2.1 Sample population

Data from 23,708 adults, over 30 years of age, from the 2005–2016 NHANES were analyzed to determine food and nutrient intake categories and their contribution of nutrients from snacking occasions in US adults stratified by diabetes status. NHANES is a federally funded surveillance system that monitors the health and nutrition status of the US population [17]. For survey data collected for this analysis, NHANES utilized a multistage, stratified sampling design to select participants from the noninstitutionalized population. Data were collected through questionnaires as well as physical and laboratory examinations performed in the mobile examination center. Subpopulations that are difficult to reach are oversampled and weighted to create a nationally representative data set. The survey collects demographic, socioeconomic, dietary, and health data via a mobile clinic. Demographic data, such as age, gender, marital status, and family income in relation to the federal poverty level, were collected during in-home interviews.

2.2 Ethical considerations

Public use data files from the National Health and Nutrition Examination Survey were obtained from the Centers for Disease Control and Prevention National Center for Health Statistics website for analysis. Participants gave written informed consent, and the National Center for Health Statistics’ Research Ethics Review Board approved all data collection methods.

2.3 Laboratory data: Classification levels of glycemic control

Glycated hemoglobin (GHb) was used to categorize diabetes status. During the visit to the NHANES mobile examination center, a blood sample was collected from consenting participants. Participants were categorized into levels of glycemic control according to their GHb percentage value: nondiabetes (< 5.7%, 39 mmol/mol); prediabetes (5.7–6.4%, 39-46mmol/mol); controlled diabetes (6.5–6.9%, 48–52 mmol/mol); and poorly controlled diabetes (≥ 7%, 53mmol/mol) [18]. Participants without data were excluded from the analyses (n = 2,314, 9%).

2.4 Dietary data

One recall per participant was assessed from 24-hour dietary recalls conducted by trained interviewers in the mobile examination center. Recalls utilized the Automated Multiple Pass Method, a validated, multistep interviewing technique shown to improve the accuracy of dietary intake data [19]. Participants reported the foods and beverages consumed from midnight to midnight from the prior day. Eating occasions were self-designated meals during the recalls and were reclassified as breakfast, lunch, dinner, or snack/other eating occasions for analysis. Individual nutrient and MyPlate equivalent intake estimates were generated using the Food and Nutrient Database for Dietary Studies [20] and the Food Patterns Equivalents Database [21]. To identify the types of foods contributing to snacking intakes, individual foods reported in the recall were classified according to the What We Eat in America food categories [22]. Participants who did not complete the dietary intake recall interview were excluded from analysis (n = 2,953, 11%).

2.5 Estimating snacking intakes

Total intakes of nutrients and MyPlate equivalents (eq) from snacking occasions were aggregated across all foods and beverages consumed from eating occasions described as snacks or other eating occasions. The proportion of the day’s intakes obtained from snacking occasions was computed as the sum of the nutrient or food group from snacks divided by the intake of the respective nutrient or food group from the total day. Average snack intakes were computed from total intakes from snacking occasions per the number of unique snack occasions reported. The highest energy snacking occasion was identified to represent the unique snacking occasion with the highest energy content per individual.

2.6 Statistical analysis

Complete dietary, demographic, and laboratory data were downloaded and imported into SPSS Complex Samples (version 24, IBM SPSS Inc., Chicago, IL). Multiple linear regressions were performed to examine intakes of snacks in relation to glycemic level among NHANES participants. Means and standard errors were generated for individual nutrient intakes and proportions of total intake for all reported snacking occasions by glycemic level. Analysis was controlled for age, race/ethnicity, income, marital status, and gender. Descriptive statistics were used to produce estimates of nutrients and food groups for an average snacking occasion, as well as for the average highest energy snacking occasion. The proportional contribution of snacking energy from USDA food categories was obtained as the snacking energy obtained from a food category divided by the total snacking energy reported, across levels of glycemic control. Sampling weights provided by the Centers for Disease Control were used to generate a nationally representative sample of population-based estimates. SPSS Complex Samples produced sample-based standard errors.

3.1 Demographic characteristics according to diabetes status

Complete dietary intake data from 23,708 adults were evaluated from 2005–2016 NHANES. In general, males and females were equally distributed across each diabetes status. Compared to those without diabetes, those with prediabetes and type 2 diabetes were older (mean age of 50.5 years compared with ≥58.6 years, respectively, p<0.001). A larger percent of those without diabetes had a college degree (34.3%) compared to those with prediabetes (23.1%), controlled type 2 diabetes (19.1%), or poorly controlled diabetes (16.5%, p<0.001). Among the population surveyed, non-Hispanic whites represented the largest number in each diabetes status category followed by non-Hispanic blacks and Mexican Americans (p<0.001).

3.2 Nutrient Intakes from snacking occasions

Estimates of intakes from snacking occasions and the proportion of the day’s snack intakes are presented in Tables 1 and 2, respectively. Those without diabetes and with prediabetes, controlled type 2 diabetes, and poorly controlled type 2 diabetes consumed an average of 2 snacks per day, with no significant differences by diabetes status (p = 0.264). On the day of intake, combined snacking occasions contributed approximately 498 kcals of energy to the day’s intakes, which accounted for 22.4% of the total daily energy intake for adults without diabetes. Adults with prediabetes and diabetes (controlled or poorly controlled) consumed significantly less, accounting for 21.3%, 19.5%, and 19.6% (467, 435, and 423 kcal, p<0.001) of the day’s energy from snacks, respectively. For both groups with type 2 diabetes (controlled and poorly controlled), intakes of carbohydrates from snacks as well as the percent of the day’s carbohydrate intakes were significantly lower (59.0 g/22.6% and 57.8 g/23.1%, respectively, p = 0.002) compared to the nondiabetes group (67.1 g/25.3%). Added sugar intakes from snacking occasions followed similar trends as total carbohydrates, with the lowest intakes and contribution to the total day in those with type 2 controlled diabetes (35g/30.4%), and poorly controlled diabetes (33.8 g/31.0%) groups (p = 0.001). There were no differences in the amount of total fat and saturated fat from snacking occasions, but those in the nondiabetes group had significantly more protein than the prediabetes group (p = 0.023). The amount of protein from combined snacks was similar between the nondiabetes group and those with poorly controlled type 2 diabetes, but as a percentage of total daily protein intake, the poorly controlled type 2 diabetes group consumed significantly less protein (11.8% vs 13.0%, p = 0.004).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Energy and nutrient intakes from all snacks and the percent of the day’s intakes of energy and nutrients obtained from snacks in people aged 30+ years with different diabetes status.

https://doi.org/10.1371/journal.pgph.0000802.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Proportion of the day’s total intakes of energy and nutrients obtained from snack occasions in people aged 30+ with different diabetes status.

https://doi.org/10.1371/journal.pgph.0000802.t002

Intakes of vitamin B₆, and magnesium from snacks, were significantly lower in those in the controlled and poorly controlled type 2 diabetes groups compared to those in the nondiabetes group (p<0.001). However, as a percentage of total daily magnesium intake, there was only a significant difference between the nondiabetes and controlled type 2 diabetes group, whereas significance remained for vitamin B₆ (p<0.001).

3.3 Food group and nutrient profile of an average snack

On the day of intake, adults with poorly controlled type 2 diabetes consumed significantly less energy (p = 0.002), carbohydrate (p<0.002), and added sugars (p<0.001) per average snacking occasion compared to adults with prediabetes or without diabetes (Table 3). For the average snack, adults with controlled type 2 diabetes consumed modestly but less total protein foods (0.29 oz. eq.) compared to those without diabetes (0.36 oz. eq., p = 0.14), yet significantly higher amounts of total vegetables (including legumes) (0.1 cup eq.) than those without diabetes (0.05 cup eq.), prediabetes (0.07 cup eq.) or poorly controlled type 2 diabetes (0.05 cup eq., p = 0.004). People without diabetes and with prediabetes consumed significantly more added sugars compared to those with poorly controlled type 2 diabetes (p<0.001). Regardless of diabetes status, refined grains accounted for >80% of all grains when examining the average and highest snack. The proportion of whole grains was not significantly different by diabetes status. Additionally, adults with controlled type 2 diabetes consumed significantly less fruit juice than any other group per average snacking occasion (p<0.001). In all groups, the average snack did not contain dark green vegetables, starchy vegetables (excluding white potatoes), legumes, organ meat or seafood high in omega-3 fatty acids.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Energy and nutrient content of the average snack occasions in people aged 30+ with different diabetes status.

https://doi.org/10.1371/journal.pgph.0000802.t003

3.4 Food group and nutrient profile of highest energy snack

The highest calorie snack for those with poorly controlled type 2 diabetes had significantly less energy (p = 0.003) and carbohydrates (p = 0.027), compared to the highest calorie snack for adults with prediabetes or without diabetes (Table 4). Adults with poorly controlled type 2 diabetes consumed significantly fewer total vegetables (p = 0.002) compared to those with prediabetes and controlled type 2 diabetes. Adults with controlled type 2 diabetes consumed significantly less fruit juice during the highest energy snack compared to people without diabetes or prediabetes (p = 0.001). Like the average snacking occasion, people with poorly controlled type 2 diabetes consumed significantly less added sugar during the highest calorie snack compared to those with prediabetes and those without diabetes (p<0.010).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Energy and nutrient content of the highest calorie snack in people aged 30+ with different diabetes status.

https://doi.org/10.1371/journal.pgph.0000802.t004

3.5 Food sources of energy from snacks

Across all adults, snacks and sweets and non-alcoholic beverages accounted for approximately half of the energy consumed during snacking occasions (Fig 1). Alcoholic beverages contributed to 14.6% of the energy from snacks in the nondiabetes group, which was two-fold greater than for the other groups. The lowest proportion of non-alcoholic beverages was noted in the controlled diabetes group, but they also obtained 47% of snack energy from the snacks and sweets category. Fruits and vegetables accounted for 5% of energy from snacking occasions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Proportion of food sources of energy consumed during snacking occasions across levels of glycemic control.

https://doi.org/10.1371/journal.pgph.0000802.g001

4.1 Snacking according to diabetes status

Few studies address the association between snacking and diabetes status. Mekary, et al. analyzed cohorts of men and women and determined snacks beyond the three main meals per day increased the risk for Type 2 diabetes [7, 25]. Other research concludes that snacking beyond three main meals per day has little impact on glycemic control, but other factors such as timing and overall energy intake do affect glycemic control [26, 27]. In a study with 163 participants who had insulin-treated diabetes, there were no correlations between the use of snacks and GHb [26]. However, as elevated postprandial glucose may contribute to glycated hemoglobin (GHb) [28], it seems logical that the glycemic response after each eating occasion should be addressed. In the present study, regardless of diabetes status, adults consumed an average of 2 snacks on the day of intake. Interestingly, adults without diabetes consumed significantly more calories in the form of snacks, and snacks made up a higher portion of energy intake compared to all other groups. While the number of snacks was similar between groups, the energy content of snacks in relation to the total daily energy consumed varied on the day of intake and may play a role in glycemic control. As such, the information from this study can contribute towards making specific snack-related nutrition recommendations.

Increased intake of refined carbohydrates and sugars, in addition to decreased fiber, parallel the increased prevalence of type 2 diabetes [29–32]. In our cohort, dietary fiber intake from snack foods was similarly low ranging from 1.6–1.7 g/day along with a poor intake of whole grains while total sugar intake was relatively high (17.1–20.3 g/day), suggesting poor overall carbohydrate quality regardless of diabetes status. The average snacks consumed were higher in carbohydrates, total sugar, and added sugar for those without diabetes and with prediabetes than those with type 2 diabetes. Although the contribution of added sugars is still relatively high, this suggests that adults with type 2 diabetes may be taking steps to limit carbohydrates and sugar in their diets as stressed in medical nutrition therapy guidelines. However, those with prediabetes consumed more lower quality carbohydrates on the day of intake indicating a potential gap and opportunity for increased nutrition education in this group. The 2019 ADA Consensus Report now includes recommendations for people with prediabetes whereas previous ADA nutrition position papers did not [3], emphasizing the increased importance in prevention (not just treatment) of type 2 diabetes. The consensus report outlines the importance of implementing intensive lifestyle/behavior interventions to improve eating habits and increasing exercise to induce weight loss which is associated with a reduction in the incidence of type 2 diabetes in adults with prediabetes. Given the poor quality of carbohydrates consumed in this study, especially in those with prediabetes, it would be advantageous to incorporate specific snacking guidelines.

Protein is an essential macronutrient than can support satiety, and aid in the maintenance of lean body mass [12, 33–36]. Despite the high amount of sugar and refined carbohydrates consumed during snacks in all groups, the higher protein intake in people without diabetes during snacks may aid in satiety and reduce overeating at other meals. People without diabetes also consumed less processed meats as well as more nuts, seeds, and soy, while those with type 2 diabetes consumed more cured and luncheon meats during snacks. In addition to protein quantity, the sources of protein and their contributions with other macronutrients may impact glycemic response. In fact, it’s been shown that consuming glucose along with lean protein (such as lean beef and turkey) results in a lower blood glucose response compared to consuming glucose alone in people with type 2 diabetes [37]. Although glucose is the primary stimulus for insulin release, protein and amino acids can enhance the release of insulin leading to faster clearance of glucose from the blood [38]. Higher protein intake (30% of total daily calories) in people with type 2 diabetes is associated with lower GHb [39]. Thus, people with type 2 diabetes may benefit from increasing the protein content of their snacks, focusing on lean protein choices and consuming less processed meats, while maintaining overall energy density of their snack.

The literature is lacking to define the snack consumption, composition and contribution to daily calories and nutrition over a 24-hour period stratified by levels of glycemic control in a US population cohort. Overall, adults with type 2 diabetes consumed less energy, carbohydrates, and total sugars from snacks than adults without diabetes. Those with controlled type 2 diabetes consumed more vegetables and less fruit juice than other groups, yet adults with diabetes in general consumed more cured and luncheon meats than adults without diabetes and with prediabetes. Finally, adults without diabetes consumed more protein from snacks, including less processed meats. Regardless of diabetes status, the nutrition content of snacks was poor including mostly refined grains, added sugar, and lacked dark green vegetables, legumes, or seafood high in omega-3 fatty acids.

4.2 Strengths/Limitations

This study was performed utilizing a large national dataset, which contributes to the validity and reproducibility of our findings. In the absence of consistent fasting blood glucose values, GHb was the objective measures used to determine diabetes status. While it is less influenced by fasting status, it may not be as sensitive of a measure for prediabetes.

While self-report dietary data has limitations regarding correct reporting of type and quantity of food consumed, self-defining snacking occasions may be a strength [13, 40]. Previous studies have defined snacking by time of day, caloric level, foods consumed outside of the standard three meals per day, or have chosen to exclude certain beverages, or beverages altogether [7, 14, 16, 25, 27]. While defining snacks in these ways is objective, it does not address snacking as a cultural construct, which allows for diverse interpretation based on cultural norms [13]. This analysis will not account for individuals who consume small, frequent meals and should be considered when interpreting the findings related to snacks. While there is potential for snacking intakes to change over this period of 2005–2016, the mean intakes of energy and macronutrients from snacks across the data cycles was generally consistent. Due to the cross-sectional nature of this data set, the scope of interpretation is limited to difference based on the foods and beverages reported on the day of intake. These intakes cannot be assumed to reflect usual intakes, including usual snacking patterns. This study contributes significantly to the literature on snacking and diabetes. Future work could explore these trends in relation to total energy intakes and obesity to evaluate the covariance with diabetes status.