Pulmonary Embolism Management Audit and Machine Learning Analysis of Delayed Anticoagulation in a Swiss Teaching Hospital

Background/Objectives: Diagnosing acute pulmonary embolism (PE) is challenging due to its wide range of symptoms and numerous differential diagnoses. Medical professionals must balance performing all essential examinations and avoiding unnecessary testing. This study aimed to retrospectively audit the diagnosis and treatment of acute PE at a Swiss public teaching hospital to determine the adherence to current guidelines and to identify the factors associated with the delayed initiation of anticoagulation in PE patients. Methods: In this retrospective observational cohort study, we included all adult patients hospitalized with PE at the Cantonal Hospital Baselland (KSBL) between November 2018 and October 2020, where the diagnosis was made within the first twelve hours of their arrival to the emergency department (ED). LASSO regression was employed to identify clinical characteristics associated with delayed anticoagulation initiation. Results: A total of 197 patients were included (mean age: 70 years, 54% female). The audit revealed that diagnostic workup was conducted according to guidelines in 57% of cases. Often, D-dimer levels were measured although not strictly necessary (70%). Pretest probability was assessed and documented using the Wells or Geneva score in only 3% of patients, and risk assessment via the Pulmonary Embolism Severity Index (PESI) score was documented in 21% of patients. The median time from ED arrival to CT scan was 120 min (IQR 89.5-210.5), and the median time to anticoagulation initiation was 193 min (IQR 145-277). Factors identified by LASSO associated with delayed anticoagulation included prolonged time from ED arrival to CT scan, the presence of distended jugular veins on examination, ED arrival in the morning, and presenting symptoms of weakness or tiredness. Complementary leg ultrasound was performed in 57% of patients, with 38% of these cases lacking prior clinical examination for deep vein thrombosis. The duration of the anticoagulation treatment was not specified in the discharge report for 17% of patients. A medical follow-up after discharge was recommended in 75% of the patients. Conclusions: In conclusion, while the management of PE at the KSBL generally adheres to high standards, there are areas for improvement, particularly in the morning performance, the use of a pretest probability assessment, D-dimer measurement, risk assessment via the PESI score, the performance of complementary leg ultrasounds, clarification of the anticoagulation duration, and follow-up management.