proximate physical causes of energy exchange (e.g., contact points in vehicle interiors). As the injury field matures, research has to move away from a reliance on descriptive case studies and toward more rigorous analytical methodologies.
Randomized controlled trials (RCTs) are considered the gold standard in evaluating the effectiveness of interventions. Randomization maximizes the comparability of intervention and nonintervention groups with respect to factors that may influence the outcome of the trial. Through randomization to intervention and control groups, RCTs are designed to adjust for a variety of potential confounding factors. In addition, the investigator can control the conditions under which the trial is conducted to ensure that standardized procedures are followed. Although RCTs are expensive, time-consuming, and not always feasible, they are the standard in other fields of research and should be in injury research as well.
Cohort studies are used to compare the rate of injury of individuals exposed to a suspected risk factor with the rate of injury among individuals unexposed to the risk factor (Lilienfeld and Stolley, 1994). One major difficulty is that large numbers of individuals may have to be followed in order to capture a sufficient number of injuries for meaningful analysis of risk. The other major difficulty is the need to follow individuals over time. Together, these often make cohort studies expensive and logistically problematic. Increased attention should be devoted to the development of multi-institutional cooperation to maximize generalizability and spread the burden of subject follow-up. Better ways to ascertain exposure are needed for use in cohort studies of injury. Cost-efficient data collection methods that maintain data quality also are required.
Case-control studies compare injured individuals (cases) with those who are not injured (controls) in order to determine what characteristics are associated with the injury (e.g., lack of safety belt use in severe motor vehicle crashes). Case-control studies are especially useful in the identification of risk factors for rare events. To conduct studies over reasonable periods of time, it is often necessary for investigators at multiple sites to collaborate in order to accrue sufficient numbers of injuries for study. The limitations of case-control studies include vulnerability to recall bias, the possibility of confounding by unidentified factors, and the inability to definitively establish causality.
Recent variants of the case-control design have been used in injury research, including studies that compare persons with specified types of injuries to persons with other types of injuries and studies in which the site of death or injury is compared with a site where an injurious event did not occur (Wright and Robertson, 1976). Studies of the efficacy of helmets have compared bicyclists with head injuries to those with injuries to other regions of the body (McDermott et al., 1993; Thompson et al., 1996a). Another variant is a case-crossover study, which uses study participants as their own controls. This type of study compares behavior during the event (e.g., a crash or injury) with that preceding the event to determine whether participants were exposed to the factor of interest. For example, a case-crossover study was used to determine the risk associated with the