Efficacy of Time-Restricted Eating and Behavioral Economic Intervention in Reducing Fasting Plasma Glucose, HbA1c, and Cardiometabolic Risk Factors in Patients with Impaired Fasting Glucose: A Randomized Controlled Trial

This parallel RCT was conducted at the outpatient clinic of the Department of Family Medicine, Faculty of Medicine, Ramathibodi Hospital, from October 2021 to February 2023. This research complies with the Consolidated Standards of Reporting Trials (CONSORT) statement. The trial protocol was registered at the Thai Clinical Trials Registry (TCTR20210520002) and was published elsewhere [24]. The protocol was approved by the Ethics Committee of Ramathibodi Hospital, Mahidol University (MURA 2021/389).

This study included patients and staff of Ramathibodi Hospital using the following eligibility criteria: adults aged 18 to 65 years, diagnosed with IFG (i.e., FPG of 100–125 mg/dL and HbA1c less than 6.5%) [25], having BMI ≥ 25 kg/m², and willing to provide informed consent. Patients were excluded if they met any of the following criteria: (1) followed a ketogenic or vegetarian diet, (2) worked night shift for a minimum of 3 h between 10:00 PM and 5:00 AM on more than one day per week, (3) experienced body weight changes exceeding 5 kg in the three months prior to study enrolment, (4) were in receipt of medication to be consumed with food either before 8:00 AM or after 5:00 PM, (5) were pregnant or breastfeeding, (6) had psychiatric disorders, such as eating or mood disorders (except depression), (7) were taking corticosteroid or anti-diabetic medications, (8) had a history of bariatric surgery, or (9) had impaired nutrient absorption.

Patients were randomly assigned to (1) TRE plus BE, (2) TRE alone, or (3) usual care in a ratio of 1:1:1. A random sequence was generated using stratified-block randomization with varying block sizes (6 and 9) and age stratification (i.e., 18–59 years and 60–65 years).

A random sequence list was concealed in sequential opaque sealed envelopes.

This study focused on two specific interventions: TRE and BE-informed interventions. In the TRE groups, patients were instructed to restrict their daily food intake to a 9 h window (between 8:00 AM and 5:00 PM), without any limitation on the types of food and beverages consumed. This approach involved a fasting period of 15 h per day.

The BE interventions included financial incentives and text reminders aimed at increasing adherence to TRE. Participants were eligible to receive a monthly monetary payment of 1000 THB (approximately 85 USD) if they self-reported adherence to the TRE for at least 5 days a week over a 4-week period. In addition, text reminders were sent to participants every 2 days to reinforce their commitment to achieving the TRE goal of 5 days per week, track their personal performance, and remind them of the TRE interval. Participants in all three groups were also requested to maintain a logbook to record their first and last mealtime each day.

The primary outcomes were FPG and HbA1c levels measured at 12 weeks after randomization. Secondary outcomes included body weight, SBP, DBP, fasting insulin, TG, total cholesterol, LDL-C, HDL-C, and hs-CRP. The homeostatic model assessment of insulin resistance (HOMA-IR) was calculated by multiplying the fasting insulin (mIU/L) with FPG (mg/dL) and dividing by 405 to measure insulin sensitivity in the participants. FPG was measured using hexokinase glucose-6 phosphate dehydrogenase. HbA1c levels were measured using the turbid metric inhibition immunoassay certified by the National Glycohemoglobin Standardization Program. Serum triglyceride was measured by lipase/glycerol kinase glycerol-3-phosphate oxidase. HDL-C and LDL-C levels were measured by the accelerator selective detergent method. Chemiluminescence and nephelometry methods were used for measuring fasting insulin and hs-CRP, respectively. Body weight was measured without shoes to the nearest 100 g using a Seca 284 Wireless Measuring Station—Weight & Height 2841300109. Blood pressure was measured by the Connex Spot Monitor; Welch Allyn, Inc. (Auburn, NY, USA) with an automatic blood pressure cuff after resting for at least 15 min. Body weight and blood pressure were measured by trained research assistants.

All primary and secondary outcomes were assessed at baseline and weeks 4, 8, and 12 after randomization. Possible adverse events of TRE, such as syncope, dizziness, and light headedness, were monitored and recorded throughout the study period.

At the first visit, potential participants were screened for the eligibility criteria (e.g., FPG, HbA1c, age, and BMI) by investigators and research assistants.

One week after enrolment (2nd visit), demographic information (e.g., age, sex, educational level, and marital status), underlying diseases (e.g., hypertension, dyslipidemia, chronic kidney disease, fatty liver disease, and history of gestational diabetes mellitus), and health risk behaviors (i.e., smoking and alcohol consumption) were collected by an interview with trained research assistants. Blood pressure, body weight, and height were measured, and participants were randomly allocated to one of three groups: TRE plus BE, TRE alone, and usual care. Physical examination including blood pressure and body weight, and laboratory measurements (i.e., FPG, HbA1c, TG, total cholesterol, LDL-C, HDL-C, fasting insulin, and hs-CRP) were obtained during the 3rd visit (4th week), 4th visit (8th week), and 5th visit (12th week) after randomization or study end. Dietary intake at baseline and weeks 4, 8, and 12 post randomization were collected using 24 h food recall. Participants were asked to record their food diary for 7 days at each visit. INMUCAL-nutrients version 4.0 (developed by the Institute of Nutrition, Mahidol University, Thailand; https://inmu2.mahidol.ac.th/inmucal/index.php↗, accessed on 24 May 2023) was used for calculating nutrient intake from dietary data.

A superiority trial approach utilizing a one-way analysis of variance method was applied for estimating the sample size. The baseline mean and standard deviation (SD) of FPG in the prediabetes cohort were 105 and 9 mg/dL, respectively [26]. We anticipated that receiving TRE plus BE, and TRE alone would lead to a reduction in the FPG of approximately 7% and 5%, respectively, compared to the control group. With type I error and power set to 0.05 and 0.8, respectively, a total of 90 participants, with 30 participants per group, were required to detect these differences. However, given that the recruitment was adversely affected since it fell within COVID-19 restrictions, the final sample size was limited to 72 participants.

Mean (SD) or median (interquartile range, IQR) and frequency and percentages were used to describe continuous data and categorical data, respectively. Means of primary and secondary outcomes among the three groups were compared using a mixed-effect linear regression model by regressing an outcome of intervention and time, considering patients as a random effect (i.e., repeated measure at 4, 8, and 12 weeks after randomization) and the intervention arms (TRE plus BE and TRE alone versus usual care) as a fixed effect. Marginal means and differences between any pair of the three interventions were estimated accordingly. If baseline characteristics differed between the 3 groups, these were adjusted using multivariate regression analysis.

All analyses were performed using three approaches. First, an intention-to-treat analysis (ITT) was applied by considering interventions that were initially assigned regardless of compliance. Second, a per-protocol analysis (PPA) excluded patients who did not comply with TRE (i.e., complied < 5 days per week) in the TRE plus BE and TRE alone and patients in the usual care group who took TRE 5 days or more per week from the analysis. Third, an “as-treated” analysis kept participants in either TRE plus BE or TRE alone if they complied with TRE ≥ 5 days/week; otherwise, they were considered in the usual care group. Likewise, participants in a usual care group were moved to the TRE group if they adhered to TRE ≥ 5 days/week. If a participant missed an appointment in week 4, 8 or 12, the previous outcome measure was carried forward to impute the missing outcome to minimize the lost to follow-up.

All analyses were performed using STATA 18.0. A p-value < 0.05 was considered statistically significant.

None of the patients in TRE plus BE, TRE alone, and usual care groups reported adverse events.

The effectiveness of TRE in improving blood glucose levels, weight reduction, and lipid profiles in patients with IFG has been investigated previously. A recent RCT conducted on IFG patients found that TRE with a 16 h fasting period for three weeks led to a significant decrease in FPG levels at the three-month follow-up [27]. However, the findings from multiple studies in patients with obesity/overweight [28,29,30], non-alcoholic fatty liver disease [31], or prediabetes [13], did not show significant benefits of TRE in lowering FPG levels when compared to the designated control group. Our results are consistent with these findings, as both TRE plus BE and TRE alone did not significantly lower FPG and HbA1c levels when compared to usual care. The lack of significant results might be attributed to low adherence to the TRE regimen among our participants, as well as our low sample size.

Our hypothesis was confirmed when analyzing the data considering adherence (as-treated analysis); TRE showed significant improvements in FPG, HbA1c, fasting insulin, and HOMA-IR levels compared to usual care. Similar results were observed previously in a crossover RCT where prediabetes patients who adhered to TRE for five weeks experienced a significant reduction in fasting insulin levels [12]. Findings from both our study and that in prediabetes suggest that adhering to TRE may enhance insulin sensitivity and β-cell responsiveness in prediabetes patients, potentially reducing the risk of progression to DM for these individuals.

Body weight changes in our study were not significantly different among the three groups. However, by this study’s end, both the TRE plus BE and TRE alone groups showed a significant decrease in body weight compared to their baseline measurements. Our results support previous findings from individuals with prediabetes, which showed no significant difference in body weight reduction between the TRE and control groups. However, in both groups, there was a notable loss of body weight compared to their respective baseline measurements [12]. Similar positive outcomes related to body weight were observed in various RCTs involving patients with obesity or overweight conditions. These studies reported a significant decrease in body weight when participants followed a TRE approach compared to their baseline measurements [28,32,33]. The beneficial effect of TRE on weight loss could potentially be attributed to the unintentional reduction in energy intake during prolonged fasting periods [34]. Moreover, extending the fasting window increases the utilization of stored fat, leading to higher levels of ketone bodies, which in turn can suppress appetite and result in reduced energy intake [35].

Weight reduction can reduce future diabetic risk in individuals with prediabetes and obesity. Our study observed a significant reduction in body weight from the baseline in participants who adhered to the TRE protocol.

Consequently, it is essential to consider that the observed benefits of TRE in reducing FPG and HbA1c, as identified from the “as-treated” analysis, might be attributed to mechanisms other than weight loss. While weight reduction is well-accepted for its preventive role in type 2 diabetes onset and progression, it appears that TRE may exert its positive effects on FPG and HbA1c through different pathways, independent of weight reduction. Glucose metabolism is intricately regulated by the circadian clock. Aligning the eating period with the body’s natural circadian rhythms, such as consuming meals earlier in the day and fasting for the remaining time, can have a significant impact on metabolic processes. This eating pattern has been shown to restore cyclic AMP response element-binding protein phosphorylation, which, in turn, leads to a reduction in gluconeogenesis and contributes to improved glucose homeostasis [34].

Our study observed significantly lower SBP in the TRE plus BE group, but no significant change was found in the TRE alone group. An RCT conducted on men with prediabetes also reported significant decreases in both SBP and DBP compared to the control group [13]. However, findings from other RCTs involving diverse population groups showed inconsistent effects of TRE on blood pressure levels [33,36,37,38]. A systematic review analyzing the effects of TRE with different feeding windows suggested that the benefits of TRE might be more pronounced in early TRE, where food intake is restricted to the early period of the day [39], likely due to better alignment with circadian rhythms.

The potential benefits of TRE in suppressing inflammation, identified from animal studies [9,10,11], have yet to be fully recognized in humans. In our study, we observed a significant decrease in hs-CRP levels in the TRE alone group, but not in the TRE plus BE group. This contradicts previous findings [13,40], which reported that TRE did not significantly affect hs-CRP levels. However, the beneficial effects of TRE in improving hs-CRP levels in our study were not observed in the “as-treated” analysis. This raises the possibility that our observed results may have occurred by chance, emphasizing the need for further studies to confirm the potential impact of TRE on the inflammatory process.

Despite the proposed seemingly easier adoption of the TRE approach compared to other diet-control techniques, adherence to TRE remains challenging as observed by the low adherence rate in the TRE alone group (9%). Applying behavioral-economic informed interventions (i.e., financial incentives and text reminders in this study) could motivate some participants to higher adherence, as observed in the TRE plus BE group (46%) compared to TRE alone (9%). This may be due to the short-term financial incentives that complement future health benefits [17], while reminders make TRE salient to participants every two days [41]. However, we could not disentangle the effects of financial incentives and reminders from each other. A recent feasibility RCT comparing text reminders and text reminders plus financial incentives (in the form of a deposit contract) to increase adherence to TRE in patients with obesity and hypertension showed no difference between both groups [42], although this study was underpowered. Future studies focusing on the effect of each of these behavioral-economic informed interventions might be of interest for clinical and cost-effective intervention design.

Given the limited evidence of the efficacy of TRE in individuals with IFG, our study contributes valuable knowledge to this field. The findings from our RCT can inform the recommendation of appropriate dietary interventions to reduce future diabetic risk. Furthermore, our study assessed the efficacy of BE-informed interventions to promote intervention adherence. This evidence is valuable for encouraging adherence not only to dietary but other interventional approaches.

However, our study had several limitations. First, ongoing COVID-19 pandemic restrictions during the recruitment period resulted in a smaller sample size than anticipated which reduced the statistical power of our findings. Second, the participant adherence rate to the TRE protocol in our study was quite low. Consequently, the non-significant effectiveness of TRE in our study should be interpreted with caution given the poor intervention compliance. This is supported by the results from the “as-treated” analysis that point to potentially significant benefits of TRE for improving blood sugar and fasting insulin levels. Third, our interventions could not be blinded; hence, the findings may be subject to observer and information bias. However, the majority of our outcomes were objectively measured, and the assessors were blinded to allocation; thus, measurement error or ascertainment bias was minimized. In addition, our study did not measure oral glucose tolerant test. As a result, we were unable to determine the status of impaired glucose tolerance among the participants in the study. Lastly, only surrogate outcomes such as FPG, HbA1c, body weight, and other laboratory tests were assessed; thus, further RCTs that measure long-term or clinically important outcomes, such as DM incidence, are needed.

Data described in the manuscript, code book, and analytic code will be made available upon request pending.