Modelling risk factors for injuries from dog bites in Greece: a case-only design and analysis

We conducted a study using a newly developed dataset based on Emergency Departments records of a network of hospitals from Greece on injuries from dog bites. Our goal is three-fold: (a) to investigate if surrogate factors of leisure time are associated with increased risk of injury from bites; (b) to address recently reported contradictory results on putative association of lunar periods and injuries from dog bites; and (c) to offer a general methodology for addressing similar case-only designs with combined factors of which some can exhibit cyclical patterns. To address these goals, we used a case-only design of our dataset, and conducted an analysis where we controlled simultaneously for weekday/weekend effects, season of year (winter, spring/fall, summer), and lunar periods, because any one of these factors can contribute to the degree of exposure to injuries from dog bites. We found that increased risk of injury from bites was associated with weekends versus weekdays (RR = 1.19, 95% CI: 1.10-1.29), summer versus winter (RR = 1.24, 95% Cl: 1.11-1.39), and fall or spring versus winter (RR = 1.31, 95% CI: 1.19-1.45). The results support the hypothesis that longer leisure time at these levels of factors does increase the risk of having a bite injury. Moreover, after controlling for these factors, risk of bite injury was not associated with moon periods, thereby also helping settle a longstanding argument. (C) 2002 Elsevier Science Ltd. All rights reserved.