Development of a health care provider survey for domestic violence: psychometric properties.

ABSTRACT

BACKGROUND: Despite rapid proliferation of descriptive studies of health care providers (HCPs) and protocols for identification and management of domestic violence (DV), few reliable instruments exist for assessing HCPs' attitudes, beliefs, and behaviors regarding this practice. This study describes the development and psychometric properties of a measure of attitudes, beliefs, and self-reported behaviors related to the identification and management of DV. METHODS: We used a multiphase study design to develop items across eight content domains. We administered an initial pool of 104 items to a pilot sample of 129 primary care providers (physicians, physician assistants, nurse practitioners, and medical assistants) in a large, urban health maintenance organization. Descriptive statistics, principal components, and reliability analyses were performed on each of the eight content domains. The analyses guided the deletion of items and development of additional items, yielding a 56-item pool. The items were then administered and re-analyzed with an independent sample of 246 HCPs. RESULTS: Six separate and reliable domains were identified Perceived Self-Efficacy, System Support, Blame Victim, Professional Role Resistance/Fear of Offending Patient, Victim/Provider Safety, and Frequency of DV Inquiry. We found item domain Cronbach alpha to be acceptable, ranging from 0.73 to 0.91. The final overall measure had 39 items and an alpha of 0.88. Data are reviewed that support the measure's sensitivity to change in response to a training intervention. CONCLUSION: The measure provides a reliable method for assessing provider characteristics and training needs. It may also serve to evaluate training and policy interventions in DV.