For more than a decade researchers have documented premature mortality among people with serious mental illness. Recent work has focused on causes of early death and preventable risk factors. The lead article this month reports findings from a comparison of death records over seven years for individuals with and without serious mental illness living in the same community. Elizabeth E. Piatt, Ph.D., and colleagues found that overall, YPLL (years of potential life lost) for decedents with serious mental illness was 14.5, compared with 10.3 for those without serious mental illness. Differences in causes of death did not explain the difference in YPLL. When analyses adjusted for sociodemographic factors, differences in YPLL were found for all causes and were highest for accidents (13.1 years), followed by suicide (9.7 years), cancer and septicemia (5.0 years each), and liver disease (4.8 years) (page 663). In addition to years of life lost in this population, are there disparities in end-of-life care? When Linda Ganzini, M.D., M.P.H., and colleagues examined indicators of palliative care in medical records of veterans who died of cancer, they found that those with schizophrenia received comparable or better care than those without this disorder (page 725). To effectively treat and prevent disease in any population, physicians rely on patients' reports of their health status, and some have questioned whether symptoms bias psychiatric patients' reports. In a study of 151 African-American inpatients with schizophrenia, Arthur L. Whaley, Ph.D., Dr.P.H., used statistical modeling to construct a map of the effects of symptoms, psychological factors, and demographic variables on global self-reported health. All the variables had some effect, but the best model was one in which demographic factors were removed, and it showed that the psychological factors of self-esteem and lack of trust had a more significant effect on self-reports than psychiatric symptoms (page 669).

In this month's Open Forum, Robert E. Drake, M.D., Ph.D., and colleagues describe a vision of applying information technology in the form of comprehensive decision support systems to overcome problems that impede adoption of recovery ideology and evidence-based practices (page 714). In Taking Issue, Alexander S. Young, M.D., M.S.H.S., calls their vision "compelling" and urges managers to find the will and dedication to implement informatics tools in clinical settings in order to remake their organizations (page 643).

Two articles report on predictors of treatment dropout. A key principle of assertive community treatment (ACT) is that no time limits are imposed. However, many clients terminate ACT services, and their reasons have not been studied empirically. Somaia Mohamed, Ph.D., and colleagues compared three groups of veterans enrolled in a program modeled on ACT: those who terminated in less than one year (16%), those who terminated later (one to three years) (26%), and those who remained in the program after three years. Receipt of less intensive services in the first six months was a stronger predictor of termination than patient characteristics or clinical change (page 675). In a Seattle study of insured members of a managed health care system, Gregory E. Simon, M.D., M.P.H., and Evette J. Ludman, Ph.D., found high rates of dropout among depressed patients newly scheduled for outpatient psychotherapy—fewer than half (42%) had three or more visits. Referral to a contracted, fee-for-service therapist in an external network rather than a salaried provider within the group-model clinic was a strong predictor of dropout before the first appointment (page 684).

Increasing the effectiveness of addiction interventions, especially for groups with poor outcomes, is the focus of two studies. Substance abuse is a salient factor in homelessness, and Jonathan R. Buchholz, M.D., and colleagues examined whether housing status over one year affected outcomes of substance abuse treatment among veterans, a group with high rates of homelessness, who are often difficult to engage in treatment. Treatment retention was surprisingly good among consistently homeless veterans; however, measures of ongoing drug and psychiatric problems indicated poorer outcomes for this group, and homelessness was associated with higher use of inpatient and emergency care (page 698). Individuals with severe psychiatric illness and cocaine addiction are another difficult-to-treat group. Marc I. Rosen, M.D., and colleagues conducted a randomized trial to examine the effectiveness of ATM (advisor-teller money management), an intervention to break the link between a patient's money and cued associations with drug use. Over one year ATM clients had significantly more negative toxicology screens for cocaine and higher clinician-rated abstinence from illicit drugs (page 707).