Psychosocial Correlates of Weight Maintenance Among Black & White Adults

Abstract

Objectives

To investigate (1) weight maintenance among black and white participants and (2) psychosocial correlates (eg, healthy eating barriers, self-efficacy, stress) of weight maintenance 18 months after behavioral weight-loss treatment.

Methods

Linear and logistic regression examined weight change and unsuccessful weight maintenance (>5% weight gain) among 107 black and white adults.

Results

After controlling for socio-demographics, differences in weight maintenance between ethnicities were not generally noted. Healthy eating barriers and stressful life events were associated with weight gain, P<.04.

Conclusions

Strategies to cope with stressful events and overcome barriers to eating healthfully are needed for weight maintenance among both ethnicities.

Obesity and overweight are widespread health problems nationally and globally.¹ The latest prevalence of US adults age 20 and older are either overweight (body mass index [BMI] between 25.0 and 29.9 kg/m²) or obese (BMI of 30 kg/m² or more).² Recent indicators suggest that as many as 43% of US adults will be obese by 2018, and $344 billion or 21% of US direct health care dollars are expected to be spent on obesity-related illnesses.³ One key to successful weight management is maintaining one’s weight after weight-loss treatment.

Although weight-loss programs in the last 20 years have succeeded in promoting weight reduction among those seeking to lose weight, maintenance of lost weight has remained a significant challenge requiring long-term adherence to lifestyle and behavior change.⁴ Frequently, about one third of lost weight is regained within the first year after weight loss,⁵ and the average weight loss maintained 4 years after treatment is 1.8 kg or approximately 4 lbs.⁶ These disappointing rates of weight-loss maintenance remain a quandary to clinicians, researchers, and those who struggle with obesity.

In addition, the obesity epidemic affects a greater proportion of ethnic minority groups like black persons, who are unduly burdened by the health problems associated with obesity, eg, hypertension, cardiovascular disease, and type 2 diabetes.⁷ The overall prevalence of obesity for non-Hispanic blacks was 44.1% in 2007–2008 compared to 32.8% for non-Hispanic whites, and 49% of black women are obese as opposed to 33% of white women.² Weight gain over one’s lifetime is common to many individuals; however, the magnitude of gain is greater in black women. Sheehan and colleagues⁸ found that over a 20-year period, white women aged 25 to 35 gained 7.7 kg whereas black women the same age gained nearly 11 kg. Weight reduction and maintenance of lost weight are key factors in decreasing the disproportionate burden of disease in ethnic minority groups.⁹

Very few studies have explored weight maintenance among ethnic minorities¹⁰; however, some differences in weight management have been reported. Some researchers found that although black individuals tended to achieve less weight loss during treatment, the amount of weight regain they experienced after the loss was either the same as or less than that of white individuals.⁹ A population-based study examining 536 adults found that African Americans were more likely to maintain 75% of lost weight at the 5-year follow-up (odds ratio [OR]=1.7, 95% confidence interval [CI]= 1.1–3.0, P=.03) compared to their white counterparts, although this study did not assess whether initial weight loss was voluntary or involuntary.¹¹ Others have noted that black participants regained weight more rapidly.¹² Among 23 black women, diminished responsiveness to physical hunger and environmental cues for eating were marginally related to weight-loss maintenance (P=.057).¹³ Additional research is needed to investigate this important aspect of weight management among black individuals.

Studies have focused on consuming a low-fat diet as a means of promoting weight maintenance,^14,15 but continuing to follow a diet low in fat may prove difficult long term, and some participants have reported feeling deprived when eating a low-fat diet.¹⁶ Yet, the development of a distaste for fat has also been reported in those who are adherent to a low-fat diet.¹⁶ The role of individuals’ experiences with a low-fat diet has been largely unexplored in weight maintenance, especially among black populations.

Barriers to healthy eating that may affect weight maintenance include concerns about taste, insufficient time and/or motivation, and the belief that healthy foods are more costly.¹⁷ Specific barriers identified for black women include the cost of more healthful foods,¹⁸ the importance of traditional preparation of cultural dishes, and family members’ eating expectations.¹⁹ Barriers to healthy eating also need further examination to determine their full effect on weight maintenance.

Other factors like self-efficacy,²⁰ social support,²¹ and the ability to deal with life stress²² have been associated with successful weight maintenance. Although increases in self-efficacy for resisting eating have been linked to weight loss,²³ the role of self-efficacy after active weight-loss treatment is less clear.²⁰ Regarding social support, some have found a significant influence on weight loss and maintenance for participants recruited with friends, but the findings were confounded by the addition of financial incentives for this group.²¹ A study of 36 black women found that over 50% endorsed stress as a factor that negatively influenced weight management.²⁴ Yet, the impact of these psychosocial factors on long-term weight maintenance is not well documented.

Because weight maintenance is a complex multifactorial problem, this study was guided by social cognitive theory, wherein personal, environmental, and behavioral elements all function as interacting determinants and influences of each other.²⁵ Using that framework, personal factors (ethnicity, self-efficacy, experiences associated with following a low-fat diet) and environmental factors (barriers to healthy eating, social support, stress) were examined in relation to the outcome of weight maintenance after behavioral weight-loss treatment.

The elements of successful long-term weight maintenance after a loss have not been clearly elucidated,²⁶ particularly among black persons who are affected by obesity in greater numbers.²⁷ Therefore, the specific aims of this study were to (1) explore possible differences in percent weight change and successful weight maintenance (≤5% weight gain) between black and white participants 18 months after a behavioral weight-loss trial; (2) examine experiences associated with following a low-fat diet, barriers to healthy eating, self-efficacy for resisting the desire to eat and for exercising, social support, and stress as correlates of weight maintenance; and (3) investigate ethnicity as a moderator of the relationship between these psychosocial variables and weight maintenance.

METHODS

Design and Study Sample

This ancillary study was conducted 18 months after the completion of a randomized clinical trial of behavioral weight-loss treatment, termed PREFER. In PREFER, 3 cohorts of participants were recruited from the community for an 18-month weight-loss trial that used standard behavioral therapy for weight loss; 176 adults were first randomly assigned to 1 of 2 preference conditions (dietary preference-yes or dietary preference-no). If a participant was randomly assigned to dietary preference-no, he or she was then randomly assigned to 1 of the 2 diet conditions (a standard, reduced-calorie and reduced-fat diet or a lacto-ovo-vegetarian, reduced-calorie and reduced-fat diet). If a participant was randomly assigned to dietary preference-yes, the individual was then assigned to the diet condition that he or she indicated was preferred at the screening session. Standard behavioral therapy for weight loss was used during the trial’s 12 months of active treatment. During these 12 months, participants attended 32 group sessions (weekly for 6 months, biweekly for 3 months, and monthly for 3 months), which focused on the key elements of cognitive behavioral weight-loss treatment, eg, self-monitoring eating and activity behaviors, goal setting, stimulus control, cognitive restructuring to avoid negative or all-or-nothing thinking, problem solving, behavioral skills development, and relapse prevention.²⁸ Female participants were given a daily calorie goal of 1200 or 1500 calories/day; daily calorie goals for men were 1500 or 1800, and everyone was to consume ≤25% of calories from fat. The physical activity instruction was to achieve at least 150 minutes of moderate-intensity activity like brisk walking each week. After 12 months of active treatment, participants began a 6-month maintenance phase with no contact from the staff, and the trial ended at 18 months. Details of the design and findings from the PREFER trial have been previously published.^28,29 The mean weight loss was 4–8% for the 4 treatment groups with no significant difference in weight loss between the 2 types of diets; the dietary preference-no group lost more weight than did the dietary preference-yes group.²⁹

Everyone who completed the weight-loss trial was eligible to participate in this ancillary study, which consisted of one assessment at 18 months after the trial ended. We contacted participants by mail asking them to return for a follow-up study, and after 2 weeks of no response, individuals were contacted by phone to request that they participate. During the weight-loss trial, participants were not informed that any type of follow-up study would occur and were not contacted prior to the letter requesting that they take part in the ancillary study. Inclusion criteria, based on the eligibility criteria for the weight-loss trial, were as follows: age 18–55 years, body mass index between 27 and 43 kg/m² inclusively, agreement to random assignment to receive their dietary preference or not and one of the 2 diets, and adequate completion of a 5-day food diary. Additionally, participants also had to have completed the final assessment of the weight-loss trial, which served as the baseline measurement for this ancillary study. Between February 2006 and April 2007, 119 of the 132 persons (90%) who completed the weight-loss trial participated. Two pregnant participants and one undergoing treatment for an eating disorder were excluded. Nine persons who self-selected a race/ethnicity other than black or African American or white were also excluded from the analysis. Therefore, the total sample (N = 107) included 81 white and 26 black participants. The study was approved by the Institutional Review Board at the University of Pittsburgh, and written informed consent was obtained from all participants.

Measures

Dependent variable (weight change)

Weight was measured in pounds with participants wearing light clothing and no shoes. A self-reported weight was obtained from one out-of-town participant, as self-reported weights have been shown to be valid.³⁰ In order to be conservative and account for possible underreporting, a 2-kilogram correction³¹ was added to this self-reported weight.

Independent variables

We measured independent variables using the following scales: Experiences Associated with Following a Low-fat Diet Scale (ELF),¹⁶ Barriers to Healthy Eating Scale (BHE),³² Weight Efficacy Lifestyle Questionnaire (WEL),³³ and Self-Efficacy for Exercise scale (SEE).³⁴ Social support was assessed using a composite score of 3 items from the BHE and 4 items from the ELF that measured social support for healthful eating, as determined by factor analysis.^32,35 Stress was assessed from 4 individual items in a survey developed for this ancillary study. The ELF, BHE, and WEL were assessed in both the weight-loss trial and the ancillary study, whereas the SEE and the stress items were assessed only in the ancillary study.

Developed and used in the Women’s Health Trial,¹⁶ the ELF is a 25-item scale measuring experiences believed to be associated with low-fat dietary maintenance: wellness (feeling healthier while on the diet), distaste (for fat), cost (time and money), inconvenience (adhering to the diet when not eating at home), deprivation (denied desired foods), and family (insufficient support from family). It has a 5-point scale (1 = strongly disagree to 5 = strongly agree) and was validated during the Women’s Health Trial (r = .26 to .76). Higher scores represent a more positive experience. Internal consistency for this ancillary study, estimated using Cronbach alpha, was .81, similar to estimates reported by others.³⁵

The BHE is a 22-item scale in which participants rate various circumstances related to following the healthy eating plan (emotions, daily mechanics of following the eating plan, social support) on a scale of 1 (no problem) to 5 (very important problem). Example items include “I have trouble estimating appropriate portion sizes” or “When I am very hungry, I have trouble controlling what I eat.” Lower scores indicate fewer barriers. Psychometric testing revealed a Cronbach alpha of .86 previously³² and an alpha of .89 for this study.

The WEL questionnaire is a 20-item measure to assess self-efficacy for weight management.³³ This scale assesses one’s confidence in the ability to resist the desire to eat in different situations on a scale of 0 to 9, such as “I can resist eating when I am watching TV.” Higher scores indicate higher confidence. Psychometric properties of the WEL are well established with Cronbach alpha coefficients ranging from .70 to .90.³³ Cronbach alpha for this study was .93. The validity and reliability of this instrument have also been established in African American women.³⁶

The SEE scale is a 9-item self-efficacy measure for exercise that asks persons to rate their confidence in the ability to exercise 3 times per week for 20 minutes given a variety of circumstances, eg, being busy with other activities or feeling tired.³⁴ The scale range is 0 (not confident) to 10 (very confident), and higher scores indicate greater confidence. Good internal consistency has been reported with a Cronbach alpha of .92,³⁴ similar to this study’s alpha of .93.

Three items from the BHE and 4 items from the ELF questionnaires, determined to assess social support via factor analysis,^32,35 were combined as a measure of the participants’ perception of their family’s and friends’ support for following a healthful eating plan. Cronbach alpha for the composite, 7-item social support measure was calculated at .78, indicating satisfactory internal consistency.

Because an established scale measuring the specific effect of perceived stress on eating behaviors and weight-loss maintenance was not available, and we were interested in the impact of stress on weight management and not a global measure of stress in general, stress was assessed from the following 4 survey items:

• (1)“On a scale of 0% to 100% of the time, how often does stress influence how you eat?”
• (2)“On a scale of 0% to 100% of the time, how often do you use techniques to lower your stress level (deep breathing, journal writing, exercise, relaxing hobbies, time management, etc)?”
• (3)“Have you had a major stressful event in the last 18 months (such as marriage, new job, divorce, death in the family)? If yes, please describe.”
• (3a)“If you experienced a stressful event, how much did this stressful event affect your eating habits on a scale of 0 (no effect) to 10 (most effect)?”

Statistical Analyses

Data from the final assessment visit of the weight-loss trial and from this ancillary 18-month follow-up assessment were used to calculate percent change scores for all independent variables (ELF, BHE, WEL, social support) and the continuous dependent variable (% weight change), except for the SEE and stress items, as these variables were measured only in the ancillary study. Percent change scores were calculated by subtracting 18-month follow-up values from the baseline values, standardized by the baseline values and expressed as a percentage.

Weight maintenance was analyzed as a continuous outcome (% weight change) and binary outcome (≤5% weight gain = successful maintenance; >5% weight gain = unsuccessful maintenance), based on weight change from the end of the weight-loss trial to the ancillary assessment, using multivariate hierarchical linear and logistic regression, respectively. Exploratory data analyses revealed < 3% missing data, which were missing at random. Due to the small amount of missing data, participants with missing data were excluded from models containing variables in which data were missing. Omission diagnostics and sensitivity analyses were used to examine outlying values; 3 overly influential outliers with increases of more than 3 standard deviations above the mean on the BHE and WEL were excluded from models with these variables. Baseline differences between blacks and whites and between those who participated in the current study and those who did not were assessed using chi-square tests of independence or Fisher’s exact tests and t-tests or Mann Whitney U tests, depending on the variable’s distribution. Statistical significance was set at P<.05. SPSS (version 17.0, SPSS Inc., Chicago, IL) was used for the analysis. Measures of central tendency are presented as means and standard deviations.

Significant univariate correlates of weight maintenance were included in multivariate models; self-efficacy for exercise was the only variable that was not univariately associated with weight maintenance and therefore was not examined multivariately. All multivariate models were built hierarchically controlling for age, gender, education, income, and marital status in the first block. To examine ethnic differences in weight maintenance (specific aim 1), an indicator-coded ethnicity variable was added in the second block. For specific aim 2, the ELF, BHE, WEL, social support, and stress scores were entered in the second block; ethnicity was then included in the third block to determine its effect on weight maintenance after controlling for other variables. Separate models were built to examine the derived social support variable, based on ELF and BHE items, in order to avoid multicollinearity with the ELF and BHE scales. Because each stress item assessed different factors associated with the impact of stress on eating (eg, environmental vs behavioral), each item was examined in a separate model using the total sample for items 1 and 2 and the subsample reporting a stressful life event (n = 69) for item 3a. For specific aim 3, interaction terms were included in the fourth block of models to assess for possible effect modification, ie moderation, due to ethnic group.

RESULTS

Socio-demographic and baseline variables for the 107 participants are presented in Table 1. Most participants were female (86%) and employed full time (79.1%). There were no differences in socio-demographic or baseline variables between black and white participants except for marital status; more white persons were married or living with a partner compared to black persons, χ²(1, N=107) = 10.08, P<.01. No differences in socio-demographic or baseline values were found between those who participated in the ancillary study and those who participated only in the weight-loss trial, P>.05.

Table 1.

Socio-demographic and Baseline Variables

Variable	Total (N = 107) n (%)	White (n = 81) n (%)	Black (n = 26) n (%)	Group Difference P-valuea
Gender
Female	92 (86.0)	67 (83.0)	25 (96.0)	.11b
Male	15 (14.0)	14 (17.0)	1 (4.0)
Marital Status
Married/living with partner	75 (70.8)	63 (79.0)	12 (46.0)	<.01
Unmarried/separated	31 (29.2)	17 (21.0)	14 (54.0)
Employment
Full time	83 (79.1)	59 (74.7)	24 (92.4)
Part time	10 (9.5)	9 (11.4)	1 (3.8)	.21b
Other	12 (11.4)	11 (13.9)	1 (3.8)
Income
< $30,000/yr	16 (15.1)	12 (15.0)	4 (15.4)
$30–50,000/yr	27 (25.2)	18 (22.5)	9 (34.6)	.44b
> $50,000/yr	63 (58.9)	50 (62.5)	13 (50.0)
	Mean (SD)	Mean (SD)	Mean (SD)	Pa
Education (years)
(Range: 12–23)	15.3 (2.6)	15.4 (2.5)	14.7 (2.7)	.15
Age (years)
(Range:20–55)	46.3 (6.9)	46.1 (7.1)	46.8 (6.4)	.73
BHE
(Range: 22–110)	52.0 (15.1)	51.0 (15.3)	55.1 (14.4)	.17
ELF
(Range: 26–130)	88.2 (12.6)	87.9 (13.6)	88.2 (9.4)	.65
WEL
(Range: 0–180)	119.8 (35.3)	120.8 (35.9)	116.6 (34.1)	.52
Social support
(Range: 7–35)	26.5 (6.3)	26.4 (6.5)	26.9 (5.6)	.94
Weight (kg)
(Range: 57.4–119.9)	87.9 (15.1)	86.9 (15.2)	90.1 (14.7)	.23
BMI (kg/m2)
(Range: 22.6–47.5)	33.0 (4.9)	32.5 (5.0)	34.6 (4.6)	.07

Note.

BHE = Barriers to Healthy Eating Scale; BMI = Body Mass Index, ELF = Experiences Associated with Following a Low-Fat Diet Scale; SD = Standard deviation; WEL = Weight Efficacy Lifestyle Scale.

^aFrom χ² tests (categorical variables), Mann Whitney U (education, age, BHE, ELF, WEL, social support), and t-tests (weight, BMI).

^bFisher’s exact test

Table 2 shows the independent and dependent variables for the total sample and each ethnic group. More participants were successful weight maintainers compared to unsuccessful maintainers in the total sample and each group. The percent weight change for the total sample was a mean of 4.6 ± 5.8%, with no difference between black and white persons in either the dichotomous or the continuous outcomes, P>.65. There was little change in the independent variables during the 18-month follow-up period except for barriers to healthy eating, which increased by a mean of 12.7 ± 30.5% and was similar in black and white participants, U=1003.0, P=.86.

Table 2.

Dependent and Independent Variables for the Total Sample, Black, and White Participants

Variable	Total (N = 107) n (%)	White (n = 81) n (%)	Black (n = 26) n (%)	Group Difference P-valuea
Successful maintenance	61 (57.0)	47 (58.0)	14 (53.8)
(≤5% weight gain)				.71
Unsuccessful maintenance	46 (43.0)	34 (42.0)	12 (46.2)
(>5% weight gain)
	Mean (SD)	Mean (SD)	Mean (SD)	Pa
% Weight change
(Range: −14.81–27.02)	4.6 (5.8)	4.4 (5.6)	5.0 (6.6)	.65
% ELF change
(Range: −28.4–37.7)	0.65 (11.5)	−0.37 (10.9)	3.72 (12.8)	.12
SEEb
(Range: 0–10)	5.4 (2.5)	5.4 (2.6)	5.3 (2.5)	.89
% BHE change
(Range: −55.7–125.8)	12.7 (30.5)	13.1 (29.7)	11.5 (33.5)	.86
% WEL change
(Range: −56.2–261.9)	1.4 (42.8)	0.6 (44.5)	3.6 (38.1)	.51
% Social support change
(Range: −44.0–100.0)	3.9 (23.9)	4.7 (24.7)	1.6 (21.8)	.66
What % of time does stress influence how you eat?b
(Range: 0–100)	56.8 (30.7)	57.9 (28.9)	53.5 (36.0)	.61
What % of time did you use techniques to lower your stress level?b
(Range: 0–100)	38.7 (28.5)	38.9 (27.7)	38.1 (31.2)	.79
How much did this stressful life event affect your eating habits?b,c
(Range: 0–10)	7.0 (3.1)	7.0 (2.9)	7.2 (3.5)	.45

Note.

BHE = Barriers to Healthy Eating Scale; ELF = Experiences Associated with Following a Low-Fat Diet Scale; SD = Standard deviation; SEE = Self-Efficacy for Exercise Scale; WEL = Weight Efficacy Lifestyle Scale;

^aFrom χ² tests (categorical variables), t-tests (weight, ELF, SEE), and Mann Whitney U (BHE, WEL, social support, stress items).

^bOnly measured in ancillary study- no percent change score.

^cSubsample analysis of n = 69 who reported experiencing a stressful life event: White n = 49, Black n = 20.

For specific aim 1, ethnicity was not a correlate of percent weight change (b=0.86, SE=1.45, P=.55) or unsuccessful weight maintenance (OR=1.38, 95% CI=0.51–3.77, P=.53) in either the multivariate linear or logistic models, after adjusting for socio-demographic covariates.

Table 3 shows the results of the multivariate linear regression of ELF change, BHE change, WEL change, and ethnic group with weight change. BHE change was the only significant correlate of weight change. The greatest barriers to healthy eating were the same for the total sample and each ethnic group. Over 65% of the sample reported that “Losing weight is rewarding, but I have trouble staying motivated to keep off the weight I lost” as a somewhat or very important problem. “When I am very hungry, I have trouble controlling what I eat” was a somewhat or very important problem for 64%. Nearly half the sample (49%) rated “It is difficult to motivate myself to eat appropriately” as a somewhat or very important problem; 45% of the sample rated “It is difficult to find time to plan appropriate meals for myself” and “I use food as a reward or treat for myself” as somewhat or very important problems.

Table 3.

Hierarchical Linear Regression With BHE, ELF, WEL, and Ethnicity Predicting %Weight Change

Variable	b	SE(b)	t-value	P-value
Block 1
Age (yrs)	−0.13	0.08	−1.50	.14
Gender	−0.57	1.84	−0.31	.76
Marital status	2.36	1.41	1.67	.10
Education (yrs)	0.04	0.25	0.16	.88
Income
< $30,000/yr	−0.65	1.86	−0.35	.73
$30–50,000/yr	0.46	1.38	0.33	.74
Block 2
% BHE change	0.08	0.02	3.21	<.01
% ELF change	−0.07	0.05	−1.42	.16
% WEL change	−0.01	0.02	−0.66	.51
Block 3
Ethnicity	2.33	1.41	1.65	.10

Note.

N = 101, White n = 76, Black n = 25. Three participants with missing data and 3 with influential observations excluded from the model. b = beta; SE = Standard error; BHE = Barriers to Healthy Eating Scale; ELF = Experiences Associated with Following a Low-Fat Diet Scale; WEL = Weight Efficacy Lifestyle Scale; reference groups = females, not married/separated, > $50,000/yr, white participants.

In Table 4, the results of the multivariate logistic model including ELF change, BHE change, WEL change, and ethnicity are presented. For every one-percent increase in barriers to healthy eating, the odds of being unsuccessful at weight maintenance were 1.04 higher (95% CI=1.02–1.07, P< .01). It is only in this one model that black participants had 3.78 times the odds (95% CI=1.04–13.67, P=.04) of being unsuccessful at weight maintenance compared to white participants.

Table 4.

Hierarchical Logistic Regression With BHE, ELF, WEL, and Ethnicity Predicting Unsuccessful Weight Maintenance

Variable	Odds Ratio	95% CI	P-value
Block 1
Age (yrs)	0.90	0.83–0.97	<.01
Gender	0.72	0.13–3.88	.70
Marital status	3.68	0.98–13.84	.06
Education (yrs)	0.97	0.79–1.18	.74
Income
< $30,000/yr	0.52	0.11–2.53	.42
$30–50,000/yr	0.46	0.14–1.54	.21
Block 2
% BHE change	1.04	1.02–1.07	<.01
% ELF change	0.96	0.92–1.01	.09
%WELchange	1.00	0.99–1.02	.76
Block 3
Ethnicity	3.78	1.04–13.67	.04

Note.

N = 101, White n = 76, Black n = 25. Three participants with missing data and 3 with influential observations excluded from the model. CI = Confidence interval; BHE = Barriers to Healthy Eating Scale; ELF = Experiences Associated with Following a Low-Fat Diet Scale; WEL = Weight Efficacy Lifestyle Scale; unsuccessful weight maintenance = > 5% weight gain; reference groups = females, not married/separated, > $50,000/yr, white participants.

Change in social support was not associated with weight change in the multivariate linear model (b=−0.02, SE=0.03, P=.45) or logistic model (OR=0.99, 95% CI=0.97–1.01, P=.35), and ethnic group was not related to weight change in the models with social support, P>.33.

For the total sample, the mean percentage of time that stress influenced how participants ate was over 53% (Table 2), but this item was not associated with weight change in the linear model (b=0.02, SE=0.02, P=.31) or the logistic model (OR=1.01, 95% CI=0.99–1.03, P=.07). The mean percentage of time that participants used techniques to reduce stress was approximately 38% (Table 2); this item also was not a correlate of weight maintenance in either the linear (b= −0.03, SE=0.02, P=.17) or logistic model (OR=0.99, 95% CI=0.98–1.01, P=.29). Ethnicity was not associated with weight change in any linear or logistic models with the stress items, P>.45.

Sixty-four percent (n=69) of participants reported experiencing one or more stressful life events during the 18-month follow-up period, and these events fell into 3 main categories: family/significant other related (eg, family member illness); work related (eg, difficult boss); and personal (eg, financial problems). The mean impact the event had on eating habits was 7.0 out of 10.0 (on a scale of 0 [no effect] to 10 [most effect]). See Table 2. The degree to which the stressful event(s) affected eating habits was significantly associated with weight gain in both the linear (b=0.51, SE=0.24, P=.04) and logistic models (OR=1.30, 95% CI=1.04–1.63, P=.02). A one-unit increase in the effect of the stressful life event on eating was associated with a 0.51% increase in weight in the linear model. The odds of being an unsuccessful weight maintainer were 1.3 times higher for each unit increase in the effect of the stressful life event on eating, adjusting for the effects of other covariates in the model.

Interaction terms assessing for moderation of the relationship between each psychosocial variable and weight maintenance due to ethnic group were not significant in any models (P>.10); therefore, effect modification due to ethnic group was not supported.

We also wished to examine the roles of dietary intake and physical activity as possible mediators of the relationships between the psychosocial variables and weight maintenance. Thus, we fit regression models, according to Baron and Kenny,³⁷ analyzing both the direct effect of each psychosocial variable on weight change and its indirect effect through caloric intake, fat gram intake, and energy expenditure. Two 24-hour dietary recalls³⁸ were used to collect dietary data via the Nutrition Data System for Research software (Nutrition Coordinating Center, University of Minnesota). The Paffenbarger Activity Questionnaire³⁹ was used to measure weekly energy expenditure via the number of city blocks walked, flights of stairs climbed, and minutes spent in leisure-time activities. An increase in barriers to healthy eating was associated with an increase in fat gram intake, but when the BHE and fat gram intake were regressed on percent weight change together, statistical mediation was not supported (b=.05, P=.58). Overall, dietary intake and physical activity were not mediators of the relationships between the psychosocial variables and weight maintenance.

Discussion

Weight maintenance was moderate with participants gaining an average of 4.6% of their weight in the 18 months between the weight-loss trial and this ancillary study. The majority was successful at weight maintenance with 58% of white participants and 54% of black participants gaining ≤5%. The proportion of individuals who gained ≤5% is similar to what others have found following behavioral weight-loss treatment. Of 179 former weight-loss participants, 42% had regained < 5% of their end-of-treatment weight an average of 14 months after completing behavioral treatment.⁴⁰ Others have reported that 78% of participants gained <5% of their weight 2.2 years later.⁴¹ The multicenter Weight Loss Maintenance trial found that 37% of participants maintained at least a 5% weight loss after 30 months.⁴²

Ethnic differences in weight maintenance were generally not noted. White persons gained about 4.4% of their weight whereas black individuals gained 5%. Ethnicity was not associated with weight maintenance except in one multivariate logistic regression model (Table 4) in which black participants were more likely to be unsuccessful weight maintainers compared to their white counterparts. An explanation for this finding is not readily apparent, although categorizing the continuous dependent variable of percent weight change might have reduced measurement precision⁴³ and affected the significant findings for ethnic group in this single model.

Despite sparse existing evidence, findings regarding black individuals and weight maintenance after a loss are inconsistent in the literature. Two large clinical trials examining weight loss and maintenance to control blood pressure found that black participants tended initially to lose less weight than white participants did, but then to regain less weight, so that the difference in weight loss between the groups was minimal at 36 months.^9,44 For example, in the Trials of Hypertension Prevention- Phase II, white participants lost 2.3 kg more weight than did black participants at 6 months, but the difference in weight loss between blacks and whites was only 0.5 kg at 36 months. Others have reported greater weight regain among black diabetic participants after a weight-loss program.¹² In the large, Weight Loss Maintenance randomized controlled trial, which had an enrollment of 38% black participants (n=388), the effects of the interventions to prevent weight gain did not differ by ethnicity.⁴² Our results suggest that weight gain after weight-loss treatment is similar for white and black persons.

Increases in barriers to healthy eating were related to weight gain, similar to what others have found.^17,40 In a follow-up study of a university-based behavioral weight-loss program, the percentage of unsuccessful weight maintainers frequently experiencing the barriers of healthy eating’s being too expensive and healthy eating’s being too time-consuming was higher compared to successful maintainers (7.5% vs 0% and 19.0% vs 8.8%, respectively).⁴⁰ In line with these results, finding time for appropriate meal planning was difficult for a greater proportion of the unsuccessful maintainers compared to the successful maintainers in our study. Befort and colleagues also noted that 88% of unsuccessful maintainers reported frequently experiencing the barrier “too easy to slip back into old habits” compared to 53% of successful maintainers.⁴⁰ This barrier could be perceived as a lack of motivation for sustaining weight management behaviors, similar to our finding that a majority had difficulty with motivation for keeping off the weight. Also in regard to maintaining motivation for behavior changes, others have found that, given the required effort, satisfaction with weight loss decreased significantly in the latter half of a weight-loss program.⁴⁵ Although the price of healthy foods has been reported to be a barrier to healthy eating,^17,40 only 17% of participants in our study reported that the cost of low-fat/low-calorie foods was a somewhat or very important problem. This finding might be due to the fact that approximately half of the participants in both ethnic groups reported an annual household income of ≥$50,000.

Despite the relatively small sample size, the impact of a stressful life event on eating habits was significantly associated with weight gain among the subsample of 69 participants who reported experiencing a stressful life event. Individuals who are more physiologically responsive to stress are at higher risk of becoming obese and developing central adiposity.^46,47 Although stress has been linked to weight change,⁴⁸ weight is not always gained. Individual stress responses might include decreased appetite and food intake.⁴⁹ We found that a greater impact of a stressful life event on eating was related to weight gain among the participants who experienced a stressful life event. Yet, in the entire sample, the influence of stress on eating was not associated with weight change, perhaps because those who did not report a stressful event had a lower overall stress level or had a physiologic response to stress that did not result in increased food intake.

Some factors—self-efficacy, social support and low-fat diet experiences—were not associated with weight maintenance, despite being associated with weight management in previous studies.^14,21,22 Similar to our findings, Linde et al²⁰ noted that self-efficacy did not predict weight change during the post-weight-loss treatment time frame. Self-efficacy for exercise was unrelated to weight maintenance, perhaps because some items might not have been relevant to our participants, eg, “How confident are you that you could exercise 3 times per week for 20 minutes if the weather was bothering you?” Moreover, self-efficacy for initially changing eating and activity behaviors may differ from the self-efficacy needed to maintain those changes over time.⁵⁰ The level of social support was relatively high at baseline (Table 1) and increased minimally, which could explain why social support was not related to weight change. Also, participants reported positive experiences following a low-fat diet at baseline, and there was essentially no change; this finding might explain the ELF’s lack of relationship to weight maintenance.

A few limitations to our study existed. Because this was a follow-up study of weight maintenance, the selection of participants was restricted to those who completed the weight-loss trial. Thus, the sample of 26 black individuals might be insufficient to draw definitive conclusions about weight maintenance with this ethnic group, and larger studies are needed to confirm these findings. Yet, a grossly inadequate amount of information exists in the literature about weight-loss maintenance among minority adult populations,^10,51 and this study adds to that small body of knowledge. In fact, so few studies have examined this topic that a recent meta-analysis of behavioral interventions for weight-gain prevention or weight loss among minority populations included studies in the review for which there were at least 25 minority participants.⁵² Our study would have met that inclusion criterion. Moreover, 24% of our study sample was black or African American, which is comparable to the population of blacks or African Americans in the closest large city (27.1%)⁵³ and more than double the population of the county (10.9%).⁵⁴ Additionally, over half of all participants had a mean household income greater than $50,000/year, and the mean years of education was about 15, limiting the generalizability of our findings to individuals of a higher education level and socioeconomic status. Because health disparities are often associated with a lower socioeconomic level,⁵⁵ it would be important to examine weight maintenance in a more socioeconomically diverse group. Given the modest sample size and single-site nature of the study, the findings may not be generalizable beyond this geographic region. Finally, our sample consisted of only 15 males, one of whom was black, suggesting that our findings are generalizable only to women; the representation of one black male prohibits generalizability to black men.

The strengths of our study include a long-term assessment of weight maintenance, an examination of weight maintenance among black individuals, an objective assessment of weight, and a high participation rate. This study adds to the limited information about long-term weight maintenance, which is not often examined beyond a year after treatment.⁴ The study also contributes to the small body of knowledge about weight maintenance among black individuals and suggests that maintenance of weight after weight loss does not significantly differ for black and white persons. Strategies for maintaining weight loss need to be developed and disseminated to individuals of both ethnicities. Lastly, 90% of participants who completed the weight-loss trial took part in this ancillary study, and no differences existed between the few participants who did not return and those who did. This high participation rate reduces the likelihood that our findings are biased as a result of individuals who were less successful at weight maintenance not taking part.

In conclusion, a majority of the participants were able to maintain their weight 18 months after completing a behavioral weight-loss trial with a similar amount of weight gain experienced by black and white participants. Future investigations are needed to determine the most appropriate strategies to help individuals reduce barriers to healthy eating and cope with the negative impact of stressful circumstances, so weight loss may be sustained long-term.