Investigating the association between household firearm ownership and suicide rates in the United States using spatial regression models. (November 2020)

• Record Type: Journal Article
• Authors: Tu, Wei; Ha, Hoehun; Wang, Weifeng; Liu, Liang
• Is Part Of: Applied geography. Volume 124(2020)
• Publication Date: 2020-11
• Subjects: Suicide rates -- Household firearm ownership -- Spatial dependence -- Regression models -- U.S. States and counties; Geography -- Periodicals; Human geography -- Periodicals; Human ecology -- Periodicals; 910
• DOI: 10.1016/j.apgeog.2020.102297
• Languages: English
• ISSNs: 0143-6228

Abstract

Abstract: Past studies have rather consistently shown a significant positive association between firearm ownership and suicide rates (particularly firearm-related) in the United States. However, the impact of spatial dependence of suicide rates has not been considered in the existing research that took an ecological studies approach. To bridge this gap in the literature, we estimated and compared the association using Ordinary Least Square (OLS), spatial autoregressive (SAR), and hierarchical spatial autoregressive (HSAR) regression models. The outcome variable was the average age-adjusted and smoothed suicide rate (all, firearm, and non-firearm) at the county level in the United States between 2008 and 2014. The covariates included the state-level firearm ownership and several key demographic, geographic, religious, psychopathological, and suicide-related variables at both the state and county levels. Our main findings were: 1) the spatially-informed models (SAR and HSAR) were significantly outperformed the OLS model. The SARlag model was a better choice than the SARerr model in fitting our data. However, the HSAR model was not statistically better than the SAR model; 2) based on the results from the three final SARlag models, firearm ownership was significantly associated with firearm suicide rates, but not with all or non-firearm suicide rates, and such relationships were also observed in the SARlag models using contiguity-based spatial weight matrix; and 3) The final SARlag models explained 82.6%, 84.9%, and 76.1% of the total variance in all, firearm, and non-firearm suicide rates, respectively. In conclusion, the SARlag model provided more robust evidence supporting the positive relationship between firearm suicide rate and firearm ownership. Results from this study confirmed that access to household firearms was a significant risk factor of firearm suicide in the United States and hence means reduction should be included in the suicide prevention strategy in the United States. In the future, the effect of state firearm laws and regulations on suicide rate should also be investigated using spatially informed models to gain insights on the connection among firearm ownership, firearm laws and regulations, and population-based suicide rates. Highlights: State-level firearm ownership was significantly associated with county-level firearm suicide rates. State-level firearm ownership was not significantly associated with county-level all or non-firearm suicide rates. Access to household firearms was a significant risk factor of firearm suicide in the United States. Means reduction should be included in the suicide prevention strategy in the United States.