The major objective of this project is to assess the potential long-term mental health consequences, including symptoms of generalized anxiety, depression, substance use/abuse, and Post-traumatic Stress Disorder, associated with qualitative and quantitative aspects of exposure to two distinct types of natural disaster (hurricane and earthquake) occurring in 1989 in South Carolina and California, respectively. Approximately eight million people reside in federally-declared disaster areas in these two states, and an estimated two million U.S. residents are affected by natural disasters each year. This project proposes to test a multivariate risk factor model for long-term disaster-related mental health problems of adults and adolescents. The hypothesized model identifies five constructs that are primary risk factors: 1) Predisaster Characteristics and History, 2) Primary and Secondary Disaster Exposure, 3) Cognitive Appraisal of Disaster, 4) Attempts to Adjust to and Overcome Disaster Losses, and 5) Other Disaster Recovery Period Events. The main project study will use random digit dialing (RDD) telephone survey methodology to interview a representative household sample of adult (n=4,000) and adolescent (n=1,600) residents of high disaster impact sites in California and South Carolina and of demographically matched no impact sites. Respondents will be assessed at two and three years post-disaster. Before the main study is conducted, a methodological pilot study will be conducted using an area probability sample of 800 households in high disaster impact sites in South Carolina and California. A randomly selected adult and adolescent (if any) in each household will be assessed in person, and the resulting data will be used to examine possible limitations of the telephone sampling and to assist in constructing short form scales for use in the main telephone survey. A variety of multivariate data analytic procedures will be used to test hypothesized relationships among variables of interest. Structural modeling procedures will be used to test causal hypotheses using the longitudinal data.
