This project was completed with Jessie Bierschenk, Leona Lu, and Shari Tian for the Fall 2023 section of STA 440: Case Study at Duke University. Due to privacy and security regulations, code source and data cannot be shared publicly. Please refer to our report at the bottom for details.
About
For diagnosing or treating heart problems, cardiac catheterizations are common procedures used by doctors. In this procedure, a thin, flexible tube (catheter) is guided through a blood vessel to the heart to diagnose (diagnostic catheterization) or treat (interventional catheterization) certain heart conditions, such as clogged arteries or irregular heartbeats; this method renders important informaton about the heart muscle, heart valves, and blood vessels in the heart. This research specifically explores the use of catheterizations as a clinical intervention in the context of heart disease and cardiac events among human patients. Even with interventional cardiology existing as a well-established field of medicine, access to such care and resources is not universal. Today, many heart conditions can be treated during an interventional heart catheterization that would have otherwise needed a separate, more intense surgery. We aim to use data from Duke Hospital, a well-established medical institute, to better inform doctors/hospitals of the factors to consider in the context of (interventional) catheterizations in the hopes of motivating an introduction or improvement of the practice.
Data
To gain a deeper understanding of variables associated with this procedure, we begin by analyzing a dataset that includes adult patients catheterized at Duke Hospital starting in 1985 through 2013, and only includes patients that have evidence of significant extant coronary artery disease. The data contain medical and demographic records for 39,098 unique patients, with one record per procedure. In total, there are 83,320 observations, each representing a unique procedure.
The data include records of procedures involving both diagnostic and interventional catheterizations. While almost all procedures in the dataset involve diagnostic catheterizations, only about 33.7% involve interventional catheterizations. A study done in 2010-2011 shows that within approximately 85% of the cardiac catheterization laboratories in the United States, “on-site cardiac surgery was not available in 83% of facilities performing fewer than 200 interventional catheterizations annually, with these facilities representing 32.6% of the facilities reporting, but performing only 12.4% of the interventional catheterizations in this data sample.” This finding highlights a potential lack of interventional catheterization procedures available as well as the limited amount of relevant research on interventional procedures. Motivated by this, our analysis aims to focus on analyzing patient survival rates associated with procedures involving an interventional catheterization vs. patient survival rates associated with procedures that did not involve an interventional catheterization, all the while controlling for and analyzing other factors associated with patient survival. By doing so, we hope to provide information to be used by doctors and hospitals that they can use to improve or introduce the practice of interventional catheterizations within their health services. This would allow clinicians and medical institutions to better isolate the association between interventional catheterization procedures and patient survival rates, thus motivating future efforts to reduce wasteful and inefficient care variations and implement effective, continuous quality improvement processes.
Objectives
Our primary (1) and secondary (2) research questions are as follows:
(1) When controlling for other patient/procedure characteristics, what is the association between patient survival and the use of interventional coronary catheterizations?
(2) In regards to patient data that a physician can access on the day of a given procedure, what factors show evidence for an association with patient survival?
In addition to informing the use of interventional catheterizations across healthcare systems, our research aims to answer these questions in order to give physicians a more comprehensive understanding of appropriate or optimal procedures for a given patient. This means that we want to synthesize information to inform physicians about past research on survival rates of patients receiving an interventional catheterization while controlling for other variables. With this, we will be focusing on variables associated with information that a physician can access about a patient in real-time and not on information that can only be gathered retrospectively; these variables will be our predictor variables.