Mapping Opioid Use Trajectories in Vetrans Undergoing Thoracic Surgery via Latent Classes

• Activity Number: 475 - Understanding Threats to People, Data, and Privacy
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Social Statistics Section
• Author(s): Michael Bishop* and Emine Bayman
• Companies: University of Iowa, College of Public Health and University of Iowa, Carver College of Medicine
• Keywords: Cluster Analysis; Latent Variables; Veterans Administration; Chronic Opioid Use; Opioid After Surgery

Abstract

Due to the opioid epidemic, greater efforts are now placed on early intervention, especially for vulnerable populations. Veterans Administration data containing information on veterans who underwent thoracic surgery included pre-surgery opioid-use information to stratify the veterans to two groups, pre-surgery non-chronic opioid use (n1 = 16,612) vs. chronic pre-surgery opioid use (n2 = 2,328). Additional variables including preoperative medication use, and psychosocial diagnoses were also used. A Latent Class Analysis (LCA) model with 3 clusters provided the best fit. Classes were well differentiated, and separated based on use of antidepressants, antiepileptics, benzodiazepine, diagnoses of depression, anxiety and rate of pre-surgery chronic pain. Chronic opioid use rates at 1 year varied between the clusters in the 2 strata (1: 9.2%, 8.5%, 4.9% and 2: 66.8%, 61.8%, 46.1%), and were our primary endpoints for the opioid-use trajectories. Clusters were created based on the variables that can be observed at the time of the surgery. Thus, at the time of surgery, the model can be used to help identify patients belonging to each cluster of chronic opioid use at 1 year after surgery.