Purpose: We sought to evaluate the performance of 2 different screening questionnaires for intimate partner violence (IPV), in conjunction with injury location, as markers for IPV-related injuries.
Methods: We implemented a cross-sectional study and enrolled a sample of women presenting to the emergency department for evaluation and management of nonverifiable injuries. Study subjects were randomly assigned to receive 1 of 2 IPV screening questionnaires: the Partner Violence Screen (PVS) or the short-Woman Abuse Screening Tool (short-WAST). We evaluated a combination of 2 markers of IPV-related injury: (1) injury location, classified as head/neck/facial (HNF) or other and (2) responses to the IPV questionnaires: positive or negative. Our predictor variable was the probability of self-report of IPV-related injury defined as (1) high probability (HNF injuries were present and there was a positive response to the IPV questionnaire) or (2) low probability (all other combinations of injury location and responses to the questionnaires). The outcome variable was self-reported injury etiology, IPV or other. Demographic variables were also recorded for our study sample. Univariate and bivariate statistical analyses were computed for the sample. Sensitivity, specificity, and positive and negative predictive values and odds ratios were calculated (P < .05).
Results: The sample was composed of 200 women. The sensitivities/specificities for the PVS-injury location and short-WAST-injury location combinations were 0.75/0.70 and 0.77/0.61, respectively. The odds ratios for IPV-related injury etiology were 10.2 (3 < OR < 41, P = .01) for the PVS-injury location combination and 3.7 (0.90 < OR < 15, P = .07) for the short-WAST-injury location combination.
Conclusions: A set of markers composed of injury location and the PVS was statistically associated with the likelihood of reporting IPV-related injuries. The short-WAST did not perform as well as the PVS in the study's clinical setting.