In contrast to what is often assumed, the present study showed that increased frequency of candy consumption was not associated with obesity or cardiovascular risk factors including blood pressure, lipid profile, and insulin sensitivity.
Adults with higher frequency of candy consumption consumed diets higher in energy, energy-adjusted carbohydrates, total sugars, added sugars, total fat, and saturated and monounsaturated fatty acids, and diets lower in protein, cholesterol and alcohol. Frequency of candy consumption was not, however, associated with objective measures of adiposity including BMI, waist circumference and skinfold measures, or with objectively measured cardiovascular risk factors including blood pressure, HDL or LDL cholesterol, or insulin sensitivity. No association was observed between frequency of candy consumption and the adiposity and cardiovascular parameters in models adjusted only for age, sex and race/ethnicity or in models adjusted for additional covariates with potential associations with the outcomes, including socioeconomic status (education, income), physical activity and sedentary activity, other dietary components including nutrients provided by foods other than candy, smoking, weight status, and use of medications for health conditions including elevated blood pressure, elevated cholesterol, and diabetes. The 95% confidence intervals (CI) of the odds ratio (OR) and the differences in point estimates of continuous variable combined with the standard error (SE) suggest that large undetected differences between groups or type 2 errors are unlikely.
The absence of any association between frequency of candy consumption and measures of adiposity or cardiovascular risk factors despite the fact that candy is a source of added sugars and saturated fat may in part be due to the relatively minor contribution of candy consumption to these dietary components. Based on NHANES data used in our assessment, adults were estimated to consume an average of 44 kilocalories (kcal) daily from candy . Candy accounted for slightly more than one teaspoon of added sugars (approximately 5 g) or 20 kcal in the diets of adults on a daily basis , which corresponds to a small fraction of the 100–150 calorie prudent upper limit of added sugars recommended by the AHA . Similarly, based on NHANES 2007–2008, Welsh and colleagues recently reported that candy and gum provided 4.5 to 6.4 g of added sugars in the diets of adults . Candy accounted for 3.1% of the total saturated fat intake by the US population aged 2 years and older in 2005–2006, or slightly less than 1 g based on a total saturated fat intake of 27.8 g/day [46, 47].
In contrast to the relatively modest contributions of candy to added sugars and saturated fat intakes, the top three dietary sources of added sugars for adults – sugary drinks, grain-based desserts, and sweetened fruit drinks – account for approximately 60% of the total added sugars intake . Sugary drinks alone, including sodas, energy drinks, sports drinks and fruit drinks, provide approximately 9.4 teaspoons of added sugars intake among adults, or approximately eight times the average amount provided by candy . Grain-based desserts provide approximately 2.8 teaspoons of added sugars in the diets of adults, or slightly more than twice the amount provided by candy, while dairy desserts provide approximately 1.2 teaspoons of added sugars, which is comparable to the amount provided by candy . The dietary sources of saturated fat are more diverse than the sources of added sugars. Cheese is the top ranked source of saturated fat in the US diet, accounting for 8.5% of total saturated fat intake, followed by pizza (5.9%), of which the saturated fat is presumably attributable primarily to cheese and a lesser extent to meat toppings .
Another possible reason that candy consumption was not found to be associated with unfavorable cardiovascular risk factors may be that although candy is a relatively minor component of the diet, cocoa-containing candy specifically can be a significant source of flavanols . Flavanols have been associated with beneficial effects on cardiovascular risk factors [12–14]. Additionally, stearic acid accounts for approximately one third of the total fat in cocoa butter and the majority of cocoa butter’s saturated fat . Unlike other saturated fatty acids, stearic acid is not known to raise LDL cholesterol levels . The lack of an association between frequency of candy consumption and cardiovascular risk factors could also be due to reverse causality, namely individuals identified with cardiovascular risk factors may be advised by their health care professionals to limit intake of saturated fat and added sugars, including candy. This possibility cannot be excluded in a cross-sectional study such as the present one when the outcome (e.g., bodyweight status) causes people to behave differently with respect to the exposure, namely candy intake.
As reporting of energy intake in dietary assessment tools is mostly dependent on body size, physical activity and under-reporting, the observed differences in energy intake between groups are not necessarily suggesting a more positive energy balance in frequent consumers of candy. Additionally, associations between frequency of candy consumption and usual macronutrient intakes cannot be attributed solely to candy. In order to better understand diet quality across the three categories of candy consumption, we conducted a post-hoc assessment of Healthy Eating Index-2005 (HEI-2005) scores  by frequency of candy consumption. Mean HEI-2005 scores of infrequent, moderate, and frequent candy consumers were 59.9 (95% CI: 58.0, 61.7), 57.5 (95% CI: 55.9, 59.2), and 56.7 (95% CI: 54.7, 58.7), respectively; the scores were not significantly different from one another, and are comparable to the mean HEI-2005 score of 57.2 for all adults in the US . Therefore, despite the observed associations between macronutrient intakes and frequency of candy consumption, diet quality as assessed by a comprehensive measure was not associated with frequency of candy intake. Although dietary factors have strong relationships with weight and cardiovascular risk factors, individual genetic and non-nutritional lifestyle factors also contribute to these factors. Furthermore, although the differences in intake of added sugars and saturated fat between groups were statistically significant, due to the large sample size, these differences may be too small to have a clinically significant impact on obesity and cardiovascular outcomes.
In the current analysis, adults were categorized into one of three categories based on reported frequency of total candy consumption during the past 12 months in a FFQ, which could be more relevant to health outcomes that develop over long periods of time than intakes based on a one- or two-day diet recall. Over the course of a year, nearly all adults - approximately 96% - reported consuming candy at least once. Use of the FFQ in this analysis therefore allowed for discrimination of adults into infrequent, moderate and frequent categories of consumption based on candy consumption over an extended period of time, thus avoiding the potential for misclassification of adults as non-consumers of candy based on just one day of recall.
The three frequency categories used in the analysis distinguish infrequent candy consumers from those who typically consume candy on most days and those with typical frequency of candy consumption between the two extremes. Given the range of reported frequency of candy consumption in the top category, there is heterogeneity in consumption within this group and potentially also outcomes. In order to better understand associations between the most frequent candy consumers and measures of body weight status and cardiovascular risk factors, a post-hoc analysis was conducted in which adults consuming candy more than once per day (8% of adults) were compared to all other adults. Results from this analysis showed that the most frequent candy consumers were not significantly more likely to be overweight or to have more adverse measures on cardiovascular risk factors (data not shown). These findings may appear to be somewhat unexpected, though are not entirely surprising in that some overweight individuals may eat candy less frequently due to dieting or in response to a healthcare professional’s guidance, and consequently underscore the challenges of using cross-sectional data to study diet and health associations. Additionally, given that portion size data were not collected in the FFQ, we do not know if the most frequent candy consumers in fact ate the most candy. Furthermore, for most people who consume candy more than once per day, the contribution of candy to the overall diet is likely still small and may be insufficient to have a meaningful health impact.
There is limited information in the literature on associations between candy consumption and measures of body weight and cardiovascular risk factors among adults. In a cross-sectional assessment based on a one-day dietary recall, O’Neil and colleagues  found that body weight, BMI, waist circumference, and risk of elevated diastolic blood pressure were lower in adults who reported consumption of candy compared to those who did not. Analyses by type of candy showed lower body weight and waist circumference in chocolate candy consumers compared to nonconsumers, and in sugar candy consumers compared to nonconsumers; mean BMI was lower in adults who reported consumption of sugar candy compared to those who did not, and consumers of chocolate candy had a reduced risk of lower HDL cholesterol and metabolic syndrome compared to nonconsumers of chocolate . In contrast to the current study, however, the study by O’Neil and colleagues was based on one day of dietary recall data rather than typical frequency of candy consumption, which could be more relevant to health outcomes that develop over long periods of time; approximately 22% of adults were identified as candy consumers in that study. In a cross-sectional assessment of adult males, median BMI was slightly though significantly higher among candy consumers compared to men classified as non-consumers (24.41 kg/m2, interquartile range (IQR): 22.95-26.44 and 24.39 kg/m2, IQR: 22.69-26.22, respectively, p < 0.001) . More recently, findings from a cross-sectional assessment of 1018 healthy adults showed that greater weekly frequency of chocolate consumption was associated with lower BMI (beta coefficient = −0.208, SE = 0.06, p = 0.001 in adjusted model) .
This study is a cross-sectional study, and therefore causality cannot be determined. Because of this limitation, longitudinal studies of associations between typical consumption of candy and anthropometric and physiologic measures are needed to better understand the role of candy in measures of health among adults. If the absence of associations between anthropometrics and physiologic measures observed in the current study is confirmed in longitudinal studies, the findings may help to focus concerns on dietary components more strongly associated with obesity and cardiovascular disease risk.
Other limitations of the study must also be considered. As with all dietary surveys, the accuracy of the estimates derived from reported intakes is limited by the accuracy of responses provided by survey participants. Misreporting of dietary intakes, specifically under-reporting of energy, occurs with both FFQs and 24-hour dietary recalls, and is more likely among obese than normal-weight individuals [54–56]. An analysis of data collected in 24-hour dietary recalls indicated that candy is among the food groups less likely to be reported by low energy reporters, and when reported, reported less frequently and in smaller portions . Reported frequencies of candy consumption on the FFQ may additionally be limited by self-interpretation of what constitutes “candy” as definitions of chocolate and non-chocolate candy were not provided. For example, individuals may or may not have interpreted non-chocolate candy to include gum or mints, and chocolate covered confections may have been categorized as chocolate or non-chocolate candies. Additionally, it was not possible to classify adults based on typical amount of candy consumed given that portion size information was not collected in the FFQ. Also, it is important to note that the intent of this analysis was to identify associations between frequency of consumption of all types of candy and the selected measures of health. If different types of candy have different effects on health, this analysis would reflect only the net effect.
There are, however, several strengths to the present study. The analysis was based on a large, nationally representative sample of the US population and classification of frequency of candy consumption was derived from reported typical consumption patterns over the past year using a tested instrument, providing a measurement of candy consumption more relevant to health and less susceptible to misclassification than measurements based on a single or two-day recall. All anthropometric and physiologic measurements were collected following established protocols.