Background: This study tests the hypothesis that postoperative undertriage of high-acuity patients to hospital floor units is associated with new postoperative goals of care (Do Not Resuscitate (DNR), palliation, comfort care, and hospice) orders, representing important outcomes that are incompletely captured by standard mortality metrics. Study Design: Using longitudinal electronic health record data from three hospitals, postoperative undertriage labels were assigned to high acuity floor patients, who were propensity-matched with ICU patients who had similar risk profiles. Patient acuity was quantified by a fully supervised neural network developed and validated on preoperative and intraoperative data from 140, 608 postoperative admissions; results are reported from a holdout test cohort (N=15, 661). Physician orders and patient outcomes were compared between undertriaged (N=829) and risk-matched ICU cohorts (N=829). Results: Compared with risk-matched ICU patients, undertriaged patients had similar baseline characteristic, total direct costs (29. 7K 18. 5K-48. 9K vs. 33. 9K 21. 8K-53. 9K), value-of-care (outcomes/costs: 1. 0 0. 6-1. 6 vs. 0. 9 0. 6-1. 4), and incidence of postoperative CPR (5. 8% vs. 4. 2%), central venous catheter placement (23. 9% vs. 20. 6%), and electric cardioversion (1. 8% vs. 3. 3%, all P≥. 05), and higher incidence of unplanned intubation (12. 1% vs. 7. 1%), fiberoptic bronchoscopy (11. 7% vs. 6. 4%), and in-hospital mortality (7. 0% vs. 1. 1%), as well as new postoperative transitions to DNR status (7. 2% vs. 2. 8%) and comfort care (4. 2% vs. 0. 6%), all P<. 05). Conclusion: Postoperative undertriage of high-acuity patients to floor units is associated with increased incidence of respiratory failure and new postoperative transition to DNR status and comfort care goals, underscoring the importance of optimizing upstream triage decisions.
Loftus et al. (2026) studied this question.