Abstract
Background:
Intensive lifestyle interventions (ILI) for type 2 diabetes (T2D) improve health outcomes, but participants from different races/ethnicities or socioeconomic status may not benefit equally.
Methods:
Within the Look AHEAD trial, we examined achievement of the 7% weight loss goal, as well as secondary weight, cardiovascular, and mortality outcomes, by race/ethnicity and educational attainment (EA). Results: Among 4,640 participants (31 % Black or Hispanic, 13 % with less than a high school degree), Black and Hispanic participants were less likely than White participants to achieve 7 % weight loss in both the ILI (45.8 % v. 60.7 %, p < 0.001 and 53.0 % v. 60.7 %, p = 0.01, respectively) and diabetes support and education (DSE) arms. Contrastingly, participants with less than a high school degree were more likely in the ILI but less likely in the DSE arm to achieve this goal, with a significant arm by EA interaction. Hispanic participants and those with lowest EA also experienced decreased mortality in the ILI versus the DSE arm.
Conclusions:
All Look AHEAD participant subgroups achieved greater weight loss from ILI; however, Black and Hispanic participants lost less weight than White participants in both arms, while those with lowest EA benefited disproportionately from the ILI compared to participants with higher EA.
Keywords: Lifestyle intervention, Race/ethnicity, Socioeconomic status, Health disparity
1. Introduction
Intensive lifestyle interventions (ILI) are commonly recommended for individuals with pre-diabetes and type 2 diabetes (T2D) in order to improve weight, glycemic management, fitness, and other cardiovascular risk factors such as blood pressure and lipids [1–5]. The Action for Health in Diabetes (Look AHEAD) trial, one of the longest, largest, and most diverse ILI trials in patients with T2D, studied an ILI, as compared to diabetes support and education (DSE), among individuals with T2D who were motivated to lose weight [1]. However, despite robust improvements in cardiovascular risk factors, no differences in cardiovascular outcomes were seen in Look AHEAD after a median follow-up of approximately ten years.
Although ILI appear to benefit most if not all populations, there may not be equal benefit in all populations. Specifically, racial and ethnically minoritized populations in America who experience different life stressors (e.g., interpersonal and systemic racism) than White populations and individuals with lower socioeconomic status (SES) who are exposed to adverse social determinants of health (e.g., food insecurity, decreased healthcare access) may face greater challenges in implementing ILI successfully in usual care settings and therefore receive lesser benefit, as seen in some real-world studies of ILI [6–8].
Theoretically, the resources provided to participants within a clinical trial may serve to overcome some of these barriers (e.g., access to healthcare) and improve equity in care to all enrolled patients; however, this has not been widely studied. Specifically, in Look AHEAD, participants in the ILI were provided with meal replacements at no cost, and participants in both arms were provided with increased access to healthcare providers and educational and social support regarding healthy eating and exercise, albeit to a much greater degree in the ILI arm. These resources may have helped to overcome certain socioeconomic factors, such as low health literacy, limited access to care, and even food insecurity, but are unlikely to address all potential SES-related factors, such as housing insecurity or inadequate access to safe places to exercise.
In studies examining weight loss in only the ILI arm of the Look AHEAD trial, Black and Hispanic participants experienced less weight loss at 1 year [9,10] and decreased rates of sustained weight loss at 8 years [10] compared to White participants. However, no study has compared racial/ethnic or socioeconomic disparities in the DSE arm, of the Look AHEAD trial, which also provided improved access to educational resources and healthcare providers over usual care settings. Examining rates of successful weight loss in both arms will help to understand whether observed differences in weight loss in the ILI were related to limitations in implementing the ILI itself or more broadly reflected disparities which exist within a cohort of individuals with T2D motivated to lose weight.
Here, we examine weight-related outcomes in both arms of the Look AHEAD trial by race/ethnicity and by SES with the goal of understanding whether benefits associated with trial enrollment were equally distributed across participants in the trial. We hypothesized that Black and Hispanic participants and those of lower SES, as measured by self-reported educational attainment (EA), would be less likely to achieve weight loss goals in both the ILI and DSE arms of the trial due to greater barriers to weight loss and ILI implementation. Finally, we explore cardiovascular outcomes and mortality in each arm among participants stratified by race/ethnicity and by EA.
2. Subjects, materials, and methods
2.1. Data source, participants, and intervention
Longitudinal data from the Look AHEAD study were obtained from the National Institute for Diabetes and Digestive and Kidney Diseases (NIDDK) Central Data Repository [11]. Look AHEAD inclusion and exclusion criteria and study interventions have previously been described [12]. Briefly, Look AHEAD recruited adults aged 45–75 years of age with T2D, body mass index (BMI) ≥ 25.0 kg/m2, the ability to exercise, and willingness to lose weight, and without severely uncontrolled diabetes (hemoglobin A1c (HbA1c) ≤ 11.0 % (97 mmol/mol)), systolic blood pressure (SBP < 160 mmHg), or hypertriglyceridemia (triglycerides < 500 mg/dL). Participants were randomly assigned to receive an ILI including both group and individual sessions (weekly for months 1–6 after randomization, three times monthly for months 7–12, and at least monthly in subsequent years) focusing on caloric restriction, including the provision of meal replacements, and increased physical activity with the goal of achieving and maintaining weight loss of at least 7 %, or to DSE which included three group sessions annually, including lifestyle education and social support sessions.
2.2. Exposures, covariates, and outcomes
The primary exposures of this analysis were race/ethnicity and SES as measured by EA; EA was chosen instead of income and net worth due to low levels of missing data, common use as a measure of SES in previous literature, and strong association with outcomes in prior ILI studies [13,14]. Race/ethnicity was classified based on self-report using categories originally available in the Look AHEAD trial: White, Black, and Hispanic. Notably, participants recruited from Native American sites (approximately 5 % of the study cohort) were excluded from the Repository per consent limitations [15], and participants of Asian, Pacific Islander, Native American, and self-reported Other race/ethnicity were combined in the Repository data set as “Other” race. EA was based on self-reported highest level of education and was classified in the following categories: less than a high school degree, high school degree, some college, or college degree or higher. In this analysis, participants of “Other” or missing race/ethnicity (n = 174) or “other” EA (n = 100) were excluded due to small group size and high within-group heterogeneity.
Key covariates included age (in years), biological sex, baseline use of insulin upon study enrollment (as a measure of disease severity associated with diabetes duration and physiology), and trial arm (ILI vs DSE). All covariates were collected and coded by the Look AHEAD study team prior to distribution through the NIDDK Central Data Repository. Clinical site data are not included in the Repository data set and therefore were not included in this analysis.
The primary outcome of this analysis was achievement of the targeted 7 % weight loss goal at 1 year. Secondary outcomes included percent weight change 1 and 4 years after randomization, occurrence of the primary composite cardiovascular outcome of the Look AHEAD trial (including death from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke, or hospitalization for angina; henceforth “4-point major adverse cardiovascular event” (4p MACE)), and all-cause mortality.
2.3. Statistical analysis
Baseline characteristics are reported as median and interquartile range for continuous variables (as most variables were non-normally distributed, especially within race/ethnicity and socioeconomic strata) and as number and proportions for categorical variables. Univariate differences between trial arms and categories of race/ethnicity or EA are analyzed using pairwise chi square tests for categorical variables, t-tests for continuous variables, and log rank tests for time-to-event variables. Multivariable associations are examined using logistic, linear, and Cox proportional hazard regression models adjusted for age, sex, baseline insulin use, and randomization arm, with interactions between arm and race/ethnicity or arm and EA examined using a multiplicative interaction term. In the primary analysis, the association between race/ethnicity or EA and outcomes were examined in isolation as these variables are highly correlated; in sensitivity analyses, these associations were examined after adjustment for the other variable (e.g., the association between race/ethnicity and achievement of the weight loss goal, adjusted for EA). All analyses were performed using R version 4.0.2 (R Core Team; Vienna, Austria) [16].
3. Results
3.1. Baseline characteristics
After the exclusion of 266 participants categorized as “Other” race/ethnicity or EA, 4,640 participants were included (mean age 59 years, 58 % female, 17 % Black, 14 % Hispanic, 43 % with at least a college degree, and 13 % with less than a high school degree). Significant differences in baseline characteristics, including age, sex, diabetes duration, HbA1c, BMI, and diabetes medication use existed among participants of different race/ethnicity and of different EA (Table 1).
Table 1.
Baseline characteristics of participants enrolled in the Look AHEAD study, stratified by race/ethnicity and educational attainment.
| Stratification by race/ethnicity |
Stratification by educational attainment |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Variable | Total (n = 4,640) | White (n = 3,177) | Black (n = 793) | Hisp (n = 670) | p value | Less than HS Degree (n = 297) | HS Degree (n = 621) | Some College (n = 1,734) | College Degree or Higher (n = 1,988) | p value |
|
| ||||||||||
| Age (years), median (IQR) | 59 (55, 64) | 59 (55, 64) | 57 (54, 63) | 57 (54, 61) | <0.001 | 59 (55, 62) | 60 (55, 64) | 58 (54, 63) | 59 (55, 64) | 0.001 |
| Sex: Male, n (%) | 1940 (41.8 %) | 1558 (49.0 %) | 189 (23.8 %) | 193 (28.8 %) | <0.001 | 66 (22.2 %) | 173 (27.9 %) | 648 (37.4 %) | 1053 (53 %) | <0.001 |
| Race/ethnicity, n (%) | ||||||||||
| White | 3177 (68.5 %) | 3177 (100 %) | 0 (0 %) | 0 (0 %) | <0.001 | 44 (14.8 %) | 406 (65.4 %) | 1146 (66.1 %) | 1581 (79.5 %) | <0.001 |
| Black | 793 (17.1 %) | 0 (0 %) | 793 (100 %) | 0 (0 %) | <0.001 | 36 (12.1 %) | 106 (17.1 %) | 364 (21.0 %) | 287 (14.4 %) | <0.001 |
| Hispanic | 670 (14.4 %) | 0 (0 %) | 0 (0 %) | 670 (100 %) | <0.001 | 217 (73.1 %) | 109 (17.6 %) | 224 (12.9 %) | 120 (6 %) | <0.001 |
| Educational Attainment, n (%) | ||||||||||
| Less than HS degree | 297 (6.4 %) | 44 (1.4 %) | 36 (4.5 %) | 217 (32.4 %) | <0.001 | 297 (100 %) | 0 (0 %) | 0 (0 %) | 0 (0 %) | <0.001 |
| HS Degree | 621 (13.4 %) | 406 (12.8 %) | 106 (13.4 %) | 109 (16.3 %) | <0.001 | 0 (0 %) | 621 (100 %) | 0 (0 %) | 0 (0 %) | <0.001 |
| Some College | 1734 (37.4 %) | 1146 (36.1 %) | 364 (45.9 %) | 224 (33.4 %) | <0.001 | 0 (0 %) | 0 (0 %) | 1734 (100 %) | 0 (0 %) | <0.001 |
| College Degree or Higher | 1988 (42.8 %) | 1581 (49.8 %) | 287 (36.2 %) | 120 (17.9 %) | <0.001 | 0 (0 %) | 0 (0 %) | 0 (0 %) | 1988 (100 %) | <0.001 |
| Age at diabetes diagnosis (years), median (IQR) | 53 (47, 58) | 53 (48, 58) | 52 (46, 57) | 51 (45, 56) | <0.001 | 52 (46.5, 56) | 53.5 (47, 58) | 52 (46, 58) | 53 (47, 58) | 0.002 |
| Diabetes duration (years), median (IQR) | 5 (2, 10) | 5 (2, 10) | 5 (2, 10) | 5 (2, 10) | 0.785 | 6 (2, 11) | 5 (2, 10) | 5 (2, 10) | 5 (2, 9) | 0.010 |
| HbAlc (%; mmol/mol), median (IQR) | 7.1 (6.4, 7.9); 54 (46, 63) |
7.0 (6.4, 7.7); 53 (46, 61) |
7.2 (6.5, 8.1); 55 (48, 65) |
7.3 (6.6, 8.3); 56 (49, 67) |
<0.001 | 7.4 (6.7, 8.4); 57 (50, 68) |
7.2 (6.4, 8.0); 55 (46, 64) |
7.0 (6.5, 7.8); 53 (48, 62) |
7.0 (6.4, 7.8); 53 (46, 62) |
<0.001 |
| BMI (kg/m2), median (IQR) | 35.0 (31.6, 39.5) | 35.1 (31.6, 39.6) | 35.7 (31.9, 40.1) | 34.1 (31.2, 38.1) | <0.001 | 34.0 (31.0, 37.4) | 35.8 (32.1, 39.6) | 35.5 (32.0, 40.2) | 34.5 (31.1, 39.3) | <0.001 |
| Obese (BMI ≥ 30 kg/m2), n (%) | 3970 (85.6 %) | 2720 (85.6 %) | 692 (87.3 %) | 558 (83.3 %) | 0.096 | 249 (83.8 %) | 541 (87.1 %) | 1514 (87.3 %) | 1666 (83.8 %) | 0.010 |
| Morbidly Obese (BMI ≥ 40 kg/m2), n (%) | 1049 (22.6 %) | 729 (22.9 %) | 202 (25.5 %) | 118 (17.6 %) | 0.001 | 45 (15.2 %) | 145 (23.3 %) | 444 (25.6 %) | 415 (20.9 %) | <0.001 |
| Diabetes Medication Use, n (%) | ||||||||||
| Any insulin | 710 (15.3 %) | 442 (13.9 %) | 149 (18.8 %) | 119 (17.8 %) | <0.001 | 64 (21.5 %) | 98 (15.8 %) | 267 (15.4 %) | 281 (14.1 %) | <0.001 |
| Metformin | 2815 (60.7 %) | 1950 (61.4 %) | 465 (58.6 %) | 400 (59.7 %) | 0.024 | 173 (58.2 %) | 365 (58.8 %) | 1060 (61.1 %) | 1217 (61.2 %) | 0.024 |
| Sulfonylurea | 2109 (45.5 %) | 1401 (44.1 %) | 364 (45.9 %) | 344 (51.3 %) | <0.001 | 164 (55.2 %) | 303 (48.8 %) | 785 (45.3 %) | 857 (43.1 %) | <0.001 |
| TZD | 1226 (26.4 %) | 896 (28.2 %) | 211 (26.6 %) | 119 (17.8 %) | <0.001 | 43 (14.5 %) | 151 (24.3 %) | 488 (28.1 %) | 544 (27.4 %) | <0.001 |
| Any diabetes medication | 4061 (87.5 %) | 2735 (86.1 %) | 716 (90.3 %) | 610 (91.0 %) | <0.001 | 267 (89.9 %) | 559 (90.0 %) | 1528 (88.1 %) | 1707 (85.9 %) | 0.014 |
| Cardiovascular | ||||||||||
| Medication Use, n (%) | ||||||||||
| Any BP medication | 3426 (73.8 %) | 2332 (73.4 %) | 645 (81.3 %) | 449 (67.0 %) | <0.001 | 215 (72.4 %) | 457 (73.6 %) | 1308 (75.4 %) | 1446 (72.7 %) | 0.624 |
| Aspirin | 2079 (44.8 %) | 1571 (49.4 %) | 294 (37.1 %) | 214 (31.9 %) | <0.001 | 105 (35.4 %) | 233 (37.5 %) | 758 (43.7 %) | 983 (49.4 %) | <0.001 |
| Statin | 2088 (45.0 %) | 1573 (49.5 %) | 286 (36.1 %) | 229 (34.2 %) | <0.001 | 100 (33.7 %) | 234 (37.7 %) | 805 (46.4 %) | 949 (47.7 %) | <0.001 |
| History of Cardiovascular Disease, n (%) | 637 (13.7 %) | 498 (15.7 %) | 76 (9.6 %) | 63 (9.4 %) | <0.001 | 31 (10.4 %) | 87 (14.0 %) | 252 (14.5 %) | 267 (13.4 %) | 0.278 |
HbA1c = hemoglobin A1c; BMI = body mass index; TZD = thiazolidinedione; BP = blood pressure; CVD = cardiovascular disease.
3.2. Racial/ethnic and socioeconomic disparities in achievement of the weight loss goal and percent weight change
Within categories of race/ethnicity or EA, all groups experienced greater achievement of the weight loss goal and greater percent weight loss at 1 and 4 years in the ILI than in the DSE arm (Supplemental Table 1).
Within the ILI arm, Black and Hispanic participants were less likely to achieve the 7 % weight loss goal than White participants in univariate analyses (45.8 % v. 60.7 %, pairwise p < 0.001 and 53.0 % v. 60.7 %, p = 0.014, respectively; Fig. 1a, Supplemental Table 1). Similarly, in the DSE arm, Black participants were less likely to achieve the weight loss goal than White participants (3.8 % v. 8.3 %, p = 0.005). In models adjusted for age, biological sex, baseline insulin use, and randomization arm, both Black and Hispanic participants were less likely to achieve the weight loss goal (Table 2), with no evidence for an interaction between race/ethnicity and randomization arm. Findings were similar for percent weight loss at 1 year, although these differences were largely attenuated at 4 years (Supplemental Tables 1 and 2).
Fig. 1.
Percent achievement of the 7 % weight loss goal at 1 year, stratified by arm and (A) race/ethnicity, and (B) educational attainment.
Table 2.
Multivariable associations between achievement of the weight loss goal and cardiovascular outcomes, adjusting for age, sex, randomization arm, and race/ethnicity or SES, with examination interactions between arm and race/ethnicity.
| Multivariable Models of Race/Ethnicity | ||||||
|---|---|---|---|---|---|---|
| Outcome | 7 % Weight Loss Goal 1 yr |
4p MACE |
Mortality |
|||
| Model | Model 1 | Model 2 | Model 1 | Model 2 | Model 1 | Model 2 |
|
| ||||||
| Intercept | 0.03 (0.01, 0.05), p < 0.001 | 0.03 (0.01, 0.05), p < 0.001 | ||||
| Age | 1.02 (1.01, 1.03), p < 0.001 | 1.02 (1.01, 1.03), p < 0.001 | 1.06 (1.04, 1.07), p < 0.001 | 1.06 (1.04, 1.07), p < 0.001 | 1.08 (1.07, 1.10), p < 0.001 | 1.08 (1.07, 1.10), p < 0.001 |
| Sex: Male | 0.96 (0.82, 1.12), p = 0.619 | 0.96 (0.82, 1.12), p = 0.616 | 1.97 (1.69, 2.29), p = 0 | 1.96 (1.69, 2.28), p = 0 | 1.57 (1.26, 1.95), p < 0.001 | 1.57 (1.26, 1.95), p < 0.001 |
| Insulin use | 0.86 (0.70, 1.04), p = 0.126 | 0.86 (0.70, 1.04), p = 0.126 | 1.42 (1.20, 1.68), p < 0.001 | 1.42 (1.20, 1.68), p < 0.001 | 1.82 (1.44, 2.30), p < 0.001 | 1.82 (1.44, 2.30), p < 0.001 |
| Arm: ILI | 18.09 (15.07, 21.85), p < 0.001 | 17.35 (14.1, 21.51), p < 0.001 | 0.98 (0.85, 1.13), p = 0.792 | 0.95 (0.81, 1.12), p = 0.567 | 0.89 (0.72, 1.10), p = 0.292 | 0.94 (0.74, 1.19), p = 0.592 |
| Race/ethnicity (ref = White) | ||||||
| Black | 0.54 (0.44, 0.67), p < 0.001 | 0.45 (0.24, 0.76), p = 0.006 | 0.92 (0.74, 1.13), p = 0.424 | 0.75 (0.55, 1.03), p = 0.074 | 0.91 (0.66, 1.24), p = 0.532 | 0.92 (0.60, 1.41), p = 0.705 |
| Hispanic | 0.75 (0.60, 0.93), p = 0.011 | 0.67 (0.38, 1.1), p = 0.135 | 0.75 (0.59, 0.97), p = 0.025 | 0.85 (0.61, 1.17), p = 0.315 | 0.76 (0.53, 1.10), p = 0.141 | 0.91 (0.58, 1.44), p = 0.692 |
| Arm by Race/Ethnicity Interactions | ||||||
| ILI * Black | 1.25 (0.70, 2.39), p = 0.467 | 1.46 (0.96, 2.22), p = 0.075 | 0.96 (0.52, 1.77), p = 0.904 | |||
| ILI * Hispanic | 1.15 (0.66, 2.10), p = 0.645 | 0.76 (0.46, 1.26), p = 0.289 | 0.62 (0.29, 1.34), p = 0.225 | |||
|
| ||||||
| Multivariable Models of Educational Attainment | ||||||
| Outcome | 7 % Weight Loss Goal 1 yr |
4p MACE |
Mortality |
|||
| Model | Model 1 | Model 2 | Model 1 | Model 2 | Model 1 | Model 2 |
|
| ||||||
| Intercept | 0.02 (0.01, 0.05), p < 0.001 | 0.01 (0, 0.03), p < 0.001 | ||||
| Age | 1.02 (1.01, 1.04), p < 0.001 | 1.02 (1.01, 1.04), p < 0.001 | 1.06 (1.04, 1.07), p < 0.001 | 1.06 (1.05, 1.07), p < 0.001 | 1.09 (1.07, 1.10), p < 0.001 | 1.09 (1.07, 1.10), p < 0.001 |
| Sex: Male | 1.03 (0.88, 1.21), p = 0.682 | 1.03 (0.88, 1.21), p = 0.705 | 2.06 (1.77, 2.39), p < 0.001 | 2.06 (1.77, 2.39), p < 0.001 | 1.62 (1.30, 2.02), p < 0.001 | 1.62 (1.30, 2.02), p < 0.001 |
| Insulin use | 0.83 (0.68, 1.01), p = 0.067 | 0.83 (0.68, 1.01), p = 0.067 | 1.39 (1.17, 1.64), p < 0.001 | 1.39 (1.18, 1.65), p < 0.001 | 1.79 (1.42, 2.26), p < 0.001 | 1.79 (1.42, 2.26), p < 0.001 |
| Arm: ILI | 17.71 (14.76, 21.37), p < 0.001 | 61.82 (24.12, 210.62), p < 0.001 | 0.98 (0.85, 1.14), p = 0.823 | 0.82 (0.39, 1.73), p = 0.604 | 0.9 (0.73, 1.11), p = 0.316 | 0.36 (0.12, 1.14), p = 0.082 |
| EA (Ref = Less than HS degree) | ||||||
| HS Degree | 0.72 (0.50, 1.03), p = 0.075 | 1.75 (0.63, 6.19), p = 0.323 | 1.51 (0.99, 2.29), p = 0.054 | 1.22 (0.68, 2.19), p = 0.502 | 1.13 (0.63, 2.03), p = 0.692 | 0.69 (0.33, 1.44), p = 0.321 |
| Some College | 0.70 (0.51, 0.98), p = 0.035 | 2.17 (0.87, 7.27), p = 0.141 | 1.61 (1.09, 2.38), p = 0.016 | 1.59 (0.94, 2.70), p = 0.086 | 1.36 (0.80, 2.32), p = 0.257 | 1.00 (0.53, 1.89), p = 0.99 |
| College Degree or higher | 0.85 (0.62, 1.18), p = 0.335 | 2.61 (1.06, 8.67), p = 0.067 | 1.3 (0.88, 1.92), p = 0.187 | 1.14 (0.67, 1.94), p = 0.628 | 1.15 (0.67, 1.96), p = 0.619 | 0.8 (0.43, 1.52), p = 0.501 |
| Arm by EA Interactions | ||||||
| ILI * HS Degree | 0.26 (0.07, 0.70), p = 0.016 | 1.52 (0.66, 3.52), p = 0.327 | 3.42 (0.94, 12.46), p = 0.062 | |||
| ILI * Some College | 0.35 (0.09, 1.05), p = 0.079 | 1.03 (0.47, 2.23), p = 0.947 | 2.33 (0.71, 7.68), p = 0.164 | |||
| ILI * College Degree or higher | 0.26 (0.07, 0.70), p = 0.015 | 1.30 (0.60, 2.83), p = 0.503 | 2.60 (0.79, 8.52), p = 0.116 | |||
Model 1 = adjusted for age, sex, baseline insulin use, randomization arm, and exposure (either race/ethnicity or educational attainment).
Model 2 = adjusted for age, sex, baseline insulin use, randomization arm, exposure (either race/ethnicity or educational attainment), and arm-by-exposure (either race/ethnicity or educational attainment) interactions.
4p MACE = four-point major adverse cardiovascular events (death from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke, or hospitalization for angina); EA = educational attainment; ILI = intensive lifestyle intervention; HS = high school.
Within the ILI arm, participants with less than a high school degree were more likely to achieve the 7 % weight loss goal than those with a high school degree (66.9 % v. 56.1 %, pairwise p = 0.040, respectively) or some college (66.9 % v. 53.7 %, p = 0.005), with a non-significant trend towards greater goal achievement compared to participants with a college degree, as well (66.9 % v. 58.8 %, p = 0.080; Fig. 1b, Supplemental Table 1). No significant difference in achievement of the weight loss goal by EA was observed in the DSE arm, although participants with less than a high school degree had the numerically least likelihood of goal achievement. Findings for secondary outcomes (percent weight loss at 1 year and 4 years after randomization) were similar, with participants with less than a high school degree achieving the greatest weight loss in the ILI arm and the least weight loss in the DSE arm, although associations did not uniformly achieve statistical significance (Supplemental Table 1). In models adjusted for age, biological sex, baseline insulin use, and randomization arm, there was evidence for an arm by EA interaction suggesting that participants with less than a high school degree received relatively greater benefit (greater likelihood of achieving the weight loss goal) from the ILI arm than those with a high school or college degree (Table 2), with similar findings for secondary outcomes (percent weight loss at 1 and 4 years; Supplemental Table 3).
In sensitivity analyses adjusting for both race/ethnicity and EA in the same models, associations seen in the primary multivariable analyses persisted (Supplemental Table 4). In brief, Black race/ethnicity was associated with lower rates of weight loss goal achievement, and arm by EA interactions suggested greater relative benefit of the ILI among participants with less than a high school degree compared to those with higher EA.
3.3. Racial/ethnic and socioeconomic disparities in all-cause mortality and 4p MACE
Although no difference was found in rates of all-cause mortality or 4p MACE between the ILI and DSE arms in the primary Look AHEAD study, when examined within categories of race/ethnicity, lower rates of mortality were seen among Hispanic participants in the ILI than those in the DSE (p = 0.045; Fig. 2a; Supplemental Table 1). Similarly, participants with less than a high school education experienced lower mortality in the ILI than in the DSE (p = 0.041; Fig. 2b; Supplemental Table 1). In multivariable models, a non-significant trend suggested an arm by EA interaction with greater mortality benefit from the ILI among those with less than a high school degree compared to those with higher EA (p = 0.062 vs HS degree; p = 0.164 vs some college; p = 0.116 vs college; Table 2). This trend persisted in sensitivity analyses which were also adjusted for race/ethnicity (Supplemental Table 4).
Fig. 2.
Time to event mortality curves, stratified by arm and (A) race/ethnicity, and (B) educational attainment.
Within both the ILI and DSE arms, rates of 4p MACE were highest among White participants and lowest among those with less than a high school degree (Supplemental Table 1), likely related to baseline differences such as age and history of cardiovascular disease. In multivariable models adjusted for age, sex, baseline insulin use, and intervention arm, and examining for arm by race/ethnicity or arm by EA interactions, neither race/ethnicity nor EA was associated with 4p MACE within either arm (Table 2).
4. Discussion
In this analysis examining racial/ethnic and socioeconomic disparities in weight, cardiovascular events, and mortality in the Look AHEAD trial, we found that Black and Hispanic participants experienced lesser weight benefit from enrollment in either the ILI or DSE arms of the trial than White participants. By contrast, participants with the lowest EA experienced greater weight benefit in the ILI arm of the trial, including both greater initial weight loss and sustained weight loss at 4 years, but lesser weight loss in the DSE arm when compared to participants with higher EA, with a significant interaction between arm and EA. Although no differences in cardiovascular or mortality outcomes were seen between the ILI and DSE arms of the trial overall, Hispanic participants and those with less than a high school degree experienced lower rates of all-cause mortality in the ILI arm than the DSE arm in univariate analyses, with a trend towards an arm by EA interaction in multivariable analyses. These findings identify a population which may receive greater benefit from ILI, a critical finding given that this population already faces inequities in unhealthy lifestyle characteristics [17,18], and targeted ILI have been successful in underserved groups [19–24]. While subgroup analyses should be interpreted as hypothesis-generating with limitations based on smaller populations within strata and potential residual confounding, Look AHEAD, as one of the largest and longest randomized controlled trials of an ILI, offers a unique opportunity to study critically important variables that are less well assessed in real-world environments.
Previous studies have noted racial/ethnic disparities in weight loss in the ILI of the Look AHEAD trial but have not examined outcomes in the DSE arm [9,10]. Our analysis revealed that Black and Hispanic participants also achieved lesser weight loss in the DSE arm of the trial. As all participants were motivated to lose weight at baseline and received at least some educational and social support resources related to lifestyle change, the fact that White participants lost more weight in both arms suggests that racial/ethnic disparities in weight loss were not present solely in the intervention arm and that the benefits of trial enrollment and resources were lesser for Black and Hispanic participants, regardless of randomization arm.
By contrast, differences in weight loss by EA diverged based on the arm to which the patient was randomly assigned, with evidence of an arm by EA interaction. This may represent a greater relative benefit, as compared to less intensive support, provided by the ILI in groups with low EA. As a possible explanation, participants with low EA and health literacy may have limited access to reliable health information under usual care and/or may not have been exposed to significant counseling related to healthy lifestyle choices prior to study enrollment, leading to substantial improvements in health behaviors among those provided with resources through the Look AHEAD ILI. Specifically in this trial, the provision of meal replacements may also have helped to overcome food insecurity or exposure to neighborhood “food deserts” or “food swamps” for participants living in low SES areas. By contrast, participants with a higher level of EA may have had prior exposure and ongoing access to reliable health information, regardless of enrollment in an ILI, leading to lesser relative benefit within an ILI and/or greater likelihood of achieving weight loss outside the structure of an ILI. Nevertheless, it is important to note that all strata of EA who were enrolled in the ILI arm experienced significantly greater weight loss than in the DSE arm, suggesting benefits of structured ILI in all groups.
Race/ethnicity and EA are highly associated in the United States and in the Look AHEAD study, with Hispanic participants having the lowest EA (see Table 1). This association might suggest difficulty in identifying the true factors suggestive of greater or lesser benefit among trial participants. However, the relative benefit of ILI in these groups diverged in the Look AHEAD population, demonstrating lesser weight loss among Black and Hispanic participants but greater weight loss among those with less than a high school degree. Further, these associations, including significant arm by EA interactions, persisted in sensitivity analyses adjusted for both race/ethnicity and EA. This is a novel finding and suggests that race/ethnicity and EA represent unique factors, or perhaps proxies for other associated factors, which may contribute independently and divergently to ILI response.
Few studies have specifically examined the role of SES within clinical trials, and results have been heterogeneous. For example, in the Diabetes Prevention Program, higher educational attainment, dichotomized at college education or greater, was associated with greater risk reduction from either ILI or metformin as compared to placebo [13]. The Special Diabetes Program for Indians Diabetes Prevention Program (SDPI-DP Program) similarly found greater weight loss among those with greater household income [8]. In the REAL HEALTH study, no heterogeneity of treatment effect was seen by either food insecurity status or income [25]. Studies that evaluated educational attainment in low-income populations, especially looking at very low educational attainment, have been more consistent. In the SDPI-DP Program, although higher income was associated with greater weight loss, educational attainment was also associated with weight loss, with those with less than a high school degree experiencing the greatest benefit [8]. Lower levels of educational attainment, measured in four or more categories, were also associated with greater immediate improvements in health behaviors, especially for those with educational attainment less than high school, in the Multisite Cardiac Lifestyle Intervention Program [26] and Familias Unidas for Health and Wellness study [27]. Differences in intervention characteristics, included participants (e.g., people with pre-diabetes v. diabetes), and choice of socioeconomic measure (e.g., educational attainment cutoffs at high school v. college) may lead to these different findings regarding the efficacy of ILI in subgroups with different SES. Based on our findings, in which those with less than a high school degree received the greatest benefit, followed by those with a college degree or more, with the intermediate educational attainment categories falling last, we believe that examining educational attainment only based on the presence or absence of a college degree may obscure more granular differences between groups.
Results of this analysis related to cardiovascular outcomes and mortality must be interpreted cautiously due to small sample size of subgroups, low event rates, and p values which were not adjusted for multiple testing. Given these limitations, the finding of decreased mortality in the ILI as compared to the DSE arm among Hispanic participants and those with less than a high school degree in univariate analyses require validation in similar studies and should be considered hypothesis-generating. As those with less than a high school degree did experience the greatest weight loss, with evidence of an arm by EA interaction, and as previous studies have demonstrated cardiovascular benefits of the ILI among participants who were able to achieve > 10 % weight loss [28], it is possible that the high levels of weight loss seen in this group may underlie the observed decrease in mortality in the ILI arm. Alternatively, differences in baseline characteristics such as risk factors for non-cardiovascular mortality or differences in the ILI delivery used by study sites serving this population may explain the observed mortality benefit of ILI in this group. Finally, it cannot be ruled out that this association represents merely a false positive association due to multiple testing. Further analyses are needed in other studies of ILI in order to inform and contextualize these preliminary findings.
This analysis had several limitations. First, although Look AHEAD was a randomized trial, this secondary analysis of non-randomized subgroups can only describe associations and cannot prove causal relationships. Look AHEAD is a completed study with a fixed sample size which was not powered to detect differences between subgroups, especially for the infrequent outcome of mortality, which may lead this analysis to be underpowered to detect significant differences by race/ethnicity or EA. It is likely that race/ethnicity, EA, and the day-to-day details of the intervention differed by enrollment site; however, site data are not included in the NIDDK data repository, so we were unable to adjust for this potential confounder. Finally, as analyses were not adjusted for multiple testing and relied on nominally significant p values, associations should be interpreted with caution given an elevated risk for false positives.
In summary, analyses of weight outcomes, cardiovascular events, and mortality from both arms of the Look AHEAD trial by race/ethnicity or EA revealed significant disparities, with Black and Hispanic participants generally experiencing lesser weight loss in both the ILI and DSE arm of the trial than White participants, but with those with the lowest EA (less than a high school degree) experiencing lesser benefits in the DSE arm but superior benefits in the ILI arm compared to those with higher EA. These findings help to demonstrate the relative benefit of ILI over usual care in different populations and notably identify individuals with low EA as a group which may receive disproportionate benefit from ILI.
Supplementary Material
3
Acknowledgements
The Look AHEAD trial was conducted by the Look AHEAD Investigators and supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The data from the Look AHEAD trial reported here were supplied by the NIDDK Central Repository. This manuscript was not prepared in collaboration with Investigators of the Look AHEAD study and does not necessarily reflect the opinions or views of the Look AHEAD study, the NIDDK Central Repository, or the NIDDK.
Funding source
SJC was supported by the National Institutes of Health (grant number F32DK127545). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Abbreviations:
- T2D
type 2 diabetes
- ILII
intensive lifestyle intervention
- DSE
diabetes support and education
- Look AHEAD
Action for Health in Diabetes
- HS
high school
- SES
socioeconomic status
Footnotes
Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: ‘SJC reports a close family member employed by a Johnson & Johnson company. JBM is an Academic Associate with Quest Diagnostics. DJW reports serving on Data Monitoring Committees for Novo Nordisk.’
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://doi.org/10.1016/j.diabres.2022.110095.
References
- [1].The Look AHEAD Research Group. Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med 2013;369:145–54. 10.1056/NEJMoa1212914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Look AHEAD Research Group, Wing RR. Long-term effects of a lifestyle intervention on weight and cardiovascular risk factors in individuals with type 2 diabetes mellitus: four-year results of the Look AHEAD trial. Arch Intern Med 2010; 170:1566–75. 10.1001/archinternmed.2010.334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Mensink M, Feskens EJM, Saris WHM, de Bruin TWA, Blaak EE. Study on Lifestyle Intervention and Impaired Glucose Tolerance Maastricht (SLIM): preliminary results after one year. Int J Obesity 2003;27:377. 10.1038/sj.ijo.0802249. [DOI] [PubMed] [Google Scholar]
- [4].Goldberg RB, Temprosa M, Haffner S, Orchard TJ, Ratner RE, Fowler SE, et al. Effect of progression from impaired glucose tolerance to diabetes on cardiovascular risk factors and its amelioration by lifestyle and metformin intervention: the diabetes prevention program randomized trial by the diabetes prevention program research group*. Diabetes Care 2009;32:726–32. 10.2337/dc08-0494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].The Diabetes Prevention Program Research Group. Impact of intensive lifestyle and metformin therapy on cardiovascular disease risk factors in the diabetes prevention program. Diabetes Care 2005;28:888–94. 10.2337/diacare.28.4.888. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Ely EK, Gruss SM, Luman ET, Gregg EW, Ali MK, Nhim K, et al. A national effort to prevent type 2 diabetes: participant-level evaluation of CDC’s national diabetes prevention program. Diabetes Care 2017;40:1331–41. 10.2337/dc16-2099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Chen M, Moran LJ, Harrison CL, Ukke GG, Sood S, Bennett CJ, et al. Ethnic differences in response to lifestyle intervention for the prevention of type 2 diabetes in adults: a systematic review and meta-analysis. Obes Rev 2022;23: e13340. 10.1111/obr.13340. [DOI] [PubMed] [Google Scholar]
- [8].Jiang L, Huang H, Johnson A, Dill EJ, Beals J, Manson SM, et al. Socioeconomic disparities in weight and behavioral outcomes among American Indian and Alaska native participants of a translational lifestyle intervention project. Diabetes Care 2015;38:2090–9. 10.2337/dc15-0394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Wadden TA, West DS, Neiberg R, Wing RR, Ryan DH, Johnson KC, et al. One-year weight losses in the look AHEAD study: factors associated with success. Obesity (Silver Spring) 2009;17:713–22. 10.1038/oby.2008.637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].West DS, Dutton G, Delahanty LM, Hazuda HP, Rickman AD, Knowler WC, et al. Weight loss experiences of African American, hispanic, and non-hispanic white men and women with type 2 diabetes: the look AHEAD trial. Obesity 2019;27: 1275–84. 10.1002/oby.22522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].NIDDK Central Repository. n.d. Available from: https://repository.niddk.nih.gov/home/ (accessed January 26, 2022). [Google Scholar]
- [12].The Look AHEAD Study. A description of the lifestyle intervention and the evidence supporting it. Obesity (Silver Spring) 2006;14:737–52. 10.1038/oby.2006.84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].O’Brien MJ, Whitaker RC, Yu D, Ackermann RT. The comparative efficacy of lifestyle intervention and metformin by educational attainment in the Diabetes Prevention Program. Prev Med 2015;77:125–30. 10.1016/j.ypmed.2015.05.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Rautio N, Jokelainen J, Saaristo T, Oksa H, Keinänen-Kiukaanniemi S, FIN-D2D Writing Group, et al. Predictors of success of a lifestyle intervention in relation to weight loss and improvement in glucose tolerance among individuals at high risk for type 2 diabetes: the FIN-D2D project. J Prim Care Community Health 2013;4: 59–66. doi: 10.1177/2150131912444130. [DOI] [PubMed] [Google Scholar]
- [15].Look AHEAD Trial Public Data. n.d. Available from: https://www.lookaheadtrial.org/publicData.cfm (accessed January 26, 2022).
- [16].R Core Team. R: A language and environment for statistical computing; 2020. n.d. [Google Scholar]
- [17].Hall G, Arena R, Severin R. Social justice as it relates to healthy living behaviors and medicine: the current state of inequity and the path forward. Prog Cardiovasc Dis 2022;71:1–3. 10.1016/j.pcad.2022.04.002. [DOI] [PubMed] [Google Scholar]
- [18].Myers J, Harber MP, Johnson L, Arena R, Kaminsky LA. Current state of unhealthy living characteristics in White, African American and Latinx populations. Prog Cardiovasc Dis 2022;71:20–6. 10.1016/j.pcad.2022.05.002. [DOI] [PubMed] [Google Scholar]
- [19].Delahanty LM, Levy DE, Chang Y, Porneala BC, Goldman V, McCarthy J, et al. Effectiveness of lifestyle intervention for type 2 diabetes in primary care: the REAL HEALTH-diabetes randomized clinical trial. J Gen Intern Med 2020. 10.1007/s11606-019-05629-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Katzmarzyk PT, Martin CK, Newton RL, Apolzan JW, Arnold CL, Davis TC, et al. Promoting Successful Weight Loss in Primary Care in Louisiana (PROPEL): rationale, design and baseline characteristics. Contemporary Clin Trials 2018;67: 1–10. 10.1016/j.cct.2018.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Höchsmann C, Dorling JL, Martin CK, Newton RL, Apolzan JW, Myers CA, et al. Effects of a 2-year primary care lifestyle intervention on cardiometabolic risk factors: a cluster-randomized trial. Circulation 2021;143:1202–14. 10.1161/CIRCULATIONAHA.120.051328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Kanaya AM, Santoyo-Olsson J, Gregorich S, Grossman M, Moore T, Stewart AL. The live well, be well study: a community-based, translational lifestyle program to lower diabetes risk factors in ethnic minority and lower-socioeconomic status adults. Am J Public Health 2012;102:1551–8. 10.2105/AJPH.2011.300456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Wayne N, Perez DF, Kaplan DM, Ritvo P. Health coaching reduces HbA1c in type 2 diabetic patients from a lower-socioeconomic status community: a randomized controlled trial. J Med Internet Res 2015;17:e224. 10.2196/jmir.4871. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Van Name MA, Camp AW, Magenheimer EA, Li F, Dziura JD, Montosa A, et al. Effective translation of an intensive lifestyle intervention for hispanic women with prediabetes in a community health center setting. Diabetes Care 2016;39:525–31. 10.2337/dc15-1899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Berkowitz SA, Chang Y, Porneala B, Cromer SJ, Wexler DJ, Delahanty LM. Does the effect of lifestyle intervention for individuals with diabetes vary by food insecurity status? A preplanned subgroup analysis of the REAL HEALTH randomized clinical trial. BMJ Open Diabetes Res Care 2020;8:e001514. 10.1136/bmjdrc-2020-001514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Govil SR, Weidner G, Merritt-Worden T, Ornish D. Socioeconomic status and improvements in lifestyle, coronary risk factors, and quality of life: the multisite cardiac lifestyle intervention program. Am J Public Health 2009;99:1263–70. 10.2105/AJPH.2007.132852. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Fernandez A, Lozano A, Lee TK, Messiah SE, Prado G. A healthy lifestyle intervention for hispanic families: moderating effects of education, income, nativity. J Nutr Educ Behav 2022;54:125–34. 10.1016/j.jneb.2021.05.001. [DOI] [PubMed] [Google Scholar]
- [28].The Look AHEAD Research Group. Association of the magnitude of weight loss and changes in physical fitness with long-term cardiovascular disease outcomes in overweight or obese people with type 2 diabetes: a post-hoc analysis of the Look AHEAD randomised clinical trial. Lancet Diabetes Endocrinol 2016;4:913–21. 10.1016/S2213-8587(16)30162-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.