Childhood Blood Pressure Trends and Risk Factors for High Blood PressureNovelty and Significance
The NHANES Experience 1988–2008
The obesity epidemic in children makes it plausible that prevalence rates of elevated blood pressure (BP) are increasing over time. Yet, previous literature is inconsistent because of small sample sizes. Also, it is unclear whether adjusting for risk factors can explain longitudinal trends in prevalence of elevated BP. Thus, we analyzed a population-based sample of 3248 children in National Health and Nutrition Examination Survey (NHANES) III (1988–1994) and 8388 children in continuous NHANES (1999–2008), aged 8 to 17 years. Our main outcome measure was elevated BP (systolic BP or diastolic BP ≥90th percentile or systolic BP/diastolic BP ≥120/80 mm Hg). We found that the prevalence of elevated BP increased from NHANES III to NHANES 1999–2008 (Boys: 15.8% to 19.2%, P=0.057; Girls: 8.2% to 12.6%, P=0.007). Body mass index (Q4 versus Q1; odds ratio=2.00; P<0.001), waist circumference (Q4 versus Q1; odds ratio=2.14; P<0.001), and sodium (Na) intake (≥3450 mg versus <2300 mg/2000 calories; odds ratio=1.36; P=0.024) were independently associated with prevalence of elevated BP. Also, mean systolic BP, but not diastolic BP, was associated with increased Na intake in children (quintile 5 [Q5] versus quintile 1 [Q1] of Na intake; β=1.25±0.58; P=0.034). In conclusion, we demonstrate an association between high Na intake and elevated BP in children. After adjustment for age, sex, race/ethnicity, body mass index, waist circumference, and sodium intake, odds ratio for elevated BP in NHANES 1999–2008 versus NHANES III=1.27, P=0.069.
- blood pressure
- body mass index
- National Health and Nutrition Examination Survey
- nutrition surveys
- waist circumference
See Editorial Commentary, pp 242–243
There has been an epidemic of obesity in the past 20 years among both children and adolescents.1 Also, sodium intake has been high in both children and adults with a majority of children above the Reference Daily Intake (RDI).2 Because body mass index (BMI) and sodium intake are important risk factors for hypertension in adults,3 it is reasonable to consider whether there have been corresponding increases in prevalence of elevated blood pressure (BP) in children. Because National Health and Nutrition Examination Survey (NHANES) III and continuous NHANES have the same BP measurement protocol, are representative of the general US population, and are of adequate size, we use these populations as the study sample for the current report.
There have been 3 previous studies comparing mean BP levels and prevalence of hypertension and prehypertension in children between NHANES III and continuous NHANES. Muntner et al4 compared BP levels between NHANES III (1988–1994) and NHANES 1999–2000 and found significant differences between surveys in both mean systolic BP (SBP) and diastolic BP (DBP) after adjusting for BMI. Din-Dzietham et al5 compared prevalence of prehypertension and hypertension between NHANES III and NHANES 1999–2002. Ostchega et al6 compared prevalence of prehypertension and hypertension between NHANES III and each of NHANES 1999–2002 and NHANES 03 to 06. In both cases, prevalence was higher in continuous NHANES, more frequently statistically significant for prehypertension than hypertension. All 3 studies used BP percentiles based on Pediatric Task Force Standards.7
One recurring theme of the previous literature is small sample sizes for estimates of hypertension prevalence. In this article, we estimated percentiles using norms based on normal-weight children8 rather than the Pediatric 2004 report,7 which included both normal-weight and overweight children. This resulted in higher rates of hypertension and prehypertension. Also, to maximize power, we focus on elevated BP=either hypertension or prehypertension defined based on a normal weight population. Second, the previous analyses established that mean BP was increasing over time, and that increasing BMI was associated with some of the increase. In the current article, we also look at possible mediating effects of (1) central obesity based on waist circumference, and (2) other dietary factors that have been associated with BP in previous studies.
We use data from NHANES III (1988–1994) and continuous NHANES (NHANES 1999–2000, 2001–2002, 2003–2004, 2005–2006, 2007–2008), subsequently referred to as the NHANES 1999 to 2008 population. To be eligible for the study population, a child had to be aged 8 to 17 years (96–215 months), and have ≥1 SBP and 1 DBP measurement. A mean of 3 BP readings was obtained. If <3 readings were available, then the mean of all available readings was used. All BP measurements were obtained with a sphygmomanometer by certified examiners after children rested quietly while sitting for 5 minutes.9,10 BMI values were converted to age-sex–specific percentiles based on Center for Disease Control growth charts,11 and converted to BMI z scores using the probit transformation. Children with outlying BMI z scores (defined as <−6.0 or >6.0) were deleted. Waist circumference z scores were computed by ranking subjects by 1-year age-sex groups and using the probit transformation. Children of self-reported non-Hispanic white, black, or Mexican-American race/ethnicity were included. The numbers of children of other ethnicities were small and were not included.
Determination of BP Percentiles
We used BP percentiles based on normal weight children derived from cubic spline and quantile regression methods to (1) provide for more flexible models to express BP as a function of age and height over the entire pediatric age range and (2) relax the assumption of normality in defining percentiles.8 These percentiles both in tabular form for assessment of BP percentiles for individual children and in a Statistical Analysis System (SAS) macro for assessment of BP percentiles in batch mode for large numbers of children are available at the following website: http://sites.google.com/a/channing.harvard.edu/bernardrosner/pediatric-blood-press.
To increase power, we focus on the prevalence of elevated BP (either hypertension or prehypertension)=either SBP or DBP ≥90th percentile or SBP≥120 mm Hg or DBP≥80 mm Hg.
In all analyses, we have used sampling weights provided by NHANES to estimate prevalence of elevated BP in a representative sample of the US pediatric population. Descriptive statistics were obtained from proc surveymeans and proc surveyfreq of SAS 9.2, with standard errors accounting for sampling weights separately for NHANES III and NHANES 1999–2008, and compared using a z statistic. Nutrient intake was measured by a single day of 24-hour recall, and expressed in categorical format compared with the RDI based on nutritional guidelines for children.12 Although 2 days of 24-hour recall was available for NHANES 1999–2008, we only used the first day for comparability with NHANES III, where only a single 24-hour recall was available. In addition, sodium (Na) intake was adjusted for total caloric intake=(Na intake)×2000/total caloric intake and expressed in quintiles. Recommended caloric intake for children corrected for age and sex13 was obtained. Children were excluded from nutrient analyses if their actual caloric intake was <0.5 recommended caloric intake or >2 recommended caloric intake for their age-sex norms. Overall, there was an initial sample of 16 693 children aged 8 to 17 years, of whom 1226 children (7%) were of other races, 1515 (9%) were missing either SBP or DBP, 129 (1%) were either missing BMI or had outlying BMI, 158 (1%) were missing waist circumference, and 2029 children (12%) had either missing or out of range caloric intake. This left a study sample of 3248 children in NHANES III and 8388 children in NHANES 1999–2008. Associations between the prevalence of elevated BP and each of 9 nutrients were assessed after adjustment for sex, age, race/ethnicity, BMI z score (in quartiles), and waist circumference z score (in quartiles), separately by study and then combined using proc surveylogistic of SAS. The prevalence of elevated BP was compared between the NHANES 1999–2008 and NHANES III population by adding an indicator variable for study and study×gender after adjustment for selected sets of covariates. In addition, the association between mean SBP and DBP (as continuous variables) and adjusted Na intake (expressed in quintiles) was assessed after adjustment for the above variables using proc surveyreg of SAS.
The demographic characteristics, anthropometric characteristics, and BP levels of the study populations are presented in Table 1. For both boys and girls, there has been a shift in the ethnic distribution between studies with a significant decrease in the percentage of non-Hispanic whites (P<0.05) and a significant increase in the percentage of Mexican Americans (P<0.01). There were significant increases in weight and BMI for both boys (P≤0.016) and girls (P<0.001). The percentage of overweight children (≥85th percentile) also significantly increased (P<0.001), and there were also large increases in waist circumference for boys and especially for girls (P<0.001).
Mean SBP significantly increased for both boys (106.1 versus 107.8 mm Hg; P=0.001) and girls (102.3 versus 104.9 mm Hg; P<0.001). However, mean DBP significantly increased among girls (57.0 versus 59.0 mm Hg; P=0.003), but not among boys (57.7 versus 56.7 mm Hg; P=0.13). The prevalence of elevated BP significantly increased among girls (8.2% versus 12.6%; P=0.007), but was only of borderline significance among boys (15.8% versus 19.2%; P=0.057).
We compared nutrient intake between surveys in relation with the RDI using the percentage of children who are above the RDI for specific nutrients by survey and sex (Table 2). For total fat, saturated fat, and protein, a large majority of children (70% to 80%) were above the RDI, with a slight decline over time. Correspondingly, there was an increase in the percentage of children above the RDI for carbohydrate intake. Approximately 45% to 50% of boys and 30% of girls were above the RDI for calcium at both surveys. Less than 11% of boys and 8% of girls were above the RDI for each of fiber, magnesium, and potassium, which declined slightly over time. Over 80% of children were above the RDI for Na at both surveys. However, the percentage of children >50% over the RDI (ie, >3450 mg per 2000 calories) declined significantly for both boys (38% versus 31%; P=0.012) and girls (40% versus 31%; P<0.001). Total caloric intake declined slightly for boys (mean 2349 versus 2255 Kcal; P=0.011), but did not change for girls (mean 1868 versus 1887 Kcal; P=0.51).
A majority of children were above the RDI for total fat, saturated fat, protein, and Na (Table 2). Hence, to maximize statistical power, we represented these nutrients in 3 categories (<RDI(ref)/≥RDI, ≤1.5RDI/>1.5RDI). Also, a majority of the children were below the RDI for calcium, carbohydrates, fiber, magnesium, and potassium. Hence, for these nutrients, we used the categories (>RDI(ref)/>(2/3)RDI, ≤RDI/≤(2/3)RDI). In all analyses, we controlled for age (continuous), male sex, race/ethnicity, and study (Table 3, model 1). In Table 3, models 2 to 4, we additionally adjust for BMI age-sex–specific z score (in quartiles) based on Center for Disease Control growth charts and waist circumference age-sex–specific z score (in quartiles) based on NHANES 1999–2008 and NHANES III, combined.
In model 1 (demographic adjusted), there were significant positive associations between prevalence of elevated BP and each of elevated Na intake (>3450 versus ≤2300 mg; odds ratio (OR)=1.37; P=0.017; P trend=0.038) and reduced carbohydrate intake (≤200 versus >300 g; OR=1.33; P=0.20; P trend=0.021). However, after adjustment for BMI z score and waist circumference z score (models 2–4), only the association with Na intake remained statistically significant (>3450 versus ≤2300 mg, OR=1.36, P=0.024; 2301–3450 mg, OR=1.17, P=0.21), P trend over all 3 Na groups=0.045.
In Table 4, we present a multivariate model concerning the association between elevated BP and other risk factors. The odds for elevated BP increased 12% for every 1-year increase in age (P<0.001), was higher for boys than girls (OR=1.85; P<0.001), and was higher for black than non-Hispanic white children (OR=1.28; P=0.002). Mexican-American children had no excess risk versus non-Hispanic white children (OR=0.99; P=0.92). The ORs for elevated BP for BMI z score quartiles 2 (Q2), 3 (Q3), and 4 (Q4) versus quartile 1 (Q1) were, respectively, 1.33 (P=0.094), 1.43 (P=0.024), and 2.00 (P<0.001). After controlling for BMI, the association between waist circumference z score and elevated BP was not monotone. Relative to Q1, there was no significant difference in the prevalence of elevated BP for Q2 (OR=0.98; P=0.90) or Q3 (OR=0.96; P=0.84). However, there was a highly significant increase for Q4 versus Q1 (OR=2.14; P<0.001) even after controlling for BMI z score quartile. Thus, both BMI and waist circumference made independent contributions to the prevalence of elevated BP. Finally, after adjusting for other risk factors, there was a significant increase in the prevalence of elevated BP between children with Na intake >3450 mg versus Na intake <2300 mg per 2000 calories (OR=1.36; P=0.024).
We also looked at effect modification of Na intake by the other variables in Table 4. There was no significant effect modification of Na intake by age, sex, BMI z score, or waist circumference z score. However, there was significant effect modification of Na intake by race/ethnicity (P=0.019). For non–black children, comparing risk for Na intake >3450 mg versus ≤2300 mg, OR=1.50 (95% confidence interval [CI], 1.08–2.07), P=0.016, P trend=0.028. However, for black children, OR=0.90 (95% CI, 0.70–1.16), P=0.40, P trend=0.26.
We now explore the association between SBP and DBP when represented as continuous variables and Na intake. To minimize the effect of outliers and not make the arbitrary assumption of a linear relationship between Na intake and BP, we categorized adjusted Na intake into quintiles and controlled for the same variables as in Table 4. The results are given in Table S1 in the online-only Data Supplement. There was a significant difference in mean SBP between Q5 (≥3754 mg/2000 kcal) and Q1 (≤2332 mg/2000 kcal; β±SE=1.246±0.577, P=0.034). For DBP, no significant effects were seen at any level of Na intake.
In Table 5, we compare the prevalence of elevated BP between NHANES 1999–2008 and NHANES III, both crudely and after adjusting for the covariates in Table 4. Overall, the crude prevalence of elevated BP was significantly higher in NHANES 1999–2008 versus NHANES III (model 1; OR=1.39, P=0.007). The association was virtually unchanged after adjusting for age, sex, and race/ethnicity (model 2; OR=1.38, P=0.009). However, after further adjusting for BMI z score quartile and waist circumference z score quartile (model 3), the association weakened and became only borderline significant (OR=1.25; P=0.089), reflecting the increase in obesity between NHANES III and continuous NHANES. After further adjusting for Na intake (model 4), the association strengthened slightly but remained only borderline significant (OR=1.27; P=0.069), reflecting the slight reduction in percentage of children >1.5 RDI (ie, >3450 mg/2000 calories) in NHANES 1999–2008 versus NHANES III (Table 2). The OR was somewhat stronger in girls (OR=1.43; P=0.12) than boys (OR=1.18; P=0.24); however, the difference was not statistically significant (P=0.44). Thus, overall roughly one third of the excess prevalence for NHANES 1999–2008 versus NHANES III is explained by differences in known risk factors between the 2 surveys.
In the present article, we focus on elevated BP=combination of hypertension and prehypertension. As noted by Din-Dzietham et al,5 prehypertension is clinically relevant because children whose BP is repeatedly ≥90th percentile exhibit signs of very early target-organ damage in young adulthood.14–16 Also, because NHANES does not have repeated BP measurements over time, it is not possible to make a diagnosis of hypertension.7
Second, to maximize precision, we compared NHANES III with NHANES 1999–2008 data rather than focusing on smaller time-periods within continuous NHANES. Third, we consider the independent contribution of both BMI and waist circumference as predictors of elevated BP. Finally, we considered dietary intake as an additional predictor of elevated BP. After adjusting for the above risk factors, we found a higher prevalence of elevated BP in NHANES 1999–2008 versus NHANES III, which was of borderline statistical significance (OR=1.27; P=0.069).
As with adults, the average dietary Na intake of children exceeds nutritional needs, is well above recommended levels, and has been progressively increasing.17 Although a significant relationship between high Na intake and hypertension is well established in adults, previous studies that examined associations between Na intake and BP levels in healthy children and adolescents report mixed or no relationships.2 In an older study, Cooper et al18 demonstrated a quantitatively weak but significant linear relation between BP and Na excretion in a sample of 73 children aged 11 to 14 years of age. These observations required 7 consecutive 24-hour urine collections per subject to adjust for intraindividual variation. More recently, He and MacGregor19 performed a meta-analysis of 10 published controlled clinical trials that investigated the effect of a reduction in Na intake on BP among children aged 6 to 15 years. Changes in salt intake were monitored from urinary Na excretion and, in some studies, Na intake from food diaries. The overall meta-analysis showed a significant reduction in both SBP (mean change=−1.17 mm Hg; 95% CI, −1.78 to −0.56), P<0.001, and DBP (mean change=−1.29 mm Hg; 95% CI, −1.94 to −0.65; P<0.001). In a study of BP sensitivity to Na, Rocchini et al20 measured BP in 60 obese and 18 nonobese adolescents after successive 2-week periods on a high-salt diet (>250 mmol of Na per day) and a low-salt diet (<30 mmol per day). The obese children had a significantly greater change (±SE) in mean arterial pressure (−12±1 mm Hg) than the nonobese children (+1±2 mm Hg), (P interaction <0.001) after the change from high Na to low Na intake.
Yang et al21 also considered the association between Na intake and BP level among children in NHANES 2003 to 2008. They found that there was a positive association between Na intake (as a continuous variable) and z score of SBP, but not DBP, after adjusting for age, sex, and height using Pediatric Task Force Standards,7 similar to the findings in the present report. In addition, there was a trend toward statistical significance between quartile of Na intake and elevated BP (combined prehypertension and hypertension) based on task force standards (P=0.062), which was significant when restricted to overweight/obese subjects (P=0.013). This observation of a stronger association of elevated BP with Na intake among overweight/obese subjects in a population study is consistent with earlier findings in the clinical study on Na sensitivity by Rocchini et al.20 In the present study, (1) we consider additional NHANES surveys yielding a sample size (n=11 636) roughly double that of Yang et al21 (n=6235); (2) use BP standards based on normal weight children, thus providing additional end points for elevated BP and a resulting increase in power; (3) restrict analyses to children with caloric intake between 0.5 and 2.0 times the recommended age-sex–specific caloric intake; (4) correct for both overall obesity (BMI) as well as central obesity (waist circumference); and (5) use an index of Na intake corrected for calories rather than age and race as in Yang et al21, which is important in looking at interactions of Na intake by ethnic group.
To our knowledge, our study represents the largest study of the effect of Na intake on the prevalence of elevated BP in children. After controlling for both overall and central obesity, we found a significant increase in the prevalence of elevated BP for children >1.5×RDI for Na (OR=1.36; 95% CI, 1.04–1.77; P=0.024) versus children with intake <RDI. Children with intake ≥RDI but ≤1.5 RDI had an OR=1.17 (95% CI, 0.92–1.49; P=0.21). Also, we found significant effect modification of Na intake by race, with a stronger association of elevated BP with Na for non–black children versus black children. However, this finding was unexpected and requires confirmation in other studies. None of the other nutritional risk factors considered was significantly associated with the prevalence of elevated BP either in multivariate analyses controlling for age, sex, race/ethnicity, BMI, and waist circumference (Table 3) or additionally adjusted for Na intake (data not shown). Finally, we found a significant effect of black (OR=1.28 versus non-Hispanic White), but not Mexican-American ancestry (OR=0.99 versus non-Hispanic White) after controlling for obesity and Na intake.
A limitation of the Na findings is that dietary intake was only assessed by a single 24-hour recall, and validation by 24-hour urinalysis was not possible. Nevertheless, 80% of children had a reported Na>RDI, which is consistent with previous literature.2 Also, we did not have enough power to assess lower levels of Na intake, because only 20% of children were below the RDI. Finally, although physical activity might be a relevant confounder for pediatric BP, it was only available in NHANES for children aged 12 to 17 years and hence was not used in our analyses.
We observed that the odds of elevated BP in children increased an estimated 27% between NHANES III and NHANES 1999–2008 (P=0.069), 2 surveys ≈12 years apart with identical BP protocols after accounting for differences in age, sex, ethnicity, BMI, waist circumference, and Na intake. We also observed an increase in the odds of elevated BP of 36% between children with Na intake >3450 mg (≥1.5 RDI) versus <2300 mg (<RDI) even after controlling for age, sex, race, BMI, and waist circumference. Furthermore, we observed a significant difference in mean SBP but not DBP between children in Na Q5 (≥3754 mg/2000 kcal) versus Na Q1 (≤2332 mg/2000 kcal), cut points that are similar to ≥1.5 RDI and <RDI, respectively. Largely because of secular changes in the food supply, dietary patterns, and dependence on processed foods, dietary Na intake has increased in the US population of children as well as adults. The findings in this report demonstrate an association between high Na intake and elevated BP in childhood and provide support to an Institute of Medicine Report on Strategies to Reduce Na Intake in the United States.22
We acknowledge the programming support of Marion McPhee. Dr Rosner prepared the article. Dr Cook provided statistical expertise and critically reviewed the article. We acknowledge Drs Daniels and Falkner for providing clinical expertise in pediatric hypertension and for critically reviewing the article. This study was approved by the Brigham and Women’s Hospital Institutional Review Committee. The subjects in NHANES gave informed consent.
Sources of Funding
This work was supported by National Institutes of Health (NIH) grant HL40619.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.111.00831/-/DC1.
- Received December 10, 2012.
- Revision received January 8, 2013.
- Accepted May 13, 2013.
- © 2013 American Heart Association, Inc.
- Falkner B,
- Michel S
- Din-Dzietham R,
- Liu Y,
- Bielo MV,
- Shamsa F
- Ostchega Y,
- Carroll M,
- Prineas RJ,
- McDowell MA,
- Louis T,
- Tilert T
- Rosner B,
- Cook N,
- Portman R,
- Daniels S,
- Falkner B
- 9.↵Department of Health and Human Services. Blood pressure section of the physician examination. NHANES 1999–2008. National Health and Nutrition Examination Survey. Available at: www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. Accessed June 12, 2013.
- 10.↵Department of Health and Human Services. National health and nutrition examination survey (NHANES) questionnaire and examination protocol. National Health and Nutrition Examination Survey. Available at: www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. Accessed June 12, 2013.
- 12.↵Reference daily intake. Wikipedia. 2011. Available at http://en.wikipedia.org/wiki/Reference_Daily_Intake. Accessed 2011.
- American Heart Association. Dietary Recommendations for Healthy Children.
- Jiang X,
- Srinivasan SR,
- Radhakrishnamurthy B,
- Dalferes ER Jr.,
- Bao W,
- Berenson GS
- Cooper R,
- Soltero I,
- Liu K,
- Berkson D,
- Levinson S,
- Stamler J
- He FJ,
- MacGregor GA
- Yang Q,
- Zhang Z,
- Kuklina EV,
- Fang J,
- Ayala C,
- Hong Y,
- Loustalot F,
- Dai S,
- Gunn JP,
- Tian N,
- Cogswell ME,
- Merritt R
- 22.↵Strategies to reduce sodium intake in the united states. The National Academies Press. 2010.
Novelty and Significance
What Is New?
Prevalence of elevated blood pressure (BP) in children has significantly increased from 1988 to 2008, although part of the increase is attributable to changes in obesity and Na intake.
Na intake >1.5× the recommended daily intake is associated with increased risk of elevated BP in children.
What Is Relevant?
Pediatricians should monitor.
Overall obesity (eg, body mass index).
Central obesity (eg, waist circumference).
Na intake to prevent elevated BP among pediatric patients.
Both prevalence of elevated BP (either prehypertension or hypertension) and mean level of systolic BP are associated with elevated Na intake in children.