David A Katerndahl1. 1. Department of Family & Community Medicine, The University of Texas Health Science Center at San Antonio.
Abstract
OBJECTIVE: The purpose of this study was to determine the proportion of variance in mental health outcomes accounted for by census tract in community-dwelling adults with and without panic attacks. METHOD: This is a secondary analysis of a population-based study (conducted from August 1989 through December 1991) of subjects with and without panic attacks (DSM-III-R criteria) from 18 census tracts in San Antonio, Texas. All subjects completed measures of symptomatology, health care utilization, substance use, and quality of life. Subjects with panic attacks completed measures of panic-related disability and health care utilization, as well as measures of sense of control and secondary mental disorders. Hierarchical modeling was used to estimate the proportion of the variance of each outcome accounted for by census tract. RESULTS: Census tract accounted for less than 4% of the variance in psychiatric symptomatology and quality of life. However, census tract contributed to the proportional variance in panic-related outcomes, accounting for 13% of the variance of mental health utilization for panic symptoms when compared with individual-level variance. CONCLUSIONS: The use of more homogeneous levels such as block group could increase the measured multilevel effects found in this study. Studies of disease-specific mental health outcomes in multiple neighborhoods or clinics should consider whether multilevel effects are present.
OBJECTIVE: The purpose of this study was to determine the proportion of variance in mental health outcomes accounted for by census tract in community-dwelling adults with and without panic attacks. METHOD: This is a secondary analysis of a population-based study (conducted from August 1989 through December 1991) of subjects with and without panic attacks (DSM-III-R criteria) from 18 census tracts in San Antonio, Texas. All subjects completed measures of symptomatology, health care utilization, substance use, and quality of life. Subjects with panic attacks completed measures of panic-related disability and health care utilization, as well as measures of sense of control and secondary mental disorders. Hierarchical modeling was used to estimate the proportion of the variance of each outcome accounted for by census tract. RESULTS: Census tract accounted for less than 4% of the variance in psychiatric symptomatology and quality of life. However, census tract contributed to the proportional variance in panic-related outcomes, accounting for 13% of the variance of mental health utilization for panic symptoms when compared with individual-level variance. CONCLUSIONS: The use of more homogeneous levels such as block group could increase the measured multilevel effects found in this study. Studies of disease-specific mental health outcomes in multiple neighborhoods or clinics should consider whether multilevel effects are present.