Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Aug 1.
Published in final edited form as: Appetite. 2020 Mar 29;151:104687. doi: 10.1016/j.appet.2020.104687

Which strategies to manage problem foods were related to weight loss in a randomized clinical trial?

Liane S Roe 1, Barbara J Rolls 1
PMCID: PMC7305980  NIHMSID: NIHMS1582912  PMID: 32234531

Abstract

Individuals managing their weight are often faced with problem foods that are difficult to resist eating. In the context of a weight-loss intervention, we characterized the most commonly reported problem foods and the behavioral strategies used to manage them, and examined which strategies were related to weight loss. Women with overweight and obesity (N=186) participated in a one-year randomized trial of three interventions (NCT01474759): standard advice to eat less food, choosing portions based on energy density, and using pre-portioned foods. At Months 0, 6, and 12 of the trial, participants listed the foods they found most problematic and reported the frequency of using eight behavioral strategies to control intake of these foods, including three practices for avoiding exposure to problem foods and three for consuming them but limiting intake. The responses showed that 82% of the top three problem foods were in the categories of sweet baked items, salty snacks, starchy side dishes, chocolate and candy, and ice cream. After one year, women who reported more frequently using the strategy of limiting portions of problem foods had a greater rate of weight loss (kg/week), regardless of their intervention group (p<0.0001). Among women who limited portions of problem foods less frequently, those using pre-portioned foods had greater initial weight loss compared to the other two groups, but then regained weight at a greater rate (p<0.0001). The three avoidance strategies for problem foods were reported to be frequently used but were not found to be related to weight loss. These results suggest that adopting and maintaining strategies to manage portions of problem foods, rather than avoiding exposure to them, can be a more useful approach for weight loss.

Keywords: Weight management, obesity treatment, behavioral intervention, portion size, discretionary foods, food craving

1. Introduction

Managing body weight depends on the long-term adoption of changes in eating habits, and this can be undermined by having to avoid or limit certain favorite foods (Appelhans 2016). Such “problem foods” can be defined as those that individuals find hard to resist or to stop eating once they have started. Although it might be expected that the most problematic foods are those high in fat, sugar, and salt (Drewnowski & Almiron-Roig, 2010; Drewnowski, 1992; Fazzino, Rohde, & Sullivan, 2019), researchers have not previously characterized these foods simply by asking individuals to list them, nor is it clear whether various practices to manage problem foods have an influence on weight loss and weight loss maintenance. An opportunity to evaluate experience with problem foods was provided in The Portion-Control Strategies Trial (Rolls, Roe, James, & Sanchez, 2017), a one-year randomized controlled trial of three behavioral weight-loss interventions in women with overweight and obesity. At several points during the trial, participants were asked to identify the foods that they found the most difficult to resist and to report their frequency of using different practices to control intake of these foods. The purpose of the current analysis was to characterize the most commonly reported problem foods and the strategies used to manage them in the context of a weight-loss intervention, and to examine which of these strategies were related to weight loss.

Much of the research on foods that are problematic for overconsumption has focused on food craving, characterized as an intense desire to eat a specific food (Weingarten 1991). Craving involves cognitive, behavioral, and external factors (Rodin 1991; Cepeda-Benito 2000) and differs from hunger, as it is specific for the desired food but does not necessarily lead to consumption of the food (Hill 2007). There is debate on the most effective management of craved foods; researchers have suggested various approaches, from reducing the frequency of consumption of the foods (Apolzan 2017) to encouraging individuals to accept their thoughts about cravings (Buscemi 2017). Research on food cravings, however, has not characterized the approaches that individuals themselves report using to manage their cravings (Taylor 2019), nor how these relate to weight loss. Problem foods that are difficult to resist also share several characteristics with those designated in dietary policy documents as “discretionary foods” (i.e., those left to individual choice after food group needs are met) or as “calories for other uses” (National Health and Medical Research Council 2013; National Health Service 2019; U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). Discretionary items are often high in energy density and low in nutrient density, with high levels of saturated fat, added sugars, refined starches, or salt. Strategies for reducing consumption of discretionary foods have been explored in reviews and surveys (Grieger 2016; Klem 1997; Sciamanna 2011), but these investigations did not focus specifically on foods that are problematic for overconsumption, nor examine the influence of the reported practices on the degree of weight loss.

There has been limited research to identify foods that are problematic for overconsumption; furthermore, researchers often ask respondents to evaluate a pre-specified list of foods, rather than asking them to name the foods they find challenging. Thus, one aim of the present study was to identify the most common problem foods reported by participants during a weight-loss intervention. Another aim was to assess participants’ frequency of using different practices to manage problem foods as well as to examine the stability of the practices during dietary intervention and their relationship to participant characteristics. We were particularly interested in comparing the use of strategies for avoiding exposure to problem foods with those for consuming such foods but limiting their intake. Finally, we examined how the use of the strategies was related to weight loss and weight loss maintenance across the year-long trial, and whether this was affected by participant characteristics. The findings of these analyses will improve understanding of individual experiences with problem foods, in order to facilitate development of more effective and individualized food-focused strategies for weight management.

2. Methods

2.1. Study design

Problem foods and the practices used to control their intake were assessed longitudinally among women in the one-year Portion-Control Strategies Trial. This randomized controlled weight-loss trial compared three interventions: standard dietary advice to eat less and two different methods of portion control. An overview of the trial participants and methods is presented in the following sections; further details of the design and main outcomes have been previously published (Rolls, Roe, James, & Sanchez, 2017).

2.2. Participants

Women were eligible for the trial if they were aged 20–65 years with a body mass index (BMI) of 28–45 kg/m2; exclusion criteria are described in the primary publication for the trial (Rolls, Roe, James, & Sanchez, 2017). A total of 186 women met the criteria and were enrolled in the trial; 81% had obesity and 19% had overweight. At baseline (Month 0), many participants had experience with dieting, with 93 participants (50%) having attempted weight loss two or more times in the previous year. Ninety-seven percent of participants were white and non-Hispanic and 55% had a college degree; other baseline characteristics are shown in Table 1. Participants gave signed informed consent and received financial compensation for their time in the trial. The Office for Research Protections at The Pennsylvania State University reviewed and approved the trial protocol.

Table 1.

Baseline characteristics of 186 women with overweight and obesity in The Portion-Control Strategies Trial

Characteristic Standard Advice Group (N=62) a Portion Selection Group (N=62) a Pre-portioned Foods Group (N=62) a
Mean SD Mean SD Mean SD
Age (y) 49.5 12.0 50.4 9.6 50.1 10.1
Body mass index (kg/m2) 34.1 4.3 33.6 4.2 34.2 4.1
Energy expenditure (kJ/d)b 8874 802 8711 914 8819 775
Energy expenditure (kcal/d)b 2121 192 2082 218 2108 185
Dietary restraint scorec (range 0–21) 9.3 3.5 8.6 3.6 8.8 3.8
Disinhibition scorec (range 0–16) 9.7 3.1 9.8 3.9 9.8 3.5
Hunger tendency scorec (range 0–14) 6.4 3.1 6.7 3.6 5.5 3.3
Depression scored (range 0 – 60) 6.1 5.4 6.0 4.1 5.2 4.7
Open in a new tab
a

None of the characteristics were significantly different across intervention groups (all p>0.15).

b

Energy expenditure was estimated from height, weight, age, sex, and activity level (Institute of Medicine, 2002).

c

Scores were from the Three-factor Eating Questionnaire (Stunkard & Messick, 1985).

d

Score were from the Beck Depression Inventory (Beck, Steer, & Brown, 1996); because of the inclusion criteria, all participants had a score ≤ 25.

2.3. Interventions and assessments

Trial participants were randomly assigned to one of three interventions: the Standard Advice Group, the Portion Selection Group, or the Pre-portioned Foods Group. The Standard Advice Group was instructed to follow dietary guidelines focusing on eating less while incorporating healthy options from all food groups. The Portion Selection Group was trained to use energy density to select food portions and was provided food scales and other portion-control tools. The Pre-portioned Foods Group was instructed to structure meals around pre-portioned foods and was given vouchers for single-serving main dishes, at a rate that decreased over the duration of the trial. Participants in all groups were not discouraged from consuming any specific foods, but were encouraged to limit portion sizes using the principles of their intervention and to substitute healthier alternatives where possible. None of the intervention groups received specific instruction on dealing with problem foods, but all participants had one lesson on snacking and one lesson on comfort foods. All intervention groups received the same instruction about increasing physical activity, completing self-monitoring records, and implementing behavioral change. Trained interventionists (Registered Dietitians or Dietetic Technicians, Registered) conducted individual sessions with participants weekly during Month 1, biweekly during Months 2–6, and monthly during Months 7–12. There were 19 intervention sessions and 4 assessment sessions provided during the trial.

Body weight was measured to the nearest 0.1 kg at 23 sessions over the year; for consistent measurement, participants wore a lightweight outfit that was kept for them at the research center. Height was measured at the baseline assessment session to the nearest 0.1 cm using a stadiometer. At multiple time points during the trial, participants completed questionnaires on computers at the research center using online software (Qualtrics, Provo, Utah, USA). The questionnaires included the Three-factor Eating Questionnaire for dietary restraint, disinhibition, and hunger tendency (Stunkard & Messick, 1985), the Beck Depression Inventory (Beck, Steer, & Brown, 1996), and the Problem Foods Questionnaire.

2.4. The Problem Foods Questionnaire

The Problem Foods Questionnaire was developed for The Portion-Control Strategies Trial as a brief instrument to collect information regarding the foods that participants found the most problematic for overconsumption. The questionnaire consisted of two sections: the identification of problem foods and the use of practices to control their intake (see Supplementary Material). The first section asked participants to report their most common problem foods, defined as “foods that you can’t resist and that you find hard to stop eating once you have started”. Respondents were asked to identify at least one and up to ten foods, ordered from the most problematic to the least, by typing the food names on the computer-administered questionnaire. The questionnaire did not request information on beverages, including alcoholic drinks.

In the second section, participants were presented with a list of eight strategies that might be used to manage intake of problem foods (Table 2). The list was based on published research as well as on consultation with dietitians and interventionists. The most comprehensive previous survey of practices associated with weight loss and weight loss maintenance was Sciamanna et al. (2011). The practices from this survey that concerned intake of specific foods were modified and included on the questionnaire: three strategies for avoiding exposure to problem foods (Items 1, 6, & 8 on Table 2) and three strategies for consuming problem foods but limiting their intake (Items 2, 4, & 7 on Table 2). Two additional strategies were added that had not been examined in the survey but were considered by interventionists to be relatively common (Items 3 & 5 on Table 2). Respondents were asked how often they used each method to control their intake of problem foods, indicating their responses on visual analog scales with the anchors of “Never” and “Always”; the responses were rated from 0 to 100 by the software. Participants completed the Problem Foods Questionnaire at Months 0, 6, and 12 of the trial.

Table 2.

Mean reported frequency1 of using strategies for managing problem foods and correlation with baseline dietary restraint score for 186 women in The Portion-Control Strategies Trial

Item Strategy Month 0 (N=186) Month 6 (N=154) Month 12 (N=143)
Mean SEM Correlation with dietary restraint score2 (Pearson r) Significance of correlation (p value) Mean SEM Mean SEM
1 I avoid keeping problem foods in my house, car, or workplace 55.7 a 2.1 0.31 <0.0001 72.2 * 2.1 72.5 * 2.0
6 I avoid buying my problem foods 54.4 a 2.3 0.30 <0.0001 73.5 * 2.1 69.8 * 2.4
2 I allow myself to eat small amounts of problem foods as a treat or reward 45.1 b 2.2 0.12 0.11 44.5 2.2 52.1 2.3
4 I limit the portion of problem foods that I eat 42.8 b 2.0 0.33 <0.0001 73.4 * 1.9 71.8 * 2.2
5 I eat problem foods less frequently, but I don’t worry about the amount that I eat 39.2 b,c 2.1 − 0.19 0.009 32.4 2.2 29.6 1.9
8 I avoid eating my problem foods 34.3 c 2.1 0.38 <0.0001 55.1 * 2.5 50.7 * 2.4
7 I eat a lower-calorie version of my problem foods instead 33.9 c 2.1 0.36 <0.0001 51.4 * 2.6 48.2 * 2.6
3 I rely on my friends, family, and co-workers to help avoid eating problem foods 9.8 d 1.2 0.14 0.056 16.3 1.9 16.7 1.8
Open in a new tab
1

The frequency of using each strategy was assessed on a visual analog scale with the anchors of “Never” (rated 0) and “Always” (rated 100).

Since mean frequencies did not differ across the three intervention groups, the combined results are presented.

2

The dietary restraint score was from the Three-factor Eating Questionnaire (Stunkard & Messick, 1985).

a,b,c,d

Means at Month 0 with different superscripts are significantly different (adjusted p<0.015).

*

Means at Months 6 or 12 are significantly different from the same strategy at Month 0 (adjusted p<0.0001).

2.5. Statistical analysis

Data from the Problem Foods Questionnaire was a pre-defined secondary outcome of the Portion-Control Strategies Trial (registered at ClinicalTrials.gov as NCT01474759). One pre-planned aim of the present analysis was to characterize the problem foods and management strategies reported by trial participants, and to test the hypotheses that the reported foods and strategies differed across time, across intervention groups, or by participant characteristics. A further pre-planned aim was to extend the findings of the trial by testing the hypothesis that the rate of weight loss was influenced by the frequency of using problem food strategies.

The number of foods reported on the Problem Foods Questionnaire by each participant was analyzed using a linear mixed model with repeated measures. The fixed effects in the model were time point and intervention group; participants were treated as a random effect. For significant effects, the Tukey-Kramer method was used to adjust for multiple pairwise comparisons between means. Participant characteristics were also tested as covariates, to assess their influence on the number of problem foods reported.

To characterize the types of foods most commonly reported on the questionnaire, we included the top three foods listed by each participant at each time point. The inclusion of multiple foods in the analysis allowed for more consistent assessment across time points than using only the first food listed, and 89% of the questionnaires reported at least three foods. The top three foods at each time point were classified into categories based on their nutrient composition and role in the diet (Table 3); in addition, each food was given a binary classification of either sweet or savory. Repeated measures logistic regression was used to examine the distribution of problem food categories across time points and intervention groups, as well as to assess the influence of participant characteristics on the distribution of problem food categories. A Chi-squared statistic was used to test the proportion of participants whose top three problem foods were either all sweet or all savory, compared to the proportion expected by chance.

Table 3.

Categories and frequencies of the top three problem foods listed by 186 women at three time points in The Portion-Control Strategies Trial

Categories & subcategories Category total Subcategory total
N1 %1 N
Sweet baked items 294 21.6%
  Cookies 120
  Cake 67
  Baked items, unspecified 61
  Pastries 38
  Other baked desserts 8
Chips & salty snacks 249 18.3%
  Chips (potato, tortilla, corn) 144
  Salty snacks, unspecified or other 25
  Nuts 23
  Crackers, cheese & crackers 20
  Chips & dip or nachos 19
  Pretzels 11
  Popcorn 7
Starchy side dishes 222 16.3%
  Pasta 80
  Bread 76
  Potatoes 31
  French fries 24
  Rice 6
  Other starchy side dishes 5
Chocolate & candy 190 14.0%
  Chocolate 130
  Candy, unspecified 38
  Candy, non-chocolate 17
  Other sweets 5
Ice cream 153 11.3%
Pizza 86 6.3%
Cheese 46 3.4%
Meat & meat dishes 33 2.4%
Miscellaneous2 86 6.3%
Total 1359 100.0%
Open in a new tab
1

The category totals report the number and percent of problem foods that were in the given category, from among the top three listed by each participant at Months 0, 6, and 12 of the trial. Since the distribution of problem foods among categories did not differ significantly across time points or intervention groups, the combined results are presented.

2

The Miscellaneous category included spreads and dips, beverages, unspecified foods (ethnic food, party food, fast food, fried food), and fruit.

Differences in the frequencies of using the eight strategies for managing problem foods during the trial were examined by a linear mixed model with repeated measures, with all strategies included together in the same model. The fixed effects in the model were strategy type, time point, and intervention group; participants were treated as a random effect. For significant effects, the Tukey-Kramer method was used to adjust for multiple pairwise comparisons between means. The Pearson correlation coefficient was used as a descriptive measure of the linear relationship between usage frequencies and dietary restraint scores, and also between mean usage frequencies for the avoidance and limiting strategies.

The influence of strategy use on the trajectory of weight loss during the trial was assessed using random coefficients models (Littell, 2006), which analyzed all available weight-loss measurements. Time was included as a continuous covariate and also as a random effect, which caused a separate weight-loss trajectory to be modeled for each participant. Trajectories were modeled using polynomial factors of time, with the linear term representing the rate of weight loss and the quadratic term representing the change in the rate of weight loss. For all participants across the entire 12-month trial, the mean linear coefficient was negative, indicating weight loss, and the mean quadratic coefficient was positive, indicating a deceleration of weight loss and a degree of weight regain (Rolls, Roe, James, & Sanchez, 2017). The frequencies of using the problem food strategies at Month 12, as well as their significant interactions with time and intervention group, were all included together as covariates in the random coefficients model, in order to assess their influence on the trajectory of weight loss from baseline to Month 12.

All models were adjusted for participant baseline age and BMI, and also for the number of intervention sessions attended, which strongly influenced the rates of weight loss (Rolls, Roe, James, & Sanchez, 2017). Participant characteristics of scores from the Three-factor Eating Questionnaire and the Beck Depression Inventory were tested as covariates in the linear mixed models. Analyses were performed using SAS software, primarily PROC MIXED (version 9.4, SAS Institute Inc., Cary, North Carolina, USA). Values are reported as means ± SEM and differences were considered significant at p<0.05.

3. Results

3.1. Identification of problem foods

Number of problem foods.

At baseline, participants listed a mean of 6.5±0.2 foods for which they found it difficult to manage the amount they ate. After intervention, the reported number of problem foods decreased [F(2,307)=18.79, p<0.0001] to 5.4±0.2 foods at Month 6 and 5.2±0.2 at Month 12. Across all three time points, the allowed maximum of ten problem foods was listed on 24% of the questionnaires. The reported number of problem foods was positively related to the participants’ number of recent weight-loss attempts [F(1,162)=6.40, p=0.012; β=0.17] and to baseline scores on the Three-factor Eating Questionnaire for dietary restraint [F(1,176)=4.32, p=0.039; β=0.10] and hunger tendency [F(1,175)=6.63, p=0.011; β=0.13]. There was no significant relation between the reported number of problem foods and participant age, baseline BMI, baseline depression score, or intervention group.

Categories of problem foods.

The categories of reported problem foods are shown in Table 3. Overall, 65% of the top three problem foods were in the high-energy-dense discretionary food categories of sweet baked items, salty snacks, chocolate and candy, and ice cream. The non-discretionary food categories of starchy side dishes, pizza, cheese, and meat accounted for a further 28% of the top problem foods. The individual foods most commonly named were ice cream, potato chips (UK: crisps), chocolate, cookies (UK: biscuits), and pizza (Table 3). The distribution of problem foods among the categories did not differ across the three time points of the trial [F(12,1291)=0.85, p=0.60], nor across trial intervention groups [F(12,637)=0.36, p=0.98]. Across all participants and time points, the top three problem foods were approximately evenly divided between savory foods (52%) and sweet foods (48%). Within participants, however, an evaluation of the clustering of sweet and savory problem foods showed that the top three foods were more likely to be either all savory (17.2% of participants) or all sweet (17.0%) than the proportion expected by chance (12.5% each; p=0.024).

Influence of participant characteristics on food categories.

The distribution of the top three problem foods across categories was not significantly influenced by the participant characteristics of age, baseline BMI, number of recent weight-loss attempts, or baseline scores on the Three-factor Eating Questionnaire. In contrast, the baseline score for severity of depressive symptoms was related to the categories of problem foods, even though all participants had scores below the value indicating severe depression. Across the trial, participants with higher scores listed a greater number of starchy side dishes and pizza among their top three problem foods than individuals with lower scores [F(6,487)=3.14, p= 0.005]. For example, participants with a depression score in the highest tertile (score 7–25), identified 32% of their top three problem foods from these categories, whereas those with depression scores in the lowest tertile (score 0–2), identified 15% of their problem foods from these categories.

3.2. Use of strategies for managing problem foods

As shown in Table 2, there were significant differences in the reported frequency of using various strategies to control intake of problem foods prior to intervention [F(7,1272)=62.18, p<0.0001]. The baseline practices most frequently reported for problem foods were avoiding keeping them available (in the house, car, or workplace) and avoiding buying them; the next most frequent were eating a small amount as a treat and limiting their portion size. The least frequently used strategy for problem foods was relying on social support. At baseline, the frequency of using five of the strategies was positively correlated with the dietary restraint score and one further strategy was negatively correlated (Table 2). The baseline disinhibition score was significantly correlated with the frequency of only one strategy: avoiding buying problem foods (r=0.16; p=0.03).

The reported frequency of using problem food strategies changed across the trial (Table 2; [F(2,3737)=102.33, p<0.0001]). For five of the eight practices, their reported use increased from baseline to Month 6 (p<0.001) and the increase was maintained at Month 12 (p<0.001); these strategies were the same as those that were positively correlated with dietary restraint at baseline. By the end of the trial, the strategy of limiting the portion of problem foods was reportedly used as frequently as the most common avoidance strategies. The frequency of using the problem food strategies did not differ across intervention groups [F(2,182)=0.65, p=0.52].

Overall, the mean usage frequency for the three avoidance strategies and the mean for the three limiting strategies had a positive correlation with each other, ranging from 0.24 to 0.30 over time (all p<0.002). This suggests that participants did not use the avoidance and limiting strategies as alternatives, but as allied approaches.

3.3. Relationship of problem food strategies to weight loss

For one of the eight strategies for managing problem foods, the frequency of use at Month 12 was related to the trajectory of weight loss over the trial in all intervention groups; three other strategies were related to weight loss in only one group, as described below. The participant dietary restraint score at Month 12 had a small additional positive effect on the magnitude of weight loss (0.07±0.03 kg) when included in the model with the strategy frequencies [F(1,142)=5.71, p=0.018]. The depression score had no significant effect on weight loss in the model [F(1,141)=0.17, p=0.68].

The strategy of limiting portions of problem foods was strongly related to the trajectory of weight loss in all intervention groups (Figure 1). At the end of the trial, women who reported more frequently limiting portions of their problem foods had a greater rate of weight loss than those who used this strategy less frequently [F(2,186)=11.23, p<0.0001]. For example, women whose frequency of using this strategy was below the median lost 3.8±0.70 kg at the end of the trial, whereas women whose use was above the median lost 7.2±0.81 kg. In addition, among the women who limited portions less frequently, the pattern of weight loss differed in the Pre-portioned Foods Group (who were instructed to structure their meals around single-serving items). Participants who less frequently limited the portions of their problem foods in this group had a greater initial rate of weight loss than those in the other two groups, but subsequently regained weight at a greater rate and ended the trial with no difference from the other two groups (Figure 1; [F(2,1020)=31.30, p<0.0001]). This pattern paralleled the decreased use of pre-portioned foods in this group later in the trial, as previously reported (Rolls, Roe, James, & Sanchez, 2017).

Figure 1:

Figure 1:

Open in a new tab

The mean (±SEM) rate of weight loss in 186 women in The Portion-Control Strategies Trial was greater for those who at Month 12 reported more frequently limiting portions of their problem foods (p<0.0001). In addition, participants who infrequently limited portions of problem foods and were in the Pre-portioned Foods Group had a greater initial rate of weight loss, but subsequently regained weight at a greater rate (p<0.0001). The figure illustrates these relationships by categorizing participants as having either lower or higher frequency of using this strategy (based on the median rating), but the random coefficients model tested the effect of the continuous frequency rating.

The other problem foods strategies were related to weight loss only among participants in the Standard Advice Group (who were instructed to eat less of all foods). At the end of the trial, participants in this group who reported frequently substituting lower-calorie versions of their problem foods for higher-calorie versions had a greater rate of weight loss than participants in this group who used this strategy infrequently (Figure 2) [F(2,1003)=4.64, p=0.01]. Two other strategies were inversely related to weight loss in the Standard Advice Group (Figure 3). There were significantly lower rates of weight loss among participants in this group who reported a higher use of eating problem foods less often but not worrying about the amount (F(2,1010)=6.76, p=0.001) and in those who more frequently relied on social support to avoid eating problem foods (F(2,1023)=5.55, p=0.004), compared to participants in the Standard Advice Group who used these strategies less frequently.

Figure 2:

Figure 2:

Open in a new tab

Among participants in the Standard Advice Group, the mean (±SEM) rate of weight loss was greater for women who reported frequently substituting lower-calorie versions of their problem foods for higher-calorie versions than for participants in this group who used this strategy infrequently (p=0.01). The figure illustrates this relationship by categorizing participants as having either lower or higher frequency of using this strategy (based on the median rating), but the random coefficients model tested the effect of the continuous frequency rating.

Figure 3:

Figure 3:

Figure 3:

Open in a new tab

Among participants in the Standard Advice Group, the mean (±SEM) rate of weight loss was inversely related to weight loss for two problem food strategies: (A) eating problem foods less often but not worrying about the amount (p=0.001) and (B) relying on social support to avoid eating problem foods (p=0.004). The figure illustrates these relationships by categorizing participants as having either lower or higher frequency of using the strategy (based on the median rating), but the random coefficients model tested the effects of the continuous frequency ratings.

Although the three avoidance strategies were reported to be commonly used, after accounting for the other strategies none of them was related to weight loss: avoiding keeping problem foods available [F(1,141)=0.00, p=0.97], avoiding buying them [F(1,141)=1.31, p=0.25], or avoiding eating them [F(1,142)=1.50, p=0.22]. Thus, this analysis found that the strategy of limiting the portions of problem foods was the most strongly related to the trajectory of weight loss during the trial, and that use of the avoidance strategies was not related to weight loss.

Discussion

By assessing the experience of women with overweight and obesity in The Portion-Control Strategies Trial, this analysis enhanced understanding of specific foods that are a practical problem for weight management. The types of foods identified as the most problematic remained stable from baseline and across the year of behavioral intervention; the majority (65%) were in the categories of high-energy-dense desserts and snacks, with a substantial minority (28%) among non-discretionary foods such as pasta, pizza, bread, and potatoes. The only reported management strategy that was related to weight loss in all intervention groups was limiting the portion sizes of problem foods. Three strategies for avoiding consumption of problem foods were commonly used but were not found to be related to weight loss. Although effective weight loss strategies are likely to differ across individuals, these results suggest that for many individuals, adopting and maintaining strategies to manage portions of their particular problem foods, rather than trying to avoid them, is a more useful approach for weight loss.

The distribution of the top three problem foods across food categories remained consistent over the year-long intervention trial. Unsurprisingly, about two-thirds of problem foods were in the discretionary categories of desserts (sweet baked items, chocolate & candy, ice cream) and salty snacks, which are frequently reported to be the most palatable foods (Drewnowski & Almiron-Roig, 2010; Fazzino, Rohde, & Sullivan, 2019). It is notable that the only non-discretionary foods frequently cited as problematic were those high in starchy carbohydrates (pasta, bread); conversely, foods such as cheese, meat, and fruit were infrequently listed. In a previous study, when 398 women in obesity treatment listed their ten favorite foods, over 50% named at least one food in the categories of baked desserts, bread & crackers, and ice cream; the next most common favorites were non-discretionary (beef, pasta, cheese, poultry), followed by chocolate (Drewnowski, 1992). In contrast, when asked about craving, one of the most commonly named foods among women is chocolate (Rozin 1991). For example, a survey of craved foods among 758 female undergraduates found that by far the most commonly craved food was chocolate (39%), followed by salty snacks, pizza, ice cream, and sweets (6–8% each; Weingarten, 1991). These findings highlight the distinctions between foods that are difficult to stop eating (and thus problematic for weight loss) and those identified as favorite or craved (which may not be identified as the most challenging for weight loss). For the purposes of weight management, it is important to address the foods that cause the most practical problems for behavioral control, which individuals managing their weight are readily able to name.

Although the types of foods identified as problematic were consistent across participant characteristics of age, body size, and dieting history, there were differences found according to reported symptoms of depression. None of the participants were severely depressed, but those who reported more intense symptoms identified a greater number of their top problem foods as starchy side dishes and pizza. This contrasts with a study of 4655 middle-aged women, in which depressive symptoms were positively associated with greater consumption of sweet foods but negatively associated with high-calorie non-sweet foods, which were mostly high in starch and fat (Jeffery 2009). Frequently, however, studies of “carbohydrate-craving” and mood have not distinguished between sweet and starchy carbohydrates (Corsica 2008; White 2002; Wurtman 1995). An added complication for understanding these relationships is that in many studies, the foods categorized as high in carbohydrates are frequently also high in fat (Markus 2017; Drewnowski 1992). For participants in the current study with more intense depressive symptoms, the problematic nature of starchy non-discretionary foods may reflect the complex interactions between depression, carbohydrate-rich foods, and obesity.

During the weight-loss trial, participants increased their reported use of five strategies for managing problem foods: three avoidance strategies (not keeping them available, buying them, or eating them) and two limiting strategies (reducing portions and substituting lower-calorie versions). Similar practices were reported to be used often or very often by 36–44% of a U.S. sample of 1165 U.S. adults who were trying to manage their weight (Sciamanna 2011). Although the interventions in the present trial focused on portion control for the entire diet, the strategies that participants reported using particularly for problem foods were more diverse. Participants concurrently used practices for avoiding exposure to problem foods along with those for consuming such foods while limiting their intake; thus, these were used as complementary methods rather than alternatives. A similar result was observed in the U.S. survey; the authors noted that for weight control, “many alternative solutions typically are implemented at the same time, which makes it challenging [for individuals] to measure the impact of each” (Sciamanna 2011). In the current study it was notable that over time, participants did not increase their use of three problem food strategies: allowing some as a treat, reducing frequency but not amount, and relying on social support. These were also the three strategies that were uncorrelated, or negatively correlated, with dietary restraint scores. These findings suggest that the strategies for problem foods that participants found feasible and sustainable were those aimed at controlling the amount of exposure to the foods, rather than less specific approaches.

The present investigation provides some of the first findings on the effects of problem food strategies on weight management. There has been limited research examining foods that are problematic for overconsumption, and much of the related research has investigated foods that are identified as craved. Most studies find that food cravings decrease during energy restriction for weight loss (Kahathuduwa 2017, Apolzan 2017) and that improvement in control of cravings is associated with greater long-term weight loss (Dalton 2017; Buscemi 2017). Although not all of the foods that individuals find challenging can be categorized as craved, as they do not evoke an intense and specific desire, these foods still present practical difficulties for managing energy intake. Other previous investigations have focused on strategies for reducing consumption of discretionary foods. In a scoping review of such strategies, the key findings for decreasing energy intake were that limiting portion size was beneficial in the short term and that substitution with other foods may be effective (Grieger 2016). Several surveys of individuals who maintained long-term weight loss indicated that among the most widely used dietary practices were limiting intake of certain types or classes of foods and controlling portions of all foods (Klem 1997; Sciamanna 2011). These surveys, however, did not compare the effect of these practices on the magnitude of weight loss.

In the present trial, not all of the strategies that participants reported using for problem food management were found to be associated with weight loss. Among women in all intervention groups, only the practice of frequently limiting the portions of problem foods was related to increased weight loss. In addition, participants who less frequently limited these portions but were in the Pre-portioned Foods Group had greater weight loss early in the intervention. These results indicate the efficacy of directing portion-control efforts at the most problematic foods; for individuals who have difficulty doing this on their own, use of pre-portioned items in the daily diet can provide additional support. In the Standard Advice Group alone, the finding that weight loss was inversely related to two of the problem food strategies (eating less frequently but not worrying about the amount, and relying on social support) implies that these strategies were less effective in the absence of food-specific portion-control practices. Although participants in all intervention groups reported that they regularly used several strategies for avoiding exposure to problem foods, these behaviors were not found to be related to weight loss. It is likely that the effectiveness of avoidance strategies varies across individuals; these practices may be difficult to maintain in shared homes and in the face of ubiquitous exposure in the current eating environment.

This analysis of experience with problem foods among individuals managing their weight fills a gap in the research by providing food-specific information on problem foods and on the behavioral practices used to manage them. Information on problem foods was consistently evaluated by women before and during weight-loss intervention, and the foods were self-identified, rather than generated from a predefined list. Although the characteristics of the trial population are typical of many individuals managing their weight, the generalizability of the results is limited among other populations, such as men and those with lower levels of education. The use of problem food strategies was self-reported and assessed by a questionnaire that was not validated against another assessment method, and these characteristics of the questionnaire may constrain the interpretation of the results. Because the trial interventions were focused on portion control, it is possible that demand characteristics influenced the reporting of portion-control practices compared to other strategies for problem foods. This possibility is not supported, however, by the finding that portion-control practices were reported to be used less frequently than avoidance strategies. Furthermore, demand characteristics do not explain the greater rate of weight loss among the participants in the Pre-portioned Foods Group who reported less frequently limiting portions of their problem foods.

This study used data from the Portion-Control Strategies Trial to characterize the foods that are problematic for individuals managing their weight and to investigate the potential of behavioral strategies to facilitate weight management. The majority of problem foods identified were in the expected categories of high-energy-dense sweets, baked goods, and salty snacks; the sizeable minority of problematic foods among non-discretionary starchy carbohydrates deserves further investigation. Participants used multiple strategies to deal with their problem foods and maintained this use over time, but analysis showed that not all of these practices were related to weight loss. Although effective weight loss strategies are likely to differ across individuals, the results of this study suggest that for many individuals, adopting and maintaining strategies to manage portions of their particular problem foods, rather than trying to avoid them, is a more useful approach for weight loss.

Supplementary Material

1

Highlights.

  • Individuals are often faced with problem foods that are difficult to resist eating

  • During a one-year weight-loss trial, 186 women reported their top problem foods

  • Women also reported the frequency of using management strategies for problem foods

  • Limiting portions of problem foods was related to a greater rate of weight loss

  • Strategies for avoiding problem foods were not related to weight loss

Acknowledgements:

We thank all of the participants of The Portion-Control Strategies Trial for their time and commitment. We also thank the research team who administered the trial, provided the interventions, collected the data, and helped to interpret the results, particularly Christine Sanchez, Cara Meehan, and Alyssa Spaw.

Funding: This work was supported by the National Institutes of Health [grant number R01 DK059853]. The funding source had no involvement in the design, conduct, or interpretation of the research.

Participants gave signed informed consent and received financial compensation for their time in the trial. The Office for Research Protections at The Pennsylvania State University reviewed and approved the trial protocol.

Footnotes

Declaration of interest: The authors declare no competing interests regarding this research.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Apolzan JW, Myers CA, Champagne CM, Beyl RA, Raynor HA, Anton SA, Williamson DA, Sacks FM, Bray GA, Martin CK. Frequency of consuming foods predicts changes in craving for those foods during weight loss: the POUNDS lost study. Obesity 2017; 25: 1343–1348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Appelhans BM, French SA, Pagoto SL, Sherwood NE. Managing temptation in obesity treatment: A neurobehavioral model of intervention strategies. Appetite 2016; 96: 268–279. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Beck AT, Steer RA, Brown GK. Manual for the Beck Depression Inventory-II. Psychological Corporation: San Antonio, TX, USA, 1996. [Google Scholar]
  4. Buscemi J, Rybak TM, Berlin KS, Murphy JG, Raynor HA. Impact of food craving and calorie intake on body mass index (BMI) changes during an 18-month behavioral weight loss trial. Journal of Behavioral Medicine 2017; 40: 565–573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cepeda-Benito A, Gleaves DH, Williams, TL, Erath SA. The development and validation of the State and Trait Food-Cravings Questionnaire. Behavior Therapy 2000; 31:151–173. [DOI] [PubMed] [Google Scholar]
  6. Corsica AJ, Spring BJ. Carbohydrate craving: A double-blind, placebo-controlled test of the self-medication hypothesis. Eating Behaviors 2008; 9:447–454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Dalton M, Finlayson G, Walsh B, Halseth AE, Duarte C, Blundell JE. Early improvement in food cravings are associated with long-term weight loss success in a large clinical sample. International Journal of Obesity 2017; 41:1232–1236. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Drewnowski A, Kurth C, Holden-Wiltse J, Saari J. Food preferences in human obesity: carbohydrates versus fats. Appetite 1992; 18: 207–221. [DOI] [PubMed] [Google Scholar]
  9. Drewnowski A, Almiron-Roig E. Human Perceptions and Preferences for Fat-Rich Foods In: Montmayeur JP, le Coutre J, editors. Fat Detection: Taste, Texture, and Post Ingestive Effects. Boca Raton (FL): CRC Press/Taylor & Francis; 2010. [PubMed] [Google Scholar]
  10. Fazzino TL, Rohde K, Sullivan DK. Hyper-palatable foods: Development of a quantitative definition and application to the US food system database. Obesity 2019; 27: 1761–1768. [DOI] [PubMed] [Google Scholar]
  11. Garner DM, Olmsted MP, Bohr Y, Garfinkel PE. The Eating Attitudes Test: psychometric features and clinical correlates. Psychological Medicine 1982; 12: 871–878. [DOI] [PubMed] [Google Scholar]
  12. Grieger JA, Wycherley TP, Johnson BJ, Golley RK. Discrete strategies to reduce intake of discretionary food choices: a scoping review. International Journal of Behavioral Nutrition and Physical Activity 2016; 13: 57. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hill AJ. The psychology of food craving. Proceedings of the Nutrition Society 2007; 66(2):277–85. [DOI] [PubMed] [Google Scholar]
  14. Institute of Medicine, Food and Nutrition Board. Dietary Reference Intakes for Energy, Carbohydrates, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids. The National Academies Press: Washington, DC, USA, 2002. [Google Scholar]
  15. Kahathuduwa CN, Binks M, Martin CK, Dawson JA. Extended calorie restriction suppresses overall and specific food cravings: a systematic review and a meta-analysis. Obesity Reviews 2017; 18:1122–1135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Klem ML, Wing RR, McGuire MT, Seagle HM, Hill JO. A descriptive study of individuals successful at long-term maintenance of substantial weight loss. American Journal of Clinical Nutrition 1997; 66:239–246. [DOI] [PubMed] [Google Scholar]
  17. Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O. SAS for Mixed Models. 2nd ed SAS Institute, Inc.: Cary, NC, USA, 2006. [Google Scholar]
  18. Markus CR, Rogers PJ, Brouns F, Schepers R. Eating dependence and weight gain: no human evidence for a ‘sugar-addiction’ model of overweight. Appetite 2017; 114:64–72. [DOI] [PubMed] [Google Scholar]
  19. National Health and Medical Research Council. Eat for Health. Australian Dietary Guidelines. Canberra: National Health and Medical Research Council, 2013. Available at: https://www.eatforhealth.gov.au/sites/default/files/files/the_guidelines/n55_australian_dietary_guidelines.pdf. Accessed 24 Feb 2020. [Google Scholar]
  20. National Health Service. Health Scotland. Eatwell Guide: how to eat a healthy balanced diet. Available at: https://www.nhsinform.scot/healthy-living/food-and-nutrition/eating-well/eatwell-guide-how-to-eat-a-healthy-balanced-diet#foods-high-in-fat-salt-or-sugar. Accessed 24 Feb 2020.
  21. Rodin J, Mancuso J, Granger J, Nelbach E. Food cravings in relation to body mass index, restraint and estradiol levels: a repeated measures study in healthy women. Appetite 1991; 17:177–185. [DOI] [PubMed] [Google Scholar]
  22. Rolls BJ, Roe LS, James BL, Sanchez CE. Does the incorporation of portion-control strategies in a behavioral program improve weight loss in a 1-year randomized controlled trial? International Journal of Obesity. 2017; 41, 434–442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Rozin P, Levine E, Stoess C. Chocolate craving and liking. Appetite 1991; 17:199–212. [DOI] [PubMed] [Google Scholar]
  24. Sciamanna CN, Kiernan M, Rolls BJ, Boan J, Stuckey H, Kephart D, et al. Practices associated with weight loss versus weight-loss maintenance results of a national survey. American Journal of Preventive Medicine 2011; 41:159–166. [DOI] [PubMed] [Google Scholar]
  25. Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. Journal of Psychosomatic Research 1985; 29: 71–83. [DOI] [PubMed] [Google Scholar]
  26. Taylor M. A review of food craving measures. Eating Behaviors 2019; 32:101–110. [DOI] [PubMed] [Google Scholar]
  27. U.S. Department of Health and Human Services and U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 8th Edition. December 2015. Available at: https://health.gov/dietaryguidelines/2015/guidelines/. Accessed 24 Feb 2020. [Google Scholar]
  28. Weingarten HP, Elston D. Food cravings in a college population. Appetite 1991; 17:167–175. [DOI] [PubMed] [Google Scholar]
  29. White MA, Whisenhunt BL, Williamson DA, Greenway FL, Netemeyer RG. Development and validation of the food-craving inventory. Obesity Research 2002; 10:107–114. [DOI] [PubMed] [Google Scholar]
  30. Wurtman RJ, Wurtman JJ. Brain serotonin, carbohydrate-craving, obesity and depression. Obesity Research 1995; 3: 477S–480S. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1582912-supplement-1.pdf (106.9KB, pdf)

RESOURCES