| ObjectivesTo understand the associations of total body fat mass and fat distribution with subtotal and lumbar spine bone mineral density(BMD)in children and adolescents aged 6-17 years in Guangzhou,thus to provide a scientific basis for promoting healthy bone development in childhood and preventing osteoporosis in adulthood.MethodsIn this cross-sectional study,a total of 2500 students aged 6-17 years from one secondary and one primary school in Guangzhou were selected in April to May 2019.Dual energy X-ray absorptiometry was used to measure the total body fat,fat distribution,and subtotal BMD.Demographic characteristics,diet and life behavior were collected through questionnaires.Of all subjects,553 students were selected for lumbar spine BMD measurement using the systematic sampling method.The total body fat was assessed using the fat mass index(FMI)and body fat percentage(BF%),while the fat distribution variables included trunk-to-limb fat ratio(T/L)and android-to-gynoid fat ratio(A/G).The age dependent changes in total body fat and fat distribution,lumbar spine and subtotal BMD were described by mean ± standard deviation(SD)and line chart by genders.Their trends with age were analyzed using the Cox-stauart trend test.To account for differences in body composition and BMD by age and gender,we calculated age-and gender-standardized Z-scores for fat and BMD variables.The associations analysis of fat variables with lumbar spine and subtotal BMD were performed separately according to gender.Directed acyclic graph was used to screen out the potential confounding factors between fat and BMD,including parental education level,family income,frequency of eating dark leafy vegetables,drinking sugary drinks,dairy products,extracurricular activities,sleeping duration,screening duration,extracurricular learning duration,and lean mass.Inverse probability weighting based on generalized propensity score(GPS)was applied to balance confounders.GPS was estimated using least square estimation,the ability of GPS to balance confounders was diagnosed by comparing the changes of the average correlation coefficient before and after the application of GPS.The associations between fat variables and BMD were performed by the multiple linear regression model with inverse probability weighting.Stratified analysis of fat variables with BMD were performed by age in boys and girls,respectively(Boys were divided into <12 year and ≥12 year;girls were divided into <10 year and ≥10 year).In order to evaluate the robustness of the GPS-weighting regression results,traditional adjustment regression model was conducted as sensitivity analysis,with adjusting for demographic characteristics,diet and life behavior.Results1.Fat variables and BMD status: In this study,1929 children and adolescents aged 6-17 years in Guangzhou were included for fat and subtotal BMD analysis,and 539 for fat and lumbar spine BMD analysis.The total body fat and fat distribution of children and adolescents showed gender differences with age.In girls,the FMI、BF% and T/L shown an increasing trend with age and the increasing trend of FMI and T/L in girls were shown statistically significant with age(both P < 0.05).However,the A/G showed an initial increase followed by a decrease after 11 years old.In boys,the T/L and A/G shown an increasing trend with age(both P < 0.05),but the FMI and BF% were both shown a decreasing trend with age after 11 years old.The decreasing trend of BF% were shown statistically significant with age(P < 0.05).Besides,both the FMI and BF% in boys were lower compared with girls of the same age after the age of 11 years.Lumbar spine and subtotal BMD increased with age in children and adolescents,and the trends were statistically significant(P < 0.05).The age change curve of subtotal BMD for boys and girls showed a "double crossover " around 10 and 13 years old,boys had higher level of subtotal BMD than girls after 13 years old(P < 0.05).The age change curve of lumbar spine BMD for boys and girls crossed at 15 years old.Boys had lower level of lumbar spine BMD than girls before the age of 15 years and higher than girls aged 15 to 16 years,whereas at 17 years old,the lumbar spine BMD levels were generally consistent between genders.2.Association analysis between fat and subtotal BMD: In both boys and girls,fat variables were all positively correlated with subtotal BMD Z-score(all P < 0.001).In simple linear regression model,positive associations were found between all fat variables and subtotal BMD in both boys and girls(all P < 0.001).The average correlation coefficients were all lower than 0.10 in both genders after using GPS.Inverse probability weighted regression showed that,the BF%and A/G Z-scores were both significantly and negatively associated with subtotal BMD Z-scores in boys,for each SD increment in BF% and A/G,the BMD average decreased by 0.12 and 0.16 SD,respectively.(BF%: β=-0.12,95%CI=-0.21～-0.04;A/G: β=-0.16,95% CI=-0.24～-0.08;both P < 0.05).When stratified by age group,the negative associations were also seen between the BF%,A/G Z-scores and subtotal BMD Z-score in boys aged 12-17 years(BF%:β=-0.23,95% CI=-0.35～-0.12;A/G: β=-0.25,95% CI=-0.38～-0.11;both P <0.001),but null-association in boys aged 6-11 years(P > 0.05).For girls,the A/G Z-score was negative associated with subtotal BMD Zscore(β=-0.08,95% CI=-0.16～0.00,P=0.050),but there were no statistical association between total body fat and subtotal BMD(P > 0.05).When stratified by age group,the FMI Z-score was positive associated with subtotal BMD Zscore of girls aged 6-9 years,for each SD increment in FMI,the BMD average increased by 0.33 SD(β=0.33,95% CI=0.19～0.47,P < 0.001).In girls aged 10-17 years,the A/G Z-score was negative associated with subtotal BMD Z-score,for each SD increment in A/G,the BMD average decreased by 0.10 SD(β=-0.10,95% CI=-0.20～0.00,P < 0.05).3.Association analysis between fat and lumbar spine BMD: In girls,fat variables were all positively correlated with lumbar spine BMD Z-score(all P< 0.05).As for boys,except for A/G,fat variables were positively correlated with lumbar spine BMD Z-score(all P < 0.05).In simple linear regression model,positive associations were found between FMI,T/L Z-scores and lumbar spine BMD Z-scores in boys(all P < 0.05)and positive associations were found between all fat variables and lumbar spine BMD in girls(all P < 0.001).The average correlation coefficients were all lower than 0.10 in both genders after using GPS.Inverse probability weighted regression analysis shown that,fat variables were all negative associated with lumbar spine BMD in boys(FMI:β=-0.10,95% CI=-0.15～-0.06;BF%: β=-0.27,95% CI=-0.42～-0.13;T/L: β=-0.11,95% CI=-0.21～-0.02;A/G: β=-0.12,95% CI=-0.19～-0.06;all P < 0.05).When stratified by age,the negative associations were also shown between FMI,A/G Z-scores and lumbar spine BMD Z-score in boys aged 7-11 years(FMI:β=-0.11,95% CI=-0.15～-0.06;A/G: β=-0.13,95% CI=-0.19～-0.06;both P <0.001);the BF% Z-score was negative associated with lumbar spine BMD Zscore of boys aged 12-17 years(β=-0.33,95% CI=-0.54～-0.12,P < 0.05).For girls,the A/G Z-score was negative associated with lumbar spine BMD Z-score(β=-0.16,95% CI=-0.28～-0.03,P<0.05).Stratified by age,the FMI Zscore was positive associated with lumbar spine BMD Z-score of girls aged 7-9 years(β=0.50,95% CI=0.15～0.85,P < 0.05);the A/G Z-score was negative associated with lumbar spine BMD Z-score of girls aged 10-17 years(β=-0.17,95% CI=-0.31～-0.02,P < 0.05).4.Sensitivity analysis: Among girls of different age groups,the results of applying traditional regression model and GPS-inverse probability weighted regression model to analyze the association between fat and BMD were showed good stability.For the associations between fat and subtotal BMD in boys,the associations between total body fat and subtotal BMD were negative in both methods of boys aged 12-17 years;for boys aged 6-11 years,the null association between A/G and subtotal BMD was shown in GPS-inverse probability weighted regression analysis,but negative in traditional regression model.The confidence interval was narrower in traditional regression model.For the associations between fat and lumbar spine BMD in boys,the association between BF% and lumbar spine BMD was negative in both methods of boys aged 12-17 years;for boys aged 7-11 years,the associations of FMI,A/G with lumbar spine BMD were inconsistent,but the confidence intervals in GPSinverse probability weighted regression were narrower.ConclusionsThe association between fat and BMD in children and adolescents is not only gender-specific and age-specific,but also affected by the site of fat distribution.1.In boys,the total body fat was negative associated with lumbar spine and subtotal BMD,the negative association was stronger particularly among older boys.2.The total body fat was positive associated with lumbar spine and subtotal BMD in younger group,but there was null-association between total body fat and BMD in older girls.3.In both boys and girls,the abdominal adipose was negative associated with lumbar spine and subtotal BMD,this negative association was stronger after the onset of puberty. |
PDF Full Text Request