The relationship between BMI and percent fat (PF) is well documented for Caucasians. This study examined the relationship between BMI and PF for a cohort of predominantly Hispanic youth.
Method
BMI and PF were measured on 630 students (Hispanic, 77%; Caucasian, 22%; Other, 1%) from two small school districts. Skinfold measures were determined by an experienced tester using established procedures. Height and weight were measured with a portable medical grade stadiometer and digital scale.
Analysis/Results
Descriptive statistics and correlations were determined. Two chi square analyses were performed to determine statistical differences between expected and observed frequencies for PF and BMI classifications. Participants were placed into one of four PF classifications: Very Low/Low, Optimal, Moderately High, and High/Very High. The four classifications for BMI were: Underweight, Normal, At Risk, and Overweight. The Bonferronni adjustment (.05/#statistical tests) was used to control Type I error. Alpha was set at .025 for statistical significance. The descriptive analysis revealed mean PF levels for Caucasian and Hispanics to be 23.39 (9.81) and 22.86 (9.29) respectively. Mean BMI values for Caucasian and Hispanics were 20.55 (5.43) and 22.38 (6.31) respectively. The correlation between PF and BMI for Caucasian and Hispanic students was significant (r =.720, p<.01). When combinations of gender and ethnicity were examined, all correlations were significant (p<.01). Since BMI is age and gender specific, correlations were determined separately for males and females by grade levels, yielding all combinations significant (p<.01). Results of the chi square analyses revealed significant differences between expected and observed frequencies within classifications for PF X2(3, N=630) =311.5, p< .001 and BMI X2(3, N=630) =372.1, p< .001. Expected frequency per classification was 157.5. The most pronounced differences for PF were in the Very Low/Low (n=30), Optimal (n=332) and for BMI were in the Underweight (n=11) and Normal (n=345) classifications. Smaller differences between expected and observed frequencies were found for PF in Moderately High (n=109) and High/Very High (n=159) classifications and for BMI in At Risk (n=113) and Overweight (n=161) classifications.
Conclusions
Results showed that as expected, there is a strong relationship between PF and BMI. When the PF categories for children outlined by Slaughter et al (1989) are reduced to four categories, frequencies in these categories are very similar to frequencies occurring in the four categories for BMI outlined by the CDC. Even though BMI is not a measure of PF, when compared categorically, both BMI and PF yield similar results.