Clinical Projects

Clinical Projects

Epidemiology of Left Heart Disease

PI: Pirooz Eghtesady, MD, PhD, St. Louis Children's Hospital

(Co-investigators: Madeleine Cunningham, PhD, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma; Matthew A. Hall, PhD, Senior Statistician, Child Health Corporation of America, Shawnee Mission, Kansas; Dr. Caroline Lee, Division of Cardiology, Department of Pediatrics, Washington University Medical Center  in St Louis, Missouri).

This research focuses on elucidating etiologic mechanisms for a group of heart defects characterized by involvement of the left side of the heart, including hypoplastic left heart syndrome (HLHS) and aortic stenosis.

Hypoplastic left heart syndrome (HLHS) is a severe and devastating heart defect that affects ~ 1 in 5-10,000 children born each year and accounts for 25% of all neonatal deaths from congenital heart disease. Much research and significant resources continue to be devoted to understanding HLHS and helping affected families.  However, the mechanisms leading to HLHS are unknown. Clinical observations and research studies suggest that HLHS exhibits complex inheritance with contributions from both genetic and environmental factors. On the basis of these studies, we proposed a hypothesis that HLHS occurs via a mechanism involving an immunologic reaction of mom to a common infectious etiology that subsequently affects her unborn baby. In the simplest terms, the hypothesis proposes that these heart defects represent manifestation of a form of rheumatic heart disease in the genetically susceptible fetus. Much circumstantial evidence has been gathered from clinical data in support of the hypothesis, and this NIH-funded study is in progress to further test the hypothesis.  Subsequently, and perhaps more significantly, it might be feasible to prevent or treat the heart defects early during pregnancy.

Quality Improvement (QI) Initiatives for Pediatric Cardiac Surgery

PI: Pirooz Eghtesady, MD, PhD

The objective of these studies is to first, prioritize QI efforts in pediatric cardiac surgery (procedure specific), then second, to focus on what measures or care practices lead to better or worse outcomes (comparative effectiveness research).

A QI initiative to reduce surgical site infection (SSI) rates was created 5 years ago to identify risk factors associated with pediatric cardiac surgery. A multi-disciplinary QI initiative was set up to identify key drivers and devise outcome measures using the Centers for Disease Control and Prevention criteria for SSI counts. Using Failure Mode and Effects Analysis (FMEA), we implemented interventions and tracked data using control charts. We then developed a bundle of interventions focused on modifiable risk factors specific for pediatric cardiac patients to prevent post-operative SSI. Ongoing studies are being conducted using this QI improvement methodology.

Targeted QI improvement in Pediatric Cardiac Surgery

PI: Pirooz Eghtesady, MD, PhD

(Collaborator: Matt Hall, PhD, Children's Hospital Association, Overton Park, KS).

The goal of this study is to analyze outcome data related to commonly performed procedures in pediatric cardiac surgery.

Retrospective analyses of data on pediatric cardiac surgery patients in the Pediatric Health Information Systems (PHIS) are being performed to determine prioritization and for targeting QI efforts. In addition, the study is assessing current risk stratification systems in pediatric cardiac surgery. Dr. Eghtesady is a member of The Society of Thoracic Surgeons (STS) committee for the Access and Publications Task Force that uses the STS National Databases to evaluate and improve outcomes for congenital heart surgery patients. In collaboration with the Duke Clinical Research Institute, the data extracted are analyzed, which is critical to support QI initiatives and patient safety.  

Systemic to pulmonary artery shunting (BT) is a common palliative procedure used in children with congenital heart defects. The optimal postoperative anticoagulation regimen for this procedure is unknown. The PHIS database was retrospectively queried to identify patients (<30 days old) with isolated BT shunt procedures. The primary end point was in-hospital mortality; secondary endpoints included the need for shunt revision, postoperative catheter-based intervention and extra-corporal membrane oxygenation support. Postoperative anticoagulation regimens were compared and Cox proportional hazard analysis was used to identify predictors of adverse events after surgery. Our studies showed that the anticoagulation regimen was the most important predictor of a complication. Our results demonstrate the role for Aspirin, particularly when initiated early, in lowering rates of adverse events in the immediate perioperative period after BT shunt surgery.

Machine Learning of Best Practice Patterns in Pediatric Cardiac Surgery

PI: Pirooz Eghtesady, MD, PhD

(Collaborators: Sanmay Das, PhD, Department of Computer Science and Engineering, Washington University in St. Louis, MO, and Matt Hall, PhD, Principal Biostatistician, Children’s Hospital Association, KS)

The goal of this collaborative venture is to use Machine Learning to apply algorithms to pediatric administrative health databases to improve decision-making for pediatric cardiac surgery.

Our long-term aim is to improve outcomes for children who require congenital heart surgery, through identifying and learning the best practice patterns underlying complex decision-making: a tool for indirectly evaluating “judgment” in the clinical setting. Based on large amounts of data from prior clinical outcomes in children who have undergone complex surgeries for congenital heart disease, we are applying Machine Learning to develop algorithms to determine best practice patterns. Data is abstracted from the PHIS database with pharmacy, clinical, and imaging resource utilization data incorporated into the learning algorithms. By using standard machine learning classifiers and feature scoring techniques, we are able to achieve high accuracy and identify several factors that are correlated with a positive outcome.