Dr. Rosenheck is affiliated with the Department of Psychiatry, Yale University School of Medicine, New Haven, Connecticut, and with the Northeast Program Evaluation Center (182), Veterans Affairs Connecticut Healthcare System, 950 Campbell Avenue, West Haven, CT 06516 (e-mail: robert.rosenheck@yale.edu). Dr. Swartz is with Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, Durham, North Carolina. Dr. Lieberman is with the Department of Psychiatry, Columbia University, New York.

Much of what is known about the effectiveness of drugs used in clinical practice is based on short-term randomized trials that involve carefully selected patients with few comorbid conditions who are treated under highly controlled circumstances, often with the goal of winning approval from the Food and Drug Administration. To generate information that can more appropriately guide real-world clinical practice, the National Institutes of Health has fostered the development of large networks of cardiologists and oncologists to conduct practical clinical trials (PCTs). These trials are characterized by large samples, prolonged treatment, a focus on clinical treatment choices of practical relevance, and diverse and representative samples. In 1999 the National Institute of Mental Health established three networks to conduct PCTs in the areas of bipolar disorder, major depressive disorder, schizophrenia, and Alzheimer's disease. The studies of schizophrenia and Alzheimer' disease evaluated treatment with second-generation antipsychotics and were called the Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE).

PCTs eventually yield longitudinal data on drug effectiveness. In addition, data from PCTs may also be used in clinical epidemiologic studies because they gather extensive data on large samples of patients with few exclusion criteria. This special section presents four studies from CATIE that illustrate use of data collected in a PCT to understand patterns of polypharmacy, clinical and comorbid factors (substance abuse and general medical illness) that effect community functioning, and the ripple effects of schizophrenia symptoms on family life.

In the first article Chakos and colleagues explore patterns and predictors of polypharmacy. At the time of entry into the CATIE trial, 26 percent of study entrants were not taking any antipsychotics and those who were taking medication were taking an average of two different agents. However, the most common forms of polypharmacy involved not the use of multiple antipsychotics but the use of concomitant antidepressant and antianxiety agents. Use of these adjunctive agents was more strongly correlated with general psychiatric symptoms, such as anxiety and depression, than with schizophrenia symptoms.

The next two articles focus on the co-occurrence in schizophrenia of substance use and general medical disorders and their association with deficits in community functioning. Swartz and colleagues used a rich combination of biochemical and psychometric measures to identify high levels of substance use alone (23 percent) and substance use disorder (37 percent). Both substance use and substance use disorder were associated with higher levels of functioning. Chwastiak and colleagues used a similar approach and found no significant correlation between severity of schizophrenia symptoms and the number of comorbid general medical illnesses and little evidence that medical comorbidity impairs functioning beyond that attributable to schizophrenia symptoms, depression, and neurocognitive deficits.

In the final article, Perlick and colleagues consider the impact of schizophrenia on patients' families. They identified four independent components of family burden, each correlated with a distinctive configuration of patient and caregiver sociodemographic and clinical characteristics. Positive and negative symptoms of schizophrenia were found to be the strongest correlates of family burden, albeit with different components of burden—positive symptoms correlated with "problem behavior," negative symptoms with "perceived functional impairment," and both equally with "resource demands and routine disruption."

Although recruitment is especially challenging in the context of PCTs, the four studies presented here clearly demonstrate that recruitment for PCTs can successfully engage people with severe mental illness and that even baseline data can have important research value.