Use of epidemiological data in the diagnosis of physical child abuse: variations in response to hypothetical cases

This study was undertaken to determine whether epidemiological data describing the injuries associated with unintentional trauma could help physicians differentiate intentional from unintentional injury. The authors also wished to determine if case and physician-specific factors altered how epidemiological data were used. Study subjects were 280 physicians who had registered for an advanced course in pediatric life support. Responses were received from 166 (59%); 48% were pediatricians and 37% had trained in emergency medicine. Case vignettes were written describing a child's fall from a highchair. The vignettes systematically varied the type of injury sustained, the presence of a social risk factor, and whether the child was followed by a primary care provider. Vignettes were administered with and without provision of epidemiological data describing injuries associated with highchair falls. Each study participant received one vignette, and was asked to mark, on a 0-100 scale, their confidence in the injury history given. In vignettes where the presenting injury (femur fracture) would not be expected based on the epidemiological data, the availability of data appropriately decreased confidence that the injury was unintentional (average decrease = 14.3, 95% confidence limits = 3.8 to 25.9). When the presenting injury (skull fracture) matched the epidemiological data, its availability did not alter confidence (average change = 0.5, 95% to -13.1 to + 12.1). The impact of a contrast between presenting injury and data varied with specialty: It caused a marked drop in confidence among pediatricians but caused no change among emergency room/intensive care unit (ER/ICU) physicians. Data did, however, lead ER/ICU physicians to respond more strongly to a social risk cue (mean confidence 14.9 vs. 42.4, p less than .05) than they had in the absence of data. We concluded that epidemiological data has the potential for influencing physician decision-making in cases of suspected maltreatment, but that its impact may vary among physicians with differing training.