Computable phENotype To Identify Pulmonary Embolism in chilDrEn (CENTIPEDE), Phase II
Variable Name
Study Dates
Last Modified
Tags
Publisher
Abstract
This phase II study is intended to further validate a computable phenotype (CP) developed in phase I which identifies a pediatric pulmonary embolism (PE) cohort from EHR. The accuracy of diagnostic and treatment code combinations using EHR data to diagnose pediatric PE was tested at a single center, the Children’s Hospital of Philadelphia. However, this CP had a high rate of false negatives in children with cardiac disease, likely due to differences in diagnosis and management. This study aims to validate the CP for PE at two additional PEDSnet centers and identify risk categories, develop a CP specific for PE in cardiac patients, and identify PE interventions and outcomes in PEDSnet. This work will allow for comparative effectiveness research and prospective studies using the CP for cohort identification.
Affiliation(s)
Funder(s)
Provenance
Description
The rate of pediatric pulmonary embolism (PPE), the most serious and potentially deadly form of venous thromboembolism, has increased by 200% over the last decade. But it remains rare at individual centers and consequently is challenging to study. Currently, the management of PPE is empiric and not derived from scientific evidence. Electronic health records (EHR) are a rich source of clinical information that can be used to study rare diseases provided a disease cohort is accurately identified. Previous studies in adults have revealed that using appropriate combinations of diagnostic and treatment codes leads to accurate identification of a PE cohort from EHR.
The purpose of this study is to identify the best combination of diagnostic and treatment codes using EHR data that can accurately diagnose PPE at a single center.
Study Aims
- Validate the previous PE CP in PEDSnet at Cincinnati Children’s Hospital Medical Center (CCHMC) and Nationwide Children’s Hospital (NCH). Compare CP predicted cases and predicted non-cases in PEDSnet to adjudicated chart review to determine performance characteristics (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV]). Risk stratify PE patients into low-risk, intermediate-risk, or high-risk at each center in PEDSnet using vital signs, lab values, and imaging results using current pediatric PE risk categories.
- Develop a new PE CP that is specific for pediatric patients with cardiac disease using CHOP, NCH, and CCHMC PEDSnet data. Test the new CP in CHOP data after adding additional diagnostic imaging and procedures, including echocardiogram and cardiac catheterization codes, and diagnostic codes for cardiac disease, then validate it in NCH and CCHMC data.
- Develop CPs for PE outcomes and interventions in CHOP, CCHMC, and NCH PEDSnet data. Using combinations of pre-selected code sets, develop CPs for outcomes (bleeding, mortality, length of stay, chronic thromboembolic pulmonary hypertension) and interventions (systemic and catheter directed thrombolysis). Assess each with sensitivity, specificity, PPV and NPV.
Study Design
This project is a retrospective study to validate a previously developed CP to accurately identify pediatric patients with acute PE at 2 additional PEDSnet centers (Aim 1), risk stratify PE patients using pediatric-adapted risk categories4 in CHOP data, by piloting NLP to interpret imaging reports for signs of right ventricle (RV) dysfunction or dilation (Aim 1), develop a CP for cardiac PE (Aim 2), and identify PE interventions and outcomes (Aim 3).
Development Code
Vocabulary
Clinical Subjects Headings
Creative Commons license
