Baseline characteristics of participants
Table 1 shows the characteristics of the subjects according to the tertiles of serum trans-β-carotene among the 8,330 participants aged 6–18 years in NHANES 2001–2006; 1,558 of them were obese (19.40 ± 0.85%) after being weighted to the US population. Serum trans-β-carotene level increased with increasing percentages of TC, LDL-C, but lower percentage of HDL-C, HOMA-IR, CRP. Subjects with higher Serum trans-β-carotene were found to be younger, in the low serum cotinine levels group, and less likely to be the Non-Hispanic Black.
Table 1
Characteristics of subjects according to the tertiles of serum trans-β-carotene in US children and adolescents, National Health and Nutrition Examination Survey, 2001–2006a
Characteristics | Total n = 8030 | Tertiles 1 | Tertiles 2 | Tertiles 3 | p for trend |
Age,y | 12.27(0.08) | 13.66(0.1) | 12(0.11) | 11.36(0.1) | < 0.001 |
Age, % | | | | | |
Age, ≥ 11yr | 64.68(0.95) | 80.94(1.18) | 61.99(1.41) | 53.41(1.35) | < 0.001 |
Age, < 11yr | 35.32(0.95) | 19.06(1.18) | 38.01(1.41) | 46.59(1.35) | |
Gender, % | | | | | |
Female | 48.49(0.77) | 47.03(1.1) | 47.84(1.31) | 50.32(1.42) | 0.097 |
Male | 51.51(0.77) | 52.97(1.1) | 52.16(1.31) | 49.68(1.42) | |
Height, cm | 152.98(0.4) | 160.66(0.53) | 151.46(0.49) | 147.88(0.55) | < 0.001 |
Weight, kg | 51.82(0.44) | 63.58(0.65) | 49.53(0.54) | 43.98(0.47) | < 0.001 |
Height-for-age z-scores, SD | 0.61(0.02) | 0.65(0.04) | 0.59(0.03) | 0.61(0.03) | 0.448 |
Weight-for-age z-scores, SD | 1.13(0.04) | 1.87(0.1) | 1.14(0.06) | 0.86(0.05) | < 0.001 |
BMI-for-age z-scores, SD | 0.82(0.03) | 1.28(0.04) | 0.77(0.05) | 0.48(0.03) | < 0.001 |
Obesity, % | 19.40(0.85) | 36.05(1) | 17.71(1.06) | 9.65(0.72) | < 0.001 |
Ethnicity, % | | | | | |
Non-Hispanic White | 60.95(2.18) | 59.5(3.04) | 59.82(2.36) | 63.2(2.13) | 0.105 |
Non-Hispanic Black | 14.8(1.42) | 16(1.51) | 15.55(1.59) | 13.1(1.49) | 0.020 |
Mexican American | 12.34(1.19) | 12.38(1.69) | 12.49(1.22) | 12.16(1.06) | 0.851 |
Other Race/Ethnicity | 6.17(0.74) | 5.39(0.96) | 6.22(0.92) | 6.78(1.06) | 0.243 |
Cotinine, % | | | | | |
Cotinine, > 10 ng/mL | 7.62(0.5) | 14.25(1.17) | 7.34(0.8) | 2.28(0.3) | < 0.0001 |
Cotinine, ≤ 10 ng/mL | 92.38(0.5) | 85.75(1.17) | 92.66(0.8) | 97.72(0.3) | |
PIR, % | | | | | |
PIR, ≥1 | 78.77(1.09) | 75.64(1.5) | 79.21(1.2) | 81(1.43) | 0.002 |
PIR, <1 | 21.23(1.09) | 24.36(1.5) | 20.79(1.2) | 19(1.43) | |
Education, % | | | | | |
Education, ≥HS | 5.49(0.48) | 7.68(0.8) | 4.88(0.58) | 4.21(0.6) | < 0.001 |
Education, <HS | 94.51(0.48) | 92.32(0.8) | 95.12(0.58) | 95.79(0.6) | |
Diabetes, % | 0.45(0.09) | 0.42(0.13) | 0.28(0.1) | 0.63(0.2) | 0.340 |
TC, ≥ 200 mg/dL | 9.92(0.57) | 6.35(0.69) | 9.54(0.98) | 13.29(0.87) | < 0.001 |
TG, ≥ 150 mg/dL | 11.21(0.91) | 14.77(1.75) | 8.31(1.36) | 10.48(1.5) | 0.080 |
LDL-C, ≥ 130 mg/dL | 6.9(0.63) | 3.58(0.55) | 7.5(1.31) | 9.51(1.25) | < 0.001 |
HDL-C, < 35 mg/dL | 5.43(0.4) | 10.47(0.92) | 4.25(0.74) | 2.11(0.56) | < 0.001 |
Glycohemoglobin, > 5.7% | 1.74(0.2) | 2.11(0.4) | 1.52(0.31) | 1.47(0.35) | 0.231 |
FPG, ≥ 100 mg/dL | 14.92(1.28) | 15.11(1.61) | 14.04(1.84) | 15.56(2.54) | 0.906 |
HOMA-IR, ≥ 3.29 | 25(1.44) | 33.4(2.32) | 22.32(2.19) | 16.29(1.81) | < 0.001 |
CRP, ≥ 0.211 mg/dL | 15.56(0.6) | 23.32(0.9) | 14.36(1.03) | 10.11(0.71) | < 0.001 |
NOTE. Results are mean (± SD), median (interquartile range), or percentage (n = number of individuals). |
a Data were weighted according to the National Health and Nutrition Examination Survey protocol. |
Abbreviations: |
BMI body mass index, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, CRP C-reactive protein, TC total cholesterol, TG triglyceride, HOMA-IR homeostasis model assessment of insulin resistance; HS High school, PIR, poverty income ratio, FPG, fasting plasma glucose. |
The Results Of Relationship Between Trans-β-carotene And Obesity
In the multiple linear regression analysis, serum trans-β-carotene level was significantly associated with obesity after being adjusted for confounding factors (Table 2). The non-adjusted and adjusted models are also set out in Table 2. In crude model, trans-β-carotene showed a negative correlation with obesity (0.32,95%CI: 0.28–0.37). In the adjusted model I (adjusted for age, sex, race/ethnicity), the result had not obviously changed (0.32, 95%CI: 0.28–0.37). In addition, we detected the same negative correlation in fully adjusted model (adjustment variables: age, sex, race/ethnicity, family PIR, education levels, serum cotinine status, diabetes, TC, TG, HDL-C, LDL-C, HOMA-IR score, Glycohemoglobin, FBG, CRP) (0.42, 95%CI: 0.29–0.61). For the sensitivity analysis, we also handled trans-β-carotene as Categorical variable (tertiles) and found the same trend (p for trend = 0.002). The higher tertile of serum trans-β-carotene level was significantly associated with reduction of obesity compared with the lower tertile after adjustment for all confounding factors (compared to 1st tertile, 2nd tertile: OR = 0.50, 95% CI, 0.31–0.81; 3rd tertile: OR = 0.30, 95% CI, 0.17–0.55).
Table 2
Association of serum trans-β-carotene level with obesity in NHANES 2001–2006 (n = 8030). a
| Odds ratio (95% CI) P |
| Model 1 | Model 2 | Model 3 | Model 4 |
Trans-β-carotene level (ug/dl) ln transform |
ln (trans-β-carotene, ug/dl) | 0.32(0.28,0.37) <0.001 | 0.29(0.25,0.33) <0.001 | 0.28(0.24,0.32) <0.001 | 0.42(0.29,0.61) <0.001 |
Trans-β-carotene Level (ug/dl) | | | |
Tertiles 1 | 1.00 (referent) | 1.00 (referent) | 1.00 (referent) | 1.00 (referent) |
Tertiles 2 | 0.44(0.38,0.52) <0.001 | 0.40(0.34,0.48) <0.001 | 0.39(0.33,0.47) <0.001 | 0.50(0.31,0.81) 0.018 |
Tertiles 3 | 0.22(0.18,0.27) <0.001 | 0.19(0.16,0.23) <0.001 | 0.19(0.16,0.23) <0.001 | 0.30(0.17,0.55) 0.003 |
p for trend | < 0.001 | < 0.001 | < 0.001 | 0.002 |
a Data were weighted according to the National Health and Nutrition Examination Survey protocol. |
Table 3
Effect sizes of serum trans-β-carotene level and obesity between two groups.
| FPG < 100 (mg/dl) | FPG ≥ 100 (mg/dl) |
Crude model (β, 95%CI, P) |
Trans-β-carotene level (ug/dl) ln transform |
ln (trans-β-carotene, ug/dl) | 0.39 (0.32, 0.47) < 0.001 | 0.20 (0.13, 0.32) < 0.001 |
Trans-β-carotene Level (ug/dl) |
Tertiles 1 | 1.00 (referent) | 1.00 (referent) |
Tertiles 2 | 0.48 (0.37, 0.62) < 0.001 | 0.20 (0.11, 0.36) < 0.001 |
Tertiles 3 | 0.32 (0.24, 0.42) < 0.001 | 0.10 (0.05,0.20) < 0.001 |
p for trend | < 0.001 | < 0.001 |
Model I (β,95%CI, P) |
Trans-β-carotene level (ug/dl) ln transform |
ln (trans-β-carotene, ug/dl) | 0.36 (0.29, 0.44) < 0.001 | 0.18 (0.11, 0.30) < 0.001 |
Trans-β-carotene Level (ug/dl) | |
Tertiles 1 | 1.00 (referent) | 1.00 (referent) |
Tertiles 2 | 0.45 (0.35, 0.59) < 0.001 | 0.18 (0.10, 0.33) < 0.001 |
Tertiles 3 | 0.30 (0.22, 0.40) < 0.001 | 0.09 (0.18, 0.38) < 0.001 |
p for trend | < 0.001 | < 0.001 |
Model II (β, 95%CI, P) |
Trans-β-carotene level (ug/dl) ln transform |
ln (trans-β-carotene, ug/dl) | 0.36 (0.29, 0.44) < 0.001 | 0.17 (0.10, 0.29) < 0.001 |
Trans-β-carotene Level (ug/dl) | |
Tertiles 1 | 1.00 (referent) | 1.00 (referent) |
Tertiles 2 | 0.44 (0.34, 0.58) < 0.001 | 0.20 (0.11, 0.37) < 0.001 |
Tertiles 3 | 0.31 (0.23, 0.42) < 0.001 | 0.07 (0.03, 0.16) < 0.001 |
p for trend | < 0.001 | < 0.001 |
Model III (β, 95%CI, P) |
Trans-β-carotene level (ug/dl) ln transform |
ln (trans-β-carotene, ug/dl) | 0.50 (0.37, 0.69) < 0.001 | 0.10 (0.03, 0.31) < 0.001 |
Trans-β-carotene Level (ug/dl) | |
Tertiles 1 | 1.00 (referent) | 1.00 (referent) |
Tertiles 2 | 0.54 (0.36, 0.82) 0.003 | 0.07 (0.02, 0.24) < 0.001 |
Tertiles 3 | 0.46 (0.29, 0.73) 0.001 | 0.04 (0.01, 0.26) < 0.001 |
p for trend | < 0.001 | < 0.001 |
The Subgroup Analyses
Interaction tests were carried out to examine the effect of relevant stratification factors on the relationship between trans-β-carotene and obesity. A significant interaction (P = 0.01) between trans-β-carotene and plasma glucose was observed. The effect sizes of trans-β-carotene and obesity showed a significant difference between two plasma glucose groups. A stronger association between trans-β-carotene and obesity was found in higher plasma glucose levels (more than100 mg/dl)(OR = 0.10; 95% CI, 0.03–0.31).With each one-unit of natural log-transformed serum trans-β-carotene unit, there was an 90% decrease in obesity among high-plasma glucose population. For sensitivity analysis, a higher serum trans-β-carotene level was significantly associated with a reduced probability of obesity in higher plasma glucose levels (compared to 1st tertile, 2nd- tertile: OR = 0.07, 95% CI, 0.02–0.24; 3rd tertile: OR = 0.04, 95% CI, 0.01–0.26).
The Analyses Of Linear Relationship
Because trans-β-carotene was a continuous variable, analyses of linear relationship were necessary. In the present study (Fig. 1A), we found that the relationship between trans-β-carotene and obesity was linear (adjustment variables: age, sex, race/ethnicity, family PIR, education levels, serum cotinine status, diabetes, TC,TG,HDL-C,LDL-C, HOMA-IR score ,Glycohemoglobin, FBG,CRP)( 0.45, 95%CI: 0.34–0.60). We also used generalized additive model to identify the linear relationship, and the inflection point was 2.34 (likelihood ratio test p = 0.198). However, the relationship between trans-β-carotene and obesity is non-linear at higher plasma glucose level(Fig. 1B). We also utilized GAM to identify the non-linear relationship and the inflection point was 2.7 (trans-β-carotene = 14.88 ug/dl) (likelihood ratio test p = 0.012) (Table 4). By two-piecewise linear regression model, effect sizes and confidence intervals for the left and right sides of the inflection point were 0.10 (0.00 to 0.2) and 6.7 (0.1 to 348.2), respectively.
Table 4
Results of two-piecewise linear regression model in higher plasma glucose level group.
Inflection point of ln (trans-β-carotene, ug/dl) | Effect size (β) | 95%CI | P value |
< 2.7 | 0.1 | (0.00, 0.2) | < 0.001 |
≥ 2.7 | 6.7 | (0.1, 348.2) | 0.345 |
Adjusted for age, sex, race/ethnicity (non-Hispanic white, non-Hispanic blacks, Mexican Americans, and others), family PIR, education levels, serum cotinine status, diabetes, TC, TG, HDL-C, LDL-C, HOMA-IR score, glycohemoglobin, CRP; |