained consistent.”

The authors’ Exhibit 1, shown above, supports their contention that there are two epochs:  a relatively stable period 1999-2014 and then three years of increased mortality in 2015-17.  They go on to offer a range of informative tables, graphs and a detailed discussion of changes in mortality rates by sex, race, ethnicity, and mechanism.

A More Detailed Look at Mortality from 1999-2017

I accessed the Centers For Disease Control and Prevention’s mortality data to see for myself.

Here’s a plot of the crude rate of firearm mortality, unadjusted for changes in age distribution.

The pattern of the rates matches the authors’ Exhibit 1, with slight reductions in adjusted rates relative to crude rates for the later years of the series.

Using the ggplot2 package in R statistical software, it is easy to visualize a ‘small multiples’ display of crude rates by mechanism of mortality along with the overall plot; in ggplot2, the function that makes small multiples is called facet.

I use facet plots regularly; the built-in facet function makes it easy to explore patterns related to factors (strata) in data sets.

Examination of the plots by mechanism supports a more nuanced interpretation of the mortality series than a simple ‘two-epoch’ summary.  

There are signals of non-random patterns using run chart rules for the Accidents, Suicide and Homicide plots. 

The Accidents and Suicide plots are especially easy to summarize. Accidental firearm mortality drops in half almost linearly from 1999 to 2017; suicide mortality increases monotonically from 2006 to 2017.  Homicide rates are a bit more complex, appearing to decrease in 2009 to 2015 relative to the first 10 years of the series and then increasing in the last three years of the series.

Of course, the contribution of accidental deaths to the overall rate is small relative to homicides and suicides.  The overall rates are driven by the combination of homicide and suicide deaths.  However, focus on just these two plots suggests that a two-epoch description of firearm mortality does not adequately describe the data.  

Notes on the data set

Data Source:  Centers for Disease Control and Prevention, National Center for Health Statistics. Multiple Cause of Death 1999-2017 on CDC WONDER Online Database, released December,2018. Data are from the Multiple Cause of Death Files, 1999-2017, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. Accessed at http://wonder.cdc.gov/mcd-icd10.html on Oct 13, 2019 8:36:25 AM

 The definitions of the mechanisms of death may be found in this summary document. Here are the descriptions of the categories I used in this post:

Accidents: "Accidental discharge of firearms (W32-W34)"                       

Homicide: "Assault (homicide) by discharge of firearms (*U01.4, X93-X95)"     

Undetermined: "Discharge of firearms, undetermined intent (Y22-Y24)"             

Suicide: "Intentional self-harm (suicide) by discharge of firearms (X72-X74)"

  The alphanumeric strings reference ICD-10 codes used in the National Violent Death Reporting System.