On 2018 Jan 06, DAVID LUDWIG commented:

A recent NY Times investigative article raised questions about the interpretation of this study, and possible harms to public health in developing nations from consumption of sugary beverages such as the ones examined here. However, the NY Times article did not provide a detailed critique of the scientific methods.

The authors of this study conclude, "These findings suggest that malted drinks are a micronutrient-rich beverage which are unlikely to promote excess energy intake and obesity risk, at the consumption pattern in the population assessed." I believe that this conclusion is overstated, for the following reasons:

First: The study design, a cross-sectional survey, is weak and especially susceptible to confounding.

Second: The final participation rate was only about 1/5 the total invited sample, potentially leading to major selection bias (thus, undermining intent to obtain a nationally representative sample).

Third: The validity of the method to measure diet, (e.g., 10 to 12 year olds responded on their own) was not demonstrated.

Fourth: Malt beverage consumers were more physically active and watched less screen time: thus, they likely came from families with greater health consciousness. Therefore, the associations with higher intakes of a few micronutrients may simply reflect confounding. That is, they may have gotten the extra nutrients from other components of the diet. Unfortunately, this point was not examined in the study.

Fifth: Concerningly, the statistical models were not adjusted for physical activity and screen time. Doing so could have unmasked higher body weight among the malt beverage consumers. That is, perhaps the increased PAL and less screen time counteracted the adverse effects of beverage consumption.

Indeed, the true associations between sugar-sweetened beverages and body weight shown in higher quality prospective studies Schulze MB, 2004 are often not seen in cross-sectional surveys Forshee RA, 2004 – demonstrating the weak nature of this study type.

On 2018 Jan 06, DAVID LUDWIG commented:

A recent NY Times investigative article raised questions about the interpretation of this study, and possible harms to public health in developing nations from consumption of sugary beverages such as the ones examined here. However, the NY Times article did not provide a detailed critique of the scientific methods.

The authors of this study conclude, "These findings suggest that malted drinks are a micronutrient-rich beverage which are unlikely to promote excess energy intake and obesity risk, at the consumption pattern in the population assessed." I believe that this conclusion is overstated, for the following reasons:

First: The study design, a cross-sectional survey, is weak and especially susceptible to confounding.

Second: The final participation rate was only about 1/5 the total invited sample, potentially leading to major selection bias (thus, undermining intent to obtain a nationally representative sample).

Third: The validity of the method to measure diet, (e.g., 10 to 12 year olds responded on their own) was not demonstrated.

Fourth: Malt beverage consumers were more physically active and watched less screen time: thus, they likely came from families with greater health consciousness. Therefore, the associations with higher intakes of a few micronutrients may simply reflect confounding. That is, they may have gotten the extra nutrients from other components of the diet. Unfortunately, this point was not examined in the study.

Fifth: Concerningly, the statistical models were not adjusted for physical activity and screen time. Doing so could have unmasked higher body weight among the malt beverage consumers. That is, perhaps the increased PAL and less screen time counteracted the adverse effects of beverage consumption.

Indeed, the true associations between sugar-sweetened beverages and body weight shown in higher quality prospective studies Schulze MB, 2004 are often not seen in cross-sectional surveys Forshee RA, 2004 – demonstrating the weak nature of this study type.