330

Abstract

Introduction: Sepsis guidelines recommend early goal-directed therapy based on patient’s hemodynamics to reduce morbidity and mortality in children. A few studies have suggested physicians inaccurately estimate the hemodynamic profile of acutely ill patients. Objective measurements of cardiac index (CI) and systemic vascular resistance index (SVRI) are still not routine in most ICUs. Methods: Single-center, prospective, double blinded observational study at the Children’s Hospital Los Angeles PICU in children with shock, septic shock or severe sepsis. Patients with intra-cardiac shunt or valvular abnormalities were excluded. Patients had serial clinical assessments of CI and SVRI by a minimum of two physicians. At the same time, objective measures of CI and SVRI were obtained using the Ultrasonic Cardiac Output Monitor (USCOM) device. Physicians assessed CI and SVRI as low, normal and high and USCOM values were categorized into the same classes using published values. Inter-observer agreement amongst physicians as well as between physicians and USCOM were assessed using the kappa (κ) statistic. Results: Twenty six patients were enrolled, with 175 assessments. Overall, physician to USCOM agreement is poor, for CI it was 50.84% agreement (κ 0.21, 95% CI(0.16–0.25)) and for SVRI it was 46% (κ 0.14 (0.12 - 0.20)). When sub grouped by level of training, for CI attending physicians (n=46) had 41% agreement (κ 0.07 (0.04 – 0.09)) with USCOM; fellows (n=102) 51% agreement (κ 0.22 (0.14 - 0.31)); residents 67% agreement (n=27) (κ 0.42 (0.28 - 0.57)). For SVRI, attending physicians (n=46) had 52% agreement (κ 0.27(0.21 – 0.38)) with USCOM; fellows (n=102) 43% agreement (κ 0.10 ((-0.01 - 0.20)); residents 44% agreement (n=27) (κ 0.13 ((0.13 - 0.42)). Conclusions: There is significant inter-observer variability in physician assessment of CI and SVRI, with poor agreement with objective measures. Future study is needed to determine if this variability results in differences in patient management, which may impact outcome.