C-Reactive Protein Concentration and Cardiovascular Disease Risk Factors in Children
Findings From the National Health and Nutrition Examination Survey 1999–2000
Background— C-reactive protein is an emerging risk factor for cardiovascular disease. Although the relations between C-reactive protein and other risk factors for cardiovascular disease have been extensively studied in adults, the relations in children are not well understood.
Methods and Results— Data from 2846 boys and girls 3 to 17 years of age who participated in the National Health and Nutrition Examination Survey, 1999 to 2000, a cross-sectional survey of the US population, were used. In univariate analyses, significant associations were observed between C-reactive protein concentration—measured with a high-sensitivity assay—and age, body mass index, systolic blood pressure, and triglyceride concentrations in both sexes. In multiple linear regression analyses, body mass index was the best predictor of C-reactive protein concentration. Age was positively associated with C-reactive protein concentration among boys 3 to 17 years of age. Some race or ethnic differences were present as well among boys 8 to 17 years of age and girls 8 to 11 years of age. Systolic blood pressure was positively associated with C-reactive protein among girls 12 to 17 years of age.
Conclusions— Among the sociodemographic and cardiovascular disease risk factors, body mass index was the best predictor of C-reactive protein concentration in children.
Received January 21, 2003; de novo received March 11, 2003; revision received April 23, 2003; accepted April 25, 2003.
C-reactive protein has garnered a great deal of interest from the medical community because its concentration is a significant predictor of the incidence of cardiovascular disease and because it has prognostic value among people with existing cardiovascular disease. As with any emerging risk factor, a thorough understanding of its determinants is needed. Such information will help to identify high-risk populations and to assess which variables need to be considered in analyses of risk factor-disease relations.
Among adults, research has shown numerous significant correlates of C-reactive protein concentration.1 Variations in C-reactive protein concentration have been found by sociodemographic indicators such as age, race or ethnicity, and sex, and risk factors for cardiovascular disease such as anthropometric measures, blood lipids, blood pressure, smoking status, physical activity, glucose tolerance, and insulin sensitivity.
Because cardiovascular disease often has origins in childhood and because several risk factors for cardiovascular disease track from childhood to adulthood, understanding the distribution and implications of such risk factors among children is of considerable interest. At this time, the study of C-reactive protein as a risk factor or marker for cardiovascular disease in children is in its early stages. A recent study showed that C-reactive protein concentration was associated with intima-media thickness in Finnish children.2 To foster our knowledge about the relations between C-reactive protein concentration and risk factors for cardiovascular disease in children, data from the National Health and Nutrition Examination Survey, 1999 to 2000, were examined.
Detailed information about the methods and procedures of NHANES 1999 to 2000 is available elsewhere.3 In brief, a representative sample of the noninstitutionalized civilian US population was selected through a stratified multistage design. Trained interviewers, using a computer-assisted personal interview system, interviewed participants at home. Participants were asked to attend the mobile examination center, where they completed additional questionnaires, underwent various examinations, and provided a blood sample. The study received human subject approval from the Centers for Disease Control and Prevention.
C-reactive protein was measured for children 3 to 17 years of age by latex-enhanced nephelometry (N High Sensitivity CRP assay) on a BN II nephelometer (Dade Behring Inc) at the University of Washington Medical Center, Seattle. Two levels of control materials from Bio-Rad Laboratories, Inc were used for quality-control purposes, and day-to-day coefficients of variation ranged from 4.93% to 7.84%. The means for the control materials were 1.67 mg/L and 3.82 mg/L for samples analyzed during an initial 9-month period and 1.84 mg/L and 3.95 mg/L for a subsequent 12-month period. The performance of this assay has been shown to be good.4
Not all study variables were collected for all age groups. For children 3 to 7 years of age, age, sex, race or ethnicity, body mass index, total serum cholesterol concentration, and homocysteine concentration were included in the analyses. For children 8 to 17 years of age, systolic and diastolic blood pressure were added. For children 12 to 17 years of age, smoking status, glycosylated hemoglobin, and triglyceride and glucose concentrations were included as well. Body mass index percentiles, calculated from measured weight and height according to the Centers for Disease Control and Prevention growth charts,5 and not body mass index were used in all analyses. Cholesterol was measured enzymatically. Homocysteine was determined through a fluorescence polarization immunoassay (Abbott Diagnostics). Smoking status was determined from the following questions: “Have you ever tried cigarette smoking, even one or two puffs?”; “How old were you when you smoked a whole cigarette for the first time?”; and “During the past 30 days, on how many days did you smoke cigarettes?” From the responses, 6 categories were created: never tried smoking, tried smoking but did not smoke an entire cigarette, did not smoke during the past 30 days, smoked for 1 to 7 days, smoked for 8 to 25 days, and smoked for 26 to 30 days. Glycosylated hemoglobin was measured on Primus CLC330 and Primus CLC385 instruments (Primus Corporation) by boronate affinity high-performance liquid chromatography. Concentrations of glucose and triglycerides were measured on a Hitachi 704 multichannel analyzer (Boehringer Mannheim Diagnostics). Glucose concentration was measured with the use of the glucose hexokinase method, and triglyceride concentration was determined after hydrolysis of triglycerides to glycerol and oxidation to dihydroxyacetone phosphate and hydrogen peroxide.
The analyses were limited to participants 3 to 17 years of age who attended the mobile examination center. Pearson correlation coefficients between C-reactive protein concentration and other continuous variables were calculated by using the sampling weights with SAS version 8.2. These univariate associations were also examined with unadjusted linear regression analyses. In addition, the independence of the study variables to C-reactive protein concentration was examined in multiple linear regression analyses. For correlation and regression analyses, C-reactive protein concentration was log-transformed to improve the distribution of this variable. To account for the complex sampling design, SUDAAN version 8.0 was used to calculate means and proportions and to perform linear regression analyses.6
C-reactive protein concentrations were available for 1479 boys and 1367 girls who attended the mobile examination center. Values ranged from 0.1 mg/L to 65.2 mg/L (geometric mean: 0.4 mg/L; median: 0.3 mg/L) among boys and 0.1 mg/L to 46.6 mg/L (geometric mean: 0.5 mg/L; median: 0.4 mg/L) among girls. Information about study variables for boys and girls are provided in Table 1.
Among both male and female participants 3 to 17 years of age, age, body mass index percentile, and systolic blood pressure were significantly associated with ln (C-reactive protein concentration) (Tables 2 and 3⇓). The largest correlation coefficients among both boys and girls were observed for body mass index. Weak positive associations were present between homocysteine and C-reactive protein concentration among girls and between diastolic blood pressure and C-reactive protein concentration among boys. When stratified by age, body mass index percentile, systolic blood pressure, and triglyceride concentration were consistently associated with C-reactive protein concentration among both boys and girls. Correlations for triglyceride concentrations and C-reactive protein concentration were larger among girls (0.25) than boys (0.16). Among boys, the correlation coefficient for systolic blood pressure was higher among the older group than those 8 to 11 years of age. Among girls, the correlation coefficients were similar for those 8 to 11 years of age and those 12 to 17 years of age.
Unadjusted geometric mean C-reactive protein concentrations ranged from 0.4 mg/L to 0.6 mg/L for the five race or ethnic groups. Geometric mean C-reactive protein concentrations were 0.4 mg/L for participants who had never smoked; 0.5 mg/L for participants who had never smoked a whole cigarette; 0.6 mg/L for participants who had not smoked during the previous 30 days; 0.5 mg/L for participants who had smoked for 1 to 7 days during the previous 30 days; 0.6 mg/L for participants who had smoked for 8 to 25 days during the previous 30 days; and 0.6 mg/L for participants who had smoked ≥26 days out of the previous 30 days. In a linear regression model, smoking status was not significantly associated with C-reactive protein concentration (P for Wald χ2=0.091).
Few statistically significant associations were noted in the multiple linear regression analyses. Among 1447 boys 3 to 17 years of age, significant associations were present for age (in months) (β=0.004, SE=0.001, P=0.003), ethnicity (geometric means: Mexican Americans=0.49 mg/L; whites=0.38 mg/L, P=0.012), and body mass index percentiles (β=0.019, SE=0.002, P<0.001). Among 1340 girls, ethnicity (geometric means: Mexican Americans=0.66 mg/L; whites=0.42 mg/L, P<0.001) and body mass index percentile (β=0.020, SE=0.003, P<0.001) were significantly associated.
Among boys 3 to 7 years of age, body mass index percentiles were significantly and positively associated with C-reactive protein concentration (Table 4). Among boys 8 to 11 years of age, Mexican American boys (geometric mean=0.60 mg/L, P=0.023) had higher adjusted C-reactive protein concentrations than did white boys (geometric mean=0.35 mg/L), and body mass index was positively associated with C-reactive protein concentration. Among boys 12 to 17 years of age, body mass index was positively associated with C-reactive protein concentration.
Among girls, body mass index was significantly associated with C-reactive protein concentration in all three age groups (Table 5). Among girls 8 to 11 years of age, Mexican Americans (geometric mean=0.76 mg/L, P=0.015) had higher C-reactive protein concentrations than did whites (geometric mean=0.39 mg/L). Among girls 12 to 17 years of age, Mexican Americans (geometric mean: 0.72 mg/L, P=0.030) had higher C-reactive protein concentrations than did whites (geometric mean=0.51 mg/L), and systolic blood pressure was positively associated with C-reactive protein concentration.
To examine the associations between C-reactive protein concentration and concentrations of fasting glucose and triglycerides, linear regression models that excluded glycosylated hemoglobin concentration for participants 12 to 17 years of age who had fasted ≥8 hours were examined. Concentrations of glucose (boys: β=−0.126, SE=0.248, P=0.614; girls: β=0.069, SE=0.326, P=0.834) and triglycerides (boys: β=0.026, SE=0.184, P=0.888; girls: β=0.441, SE=0.326, P=0.834) were not significantly associated with C-reactive protein concentration among boys or girls. Among boys (n=529), significant associations included age (β=−0.139, SE=0.071, P=0.055), age-squared (β <0.001, SE<0.001, P=0.040), race or ethnicity (geometric means: white=0.41 mg/L; Mexican American=0.58 mg/L, P versus whites=0.021; other=0.25 mg/L, P versus whites <0.001), body mass index percentile (β=0.017, SE=0.004, P<0.001), and homocysteine (β=−0.109, SE=0.031, P=0.001). Among girls (n=510), only body mass index percentile was significantly associated with C-reactive protein concentration (β=0.023, SE=0.003, P<0.001).
In this large study of the interrelations of C-reactive protein, measured with the use of a high-sensitivity assay, with other risk factors for cardiovascular disease in children, body mass index was the most consistent and strongest predictor of C-reactive protein concentration. In some age- and sex-matched groups, race or ethnicity, homocysteine concentration, and systolic blood pressure were significantly associated with C-reactive protein concentration. C-reactive protein concentration was not independently associated with concentrations of total serum cholesterol, glycosylated hemoglobin, glucose, or triglycerides or with smoking.
In previous studies of children, the median C-reactive protein concentration was 0.15 mg/L (interquartile range, 0.06 to 0.47 mg/L) among 699 British children,7 the mean C-reactive protein concentration was 0.7 mg/L (range, 0 to 8.6 mg/L) among 79 Finnish children,2 and means ranged from 2.1 to 3.8 ng/mL among 74 obese US children.8 In comparison, the median C-reactive protein concentration was 0.3 mg/L among boys and 0.4 mg/L among girls in the present study.
Few other studies have examined the relations between C-reactive protein concentration and other cardiovascular disease risk factors in children. Among 699 British children who were 11 years old, C-reactive protein concentration was associated with systolic blood pressure, diastolic blood pressure, pulse, fibrinogen concentration, factor VIIc, HDL cholesterol concentration, triglyceride concentration, apolipoprotein B concentration, and postload insulin concentration after adjustment for age, sex, ethnicity, and town.7 Further adjustment for ponderal index attenuated most of the Pearson correlation coefficients somewhat and caused systolic blood pressure, factor VIIc, triglyceride concentration, apolipoprotein B concentration, and postload insulin concentration not to remain significant. Analyses of NHANES III data showed that body mass index was positively associated with C-reactive protein concentration in large numbers of US children.9,10 Another analysis of NHANES III data showed that C-reactive protein concentration was inversely and independently associated with serum carotenoid concentrations among US children.11 In addition, among 79 Finnish children (mean age, 10.5 years), C-reactive protein concentration was significantly associated with body mass index but not age, systolic blood pressure, diastolic blood pressure, total cholesterol concentration, HDL cholesterol concentration, triglyceride concentration, and glycosylated hemoglobin in univariate analyses.2 Although percent body fat correlated positively and physical fitness inversely with C-reactive protein concentration at baseline among 74 US children 12 to 16 years of age, 8 months of physical training had little effect on C-reactive protein concentrations despite weight loss and improved fitness.8
In children as in adults, excess weight is a consistent and an important determinant of circulating C-reactive protein concentration. Increases in the concentrations of tumor necrosis factor (TNF) and interleukin-6 (IL-6) accompany increases in weight.12–14 These cytokines stimulate hepatic production of C-reactive protein.15,16
In contrast to studies in adults, in which C-reactive protein concentration has been associated with smoking, blood pressure, blood lipid concentrations, and hyperglycemia,1 C-reactive protein concentration did not correlate consistently with these risk factors in children in this study. Perhaps some associations with various risk factors for cardiovascular disease emerge after childhood, when a critical threshold of inflammation from atherosclerosis has been reached.
Several limitations of this study deserve to be acknowledged. First, it was a cross-sectional analysis, and thus, the directionality of the associations cannot be established. Second, smoking was based on self-reported data. No biochemical markers of smoke exposure were available. Although the lack of an association between C-reactive protein concentration and smoking may be a true finding, inaccurate reporting by the participants may have accounted for the null finding. Finally, the sample sizes for some age groups may have been insufficient to detect weak associations.
In conclusion, body mass index was the most consistent and strongest predictor of C-reactive protein concentration in a representative sample of US children. In adults, the use of C-reactive protein measurements in assessing the risk for cardiovascular disease is transitioning from the research arena into clinical practice. Whether a role exists for measuring C-reactive protein in children to assess their risk for cardiovascular disease remains to be established.
Jarvisalo MJ, Harmoinen A, Hakanen M, et al. Elevated serum C-reactive protein levels and early arterial changes in healthy children. Arterioscler Thromb Vasc Biol. 2002; 22: 1323–1328.
Centers for Disease Control and Prevention. NHANES 1999–2000 public data release file documentation. Hyattsville, Md: National Center for Health Statistics: http://www.cdc.gov/nchs/about/major/nhanes/currentnhanes.htm. Last accessed January 15, 2003.
Research Triangle Institute. SUDAAN User’s Manual, Release 8.0. Research Triangle Park, NC: Research Triangle Institute; 2001.
Ford ES, Gillespie C, Ballew C, et al. Serum carotenoid concentrations in US children and adolescents. Am J Clin Nutr. 2002; 76: 818–827.