Early attachment quality moderates eating disorder risk among adolescent girls

Abstract

Objective: There is growing evidence that children’s early relational environment has lasting implications for physical and mental health. In this paper, we test whether attachment insecurity in early childhood is associated with increased responsivity to risk factors for eating disorders (EDs; e.g. pubertal weight gain, maternal negative affect) during adolescence.

Design: Hypotheses were tested with longitudinal data from 447 girls (final mean age = 15.1 years) over a 12-year period. Tests of direct effects, moderation and moderated mediation were conducted using nested structural equation models and bootstrapped estimates of direct and indirect effects.

Results: Early attachment quality was not directly associated with disordered eating attitudes and behaviours (DEABs), but did moderate relations between adolescent ED risk factors and DEABs. Specifically, among girls with an insecure attachment history, higher BMI at age 15 directly predicted more DEABs, while maternal negative affect and pubertal weight gain indirectly predicted DEABs via greater preoccupation with parental relationships. These same direct and indirect paths did not emerge among adolescent girls with a secure attachment history.

Conclusion: Results delineate one way early attachment quality may contribute to EDs among some adolescent girls, and support recent efforts to incorporate relational components into obesity and ED prevention programmes.

Keywords:

• eating disorder
• attachment
• adolescence
• BMI
• obesity
• NICHD

Acknowledgements

The NICHD Study of Early Child Care and Youth Development was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) through a cooperative agreement (U10) that calls for a scientific collaboration between NICHD staff and participating investigators. We gratefully acknowledge the data-set provided by this study that is made available to researchers, and express our appreciation to NICHD and study investigators, personnel and participants.

Notes

1. Because delta variables can cause problems of interpretation because of multicollinearity, models were run with just BMI 15 included and with BMI 15 and BMI 9 included. The pattern of results was consistent across models.