Abstract
BACKGROUND:
Schizophrenia is a chronic, debilitating disease that affects approximately 1% of the U.S. population and has disproportionately high costs. Several factors, including age, gender, insurance status, and comorbid conditions, have been hypothesized to be associated with schizophrenia-related costs.
OBJECTIVES:
To identify demographic and clinical characteristics of community-dwelling schizophrenia patients experiencing high schizophrenia-related direct medical costs.
METHODS:
Community-dwelling patients with a diagnosis for schizophrenic disorder (ICD-9-CM code 295) and other nonorganic psychoses (ICD-9-CM code 298) were identified from the 2005-2008 Medical Expenditure Panel Survey (MEPS). Schizophrenia-related direct medical costs were calculated for (a) inpatient hospitalizations; (b) prescription medications; and (c) outpatient, office-based physician, emergency room, and home health care visits. Using Andersen’s Behavorial Model of Health Services Use and the literature, factors that could potentially affect schizophrenia-related direct medical costs were identified. Based on the distribution of their mean annual costs, patients were classified into high- and low-cost groups. Logistic regression was used to determine the likelihood of high-cost group membership based on age, sex, race, insurance status, marital status, region of residence, family income as a percentage of poverty line, number of medical comorbidities, number of mental health-related comorbidities, patient-perceived general health status, patient-perceived mental health status, and year of inclusion in MEPS. In addition, a generalized linear model (GLM) regression (gamma distribution with a log-link function) was used to evaluate the relationships between the independent variables and total schizophrenia-related direct medical costs as a continuous variable.
RESULTS:
From the MEPS database, we identified 317 patients with schizophrenia who represented 2.75 million noninstitutionalized, community-dwelling schizophrenia patients in the United States between 2005 and 2008. The logistic regression procedure showed that older patients (OR=0.933, 95% CI=0.902-0.966) and patients with a spouse (OR=0.150, 95% CI=0.041-0.555) were less likely to be in the high-cost group, while those who reported having “poor” perceived general health status (OR=15.548, 95% CI=1.278-189.127) were more likely to be in the high-cost group. The GLM regression procedure showed that younger patients (compared with older patients), African Americans (compared with Caucasions), patients with private insurance (compared with the uninsured), and those living in the northeastern United States (compared with those living in the southern United States) had higher schizophrenia-related direct medical costs.
CONCLUSIONS:
Identification of factors associated with a high-cost population may help decision makers in managed care, government, and other organizations allocate resources more efficiently and health care providers manage patients more effectively through assignment of these patients to case managers and appropriate monitoring and treatment.