Dear editor: The article by Vázquez and colleagues on “Social determinants of overweight and obesity in a highly marginalized population in Comitán, Chiapas, México” published in Salud Pública de México had a sample of 1 858 participants in urban and rural areas,¹ making known the epidemiology of obesity.

In their across sectional study, we find the samples of rural areas chosen for convenience and the urban areas, randomly. Álvarez and colleagues demonstrated the importance of simple random sampling, which allows the study to be highly representative.² Therefore, we believe the choice of population density in this study could cause an important selection bias when extrapolating data from rural areas.

The study used anthropometric equations “Durning-Womersley”, having to be corrected by the age, sex, and ethnic group, which is justified for the Caucasian population; being little suitable for the population of Mexico where the Crossbreeding prevails (indigenous, European, and African lineages).³ Thus, a study in America (Colombia) evaluated anthropometric equations, proving to be significant for the population of Medellin, being so far the most appropriate methodology to assess the Latin population.⁴

They developed the semi-quantitative questionnaire on food frequency consumption (FFQ) using the methodology of Walter Willet, being validated for a population with a level of literacy and numerical skills relatively intermediate for its resolution. The study does not mention cultural adaptations for these instruments; 44.5% of the population surveyed does not have complete primary education, leaving the adaptation methodology unclear for the understanding of data collection.⁵ Likewise they used the international questionnaire for physical activity (IPAQ, short version), which is recommended when population monitoring is intended but the ideal in epidemiological research studies is the long version.⁶

Finally, it’s necessary to validate instruments with cultural adaptation for vulnerable populations. We believe that communication through the use of photographs, videos, and interviews are more effective. Therefore, it is not possible to extrapolate the results obtained to the native community in general.