Sugar-sweetened beverages and health: where does the evidence stand?

The relation between sugar-sweetened beverages (SSBs), obesity, and related health outcomes has increasingly attracted public and scientific interest. Since the late 1970s, intake of SSBs has increased more than 2-fold (1), and currently they are now the primary source of added sugar in the US diet (2). On average, SSBs contain 140–150 kcal and 35–37.5 g sugar per 12-oz serving. The prevailing mechanisms linking SSB intake to weight gain are low satiety of liquid calories and incomplete compensatory reduction in energy intake at subsequent meals, leading to an increase in total energy intake (3). SSBs also induce independent metabolic effects through their contribution to a high dietary glycemic load, leading to inflammation, insulin resistance, and impaired b cell function (3, 4). In addition, regular consumption of SSBs is associated with high blood pressure and accumulation of visceral adipose tissue and dyslipidemia through increased hepatic de novo lipogenesis. Numerous epidemiologic studies have examined the relations between SSBs, obesity, and related cardiometabolic diseases. Whereas most studies have found positive associations, some have yielded inconsistent results, sparking controversy in the field. In general, associations are strongest and most consistent in large prospective cohort studies with long durations of follow-up and repeated measures of intake that capture long-term dietary patterns. Studies that do not adjust for the potential mediating effect of total energy intake in analyses tend to yield stronger associations; adjustment for total energy is equivalent to removing any effect of SSBs on body weight that occurs through energy intake and may thus attenuate the association (5). Furthermore, studies funded by the food industry tend to report significantly weaker associations than do non–industry-funded studies (6). A number of reviews have been conducted to summarize the current evidence in an effort to guide clinicians, public health experts, and policymakers. However, individual reviews may be prone to bias and may vary in terms of quality. For these reasons, rigorous evaluation of published reviews can be useful for guiding public health decisions, especially for controversial topics with far-reaching implications. In this issue of the Journal, Weed et al (7) conducted a systematic review to assess the methodologic quality of reviews or meta-analyses published in the past decade that examined the relation between SSB and health outcomes. However, rather than shedding more light on this pressing public health issue, the report obscured important relations between SSB consumption and harmful health consequences. To assess study quality, Weed et al used the Assessment of Multiple Systematic Reviews (AMSTAR) tool to assess the quality of the studies. AMSTAR is a validated scoring system based on 11 questions specifically designed to evaluate the methodologic quality of systematic reviews (8). Systematic reviews, which include meta-analyses, use a comprehensive and criterion-based selection of relevant evidence with methods clearly stated and reproducible by others, an appraisal of validity, and an objective or quantitative summary of the evidence. In contrast, narrative reviews (including position statements by professional organizations), often written by experts in the field, are usually broader in scope but do not necessarily state criteria relevant to assessing the methodologic quality of systematic reviews (9). For example, narrative reviews typically do not include a comprehensive literature search strategy, detailed tabulation of study characteristics, data extraction methods, or an assessment of publication bias. For this reason, some studies that use AMSTAR for quality appraisal of systematic reviews have excluded narrative reviews (10). The maximum AMSTAR score a review can receive is 11 (11 for meta-analyses and 10 for systematic reviews), with scores of 0–4 indicating low quality, 5–8 moderate quality, and 9–11 high quality (9). Overall, 17 reviews of obesity, diabetes, metabolic syndrome, and coronary heart disease were included in the report and found to be of moderate to low quality on the basis of AMSTAR scoring (mean 1⁄4 4.4, median 1⁄4 4, range 1–8.5) (7). Of the 17 reviews, 7 were narrative and thus did not possess many of the components relevant to the AMSTAR score. By applying the AMSTAR instrument equally to narrative and systematic reviews, the authors underestimated the quality of the reviews on SSBs and health, leading to a spurious conclusion. Not surprisingly, the studies that used a systematic or metaanalysis approach received high scores, whereas the nonsystematic narrative reviews received low scores. By lumping these 2