Blood clots often form in the veins of the legs or in the lungs and are more common in patients with cancer. The risk is highest following the diagnosis or treatment of the cancer with drugs, radiation or surgery. Blood clots are usually treated with “blood thinning” drugs called anticoagulants. Guidelines recommend treatment for 3-6 months for some patients with longer courses of therapy often recommended for patients with cancer and clots. Whilst taking anticoagulants patients are more likely to have bleeding complications and patients with cancer are more likely to bleed than those without cancer.
The fear of severe bleeding from anticoagulants could result in doctors avoiding anticoagulants in patients who are likely to benefit from them. Bleeding in patients initially treated with anticoagulants could result in discontinuation of anticoagulant therapy and consequently increase the potential for new clots. Our aim is to be able to advise doctors more accurately, with a medical calculator, on the things that increase the risk of bleeding so that these factors can be dealt with and treatments can be optimised.
This observational cohort study aims to investigate predictors for the risk of significant bleeding during anticoagulant (AC) treatment for active cancer associated venous thromboembolism (CAT).
A cohort of patients with venous thromboembolisms (VTEs), defined as deep vein thrombosis (DVT) or pulmonary embolism (PE), between 01/04/2008 and 31/10/2020, a cancer recording within 180 days before the VTE and recorded AC use within 90 days after the CAT will be defined.
The outcome of interest will include major bleeding and clinically-relevant non-major bleeding requiring hospitalization (CRNMB-H) within 180 days after CAT and will be defined from general practitioner (GP) diagnoses, hospital discharge diagnoses and causes of death.
Potential predictors for bleeding events will include clinical and laboratory variables identified as components in published risk scales for bleeding in AC use and other potential predictors mentioned in published literature.
Multivariate Fine and Gray regression models accounting for the competing risk mortality will be used to estimate subdistribution hazard ratios (SHRs) for all potential bleeding event predictors. The scoring scheme for the prediction of bleeding events will be developed from these SHRs. The discrimination of the new score, will be assessed by estimation of C statistics using cross validation. In addition, discrimination will be compared directly to existing clinical scores to predict bleeding in patients with VTE. To assess the calibration of the score, the study cohort will be split into 5 subgroups based on quintiles of risk. Within each of these 5 subgroups the observed and expected number of bleeding cases will be compared and a goodness of fit test will be performed.
Significant bleeding including major bleeding in outpatients;
Clinically-relevant non-major bleeding requiring hospitalization;
Ayman Nassar - Chief Investigator - Bristol Myers Squibb - Europe ( BMS )
Carlos Martinez - Corresponding Applicant - Institute for Epidemiology, Statistics and Informatics GmbH (Pharma Epi)
Alexander Cohen - Collaborator - King's College London (KCL)
Christopher Wallenhorst - Collaborator - Institute for Epidemiology, Statistics and Informatics GmbH (Pharma Epi)
Raza Alikhan - Collaborator - University Hospital of Wales
Sarah Grundy - Collaborator - Bristol Myers Squibb - Europe ( BMS )
Satarupa Choudhuri - Collaborator - NHS England
Ayman Nassar - Collaborator - Bristol Myers Squibb - Europe ( BMS )
Stephan Rietbrock - Collaborator - Institute for Epidemiology, Statistics and Informatics GmbH (Pharma Epi)
HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation