Indian Pediatrics - Editorial

Letters to the Editor

Indian Pediatrics 2003; 40:1018

Prevalence of Childhood Obesity in Chennai

We were quite surprised to note the conclusions drawn by authors that the prevalence of overweight and obesity in Chennai amongst affluent adolescent girls has not changed(1). The authors have computed percentiles of their study population on both groups separately and rather than comparing every observation against an external standard, e.g., Agarwal’s(2) or Cole’s(3), only the final comparison of percentile values with external standards is shown in Table III. Computing percentile values of the study groups separately is like doing internal ranking and then it is not at all surprising that the percentage of children above 85th and 95th percentile has not changed from 81 to 98 because the population will get distributed uniformly around the mean keeping the prevalence same at a given percentile within that group.

The authors have shown that the 95th percentile in 1998 group is higher than that of 1981 study at all ages. It is seen from their Table I and Table III that at 11 years of age in the 81 group there were 2 girls who had a BMI of >19.87 (95th percentile for 81 group at 11 years) whereas in 98 study there were 3 girls with a BMI of >22.95(95th percentile for 98 group at 11 years). As the cut off used to define obesity at 11 years in 1981 was 19.87 and in 1998 was 22.95 there were possibly more girls between BMI of 19.87 and 22.95 in the 98 groups which have not been counted. The same may be true at all ages. This means that the whole population has shifted to the heavier side but as internal ranking was done the prevalence appears to remain constant, and thus gives the reader a wrong message.

The authors have shown that the 50th percentile values have remained the same and yet the 95th percentile values are higher at all ages. It would be important for the reader to also know the means of BMI at various ages rather than the 50th percentile because when percentiles are calculated on small numbers such as 27 and 31 there may be a substantial difference between the mean and the 50th percentile. Creating your own percentile values is like creating a yardstick from your measurements rather than using a yardstick to measure your observations. To compare any two sets of data the standard has to be the same and different from the study population (external) which is not done in this study and hence authors should consider re-publishing their data using an external standard with which two data sets should be compared separately.

Vaman Khadilkar,
Anuradha Khadilkar,
Hirabai Cowasji Jehangir
Medical Research Institute,
32 Sassoon Road, Pune 411 001,
Maharashtra, India.
E-mail: [email protected]

References

1. Subramanyam V, Jayashree R, Raft M. Prevalence of overweight and obesity in affluent adolescent girls in Chennai in 1981 and 1998. Indian Pediatr 2003; 40: 332-336.

2. Agarwal KN, Saxena A, Bansal AK, Agarwal DK. Physical growth assessment in adolescence. Indian Pediatr 2001, 38: 1217-1235.

3. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing standard definition for child overweight and obesity world wide. International survey. BMJ 2000, 320: 1240-1255.