1302: SERUM BIOMARKERS AND VITAL SIGNS DO NOT PREDICT THE NEED FOR 6-HOUR HEMORRHAGE CONTROL INTERVENTIONS

Introduction: The literature is unclear about reliable markers of ongoing bleeding requiring intervention. Selecting patients who require operative or interventional radiology (IR) hemorrhage control intervention (HCI) is challenging. We hypothesized that our “Hemorrhage Watch” protocol (HW) consisting of serial biomarkers (SB) and vital signs (VS) accurately discriminates these patients from those without the need for HCI. Methods: This is a retrospective study of patients admitted to one surgical ICU between 8/2017 - 3/2022 who underwent HW. HW protocol consisted of hourly hemoglobin (Hgb) and blood gas, and VS every 15 minutes for 3 hours. Inclusion criteria were adult patients with imaging indicative of ongoing hemorrhage or physician discretion. Demographics and background data were recorded. The primary outcome was the need for early (within 6 hours of HW start) HCI. Secondary outcomes included in-hospital mortality, blood transfusions, ICU and hospital length-of-stay. A logistic regression model adjusting for demographics, bleeding disorders, current anti-coagulant (AC) and anti-platelet (AP) use, trends (defined as average minus last measurement) of VS and SB, and ICU blood transfusions during HW was developed. Results: A total of 888 patients were included; 225 had early HCI (218 operative, 7 IR) and 663 did not (no-HCI). The characteristics of the groups (HCI/no-HCI) were: mean age 67/67, male 54/57%, white 47/50%, bleeding disorders 9/9%, AC 27/27%, AP 34/30%, pre-ICU Hgb 10/11 g/dL, platelet count 261/239 K/µL, lactate 4.0/3.5 mmol/L, base deficit 2.4/2.0 mmol/L, PT 16/16 sec, PTT 33/37 sec, AP/AC reversal agents 2.7/4.7%, and red blood cell 3.0/5.7% (mean volumes: 435/333mL), plasma 0.4/0.6% (346/355mL), platelet 0.4/0.1% (181/63mL) transfusions. After adjusting for baseline characteristics and transfusions, neither individual SB nor VS, or their trends, were predictive of the need for HCI (OR=1.04, p=0.47). Of the secondary outcomes, only in-hospital mortality was lower in the HCI vs no-HCI group (16 vs 28%, p< 0.001). Conclusions: In this study of patients suspected of ongoing bleeding, serial SB and VS were unable to predict the need for HCI in surgical ICU patients.