|
Activity Number:
|
440
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305450 |
|
Title:
|
Distributions of Pediatric Blood Pressure Index and Percentiles: Implications as Diagnostic Tools
|
|
Author(s):
|
Cynthia Bell*+ and Alisa Acosta and Karen McNiece and Ronald Portman and Joshua Samuels
|
|
Companies:
|
The University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston and University of Arkansas Medical School/Arkansas Children's Hospital and The University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston
|
|
Address:
|
Division of Pediatric Nephrology and Hypertension, Houston, TX, 77030,
|
|
Keywords:
|
reference values ; percentiles ; blood pressure ; distribution ; hypertension ; skewed
|
|
Abstract:
|
Blood pressure (BP) reference values defining hypertension in children are set at gender-, age-, and height-specific limits, generally the 95th percentile. Current published reference values are based on parametric models that assume normally distributed BP without correcting for skewness. To evaluate a patient's hypertension severity, clinicians frequently use 2 classification indicators: relative BP percentile or BP index. If BP was normally distributed, these indicators would have a uniform and normal distribution, respectively. The violation of BP normality assumptions yields skewed distributions for the indicators hindering their diagnostic interpretation. We use data from a school-based BP screening of 6,390 children to demonstrate the effect of non-normally distributed BP on the 2 diagnostic indicators and determine if T.J. Cole's LMS transformation is a feasible smoothing method.
|
