Principal Investigator: Patricia Dykes
Blood clots in the leg and lung can be dangerous. They are also hard for doctors to diagnose. Researchers use codes to study cases of blood clots. However, these codes are often not accurate. These codes are only accurate 64% of the time. Therefore, our team sought to find a way to improve these codes. We did this by looking at the codes as well as other items. The other items we used were medication data and imaging data. Using all of these factors improved how we could define blood clots in health records. With the new method, our accuracy is now 94%. This is a big and important improvement. This work will allow researchers to better understand blood clots. As a result, patients with blood clots can be treated better.
Venous thromboembolism (VTE), consisting of pulmonary embolism (PE) and deep vein thrombosis (DVT), is a common, preventable public health problem affecting approximately 300,000 – 600,000 individuals in the United States each year, and requires timely and adequate treatment. To improve VTE detection and diagnosis, informaticians and researchers leverage healthcare databases to identify instances of VTE for research and quality reporting purposes. Traditionally, identifying cases of VTE has relied mainly on the use of ICD codes. However, previous studies have demonstrated that the use of ICD codes alone are often subject to error due to limitations in available clinical data, diagnostic errors, and coding errors made by human operations. As a result, using ICD codes alone to determine the presence of a VTE provides a poor Positive Predictive Value (PPV) of approximately 64%. The team at Brigham and Women’s Hospital has sought to improve this PPV by defining a VTE as the combination of an ICD, imaging, and RxNorm code within a given hospital encounter. Defining a VTE as the combination of these three features results in a notable improvement in defining a true VTE, raising the PPV to 94%.