Exercise and Protein Effects on Strength and Function With Weight Loss in Older Women

Ellen M. Evans; Chad R. Straight; Rachelle A. Reed; Alison C. Berg; David A. Rowe; Mary Ann Johnson


Med Sci Sports Exerc. 2021;53(1):183-191. 

In This Article


Study Design, Overview, and Participants

Relatively healthy but overweight/obese inactive older women were recruited with the following primary inclusion criteria: 65 to 80 yr old, body mass index (BMI) ≥25.0 kg·m−2, and physically inactive (<1 h·wk−1 of physical activity or less than two exercise sessions per week) over the past 6 months. The primary exclusion criteria were as follows: habitual smoker, dietary restrictions that precluded adherence to the dietary protocol, history of unstable cardiovascular disease, self-report of active cancer or treatment within previous 5 yr, psychiatric or cognitive disorder that precluded the ability to adhere to study protocols, severe arthritis, or other conditions that precluded safe adherence to the exercise prescription. Women randomized to the exercise interventions were also required to complete a graded exercise test administered by the study physician to ensure safety for exercise participation. All participants obtained medical clearance from their personal physician. Participants were stratified by age and BMI and randomized using an online program (www.randomnumbergenerator.com) to one of three treatment groups: high-protein diet with exercise (PRO + EX), high-protein diet without exercise (PRO), or conventional protein control diet with exercise (CON + EX). The study coordinator who was not involved with the intervention or data collection, reduction, or analysis completed the stratification and randomization procedures. All procedures of this study were approved by the University of Georgia's Institutional Review Board (protocol ID no. 2012109023), and participants provided written informed consent before enrollment.

Dietary Intervention and Dietary Intake

Energy Goals and Macronutrient Distribution. All individuals, regardless of randomized group, were prescribed diet programs supervised by a registered dietitian nutritionist (RDN) based on a reduction of ~500 kcal from calculated energy needs (Mifflin St. Jeor equation using baseline body weight; activity factor 1.3) to promote loss of ~0.25 to ~1 kg·wk−1.[13] The recommended PRO diet was designed to provide 30% of daily energy from protein, targeting at a minimum 1.0 g·kg−1 BW·d−1[1] with a prescribed level of ~1.6 g·kg−1 BW·d−1, and the recommended CON diet was designed to provide 18% of daily energy from protein (0.8 g·kg−1 BW·d−1). Both diets were designed to provide 30% of energy from fat, and the remainder of energy from carbohydrate (PRO = ~40%, CON = ~52%). Participants in the PRO intervention groups were instructed to consume at least one serving of cooked lean beef (3 oz) per day. In addition, they were recommended to select high-quality low-fat protein foods as preferred including other lean meats (e.g., poultry, pork), low-fat/fat-free dairy, eggs, and protein bars and powder supplements as preferred for adherence to protein goals, the latter being less encouraged than whole foods. Based on recommended clinical practice given the risk for deficiencies with calorie restriction and under direct oversight by the RDN, all participants were provided a multivitamin mineral supplement (Centrum® Silver® Women, Pfizer, Inc., Madison, NJ). Similarly, additional calcium supplements (Regular Strength TUMS®, 200 mg elemental calcium; GlaxoSmithKline, St. Louis, MO) were prescribed on an individual basis to meet the RDA (1200 mg·d−1).

Dietary Counseling Sessions. At the beginning of the intervention, participants attended a minimum of two individual sessions with the RDN or nutrition graduate student under supervision of the RDN for instruction regarding energy restriction, macronutrient distribution of the intervention diet, and self-monitoring methods. For the remainder of the study, participants attended weekly educational and/or motivational group sessions (45–60 min) and individual sessions as necessary to meet weight loss goals. The group session topics included both general nutrition education and behavioral strategies for weight management. All participants attended one individual session at the midpoint of the intervention to assess progress toward weight loss and dietary goals. Participants were instructed to use a free online dietary intake monitoring application (MyFitnessPal.com) to enhance adherence. Interventionists set up the program with the individualized energy restriction goal and macronutrient distribution per protocol. Entries were monitored weekly, along with weight changes, and participants were provided individualized feedback to help meet weight loss and nutrient goals.

Dietary Intake Assessment. Participants completed diet records for 3 d, including two weekdays and one weekend day, recording all food and beverages consumed each 24-h period, at baseline, midpoint, and postintervention. The RDN or the trained interviewer reviewed the record with the participant to clarify information provided, allowing the participant multiple opportunities to recall food and beverage intake. Food models were used as needed to verify portion sizes. Dietary intake data were assessed using the Nutrition Data System for Research (NDSR 2013, Minneapolis, MN) to determine 3-d average energy intake and diet composition at each data collection time. Note that the MyFitnessPal dietary intake records acquired above were for adherence enhancing purposes only and were not research outcomes.

Exercise Intervention

All participants in the PRO + EX and CON + EX groups were asked to complete 3 × 75-min sessions of supervised exercise on nonconsecutive days during each week of the intervention for 6 months. The exercise intervention was a multicomponent program that integrated cardiorespiratory training (30 min at moderate intensity), resistance training for all major muscle groups (two sets of 8–10 repetitions at 65% of 1-repetition maximum), and balance and functional exercises per established recommendations.[14] All participants were asked to maintain usual activities outside the intervention. Habitual physical activity was assessed objectively using an accelerometer (NL-1000; New Lifestyles, Inc., Lee's Summit, MO) at baseline, midpoint, and postintervention. Step counts and moderate-to-vigorous physical activity (MVPA) minutes were calculated using the average step count from valid wear days.

Outcome Measures

NCT01893684; clinicaltrials.gov The authors acknowledge the following deviations from our registered protocol: (a) omission of body composition outcomes assessed using MRI due to common knee implants in our older adult sample, which were not deemed safe for scanning on our university magnet, and (b) reporting on the secondary outcome measure, physical functional performance (LEPF), within the primary aim in the present manuscript due to increased interest in these end points in recent literature.

Weight Status and Body Composition. Standing height and weight were measured using conventional clinical methods. Whole-body and regional soft tissue composition (absolute [kg] and relative fat mass [%Fat] and mineral-free lean mass) was assessed via dual-energy x-ray absorptiometry (iDXA; GE Healthcare-Lunar, Madison, WI). Regional analyses were performed per manufacturer's guidelines and involved bisecting the femoral neck and patella to measure mineral-free lean mass of the upper leg.

Muscle Strength. Concentric isokinetic knee torque was assessed via isokinetic dynamometry (System 4 Pro; Biodex Medical Systems Inc., Shirley, NY) at 60°·s−1 using 2 sets of 4 extension and flexion repetitions. The greatest extension and flexion peak torque values for each leg were summed to calculate maximal isokinetic knee torque (IK-60).

LEPF. Objective LEPF was measured using three well-established functional tests: the 8-ft up and go (UPGO), 30-s chair stand (CHAIR), and 6-min walk (WALK).[15]

Statistical Analysis

All statistical analyses were conducted using SPSS for Windows version 22.0 (IBM Corp., Armonk, NY) and/or SAS Version 9.4 (SAS Institute, Inc., Cary, NC). Data were examined for distributional characteristics via skewness and kurtosis, and visual inspection of histograms and box plots was used to identify outliers (defined as 3 × the semi-interquartile range from the median). One-way independent-groups ANOVA was used to test for group differences at baseline, and chi-square tests were used to compare groups on categorical variables. Two-way 3 (group) × 3 (time) mixed factorial ANOVA was used to compare change over time among the three study groups. Where the sphericity assumption was violated, the Greenhouse–Geisser adjustment of the degrees of freedom was used to correct for the violation. Statistical significance was set at P < 0.05 for all tests. Where interaction effects were deemed to be significant, the pattern of interaction effects was interpreted using Cohen's d effect sizes[16] with suggested criteria being used to interpret effect sizes as being small (d = 0.2), medium (d = 0.5), or large (d = 0.8). Furthermore, in the presence of a significant interaction effect for a primary outcome, both a per-protocol/completer analysis (n = 61) and an intention-to-treat (ITT) analysis (n = 72) with a last value/observation carried forward approach were used to compare group changes from baseline to posttest using ANOVA and post hoc analysis, LSD.

Power Calculation. Although changes in body composition and muscle strength were of interest, the overarching goal of this study was to assess the relative efficacy of the treatments for obesity to prevent risk of physical disability in older women. Based on their excellent translational meaning regarding daily LEPF and subsequent risk for disability, UPGO and CHAIR were deemed as the primary outcomes of interest for power calculations. Accepted clinically meaningful change in LEPF varies based on type of objective; therefore, similar to other health outcome research, it was accepted that a moderate effect (Cohen's d = 0.5) is the minimal effect to be of meaning. Based on our pilot data,[9] it was determined that a sample of 25 per group with a retention of 80% providing 20 participants per group would provide power of ~0.85 to detect a group–time interaction effect size of 0.50 SD with the three measurement time points (0, 3, and 6 months), an alpha level of 0.05, and an average correlation between repeated measurements of ≥0.80.