• eating disorders;
  • IDDM;
  • co-morbidity;
  • meta-analysis;
  • ‘exact’ methods



Firstly to provide a quantitative summary of existing studies on the occurrence of eating disorders (ED) in females with type 1 diabetes (IDDM), with the focus on retinopathy and insulin misuse for the different eating disorders. Secondly to disseminate knowledge about useful statistical tools.

Research Design and Methods:

Data were extracted from the relevant case–control and follow-up studies. Odds ratios (OR) and risk differences (RD) were the main effect sizes analysed. Analyses were based on ‘exact’ methods as many studies are sparse. Data and findings are presented in sufficient detail for re-analysis.


An hypothesis of an increase in Anorexia Nervosa (AN) in IDDM is not supported by existing evidence. Bulimia Nervosa is increased (OR = 2.9 (95%CI: 1.03 to 8.4); pOR = 0.04) in IDDM. Both ED-NOS and subthreshold ED is increased (OR ∼2; pOR < 0.001) in females with IDDM. Co-existing ED in IDDM increases the overall common OR for retinopathy to 4.8 (95%CI: 3.0 to 7.8); pOR < 0.00001, and the overall mean RD is 33% (95%CI: 25% to 42%); pRD < 0.001. Insulin misuse (IM) is increased when ED co-exists with IDDM: OR 12.6 (95%CI: 7.8 to 21.1); pOR < 0.00001, and mean RD is 40% (95%CI: 29% to 50%); pRD < 0.001.


ED-NOS and subthreshold ED seem to be the quantitatively most important EDs in type 1 diabetic females. Mismanagement of diabetes in the form of IM is frequent in eating disordered IDDM probands. Early occurrence of retinopathy and other complications is an increased risk in concurrent cases, as is premature death. The implications of Binge Eating Disorder (BED) and overweight needs to be elucidated for both type 1 and type 2 diabetes. Copyright © 2002 John Wiley & Sons, Ltd and Eating Disorders Association.