| Although earthquakes have killed over a million people and have injured many more in this century, they have received little attention from epidemiologists. A case-control study examined how the physical environment and behaviors of County of Santa Cruz (CSC), California residents contributed to their risk of being physically injured in the 1989 Loma Prieta earthquake. Cases either died, visited a CSC hospital or were helicoptered to a hospital outside CSC because of a mainshock injury. A stratified, random sample of controls was selected using a random digit dial of CSC residential phone numbers. Non-injured controls were frequency matched to cases on general area of residence. To examine selection factors for seeking medical care, cases were compared to untreated injured controls. Subjects or proxies were interviewed. Bivariate analyses stratified by residential area were performed, followed by unconditional logistic regression. Injuries were characterized as present or not, and by overall level of severity.;Most cases incurred mild injuries (82%). For all injuries combined, the physical environment posed the greatest hazards. Residence-adjusted odds ratios (OR) and 95% confidence intervals (CI) for being trapped, in a building, or in an earthquake-damaged place were 14.35 (8.37, 24.60), 2.91 (1.81, 4.67), and 3.64 (2.32, 5.71), respectively. Rescuing people elevated risks while holding onto something was protective. Exiting risks varied by residential area, and were highest for North Mountain dwellers (OR (95% CI) = 5.64 (2.07, 15.41)). In multivariate models, occupying a building became insignificant when location damaged was included, suggesting exposures operated sequentially in time. Thus, models were developed for the start (P1) and duration (P2) of the shaking. Being in a building regained significance in P1 models, while P2 models revealed negative interactions between exiting and: (1) initial location damaged, and (2) rescuing people. The risks from exposures generally increased with temporal proximity to injury.;Among building occupants, the greatest risks at the earthquake's start were to people inside concrete-exterior buildings (OR (95% CI) = 2.45 (1.30, 4.62)) or kitchens, dining rooms, shopping areas, or factory floors (OR (95% CI) = 2.45 (1.61, 3.74)). During the shaking, the highest risks were associated with damage to structural components (e.g., collapsing walls, ceilings and floors, OR (95% CI) = 17.30 (1.61, 186.23)) and contents (e.g., spilling chemicals, OR (95% CI) = 3.59 (1.80, 7.18)) of the room initially occupied. Standing under a doorway or holding on reduced risks by 61% and 46%, respectively. Either rescuing people or exiting was hazardous, but doing both lowered their combined expected risk. Interactions between residence and: (1) exiting, and (2) overturning furniture/machinery raised risks for North Mountain dwellers.;Although more seriously injured survivors obtained needed treatment, self-perception of injury severity appeared to govern care-seeking behavior among the most mildly injured.;The preponderance of minor injuries suggests seismic building codes are effective in a moderate earthquake. Nonetheless, reducing building damage and responding appropriately in future mainshocks should prevent injuries. The effect of exiting may depend on the degree of earthquake-caused damage to one's-initial surroundings. Emergency response plans should be developed for buildings based on their earthquake injury potential. |
