Résumé
ABSTRACT Objective: Screening for childhood obesity is a necessary step in developing appropriate and effective interventions. We evaluated the diagnostic performance of various recommended international anthropometric cut-offs based on body mass index (BMI), waist circumference (WC), waist-to-height ratio (WHtR), triceps skinfold (TSF), and mid-upper arm circumference (MUAC) in predicting excess adiposity (body fat ≥ 25%) in a random sample of Trinidadian preschoolers. Methods: After obtaining written parental consent, weight, height, WC, TSF, and MUAC were measured in 596 children using standard procedures. These were used to calculate BMI for age, WHtR, TSF-for-age z-scores, and MUAC-for-age z-scores. Percentage body fat was measured using a Tanita-531 foot-to-foot bioelectrical impedance analyser (BIA). Sensitivities, specificities and area under the receiver-operating curve analysis and predictive values were then computed in reference to BIA estimates. Results: The prevalence of excess adiposity was 12.2% and 5.1% among males and females, respectively. Sensitivities for the various cut-offs ranged from 20.0% to 75.0% and 57.1% to 96.9% among males and females, respectively. WHO-BMI recommended cut-offs and those based on MUAC z-scores had significantly higher sensitivities in females than in males. TSF z-scores had significantly lower sensitivities compared to those based on BMI and WHtR among males. Similarly, specificities ranged from 81.3% to 99.9% and 79.8% to 99.9% among males and females, respectively. In girls, cut-offs based on TSF z-scores had a higher likelihood ratio than cut-offs from Centers for Disease Control, International Obesity Task Force and WHtR. Diagnostic performance was not associated with ethnicity. Conclusion: Our results suggest that diagnostic performance was associated with gender and the cut-offs used; however, it was not associated with ethnicity.