Risk Factors Associated with Moderate and Serious Injuries Attributable to the 1994 Northridge Earthquake, Los Angeles, California

PURPOSE: The purpose of this study was to use emergency department data to estimate levels of morbidity and risk factors due to earthquake-related mechanisms of injury subsequent to an urban night-time earthquake. METHODS: Data were abstracted from 4190 medical records for the month of January, 1994. Injuries attributable to the earthquake were identified through emergency department and medical records. These injuries were: (a) categorized by mechanism of injury; (b) assigned an injury severity score; and (c) linked to structural and geologic data. Proportional polytomous and dichotomous logistic regression were used to estimate risk of more severe injury associated with demographic characteristics, injury characteristics, structural characteristics, and geologic factors. RESULTS: More severe earthquake-related injuries (serious versus moderate and moderate versus minor) were statistically significantly associated with patient age (greater-or-equal, slanted 60 years old), upper extremities, falling, multi-family structures, pre-1960 housing, and the 50th percentile of Peak Ground Acceleration, after adjusting for all other available demographic, injury, structural, and geologic characteristics. CONCLUSIONS: The current recommendation of ‘duck, cover, and hold’ might not be optimal during a nighttime earthquake, particularly if individuals are in the padded environment of the bed. Actions such as reaching for or catching objects, bracing, or holding onto perceived stable objects may increase risk for more serious injury. Alternate responses include assuming a tucked position (as in airline crashes) or staying in bed for non-ambulating people. Structural damage and structure size were not associated with more serious injuries, but structure use and age were, leading the authors to suspect that unmeasured socioeconomic factors might impact risk factors. The importance of including population demographic characteristics in hazard modeling is emphasized.