Self-reporting of outcomes can introduce reporting bias. Reporting bias, which occurs when the group being studied reports more frequently what it remembers than a comparison group, can potentially lead to an overestimation of the incidence or prevalence of symptoms or diagnoses in the exposed populations. Self-reporting of outcomes based solely on symptoms might also introduce misclassification bias, in which there are errors in how symptoms are classified into outcomes.
Low participation rates, which can introduce selection bias, can severely limit the ability to generalize study results because the study population may not be representative of the larger population to which the results are meant to be generalized. A related issue is the use of inappropriate controls, such as comparison of military populations with civilian populations; military personnel may be healthier than the general population, so the two populations may be noncomparable. That is referred to as the healthy-warrior effect; there may have been nonrandom assignment of those selected and not selected for participation in the military. It is possible to measure the potential for such biases and to adjust for them in the analysis.
Another important limitation of some of the cohort studies is that they lack unexposed control groups. An unexposed group is a necessary component of a well-designed cohort study because it permits comparisons of rates of disease between exposed and unexposed populations and understanding of how an exposure affects the incidence of an outcome.
Some of the studies discussed below are registries of participants who presented for care. These studies are not intended to be representative of the symptoms and diagnoses of an entire population.
Although this is not necessarily a limitation, many studies discussed below were not designed with the committee’s research question in mind. It was therefore difficult to use their findings to assess the broader question of the relation of long-term health outcomes to TBI.
This chapter has sections on military cohort studies, population-based studies, other cohort studies, and sports-related studies. For each major cohort study, the methods for selecting the study population, the outcomes assessed, and the general findings are discussed. The committee was most interested in studies of long-term health outcomes related to TBI in military and veteran populations, so this group of studies is given primary consideration below.
Studies of TBI have been conducted in nearly all the major conflicts of the 20th century, including World Wars I and II, the Korean War, and the Vietnam War; many of the studies evaluated seizure as the outcome of interest. Meirowsky (1982) noted that studying military populations “offers the advantage of similarity in age and general health of the subjects at the time of injury and the relative ease with which they can be followed in subsequent years.” The committee paid particular attention to studies that assessed TBI in military populations because these were generally long-term prospective assessments of the population of interest.