{"title":"Data-driven identification of urgent surgical procedures for use in trauma outcomes measurement.","authors":"Matthew Miller, Louisa Jorm, Blanca Gallego","doi":"10.1136/tsaco-2025-001783","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>No standardized list of urgent-trauma-surgery exists for analysis in injury studies. If coded by a standard classification system, such a list could facilitate the standard evaluation and comparison of trauma systems. Solving this problem using Delphi methods or expert opinion incorporating all surgical specialties would be resource-intensive. Instead, we describe a flexible data-driven method for generating a list of urgent surgical procedures from routine administrative data.</p><p><strong>Methods: </strong>We linked perioperative and inpatient data for trauma patients with procedures booked within 24 hours of admission from a single Australian hospital (July 2018-July 2023). Surgical procedure codes were extracted where booked free-text and coded procedures matched. Procedures were labeled urgent-by-agreement if over 75% were needed within 4 hours, or urgent-by-consensus if 50-75% met this time frame with consensus below 0.7. Our method also allows adjustment for urgency time frame.</p><p><strong>Results: </strong>Of 567 unique procedures from 6,750 total in 4,737 trauma admissions, 161 were classified as urgent-by-agreement and 6 as urgent-by-consensus. 15 surgical specialties were represented on this list.</p><p><strong>Discussion and conclusions: </strong>Using routinely collected data, we outline a method for generating and updating urgent surgical procedure lists for trauma patients that could be applied at the institution level or across trauma networks. In addition, different urgency periods can be accommodated. Future work could look at further automating these processes by incorporating deep learning.</p>","PeriodicalId":23307,"journal":{"name":"Trauma Surgery & Acute Care Open","volume":"10 2","pages":"e001783"},"PeriodicalIF":2.1000,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12097093/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Trauma Surgery & Acute Care Open","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1136/tsaco-2025-001783","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/1/1 0:00:00","PubModel":"eCollection","JCR":"Q3","JCRName":"CRITICAL CARE MEDICINE","Score":null,"Total":0}
引用次数: 0
Abstract
Background: No standardized list of urgent-trauma-surgery exists for analysis in injury studies. If coded by a standard classification system, such a list could facilitate the standard evaluation and comparison of trauma systems. Solving this problem using Delphi methods or expert opinion incorporating all surgical specialties would be resource-intensive. Instead, we describe a flexible data-driven method for generating a list of urgent surgical procedures from routine administrative data.
Methods: We linked perioperative and inpatient data for trauma patients with procedures booked within 24 hours of admission from a single Australian hospital (July 2018-July 2023). Surgical procedure codes were extracted where booked free-text and coded procedures matched. Procedures were labeled urgent-by-agreement if over 75% were needed within 4 hours, or urgent-by-consensus if 50-75% met this time frame with consensus below 0.7. Our method also allows adjustment for urgency time frame.
Results: Of 567 unique procedures from 6,750 total in 4,737 trauma admissions, 161 were classified as urgent-by-agreement and 6 as urgent-by-consensus. 15 surgical specialties were represented on this list.
Discussion and conclusions: Using routinely collected data, we outline a method for generating and updating urgent surgical procedure lists for trauma patients that could be applied at the institution level or across trauma networks. In addition, different urgency periods can be accommodated. Future work could look at further automating these processes by incorporating deep learning.