Latent Anastomotic Leak Susceptibility Phenotypes Across Gastrointestinal Operations and Postoperative Care Environments

• Fahad Siddiqui

  Karakoram International University, University Road, Gilgit 15100, Pakistan

  Author

• Kamran Ashraf

  University of Sahiwal, Farid Town Road, Sahiwal 57000, Pakistan

  Author

Anastomotic leak after gastrointestinal reconstruction is commonly reported as an average complication rate attached to a procedure or a hospital. That practice is clinically convenient but biologically coarse, because patients who experience leak do not appear to travel through a single pathophysiologic route. Some deteriorate through early inflammatory and perfusion failure, some through technically stressed local anatomy, and others through chronic nutritional or frailty-related vulnerability that becomes visible only after the operation has already begun to tax physiologic reserve. If those hidden susceptibility states are unevenly distributed across surgery types and care environments, average rates will conceal an important layer of risk structure. This investigation estimated latent leak-susceptibility phenotypes in a large multicenter electronic health record cohort and then examined how those phenotypes were sorted across gastrointestinal procedures and postoperative care environments. The cohort contained 53,418 adult restorative gastrointestinal operations from 21 hospitals. Anastomotic leak within 30 days occurred after 3,446 procedures, producing an overall incidence of 6.5%. A joint latent class outcome model selected four phenotypes. Their estimated prevalences were 38.9%, 22.7%, 17.1%, and 21.3%, with corresponding leak incidences of 2.7%, 10.9%, 11.6%, and 7.8%. Procedure families and care environments differed substantially in phenotype composition. Low pelvic colorectal surgery was strongly enriched for a mechanically stressed phenotype, whereas esophagogastric reconstruction accumulated a depletion-dominant phenotype. Emergency overnight care contained the highest proportion of the inflammatory-perfusion phenotype. Across the procedure-by-setting matrix, 51.8% of leak-rate dispersion was attributable to phenotype sorting, 28.4% to differential phenotype expression within the same phenotype, and 19.8\% to residual interaction. The findings indicate that cross-setting heterogeneity in leak is partly a problem of hidden clinical states being redistributed across pathways rather than only a problem of visible case mix.

• PDF