[An alternative method to describe an "obesogenic" environment for children: Relevance of a multiple correspondence analysis (MCA) (corrected)]

The purpose of this study was to describe an "obesogenic" environment for a group of schoolchildren using a multiple correspondence analysis (MCA) as an alternative approach to traditional methodological choices. MCA is applicable even for small samples. Ninety-one children (39 girls and 52 boys) aged 10.0 +/- 0.9 years were randomly recruited from two French public schools. Data on their family context, parental involvement, television time and their eating habits were obtained through questionnaires. Their level of physical activity and sedentary time were assessed using an accelerometer (MTI Actigraph model 7164) for three days, including a holiday. The data were processed using an MCA together with a technique for estimating relative risks (RRs) of overweight/obesity according to the distribution of children in the factorial plane produced by the MCA. The "obesogenic" factors appeared as four possible combinations between family environments and various behaviours with regard to physical activity, sedentary behaviour and diet. The RR of overweight/obesity was 2.64 [1.52, 4.57] (P < 0.0001) for a combined association of a "disadvantaged" family environment + low physical activity and high fat diet. The RR of overweight/obesity was 0.36 [0.14, 0.94] (P < 0.05) for an association of a "privileged" family environment + high physical activity and low fat diet. Thus, MCA appears sufficiently robust and relevant to effectively guide etiological hypotheses and decisions about individual and collective intervention strategies.

Diagnostic quality of Actigraph-based physical activity cut-offs for children: what overweight/obesity references can tell?

BACKGROUND

There is conflicting literature about Actigraph-based cut-off points (CoP) that define the lower limit of moderate-to-vigorous physical activity (MVPA) among children. The aim of the present study was to compare the diagnostic quality of Actigraph-based CoP relative to different body mass index (BMI) classification systems of the weight status of children.

METHODS

One hundred and thirteen children aged 8-11 years recruited from three randomly selected elementary schools wore the Actigraph, model 7164 (Actigraph Manufacturing Technology) for 3 days. Five CoP, expressed in counts per minute (c.p.m.; CoP 1000 CoP 2000, CoP 3000, CoP 3200, and CoP 3600) were used to compute their MVPA. The area under the receiver operating characteristic curves (AUC) together with specific indices allowed assessment of the performance of these CoP in reference to overweight/obesity status as defined using BMI-based criteria from the International Obesity Task Force (IOTF), French References (FR) and World Health Organization (WHO) standards.

RESULTS

Overweight/obesity frequency ranged from 30% to 42%, with no sex-related difference. All AUC (range, 0.623-0.660) were significantly higher than 0.500 except from those of CoP 1000 and 2000 using the IOTF criteria (0.602 +/- 0.058 and 0.601 +/- 0.057, respectively) and the FR (0.608 +/- 0.060 for CoP 1000). Furthermore, with the WHO standards, all the CoP provided non-significant AUC (range, 0.566-0.597). According to the IOTF and the FR criteria, respectively, CoP 3600 had the highest probability of correct decision (0.62 and 0.68), the lowest misclassification errors (0.38 and 0.32), the highest validity coefficient (0.21 and 0.29), and the highest expected maximum utility (59 and 83).

CONCLUSION

When children are classified using BMI-based criteria, the threshold of 3600 c.p.m. should be more appropriate in discriminating non-overweight from overweight/obese.