Key result
EHR-derived pediatric BP percentiles align closely with 2017 AAP guideline standards.
Why the study?
Identification of abnormal BP in children requires normative data, prompting evaluation of the feasibility of using real-world EHR office BP data to generate paediatric BP percentiles.
Does a mixed-effects polynomial regression model applied to real-world paediatric electronic health record blood pressure data generate age-, sex-, and height-specific blood pressure percentiles consistent with the 2017 American Academy of Paediatrics Clinical Practise Guidelines in normal-weight children aged 3-17 years?
Cohort (n=292,412)
Yes
Does a mixed-effects polynomial regression model applied to real-world paediatric electronic health record blood pressure data generate age-, sex-, and height-specific blood pressure percentiles consistent with the 2017 American Academy of Paediatrics Clinical Practise Guidelines in normal-weight children aged 3-17 years?
Real-world electronic health record data from nearly 300,000 normal-weight children can feasibly generate contemporary paediatric blood pressure percentiles that demonstrate strong agreement with the 2017 AAP Clinical Practice Guidelines, though slightly higher systolic values would reclassify 11.2% of children from ≥95th to <95th percentile.
Answers are grounded in this paper and cite their sources.
Supports EHR-derived pediatric BP percentiles as feasible; leaves open prospective outcome validation before clinical adoption.
BACKGROUND: Identification of abnormal blood pressure (BP) in children requires normative data. We sought to examine the feasibility of using "real-world" office BP data obtained from electronic health records (EHR) to generate age-, sex- and height-specific BP percentiles for children. METHODS: Using data collected 01/01/2009-8/31/2021 from eight large children's healthcare organisations in PEDSnet, we applied a mixed-effects polynomial regression model with random slopes to generate Z-scores and BP percentiles and compared them with currently used normative BP distributions published in the 2017 American Academy of Paediatrics (AAP) Clinical Practise Guidelines (CPG). FINDINGS: We identified a study sample of 292,412 children (1,085,083 BP measurements), ages 3-17 years (53% female), with no chronic medical conditions, who were not overweight/obese and who were primarily seen for general paediatric care in outpatient settings. Approximately 45,000-75,000 children contributed data to each age category. The PEDSnet systolic BP percentile values were 1-4 mmHg higher than AAP CPG BP values across age-sex-height groups, with larger differences observed in younger children. Diastolic BP values were also higher in younger children; starting with age 7 years, diastolic BP percentile values were 1-3 mmHg lower than AAP CPG values. Cohen's Kappa was 0.90 for systolic BP, 0.66 for diastolic BP, and 0.80 overall indicating excellent agreement between PEDSnet and 2017 AAP CPG data for systolic BP and substantial agreement for diastolic BP. INTERPRETATION: Our analysis indicates that real-word EHR data can be used to generate BP percentiles consistent with current clinical practise on BP management in children. FUNDING: Funding for this work was provided by the Preserving Kidney Function in Children with Chronic Kidney Disease (PRESERVE) study; Patient-Centred Outcomes Research Institute (PCORI) RD-2020C2020338 (Principal Investigator: Dr. Forrest; Co-Principal Investigator: Dr. Denburg).
No takes yet. Share an insight, caveat, or question.
Mitsnefes et al. (2023) conducted a cohort in Pediatric blood pressure normative distribution (n=292,412). EHR-derived mixed-effects polynomial regression BP percentiles (PEDSnet) vs. 2017 American Academy of Pediatrics Clinical Practice Guidelines (AAP CPG) normative BP percentiles was evaluated on Agreement in blood pressure percentile distributions and classification between PEDSnet EHR-derived values and 2017 AAP CPG reference values (Cohen's Kappa 0.80 overall (0.90 systolic, 0.66 diastolic)). EHR-derived pediatric BP percentiles showed excellent agreement with 2017 AAP CPG standards for systolic BP and substantial agreement for diastolic BP.
Synapse has enriched 10 related papers. Consider them for comparative context: