A Pilot Group-based Lifestyle Intervention to Promote Weight Loss among Mexican Immigrants

Abstract

Background:

Mexican Americans are at increased risk for obesity upon immigration to the U.S., increasing their risk for diabetes, cardiovascular disease, and cancer. Our pilot individual lifestyle intervention culturally tailored for Mexican Americans, COMIDA (Consumo de Opciones Más Ideales De Alimentos) (Eating More Ideal Food Options), showed promising results. This paper presents outcomes from the group-based version of COMIDA.

Methods:

129 overweight/obese Mexican Americans were enrolled in ‘Group COMIDA’. Participants’ weight (primary outcome), dietary intake and nutrition knowledge (secondary outcomes) were assessed pre- and post-intervention.

Results:

Seventeen percent of participants experienced 5% weight loss at follow-up. Post-intervention, participants consumed more fruit and vegetables and fewer sweets and fried foods.

Discussion:

A group-based, culturally adapted lifestyle intervention may be a more cost-effective approach than individual interventions to improve dietary behavior among underserved populations, though additional modifications may be considered to increase the intervention’s effectiveness in promoting significant weight loss.

Background

The national obesity crisis is disproportionately borne by Hispanics and especially by Mexican Americans (MAs), the largest U.S. Hispanic subpopulation. Among MAs, 78.3% of women and 81.9% of men are overweight or obese, compared to 60.9% of women and 73.2% of men among non-Hispanic Whites (NHWs).¹ Overweight/obesity is strongly correlated with an increased risk of all-cause mortality and is a significant risk factor for cardiovascular disease (CVD) and diabetes, cancer, and other chronic diseases, including osteoarthritis, sleep apnea, and depression.²

While CVD and cancer encompass two broad disease groups, they have overlapping preventable behavioral (unhealthy diet, physical inactivity) risk factors.^3–5 Obesity-related health risks can be significantly attenuated with a 5% reduction in body weight.⁶ Evidence-based lifestyle interventions that target behavior change, such as the Diabetes Prevention Program (DPP) and LOOK AHEAD have been found to be effective for weight loss, weight loss maintenance, and prevention of diabetes and CVD.^{7, 8} Lifestyle interventions continue to remain frontline anti-obesity treatments as they are less intrusive than pharmacological and surgical treatments.^{9, 10} However, these programs are resource intensive and can be burdensome to participants (requiring 16 in-person sessions in the initial phase) and therefore have not been widely adopted in community settings.¹¹

There is strong evidence that group-based lifestyle interventions are time-efficient, economical, and effective in promoting weight loss and lead to related health benefits (e.g., improvements in blood pressure, cholesterol, and blood sugar; prevention of Type 2 diabetes and cardiovascular disease) for overweight/obese individuals.¹² Group-based delivery facilitates opportunities for social support and sharing of weight loss strategies¹² and may have special relevance for individuals who share background and lifestyle characteristics. For example, group-based lifestyle interventions have focused on workers employed in shift-based work (i.e., any work schedule that falls outside the hours of 7 am and 6 pm, a risk factor for chronic diseases),¹³ as well as male- or female-only groups.¹⁴ However, few have specifically focused on minority/cultural groups, despite the fact that members of minority groups may share culturally based food and physical activity preferences.¹⁵ A systematic review of group-based lifestyle interventions for overweight/obese individuals found only four of 47 studies included interventions adapted for Hispanic populations.¹² The paucity of such programs is concerning given findings demonstrating that Hispanics who do participate in weight loss interventions designed for the general population lose less weight and are less likely to maintain weight loss post-intervention compared to NHWs.¹⁶

We developed Group COMIDA (Consumo de Opciones Más Ideales De Alimentos) (Eating More Ideal Food Options) to fill this critical gap in group-based, culturally responsive lifestyle interventions for this large, growing, and underserved group. The Group COMIDA program was based on the Individual COMIDA program (manuscript under review)¹⁷ to address obesity/overweight and dietary educational needs among Mexican Americans in a community setting, i.e., the New York metropolitan area Ventanillas de Salud (Health Windows) program at the Mexican Consulate in New York City.¹⁸

The current study presents pilot findings of the Group COMIDA intervention on participant changes in weight (primary outcome) and dietary intake, nutrition knowledge, attitudes, and beliefs, food insecurity, and physical activity (secondary outcomes).

Methods

Study Design

This was a single-arm, pre-post 12-week pilot of a group-based lifestyle intervention. Study approval was obtained from the Memorial Sloan Kettering Cancer Center Institutional Review Board. The study was conducted between January 2016 and January 2020.

Participants

Recruitment was conducted through the Ventanillas de Salud (VDS; Health Windows) program at the Mexican Consulate in New York City (NYC). The VDS provides basic health screening and education, and navigation into health care, and assists hundreds of thousands of underserved MAs annually. The New York metropolitan area VDS is an academic-community partnership led by the MSKCC Immigrant Health and Cancer Disparities (IHCD) Center and the Mexican Consulate in NYC. Potential participants were approached for study enrollment while visiting the VDS. Inclusion criteria were: 1) self-identifies as Mexican American or Mexican; 2) Spanish-speaking; 3) overweight (BMI 25 – 29.9 kg/m²) or obese (BMI ≥ 30 kg/m²); 4) ≥ 18 years of age. Exclusion criteria were: 1) not being in the NYC area for the 3-month study duration; 2) being pregnant or lactating; 3) presence of a chronic disease such as cancer, kidney disease, liver disease; 4) having dietary restrictions; 5) having a phone without text messaging capabilities or unwilling to accept text messages; 6) presence of a serious psychiatric or cognitive impairment; and 7) having another family member already enrolled in Group COMIDA.

Description of the Intervention

Group COMIDA is a 12-week culturally and linguistically tailored community-engaged dietary counseling and education program developed to help participants eat more healthfully, increase physical activity, and lose weight. Group COMIDA is delivered in Spanish by trained bilingual outreach staff and includes, at baseline: BMI measurement and intake survey; an in-person group educational workshop session; low literacy Spanish-language written materials with information and resources for food, nutrition and exercise; and healthy eating tools (measuring cups and spoons, water bottle, digital scale, and a Spanish-language MyPlate portion plate¹⁹). After baseline activities are completed in-person at the Mexican Consulate, participants continue to receive thrice weekly diet/exercise text messages (e.g., “Eat fruit instead of dessert or pan dulce”) for 12 weeks. Twelve weeks post-enrollment, participants completed a follow-up BMI measurement and exit survey. Participants received a $30 incentive at intake and exit.

The COMIDA group nutrition workshop session was based on the USDA MyPlate program¹⁹, which is available in Spanish, and is described in more detail in a previous publication¹⁸. The group workshop differed from the individual counseling sessions in that information was shared in a slide presentation to accommodate groups and incorporated discussion to facilitate exchange of experiences and information to encourage behavioral change. Additional information was also included in the group sessions based on participants’ interests (e.g., more information on carbohydrate content of potatoes vs. sweet potatoes vs. yucca, the nutritional value of fruit juices, etc.). Workshops were designed to be appropriate for small groups of 2–3 to up to 10–15 participants and took approximately 45–60 minutes to deliver.

Study Measures

Primary Outcome Measure

Change in Weight.

Weight was measured using a Taylor 7009 Lithium Electronic Scale at intake and exit.

Secondary Outcome Measures

Dietary Intake.

The California Health Interview Survey (CHIS) was developed to capture the diversity of the California population, including a large sample of Hispanic respondents and Hispanics of Mexican origin²⁰. The CHIS is available in Spanish and includes self-reported measures on dietary intake per week during the month before the interview: fruit (excluding juices), fried potatoes (including French fries, home fries, and hash browns), other vegetables (excluding fried potatoes), regular soda or pop that contains sugar (excluding diet soft drinks), and fast food (e.g., “During the past month [“or in the past 7 days” for fast-food consumption], how often did you eat [food item name]?”). The scale was modified by adding 7 items: sports and energy drinks, sweetened fruit drinks, coffee and tea with sugar or honey added, ice cream or frozen desserts, sweet pastries (including cookies, cake, pie, or brownies), fried foods, and red meats. Responses were standardized to reflect mean per-week intake frequency. The modified CHIS was administered at intake and exit.

Nutrition Knowledge.

Participants were administered a nutrition knowledge, attitudes, and beliefs (KAB) survey based on a tool used in previous research to assess nutrition knowledge among low-income Hispanics and African Americans²¹. The survey was modified to include questions related to specific content in the group counseling and written materials. Ten items on nutrition knowledge (i.e., identifying foods high in saturated fat, good sources of calcium, fiber, etc.) comprised a nutrition knowledge score. Correct answers were scored as 1 and incorrect answers were scored as 0. A higher score indicated greater knowledge of nutrition.

Food Insecurity.

Food insecurity was assessed using the Spanish version of the validated USDA Household Food Security Survey (HFSS) module²². In this module, participants were asked 18 questions regarding their household’s experience with food insecurity using the 12-month food-security scale at intake and 30-day scale at exit. Households with a raw score ≥3 were classified as food insecure, as defined by the USDA.²²

Satisfaction with Program.

The participant satisfaction survey was used to capture information about the usefulness of, and recommendations to improve, the Group COMIDA program. Participants were asked to complete the satisfaction survey at intake after the group session and at exit.

Statistical Analyses

Sociodemographic and health care access characteristics were summarized as follows: categorical characteristics were summarized using frequencies and percentages, while continuous characteristics were summarized with mean ± standard deviation. Comparisons were carried out using chi-square tests for categorical characteristics, and t-tests for continuous characteristics.

Pre- and post- intervention weight (in pounds) was reported using mean ± standard deviation and achievement of 5% weight loss at follow-up determined based on baseline weight and summarized using frequency and percent. Mean and median servings per week for each food item on the CHIS were reported pre- and post- intervention. The frequency and percent of correct responses for each item on the KAB survey were reported pre- and post- intervention, as well as the percent difference between the two. Food security status was measured using the USDA Household Food Security Survey (HFSS) module²². Participant satisfaction surveys were summarized with frequencies and percentages. Univariate analyses were performed to determine whether primary and secondary outcomes differed by sociodemographic characteristics.

To account for statistical dependency due to participants receiving the intervention in groups, mixed effects modeling was used to examine the effect of the intervention on the continuous outcomes of weight, CHIS domains, and KAB overall score. Linear mixed effects models were used to model changes in weight and each CHIS domain, as a function of time (pre- vs. post- intervention). Based on the distribution of the KAB overall count score, a Poisson mixed effects model was used to assess change in nutrition knowledge as a function of time (pre- vs. post- intervention). A generalized linear mixed model was used to model odds of USDA-defined food insecurity, as a function of time (pre- vs. post- intervention). To adjust for potentially inflated type I error due to multiple comparisons, the Benjamini-Hochberg procedure²³ was used with a false discovery rate of 5%.

Results

Sociodemographic and Clinical Characteristics of the Sample

Among all participants (N=129), average age was 40 years (SD=8.96), 71% were female, 56% had between 7 and 12 years of education, 94% were limited English proficient, and mean duration of years living in the U.S. was 17 (SD=8.02). The most common occupations were in the restaurant and cleaning services industries. The mean monthly household income was $1,799 (SD=$951). Eighty seven percent of participants were living below the federal poverty line. Fifty-three percent did not have a primary care physician and 74% were uninsured. Thirteen percent of participants reported a history of heart disease or stroke, 12% hypertension, 19% diabetes, 2% cancer and 11% depression. Two percent were current smokers and 3% were past smokers. Eight percent of participants reported current drinking; those who currently drank reported drinking alcohol an average of 2 days/month (Table 1). There were no significant differences found when comparing all enrolled participants (n=129) with those that completed follow-up (n=100) across demographic and healthcare access characteristics.

Table 1.

Sociodemographic and clinical characteristics of the sample (N=129)^a

Variable		n (%)
Demographics
Gender	Male	37 (29%)
	Female	91 (71%)
Education	0–6 years	43 (34%)
	7–12 years	72 (56%)
	> Some college	13 (19%)
English	Very well	7 (6%)
proficiency	Well	28 (23%)
	Not well	70 (55%)
	Not at all	20 (16%)
Spanish	Very well	94 (73%)
proficiency	Well	32 (25%)
	Not well	2 (2%)
	Not at all	0 (0%)
Employment	Not currently working	46 (36%)
	Restaurant worker	24 (19%)
	Cleaning	24 (19%)
	Driver	7 (5%)
	Home attendant/nanny	7 (5%)
	Retail	5 (4%)
	Agriculture	4 (3%)
	Construction	4 (3%)
	Mechanics	4 (3%)
	Other	4 (3%)
Medical Conditions and Risk Behaviors
Heart disease or stroke	Yes	17 (13%)
	No	112 (87%)
Hypertension	Yes	15 (12%)
	No	112 (88%)
Diabetes	Yes	25 (19%)
	No	103 (81%)
Cancer	Yes	2 (2%)
	No	125 (98%)
Depression	Yes	14 (11%)
	No	113 (89%)
Smoked more than 100	Current smoker	2 (2%)
cigarettes in life	Past smoker	3 (2%)
	No	124 (96%)
		M (SD)
Alcohol consumption	days per month	2 (3.928)

^aCells do not all sum to n=129 due to missing data

Results for the Grouped Analysis Outcomes

There were 53 groups in total, with a mean group size of 2 (SD=0.57, min=2, max=4).

Primary Outcome (Change in Weight)

Overall mean weight loss (N=100) was 0.96 +/− 6.70 pounds. While accounting for the data structure of participants nested within groups, a linear mixed effects model indicated no significant change in weight from pre to post intervention. When looking at weight as a categorical outcome (achievement of 5% weight loss), we found that 17% (n=17) of participants experienced 5% weight loss at follow up. For those who achieved 5% weight loss, the mean weight loss was 8.89 +/− 4.07 pounds. Univariate analyses showed no significant differences in sociodemographic characteristics between those who lost at least 5% of their baseline body weight and those who did not.

Secondary Outcomes

Univariate analyses showed no significant differences in sociodemographic characteristics by secondary outcomes (dietary intake, nutrition knowledge, food insecurity, nor satisfaction).

Dietary Intake.

While accounting for the data structure of participants nested within groups, a series of univariate linear mixed effects models indicated a significant increase in self-reported servings of fruit (β=2.07, p=.006) and vegetables (β=2.16, p=.041), and significant decreases in self-reported servings of fried potatoes (β=0.66, p<.001), soda (β=−1.34, p=.006), coffee/tea with sugar (β=−2.23, p=.002), cookies (β=−1.75, p=.006), frozen dessert (β=−0.44, p.006), fast food (β=−0.52, p.011), fried food (β=−0.54, p=.017), and red meat (β=−0.63, p=.022), as a function of time (pre vs. post intervention) (Table 2).

Table 2.

Self-reported food servings per week pre- vs. post-intervention (n=100, 53 groups)^*

	Pre-intervention		Post-intervention		LMEa Results
	Mean	Median	Mean	Median	B b	SE c	p-valued
	Self-reported food servings/week
Fruit	5.16	3.00	7.23	7.00	2.07	0.66	0.006
Fried potatoes	1.15	0.46	0.49	0.23	−0.66	0.16	<.001
Other vegetables	4.48	3.00	6.64	4.00	2.16	1.01	0.041
Soda	2.57	1.00	1.23	0.58	−1.34	0.44	0.006
Fast food	0.89	0	0.97	0	0.07	0.30	0.811
Sports/energy drinks	1.60	0.12	1.75	0.46	0.14	0.5	0.811
Sweetened fruit drinks	5.46	7.00	3.26	2.00	−2.23	0.61	0.002
Coffee/tea with sugar	2.95	1.15	1.2	0.69	−1.75	0.57	0.006
Frozen desserts	0.98	0.46	0.44	0.23	−0.44	0.12	0.006
Sweet pastries	1.81	1.00	1.32	1.00	−0.52	0.19	0.011
Fried foods	1.13	0.46	0.58	0.23	−0.54	0.21	0.017
Red meats	2.09	1.16	1.47	1.00	−0.63	0.26	0.022

NB:

^*5 groups were excluded due to missing data.

^aLinear mixed effects model.

^bBeta coefficient for univariate linear mixed model.

^cStandard error.

^dAdjusted p-values were corrected for multiple testing using the Benjamini-Hochberg procedure with a 5% false discovery rate

Nutrition Knowledge.

Nine out of 10 items on the KAB survey were included in the analysis, with 1 item excluded due to missing data. Item-wise descriptive statistics indicate that participants had greater knowledge of nutrition post-intervention vs. pre-intervention. Improvements were seen for prompts regarding whole vs. skim milk (10.6% increase), whole grains vs. refined grains (7.6%), recommended frequency of meat and poultry consumption (7.1%), lean meats and poultry (20%), and vegetable/fruit consumption (9.7%), though differences were not significant. There was no significant change in overall KAB score (Table 3).

Table 3.

Nutrition knowledge pre- vs. post-intervention (n=100, 53 groups)^*

KAB prompts	Pre-intervention		Post-intervention		% Change
	N	%	%	N
1. It is healthier to drink whole milk than skim or 1% milk	66	71	78.5	73	10.6%
2. I should choose 100% whole-grain cereals, breads, crackers, rice, and pasta instead of refined grains to have a healthy diet.	66	71	76.3	71	7.6%
3. When considering portions in my meal, a quarter of my plate should be fruits and vegetables.	83	89	91	85	2.4%
4. I should eat meat and/or poultry everyday to have a healthy diet.	70	75	81	75	7.1%
5. It is healthier to eat lean cuts of meats or poultry than regular meats and poultry	60	65	77	72	20.0%
6. It is better to eat fruit for dessert than pastries.	89	96	97	90	1.1%
7. Sodas have more nutritional value than water.	90	97	97	90	0.0%
8. It is better to use butter than olive or canola oil.	83	89	85	79	−4.8%
9. It is important to eat vegetables, but not fruits.	62	67	73	68	9.7%

NB:

^*5 groups were excluded due to missing data.

Food Insecurity.

Thirty-seven percent of participants were food insecure at baseline, compared to 31% at follow up. There was no significant change in food insecurity as a function of time (pre vs. post intervention).

Satisfaction.

At follow up, 100% participants rated the program as good or excellent, 100% rated the health coach performance as good or excellent, and 97% stated they would not change anything about the program.

Discussion

Group COMIDA had a limited impact on weight loss. Mean weight loss was 0.96 +/− 6.70 pounds and overall, participants did not evidence a significant change in weight pre vs. post intervention. When weight was examined as a categorical outcome (5% weight loss from baseline), however, the intervention was effective in a subgroup (17%) of participants. This subgroup’s mean weight loss of 8.90 +/− 4.07 pounds was within the range of that observed in other more intensive interventions with underrepresented populations (a meta-analysis of obesity interventions for ethnic minority adults in the U.S. found a range of −16.3 to +1.3 pound mean weight change among 13 studies, with an overall mean weight change of 7.7 pounds)²⁴. On the other hand, the intervention was effective in producing changes in dietary intake.

Compared with the individual COMIDA intervention, Group COMIDA was less effective in achieving 5% weight loss (17% in Group COMIDA vs 28% in the individual format¹⁷). However, changes in dietary intake were similar across the intervention formats, with both showing significant differences in self-reported dietary intake pre- and post-intervention. Changes in nutrition knowledge, while notable and showing positive trends, were not statistically significant among group-based participants but were significant among participants in the individual COMIDA intervention.

One factor that may have limited Group COMIDA’s impact is the mixed gender of group participants. A meta-analysis of group-based weight loss interventions found that interventions delivered to men only were twice as effective as interventions delivered to women only¹², suggesting that men in particular may benefit from male-only weight-loss programs. Researchers have suggested that cultural and socioeconomic factors (e.g., cultural views of masculinity, fatigue due to strenuous manual labor occupations, fear of having to pay for research studies) may contribute to the need for tailoring for Hispanic men²⁵.

The meta-analysis also demonstrated that group interventions providing individualized feedback on dietary plans showed better results than those that did not¹². It is possible that the individual COMIDA intervention included more opportunities for individualized feedback than the Group COMIDA intervention. Incorporating individualized elements such as telephone check-ins and tailored text messages based on baseline dietary intake and nutrition knowledge^{26, 27} may maximize the benefits of group-based lifestyle interventions while minimizing cost. Increasing the intervention’s duration (i.e., by lengthening the workshop from 45–60 minutes to 90 minutes) might also increase opportunities for individualized elements (such as Q&A).

Prior studies also suggest that a unique benefit of group-based lifestyle interventions is the provision of social support in meeting behavioral goals.²⁸ Given the Hispanic cultural value of familismo or familism, which emphasizes commitment to one’s family, Hispanic populations may benefit from family-oriented interventions, which provide both social support and opportunities for modeling healthy behaviors.²⁹ In our recent focus group study with overweight/obese Mexican Americans, participants described family dynamics as both a contributor to obesity and a motivation for behavior change.³⁰ Including family members in the group COMIDA intervention could improve efficacy through the provision of family-based social support. In addition, including an online group forum or social media component to increase social support from peers could enhance accountability and help sustain behavior change.³⁰

A common challenge in developing community-based lifestyle interventions for underserved populations is time and resource constraints, as many interventions feature high cost and participant burden¹¹. Most Group COMIDA participants (87%) had incomes below the federal poverty line, 74% were uninsured, and were employed in shift-based work, which limited their ability to engage in a more time-intensive program. Group COMIDA was designed to address common barriers to participation among underserved MAs (e.g., time constraints, cultural and linguistic barriers) by integrating SMS content³¹, requiring in-person participation only at study intake and exit, and by cultural and linguistic tailoring of lifestyle modification recommendations³². These adaptations likely aided in retaining participants, as 100% of participants rated the program as “good” or “excellent” and loss to follow-up (22%) was much lower than in other community-based weight loss interventions for Hispanic populations, which ranges from 45–60%¹⁶. Attrition was similar to other group-based interventions for the general population¹².

Study limitations include the use of self-reported data; participants may have inaccurate recall of dietary intake or have over-estimated changes to their diet due to a desire to report improvements post-intervention. The lack of a control group limits conclusions we can draw from this single-arm pilot; a randomized controlled trial is needed to further evaluate the impact of the intervention.

Still, the present study, together with results from the individual pilot COMIDA intervention, yields important insights concerning the impact of a group-based version of COMIDA to reduce obesity among Mexican Americans. Group COMIDA was feasible to deliver, highly satisfactory to participants, and resulted in improvements in dietary intake, but was less effective in weight reduction than individual-based COMIDA. A next step could be an optimization trial³³ to determine which specific program elements (e.g., group workshops, tailored SMS messages, telephone check-ins, written materials, healthy eating tools) and delivery variables (e.g., single vs. mixed gender groups) were most effective in producing change. Further research could also evaluate the cost effectiveness of group versus individual formats, explore entirely remote delivery of the intervention to further reduce barriers to participation (e.g. group workshops held via teleconference call), and assess longer term impact (assess outcomes at 6 months/1 year). Our Group COMIDA pilot study thus provides preliminary results to inform the development of future trials to build and evaluate disseminable, culturally adapted, community-based lifestyle interventions for underserved Hispanic populations.

Availability of data and material

Data available upon request.