Abstract

1. Introduction

Parallel with increasing rates of obesity, the prevalence of prediabetes in the US has grown to 38% of the adult population [1]. Although estimates vary by population characteristics and the definition of prediabetes, about 10–50% of individuals with prediabetes will progress to overt diabetes within the next 5 to 10 years [2]. Notably, overweight/obesity can double the lifetime risk of developing diabetes [3]. Accumulating evidence also points to prediabetes being a high-risk metabolic condition associated with a multitude of current and future diseases, including cardiovascular disease, non-alcoholic fatty liver disease, neuropathy, chronic kidney disease, cancer, dementia, and all-cause mortality [2,4,5].

Data from multiple intensive lifestyle intervention studies directed toward diet modification, increased physical activity, and weight loss in diverse patient populations have shown a 40–70% relative risk reduction in developing diabetes compared with standard care [6,7,8,9]. While there is evidence to support a variety of eating patterns tailored to individual preferences and metabolic goals in individuals with prediabetes [10], reducing overall calorie intake and glycemic load remains an important component of nutrition counseling. However, adherence to low-carbohydrate diets is challenging. Some of these challenges include the hedonic response to sugar triggered by the release of dopamine possibly leading to compulsive eating, as well as cultural and economic barriers [11,12]. An alternative, novel approach, researched by our group and others, to attenuating the glycemic effect of a meal is to consume concentrated carbohydrates at the end of a meal, after protein and non-starchy vegetables (“carbohydrate-last food order intervention”) [13,14,15,16]. In a previous crossover study of patients with metformin-treated type 2 diabetes (T2DM), the ingestion of protein-rich food plus non-starchy vegetables before concentrated carbohydrates reduced the incremental areas under the curve for glucose (iAUC0–180) and incremental glucose peaks by over 50%, compared to eating the same foods in the reverse order [17]. In addition, our group also found that levels of the hunger hormone ghrelin were more suppressed 3 h after a carbohydrate-last meal, suggesting this strategy may have implications for satiety and weight management [17]. Furthermore, the glycemic effects of a carbohydrate-last food order were replicated in a subsequent study in individuals with prediabetes [18]. This study additionally showed that post-meal glucose spikes were similarly reduced by ~40% when non-starchy vegetables alone or with protein were consumed prior to the carbohydrate-rich component of the meal.

This prospective pilot study expands what is known from previous research on nutrient order and tests the feasibility and metabolic effects of this behavioral intervention in the real world. A 16-week carbohydrate-last food order behavioral intervention was implemented in adults with prediabetes to assess its feasibility and impact on glucose tolerance (GT), HbA1c, weight, and nutrient intake. We postulated that a carbohydrate-last food order was a feasible behavioral strategy in individuals with prediabetes and that the addition of food order counseling to standard nutritional counseling would improve glucose tolerance, HbA1c, and nutrient intake, while reducing body weight.

2. Materials and Methods

2.3. Intervention Overview

2.3.1. Standard Counseling

Participants in both groups received 30 min of standard nutritional education and counseling at the baseline visit by a registered dietitian who was blinded to study assignments. The dietitian reviewed the baseline dietary intake record completed by each participant and counseled them on the tenets of a balanced, healthful diet in accordance with the USDA MyPlate framework [19]. Both groups received 15 min of standard nutritional counseling from research assistants at weeks 4, 8, and 12, consisting of education on portion sizes and macronutrients, healthy beverages, and limiting salt intake, respectively. The participants in both groups were advised not to alter their level of physical activity for the duration of the study.

2.3.2. Food Order Counseling

Participants in the carbohydrate-last (FO) group received additional carbohydrate-last food order counseling for 15 min by research assistants at baseline and weeks 4, 8, and 12. There were three graphs illustrating the results of previous food order intervention studies shown to participants to enhance their understanding of the influence of dietary intervention on blood glucose, insulin, and ghrelin responses in acute conditions. The food order counseling initially focused on the clear categorization of foods into the three categories of interest: vegetable, protein, and carbohydrate. This required recategorization of some traditional food types with a high carbohydrate content to be moved around within the traditional food category placements. For example: peas, corn, fruit, milk, and non-Greek yogurt were placed into the carbohydrate category whereas other dairy products such as cheeses, cottage cheese, and Greek yogurt were placed into the protein category (Figure 1).

Open in a separate window

Figure 1

Handout given to patients categorizing foods into macronutrient groups.

The second component of food order counseling focused on advising patients to consume their vegetables and protein portions before their carbohydrates at each meal. For meals where this was not feasible, such as when consuming a sandwich, patients were advised to start the meal with a salad or other non-starchy vegetables. Handouts were provided to patients outlining the carbohydrate-last food order intervention using the basic framework of USDA MyPlate and categorizing common foods into macronutrient groups (Figure 1 and Figure 2).

Open in a separate window

Open in a separate window

Figure 2

Handout given to patients outlining the carbohydrate-last food order intervention using the basic framework of the USDA MyPlate.

Additionally, the participants in the FO group were instructed to record after every meal consumed “yes” if they followed the food order for the meal and “no” if they did not follow the food order. Subjects were told to carry this out daily for the entirety of their study participation. These records, which were termed food order logs, were analyzed to determine adherence to and the feasibility of the carbohydrate-last food order intervention.

A study exit questionnaire was administered to the participants in the FO group at the end of 16 weeks to assess the overall feasibility of the intervention (Supplementary S1). Participants in both groups attended in-clinic counseling sessions until the declaration of the COVID-19 pandemic in March 2020. Thereafter, counseling sessions were delivered via telemedicine (telephone/video visits) or in clinics based on participant preference and the prevailing pandemic-related restrictions.

3. Results

A total of 80 individuals were screened, 45 were randomized (C 24, FO 21), and 39 completed the study (C 21, FO 18). In the FO group, three subjects withdrew due to the COVID-19 pandemic, and in the C group, three subjects withdrew due to personal reasons (Supplementary S3). The demographics and baseline variables including weight (191.1 lbs vs. FO 190.8 lbs), HbA1c (6% vs. 6%), and caloric intake (2131.9 kcal vs. FO 1972.2 kcal) were similar between groups, except fiber intake, which was greater in the control group (24.9 ± 10.6 g vs. FO 18.3 ± 6.4 g, p = 0.023). Exercise levels did not differ between groups at baseline or week 16 (Table 1).

The number of participants achieving a 15% reduction in 2 h glucose tolerance was 2 out of 14 in the FO group and 4 out of 21 in the C group (Fisher’s exact test p = 1.000). Within the FO group, body weight decreased from baseline to week 16 (−3.6 ± 5.7 lbs, p = 0.017; −1.8% ± 2.8, p = 0.012), and there was a trend toward improvement in HbA1c (−0.1 ±0.2, p = 0.054) (Table 2). The control group weight trended lower (−2.6 ± 6.8 lbs, p = 0.102; −1.6% ± 3.5, p = 0.048) without altering HbA1c (−0.03 ± 0.3, p = 0.605). Changes in weight, HbA1c, lipids, and glucose tolerance (GT) were not significantly different between groups. There were no significant changes in the HOMA-IR or Matsuda index in either group; however, a trend toward improvement in the oral disposition index was observed in the FO group (38.4% ± 75.6%, p = 0.080).

At week 16, only the C group significantly reduced its daily intake of calories (−292 ± 506 kcal/day, p = 0.016), fat (−13.0 ± 25.4 g/day, p = 0.029), protein (−15.0 ± 26.3 g/day, p = 0.017), and grains (−1.0 ± 1.9 ounce equivalents per day, p = 0.027) whereas the FO group increased its daily intake of vegetables (1.0 ± 1.6 cups per day, p = 0.019) and protein (2.5 ± 5.1 ounce equivalents per day, p = 0.050) (Table 3).

A sensitivity analysis of the pre-COVID participants showed a weight loss trend between the FO and C groups (FO −5.9 ± 5.3 lbs vs. C −1.0 ± 6.8 lbs, p = 0.065) (Table 2). Similar to the results from all participants, only the pre-COVID participants in the C group significantly reduced their intake of calories (−344 ± 478 kcal/day, p = 0.030) and fat (−18.8 ± 26.5 g/day, p = 0.032), while the FO group increased its intake of vegetables (1.0 ± 1.5 cups/day, p = 0.041 (Table 3).

The 18 participants in the food order group completed their food order log, on average, for 60.5% of their total days in the study. Participants reported following the carbohydrate-last food order intervention for 79.2% of reported meals consumed in free-living conditions over the course of 16 weeks.

At week 16, 17 FO group subjects (94%) reported high intervention adherence and 13 (72%) reported it was easy to eat vegetables/protein before carbohydrates. Although 44% felt eating vegetables/protein before carbohydrates reduced their meal enjoyment, 94% reported they were likely to continue implementing this behavioral strategy in the future (Table 4).

4. Discussion

In this 16-week study of adults with overweight/obesity and prediabetes spanning the COVID-19 pandemic, we demonstrated that the addition of carbohydrate-last food order counseling to standard nutrition counseling improved nutrient intake, notably increasing the intake of protein and vegetables, although reductions in body weight and HbA1c were not significantly different from those observed with standard nutrition counseling alone. Participants reported high intervention adherence and likelihood of continuing to eat protein/vegetables before concentrated carbohydrates at meals, suggesting carbohydrate-last food order is a feasible behavioral strategy.

Consuming a substantial quantity of fruits and vegetables is foundational to a healthy diet and has been shown to reduce the risk of cardiovascular disease, cancer, and all-cause mortality [24,25,26,27,28]. However, even with the known benefits of consistent vegetable intake, only 9% of adults in the United States meet the federal daily vegetable recommendation [28]. Based on our findings, food order counseling is a simple strategy that can significantly enhance vegetable consumption. Interestingly, weight loss in the FO group occurred concurrently with significant increases in vegetable and protein intake without a significant change in caloric intake, unlike in the C group, which did significantly reduce its intake of calories, fat, and protein. Dietary strategies that emphasize foods to add to the diet rather than those that focus on what to avoid are highly useful in practice, and this evidence supports the validity of the carbohydrate-last approach in adults with prediabetes.

Contrary to our postulated hypothesis, we did not observe significant differences between the study groups in glucose tolerance and HbA1c. This was not entirely unexpected given the unprecedented disruptions of the COVID-19 pandemic. No changes to physical activity were observed in either study group, supporting the objective to isolate the effects of the study intervention. Improvements in glucose tolerance have previously been described in more intensive lifestyle intervention programs that included moderate to vigorous intensity exercise [29]. Considering contextual factors, a 0.1% reduction in HbA1c within the FO group is notable. Limited data from previous prospective studies suggest that manipulating the sequence of meal ingestion improves glycemic control in individuals with T2DM [13,30].

Given previous investigations of the carbohydrate-last food order intervention demonstrate perturbations in insulin, GLP-1, and ghrelin concentrations that could potentially favor weight management, we postulated greater weight loss in the FO group. Indeed, in the pre-pandemic cohort of 24 participants (12 in each group), we did observe an average weight loss of 5.9 lbs vs. 1 lb in the FO vs. C groups. This difference attenuated over the pandemic period, when the once-monthly, in-person counseling sessions were converted into telemedicine appointments. We suspect the pre-COVID group had greater weight loss in part due to the in-person counseling visits and greater daily lifestyle stability, which may have led to greater effectiveness of FO counseling. Remarkably, while it has been shown that eating behavior changed during the COVID-19 pandemic with a shift toward increased snack frequency and preference for processed foods at the expense of fruits and vegetables [31,32], we continued to observe improved diet quality in the FO group.

Our study is novel, being the first prospective study to investigate the feasibility and effectiveness of the carbohydrate-last food order behavioral intervention in prediabetes. The study design included parallel intensity arms and the use of uniform scripts for standard counseling and structured counseling materials for CHO-last food order counseling. The implementation of this intervention is simple as it builds upon the messaging of the widely used USDA MyPlate framework, emphasizing nutrient sequencing.

Our study has limitations, including sample size and short study duration. As in most nutritional studies conducted in free living conditions, we had limited oversight of participants’ dietary intake. Dietary recall via self-reporting is commonly used for nutritional studies. However, the validity and reliability of dietary recall are limitations in nutrition trials. Additionally, adherence to carbohydrate-last food order behavior may be overestimated due to respondent bias given the nature of self-reporting [11]. A limitation of generalizability is the demographic of the cohort being primarily white, as adherence to dietary interventions can vary widely by race and ethnicity as a proxy for culture. Furthermore, a significant proportion of subjects completed this study during the COVID-19 pandemic, which likely impacted individuals in different ways, adding multiple variables, such as a change in work schedule, access to food, and a change in overall routine, that could not be accounted for.

5. Conclusions

In conclusion, carbohydrate-last food order is a feasible behavioral strategy that can be implemented by individuals with prediabetes. Carbohydrate-last food order behavioral counseling led to improved diet quality, notably greater vegetable and protein intake. Further research with a larger sample size and longer duration is needed to delineate its clinical impact.

Acknowledgments

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/nu15204452/s1, Supplementary S1: Study exit questionnaire. Supplementary S2: Godin Leisure-Time Exercise Questionnaire. Supplementary S3: Participant Flowchart.

Funding Statement

The research reported in this publication was supported by the National Center for Advancing Translational Science of the National Institutes of Health under award UL1TR002384, and the Rudin Foundation.

Author Contributions

Conceptualization, A.P.S., L.J.A. and K.C.H.; methodology, A.P.S., L.J.A. and K.C.H.; formal analysis, D.D.; investigation, A.K., M.G., I.S., B.A., A.G., J.T., L.H., M.O. and A.J.C.; data curation, A.P.S., A.K., K.C.H., L.J.A., A.G., M.G., I.S., B.A. and M.O.; writing—original draft preparation, A.P.S., A.K. and K.C.H.; writing—review and editing, all authors; visualization, A.P.S.; supervision, A.P.S.; project administration, A.K. and A.J.C.; funding acquisition, A.P.S. and L.J.A. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Weill Cornell Medicine (IRB #1807019463, approval date: 10 January 2019).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The dataset pertaining to the current study is available from the corresponding author in accordance with the appropriate data use agreements and Institutional Review Board approvals for secondary analyses.

Conflicts of Interest

Louis J. Aronne has stock in Intellihealth, Inc. (equity due to interest). Katie Hootman has received consulting fees from Faeth Therapeutics, Inc. within the past 36 months. She has also received consulting/program reviewer fees from PESI, Inc. Hootman has also received honoraria for lectures at Columbia University.

Footnotes

References