Lower Fat Intake Is Associated With Lower Weight and Cholesterol at 4 Months in Adults With Impaired Glucose Metabolism

DEPLOY study methods have been described extensively in a previous publication.⁸  They are briefly described here. In partnership with the YMCA, two sites were selected based on data from a 2003 YMCA primary market area analysis.⁸  The sites were matched by location in a semi-urban area and by similar racial and socioeconomic characteristics.⁸  One of two YMCA sites was randomly assigned to receive training and support for delivering a structured, group-based adaptation of the DPP lifestyle intervention called PLAN4WARD (P4W). All trainings of YMCA wellness or fitness coaches were conducted by experienced DPP instructors using an adapted DPP curriculum for group instruction.¹⁸  The second YMCA site served as a control.

A total of 535 adults attended several diabetes risk screening fairs at two participating YMCAs in greater Indianapolis, Ind., after three waves of bulk mailings to households within the catchment areas between August 2005 and May 2006. One hundred forty-three individuals met high-risk screening criteria, including elevated BMI (≥ 24 kg/m²), an American Diabetes Association diabetes risk questionnaire score ≥ 10,¹⁹  and abnormal capillary glucose concentration of 110–199 mg/dl (or 100–199 mg/dl if fasting for ≥ 9 hours). All risk assessments were conducted by Indiana University School of Medicine (IUSM) research staff.

Exclusion criteria included self-reported diabetes or any comorbidity that could limit life span to < 3 years or contraindicate gradual adoption of light-to-moderate physical activity (e.g., a recent cardiovascular event, severe chronic obstructive pulmonary disease, advanced arthritis, or poorly controlled hypertension).⁸  Ninety-two of the 131 eligible individuals decided to enroll in the 12-month study.

All measures were collected by the IUSM research staff. The primary endpoint was percentage of change in body weight at 4–6 months. Whole blood glucose concentration, A1C, total and HDL cholesterol, and systolic blood pressure levels were secondary endpoints. An overnight fast was not required for blood measurements because screenings occurred throughout the day; therefore, LDL cholesterol was not measured.²⁰  These results were published previously.⁸ 

Additional secondary endpoints included physical activity and several dietary components. At the baseline visit, physical activity was assessed using the Modifiable Activity Questionnaire (MAQ) developed by Kriska et al.²¹  Because of participant difficulty completing the MAQ as a self-report questionnaire, the International Physical Activity Questionnaire (IPAQ) short form²²  was substituted at 4–6 and 12–14 months. IUSM research staff also administered the version of the National Cancer Institute (NCI) Multifactor Screener²³  employed in the 2000 National Health Interview Survey at baseline, 4–6 months, and 12–14 months. Using this tool, screeners do not assess total diet calories, and no portion size questions are included. Estimates of fruit and vegetable servings, fiber grams, and PEF were calculated based on NCI scoring procedures.²⁴ 

During screening events, all study participants received brief advice (2–5 minutes) that moderate weight loss of 5–10% through calorie restriction and moderate physical activity such as daily brisk walking for 30 minutes was generally safe and effective in preventing or delaying the development of type 2 diabetes. Advice was supplemented with Small Steps, Big Rewards materials from the National Diabetes Education Program.²⁵  In addition, all study participants were introduced to YMCA programs that could help with weight loss and were offered repeat testing and additional brief lifestyle counseling after 4–6 and 12–14 months.

Intervention participants—those who consented to study participation at the YMCA site randomized to receive group DPP lifestyle training—also received free access to P4W classes. The P4W core curriculum, modeled on the DPP curriculum, included 16 small-group sessions lasting 60–90 minutes and was delivered over a 4- to 6-month period. Briefly, sessions focused on basic knowledge about nutrition, physical activity, and self-monitoring and the challenges of maintaining long-term diet and physical activity behaviors.^26,27  Self-monitoring tools (a food scale, measuring cups, measuring spoons, and a pocket-sized log for fat, calories, and minutes of physical activity) were provided at the beginning of the core curriculum, in addition to a DPP Lifestyle Balance Fat Counter booklet that included the fat and calorie content for various foods.²⁸  After this core curriculum, the YMCA offered less structured large-group education and support meetings with guest presenters every 4–6 weeks. The Indiana University Institutional Review Board approved the study.

Two sample t tests for continuous variables and χ² tests for categorical variables were used to determine differences in baseline characteristics between groups. Analysis of variance was used to test group differences in dietary intake variables at visits occurring 4–6 and 12–14 months after adjusting for baseline dietary intake, sex, age, and race (white vs. nonwhite). To test for differences between participants with and without 4-month follow-up visits, two sample t tests were used for all continuous variables except for PEF. Satterthwaite's test was used for PEF to allow for different group variances, and Fisher's exact tests were used for categorical variables because of small expected cell counts.

Because this was a pilot study involving only one YMCA in each study arm, we were unable to adjust standard errors for potential intracluster correlation. Furthermore, because of the limited statistical power for a small sample size, we also did not adjust for multiple comparisons in within-treatment tests. Mantel-Haenszel χ² tests were used to test group differences in ordered levels of physical activity at each follow-up visit.

Further analyses focused on PEF at 4 months because it was the only dietary variable to change significantly from baseline. Linear regression was used to assess the associations between change in PEF at 4 months and the outcomes of percentage change in weight and change in total cholesterol. Models were adjusted for sex, age, race (white vs. nonwhite), treatment group, IPAQ physical activity measurement at 4 months, and either baseline weight or total cholesterol. R² terms from models with and without PEF were used to determine the amount of variation in percentage of weight change and change in total cholesterol explained by change in PEF.

Finally, we conducted sensitivity analyses to determine the independent benefits of reduction in PEF on percentage of change in weight and change in total cholesterol using backward stepwise regressions. Baseline values of the respective outcome measures (i.e., weight and total cholesterol) and change in PEF at 4 months were forced to remain in the model. Candidate covariates (i.e., treatment group, race, sex, age, and 4-month physical activity levels [low, moderate, and high]) remained in the final model if the P value was < 0.10.

All analyses were performed using SAS version 9.3 (SAS Institute, Cary, N.C.) and included all participants for whom complete data were collected.

Compared to the intervention group, the control group had statistically significant lower baseline total cholesterol. No other variables were significantly different at baseline (Table 1).

From baseline to 4 months, both P4W participants and control participants reduced their intake of PEF significantly from baseline after adjusting for age, sex, race, and baseline value of the dependent variable. The P4W participants reduced fat intake more than control participants (−2.9 absolute change in percentage of fat intake vs. −1.3 absolute change in percentage of fat intake from baseline, treatment difference, P = 0.02). No within- or between-group differences in grams of fiber or fruit and vegetable servings achieved statistical significance at 4 and 12 months. After 12 months, within- and between-group differences in PEF compared to baseline did not achieve statistical significance (Table 2).

We did not observe between-group differences in physical activity measured by the IPAQ at 4 or 12 months (Table 3). We also did not observe statistically significant differences for any of the baseline characteristics between participants with and without 4-month follow-up visits (i.e., all P values were > 0.05).

In multivariate models adjusting for sex, age, race (white vs. nonwhite), treatment group assignment, physical activity at 4 months, and baseline value of the dependent variable, there was no statistically significant interaction effect between treatment group assignment and changes in PEF on the change in either body weight or total cholesterol at 4 months. However, within the P4W arm, change in PEF was a statistically significant predictor of both percentage of weight change and change in total cholesterol; every 10% absolute decrease in PEF was associated with a 3.9% decrease in body weight and a decrease in total cholesterol of 21.4 mg/dl (Table 4).

In sensitivity analyses, a reduction in PEF at 4 months was associated with a related percentage of weight reduction after controlling for baseline weight and treatment group (P = 0.03 for PEF, 0.07 for baseline weight, and 0.001 for treatment). Similarly, reduction in PEF at 4 months was associated with reduction in related total cholesterol after controlling for baseline total cholesterol, treatment group, and low activity versus moderate or high activity (P = 0.02 for PEF and P < 0.05 for covariates).

Within the intervention group, the change in PEF explained 24.3% of the variation in weight change and 14.5% of the variation in total cholesterol change at 4 months after adjusting for the covariates.

Our previous research demonstrated achievement and maintenance of meaningful levels of weight loss and reductions in overall cardiometabolic risk for DEPLOY community-based lifestyle participants compared to those who received standard advice.^8,14  Here, we explored whether a dietary strategy to reduce total calories, beginning with reduction in dietary fat, is responsible for these observations.

The DEPLOY study's community-based lifestyle intervention used the same dietary behavioral approaches to weight loss as those in the DPP, including provision of self-monitoring tools, instruction on healthier meal choices and preparation, and lowering total energy consumption by initially reducing total fat grams. Lifestyle intervention participants had a significantly greater reduction in dietary fat intake at 4 months than those receiving standard advice. However, we did not see a significant change in the number of daily fruit and vegetable servings or fiber intake at 4 months. Nevertheless, these observations are consistent with the primary focus of the intervention on reducing total caloric intake through first reducing intake of dietary fat.

We also demonstrated that, for participants in the lifestyle intervention arm, the change in PEF was associated with weight change at 4 months such that for every 10% absolute decrease in PEF, weight decreased by almost 4%. Furthermore, we observed that change in dietary fat was associated with changes in body weight and total cholesterol only in the DEPLOY intervention arm. This suggests that the modest weight loss experienced in the group receiving standard advice may have occurred through dietary behavior changes other than fat reduction or that associations were too small to observe in the context of more modest weight losses and a relatively small study sample size.

One year after randomization, lifestyle participants in the DPP clinical trial reported greater reductions in their percentage of energy from dietary fat, greater increases in number of daily fruit and vegetable servings, and greater fiber intake (g/day) compared to the other groups.⁶  In contrast, our participants did not report a significant change in number of fruit and vegetable servings or fiber intake 12 months after randomization to either group. In addition, we did not observe maintenance of lower PEF intake at 12 months, although weight loss was maintained in the lifestyle intervention group, which may have been the result, in part, of the limited intensity of the intervention between 4 and 12 months. Furthermore, participants may have made other dietary changes to maintain lower total caloric intake, such as consumption of smaller portion sizes or reduced snack frequency that could not be ascertained using the Multifactor Screener. Finally, in addition to attrition at 12 months, fewer of the remaining participants completed the Screener, which reduced the statistical power to detect significant dietary changes.

As previously mentioned, our participants used dietary self-monitoring tools during the lifestyle intervention, and food logs were regularly reviewed by DEPLOY lifestyle coaches. However, the food log data were not collected and analyzed for the study. Although these data might have provided information about adherence to dietary self-monitoring and its potential association with achievement and maintenance of weight loss, as seen in other studies,^7,13,17  they would not have provided any additional information about other nutrient intakes because the log was specifically for dietary fat.

A recent systematic review and meta-analysis²⁹  estimated that the incidence of type 2 diabetes may be reduced 14% by increasing daily intake of green leafy vegetables by 1.15 servings. However, no significant relationships were observed between incidence of type 2 diabetes and intake of fruit, all vegetables, or a combination of fruit and all vegetables.²⁹ 

Because the studies included in this meta-analysis used different dietary assessment methods such as self-administered food frequency questionnaires (FFQs), diet assessment interviews, and a single 24-hour recall, the results should be interpreted with caution. Also, included studies used different grouping methods (i.e., fruit only, vegetables only, and a combination of fruit and vegetables) and different criteria for green, leafy vegetables, and they adjusted for different potential confounders.

Another recent meta-analysis found that type of dietary fat, particularly polyunsaturated fat and possibly long-chain n-3 fatty acids, and carbohydrate such as minimally processed whole grains were more important than total fat and total carbohydrate as a proportion of total energy intake in the prevention of type 2 diabetes.³⁰ 

Both studies suggest the need for further investigation of relationships between dietary intake and prevention of type 2 diabetes in community settings using a combination of biochemical measures (i.e., serum carotenoids for fruit and vegetable intake and possibly fatty acid composition of serum cholesterol esters)^30,31  and brief dietary measures that increase quantifiable preciseness, assess total diet, and limit participant burden.

This study does have some limitations. The Multifactor Screener does not assess total diet, but rather approximates the number of daily fruit and vegetable servings, fiber intake, and PEF.²³  It may also be useful for tracking dietary changes over time in individuals. Yet, it does not track changes in portion size over time. Because the DPP intervention teaches both change in frequency of consumption and reduction in portion size as strategies to reduce intake of high-fat foods, the Multifactor Screener may have underestimated changes in fat intake, particularly for participants in the intervention group.

In contrast, the DPP Research Group used a 117-item semi-quantitative FFQ that required in-person interviews lasting ~ 45 minutes and coding of food items reported in open-ended questions.⁶  Although FFQs have several benefits, including assessment of total diet and usual intake as well as no effect on eating behavior, the 117-item interviewer-administered FFQ with open-ended questions increases investigator and respondent burden as well as investigator costs.³¹ 

Other limitations include the moderately high average nonresponse rates for diet survey items from baseline to 12 months (i.e., 4.7% for fruit and vegetable servings, 6.9% for PEF, and 7.0% for grams of fiber intake) and for physical activity assessment at 4 and 12 months for participants with follow-up visits (i.e., 5.5% at 4 months and 14.5% at 12 months), which further reduced the sample size in multivariate analyses.

Finally, the overall attrition rate was 33%. Although this pilot was not designed to measure program reach or optimal enrollment strategies, we found that 57% of P4W participants attended all 16 sessions.⁸  Nevertheless, these results should be confirmed with a larger sample of individuals in a study designed to find replicable and scalable enrollment strategies.

Structured diabetes prevention programs, including behavioral changes and weight loss of 5–7% of body weight, have been successfully translated in various community settings. This is the first study to address dietary intake associated with achievement and maintenance of weight loss and lower total cholesterol in a community-based diabetes prevention program. Although we found a significant association between PEF and weight loss and total cholesterol at 4 months, these associations were not maintained 12 months after randomization even though meaningful weight loss and lower total cholesterol were maintained in the intervention group.

Our results suggest that reducing total calories beginning with dietary fat may be an effective short-term dietary strategy for lowering total cholesterol and body weight in adults with abnormal glucose metabolism. However, a different dietary strategy may be required to maintain these cardiometabolic benefits in this population.

Support for this study was provided by the National Institute of Diabetes and Digestive and Kidney Diseases (R34 DK70702-02) and Indiana University School of Medicine. The authors recognize the support and participation of the YMCA of Greater Indianapolis and the involvement of all DEPLOY study participants.