Risk Factors for Eating Disorders
The causation of eating disorders is widely thought to be “multifactorial,” a term so broad as to render it useless without further qualification. Myriad individual risk factors have been studied. As in other areas of research, there are definite fashions here. For example, during the 1960s and 1970s, an “anorexogenic family environment” was thought to be crucial for the development of anorexia nervosa, and during the 1980s and 1990s, childhood trauma, in particular, childhood sexual abuse, was promoted as causally important, mainly for bulimia nervosa. With the advent of new biotechnologies (molecular biology, brain scanning), we are seeing a revival of the interest in biological factors. Several important classes of risk factors are discussed in detail elsewhere in this book (e.g., genetic factors—see Chapters 42 and 43; sociocultural risk factors—see Chapters 45 and 47; personality factors—see Chapter 36; family factors—see Chapter 38). The focus of this chapter is threefold: on some of the methodological issues affecting this kind of research; on research that attempts to integrate our knowledge about the relative contribution of different risk factors; and on some of the newer lines of investigation in this area.
In order to appreciate the methodological difficulties in studying risk factors, it is useful to remember the definition of a “risk factor” a factor that is associated with a disorder and may support a causal connection. Evidence supporting a causal link between a risk factor and a disorder is provided by (1) the factor preceding the disorder being studied; (2) the repeated appearance of the same risk factor in multiple risk factor studies; (3) the risk factor being associated with one disorder only; and, most importantly, (4) the finding that an experimental intervention that eliminates the risk factor also eliminates the disorder. Many studies of eating disorders, especially those using cross-sectional designs, violate the first point regarding temporal precedence and, while purportedly being interested in identifying risk factors, do not attempt to identify what occurred first—the putative