Using an Electronic Medical Record to Evaluate Pediatric Blood Pressure Follow-Up: Limitations of Automated Data Extraction in a Nephrology Cohort

Abstract

Hypertension is a prevalent condition in the pediatric population. Diagnosis and management can be challenging. The widespread adoption of electronic medical records (EMRs) has revealed their potential for improving patient care and research. This study used EMR data from the nephrology clinic at a single center to identify, evaluate, and monitor children with elevated blood pressure. Specifically, we attempted to assess whether documented clinical management following elevated blood pressure readings in clinic could be accurately identified using automated EMR-derived data, including whether a follow-up visit was scheduled, if the patient was sent back to their family physician, if a diagnosis of white-coat hypertension was made, or if pharmacological or non-pharmacological management for hypertension was initiated. We performed a single center, retrospective cohort study using EMR data. Children who had their initial referral visit at the nephrology clinic between January 1st, 2018, and January 1st, 2022. Outpatients were identified using ICD-10 codes related to nephrology diseases. The EMR was reviewed to extract anthropometric and blood pressure (BP) data. A BP index was calculated using systolic and diastolic BP values and the 2017 American Academy of Pediatrics (AAP) hypertension guidelines. Supplemental chart review was performed only for patients with incomplete follow-up documentation following automated data extraction to assess concordance between EMR-derived data and documented clinical management. The primary analysis examined EMR data to determine whether elevated BP was appropriately managed based on available documentation, including scheduling follow-up visits, diagnosing white-coat hypertension, or initiating pharmacological or non-pharmacological interventions. A total of 1,469 children aged 1–18 (median age 9.8 years) were newly referred to the nephrology clinic with complete data for BP index calculation. Many children (71/147 after visit 2, and 10/44 after visit 3) who initially had elevated BP were normotensive on their subsequent visits. Among patients with elevated BP across multiple visits, a substantial proportion had missing or incomplete longitudinal data following EMR extraction (∼20%). In addition, among patients with elevated BP after the first visit, approximately 30% did not have subsequent follow up visits documented in the EMR. Targeted manual chart review revealed that a substantial proportion of these patients had documented follow-up or BP-related management that was not captured through automated EMR extraction. Children presenting with isolated elevated diastolic BP elevations had fewer BP measurements and were less likely to have BP-related follow up, potentially reflecting the perceived benign nature of this phenomenon. This study’s retrospective, non-randomized design limits generalizability. This study identifies important challenges in using EMR-derived data to study pediatric hypertension, particularly with respect to incomplete longitudinal capture and insufficient BP measurements. By integrating EMR-derived data with targeted manual chart review, we demonstrated that a substantial proportion of appropriate clinical management was not captured through automated data extraction. This underscores the importance of validating EMR-based analyses when assessing clinical care processes, as highlighted in relation to BP.