This is an HTML version of an attachment to the Freedom of Information request 'obesity'.

Ref. Ares(2015)3265873 - 04/08/2015
A dose-response study of consuming high-fructose corn syrup-sweetened 
beverages on lipid/lipoprotein risk factors for cardiovascular disease in young 
Stanhope KL, Medici V, Bremer AA, Lee V, Lam HD, Nunez MV, Chen GX, Keim NL, Havel PJ. 
Am J Clin Nutr. 2015 Jun;101(6):1144-54. doi: 10.3945/ajcn.114.100461. 
  The aim of this study was to determine the dose-response effects of consuming beverages 
providing 0%, 10%, 17.5%, or 25% of energy requirements (Ereq) from HFCS on circulating 
concentrations of lipid/lipoprotein risk factors for cardiovascular disease (CVD) and uric acid in 
normal to overweight adults.  
  This was a parallel-arm, double-blinded diet intervention study with 3 phases:  
o  1) a baseline period during which subjects consumed an isocaloric standardized baseline 
diet in a controlled setting (inpatient);  
o  2) a 12-d outpatient intervention period during which subjects consumed their assigned 
sweetened beverages along with their usual ad libitum diets (non-isocaloric); and  
o  3) an inpatient intervention period during which subjects consumed isocaloric 
standardized diets that included the sweetened beverages.  
  Examining the difference in responses after 2 weeks, the authors concluded that “this study 
demonstrates for the first time that established risk factors for CVD, plasma concentrations of 
non–HDL cholesterol, LDL cholesterol, and apoB, increase in a dose-dependent manner in young 
adults consuming beverages providing 10%, 17.5%, or 25% Ereq from HFCS for 2 wk. The dose-
dependent increases of these risk factors for CVD, which were shown to be statistically 
independent of body weight gain, provide mechanistic support for the recent epidemiologic 
findings that there is increased risk of CVD mortality with increased intake of added sugar across 
  Authors report a strength of the current study was the presence of a biomarker in the study 
beverages providing an objective measure of compliance. The 3.5-d inpatient periods during 
baseline and at the end of intervention ensured that the study results were not confounded by 
noncompliance or variations in diet or physical activity during the days immediately preceding 
the blood collection procedures, were an additional strength.  
  The authors report that the study was not randomized increasing risk of bias and participants 
were given the study beverage in addition to an ad libitum diet throughout the 12-d outpatient 
period. Authors further state, this has been shown to result in an increase in total calories in 
previous studies and therefore it is not clear what the final responses actually measured (a few 
day or 2-week response to changes in HFCS beverage intake). 
 Please see ABA and CRA statements.  

Sugar-sweetened beverages, vascular risk factors and events: a systematic 
literature review. 
Keller A, Heitmann BL, Olsen N. Public Health Nutr. 2015 May;18(7):1145-54. doi: 

  Researchers Keller, Heitmann and Olsen conducted a systematic review of literature and 
concluded that there is a relationship between SSB intake and vascular risk factors, and a less 
consistent relationship with vascular events (i.e., stroke, fatal and non-fatal myocardial 
infarction, vascular death).  
  The evaluation of full-text articles resulted in the inclusion of ten prospective studies and one 
RCT in the review. The quality of these studies was assessed using the Academy of Nutrition and 
Dietetics’ (US) Quality Criteria Checklist: Primary Search from the ADA Evidence Analysis 
Manual, and ranged from good to medium quality.  
  Of the five identified prospective studies using vascular events as outcomes, two found direct 
associations between SSB consumption and CHD, one found an association between SSB 
consumption and combined vascular events, and the remaining two reported an association 
between SSB consumption and stroke. Six studies using vascular risk factors as outcomes found 
direct associations between baseline or change in SSB consumption and changes in blood 
pressure or lipid metabolism. However, one study only found a small direct association with 
diastolic blood pressure but not systolic blood pressure, while examining baseline SSB intake.  
  The strength of the evidence was graded as ‘fair’ for the association between SSB and vascular 
risk factors, and as ‘limited to fair’ for the association between SSB and vascular events as well 
as for diabetes, hypertension, BMI and energy intake as mediators.  
  The authors concluded that strength of the evidence relating SSB and CVD is still limited, and 
warrants further study. The articles reviewed “generally showed discrepant results for the 
association between SSB intake and vascular events, while the evidence for an association 
between SSB and vascular risk factors was stronger”. 
  This review included only published articles, which presents the possibility of publication bias, in 
which studies showing an association between the variables in question are favored. It is 
possible that the positive associations shown may have been limited, had the authors included 
the results of unpublished articles.  
  Most of the studies included were from the United States, limiting the generalization of the 
results to other racial and ethnic populations.  
  The use of Food Frequency Questionnaires (self-reported dietary intake assessment tool), is 
prone to measurement errors or recall bias, leading to over- or underestimation of dietary 
intake. The majority of the studies included in this review utilized this method. 

Sugar-sweetened beverage consumption and central and total adiposity in older 
children: a prospective study accounting for dietary reporting errors. 
Bigornia SJ, 
LaValley MP, Noel SE, Moore LL, Ness AR, Newby PK. Public Health Nutr. 2015 May;18(7):1155-63. doi: 
  Researchers Bigornia et al., conducted a prospective study to determine the relationship 
between changes in sugar-sweetened beverage (SSB) intake and waist circumference (WC), BMI 
and body fat in children at age 13. A stronger proportional association was seen between higher 
consumption of SSB and larger waist circumference, independent of differences in total 
adiposity, once errors in dietary reporting were accounted for.  
  Dietary intake, WC and BMI (determined from weight and height measurements) were 
measured at ages 10 and 13 years. Total body fat mass and physical activity levels were assessed 
at 13 years. Covariates including maternal height and weight, and educational level (as a proxy 
for socio-economic status), in addition to the participants’ pubertal stage, dieting status, and 
change in fruit juice, fruit, vegetable, and total fat intakes were also assessed.  
  At baseline, less than a quarter of participants were overweight or obese, and approximately 
60% of the boys and girls were consuming SSB at age 10. At age 13, change in SSB consumption 
was positively associated with higher weight and BMI, and a weaker effect was shown on WC 
and total body fat mass.  
  The authors concluded that the results of this study “provide evidence that SSB consumption in 
children influences total fat mass accumulation” and “add further support to recommendations 
to curtail intakes of SSB as a means to combat excess weight gain in children”. 
  The interpretation of these finding should be done with caution, given the concern for residual 
confounding due to the fact that the cause of weight gain is multifactorial. Several potential 
confounders (such as, meals consumed outside of the home, dieting status, and fruit, vegetable 
and fat intake) were either not assessed, or were not available for all participants.  
  The observational nature of this study, as well as the implementation of dummy variables in 
order to preserve the sample size and account for missing data, also adds to the concern for 
residual confounding. Dummy coding likely resulted in underestimation of the results of this 
study, once potentially confounding variables were collapsed into dichotomous measures 
(no=never and yes=all other responses).  
  Lastly, maternal BMI was calculated using self-reported height and weight, and errors in reports 
may result in misclassification which will inevitably impact the findings of the study.