Machine learning analysis of bleeding status in venous thromboembolism patients.

Background: Anticoagulation therapy is the mainstay of therapy for patients with venous thromboembolism (VTE). However, continuing or stopping anticoagulants after the first 3 to 6 months is a difficult decision that requires ascertainment of the risk of bleeding and recurrent VTE. Despite the development of several statistical models to predict bleeding, the benefit of machine learning (ML) models has not been investigated in depth. Objectives: To assess the benefits of ML algorithms in bleeding risk evaluation in VTE patients and gain insight into their baseline information. Methods: The baseline clinical, demographic, and genotype information was collected for 2542 patients with VTE who were on extended anticoagulation therapy. Six unsupervised dimensionality reduction and clustering ML algorithms were used to visualize and cluster the data for patients with major bleeding (118 patients) and nonbleeders. Eight supervised ML algorithms were trained and compared with the previously derived clinical models using a 5-fold nested cross-validation scheme. Results: The baseline dataset for bleeders and nonbleeders showed a high degree of similarity. Two novel clusters were discovered within the dataset for bleeders based on the presence of isolated pulmonary embolism or isolated deep vein thrombosis, though the difference in bleeding risks was not statistically significant (P = .32). The supervised analysis showed that the ML and clinical models have similar discrimination (c-statistics, ∼62%) and calibration performance (Brier score, ∼0.045). Conclusion: The clinical variables recorded at baseline are not distinctive enough to improve bleeding prediction beyond the performance of the existing models, and other strategies or data modalities should be considered.

A deep-learning approach to predict bleeding risk over time in patients on extended anticoagulation therapy.

BACKGROUND: Thus far, all the clinical models developed to predict major bleeding in patients on extended anticoagulation therapy use the baseline predictors to stratify patients into different risk groups. Therefore, these models do not account for the clinical changes and events that occur after the baseline visit, which can modify risk of bleeding. However, it is difficult to develop predictive models from the routine follow-up clinical interviews, which are irregular sequences of multivariate time series data. OBJECTIVES: To demonstrate that deep learning can incorporate patient time series follow-up data to improve prediction of major bleeding. METHODS: We used the baseline and follow-up data that were collected over 8 years in a longitudinal cohort study of 2542 patients, of whom 118 had major bleeding. Four supervised neural network-based machine-learning models were trained on the baseline, follow-up, or both datasets using 70% of the data. The performance of these models was evaluated, along with modified versions of 6 previously developed clinical models, on the remaining 30% of the data. RESULTS: An ensemble of feedforward and recurrent neural networks that used the baseline and follow-up data was the best-performing model, achieving a sensitivity and a specificity of 61% and 82%, respectively, in identifying major bleeding, and it outperformed the previously developed clinical models in terms of area under the receiver operating characteristic curve (82%) and area under the precision-recall curve (14%). CONCLUSION: Time series follow-up data can improve major bleeding prediction in patients on extended anticoagulation therapy.

Qualitative experiences, values, and decisional needs of patients with unprovoked venous thromboembolism who suffer bleeding-"This pill will keep you alive tonight".

Background: Contemporary guidelines recommend extended-duration anticoagulation among patients with a first unprovoked venous thromboembolism (VTE). Little is known about whether this recommendation aligns with patient values after a bleeding complication. Objectives: To explore the experiences, values, and decisional needs of patients with unprovoked VTE related to extended-duration treatment after an anticoagulant-associated bleed. Methods: In this descriptive, qualitative study, face to face online semistructured interviews were conducted with patients with unprovoked VTE who had experienced bleeding and continued anticoagulant treatment in one academic hospital in Canada. Data were analyzed using directed content analysis to identify themes. Themes were mapped onto the Ottawa Decisional Support Framework to identify decisional needs. Results: Between September and December 2021, 14 patients were interviewed (age 41-69 years; 9 females). Many patients were not aware of the option to stop anticoagulation and had limited understanding of the decision about treatment duration. Despite the negative quality-of-life impact of clinically relevant bleeding during VTE treatment, the majority continued anticoagulation due to emotional trauma of VTE diagnosis, a perception that bleeding would be more manageable than VTE recurrence, a desire to maintain a connection to subspecialty care or non-VTE related benefits (eg, cancer diagnosis, protection from COVID-19). Patients' decisional needs included lack of choice awareness, inadequate support for participation, lack of personalized risk stratification, and inadequate information on monitoring and managing heavy menstrual bleeding. Conclusion: Despite the impact of anticoagulant-associated bleeding on quality of life, patients preferred continuing with anticoagulation for reasons extending beyond secondary VTE prevention. Effective decision-support interventions are needed to address unmet decisional needs.

Balancing risks of recurrent venous thromboembolism and bleeding with extended anticoagulation: a decision analysis.

Background: A decision to stop or continue anticoagulation after 3 months of anticoagulation for venous thromboembolism (VTE) should be made by weighing individual risks of recurrence and bleeding. Objectives: To determine the optimal ratio of recurrence risk reduction to increase the risk of bleeding in terms of maximizing quality-adjusted life years (QALYs) gained. Methods: Using a microsimulation model, outcomes within 5 years were simulated after assigning extended treatment if absolute recurrence risk reduction outweighed absolute increase in clinically relevant bleeding risk (International Society on Thrombosis and Haemostasis definition), weighted by a certain ratio. Data were simulated based on the Bleeding Risk Study, a prospective cohort including patients after ≥3 months of anticoagulation for unprovoked VTE or provoked VTE with history of VTE. The VTE-PREDICT risk score was used to estimate 5-year risks of recurrent VTE and clinically relevant bleeding. Results: Among 10,000 individuals (mean age, 60.2 years, 36% female), the ratio of 0.90 (95% CI, 0.51-3.40; ie, bleeding is considered 0.90 the severity of recurrent VTE), with 99% of patients assigned extended anticoagulation, was considered optimal and resulted in 93 (95% CI, -23 to 203) additional QALYs compared with the least favorable ratio (5.10, 0% extended anticoagulation). At the optimal ratio, treatment based on VTE-PREDICT yielded 44 (95% CI, -69 to 157) additional QALYs versus standard of care. Conclusion: With the current evidence, the optimal ratio between relevant bleeding risk and absolute recurrence risk reduction remains uncertain. Our results confirm that clinical equipoise exists regarding the decision to stop or continue anticoagulation after initial VTE treatment, emphasizing the importance of shared decision-making.

Quality and Patient Safety Metrics: Developing a Structured Program for Improving Patient Care in the Department of Medicine at The Ottawa Hospital.

PROBLEM: Despite increasing recognition of the importance of quality and patient safety in academic medicine, challenges remain with ensuring physician participation in quality assurance and quality improvement efforts, such as lack of compensation and enabling resources. An organizational culture that includes physician leadership and a supportive infrastructure is needed to encourage physician backing of quality and patient safety initiatives. APPROACH: The authors describe the development of a robust quality and patient safety program in the Department of Medicine at The Ottawa Hospital over the past 7 years and highlight how the department changed its organizational culture by prioritizing quality and patient safety and establishing the necessary infrastructure to support this program. Program development was characterized by 4 overarching themes: incentives, administrative structure and physician leadership, training and support, and system enhancements. OUTCOMES: As a result of the program, the department broadly implemented a standardized framework for conducting quality committee meetings and morbidity and mortality rounds and reviewing patient safety incidents and patient experience across its 16 divisions. This has led to 100% departmental compliance on corporate quality assurance metrics each year (e.g., regular multidisciplinary divisional quality committee meetings), along with physician participation in formal quality improvement initiatives that align with larger corporate goals. NEXT STEPS: The authors reflect on lessons learned during the implementation of the program and the essential elements that contributed to its success. Next steps for the program include using a centralized repository of quality and patient safety data, including patient safety incident dashboards, to encourage greater divisional collaboration on quality improvement initiatives and continuous institutional learning over time. Another important avenue will be to create an academic hub for excellence in quality and a formal approach to reward and promote physicians for their quality work.

Antithrombotic Therapy for VTE Disease: Compendium and Review of CHEST Guidelines 2012-2021.

The American College of Chest Physicians (CHEST) Antithrombotic Therapy for Venous Thromboembolism Disease evidence-based guidelines are now updated in a more frequent, focused manner. Guidance statements from the most recent full guidelines and two subsequent updates have not been gathered into a single source. An international panel of experts with experience in prior antithrombotic therapy guideline development reviewed the 2012 CHEST antithrombotic therapy guidelines and its two subsequent updates. All guideline statements and their associated patient, intervention, comparator, and outcome questions were assembled. A modified Delphi process was used to select statements considered relevant to current clinical care. The panel further endorsed minor phrasing changes to match the standard language for guidance statements using the modified Grading of Recommendations, Assessment, Development, and Evaluations (ie, GRADE) format endorsed by the CHEST Guidelines Oversight Committee. The panel appended comments after statements deemed as relevant, including suggesting that statements be updated in future guidelines because of interval evidence. We include 58 guidance statements from prior versions of the antithrombotic therapy guidelines, with updated phrasing as needed to adhere to contemporary nomenclature. Statements were classified as strong or weak recommendations based on high-certainty, moderate-certainty, and low-certainty evidence using GRADE methodology. The panel suggested that five statements are no longer relevant to current practice. As CHEST continues to update guidance statements relevant to antithrombotic therapy for VTE disease, this article serves as a unified collection of currenrtly relevant statements from the preceding three guidelines. Suggestions have been made to update specific statements in future publications.