To evaluate whether snacking would improve weight loss and weight maintenance in overweight individuals within the context of a structured meal replacement (MR) weight loss program.
A prospective 24 week, 2 (snacking vs nonsnacking) × 2 (MR vs meal replacement augmented with snacks (MRPS)) randomized trial. Participants were instructed to limit their total daily intake to 1200 (women) or 1500 (men) kcals. Those receiving the MR program were instructed not to snack while those in the MRPS program were told to snack three times per day.
A total of 100 participants were block-randomized, based on prestudy snacking status (high vs low), to receive a standard meal replacement program (MR) or MRPS.
Weight, height, blood pressure, lipid fractions, glucose, and insulin were assessed at the baseline, 12-, and 24 weeks.
Completers analysis at 24 weeks demonstrated a significant time effect (F(1,46)=44.6, P<0.001), indicating that all participants lost significant amounts of weight regardless of group assignment. An intention-to-treat model resulted in similar results. By week 24, the average weight loss across groups was 4.6 kg. There also were significant improvements across all groups among completers for systolic blood pressure (P=0.047), cholesterol (P=0.001), LDL (P=0.001), glucose (P=0.004), and insulin (P=0.001) at week 12, and glucose (P=0.001) and insulin at week 24 (P=0.003).
Our results suggest that a participant's preferences for snacking did not affect their response to treatment. Snackers and nonsnackers responded equally well whether they received a standard meal replacement program or one augmented with snacks.
Increased meal frequency may improve appetite control and prevent individuals from overeating at the next meal.1, 2 Fabry and co-workers3, 4, 5 and Kant et al6 demonstrated significant inverse relationships between meal frequency and body weight. Ma and co-workers7 also found that participants who reported four or more eating episodes per day experienced a 45% decreased risk of obesity. Meal frequency and snacking also may affect plasma lipids.7 For example, participants in a prospective cohort study who reported eating more frequently during the day experienced significant decreases in their overall total cholesterol and LDL cholesterol levels, but no decrease in BMI or weight, even though they experienced a higher daily intake of energy from fat, fatty acids, carbohydrates, and proteins.8 Other studies9, 10 have found similar results, that is, total cholesterol and LDL cholesterol levels were lowered regardless of the intake of total and saturated fats, leading some investigators to conclude that greater meal frequency may play an important role in changing fasting serum lipid levels due to changes in insulin secretion.9 However, these studies suffer from methodological limitations such as dietary under-reporting, misclassification of meal frequency, and post hoc changes in diet patterns that may affect any conclusions about snacking and weight.11
The aim of the present study was to evaluate whether snacking would improve weight loss and weight maintenance in overweight individuals within the context of a prospective, randomized trial using a meal replacement (MR) weight loss program. Secondary objectives included whether snacking improved other markers of cardiovascular disease risk, including blood pressure, plasma lipoproteins, fasting blood glucose, and insulin. While MR strategies have been demonstrated to be extremely effective in producing significant weight losses,12, 13, 14, 15, 16, 17 energy deficit diets frequently suffer from the liability of leaving individuals feeling hungry, a significant risk for relapse.18, 19 In addition, no studies to date have evaluated the use of snacks within the context of meal replacement programs or with individuals who are high-frequency snackers.
A total of 100 participants (50 ‘snackers’ and 50 ‘nonsnackers’) were recruited through the Nutrition Research Clinic's Recruitment Database and by word-of-mouth. Study inclusion criteria required that participants be between the ages of 35 and 55 y, have a BMI between 25 and 30, blood pressure less than 140/90 mmHg, fasting blood glucose less than 126 mg/dl, prescriptions for current medications must have begun at least 2 months prior to screening, and have plans to remain in Houston, TX for 1 y after beginning the study. Exclusion criteria included the presence of insulin dependent diabetes, pregnancy, significant conditions, diseases, or medications which could impact the results of the study, previous participation in any weight loss study by the Nutrition Research Clinic, or weight loss greater than 10 pounds during the last 6 months. Participants who reported current use of antihypertensive or antidiabetic medications (insulin or oral agents) were allowed to participate.
In total, 323 potential participants were prescreened via the Weight Loss Study Screening Questionnaire. Of those prescreened, 191 were screen failures. The main reasons for failure were having a BMI exceeding 30 kg/m2 or not meeting the criteria for the snacking or nonsnacking categories. Snacking status was determined by how participants answered questions about the number of meals and snacks that they eat on weekdays and weekends on a snacking habit questionnaire. Participants with an average of 4.5 daily eating episodes or less were classified as ‘nonsnackers’, while those whose average was 5.5 or more were classified as ‘snackers’. Participants whose average daily eating episodes were between 4.6 and 5.4 were ineligible for the study. These cutpoints were based on data on meal frequency in the US from NHANES, which demonstrated that men and women ate at an average of 5.3 and 4.9 meals per day, respectively.5 Thus, our definition of ‘snackers’ exceeds the average meal frequency of adults. We used this extreme groups approach to maximize potential differences in how the snacking products might be used, reasoning that high-frequency snackers might benefit most from a weight loss program augmented with snacks.
In total, 132 individuals were asked to attend a fasting screening visit. Of those, 26 did not attend, six came but were screen failures and 100 passed the screening process. Randomization was in a block design balancing snackers and nonsnackers into receiving either MR alone or MR+snacks. The recruitment and randomization of participants are presented in Figure 1.
Baseline characteristics of participants in the four groups are presented in Table 1.
Weight was measured at each clinic visit on a calibrated balance beam scale with participants wearing light clothing and without shoes at all measurement and brief clinical visits. Height (baseline only) was assessed using a wall mounted stadiometer. BMI was calculated as weight (kg)/height (m)2. Lipid fractions (total cholesterol, HDL, LDL, VLDL, and triglycerides) were assessed at the baseline, 12-, and 24 weeks with 12-h fasting blood samples drawn by the Atherosclerosis Clinical Research Laboratory of The Methodist Hospital, Houston, TX, USA.
Blood pressure was measured in the seated position using the same arm and the same appropriately sized cuff was used throughout the study. In addition, all readings were taken at approximately the same time of day (±2 h). For each indicated measurement of blood pressure, three separate readings at approximately 2-min intervals were recorded and the last two were averaged.
Snacking and nonsnacking participants were assigned to a 24-week program of MR alone (Slim Fast) or meal replacement augmented with snacks (MRPS) (Slim Fast+snacks). The efficacy of the Slim Fast MR program for weight loss has been established in a number of short- and long-term studies (eg Ditschuneit et al,13 Flechtner-Mors et al,14 Heymsfield et al,15 Rothacker,16 and Noakes et al17). Women were instructed to limit their total daily intake to 1200 kcals and men to 1500 kcals per day and were given a book with kcals counts for individual foods including fast-foods. Slim Fast coupons for shakes, meal bars and snack bars were dispensed at each 2-week visit to each group. Also, Slim Fast pasta and soup products not available on the market when the study started were furnished to all of the study participants. Participants in the snacking conditions were told to use the products for both meals and snacks while those in the nonsnacking groups were told to only use the products for meals.
All participants were asked to use a minimum of two Slim Fast products a day. Those participants in the MR+snack intervention were instructed to use Slim Fast snack bars or other foods such as fruit, low-calorie yogurt, etc, for their three snacks a day. Participants in this group also received instructions for consuming healthy meals and how to include snacks throughout the day as part of the Slim Fast Easy Options Plan. In contrast, those in the standard MR program without snacks were instructed about how to avoid excess calories between meals and were explicitly encouraged to not use any snacks as part of their diet. All participants visited the center once per week and received brief counseling (20 min) from a registered dietitian. Daily diaries were given out to both groups at each visit to be completed and returned at the next visit. The diaries were used for adherence counseling only. At each session with the dietitian, problems were discussed, behaviorally- and/or food-oriented handouts were dispensed and goals were set to be completed by the next visit.
Statistical analysis plan
Separate statistical outcome models were developed for participants who completed the study at 12 and 24 weeks and for intention-to-treat models by using last observation carried forward (LOCF) for both time points, consistent with the CONSORT guidelines for randomized clinical trials. 20Participants were randomized to treatment based on their snacking preference, with snackers and nonsnackers both randomized to either MR only (MR) or MR plus Snacks (MRPS). Thus, four treatment groups were formed, snackers in the MR group (MR/S), snackers in the MRPS group (MRPS/S), nonsnackers in the MR group (MR/NS), and nonsnackers in the MRPS group (MRPS/NS). Data were analyzed using general linear models for repeated measures at each time point and for both completers and LOCF, resulting in four outcome models. For cardiovascular outcomes, only completers analyses were performed. Otherwise, a similar analytic strategy was followed for cardiovascular parameters as was used with weight.
Intervention fidelity check
In order to determine if participants' snacking patterns matched their preassignment snacking status and their group assignment, all were asked about how many snacks they consumed on an average weekday and weekend. There were significant group differences in snacking frequency for both average weekday and weekend snacking at baseline (P<0.001 weekday; P<0.001 weekend day), 12- (P<0.001 weekday; P<0.001 weekend day), and 24-weeks (P<0.040 weekend day only; NS weekday P=0.134) (see Figure 2).
As would be expected, both groups of self-reported snackers (MR/S and MRPS/S) reported significantly more snacks than both self-reported nonsnacking groups (MR/NS and MRPS/NS; P< 0.001 for all post hoc tests) at baseline for both weekday and weekend day snacking. By week 12, assignment to one of the snacking groups was the primary determinant of weekday or weekend day snacking, with both the MRPS/S and MRPS/NS eating significantly more snacks than those assigned to a nonsnacking treatment condition, regardless of self-reported snacking status (P<0.001 for all post hoc tests). At week 24, both the MRPS/S and MRPS/NS reported greater snacking frequency than MR/NS (P=0.007 and 0.041, respectively). No other post hoc comparisons were statistically significant.
Overview of weight loss
There were no significant differences at baseline between patients randomized to the four groups with exception to differences in total cholesterol (P=0.050) between the snackers receiving MRPS (219.08±30.60) and the nonsnackers receiving MRPS (193.60±30.83). There were no significant differences in attrition across groups (P=0.923), with attrition ranging from 40 to 48%. All of the groups experienced weight loss at 12 and 24 weeks, yet there were no significant differences between the groups. Figure 3 presents weight changes (kg) for the four treatment groups based on completers at each end point.
Treatment outcomes at week 12
Table 2 presents weight losses at 12- and 24 weeks.
At 12 weeks based on a completers analysis, there was a significant time effect for weight loss (F(1,53)=106.3, P<0.001) indicating that there was significant weight loss across groups. There was no significant interaction of treatment group with time, suggesting that groups did not differ on weight loss from baseline to week 12 (F(3,53)=0.82, P=0.983).
The LOCF analysis at week 12 produced similar outcomes as the completers analysis. Although all groups demonstrated significant weight losses (F(1,96)=77.5, P<0.001), the treatment group by time interaction was not significant (F(3,96)=1.6, P=0.185).
Treatment outcomes at week 24
A completers analysis at week 24 (see Table 2) demonstrated a significant time effect (F(1,46)=44.6, P<0.001), indicating that participants lost significant weight across groups. The treatment group by time interaction was not significant (F(3,46)=0.29, P=0.835), suggesting no differential treatment effect.
As with the completers analysis, there was a significant time effect (F(1,96)=46.69, P<0.001), while the treatment group by time interaction was not significant (F(3,96)=0.58, P=0.629).
Table 3 presents cardiovascular outcomes for completers at each of the study measurement points.
General linear models with repeated measures were used to examine treatment effects at 12 and 24 weeks. There were significant improvements across groups among completers for systolic blood pressure (P=0.047), cholesterol (P=0.001), LDL (P=0.001), glucose (P=0.004), and insulin (P=0.001) at week 12 and glucose (P=0.001) and insulin at week 24 (P=0.003). There were no significant differences among the groups on changes in the cardiovascular parameters.
We also examined the CVD outcomes using LOCF values. There were significant improvements across groups at 12 weeks for systolic blood pressure (P=0.042), cholesterol (P=0.001), LDL (P=0.001), glucose (P=0.001) and insulin (P=0.001) at week 12. At week 24, cholesterol (P=0.032), LDL (P=0.048), glucose (P=0.001), and insulin (P=0.002) improved. However, there were no significant changes by group status at either time point.
Discussion and Conclusion
The results of our study suggest that the participants' preferences for snacking did not affect their response to treatment. Snackers and nonsnackers responded equally well whether they were in the MR or MRPS intervention with all groups losing between 4.6 and 6.1% of initial body weight. As noted earlier, we classified participants as snackers or nonsnackers using cutpoints that were based on data on meal frequency in the US from NHANES. This approach represented what Shadish et al21 have referred to as sampling based on heterogeneous instances or extreme groups, which attempts to maximize potential group differences on an independent variable of interest. To the best of our knowledge, no previous studies have attempted to classify and randomize participants into a weight loss study based on their self-reported snacking status, so there is little precedent to rely upon for guidance.22 Hampl et al22 attempted to classify snacking status in an observational study, but only used a crude frequency measure to classify individuals as multiple or never snackers. It is possible that our approach was insensitive or inaccurate in classifying snackers, thus leading to null results with respect to matching snacking status with the use of snacks.23 Another possibility is that our weight loss program was too intensive and a less intensive program may have yielded differences across conditions and by snacking status.
Our weight loss results are similar to those reported in other meal replacement studies. For example, a meta-analysis of meal replacement studies found that participants lost an average of 6.28 and 7.31% of initial body weight at 3 and 12 months, respectively, across six studies.15 These studies have reported similar short-term improvements in other cardiovascular parameters such as blood pressure, lipids, and glucose. Thus, the inclusion of snacks in our study did not mitigate weight losses produced by a meal replacement program or appear to increase risk for weight gain. However, the increased meal frequency in the MRPS group did not result in greater improvements in lipids or other CVD risk factors (eg, glucose, insulin, blood pressure), as sometimes seen in previous studies evaluating benefits associated with greater meal frequency (eg Ma et al,7 Cunnane and Rao, 9 and Arnold et al23).
A limitation of our study was the attrition experienced at both 12 and 24 weeks, a total of 50% at 24 weeks. Thus, our ability to detect group differences based on snacking status was clearly hampered by attrition, as well as the high variability in weight changes. However, only six participants complained about the use of the meal replacement products or their group assignment. Thus, while our attrition was similar to that reported in other MR or other recent caloric restriction studies (eg Ashley et al,12McManus et al,24 and Samaha et al25), attrition across groups was not differential and our results were only modestly changed when using conservative ITT models to address loss to follow-up.
In conclusion, our results demonstrated that the addition of snacks to a meal replacement program did not negatively impact weight loss or CVD risk factors when compared to a standard meal replacement program alone. Classifying participants based on their self-reported status as high vs low-frequency snackers did not improve treatment outcomes, that is, snackers who received the MRPS program did not experience greater weight losses when compared to snackers receiving MR alone. However, our study was limited by attrition and high variability in weight changes. Thus, while we could not demonstrate that treatment matching based on self-reported snacking status enhanced outcomes, our lack of significant findings also might be due to type II error. Future research should attempt to replicate our study with larger samples. To the best of our knowledge, this is the first study that actually assessed and randomized participants to treatment based on their snacking status in a weight loss study paradigm.
Drummond S, Crombie N, Kirk T . A critique of the effects of snacking on body weight status. Eur J Clin Nutr 1996; 50: 779–783.
Bellisle F, McDevitt R, Prentice AM . Meal frequency and energy balance. Br J Nutr 1997; 77 (Suppl 1): S57–S70.
Fabry P, Fodor J, Hejl Z, Braun T, Zvolankova K . The frequency of meals: its relation to overweight, hypercholesterolaemia, and decreased glucose tolerance. Lancet 1964; ii: 614–615.
Fabry P, Hejda S, Cerny K, Osancova K, Pechar J . Effect of meal frequency in schoolchildren: changes in weight-height proportion and skinfold thickness. Am J Clin Nutr 1966; 18: 358–361.
Hejda S, Fabry P . Frequency of food intake in relation to some parameters of the nutritional status. Nutr Dieta 1964; 6: 216–221.
Kant AK, Schatzkin A, Graubard BI, Ballard-Barbash R . Frequency of eating occasions and weight change in the NHANES I Epidemiologic Follow-up Study. Int J Obes Relat Metab Disord 1995; 19: 468–474.
Ma Y, Bertone ER, Stanek 3rd EJ, Reed GW, Hebert JR, Cohen NL, Merriam PA, Ockene IS . Association between eating patterns and obesity in a free-living US adult population. Am J Epidemiol 2003; 158: 85–92.
Titan SMO, Bingham S, Welch A, Luben R, Oakes S, Day N, Khaw KT . Frequency of eating and concentrations of serum cholesterol in the Norfolk population of the European prospective investigation into cancer (EPIC-Norfolk): cross sectional study. BMJ 2001; 323: 1–5.
Cunnane SC, Jenkins DJ, Wolever TM, Vuksan V, Brighenti F, Venketeshwer R, Jenkins AL, Buckley G, Patten R, Singer W, Corey P, Josse R . Nibbling versus gorging: Metabolic advantages of increased meal frequency. N Engl J Med 1989; 321: 929–934.
Edelstein SL, Barrett-Connor EL, Wingard DL, Cohn BA . Increased meal frequency associated with decreased cholesterol concentrations. Am J Clin Nutr 1992; 55: 664–669.
McGrath SA, Gibney MJ . The effects of altered frequency of eating on plasma lipids in free-living healthy males on normal self-selected diets. Eur J Clin Nutr 1994; 48: 402–407.
Ashley JM, St Jeor ST, Schrage JP, Perumean-Chaney SE, Gilbertson MC, McCall NL, Bovee V . Weight control in the physician's office. Arch Intern Med 2001; 161: 1599–1604.
Ditschuneit HH, Flechtner-Mors M, Johnson TD, Adler G . Metabolic and weight-loss effects of a long-term dietary intervention in obese patients. Am J Clin Nutr 1999; 69: 198–204.
Flechtner-Mors M, Ditschuneit HH, Johnson TD, Suchard MA, Adler G . Metabolic and weight loss effects of long-term dietary intervention in obese patients: four-year results. Obes Res 2000; 8: 399–402.
Heymsfield SB, van Mierlo CAJ, van der Knaap HCM, Heo M, Frier HI . Weight management using a meal replacement strategy: meta and pooling analysis from six studies. Int J Obes Relat Metab Disord 2003; 27: 537–549.
Rothacker DQ . Five-year self-management of weight using meal replacements: comparison with matched controls in rural Wisconsin. Nutrition 2000; 16: 344–348.
Noakes M, Foster PR, Keogh JB, Clifton PM . Meal replacements are as effective as structured weight-loss diets for treating obesity in adults with features of metabolic syndrome. J Nutr 2004; 134: 1894–1899.
Teixeira PJ, Going SB, Sardinha LB, Lohman TG . A review of psychosocial pre-treatment predictors of weight control. Obes Rev 2005; 6: 43–65.
National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute (NHLBI). Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity: the evidence report. US Government Press: Washington, DC; 1998.
Moher D, Schulz KF, Altman D, CONSORT Group (Consolidated Standards of Reporting Trials). The CONSORT Statement. Revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA 2001; 285: 1987–1991.
Shadish WR, Cook TD, Campbell DT . Experimental and quasi-experimental designs for generalized causal inference. Boston: Houghton-Mifflin; 2002.
Hampl JS, Heaton CLB, Taylor CA . Snacking patterns influence energy and nutrient intakes but not body mass index. J Hum Nutr Diet 2003; 16: 3–11.
Arnold L, Ball MJ, Duncan AW, Mann J . Effect of isoenergetic intake of three or nine meals on plasma lipoproteins and glucose metabolism. Am J Clin Nutr 1993; 46: 611–620.
McManus K, Antinoro L, Sacks F . A randomized controlled trial of a moderate-fat, low-energy diet compared with a low fat, low-energy diet for weight loss in overweight adults. Int J Obes Relat Metab Disord 2001; 25: 1503–1511.
Samaha FF, Iqbal N, Seshadri P, Chicano KL, Daily DA, McGrory J, Williams T, Williams M, Gracely EJ, Stern L . A low-carbohydrate as compared with a low-fat diet in severe obesity. N Engl J Med 348; 21: 2074–2081.
This study was funded by a grant from the Slim Fast Corporation.
About this article
Effect of meal frequency on glucose and insulin levels in women with polycystic ovary syndrome: a randomised trial
European Journal of Clinical Nutrition (2016)
Effectiveness of a Medifast meal replacement program on weight, body composition and cardiometabolic risk factors in overweight and obese adults: a multicenter systematic retrospective chart review study
Nutrition Journal (2015)
Eating frequency is inversely associated with blood pressure and hypertension in Korean adults: analysis of the Third Korean National Health and Nutrition Examination Survey
European Journal of Clinical Nutrition (2014)
Nutrition Journal (2012)