Week 7: More results

As the summer immersion is coming to an end, I spent a lot of time trying to finish my research project and reading through the literature to understand the clinical significance of my results.

Typically, transcatheter aortic valve implantation (TAVI) is recommended to patients with aortic stenosis that are too at risk for surgery but is contraindicated to patients with left atrial thrombosis (LAT). LAT was frequently found associated with an increase risk of cardioembolic events and, in many clinical trails, was an exclusion criteria for TAVI. In other words, patents with LAT have blood clots in their left atrial appendage which can break free during the intervention and provoke an embolic stroke. LAT patients outside clinical trails, however, have still undergone TAVI with generally positive outcomes. Data about those patients have been collected by my mentor and his team. The hypothesis of my research is that the mortality of TAVI patients may not due mainly to cardioembolic events. If true, surgeons could safely perform TAVI on LAT patients over an invasive surgery.

As I explained last week, I did a series of univariate/multivariate logistic regression to model which clinical parameter contributes to 30 day or 1 year mortality of TAVI patients. I first compared patients with LAT against patients with SEC (cf. week 6) and patients with neither LAT and/or SEC and demonstrated that only pre-procedural atrial fibrillation (AF), the ejection fraction (a measurement of how efficient a heart if pumping blood out), and the mortality at 1 year were significantly different between the three groups. Then, I did a multivariate logistic regression for 30 day mortality and showed that patients with pre-procedural AF have more than 12 times the odds of dying (odd ratio = 12.5). Patients with aortic insufficiency equal or greater than moderate and life threatening bleeds (as defined by cardiologist) also have about 7 to 20 times the odds of dying after 30 days. Similarly, 1 year mortality can be significantly explained by pre-procedural AF, the presence of thrombus, and life-threatening bleed (odd ratios of respectively 2, 3, 19).
Collectively, these results suggest that pre-procedural AF may be one of the main causes of mortality and that TAVI may be performed in patients with LAT.

I still have a couple of statistical testing to do next week. For instance, I want to estimate and compare patient survival curves using the Kaplan-Meier method and log-rank test, as well as model overall mortality using a Cox regression.