Effect of the 2010 Chilean Great Earthquake on Post-Traumatic Stress: An Analysis Using Multivariate Matching and Sensitivity Analysis

• Activity Number: 304
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jose Zubizarreta*+ and Magdalena Cerdá and Paul R Rosenbaum
• Companies: The Wharton School and Columbia University and University of Pennsylvania
• Keywords: Causal inference ; Earthquakes ; Posttraumatic Stress Disorders ; Observational studies ; Optimal matching ; Sensitivity analysis

Abstract

In 2010, an earthquake of magnitude 8.8 hit Chile, causing devastating certain parts of the country. With prospective longitudinal data, free of recall bias, we used new optimal matching methods to pair respondents who were similar prior to the earthquake and who had vastly different exposures to the earthquake, as measured objectively by peak ground acceleration. Posttraumatic stress symptoms (PTS) were measured using the Davidson Trauma Scale. PTS was dramatically elevated among some residents of strongly shaken areas of Chile when compared to similar individuals in largely untouched parts of the country. This comparison was less sensitive to unmeasured biases than studies linking heavy smoking with lung cancer, hence, less sensitive than some of the least sensitive studies ever reported in epidemiology. Moreover, the earthquake effect on stress was not uniform, but rather affected some severely exposed individuals far more than others with similar exposure. Our presentation illustrates recent statistical methodology aimed at more effectively removing measured biases using matching and reducing sensitivity to unmeasured biases through design choices guided by design sensitivity.