Validation of alternative indicators of social support in perinatal outcomes research using quality of the partner relationship.

Aim: To test alternatives to the current research and clinical practice of assuming that married or partnered status is a proxy for positive social support.

Background: Having a partner is assumed to relate to better health status via the intermediary process of social support. However, women's health research indicates that having a partner is not always associated with positive social support.

Design: An exploratory post hoc analysis focused on posttraumatic stress and childbearing was conducted using a large perinatal database from 2005-2009.

Methods: To operationalize partner relationship, four variables were analysed: partner ('yes' or 'no'), intimate partner violence ('yes' or 'no'), the combination of those two factors, and the woman's appraisal of the quality of her partner relationship via a single item. Construct validity of these four alternative variables was assessed in relation to appraisal of the partner's social support in labour and the postpartum using linear regression standardized betas and adjusted R-squares. Predictive validity was assessed using unadjusted and adjusted linear regression modelling.

Results: Four groups were compared. Married, abused women differed most from married, not abused women in relation to the social support, and depression outcomes used for validity checks. The variable representing the women's appraisals of their partner relationships accounts for the most variance in predicting depression scores.

Conclusions: Our results support the validity of operationalizing the impact of the partner relationship on outcomes using a combination of partnered status and abuse status or using a subjective rating of quality of the partner relationship.