Predictors of Self-Rated Health in Disaster Prone Communities: Measuring Individual Disaster Resilience in Louisiana

Background and Purpose: A great deal of disaster research has focused on community disaster resilience, not taking into account the individual and his or her resilience within the community. Disaster resilience scholars have studied the impact of individual and contextual indicators on the ability of individuals and communities to adjust to, moderate the effects of, and cope with disaster disturbances (Cutter et al. 2008, Cutter, Burton & Emrich 2010). More specifically, attention has shifted to consider both the individual and the contextual determinants of good health outcomes for individuals when faced with disasters. This research is of importance to a state like Louisiana. Recent disaster events have become more common and more severe, mainly due to South Louisiana subsiding, coastal wetlands eroding and warmer oceans causing rising sea levels (GOSHEP, 2008). From 1953 to 2013 Louisiana experienced 60 federally declared disasters ranking it 6^th in the nation (FEMA, 2013).

This study was conceptually guided by a modified version of Andersen’s Behavioral Model of Health Utilization (Andersen, Rice & Kominiski, 2007). The model was modified with the disaster resilience of place model (DROP) (Cutter et a., 2008) and the hazards-of-place model of vulnerability (Cutter, Boruff & Shirley, 2003). According to this model, equitable disaster mitigation would occur when demographic (age, gender) and need variables (contextual disaster risk and population health) accounted for most of the variance in individual health outcomes, while inequitable disaster mitigation would occur when social structure (contextual social resilience, community capital, individual social vulnerability), and enabling resources (institutional resilience, contextual and individual economic resilience, infrastructure resilience) determined health outcomes for individuals exposed to disaster events.

Methods: The study was a multilevel, repeated cross-sectional design with a three-level, nested structure (individuals nested within Louisiana parish cohorts and Louisiana parish cohorts nested within parishes). Secondary data was gathered from the Behavioral Risk Factor Surveillance System for 7 cohorts from 2004-2010. Parish-level data came from 12 additional sources. A representative sample of 34,685 individuals was used to test the model. Because of the nature of the outcome variable (self-rated health, ordinal level), a nonlinear analysis was done using an ordered multinomial response model with a log-link function and empirical Bayes Markov chain Monte Carlo (MCMC) estimation (Raudenbush & Bryk, 2002). 

Results: Overall, 20% of the sample reported fair to poor health in 2004. By 2010, this number increased to 25%. The results of the model testing clearly indicated inequitable disaster mitigation, with the social resilience, individual social vulnerability, individual economic resilience and contextual economic resilience indicators explaining most variance in the self-rated health of parish residents. Disaster risk indicators did play a role in explaining health outcomes, but not as strong as the variables mentioned.

Conclusions and Implications: The need and value of multilevel modeling in disaster research has been underlined. The findings from this study can advance disaster research, and educate social work decision makers on the importance of building resilience not only within communities but within individuals as well, and raising the awareness of the problem of inequitable disaster mitigation.