Predictors of injury-related and non-injury-related mortality among veterans with alcohol use disorders.

AIMS To describe the association between alcohol use disorders (AUDs) and mortality and to examine risk factors for and all-cause, injury-related and non-injury-related mortality among those diagnosed with an AUD. SETTING Department of Veterans Affairs, Veterans Health Administration (VHA). PARTICIPANTS A cohort of individuals who received health care in VHA during the fiscal year (FY) 2001 (n = 3,944,778), followed from the beginning of FY02 through the end of FY06. MEASUREMENTS Demographics and medical diagnoses were obtained from VHA records. Data on mortality were obtained from the National Death Index. FINDINGS Controlling for age, gender and race and compared to those without AUDs, individuals with AUDs were more likely to die by all causes [hazard ratio (HR) = 2.30], by injury-related (HR = 3.29) and by non-injury-related causes (HR = 2.21). Patients with AUDs died 15 years earlier than individuals without AUDs on average. Among those with AUDs, Caucasian ethnicity and all mental illness diagnoses that were assessed were associated more strongly with injury-related than non-injury-related mortality. Also among those with AUDs, individuals with medical comorbidity and older age were at higher risk for non-injury related compared to injury-related mortality. CONCLUSIONS In users of a large health-care system, a diagnosis of an AUD is associated significantly with increased likelihood of dying by injury and non-injury causes. Patients with a diagnosis of an AUD who die from injury differ significantly from those who die from other medical conditions. Prevention and intervention programs could focus separately upon selected groups with increased risk for injury or non-injury-related death.