Using existing data for epidemiologic research of cardiovascular diseases

Cardiovascular diseases (CVDs) are the leading cause of morbidity and mortality and have placed a significant burden on our society both regionally and globally. With the development of information technology, we are accumulating digitalized data at an exponential speed. This growing volume of existing data has provided researchers tremendous opportunities and resources to generate real-world evidence for healthcare decision making. Leveraging data from overwhelming numbers of available databases could effectively speed up the progress of scientific advances and provide a significant impact on clinical practice and public health. Despite that plenty of resources and efforts have been put to address health-related questions, countless unanswered questions remain and new hypotheses keep arising with the advance in healthcare system. Many of them can be addressed by epidemiologic studies using the available data without the primary data collection. Meanwhile, the challenges CVD researchers are facing when conducting research using existing data are less frequently discussed. In the thesis, I aim to address several clinical epidemiologic research questions and controversies in the field of cardiovascular diseases using available data, and discuss the challenges during the research process from a data user’s perspective to raise the awareness and to stimulate discussions. Chapter 2 is a diagnostic and community-wide study that evaluated the feasibility of using administrative health data to identify peripheral arterial disease (PAD) patients. The results show that using administrative data to identify PAD patients was highly specific but not sensitive in the community. Chapter 3 is a systematic review that compared and summarized the admission rates and length of stay (LOS) for venous thromboembolism (VTE) patients with different anticoagulant therapies. We found that the admission rates were lower and the LOS was shorter comparing low molecular weight heparin (LMWH) to unfractionated heparin (UFH) and comparing oral therapy to parenteral therapy for acute VTE in RCTs. These crucial clinically relevant outcomes were underreported in the existing VTE clinical trials. Chapter 4 is a retrospective and population-based cohort study that measured the association between chronic obstructive pulmonary disease (COPD) and angiographically diagnosed CAD. COPD was found to be negatively associated with CAD with and without adjustment for classic cardiovascular risk factors. Chapter 5 is a case-crossover study that evaluated the association between acute atmospheric pressure changes and the occurrence of ST-Elevation Myocardial Infarction (STEMI). We found that there was no association between acute air pressure changes and the onset of STEMI 1 to 6 days after the exposure, while there was a higher likelihood of STEMI occurrence for patients experiencing acute air pressure decrease in a lag time of 7 days. This thesis shows that reutilizing existing data is time-saving and economical to generate evidence for important clinical epidemiologic research questions in the field of cardiovascular diseases despite some practical challenges. Timely access to data, lacking detailed data documentation, establishing causality, data conversion, data consistency, and data validity are the main challenges for CVD researchers conducting epidemiologic research using existing data. Overcoming these challenges would streamline and speed up the research process for investigators conducting epidemiologic research.