Denis G. Birgenheir, Ph.D.
Zongshan Lai, M.P.H.
Amy M. Kilbourne, Ph.D., M.P.H.
crease the likelihood of early mortal-
ity. We determined years of potential
life lost (YPLL) over an 11-year
period among Department of Veter-
ans Affairs (VA) patients using home-
less services, assessing for the impact
of severe mental illness.
Using the VA National Patient Care
Database (NPCD), we analyzed all-
cause mortality for fiscal years (FYs)
2000–2009 among VA patients by
severe mental illness status and use
of VA homelessness services for each
FY. All patients with a severe mental
illness diagnosis from each FY were
identified by ICD-9 codes. We also
used NPCD data to identify a 5%
random sample of individuals without
FY. The primary analyses were hazard
ratio calculations on all-cause mortality,
comparing differences in YPLL ad-
justed for age and gender by homeless-
ness and severe mental illness status.
Regardless of severe mental illness
diagnosis, homeless veterans died
younger than nonhomeless veterans
in all years, with YPLL ranging from
18.9 to 24.3 (Figure 1). For non-
homeless veterans with and without
severe mental illness, YPLL ranged
from 9.1 to 14.0 years. Homeless
veterans with severe mental illness
oth homelessness and severe
mental illness are known to in-
had the greatest YPLL over time.
Nonhomeless veterans with severe
mental illness died younger than
those without a severe mental illness.
Using time-series analyses, we found
a statistically significant increase in
YPLL only for nonhomeless veterans
without severe mental illness.
Overall, having a severe mental
illness diagnosis increased YPLL re-
gardless of housing status. Homeless-
ness was an additional and more
severe contributor to YPLL, above
and beyond a severe mental illness
diagnosis. Thus, in this study, home-
lessness had the strongest effects on
YPLL. However, VA patients who
were not homeless and who did not
have a severe mental illness diag-
nosis were also dying younger over
time. The VA has made ending
homelessness among veterans a pri-
ority, with a particular focus on those
with severe mental illness. Recent
VA efforts to help homeless veterans
and integrate services for those with
severe mental illness may have offset
this trend for these vulnerable pop-
ulations. Further research is needed
to assess the long-term impact of
emerging initiatives, such as medical
home models for homeless veterans
and those with severe mental illness,
on reducing the mortality gap.
Acknowledgments and disclosures
This work was supported by the Health Services
Research and Development Service (IIR 11-
expressed are those of the authors and do not
necessarily represent the views of the VA.
The authors report no competing interests.
Years of potential life lost (YPLL) from all-cause mortality among veterans
(N=575,194), by homelessness and severe mental illness (SMI) statusa
Homeless with SMI (N=18,303)
Homeless without SMI (N=2,780) Not homeless without SMI (N=251,483)
Not homeless with SMI (N=302,628)
aSMI: inpatient or outpatient ICD-9 code for schizophrenia, bipolar disorder, or other psychosis.
Homelessness: outpatient or inpatient service code ICD-9 V60.0 (lack of housing) or clinic stop
code 501 or 528 (homeless mentally ill outreach), 522 or 530 (Housing and Urban Development–
Veterans Affairs Supported Housing initiative), 529 (health care for homeless veterans), 590
(community outreach for homeless veterans), or inpatient homeless stay (37, domiciliary; 28,
homeless compensated work therapy)
The authors are affiliated with the De-
partment of Veterans Affairs Ann Arbor
Health Center for Clinical Management
Research. Send correspondence to Dr.
Kilbourne at North Campus Research
Arbor, MI 48108 (e-mail: amykilbo@med.
umich.edu). Dr. Kilbourne and Tami L.
Mark, Ph.D., are editors of this column.
PSYCHIATRIC SERVICES ' ps.psychiatryonline.org ' July 2013 Vol. 64 No. 7