Background Eating disorders are disabling, unpredictable, and difficult to
treat. We did a prospective 5-year investigation of a representative sample
of patients with eating disorders. Our aim was to identify predictors of o
utcome and to assess effects of available treatments.
Methods We prospectively investigated 95 patients with anorexia nervosa, 88
with bulimia nervosa, and 37 with eating disorders not otherwise specified
(EDNOS), who sought treatment in Adelaide, South Australia. We divided pat
ients into those who had, and had not, received treatment in specialist uni
ts and reached a safe body weight. Individuals were then further classified
dependent on intensity of any treatment received. We assessed clinical sym
ptoms, body-related attitudes, and psychosocial function.
Findings 216 (98%) patients were available for follow-up after 5 years. Thr
ee patients with anorexia nervosa and two with EDNOS died. 65 (74%) bulimic
, 29 (78%) EDNOS, and 53 (56%) anorexic patients had no diagnosable eating
disorder. A small proportion of patients in every group had poor Morgan-Rus
sell-Hayward scores at outcome. Final outcome was predicted by extent and i
ntensity. but not duration, of initial symptoms in patients with anorexia n
ervosa, and by initial body-related attitudes and impaired psychosocial fun
ctioning in bulimia patients. We were unable to predict EDNOS outcome. Trea
tment did not affect outcome for any group.
Interpretation Deaths in the study confirm the serious nature of eating dis
orders. However. our results suggest that the efficacy of existing interven
tions is questionable.