A latent class model for defining severe hemorrhage

• Activity Number: 184
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Mohammad Rahbar*+ and Hanwen Huang and Jing Ning and Deborah del Junco and Erin Fox and John Holcomb
• Companies: The University of Texas Health Science Center at Houston and The University of Texas Health Science Center and MD Anderson Cancer Center and The University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston
• Keywords: Latent class models ; Trauma ; Blood product utilization ; Massive transfusion ; Survival ; Hemorrhage

Abstract

There is no diagnostic test to identify trauma patients who have had severe hemorrhage (SH) and may need a massive transfusion protocol (MTP). However, several predictive models have been developed based on the traditional definition of massive transfusion, which is transfusion of 10 units of red blood cells (RBCs) within 24 hours of Emergency Department (ED) admission. This definition excludes patients with severe bleeding who died before a 10th unit of RBCs could be transfused, resulting in survival bias. The lack of a valid definition for severe hemorrhage calls these prediction models into question. We proposed a latent class model for identifying a subgroup of patients with SH.We developed an EM algorithm for estimating the posterior probability of being an SH patient based on information at ED admission, blood product utilization, and survival status during the first 24 hours. We assessed the performance of our latent class model in classifying SH patients and compare to the traditional massive transfusion definition using data from a retrospective trauma transfusion study.