The Impact of Exercise and Cumulative Physical Activity on Energy Intake and Diet Quality in Adults Enrolled in The Midwest Exercise Trial for The Prevention of Weight Regain

Abstract

The purpose of this study was to assess impact of different volumes of exercise as well as cumulative moderate to vigorous PA (MVPA), on energy intake and diet quality, as assessed by the Healthy Eating Index 2010 (HEI-2010), across a 12-month weight maintenance intervention. Participants were asked to attend group behavioral sessions, eat a diet designed for weight maintenance, and exercise either 150, 225, or 300 mins/week. Dietary intake was assessed by 3-day food records, and MVPA was assessed by accelerometry. Two hundred and twenty-four participants (42.5 years of age, 82% female) provided valid dietary data for at least one time point. There was no evidence of group differences in energy intake, total HEI-2010 score, or any of the HEI-2010 component scores (all p > 0.05). After adjusting for age, sex, time, group, and group-by-time interactions there was an effect of cumulative MVPA on energy intake (1.08, p = 0.04), total HEI-2010 scores (−0.02, p = 0.003), sodium (−0.006, p = 0.002) and empty calorie scores (−0.007, p = 0.004. There was evidence of a small relationship between cumulative daily EI and weight (β: 0.00187 95% CI 0.001, p= 0.003). However, there was no evidence for a relationship between HEI total score (β: −0.006, 95% CI −0.07-0.06) or component scores (all p > 0.05) and change in weight across time. The results of this study suggests that increased cumulative MVPA is associated with small, clinically insignificant, increases in energy intake and decreases in HEI scores.

INTRODUCTION

Obesity is a global health problem that affects individuals of every age, sex, ethnicity, race, and socioeconomic status, with 39.6% of adults in the US having obesity ¹. Numerous combinations of energy restriction and increased exercise have shown moderate short-term success in producing clinically significant (≥5%)² reductions in body weight ^3-6. However, regardless of the weight loss program employed, preventing weight regain is difficult. In general, 50% of individuals who initially lose weight will regain more than 45-75% of the weight lost within 12 to 30 months from the end of treatment ^4,7-10. A 2015 NIH working group ¹¹ identified two primary reasons to explain the high rates of weight regain. First, that weight loss achieved by dieting induces a number of physiological and behavioral adaptations that collectively result in an elevation of appetite ^12,13, and as well as enhance the incentive salience (rewarding value) of food ^14,15 resulting in greater preference for more palatable foods leading to food seeking and ingestive behaviors. Thus, weight loss may lead to increases in energy intake and decreases the quality of an individual’s diet post weight loss. Secondly, the behavioral strategies used to induce weight loss are commonly employed in a transient manner and adherence generally wanes as weight loss plateaus.

Exercise has been suggested to play an important role in the prevention of weight regain ¹⁶. However, the clinical effectiveness of physical activity as a weight loss maintenance strategy, in the form of regimented exercise or increased moderate to vigorous physical activity (MVPA), has yet to be demonstrated. For example, recent work by our group demonstrated that higher levels of prescribed exercise did not prevent weight-regain ¹⁷. Increasing energy expenditure through increased physical activity can result in behaviors (performed consciously or unconsciously), such as changes in dietary intake (increased energy intake, decreased diet quality), that minimize the exercise-induced negative energy balance or, in some cases, even result in a positive energy balance (i.e weight gain) ^18-23.This topic is highly debated, and we are unaware of previous studies which have examined the role of exercise or MVPA on dietary intake in individuals attempting to maintain their weight loss.

The Midwest Exercise Trial for the Prevention of Weight Regain (MET-POWeR) was designed to evaluate the effectiveness of 3 levels of exercise (150, 225, or 300 min/week) on the prevention of weight regain over 12 months following clinically meaningful weight loss (≥ 5%) in a sample of adult men and women with overweight and obesity. The previously published results for the primary outcome, weight change over 12 months, indicated no significant difference between randomized group (150 min/week (1.1 ± 6.5 kg), 225 min/week (3.2 ± 5.7 kg) or 300 mins/week (2.8± 6.9 kg) groups (p=0.21)¹⁷. However, it should be noted that in all three groups, average weekly exercise minutes fell short of the recommendations, with participants randomized to the 150, 225 and 300 min/week groups only completing 129, 153 and 179 mins/week, respectively. Data collected across the 12-month intervention on self-reported dietary intake (3-day food records) provided an opportunity to assess the impact of different volumes of exercise on change in energy intake and diet quality, as assessed by the Healthy Eating Index 2010 (HEI-2010), during a weight maintenance intervention. Additionally, accelerometer data collected across the 12-month intervention provided to opportunity to assess the impact of cumulative MVPA, irrespective of group randomization on energy intake and diet quality.

METHODS

Overview

A detailed description of the design and methods for MET-POWeR have been previously published ²⁴. Briefly, 298 men and women initiated a 3-month behavioral weight loss intervention. The 235 participants who met the 5% weight loss goal were randomized by gender and level of weight loss (i.e., 5 - 9.9%, 10 - 14.9%, ≥ 15%) in a 1:1:1 allocation to 1 of 3 exercise groups (i.e., 150, 225, or 300 minutes/week of exercise) for a 12-month weight maintenance intervention. Participants were also asked to attend group behavioral sessions and follow dietary recommendations for weight maintenance. Approval for this study was obtained from the Human Subjects Committee at the University of Kansas-Lawrence. All participants provided written informed consent and were compensated monetarily for time and travel associated with outcome evaluations. For the present investigation, we have opted to explore the effect of different volumes of exercise and cumulative MVPA, irrespective of group randomization, on energy intake and diet quality, as assessed by the HEI-2010, during the 12 month weight maintenance intervention only.

Participants

Participants were adults (BMI 25-44.9 kg/m², age 21-55 years) who were able to exercise and willing to be randomized to one of 3 exercise groups. Clearance from their primary care physician was required. Exclusion criteria included: participation in a research project involving weight loss or exercise in the previous 6 months; currently participating in a regular exercise program, i.e., > 500 kcal/week. of planned activity assessed by questionnaire ²⁵ not weight stable (±4.5 kg) for 3 months prior to intake; pregnant during the previous 6 months, lactating, or planned pregnancy in the following 15 months; serious medical risk such as type 1 diabetes, cancer or recent cardiac event, i.e., heart attack, angioplasty, etc.; eating disorder ²⁶; current treatment for psychological issues or taking psychotropic medications; taking medications known to affect weight; adherence to specialized diets; no access to grocery shopping and meal preparation, i.e., military, college students with cafeteria plan, etc.

Diet Recommendations

Energy intake was prescribed as resting metabolic rate estimated using the equation of Mifflin-St. Jeor multiplied by 1.2 to account for activities of daily living (30). Daily meal plans including suggested servings of grains, proteins, fruits, vegetables, dairy, and fats, based on participants’ energy requirements and in compliance with the USDA/HHS Dietary Guidelines for Americans were provided to all participants (31). Participants were recommended but not required to consume of 2 commercially available portion-controlled entrées (180 to 270 kcals each) or low-calorie shakes, and 5 servings of fruits and vegetables per day. Participants were asked to purchase portion-controlled entrées and shakes with acceptable energy and macronutrient content available at local supermarkets from a list provided by the trial.

Exercise Recommendations

Participants were randomized to prescribed exercise groups of 150, 225 or 300 min/week. Exercise progressed from 100 min/week at 70% heart rate max (HR_max) at the end of weight loss to the prescribed goals of 150, 225 or 300 min/week at month 2 and remained at goal for the duration of the 12-month intervention. Exercise intensity was prescribed at 70% of age-predicted maximal heart rate (HR_max = 220 – age in years). Participants were asked to exercise 5 days/week, with 3 days under supervision by research staff at one of our dedicated exercise facilities which were open Monday thru Friday, 6-10 AM and 4-8 PM to accommodate participants schedules. Treadmill walking was the primary mode of supervised exercise; however,1 session/week of supervised exercise using an alternative mode (elliptical, bike, etc.) was permitted to provide variety and prevent overuse injuries. The duration and intensity of all exercise sessions (supervised +unsupervised) were verified by heart rate (HR) monitors (RS 400; Polar Electro, Woodbury, NY). Each exercise session was preceded by a brief (5 minute) warm up following which participants were instructed to increase exercise intensity to reach their prescribed target HR prior to starting the HR monitor and to remain at the target intensity for the duration of each exercise session. Trained research staff monitored participants exercise HR to ensure that it remained at target during all supervised exercise sessions. A valid exercise session (supervised or unsupervised) was defined as a HR ± 4 beats/ min of target HR for the prescribed exercise duration. Prior to each supervised exercise session (minimum of 3 sessions/week) research staff downloaded HR data from all unsupervised exercise sessions completed since the last exercise laboratory visit using the Polar HR Software. Research staff reviewed intensity and duration of the unsupervised exercise sessions. Participants not completing a valid exercise session were counseled to meet both the intensity and duration prescriptions. Participants who failed to meet the exercise recommendations were not dismissed from the trial.

Behavioral sessions

Participants attended 60-minute group sessions based on Social Cognitive theory to promote adherence to diet and exercise recommendations (28). Behavioral sessions were held weekly and in person during the first 3 months and twice per month over group phone call during the final 9 months of the maintenance intervention. Health educators delivered interactive 30-45-minute lessons on topics including topics such as meal planning, environmental control, eating on the go, and behavior change strategies for maintaining motivation for diet and exercise with the remaining time devoted to discussion and problem solving. Participants were asked to complete homework assignments designed to increase self-efficacy for both diet and exercise and to provide practice of behavioral skills and to selfmonitor daily diet (study designed web form) and exercise (paper and pencil log) and weekly body weight.

Outcomes

All outcome assessments were completed by trained research staff in the Center for Physical Activity and Weight Management, Energy Balance Laboratories at either The University of Kansas- Lawrence or The University of Kansas Medical Center, Kansas City, Kansas. The assessment schedule was as follows: anthropometrics (weight/height) 0, 3, 6, 9, and 12 months; energy intake 0,3,6,9, and 12 months; and total daily MVPA 0, 6 and 12 months. Minutes of prescribed exercise were assessed weekly across the 12-month intervention. Prior to baseline testing participants completed a health history and demographic survey, i.e., age, sex, household income, and education.

Anthropometrics (Weight/height).

Weight was obtained using a digital scale accurate to ± 0.1 kg (Befour Inc. Model #PS6600, Saukville, WI) between the hours of 6 and 10 AM, after an overnight fast. Participants were weighed prior to breakfast and after attempting to void wearing a standard hospital gown. Height was measured using a stadiometer (Model PE-WM-60-84, Perspective Enterprises, Portage, MI). BMI was calculated as weight (kg)/height (m²)

Dietary Intake.

Self-report 3-day food records (2-week days/1 weekend day) were used to assess energy intake and diet quality. Subjects were instructed on how to complete the records individually at their baseline testing appointment. Food record templates prompted for specific details regarding foods consumed and included location, time of meal/snack, generic food description, food description details, and portion size. Subjects were provided an instruction sheet noting the level of detail desired and specific instructions related to food type and quantity and instructed to record this information immediately after each meal for all three days. Participants were asked to provide weight or household measurements (e.g. cups, tsp, etc.) when available; a measurement book was provided to increase the accuracy in portion size measurements and a food mapping grid was provided to ensure accurate descriptions on food descriptions. Upon receipt of the records, data were reviewed with participants by a registered dietitian to ensure completeness of records and to gather additional information on food details or amounts. Staff reviewing records did not have regular contact with participants during the intervention to reduce the likelihood of participants altering diet information to comply with the guidelines of their diet prescriptions. The Nutrition Data System for Research (NDS-R, version 2014, University of Minnesota, Minneapolis, MN) was used for determining nutrient content. Data were entered by research staff and checked by a registered dietitian with at least one year of experience using NDS-R. Diet outcome data were reviewed for plausibility and accuracy by the study team prior to analysis (e.g records were excluded if they did not represent a typical day (holiday) or if the average daily energy intake across the 3 days was < 500 kcals/day). Data obtained from the food records were used to calculate diet quality index scores using the HEI-2010. The HEI-2010 was developed by the United States Department of Agriculture to assess conformance to the 2010 Dietary Guidelines for Americans²⁷. The HEI-2010 assesses the intake of total fruit, whole fruit, total vegetables, dark green vegetables and beans, refined grains, whole grains, milk, meat and beans, seafood and plant protein, fatty acid ratio, sodium, and empty calories, and provides a point value based on how well a person meets the dietary guidelines (expressed as a percent per 1000 kcals)²⁸. Point values for each category are summed to obtain the total HEI-2010, with 100 points representing the maximum score. The HEI-2010 was chosen over the more recent HEI-2015, as the diet recommendations for the study intervention were based on the 2010 dietary guidelines.

Physical activity.

Participants were asked to wear a portable accelerometer (Actigraph GT3X, ArchiMed Inc. Lyon, France) on a belt over the non-dominant hip for 7 consecutive days. Accelerometer data was collected in 1-min epochs with a minimum of 8 hours constituting a valid monitored day. Accelerometer data was processed using the protocol for adults used in the 2003-2004 and 2005-2006 cycles of NHANES ^29,30. The following intensity cut-points were used: sedentary (< 1.0 MET; ≤100 counts/min), light (1.1-2.99 METs; 101-2019 counts/min.), moderate (3.0-5.99 METs; 2020-5988 counts/min) and vigorous ≥ 6 METs; ≥ 5999 counts/min) ^29,30. Non-wear time was defined as at least 60 consecutive minutes of zero counts, with allowance for 1-2 min. of counts between 0 and 100. Participants with less than 4 valid days were excluded from the accelerometer data analysis, as 4 days represents the minimum number of days to reliably estimate habitual physical activity ³¹.

Statistical Analysis

Baseline measures and demographic characteristics were summarized using means and standard deviations for continuous variables, and frequencies and percentages for categorical variables. Linear Mixed Models (LMMs) were used to examine difference between randomized groups (group effects) changes over time (time effects) and group-by-time interactions for energy intake and HEI. The models were adjusted for age and sex and included a random intercept of each subject. All LMMs were fit with an autoregressive-1 correlation structure was degrees of freedom were estimated using the Satterthwaite approximation. Additionally, LMMs were used to examine the effect of total daily MVPA (accelerometer) on HEI-2010 total score and sub scores irrespective of randomized group. MVPA was measured at every time point and is thus a time-varying covariate, to avoid using MVPA data from future observations to predict diet quality we calculated the cumulative MVPA for each participant (MVPA at current time point plus MVPA at all previous time points) and included it as a covariate in the model. As a sensitivity analysis, most recent MVPA proceedings measurement of diet quality was used in the models instead of cumulative MVPA. This formulation assumes that any relationship between diet quality and MVPA is only dependent on the most recent exercise data (e.g. a cross-sectional relationship independent of previous MVPA). Linear mixed models were also used to assess the effect of cumulative and cross-sectional energy intake and cross-sectional HEI-2020 total and sub scores on weight loss. All analyses were conducted using R 3.6.3, mixed models were fit using the nlme package^32,33.

RESULTS

Participants

Two hundred ninety-eight participants (53 men, 245 women) initiated the 3-month weight loss intervention. During weight loss 46 participants volitionally left the trial (9 men, 37 women) and 17 participants (1 man, 16 women) did not achieve the 5% weight loss criteria required for randomization. Thus, 235 participants were randomized to one of the 3 exercise groups: 150 min/week, n = 76, 225 min/week, n= 80, 300 min/week, n = 79. This analysis is based on 224 participants (150 min/wk. n= 72, 225 min/wk. n= 77, 300 min/wk. n= 75) who had valid dietary data for at least one time point. Demographic characteristics of the participants at the start of the weight maintenance intervention are presented in Table 1.

Table 1:

Demographic characteristics of adults who had previously lost ≥5% weight and were randomized to either 150, 225, or 300 mins/wk of exercise during a 12-month weight maintenance intervention.

Variable	Combined (n = 224)	150 min/week (n = 72)	225 min/week (n = 77)	300 min/week (n = 75)
Age (yrs.)*	42.5 ± 8.4	42.3 ± 8.2	43.0 ± 8.1	42.1± 9.0
Gender - women (%)	184 (82.1 %)	58 (80.6%)	65 (84.4%)	61 (81.3%)
Weight (kg)*	89.2 ± 15.8	88.9 ± 15.6	89.5 ± 17.2	89.2 ±14.8
BMI (kg/m2)*	32.0 ± 4.7	32.2 ± 4.7	32.0 ± 5.0	32.0 ± 4.5
Waist circumference (cm)*	92.9 ± 11.2	93.5 ± 11.1	92.9 ± 11.2	92.4 ±11.3
Minority (%)**	67 (29.9%)	18 (25%)	28 (36.4%)	21 (28%)
Education (Yrs.)*	15.3 ± 3.7	15.4 ± 4.1	15.1± 3.6	15.3 ± 3.6
Annual Household Income (%)
< = 39k	34 (15.2%)	13 (18.1%)	12 (15.6%)	9 (12.0%)
>39 - <=79k	85 (37.9%)	25 (34.7%)	27 (35.1%)	33 (44.0%)
>79k	87 (38.8%)	30 (41.7%)	35 (45.5%)	22 (29.3%)
No response	18 (8.0%)	4 (5.6%)	3 (3.9%)	11 (14.7%)
Energy Intake (kcals)	1321 ± 293	1340 ± 307	1283 ± 252	1340 ± 318
HEI-2010 Total Score	58.8 ± 7.6	58.4 ± 7.3	58.9 ± 8.6	59.2 ± 7.0

^*Mean (Standard Deviation)

^**Select a race other than non-Hispanic white.

Exercise completed.

Participants randomized to the 150, 225 and 300 min/week groups completed 129 ± 30, 153 ± 49 and 179 ± 62 min/week, respectively. Weekly exercise was higher in the 300 min/week group compared with 150 and 225 min/week groups, and in the 225 min/week compared with the 150 min/week group (all p <0.0001). Weekly exercise minutes did not differ by sex (p = 0.63) and there was no evidence for a group by sex interaction (p = 0.67). The number of completed supervised (p = 0.16) or unsupervised exercise sessions (p = 0.054) did not differ by group. Approximately 60% of weekly exercise minutes were completed under supervision. The average exercise intensity of all exercise sessions (supervised + unsupervised) was 76%, 75% and 76% in the 150, 225, and 300 min/week groups, respectively. Exercise session attendance across the 12-month intervention is presented in Figure 1.

Figure 1.

Mean exercise session attendance in adults randomized to either 150, 225, or 300 mins/wk of exercise across a 12-month weight maintenance period.

Behavioral sessions completed.

Participants attended −67% of the 30 scheduled behavioral sessions across 12 months: 150 min/week =62%, 225 min/week =60%, 300 min/week=62%. Behavioral session attendance did not differ by group (225 vs 150: p=0.84; 300 vs 150: p= 0.26) and there was no impact of session attendance on energy intake (p=0.87) or total HEI-20210 score (p = 0.75) at 12 months. Behavioral session attendance across the 12-month intervention is presented in Figure 2.

Figure 2.

Mean behavioral session attendance in adults randomized to either 150, 225, or 300 mins/wk of exercise across a 12-month weight maintenance period.

Energy intake.

Self-reported energy intake (kcals/day) over the 12-month intervention is presented in Figures 3-4 and Table 2. Mixed modeling revealed that after adjusting for age and sex there was a time effect with energy intake increasing across time (β; 25.6; 95% CI 16.3-34.9; p < 0.0001). However, there was no evidence for group differences (225 mins/week vs 150 mins/week: −20.1; 95% −160.6-120.5 p = 0.77, 300 min/week vs 150 min/week: 25.3; 95% −116.6-167.1; p = 0.73) and only evidence for a group-by-time interaction in the 225 min/week group relative to 150 min/week group (225 vs 150: −12.9; 95% −25.7-0.22 ; p = 0.04, 300 min/week vs 150 min/week: −3.9; 95% −16.8-9.0; p = 0.55) in self-reported energy intake. However, when we combined all groups and examined the impact of cumulative MVPA assessed by accelerometer on changes in energy intake across time mixed modeling revealed that after adjusting for age, sex, and group there was evidence for a time effect (20.4, p =0.0002). While the estimate for the effect of cumulative MVPA on energy intake was positive, it was relatively small (1.08; 95% CI 0.07-2.1, p = 0.04).

Figure 3.

Changes in (a) weight (kg), (b) self-report energy intake (kcals/day), (c) HEI-2010 total score, and (d) MVPA (mins/day) in adults randomized to either 150, 225, or 300 mins/wk of exercise across a 12-month weight maintenance period.

Figure 4.

Individual changes in in self-report energy intake (kcals/day) from 0 to 12 months, in adults randomized to either 150, 225, or 300 mins/wk of exercise across a 12-month weight maintenance period.

Table 2.

Energy intake and HEI-2010 scores in adults randomized to either 150, 225, or 300 mins/wk of exercise across a 12-month weight maintenance period.

	Total sample (n = 224)	150 min/wk (n = 72)	225 min/wk (n = 77)	300 min/wk (n = 75)	Group Effect	Group-by- Time Effect	Cumulative MVPA
Energy Intake					−57.6 (58.1)	−12.3 (6.5)	1.08 (0.51)
(kcals/day)					16.4 (58.5)	−3.2 (6.6)
0 mos.	1321 ± 293	1340 ± 307	1283 ± 252	1340 ± 318
6 mos.	1637 ± 494	1665 ± 513	1527 ± 449	1731 ± 510
12 mos.	1568 ± 470	1639 ± 479	1449 ± 414	1631 ± 501
Total HEI-Score					0.75 (1.08)	−0.12 (0.16)	−0.02 (0.01)
(Potential Max Score 100)					0.59 (1.10)	−0.07 (0.16)
0 mos.	58.8 ± 7.6	58.4 ± 7.3	58.9 ± 8.6	59.2 ± 7.0
6 mos.	59.2 ± 8.4	58.0 ± 7.4	59.9 ± 8.3	59.5 ± 9.2
12 mos.	59.1 ± 7.5	60.0 ± 7.9	58.4 ±6.3	59.2 ± 8.4
Total Fruit					0.002 (0.16)	−0.001 (0.02)	−0.0012
(Potential Max Score 5)					−0.11 (0.16)	0.01 (0.02)	(0.0014)
0 mos.	2.81 ± 1.20	2.75 ± 1.10	2.87 ± 1.28	2.81 ± 1.24
6 mos.	2.91 ± 1.19	2.97 ± 1.14	3.02 ± 1.23	2.74 ± 1.19
12 mos.	2.84 ± 1.15	2.85 ± 1.20	2.78 ± 1.06	2.91 ± 1.23
Whole Fruit					−0.001 (0.17)	−0.001 (0.02)	−0.001
(Potential Max Score 5)					−0.03 (0.17)	0.0001 (0.02)	(0.0014)
0 mos.	3.39 ± 1.24	3.26 ± 1.15	3.47 ± 1.32	3.42 ± 1.23
6 mos.	3.53 ± 1.21	3.59 ± 1.21	3.58 ± 1.23	3.43 ± 1.18
12 mos.	3.41 ± 1.15	3.47 ± 1.18	3.37 ± 1.01	3.40 ± 1.27
Total Vegetables					−0.07 (0.13)	0.02 (0.02)	0.0007
(Potential Max Score 5)					−0.17 (0.13)	0.03 (0.02)	(0.001)
0 mos.	3.80 ± 0.96	3.94 ± 0.96	3.81 ± 0.93	3.65 ± 0.98
6 mos.	3.63 ± 1.01	3.61 ± 1.03	3.59 ± 1.01	3.70 ± 1.00
12 mos.	3.90 ± 0.80	3.81 ± 0.80	4.01 ± 0.70	3.88 ± 0.91
Greens and Beans					−0.13 (0.18)	0.016 (0.03)	−0.00005
(Potential Max Score 5)					−0.17 (0.18)	0.01 (0.03)	(0.002)
0 mos.	2.48 ± 1.40	2.56 ± 1.35	2.46 ± 1.50	2.43 ± 1.36
6 mos.	2.43 ± 1.34	2.34 ± 1.25	2.58 ± 1.40	2.36 ± 1.37
12 mos.	2.54 ± 1.36	2.47 ± 1.30	2.67 ± 1.26	2.47 ± 1.51
Refined Grains					0.24 (0.27)	−0.05 (0.04)	−0.003
(Potential Max Score 10)					−0.05 (0.27)	0.001 (0.04)	(0.002)
0 mos.	7.38 ± 1.94	7.4 ± 1.98	7.5 ± 2.05	7.23 ± 1.81
6 mos.	7.58 ± 2.03	7.51 ± 2.04	7.66 ± 1.97	7.55 ± 2.11
12 mos.	7.44 ± 1.92	7.58 ± 1.64	7.09 ± 2.10	7.68 ± 1.92
Whole Grains					0.64 (0.31)	−0.06 (0.04)	−0.002
(Potential Max Score 10)					0.40 (0.32)	−0.002 (0.05)	(0.003)
0 mos.	4.36 ± 2.37	3.99 ± 2.67	4.52 ± 2.30	4.58 ± 2.09
6 mos.	4.16 ± 2.35	3.46 ± 2.39	4.41 ± 2.29	4.52 ± 2.36
12 mos.	4.14 ± 2.38	4.24 ± 2.70	3.92 ± 2.00	4.29 ± 2.48
Dairy					−0.09 (0.29)	0.02 (0.04)	−0.0006
(Potential Max Score 10)					0.11 (0.29)	0.004 (0.04)	(0.002)
0 mos.	5.36 ± 2.16	5.50 ± 1.89	5.24 ± 2.39	5.33 ± 2.19
6 mos.	5.60 ± 2.29	5.28 ± 2.29	5.29 ± 2.33	6.24 ± 2.13
12 mos.	5.59 ± 1.96	5.56 ± 1.88	5.69 ± 2.04	5.49 ± 1.98
Total Protein Foods					−0.14 (0.10)	0.01 (0.015)	−0.0004
(Potential Max Score 5)					0.003 (0.11)	0.0002 (0.02)	(0.0008)
0 mos.	4.29 ± 0.74	4.30 ± 0.80	4.22 ± 0.80	4.36 ± 0.60
6 mos.	4.13 ± 0.83	4.14 ± 0.83	4.1 ± 0.81	4.16 ± 0.85
12 mos.	4.31 ± 0.76	4.26 ± 0.76	4.33 ± 0.71	4.32 ± 0.81
Seafood and Plant Protein					0.12 (0.17)	−0.03 (0.023)	0.0005
(Potential Max Score 5)					0.26 (0.17)	−0.04 (0.024)	(0.0015)
0 mos.	1.64 ± 1.23	1.45 ± 1.23	1.61 ± 1.27	1.85 ± 1.18
6 mos.	1.59 ± 1.21	1.76 ± 1.14	1.56 ± 1.29	1.47 ± 1.19
12 mos.	1.75 ± 1.36	1.97 ± 1.48	1.50 ± 1.22	1.81 ± 1.38
Fatty Acids					0.09 (0.30)	−0.014 (0.042)	−0.0007
(Potential Max Score 10)					0.17 (0.30)	−0.02 (0.043)	(0.003)
0 mos.	4.98 ± 2.10	4.91 ± 1.85	4.94 ± 2.38	5.07 ± 2.07
6 mos.	4.80 ± 2.23	4.50 ± 2.18	5.00 ± 2.41	4.85 ± 2.08
12 mos.	5.13 ± 2.17	5.33 ± 1.99	5.07 ± 1.95	5.02 ± 2.56
Sodium					0.06 (0.25)	−0.07 (0.035)	−0.006
(Potential Max Score 10)					−0.13 (0.25)	−0.02 (0.036)	(0.002)
0 mos.	2.61 ± 1.83	2.74 ± 2.07	2.67 ± 1.71	2.43 ± 1.69
6 mos.	3.13 ± 1.97	3.35 ± 2.09	3.24 ± 2.09	2.80 ± 1.69
12 mos.	2.81 ± 1.79	3.27 ± 1.99	2.27 ± 1.42	2.98 ± 1.86
Empty Calories					0.004 (0.39)	0.03 (0.056)	−0.007
(Potential Max Score 20)					0.30 (0.40)	−0.04 (0.056)	(0.003)
0 mos.	15.72 ± 3.07	15.6 ± 2.96	15.6 ± 3.40	16.0 ± 2.85
6 mos.	15.71 ± 3.05	15.5 ± 2.92	15.9 ± 3.27	15.7 ± 3.00
12 mos.	15.27 ± 3.16	15.2 ± 3.29	15.7 ± 2.99	14.9 ± 3.24

Values for the Group and Group-by-time interaction are listed as first row (225 min/wk relative to 150 min/wk) and second row (300 min/wk relative to 150 min/wk). The models are adjusted for sex and age and have a random intercept for each subject.

Diet Quality

Total HEI-2010 Scores

Total HEI-2010 scores for the total sample and by group over the 12-month intervention are presented in Table 2. Mean baseline score for the total sample was 58.8 ± 7.6. Mixed modeling revealed that after adjusting for age and sex there was no change in total HEI-2010 scores across the intervention (β; 0.08; 95% CI −0.15-0.32 p =0.49). Additionally, there was no evidence for group differences (225 mins/week vs 150 mins/week: 0.75; 95% −1.4-2.9 ; p = 0.49, 300 min/week vs 150 min/week: 0.60; 95% −1.6—2.8.; p = 0.58) or group-by-time interactions (225 mins/week vs 150 mins/week: −0.12; 95% −0.42-0.19; p = 0.44, 300 min/week vs 150 min/week: −0.07; 95% −0.38-0.24.; p = 0.66). However, when we combined all groups and examined the impact of cumulative MVPA, assessed by accelerometer on changes in total HEI-2010 scores across time mixed modeling revealed that, after adjusting for age, sex, time, group, and the group-by-time interaction there was evidence for an effect of cumulative MVPA (−0.02, p = 0.003). Although, this effect was small, for every standard deviation increase in cumulative MVPA (~ 46 minutes) we would expect HEI total score to decline by about 0.92 points (~ 1.5%).

HEI-2010 Component Scores

Each of the HEI-2010 components for the total sample and by group over the 12-month intervention are presented in Table 2. Mixed modeling revealed that after adjusting for age and sex there was no change in any of the HEI-2010 component scores across the intervention (all p > 0.05). Additionally, there was no evidence for group differences or group-by-time interactions for any of the component scores. When we combined all groups and examined the impact of cumulative MVPA, assessed by accelerometer, on changes in HEI-2010 cumulative scores across time, mixed modeling revealed that, after adjusted for age, sex, time, group, and the group-by-time interaction there was evidence for an effect of cumulative MVPA on sodium (−0.006, p = 0.002) and empty calorie scores (−0.007, p = 0.004). Thus, for every standard deviation increase in cumulative MVPA (~ 46 minutes) we would expect sodium scores to decline by about 0.28 points (~ 10%) and empty calorie scores to decline by 0.32 (~2%)

Impact of Energy Intake and HEI Scores on Change in Weight

We examined the impact of cumulative energy intake, HEI-2010 total scores, and HEI-2010 component scores on change in weight across the 12-month intervention. Mixed modeling revealed that after adjusting for age, sex, group randomization, and cumulative MVPA there was evidence of a relationship between cumulative daily EI and weight (β: 0.00187 95% CI 0.001, p= 0.003). There was no evidence for a relationship between HEI total score (β: −0.006, 95% CI −0.07-0.06) or component scores (all p > 0.05) and change in weight across time.

DISCUSSION

Participants increased their self-reported energy intake by ~300 kcals/day and maintained their total HEI-2010 scores across the 12-month weight maintenance intervention. This increase in energy intake was expected since individuals had just completed a weight loss program, which included reduced energy intake, and were recommended to increase their daily energy intake to levels appropriate for weight maintenance (~1800 kcals/day). This change in energy intake is similar to an 18-month (6-month weight loss, 12-month weight maintenance) trial, conducted by our group, which utilized a similar diet intervention (e.g. portion control meals), an identical education curriculum, and compared phone to in-person weight management delivery in 295 individuals with overweight and obesity (BMI = 35.1±4.9, Age = 43.8±10.2, Minority = 39.8%) and observed a mean energy intake increase of ~200 kcals/day (21). Interestingly, in this previous study participants had a mean decrease of 8.9 points in total HEI-2010 scores across the one-year weight maintenance period while in the current study participants maintained their HEI-scores across the 12-month intervention.

While the total sample demonstrated an increase in mean energy intake across the 12-month intervention, there was minimal evidence that exercise group randomization influenced self-reported energy intake in individuals enrolled in a weight loss maintenance program. The estimated group effects indicated self-reported energy intake in the 150 min/week group was higher than the 225 min/week group and lower than the 300 min/week group, providing additional evidence for the lack of a linear effect of group randomization on self-reported energy intake. However, in all three groups, average weekly exercise minutes fell short of the recommendations, with participants randomized to the 150, 225 and 300 min/week groups only completing 129, 153 and 179 mins/week, respectively. This range of exercise minutes (~50 mins/week) may have limited our ability to see group differences. To overcome this limitation and better examine the association between physical activity and diet intake, we combined all groups and examined the impact of cumulative MVPA, assessed by accelerometry, on self-reported energy intake. Our results suggest that increased MVPA is associated with a slight increase in energy intake over time. However, this association is small; for each standard deviation increase in cumulative MVPA (~ 46 minutes) energy intake increased by 50 kcals/day (~ 3%). Additionally, this increase in energy intake was associated with minimal changes in weight, for example, for every standard deviation increase in daily EI (~ 427 kcal/day) we would expect weight to increase by only 0.43kg.

We are unaware of previous studies that included individuals who were enrolled in a weight loss maintenance program, or which measured physical activity rather than exercise to be able to compare our results. Our findings are most comparable to the results by Martin et al ²² in 171 individuals with overweight or obesity (BMI = 31.5 ± 4.7, Age = 48.9 ± 11., Minority = 33.3%) which reported individuals who exercised at levels that expended 8 kcal/kg/week and 20/kcals/kg/week had a 3.9% and 5.5% increase in energy intake, assessed by doubly labeled water, across the 6 month intervention. However, the results of Martin et al differ from previous randomized and non-randomized trials conducted by our group ^34-37 and others ^38-45 which found no impact of increased exercise or physical activity on energy intake. These inconsistent findings may be due to a variety of study design differences including the measurement of exercise vs. total daily physical activity, different exercise parameters (e.g. mode, frequency, intermittent vs. continuous, time of day), participant characteristics (e.g. age, body weight/composition, race/ethnicity, and aerobic capacity), and type of energy intake assessment (e.g. self-report, direct observation, doubly labeled water).

In the current study there was no evidence of differences in HEI scores between group randomization, however, there was an effect of cumulative MVPA on changes in total HEI-2010 scores as well as sodium scores and empty calories (a measure of the percent of total calories from added sugar and saturated fat). This suggests that individuals enrolled in a weight management intervention who engage in more MVPA may be more likely to consume slightly more empty calories (i.e. added sugar and saturated fat), and sodium. However, like energy intake, this association was small; a 36-minute increase in MVPA was only associated with a 0.75 decrease in total HEI-2010 scores. Furthermore, there was no relationship between HEI and changes in weight, suggesting that any decrease in HEI that may be observed in relationship to increased MVPA does not impact weight. These results are in slight contrast with one previous study, by our group, which examined the effect of exercise training on changes in HEI-2010 ³⁷. In this previous study, 141 previously sedentary, overweight/obese young adults were randomized to 2 levels of energy expenditure (supervised exercise, 5 days/wk at 400 and 600 kcals/session) or a non-exercise control for 10 months. Dietary intake was assessed at baseline, 3.5, 7 and 10 months over 7-day periods of ad libitum eating in a university cafeteria using digital photography plus recall methodology ⁴⁶. There were no significant between-or within-group differences in HEI-2010 scores in the total sample, or in men or women. The HEI-2010 score averaged across all time periods was 37.6 ± 8.9, 35.6 ± 8.4 and 36.7 ± 8.5 for the 400 kcal/session, 600 kcal/session, and control groups, respectively. However, participants were college students who were required to eat in a university cafeteria, and thus these results may not apply to the average adult population.

One major difference between the current study and previous studies is that participants were enrolled in a weight maintenance intervention and intentionally trying to maintain their body weight through their eating and exercise habits, whereas previous trials included individuals who had not previously lost clinically significant weight (≥ 5%). Previous findings suggest that there may be physiological adaptations in response to weight loss that elevate appetite and may contribute to increased energy intake and subsequent weight regain ^12,13. Compensatory physiological adaptations following diet-induced weight loss include decreases in energy expenditure, fat oxidation and anorexigenic hormone (e.g, leptin) levels, and increases in appetite, craving and orexigenic hormone levels (e.g ghrelin)⁴⁷. Preclinical research in rat models suggest that increased exercise may counter the biological drive to increase energy intake in order to regain lost weight ^12,48. Our results showed a small increase in energy intake (300 kcals) during a weight maintenance period, following weight loss >5% of body weight. Suggesting that a weight maintenance program that includes increased MVPA may mitigate the increased energy intake and decreased diet quality typically associated with weight loss. However, all individuals in this trial were in an active weight management intervention and were only followed for 12 months, thus our results need to be interpreted with caution.

Previous research also suggesting behavioral strategies used to induce weight loss are commonly employed in a transient manner and adherence generally wanes as weight loss plateaus, often leading to weight regain ¹¹. The current study observed decreased MVPA and behavioral session attendance from 6 to 12 months, the same period when weight regain was the highest. Interestingly, HEI and energy intake appeared to remain fairly stable (±100kcals) during this time period. However, given limitations self-reported dietary intake, and the lack of additional factors that may be associated with weight gain, such as changes in mood and life stressors ⁴⁹, no conclusions can be drawn from these results. Future, research is needed to determine what factors (e.g., diet, energy intake, behavioral session attendance), are associated with weight regain across time.

Strengths of the current investigation include the use of a randomized design, a weight maintenance intervention 12 months in length, and the use of objectively measured physical activity. However, this study is limited in that it was not specifically designed to detect differences in self-reported dietary intake either within or between intervention groups. Additionally, dietary intake was assessed by self-report food records, which may not provide data of sufficient quality to adequately address this question ⁵⁰. Finally, all participants were randomized to an exercise intervention, thus there was no control group to compare these results against. Therefore, we recommend additional randomized trials to specifically evaluate the impact of exercise training on energy intake and diet quality during weight maintenance that: 1) are powered to detect clinically significant differences in energy intake and/or diet quality; 2) utilize state-of-the-art techniques for the assessment of energy intake such as direct observation weigh and measure, image-assisted recalls, or doubly labeled water and provide multiple measures across the duration of the study; 3) include a per-protocol study design where participants are required to complete their assigned exercise prescription; 4) evaluate both the effect of exercise parameters (e.g. mode, frequency, intensity duration, time course) and participant characteristics (e.g. age, gender, body weight, activity level, ethnicity) on the association between exercise and energy intake and diet quality.

In summary, we found minimal evidence that exercise group randomization (150, 225, or 300 mins/wk) influences self-reported energy intake or changes in diet quality in adults with overweight or obesity who had achieved >= 5% weight loss and were enrolled in a weight loss maintenance program. Evidence suggests that increased MVPA, measure by accelerometer, is associated with slightly increased energy intake and decreased diet quality. However, these associations are not clinically relevant and do not appear to significantly impact weight.