A Computable Phenotype Improves Cohort Ascertainment in a Pediatric Pulmonary Hypertension Registry

Abstract

Objectives: To compare registry and electronic health record (EHR) data mining approaches for cohort ascertainment in patients with pediatric pulmonary hypertension (PH) in an effort to overcome some of the limitations of registry enrollment alone in identifying patients with particular disease phenotypes.

Study design: This study was a single-center retrospective analysis of EHR and registry data at Boston Children's Hospital. The local Informatics for Integrating Biology and the Bedside (i2b2) data warehouse was queried for billing codes, prescriptions, and narrative data related to pediatric PH. Computable phenotype algorithms were developed by fitting penalized logistic regression models to a physician-annotated training set. Algorithms were applied to a candidate patient cohort, and performance was evaluated using a separate set of 136 records and 179 registry patients. We compared clinical and demographic characteristics of patients identified by computable phenotype and the registry.

Results: The computable phenotype had an area under the receiver operating characteristics curve of 90% (95% CI, 85%-95%), a positive predictive value of 85% (95% CI, 77%-93%), and identified 413 patients (an additional 231%) with pediatric PH who were not enrolled in the registry. Patients identified by the computable phenotype were clinically distinct from registry patients, with a greater prevalence of diagnoses related to perinatal distress and left heart disease.

Conclusions: Mining of EHRs using computable phenotypes identified a large cohort of patients not recruited using a classic registry. Fusion of EHR and registry data can improve cohort ascertainment for the study of rare diseases.

Trial registration: ClinicalTrials.gov: NCT02249923.

Keywords: bioinformatics; computer-based model; pediatrics; pulmonary hypertension; registry.

Figure 1. Overview of our EHR-based approach for cohort ascertainment

Figure 2. Venn diagram showing computable phenotype- and registry-based cohorts

The cohorts used for subsequent comparisons are labeled in small caps.

Figure 3. Comparison of ICD-9 coding practices between the computable phenotype and registry cohorts

Bars represent percent of patients who had each ICD-9 code recorded at least once. P < 0.0001 (****), P < 0.001 (***), P < 0.01 (**) for pairwise comparisons between cohorts.

Figure 4. ICD-9 codes with greater relative frequency among patients identified only in the EHR as compared with patients enrolled in the registry

Magenta lines indicate ICD-9 codes related to heart disease and green lines indicate ICD-9 codes related to neonatal distress. Trivial diagnoses are shown in light grey to emphasize diagnoses of interest. Only conditions with P < 1.5 are shown.

Figure 5. ICD-9 codes with greater relative frequency among registry patients as compared with patients identified only in the EHR

Magenta lines indicate ICD-9 codes related to gastrointestinal disease and malformations, blue lines indicate ICD-9 codes related to chronic respiratory disease, and green lines indicate ICD-9 codes related to prematurity, developmental disorders, and seizures. Trivial diagnoses are shown in light grey to emphasize diagnoses of interest. Only conditions with P < 0.05 are shown. Some ICD-9 code descriptions appear twice due to unique ICD-9 codes mapping to similar descriptions.