Introduction/objective: Bowlby (1984) regarded attachment as a model of psychological vulnerability to depression. Since then, a large number of studies have considered vulnerability to depression in light of the idea of attachment style. Attachment styles correspond to two dimensions observed in relationships (anxiety and avoidance) evoking ideally the internal operating models of self and other respectively, as first described by Bowlby (1984). Two types of adult attachment styles are evaluated in our study: romantic attachment (Hazan and Shaver, 1987) and interpersonal attachment (Bartholomew and Horowitz, 1991). The existing literature indicates that depression is associated with the insecure attachment styles, in both romantic an interpersonal relationships. Nevertheless, a question remains concerning the nature of the link between attachment style and depression: are attachment styles stable and independent of the depression or are they modified as the depression evolves? The aim of the present study was to investigate the relationships between attachment and depression in adult women hospitalized for depression; following up the evolution in their romantic and interpersonal attachment styles from the beginning to the end of their hospitalization.

Methodology: The study population consisted of 50 women hospitalized for an episode of major depression (Axis I, DSM IV). Individuals exhibiting bipolar disorders and other pathologies linked to depression were not included in the population. Sixty-eight percent of the depressed women in our population had previously experienced depressive episodes and 42% of them also exhibited a personality disorder (Axis II, DSM IV). The clinical group participated in two psychological investigations, one at the beginning (T1) and one at the end of the hospitalization (T2), including each time a clinical interview during which the depression as well as the romantic (ECR, 1998) and interpersonal (RQ, 1991) attachment styles were evaluated.

Study results: Our findings showed that depression is positively correlated with the "avoidant" dimension of the romantic attachment style and negatively with the "secure" dimension of the interpersonal attachment style. Between the beginning and the end of hospitalization, only the "secure" and "fearful" dimensions of the interpersonal attachment styles were modified, whereas the styles of romantic attachment remained stable. In terms of the links between attachment styles and the evolution of depression during hospitalization (between T1 and T2), we noted that an increase in security of attachment was correlated with a decrease in the depressive symptomatology. Conversely, the other romantic and interpersonal attachment styles were not linked with the evolution of the depressive symptomatology.

Discussion/conclusion: Our results confirm that romantic and interpersonal styles of attachment constitute factors of vulnerability to depression. But more importantly, these findings open up new perspectives in terms of the nature of the relationships between attachment styles and depression. They provide matter for discussion concerning the stability or the change in romantic and interpersonal attachment styles. Indeed, we have revealed the stable and independent nature of romantic attachment styles in relation to depressive symptomatology. On the contrary, security in the interpersonal attachment style was shown to be a factor of change, associated with the evolution of the depressive symptomatology in progress. In the quest to take combined account of romantic and interpersonal attachment styles and their links with the evolution of depression, the present study results in a new understanding of depression, viewed from the perspective of the model of attachment in adults.