The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #70233

Lynne Fullerton-Gleason, PhD¹, Jennifer Personius, MA², Jonathon Lavalley, BS³, Michelle Chino, PhD⁴, and G. Thomas Shires, MD¹. (1) Trauma Institute, University of Nevada School of Medicine, 2040 W. Charleston Blvd, Suite 501, Las Vegas, NV 89102, 702-671-2276, lfullerton@salud.unm.edu, (2) Nevada Institute for Children, University of Nevada Las Vegas, 4505 Maryland Parkway, Box 45-3030, Las Vegas, NV 89154-3030, (3) Department of Emergency Medicine, University of New Mexico, MSC10 5560, 1 University of New Mexico, Albuquerque, NM 87131-0001, (4) UNLV Center for American Indian Research and Education, University of Nevada Las Vegas, 4505 Maryland Parkway, Box 45-3030, Las Vegas, NV 89154-3030

Purpose: To determine the relationship between factors posited to be associated with suicide death and state suicide death rates. Methods: State-specific data were collected from internet-available government datasets. Results: Bivariate analyses indicated that state suicide death rates were positively correlated with percent of firearm suicides (p<.0001), percent of individuals living in households with firearms (p<.0001), death rates from unintentional motor vehicles (p=.0001) and unintentional injury (p<.0001), American Indian/Alaska Native population (p<.0001), and internal population increase (p=.006). Variables negatively associated with suicide death rates were: per capita income (p<.0001), population density (p=.002), emotional disturbances among children (p=.0066) and adults (p=.015), and total population (p=.0167). Legalized gambling was not correlated with suicide death rates (p=.87). Two multivariate models were constructed. The best fit model for total suicide death rates (R-squared=.74) included five variables: population density was negatively correlated with suicide death rates (p<.0001), and positively correlated were non-firearm unintentional injury rates (p=.0002), American Indian/Alaska Native population percentage (p=.0014), internal migration count (p=.013), and past month use of illicit drugs (p=.022). The second model was simpler but less explanatory (R-squared=66%) and included only population density (p=.0006), non-firearm unintentional injury rates (p<.0001), and American Indian/Alaska Native population percentage (p=.0052). Conclusion: Ecological variables significantly predict state suicide death rates when multivariate techniques are used. Bivariate analyses missed the relationship between drug use and suicide death, and indicated other relationships that disappeared in the multivariate model. The ultimate model did not include any measure of household firearm ownership or legalized gambling.

Learning Objectives:

• Understand the utility of ecological variables
• Identify on-line sources of injury data
• List three factors correlated with state suicide death rates in the multivariate model.

Keywords: Suicide, Epidemiology

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.