Interobserver Agreement in the Clinical Assessment of Children With Blunt Abdominal Trauma

Abstract

Objectives The objective was to determine the interobserver agreement of historical and physical examination findings assessed during the emergency department (ED) evaluation of children with blunt abdominal trauma. Methods This was a planned substudy of a multicenter, prospective cohort study of children younger than 18 years of age evaluated for blunt abdominal trauma. Patients were excluded if injury occurred more than 24 hours prior to evaluation or if computed tomography (CT) imaging was obtained at another hospital prior to transfer to a study site. Two clinicians independently recorded their clinical assessments of a convenience sample of patients onto data collection forms within 60 minutes of each other and prior to CT imaging (if obtained) or knowledge of laboratory results. The authors categorized variables as either subjective symptoms (i.e., patient history) or objective findings (i.e., physical examination). For each variable recorded by the two observers, the agreement beyond that expected by chance was estimated, using the kappa (κ) statistic for categorical variables and weighted κ for ordinal variables. Variables with 95% lower confidence limits (LCLs) κ ≥ 0.4 (moderate agreement or better) were considered to have acceptable agreement. Results A total of 632 pairs of physician observations were obtained on 23 candidate variables. Acceptable agreement was achieved in 16 (70%) of the 23 variables tested. For six subjective symptoms, κ ranged from 0.48 (complaint of shortness of breath) to 0.90 (mechanism of injury), and only the complaint of shortness of breath had a 95% LCL κ < 0.4. For the 17 objective findings, κ ranged from –0.01 (pelvis instability) to 0.82 (seat belt sign present). The 95% LCL for κ was <0.4 for flank tenderness, abnormal chest auscultation, suspicion of alcohol or drug intoxication, pelvis instability, absence of bowel sounds, and peritoneal irritation. Conclusions Observers can achieve at least acceptable agreement on the majority of historical and physical examination variables in children with blunt abdominal trauma evaluated in the ED. Those variables are candidates for consideration for development of a clinical prediction rule for intra‐abdominal injury in children with blunt trauma.

Interobserver Agreement in Pediatric Cervical Spine Injury Assessment Between Prehospital and Emergency Department Providers

“…Our findings are consistent with previous studies that evaluated interobserver agreement between various provider groups regarding clinically important factors in pediatric trauma . Most germane to this analysis are previous prospective PECARN studies that evaluated interobserver agreement for risk factors that were ultimately incorporated into clinical decision support tools.…”

“…showed that pediatric emergency clinicians achieved at least acceptable agreement for both subjective and objective clinical variables in children with blunt head trauma. Yen et al . showed that clinicians in pediatric EDs achieved at least acceptable agreement on the majority of historical and physical examination variables in children with blunt torso injury.…”

“…Similar to previous studies evaluating risk factors for traumatic injury in children, we sought to compare the performance of two independent medical providers in the real‐time assessment of clinical risk factors in pediatric trauma . However, in contrast to these studies, we compared the observations of the treating EMS scene response providers to those of the treating ED providers.…”

Interobserver Agreement in Retrospective Chart Reviews for Factors Associated With Cervical Spine Injuries in Children

“…8 A prospective study of blunt abdominal trauma found almost perfect agreement for injury mechanism and moderate agreement for painful injury, difficulty breathing, and tenderness. 9 A study of adults with dystonia showed moderate agreement (j = 0.52) for cervical dystonia. 10 The agreement of retrospectively obtained data is fundamentally different from agreement between prospective observers due to bias introduced by the abstractor's interpretation of the medical record.…”

Relationship of Physician‐identified Patient Race and Ethnicity to Use of Computed Tomography in Pediatric Blunt Torso Trauma

“…Additionally, whether a CT was obtained or not, the treating physician recorded their clinical suspicion for: 1) the presence of an intra‐abdominal injury identifiable on CT and 2) an intra‐abdominal injury that would require acute intervention. Additional information on the study population, methods, interobserver agreement, and outcomes of children discharged from the ED without CT scans are reported in other study publications …”