National Meeting 2009

3051 — Excess Cardiovascular Disease-Related Mortality in Veterans with Serious Mental Illness: Are Behavioral or Treatment Factors to Blame?

Kilbourne AM (VA Ann Arbor HSRD/SMITREC) , Morden N (White River Junction VAMC), Austin K (VA Ann Arbor HSRD/SMITREC), Ilgen M (VA Ann Arbor HSRD/SMITREC), Welsh DE (VA Ann Arbor HSRD/SMITREC), McCarthy JF (VA Ann Arbor HSRD/SMITREC)

Objectives:
Cardiovascular disease (CVD) is associated with substantial premature mortality among persons with serious mental illness (SMI). Efforts to explain this differential have been limited by lack of behavioral and treatment data that may confound differences in mortality by SMI diagnosis. We determined whether VA patients with SMI were more likely to die from CVD-related conditions than those without psychiatric disorders, and whether behavioral and treatment factors explained the impact of SMI diagnosis on mortality.

Methods:
VA patients completing the 1999 national Large Health Survey of Veteran Enrollees (LHSV) and who were either diagnosed with SMI (schizophrenia, bipolar disorder, other psychosis) or had no psychiatric diagnosis in 1998-99 were included. LHSV data on patient, treatment (e.g., outpatient visits), and behavioral factors (e.g., exercise, smoking, substance use) were linked to mortality data from 1999-2004 from the CDC’s National Death Index. Multivariable Cox proportional hazards modeling was used to assess risk of CVD-related mortality (defined using CDC classifications) by psychiatric diagnosis and whether treatment or behavioral factors explained mortality risk.

Results:
Of 285,065 respondents, 16% were diagnosed with SMI and 59,698 (21%) died prior to 2005, of which 34% of deaths were attributed to CVD. After controlling for patient sociodemographic, treatment, and behavioral factors, patients diagnosed with schizophrenia (HR = 1.44;P < 0.001), bipolar disorder (HR = 1.27;P < 0.001) or other psychotic illness (HR = 1.24;P < 0.001) were more likely to die from CVD than those without psychiatric diagnoses. Other significant factors affecting CVD-related mortality included diabetes (HR = 1.39;P < 0.001), smoking (HR = 1.20: P < 0.001), lack of exercise (HR = 1.41;P < 0.001), and illicit drug use (HR = 1.25;P < 0.001), yet they did not explain the association between psychiatric diagnosis and CVD mortality. Similar trends were evident for all-cause mortality.

Implications:
VA patients with SMI had a 24-44% greater risk of dying from CVD than those without a psychiatric diagnosis. Treatment or health behavior factors contributed but did not fully account for the effect of SMI diagnosis on mortality.

Impacts:
Reducing preventable mortality among veterans with SMI is a priority within the VHA. Future research should focus on specific mechanisms by which mental illness increases the risk of CVD-related mortality, and additional interventions should be implemented to reduce the risk of premature mortality in this group.