Dietary Outcomes in a Spanish-Language Randomized Controlled Diabetes Prevention Trial With Pregnant Latinas

Abstract

Objectives. We evaluated the effectiveness of a community-based healthy lifestyle intervention in improving dietary behaviors of pregnant Latinas from 2004 to 2006 in Detroit, Michigan.

Methods. The 11-week, culturally tailored, Spanish-language Healthy Mothers on the Move (MOMs) intervention offered home visits, group classes, related activities, and social support from trained community health workers (CHWs) and peers. Dietary behaviors were measured by food frequency questionnaire. Linear mixed models estimated pre- and post-intervention changes, within and between MOMs intervention and minimal intervention (MI) groups.

Results. MOMs (n = 139) and MI (n = 139) participants had similar baseline characteristics and dietary intake. Post-intervention, MOMs participants showed significant improvement in all dietary behaviors, except fruit and fiber consumption. Compared with MI participants, MOMs participants had significantly decreased consumption of added sugar (P = .05), total fat (P < .05), saturated fat (P < .01), percentage of daily calories from saturated fat (P < .001), solid fats and added sugars (P < .001), and had increased vegetable consumption (P < .001). Their increase in fiber consumption (P < .05) was significant relative to MI participants’ decrease in fiber intake.

Conclusions. We confirmed the hypothesis that a community-planned, CHW-led healthy lifestyle intervention could improve dietary behaviors of low-income Latina women during pregnancy.

Obesity, gestational diabetes, and type 2 diabetes are prevalent among Latina women of childbearing age in the United States.^1–6 Dietary intake, particularly low consumption of vegetables and fruits and high consumption of added sugars and solid fats, is associated with these conditions.^7–13 In the Diabetes Prevention Program randomized controlled trial, an intensive healthy lifestyle intervention that helped participants develop healthy diet and exercise patterns reduced the risk of developing type 2 diabetes among nonpregnant women with impaired glucose tolerance by half, including those with and without a history of gestational diabetes.¹⁴

Pregnancy may be an optimum period for interventions to improve dietary behaviors^10,15,16 that have consequences for both maternal and child health.^11,17–20 Following the success of the Diabetes Prevention Program, it is important to assess whether community-based healthy lifestyle interventions can reduce diabetes risk factors in pregnant women.¹⁸ Nonetheless, few randomized controlled studies have included dietary outcomes for pregnant women.^21,22 Only 1 pilot study was conducted in pregnant Latinas.²³

To address this gap, a community-academic partnership used community-based participatory research (CBPR) processes to develop and implement Healthy Mothers on the Move (MOMs), a community health worker (CHW)-led healthy lifestyle intervention tailored to the needs and strengths of pregnant Latinas in Detroit, Michigan.^24–26 Healthy MOMs aimed to demonstrate the effectiveness of this intervention to reduce behavioral and clinical risk factors for type 2 diabetes in pregnant and postpartum Latinas. We examined the hypothesis that women randomly assigned to the MOMs intervention group would significantly decrease their intake of added sugars, total fat and saturated fat, and significantly increase their intake of fruit, vegetables and fiber, compared with women assigned to the minimal intervention (MI) control group during pregnancy.

METHODS

The study design, staffing plans and recruitment, retention, intervention, and evaluation methods and materials were developed by a community-based steering committee in accordance with CBPR principles.²⁵ Before the start of recruitment, the study was approved by the institutional review board at the University of Michigan.

This study was conducted between 2004 and 2006 in southwest Detroit, a predominantly low-income, Mexican-origin Latino community.²⁷ Pregnant women were recruited at a federally qualified health center (FQHC), Supplemental Nutrition Program for Women, Infants, and Children (WIC) clinics, and with materials distributed to community organizations. Pregnant Latinas were eligible to participate if they were aged 18 years or older, southwest Detroit residents, and at less than 20 weeks’ gestation at eligibility screening. During an orientation session, staff explained the study purpose, meaning of randomization, program content, expectations for MOMs and MI participants, and data collection procedures. Informed primary and medical record consents were obtained, baseline data collection visits were scheduled, and information about childcare, transportation, and incentives was provided.

This 2-arm randomized control trial used a parallel-group design. Participants were allocated in a 1:1 ratio to the MOMs or MI group. A University of Michigan statistician generated the random allocation sequence in blocks of 40 for each cohort. Following baseline data collection, each woman received an incentive payment and sealed envelope containing her group assignment. Sample size calculations indicated that 100 participants per arm would result in 80% power to detect an increase from 25% to 45% of women who consumed at least 5 daily servings of fruits and vegetables.²⁸ Target recruitment sample size was increased to 160 women per arm to account for a possible 35% attrition rate.^4,29 Recruitment ended at 139 women per arm because of better than expected retention.

Interventions

The MOMs intervention was based on formative research that included semi-structured interviews with Spanish-speaking pregnant and postpartum Latinas and people who they indicated influenced their eating, exercise, and weight beliefs and behaviors.²⁵ Major themes and examples of beliefs, behaviors, barriers, and strategies were derived from these interviews, and focus groups conducted with additional pregnant and postpartum Latinas and program and policy leaders. The Steering Committee summarized themes and reviewed the resulting curriculum, staffing, and intervention design developed by project staff. The intervention was pilot tested in Detroit with similar participants in 2002.

The MOMs curriculum was conducted in Spanish with cohorts of 15 to 20 pregnant women by trained Latina CHWs, in 2 home visits and 9 group meetings during an 11-week period. Content integrated general pregnancy education and information, discussion, and activities aimed at developing knowledge and skills needed to reduce social and environmental barriers to healthy eating, regular exercise, and management of daily life stressors. Each group meeting concluded with content review and goal setting. Meeting 1, called Healthy Mom, Healthy Baby, discussed stress and dietary behaviors. During meetings 2 and 3 (home visits), CHWs encouraged women to develop and review behavioral goals. Four meetings focused specifically on diet. Meeting 3 was named Plan to Eat Healthy, and included a home visit and a discussion of the role of nutrition, beliefs about food and eating patterns during pregnancy, and goal setting. Meetings 5, 6, and 7 were group meetings titled Eat More Fiber, Eat More Fruits and Vegetables, and Eat Less Fat and Sugar, respectively. They used the US Department of Agriculture Pregnancy Food Guide Pyramid, and food label reading, food model, and taste test activities. Meeting 10, which was called Healthy Activities Together, included healthy eating problem solving based on barriers the women identified. Optional weekly group activities such as healthy cooking demonstrations corresponded to intervention topics. A key component of the MOMs intervention was informational and emotional social support from the CHWs and peers. CHWs encouraged women to problem solve, share strategies, and recognize each other’s efforts. Informal conversations about shared experiences strengthened peer support.

MI participants had 3 group pregnancy education meetings, delivered in Spanish by professional staff from a partner organization in a separate community setting, using MOMs curriculum materials related to pregnancy, childbirth, fetal, newborn, and postpartum development and care. They received March of Dimes and American College of Obstetricians and Gynecologists materials about eating and exercise during pregnancy. Participants in both groups received transportation and childcare for all study activities, monthly newsletters, attendance reminder cards and phone calls, small mother and baby care gift incentives after each intervention meeting, and $50 grocery store gift certificates after baseline and follow-up data collection. Graduation ceremonies followed program completion.

MOMs participants attended a mean of 8.1 meetings, with 98.6% attending at least 1 and 12.2% attending all 11 meetings. Of the 2 home visits, 95.7% received at least 1 and 93.5% received 2 visits. Of the 9 group meetings, the average attendance was 6.2, with 96.4% attending at least 1 and 12.2% attending all 9 meetings. MI participants attended an average of 1.5 of 3 meetings, with 84.6% attending at least 1 and 12.5% attending all 3 meetings. Additional study details were previously published.³⁰ The weekly pregnancy intervention meeting schedule and objectives are available in Appendix A (available as a supplement to the online version of this article at http://www.ajph.org). The curriculum is available at http://www.reachdetroit.org/about/HealthyMoms and http://www.michwa.org/for-chws-programs.

Data Collection

Data were collected at a community FQHC by trained interviewers employed by a separate institution with no involvement with administration or delivery of the MOMs or MI interventions and blinded to study assignment. Baseline data were collected before randomization and any intervention activities. Follow-up data were collected immediately following the last intervention session during pregnancy. Dietary intake data were collected using a food frequency questionnaire (FFQ) developed by the Nutrition Assessment Shared Resource of the Fred Hutchinson Cancer Research Center,³¹ which was validated in a Hispanic population.³² The Food and Nutrient Database of the Nutrient Data System for Research (software version 2008; developed by the Nutrition Coordinating Center, University of Minnesota, Minneapolis) was used for nutrient calculations.^33,34 At baseline (mean [SD] = 18.6 [4.5] weeks’ gestation), women described their dietary habits in the previous year, which corresponded to approximately 8 months preconception through 4 months gestation. At follow-up (mean [SD] = 29.2 [4.8] weeks’ gestation), women described their dietary habits in the previous 3 months.

The dietary outcomes were added sugars (grams), percentage of total calories from added sugars, total fat (grams), saturated fat (grams), percentage of total calories from saturated fat, fiber (grams), fruits (servings), vegetables (servings), and percentage of total calories from solid fats and added sugars (i.e., saturated fat, transfat, and added sugars). Outcomes estimated as a percentage of total calories were based on quantities reported in the FFQ.

Statistical Analysis

Baseline demographic characteristics of MOMs and MI participants were compared using Pearson’s χ² test, Fisher’s exact test, or Student’s t-test, as appropriate, with P < .05 considered statistically significant. All outcomes, except percentage of total calories from saturated fat, were log-transformed to reduce skewness.

Reported consumption of each dietary nutrient was analyzed using a linear mixed model in which baseline and follow-up values were included as the outcomes, with dummy variables for time (baseline or follow-up) and group (MOMs or MI) and their interaction included as predictors. Linear mixed models allow for correlation among observations on the same woman, and allow participants to be included in the analysis if they had data at baseline or follow-up^35–37 (Appendix B, available as a supplement to the online version of this article at http://www.ajph.org). A compound symmetric covariance structure was used for each outcome based on a comparison of the Akaike Information Criterion for competing covariance structures.^36,37

Changes in nutrient consumption (and 95% confidence intervals [CIs]) from baseline to follow-up were estimated for MOMs and MI participants using post hoc contrasts. The “intervention effect” was estimated as a contrast between the changes in nutrient consumption from baseline to follow-up for the 2 study groups. For log-transformed outcomes, these changes and the intervention effect were exponentiated and then expressed as a percentage change from baseline to follow-up and percentage change for MOMs versus MI participants, respectively.³⁸ Change in the only outcome analyzed on the original scale (percentage of calories from saturated fat) was expressed as difference between means.

The unadjusted intention-to-treat analysis included all available baseline and follow-up FFQs for all randomized participants. Adjusted analyses included covariates considered theoretically important (maternal age, education level, years lived in United States, care at FQHC, food stamp participation, WIC enrollment, parity, prepregnancy body mass index), and 1 covariate (English-speaking ability) that was different between MOMs and MI participants at baseline. The adjusted analysis excluded participants ineligible for randomization, those with missing covariates, and 1 baseline FFQ with unreasonably high values. Multicollinearity was assessed by computing the variance inflation factor for each covariate.

Logistic regression models, both unadjusted and adjusted, were used to analyze whether participants met the 2000 US dietary guidelines for vegetable consumption (≥ 3 vegetable servings) at follow-up. The data analysis was conducted using SAS version 9.3 (SAS Institute Inc., Cary, North Carolina).

RESULTS

Eligibility screening began in January 2004. The final intervention classes ended in October 2006. Although 278 women were randomized (Figure 1), 3 women were excluded from all analyses because they completed neither the baseline nor follow-up FFQ. The sample size for the unadjusted analysis was 275 women (MOMs, n = 139; MI, n = 136). Both baseline and follow-up FFQ data were available for 220 women; 48 women had only baseline FFQ data, and 7 women had only follow-up FFQ data. The adjusted analysis excluded data for 2 ineligible participants who were incorrectly included in the randomization process (1 multiple gestation and another enrolled at 35 weeks’ gestation), baseline data for 1 participant whose recorded intake was unreasonably high (8724 kcal/day, 6 SD above the mean for all participants), and 30 participants with missing covariate data.

FIGURE 1—

Flow of participants through Healthy Mothers on the Move (MOMs) study: Detroit, MI, 2004–2006.

Note. FFQ = food frequency questionnaire.

^aBaseline FFQ data removed for 1 woman with extreme values (8724 kcal/day); follow-up data included.

There were no significant differences in baseline sociodemographic characteristics, prenatal care initiation, or WIC or food stamp use between MOMs and MI participants, except for the percentage of women who spoke no English (MOMs 85% vs MI 75%; P < .05) (Table 1). MOMs and MI participants consumed similar amounts of all nutrients at baseline (Table 2).³⁹ Participants consumed an average of 2266 kcal/day, 26.6% of which were attributable to solid fats and added sugars. Participants consumed an average of 4.4 servings of fruit and 2.4 servings of vegetables per day. There was a small difference that approached significance between intervention groups in the mean percentage of total calories from saturated fat consumed (P = .06).

TABLE 1—

Selected Baseline Characteristics of Study Participants: Healthy Mothers on the Move (MOMs) Study, Detroit, MI, 2004–2006

	MOMs Group (n = 139),a Mean ±SD or No. (%)	MI Group (n = 139),a Mean ±SD or No. (%)	Pb
Age, y	27.3 ±5.3	27.1 ±5.2	.72
Education, y	9.1 ±3.2	9.4 ±3.0	.39
Country last educated			.29
Mainland United States	9 (6.7)	14 (10.7)
Mexico	117 (87.3)	105 (80.2)
Other	8 (6.0)	12 (9.2)
Birthplace			.27
Mainland United States	3 (2.2)	5 (3.7)
Mexico	128 (92.1)	118 (86.1)
Other	8 (5.8)	14 (10.2)
Age at arrival in United States,c y	21.1 ±5.5	21.3 ±5.8	.81
Years lived in United States	6.6 ±5.5	6.6 ±6.0	.99
Does not speak English at all	118 (84.9)	103 (75.2)	.04
Ethnicity: Mexican/Chicano	129 (92.8)	123 (89.8)	.37
Occupation: homemaker	125 (90.6)	120 (90.2)	.92
Married	84 (61.3)	76 (56.3)	.4
No. of previous pregnancies	1.5 ±1.3	1.4 ±1.1	.64
Prenatal care began in first trimester	98 (72.6)	106 (77.9)	.31
Prenatal care paid for by Medicaidd	129 (93.5)	119 (87.5)	.09
Received food stamps in past 6 mo	20 (14.6)	23 (17.3)	.55
Mother received WIC services in past 6 mo	112 (81.8)	101 (75.9)	.24
BMI,e kg/m2	24.2 ±5.1	24.7 ±5.0	.45f
BMIe			.64
Not overweight	79 (61.2)	70 (56.9)
Overweight	29 (22.5)	34 (27.6)
Obese	21 (16.3)	19 (15.4)

Note. BMI= body mass index; MI= minimal intervention; WIC= Special Supplemental Nutrition Program for Women, Infants and Children. The sample size was n = 278.

^aNumbers may not add to total because of missing data.

^bP values were obtained using the t-test for continuous variables, the Fisher exact test for categorical variables with small counts (ethic identity, birthplace), and the Pearson χ² test for all other categorical variables.

^cAmong women born outside of the United States (MOMs n = 134, MI n = 124)

^dIncludes those eligible only during pregnancy and delivery because of legal status.

^eBody mass index (BMI) was defined as estimated prepregnancy weight in kilograms divided by the square of height in meters. Overweight: BMI ≥ 25 and < 30 kg/m²; obese: BMI ≥ 30 kg/m².

^fP value from log transform.

TABLE 2—

Dietary Intake of Selected Nutrients for Study Participants at Baseline: Healthy Mothers on the Move (MOMs) Study, Detroit, MI, 2004–2006

	MOMs Group (n = 139), Mean (SD)	MI Group (n = 136), Mean (SD)	Pa
Calories, kcal	2194.9 (951.6)	2338.7 (1105.5)	.298
Fruit, servings	4.2 (3.0)	4.6 (3.3)	.325
Vegetable, servings	2.3 (1.4)	2.6 (1.9)	.438
Fiber, g	23.0 (10.6)	25.1 (12.4)	.219
Added sugar, g	72.0 (54.6)	74.9 (48.7)	.371
Percentage of total calories from added sugarb	12.7 (6.4)	12.9 (6.5)	.727
Total fat, g	82.0 (36.9)	86.6 (46.1)	.539
Saturated fat, g	29.8 (13.5)	31.0 (16.7)	.714
Percentage of total calories from solid fats and added sugarsc	26.7 (6.1)	26.4 (6.1)	.74
Percentage of total calories from saturated fatd	12.2 (1.9)	11.8 (1.9)	.062

Note. MI= minimal intervention. The sample size was n = 275. Mean gestation was 18.6 weeks.

^aP value of Student’s t-test; all nutrient intakes (except for percentage of calories from saturated fat) were log-transformed to reduce skewness.

^bPercentage of total calories from added sugar = [ (added sugar (g) × 4)/total calories ] x 100.³⁹

^cPercentage of total calories from solid fats and added sugars = {[(saturated fat [g] × 9) + (transfat [g] × 9) + (added sugar [g] × 4)]/total calories} x 100.³⁹

^dPercent of total calories from saturated fat = [(saturated fat [g] × 9)/total calories] x 100.³⁹

From baseline to follow-up, MOMs participants reported significantly improved intake of all selected nutrients, except fruit and fiber (Table 3). Mean consumption of added sugar decreased by 21.9% (P < .001) and saturated fat decreased by 18.1% (P < .001), whereas vegetable servings increased by 25.5% (P < .001). For MI participants, daily nutrient intake remained generally unchanged, except for decreases in mean fiber consumption by 9.0% (P = .026) and mean vegetable consumption by 11.5% (P = .052).

TABLE 3—

Unadjusted Analysis of Changes in Dietary Intake and Intervention Effect for Selected Nutrients Comparing Healthy Mothers on the Move (MOMs) Group vs Minimal Intervention (MI) Group: Detroit, MI, 2004–2006

	Changea		Intervention Effectb
	Follow-up – Baseline	P	Change (95% CI)	Pc
Calories, kcal
MOMs group	−11.0	.002	−7.3 (−16.5, 2.9)	.152
MI group	−4.0	.281
Fruit, servings
MOMs group	3.6	.525	3.3 (−11.5, 20.5)	.681
MI group	0.3	.958
Vegetables, servings
MOMs group	25.5	< .001	41.9 (19.2, 68.8)	< .001
MI group	−11.5	.052
Fiber, g
MOMs group	5.4	.21	15.9 (3.1, 30.3)	.014
MI group	−9.0	.026
Added sugar, g
MOMs group	−21.9	< .001	−16.1 (−29.6, −0.1)	.049
MI group	−6.9	.257
Percentage of total calories from added sugar
MOMs group	−12.4	.006	−9.7 (−21.0, 3.1)	.13
MI group	−3.0	.531
Total fat, g
MOMs group	−16.7	< .001	−12.9 (−22.0, −2.7)	.015
MI group	−4.4	.26
Total saturated fat, g
MOMs group	−18.1	< .001	−15.7 (−25.2, −5.0)	.005
MI group	−2.8	.508
Percentage of total calories from solid fats and added sugars
MOMs group	−10.4	< .001	−9.4 (−14.3, −4.3)	< .001
MI Group	−1.1	.585
Percentage of total calories from saturated fat
MOMs group	−0.9	< .001	−1.1 (−1.7, −0.5)	< .001
MI group	0.2	.392

Note. CI = confidence interval. Changes were measured from baseline (mean 18.6-week gestation) to follow-up (mean 29.2-week gestation). The sample size was n = 139 for the MOMs group and n = 136 for the MI group.

^aChange is defined as percent change from baseline to follow-up for all variables except percentage of calories from saturated fat, where change was the mean difference from baseline to follow-up on the original scale.

^bIntervention effect is defined as difference in percent change from baseline to follow-up for all variables, except percentage of calories from saturated fat, where intervention effect was the difference in mean change from baseline to follow-up on the original scale.

^cAnalysis is based on a linear mixed model carried out on the log scale for all variables except percentage of calories from saturated fat, which was analyzed on the original scale.

A significant intervention effect was found for vegetable consumption, with a 41.9% greater increase among MOMs than MI participants (P < .001; Table 3). Although the observed 5.4% increase in fiber intake for MOMs participants was not significant, it was significantly greater than the 9.0% decrease observed for MI participants (P = .014). Compared with MI participants at follow-up, MOMs participants significantly reduced their intake of added sugar (P = .049), total fat (P = .015), and saturated fat (P = .005); percentage of total calories from saturated fat (P < . 001); and percentage of total calories from solid fats and added sugars (P < .001). No intervention effect was observed for total calories, fruit, or percentage of calories from added sugar.

There was no significant difference between intervention groups in the percentage of women meeting daily vegetable consumption recommendations at baseline (MOMs = 30.3%, MI = 23.9%; P = .29). After adjusting for whether participants met the vegetable guideline at baseline, the odds for MOMs participants meeting this guideline at follow-up were almost twice as high as the odds for MI participants, although this result was not significant (odds ratio [OR] = 1.8; 95% CI = 0.9, 3.6; P = .09). After adjusting for participant characteristics, the result was still not significant (OR = 2.0; 95% CI = 0.9, 4.1; P = .1).

In the adjusted linear mixed model analysis (Table A3, Appendix C; available as a supplement to the online version of this article at http://www.ajph.org), conclusions were unchanged from the unadjusted intent-to-treat analysis, except the difference in added sugar consumption between MOMs and MI participants became marginally significant (P = .052 vs P = .049 in the unadjusted analysis). Because all variance inflation factors were less than 3, multicollinearity was not a problem.⁴⁰

DISCUSSION

This study supported the hypothesis that a community-planned, CHW-led healthy lifestyle intervention could improve dietary behaviors of low-income Latina women during pregnancy. MOMs participants showed improvement between baseline and follow-up for all dietary outcomes. These changes reached statistical significance for all outcomes, except for fruit and fiber. Compared with MI participants, MOMs participants had significantly increased daily consumption of vegetables and fiber and decreased consumption of added sugar, total and saturated fat, percentage of total calories from saturated fat, and percentage of total calories from solid fats and added sugars. Baseline fruit consumption was relatively high in both groups, possibly limiting the potential intervention effect on this outcome.

Average baseline vegetable consumption was below recommended levels⁴¹ for both MOMs and MI participants, as reported for Latinas and other US populations.^8,23,42–44 Mean vegetable consumption increased 25.5% among MOMs participants and increased 41.9% compared with the change among MI participants. Only 1 other randomized trial addressed dietary behaviors among pregnant Latinas.²³ Following specially designed nutrition education in an inner city clinic, vegetable consumption declined for both intervention and control participants, resulting in a nonsignificant intervention effect. Both groups had adequate levels of fruit consumption.

Adequacy of dietary intake might decline among low-income Latinas as duration of US residence increases.^{42,43,45–47} Acculturation to US dietary patterns often leads to high intakes of processed foods and high sugar beverages, particularly in low-income communities.⁴⁵ Our study participants consumed an average of 26.6% of their daily calories in solid fats and added sugars, which is twice the recommended level.⁴⁸ The American Heart Association recommends that the average adult woman consume a maximum of 25 grams (or 100 cal) per day of added sugar to achieve a healthy weight and decrease cardiovascular risk.⁴⁸ By comparison, the average daily intake of added sugar by MOMs group participants at baseline was 72.0 grams (288.0 cal); MI participants averaged 74.9 grams (299.6 cal). Although MOMs participants still exceeded the American Heart Association goal for added sugar intake at follow-up, the MOMs intervention reduced their intake by 21.9%. This effect would correspond to an 11-gram decrease for a woman consuming 50 grams of added sugar per day and a 21.9-gram decrease for a woman consuming 100 grams of added sugars per day. Overall, participants consumed an average of 30.4 grams of saturated fat at baseline. The MOMs intervention reduced participants’ intake by 18.1%. Added sugar and saturated fat intake during pregnancy are associated with gestational diabetes and postpartum weight retention, which are risk factors for type 2 diabetes.^12,13 The successful reduction of these dietary components by MOMs participants could contribute to reducing their risk of developing diabetes.

To our knowledge, this was the first full-scale randomized clinical trial that included among its aims the improvement of the dietary behaviors of pregnant Latinas. Strengths of this study included its randomized controlled design and basis in CBPR processes that led to development of its culturally and linguistically tailored curriculum, activities, structure, and staffing by CHWs.^25,26 Conducting the intervention in trusted community organizations and providing child care and transportation reduced participation barriers.^24–26 Our previous research found that positive and negative social support from husbands and other family members was an important influence on the dietary practices of pregnant Latinas.^24,26 Based on these findings, social support from CHWs and peers was integral to the MOMs intervention design. Social support was associated with diet quality during pregnancy among Latinas^42,43,49 and might counter the negative effects of acculturation on diet.⁴³ In the MOMs intervention, CHWs provided informational support during individual and group meetings. Activities fostered mutual problem solving that might have helped reduce family, social, and food access barriers.

Limitations

This study had several limitations. Dietary intake was measured by self-report, using an FFQ, which is useful for measuring intake during months-long time periods,⁵⁰ such as pregnancy.⁵¹ At baseline, we asked women to report their usual dietary intake during the previous 12 months. Because this assessment was conducted when women were, on average, in their fourth month of pregnancy, dietary changes during early pregnancy might have influenced their reporting. We would not expect this possible effect, or the effect of difference in length of recall for the baseline and follow-up FFQs, to differ between our study groups. This study did not include a food diary or biomarker of dietary intake. We could not exclude the possible impact of differential reporting bias on self-reported dietary intake between intervention groups after nutrition education. The results might not be generalizable to pregnant US-born Latinas, or those with greater levels of acculturation. More than 95% of our participants were born outside the United States, and most spoke only Spanish. Our study population was similar to a growing number of US communities with Mexican immigrant populations.⁵² Currently, 60% of US births to women of Mexican ancestry are to immigrant women.⁵³

The intention-to-treat study design specifically did not account for the number of participant contacts in assessing the impact of the intervention. Retention contacts were received in equal numbers by participants in both intervention arms. The greater number of intervention meetings and optional activity days could have contributed to greater participation and engagement by MOMs compared with MI participants. However, unlike the Diabetes Prevention Program, which provided intensive professional dietary counseling and dietary change supports to participants,⁵⁴ the MOMs intervention was successfully implemented by CHWs in community organization settings. The MOMs intervention might also be well suited to prenatal care and WIC clinics, Healthy Start programs, FQHCs, and other patient-centered medical homes. The Spanish language curriculum and 2 English language versions, aimed at English-speaking Latinas and African American women, were designed for use by CHWs and could be adapted for use by other educators and health care providers. CBPR approaches might help tailor these curricula to the social and environmental context of the communities in which they would be used.²⁵ Future research to study the impact of the MOMs intervention on the dietary behaviors of US-born Latinas, African American women, and other populations of pregnant women is recommended.

Conclusions

Our study added to the growing body of research demonstrating the effectiveness of trained CHWs in empowering residents of low-resource communities to achieve healthy behavior change,^55–57 in this case improved dietary habits during pregnancy. CHWs played essential roles in community-driven approaches to addressing health and health care disparities and were integral to the success of the MOMs intervention. CHWs are increasingly recognized as members of multidisciplinary health care teams and by health care institutions and policymakers as essential to comprehensive health care reform.^55–57