Eating alone, parents' marital status, and use of radio and girls' magazines were risk factors for eating disorders

The study’s findings are strengthened by its design. Initial levels of eating pathology were controlled for at baseline assessment, ensuring that increases in EDs were not a consequence of initial, marked disturbances. The control of confounders (eg, age at baseline assessment and socioeconomic status) adds confidence to the correlations. The prospective nature of this study and its use of logistic modelling ensure that potential interaction among factors could be investigated. However, as with all correlational research, findings do not necessarily connote causation and are vulnerable to other explanations. Solitary eating, for example, may reflect symptoms (eg, negative affect) rather than cause. Additionally, the length of follow up may not truly capture the effect or interactional quality of these risk factors. Clinicians can use the risk factors identified in this study 4 to maximise the effect of