Cancer-related venous thromboembolism: treatment choice, treatment patterns, and subsequent outcomes

This project will evaluate and compare anticoagulation treatments for cancer-related VTE. A large, national cohort of individuals will be utilized from an administrative claims database for the years 2012- 2014 which includes medical and pharmacy information for over 20 million lives. The cohort study design will index the original VTE event for people with cancer and follow these individuals forward in time to assess treatment patterns. Follow-up will continue until one of four events occur: a recurrent VTE, an adverse bleed event, death, or the end of the study period. Baseline demographics and clinical characteristics will be assessed using a pre-index look back period and will include, among others, patient age, gender, region, comorbidities, tumor site, and presence of metastases. A comparative effectiveness analysis will implement a time-varying treatment, marginal structural, competing risks model which will allow for changes in treatment group assignment as well as changes in clinical characteristics (e.g. surgery, newly diagnosed conditions, etc.). Treatment group assignments will include warfarin, low-molecular weight heparins, novel oral anticoagulants, vena cava filters, or no treatment. Study outcomes will be assessed at 1, 3, 6, and 12 month intervals to assess the effectiveness outcome (recurrent clot) and the safety outcomes (major and minor bleeds). Propensity scoring techniques will be utilized to control for selection bias between the treatment groups and to provide more valid treatment comparisons. Using the results from the competing risks models, a predictive risk score will be developed for clinical decision making. The risk score will be internally validated in the parent dataset using a random sample and the externally validated using another in-house data source. The risk score will be developed into an interactive tool to be used as a decision making tool and will assess the baseline risk of a patient to experience either a recurrent clot or a bleed event as a measure of net clinical benefit (or harm) of treatment. The tool will be further optimized with links to clinical practice guidelines and product labels for clinical use.