Differences in Blood Pressure Levels Among Children by Sociodemographic Status

Melissa Goulding, MS; Robert Goldberg, PhD; Stephenie C. Lemon, PhD


Prev Chronic Dis. 2021;18(9):e88 

In This Article


Study Design and Database

Our cross-sectional study used nationally representative data from the National Health and Nutrition Examination Survey (NHANES),[13] which is collected biennially by the National Center for Health Statistics to provide data on the health status of community-dwelling US residents. NHANES collects sociodemographic, dietary, and general health information by survey and medical, dental, and laboratory data by physical examination. We used data from 2011–2018, which consists of 4 biennial cycles. Unweighted survey response rates ranged from 53.6% to 78.5% for our study sample. Additional adjustments to weighting procedures were used to reduce the potential effects of response bias resulting from a lower response rate in the 2017–2018 NHANES cycle.[13] NHANES data collection is approved by the National Center for Health Statistics Research and Ethics Review Board. Participant and parental consent were obtained for children aged 13 years or older. Participant assent and parental consent were obtained for children aged 7 to 12 years.

Study Population

NHANES BP data comes from physical examinations.[13] For our study we included children aged 8 to 17 years for whom data on BP, height, weight, race/ethnicity, and socioeconomic characteristics were available. We excluded children who were missing BP measurements (n = 338), had fewer than 3 BP readings (n = 68), were missing data on body mass index (BMI) (weight in kg/height in m2) (n = 32), or were missing data on sociodemographic characteristics (n = 702). The final sample included 5,971 children, weighted to represent 36,612,323 children. To provide biennial prevalence estimates of hypertensive and elevated BP, the sample was defined by NHANES cycle. We used the entire sample for prevalence estimates of various BP parameters and differences in these end points according to sociodemographic factors.

Operational Definition of Pediatric Elevated and Hypertensive BP. Although clinical diagnosis of hypertension requires BP measurement across at least 3 occasions, NHANES is limited to physical examination on 1 occasion. Therefore, 3 BP measurements taken on a single occasion were averaged for each child in accordance with AAP guidelines for clinicians and common practice in pediatric hypertension studies.[1,5,7–9] NHANES BP measurement techniques have been described previously.[13] For children aged 8 to 12 years, we used age, sex, and height to determine their BP percentile according to the 2017 AAP BP tables. BP percentiles (for children aged <13 y) or average measurement (for children aged 13–17 y) were then used for categorization according to 2017 AAP guidelines. Elevated BP was defined as ranging from ≥90th percentile to <95th percentile or 120/<80 mm Hg to <95th percentile (whichever is lower) for children aged 8 to 12 years and 120/<80 to 129/<80 mm Hg for those aged 13 to 17 years. Hypertensive BP was defined as a BP percentile of ≥95 or an average BP of ≥130/80 mm Hg (whichever was lower) for children aged 8 to 12 years and ≥130/80 mm Hg for those aged 13 to 17 years.

Body Mass Index Percentile. Children's standing height and weight were measured by trained professionals during the NHANES physical examination, and their BMI was calculated. Methods and equipment used for anthropometric measures have been described previously.[14] We determined BMI percentiles according to the Centers for Disease Control and Prevention 2000 growth charts.[15] Weight status was categorized by BMI percentile to represent healthy weight (BMI percentile <85), overweight (BMI percentile ≥85 to <95), and obesity (BMI percentile ≥95). For adjusted prevalence estimates, we dichotomized weight to indicate unhealthy weight status (BMI percentile ≥85).

Sociodemographic Factors Associated With Elevated and Hypertensive BP. Age at the time of the NHANES physical examination was determined by the child's date of birth and was stratified at 8 to 9 years, 10 to 12 years, 13 to 15 years, and 16 to 17 years. Sex was determined by self-report with options of male or female. We used the more inclusive NHANES race/ethnicity variable in which children who identified as Mexican American were coded as such, those who identified as Hispanic or Latino were coded as other Latino, and those who identified as non-Latino were coded according to self-reported race of White, Black, Asian, or other (American Indian or Alaska Native, Native Hawaiian or Pacific Islander, mixed race).

We used 2 proxy measures for socioeconomic status, family poverty income ratio (PIR) and parent/guardian education level. PIR was calculated by dividing family income by the Department of Health and Human Services' poverty guidelines and then categorized as low (PIR <1.3), medium (PIR ≥1.3 and <3.5), and high (PIR ≥3.5). This categorization was used to be consistent with past obesity-related research and because a PIR of <1.3 is often used to determine eligibility for federally funded programs, including the Supplemental Nutrition Assistance Program.[16] Parent/guardian education level was measured as the highest education of the household reference person, who was the first person listed in the household aged 18 years or older who owned or rented the residence.

Statistical Analysis

We computed frequencies on our study sample. Because each of the continuous variables had nonnormal distributions (assessed via Shapiro–Wilk test), medians with interquartile range were calculated. Prevalence estimates of elevated and hypertensive BP were computed for the 2011–2018 period overall and by 4 biennial cycles. We estimated crude prevalence differences and weight status (BMI percentile ≥85) adjusted prevalence differences with 95% CIs for elevated and hypertensive BP for each sociodemographic subgroup through log binomial regression with the identity link.[17] Each sociodemographic factor was assessed separately. Models were then adjusted for weight status. Assessment of correlations between weight status and each sociodemographic variable suggested adjusted models were not collinear. All analyses were appropriately weighted and analyzed with examination sample weights and Taylor series linearization[13] accounting for the complex sampling design of NHANES.