Superiority of the case-crossover compared to the case control design for simultaneous examination of agent and host risk factors for dog bites

Background/Purpose: There are two possible types of case control studies on dog bites depending on the cases chosen: 1) Those in which cases are human dog bite victims (hosts) and, 2) those in which cases are dogs (the agents) that have bitten a human. They investigate, respectively, risk factors for being bitten by a dog and risk factors for a dog biting. Neither study type can, by virtue of design, address the other's research question. The goal of this paper is to demonstrate the utility of the case-crossover study as a means of efficiently studying both agent and host risk factors for dog bites and more generally, injuries arising when both agent and host are autonomous actors.

Approach/Methods: We discuss this feature of case-crossover studies using examples of case control studies from the dog bite literature. Emphasis is placed on demonstrating how a case-cross over study can be used to simultaneously answer both agent- and host-related research questions using one data set. We also discuss important methodological considerations and generalize this application to other types of injuries.

Results: Immediate determinants of dog bites involve both agent and host. As the case-crossover design investigates immediate determinants of dog bites, it can be used to simultaneously investigate both agent and host risk factors.

Conclusions: A case-crossover study is an efficient means of simultaneously investigating both agent and host risk factors for dog bites, and can yield important etiological insights into injuries caused by interaction between autonomously acting agents and hosts.