Highlights
-
•
Virtual BMI ≥30 group lost 8.2 lbs vs 5.2 lbs for in-person group.
-
•
Virtual group gained 575 MET-min/week physical activity; in-person group showed no gain.
-
•
Virtual group increased fruit/vegetable intake more than in-person group (1.9 vs 0.5/day).
-
•
Virtual delivery retained 81% of participants vs 51% for in-person delivery.
-
•
Both modes effectively engaged Hispanic adults in diabetes prevention.
Keywords: Virtual delivery, diabetes prevention program, weight loss, physical activity, fruit and vegetable intake, underserved populations
Abstract
Introduction
The COVID-19 pandemic posed challenges to traditional in-person methods used to deliver lifestyle-change programs. This comparative effectiveness study examined how virtual delivery of the Diabetes Prevention Program performed relative to established in-person delivery in real-world conditions, comparing changes in weight, fruit and vegetable consumption, and physical activity levels from baseline to program completion in the Rio Grande Valley, Texas.
Methods
A secondary data analysis was conducted among adults enrolled in the Rio Grande Valley Coordinated Diabetes Prevention Program Project from 2018 to 2021. During this period, programs were delivered either in person or virtually. Covariate-adjusted regression models were used to evaluate 12-month changes in study outcomes by delivery mode. Data were analyzed in 2023 using STATA (Version 17).
Results
There were 609 participants in the in-person group and 283 in the virtual group. After adjustment for baseline characteristics, participants with BMI ≥30 in the virtual group lost more weight (-8.19 lb vs. -5.19 lb, p<0.001). Across all participants, those in the virtual group had greater increases in physical activity (+575 MET minutes/week vs. −58, p<0.001) and fruit and vegetable consumption (+1.9 vs. +0.5 servings/day, p<0.001) compared with in-person participants.
Conclusions
Virtual delivery of the Diabetes Prevention Program is a viable alternative to traditional in-person delivery, with both modes achieving meaningful improvements in weight, physical activity, and fruit and vegetable consumption among underserved adults at high risk for Type 2 diabetes. These findings support consideration of virtual Diabetes Prevention Program delivery for Hispanic and other underrepresented populations living with obesity and at risk for Type 2 diabetes mellitus.
INTRODUCTION
Diabetes mellitus (DM) is among the most common chronic diseases in the U.S., affecting 14.7% of adults in 2021.1 In addition, racial and ethnic disparities in the prevalence of DM persist, with higher rates among non-Hispanic Black (17.4%), non-Hispanic Asian (16.7%), and Hispanic adults (15.5%) than among non-Hispanic White adults (13.6%).1 Prediabetes is a preventable, asymptomatic condition where blood sugar is elevated but not yet in the diabetes range.2 Notably, 38.0% of U.S. adults have prediabetes,1 which if uncontrolled, increases the risk of DM, stroke, and heart disease.3
The Diabetes Prevention Program (DPP) is an evidence-based intervention that prevents or delays the onset of Type 2 DM (T2DM) in those with prediabetes.4,5 The DPP is characterized by lifestyle changes, physical activity (PA), and healthy eating and has been implemented as a community- and group-based program in various settings across the U.S. since 2010, when Congress approved disseminating the National DPP by the Centers for Disease Control and Prevention (CDC).6, 7, 8
During 2020–2022, organizations delivering DPP faced severe program delivery changes and challenges due to the coronavirus disease 2019 (COVID-19) public health emergency. With social distancing, traditional in-person delivery mode (I-PDM) was no longer an option,9 and organizations were forced to switch to virtual delivery mode (VDM). This rapid shift aligned with a growing trend toward virtual health delivery in the U.S., with evidence suggesting comparable effectiveness of virtual interventions for weight loss and chronic disease management in various settings.10 Internationally, digital DPPs such as the National Health Service Digital Diabetes Prevention Program in the United Kingdom have also demonstrated significant improvements in weight outcomes among adults at high risk for T2DM.11 Similarly, remote chronic disease management programs for conditions such as hypertension and heart failure have shown promise in improving patient adherence and clinical indicators and reducing hospitalizations.12,13
Thus, there is considerable interest in evaluating the effectiveness of VDM relative to I-PDM, not only in the context of pandemic-related adaptations but also to inform long-term strategies for delivering DPP beyond COVID-19 restrictions. This study uses a comparative effectiveness research approach to examine whether virtual delivery of the DPP produces outcomes comparable with those of I-PDM for weight, PA levels, and healthy eating among adults residing along the U.S.–Mexico border, using data from the Rio Grande Valley (RGV) Coordinated DPP Project. Although the COVID-19 pandemic provided the impetus for utilizing new virtual platforms for delivery of lifestyle programs,14 the evidence on the impact of VDM on participant outcomes, particularly among low-resource participants at high risk for T2DM, is lacking. A better understanding of the effectiveness of VDM in DPP on lifestyle outcomes offers the opportunity to reduce costs and improve dissemination of community-based programs such as DPP.
METHODS
This study was a secondary analysis of retrospective data from adults enrolled in the original study, RGV Coordinated DPP Project, between January 2018 and December 2021. The RGV is a region located in southern Texas along the U.S.–Mexico border, characterized by a predominantly Hispanic population and high rates of diabetes, obesity, and other chronic conditions.15
This study employed a comparative effectiveness research approach to examine the real-world performance of virtual versus in-person delivery modalities. The design reflects a pragmatic comparison driven by pandemic-related delivery adaptations rather than formal superiority testing. Instead, virtual delivery was implemented in response to COVID-19 restrictions, creating a natural experiment to evaluate its potential as a viable alternative to in-person delivery.
The original DPP study was a quasiexperimental pretest–post-test, intervention-group-only design. Data were collected at baseline and at 3, 6, 9, and 12 months, with deidentified analysis focusing on baseline and 12-month outcomes. The study compared average weight, PA levels, and daily fruit and vegetable (F/V) consumption. Most in-person classes occurred before the onset of COVID-19 restrictions, whereas most virtual classes occurred during the pandemic. A small number of classes (n=11) transitioned from in person to virtual (hybrid) and were classified by the predominant delivery mode. As a result, delivery mode was collapsed into 2 groups: virtual and in person (as referred to in the remaining parts of this paper). Although some temporal overlap existed through these hybrid cohorts, the majority of each delivery mode took place in distinct time periods, which could allow for external, time-related factors (e.g., pandemic-related disruptions, changes in daily routines, or other societal shifts) to influence outcomes independently of delivery mode.
Groups were not randomized; virtual delivery was implemented only in response to COVID-19 restrictions. Participants enrolled in the mode available at their site. The null hypothesis was that there would be no significant difference in weight loss, PA levels, or F/V consumption between VDM and I-PDM at 12 months. This comparative effectiveness analysis examined whether virtual delivery achieved outcomes comparable with those of established in-person delivery, with the expectation that both approaches would demonstrate meaningful improvements in health outcomes.
Study Sample
Of the 892 participants enrolled, data from 530 individuals who had measurements recorded at both enrollment and program completion were included in the analysis. Participants were eligible for DPP if they met the inclusion and exclusion criteria established by CDC.16,17 All participants provided written informed consent prior to participation. The Committee for the Protection of Human Subjects, the IRB of the University of Texas Health Science Center at Houston, reviewed and approved the original study on June 08, 2018 and subsequently approved this secondary data analysis.
Measures
Primary study outcomes included changes in average weight (self-reported for VDM group), self-reported total weekly PA energy expenditure in MET minutes per week (MET-min/week), and self-reported daily F/V consumption. These outcomes were assessed at both baseline and program end. Although the self-report method was consistent across groups for PA and F/V, the measurement of weight differed, which may introduce recall and measurement biases. The primary independent variable was delivery mode, categorized into 2 groups: VDM and I-PDM.
Weight was measured directly at baseline, quarterly, and at 12 months for the I-PDM group using a calibrated digital scale by trained staff. The measurement was taken twice at each time point, and an average weight was calculated. For VDM participants, digital scales were provided at enrollment to self-report weight measurements during the pandemic.
Leisure-time PA was self-reported using a modified version of the Godin Leisure-Time Exercise Questionnaire.18 An example question is During a typical 7-day period, how many times on average do you do strenuous exercise where your heart beats rapidly for more than 10 minutes during your free time? The modified version additionally included duration; for example, On occasions when you do strenuous exercise, what is the average number of minutes you exercise?
F/V consumption was self-reported using the validated 2-item dietary questionnaire by Cappuccio et al.,19 2003. The 2 questions were How many portions of fruit/vegetables excluding potatoes (based on these photos) do you eat on a typical day? Responses were recorded in portions and are reported in this study as servings for consistency with U.S. dietary guidelines.
Finally, demographic and baseline data—including age, sex, ethnicity, height and weight (for BMI), education, employment status, preferred language, and insurance—were collected at enrollment using standardized intake forms. Program delivery characteristics such as language used for delivery, session timing, attendance, and retention were recorded by coaches and compiled for VDM and I-PDM.
Statistical Analysis
Descriptive statistics were used to describe participant baseline and demographic characteristics. Means and SD were used for continuous variables, and frequencies and percentages were used for categorical variables. Study completers were defined as participants with measurements at both baseline and program end. Characteristics of completers versus noncompleters were compared using independent t-tests for continuous variables and chi-square tests for categorical variables.
Covariate-adjusted regression models were used to evaluate 12-month changes in study outcomes by delivery mode. All fully adjusted models controlled for baseline characteristics, including age, sex, ethnicity, preferred language, insurance status, employment status, and education level. Analyses for average weight were further stratified by BMI category (<30 and ≥30 kg/m²). The final analytic sample included participants with complete baseline and 12-month data: 517 participants for the PA and F/V models and 153 participants in the BMI <30 group and 364 participants in the BMI ≥30 group for the BMI-stratified weight models.
Estimates of change in primary outcomes were assessed using time-by-treatment interactions within repeated measures models. A dose–response analysis was also performed to examine whether the number of sessions attended affected health outcomes within each delivery mode. Data were analyzed in 2023 using STATA IC software (Version 17). Statistical significance was determined using an alpha level (p-value) <0.05. Sample size calculations were performed for a between-groups design using change-score ANOVA. On the basis of the observed sample sizes at study initiation (n=514) and assuming 80% power at α=0.05, the study was adequately powered to detect an 11-lb difference in weight loss, a 361.5 MET-minute per week change in PA, or a 0.5-portion change in F/V consumption between delivery modes. Although the detectable effect sizes for weight and PA were larger than initially expected, the 0.5 portion difference in F/V consumption was considered feasible and meaningful for diabetes prevention outcomes.
RESULTS
Of the 892 participants (starters) enrolled between 2018 and 2021, 609 participated in I-PDM, and 283 participated in VDM. Among these, 530 (59.4%) completed the 1-year program and evaluation, with completion rates of 43.3% in I-PDM and 94.0% in VDM. Participants who withdrew before the first class were excluded from the baseline sample.
Among starters, mean age (±SD) was comparable across groups (47.3±11.3 years for I-PDM vs 47.3±10.9 years for VDM). Sex, ethnicity, employment status, preferred language, and insurance status were also comparable. When comparing completers with starters, baseline characteristics were similar overall, with differences related to education level and preferred language (p<0.05). Baseline characteristics for completers were also compared between delivery modes. Significant differences were observed for education levels (p=0.003) and preferred language (p<0.001). In the I-PDM group, 31.8% were classified as overweight (BMI≥25<30), and 63.7% were classified as obese (BMI≥30), compared with 22.2% overweight and 75.6% obese in the VDM group. Spanish was the preferred language for 53.1% of the I-PDM participants and 64.7% of the VDM participants (Table 1).
Table 1.
Baseline and Demographic Characteristics of DPP Participants by Delivery Mode, 2018–2021
| Characteristic | All starters (n=892) |
Study completers (n=530) |
||||
|---|---|---|---|---|---|---|
| In person (n=609) | Virtual (n=283) | p-value | In person (n=264) | Virtual (n=266) | p-value | |
| Demographic characteristics | ||||||
| Age, years | ||||||
| Mean (SD) | 47.3 (11.3) | 47.3 (10.9) | 0.996 | 47.2 (11.3) | 48.2 (10.6) | 0.293 |
| Range | 19–77 | 19–74 | — | 19–74 | 19–74 | — |
| Sex, n (%) | 0.395 | 0.405 | ||||
| Female | 537 (88.2) | 255 (90.1) | 229 (86.7) | 237 (89.1) | ||
| Male | 72 (11.8) | 28 (9.9) | 35 (13.3) | 29 (10.9) | ||
| Ethnicity, n (%) | 0.246 | 0.057 | ||||
| Hispanic/Latino | 588 (96.6) | 276 (97.5) | 254 (96.2) | 258 (97.0) | ||
| Non-Hispanic | 15 (2.5) | 7 (2.5) | 5 (1.9) | 8 (3.0) | ||
| Declined/missing | 6 (1.0) | 0 (0.0) | 5 (1.9) | 0 (0.0) | ||
| BMI category, n (%) | <0.001 | 0.021 | ||||
| Normal (BMI <25) | 26 (4.3) | 5 (1.8) | 12 (4.5) | 6 (2.3) | ||
| Overweight (BMI=25.0–29.9) | 187 (30.7) | 56 (19.8) | 84 (31.8) | 59 (22.2) | ||
| Obese Class I (BMI=30.0–34.9) | 195 (32.0) | 91 (32.2) | 91 (34.5) | 93 (35.0) | ||
| Obese Class II (BMI=35.0–39.9) | 114 (18.7) | 63 (22.3) | 39 (14.8) | 54 (20.3) | ||
| Obese Class III (BMI≥40.0) | 87 (14.3) | 68 (24.0) | 38 (14.4) | 54 (20.3) | ||
| Socioeconomic characteristics | ||||||
| Education level, n (%) | 0.032 | 0.003 | ||||
| 8th grade or less | 142 (23.3) | 59 (20.9) | 48 (18.2) | 53 (19.9) | ||
| Some high school | 61 (10.0) | 35 (12.4) | 24 (9.1) | 26 (9.8) | ||
| High-school diploma/GED | 128 (21.0) | 57 (20.1) | 73 (27.7) | 52 (19.5) | ||
| Some college | 109 (17.9) | 36 (12.7) | 52 (19.7) | 40 (15.0) | ||
| Technical/trade school | 37 (6.1) | 32 (11.3) | 12 (4.5) | 35 (13.2) | ||
| College degree (BA/BS) | 69 (11.3) | 44 (15.6) | 24 (9.1) | 39 (14.7) | ||
| Graduate degree (MS/PhD) | 42 (6.9) | 16 (5.7) | 22 (8.3) | 16 (6.0) | ||
| No school completed | 11 (1.8) | 3 (1.1) | 4 (1.5) | 5 (1.9) | ||
| Missing | 10 (1.6) | 1 (0.4) | 5 (1.9) | 0 (0.0) | ||
| Employment status, n (%) | 0.654 | 0.312 | ||||
| Employed | 293 (48.1) | 143 (50.5) | 137 (51.9) | 129 (48.5) | ||
| Unemployed | 306 (50.3) | 140 (49.5) | 121 (45.8) | 136 (51.1) | ||
| Missing | 10 (1.6) | 0 (0.0) | 6 (2.3) | 1 (0.4) | ||
| Preferred language, n (%) | 0.089 | <0.001 | ||||
| English | 267 (43.8) | 107 (37.8) | 135 (51.1) | 94 (35.3) | ||
| Spanish | 342 (56.2) | 176 (62.2) | 129 (48.9) | 172 (64.7) | ||
| Insurance status, n (%) | 0.428 | 0.882 | ||||
| Insured | 295 (48.4) | 147 (51.9) | 141 (53.4) | 142 (53.4) | ||
| Uninsured | 306 (50.3) | 136 (48.1) | 119 (45.1) | 123 (46.2) | ||
| Missing | 8 (1.3) | 0 (0.0) | 4 (1.5) | 1 (0.4) | ||
Note: Boldface indicates statistical significance (p<0.05).
Data are presented as n (%) for categorical variables and mean (SD) for continuous variables. Differences between in-person and virtual groups were assessed using independent t-tests for continuous variables and chi-square test for categorical variables. BMI categories are included for descriptive purposes; differences were not interpreted in the Results because BMI is an outcome measure. Missing data reflects participants who declined to answer or had incomplete information. Percentages may not sum to 100% owing to rounding.
DPP, Diabetes Prevention Program.
Program delivery characteristics are shown in Table 2. Although average attendance (70.90%) and retention (81.1%) rates were higher in VDM than in I-PDM (58.5% and 51.2%, respectively), differences were not statistically significant (p=0.272 attendance and p=0.053 retention). Significant differences were observed between the groups in class delivery time, with VDM programs more often held in the evenings (48%) than in the mornings in I-PDM (12%, p=0.005).
Table 2.
DPP Program Delivery Characteristics by Delivery Mode, 2018–2021
| Characteristics | In-person delivery, n (%) | Virtual delivery, n (%) | Total | p-value |
|---|---|---|---|---|
| Number of DPP courses delivered | 42 | 21 | 63 (100%) | — |
| Number of participants | 609 | 283 | 892 (100%) | — |
| Delivery language, n (%) | ||||
| English | 9 (21%) | 7 (33%) | 16 (25.4%) | 0.085 |
| Spanish | 18 (43%) | 12 (57%) | 30 (47.6%) | |
| Bilingual | 15 (36%) | 2 (10%) | 17 (27.0%) | |
| Time of class, n (%) | ||||
| Mornings (8:00am to noon) | 29 (69%) | 7 (33%) | 36 (57.1%) | 0.005 |
| Afternoons (noon to 5:00pm) | 8 (19%) | 4 (19%) | 12 (19.0%) | |
| Evenings (5:00 pm and after) | 5 (12%) | 10 (48%) | 15 (23.8%) | |
| Average attendance, n (%) | 58.46% | 70.90% | — | 0.272 |
| Average retention, n (%) | 51.16% | 81.06% | — | 0.053 |
Note: Boldface indicates statistical significance(p<0.05).
Differences between delivery modes were assessed using chi-square tests for categorical variables and t-tests for continuous variables (average attendance and average retention). Percentages were calculated within delivery mode (column percentages) for categorical comparisons.
DPP, Diabetes Prevention Program.
Participant-level absolute changes from baseline to 12 months for weight, PA, and F/V intake, by delivery mode, are presented in Table 3. Corresponding descriptive statistics for weight, PA, and F/V intake at baseline and 12 months, stratified by delivery mode, are provided in Appendix Table 1 (available online).
Table 3.
Changes in Weight, Physical Activity, and Fruit and Vegetable Intake From Baseline to 12 Months by Delivery Mode
| Outcome | Delivery mode | Baseline, n | Baseline, mean (SD) | 12 month, n | 12 month, mean (SD) | Absolute change, meana |
|---|---|---|---|---|---|---|
| Weight (lbs) | In person | 264 | 185.08 (42.24) | 261 | 180.88 (43.40) | −4.20 |
| Virtual | 266 | 196.20 (45.42) | 266 | 189.15 (42.48) | −7.05 | |
| Physical activity (MET-min/week) | In person | 258 | 1,083.38 (2,051.79) | 264 | 1,055.93 (1,289.07) | −27.45 |
| Virtual | 266 | 396.75 (1,032.33) | 266 | 965.38 (1,241.87) | 568.63 | |
| Fruit and vegetable intake (servings/day) | In person | 264 | 3.46 (2.15) | 264 | 3.97 (2.13) | 0.51 |
| Virtual | 266 | 2.48 (1.86) | 266 | 4.42 (1.94) | 1.94 |
Note: Sample sizes (n) reflect participants with complete data for the respective outcome at both baseline and 12 months. Values are presented as mean (SD). Physical activity is measured in MET-min/week.
Absolute change was calculated as 12-month mean − baseline mean. Negative values indicate decreases; positive values indicate increases.
After these descriptive comparisons, covariate-adjusted mixed-effects regression models were used to examine 12-month changes in health outcomes. The time-by-treatment interaction term in these models represents the formal statistical test of whether changes differed between delivery modes (VDM versus I-PDM). Full model results are provided in Appendix Table 2 (available online), and key estimates are summarized in Table 4. All models adjusted for baseline age, sex, ethnicity, preferred language, insurance status, employment status, and education level. For average weight, analyses were stratified by BMI category (<30 and ≥30 kg/m²). The PA and F/V models included 517 participants; the BMI-stratified weight models included 153 participants with BMI <30 and 364 participants with BMI ≥30.
Table 4.
Covariate-Adjusted Mixed-Effects Regression Model Results: Marginal Means, Contrasts, and Estimated Pre–Post Changes in Main Outcomes
| Outcome/stratum | Group, n | Baseline, mean (95% CI) | 12 month, mean (95% CI) | Change from baseline (95% CI) | SE | p-value (within group) | Treatment difference (12-month change, virtual versus in person) (95% CI) | p-value (between groups) |
|---|---|---|---|---|---|---|---|---|
| Average weight (lbs) | ||||||||
| BMI <30 | Virtual (n=306 obs, 153 pts) | 153.5 (150, 157) | 150.0 (146, 154) | −3.56 (−5.43, −1.69) | 0.95 | <0.001 | 0.22 (−2.25, 2.68) | 0.863 |
| In person | 150.0 (147, 153) | 146.7 (144, 150) | −3.34 (−4.95, −1.73) | 0.82 | <0.001 | |||
| BMI ≥30 | Virtual (n=726 obs, 364 pts) | 211.0 (206, 216) | 202.8 (197, 208) | −8.19 (−10.2, −6.2) | 1.02 | <0.001 | 3.00 (0.06, 5.94) | 0.045 |
| In person | 204.2 (198, 210) | 199.0 (193, 205) | −5.19 (−7.4, −3.0) | 1.12 | <0.001 | |||
| Physical activity (MET-min/week) | Virtual | 387 (214, 560) | 962 (789; 1,136) | 575 (355, 795) | 112.24 | <0.001 | −633 (−950, −317) | <0.001 |
| In person (n=1,028 obs, 517 pts) | 1,128 (948; 1,309) | 1,070 (893; 1,248) | −58 (−285, 169) | 115.82 | 0.617 | |||
| Fruit and vegetable intake (servings/day) | Virtual | 2.46 (2.22, 2.70) | 4.40 (4.16, 4.64) | 1.94 (1.69, 2.20) | 0.13 | <0.001 | −1.43 (−1.80, −1.07) | <0.001 |
| In person (n=1,034 obs, 517 pts) | 3.50 (3.25, 3.75) | 4.01 (3.76, 4.25) | 0.51 (0.25, 0.77) | 0.13 | <0.001 |
Note: Boldface indicates statistical significance (p<0.05).
Obs denotes total repeated observations used in the mixed-effects models (baseline and 12 months), and pts denotes unique participants with complete outcome and covariate data. Percentage average weight loss was calculated as ([average weight at baseline − average weight at 12 months]/baseline weight) × 100. Physical activity was measured in MET-minutes per week. Estimated marginal means and contrasts are presented for each outcome at baseline and 12 months, by delivery mode. Change from baseline is the estimated pre–post difference within each group.
For weight, age and sex were significant determinants of reduction in both BMI strata, with older participants and females showing greater decreases in weight over time. Marginal estimates indicated that in the BMI ≥30 stratum, mean 12-month weight loss was 8.19 lb in VDM compared with 5.19 lb in I-PDM (Table 4), corresponding to 3.9% and 2.5% of baseline weight, respectively. Although these changes did not meet the CDC DPP benchmark of 5%–7% weight loss, they are consistent with other real-world DPP outcomes20 and represent clinically meaningful improvements that are likely to yield important public health benefits.
For PA, adjusted models showed a gain of 575 MET-min per week in VDM (p<0.001) and no significant change in I-PDM (−58 MET-min/week, p=0.617). The ethnicity category Other (versus Hispanic/Latino, p<0.001) and being unemployed (versus being employed, p=0.010) were significant determinants of increased PA.
For daily F/V consumption, VDM participants increased intake by 1.9 servings, compared with 0.5 servings in I-PDM. Sex and education level were significant determinants of F/V increase, with females showing a statistically significant association with higher intake (p=0.007). Education level also demonstrated a clear gradient effect, with significant increases observed among participants with some high school (p=0.010), high-school diploma or GED (p=0.005), some college (p=0.001), college degree BA/BS (p=0.007), and graduate degree MS/PhD (p=0.022). Notably, individuals who declined to provide ethnicity information showed significantly lower F/V consumption than participants who provided ethnicity data.
Across outcomes, among those with obesity (BMI ≥30), VDM participants lost an average of 3 lb more than I-PDM participants over 12 months (p=0.045), whereas weight change did not differ for those with BMI <30 (p=0.863). The between-group comparison showed that VDM participants achieved significantly greater improvements than I-PDM for both PA (difference: 633 MET-min/week, p<0.001) and F/V intake (difference: 1.44 servings/day, p<0.001) (Table 4).
In the dose–response analysis, all participants experienced on average a 6.7 lb reduction in body weight over 12 months regardless of the number of sessions attended. Participants in the I-PDM experienced a slightly smaller weight reduction (–6.3 lb, SD=20.4) than those in the VDM (–7.0 lb, SD=14.7). PA improvements were greater in the VDM group (569 MET-min/week, SD=1,559) than minimal to no change in the I-PDM group (–2.8 MET-min/week, SD=2,033). Both groups showed improvements in F/V intake, but with larger increases in the VDM (1.9 servings/day, SD=2.0) than in the I-PDM (0.5 servings/day, SD=2.2) groups.
These estimates reflect raw pre–post differences and do not account for differences across delivery modes in the number of sessions attended. On average, VDM participants attended 20 sessions, whereas I-PDM participants attended 17 sessions.
To examine dose–response effects, additional multilevel mixed-effects linear regressions assessed the per-session effect on health outcomes among study completers, adjusting for potential confounders, including age, sex, ethnicity, preferred language, insurance status, employment status, and education level. For each outcome variable, the total pre–post change was computed for participants with complete data; this change was then regressed against the number of sessions (specified as a continuous variable).
Each additional DPP session was associated with a slight, nonsignificant weight increase in both modes (0.46 lb per session in I-PDM vs 0.31 lb per session in VDM). This unexpected result likely reflects confounding from participants with higher baseline weights being more likely to attend additional sessions, influencing the association. The main result is that, overall, participants lost weight across the program regardless of attendance level.
PA change per session was also not statistically significant in both modes, although the point estimate was higher in I-PDM (45.2 MET-min per session) than in VDM (2.13 MET-min per session). Notably, the per-session increase in daily F/V intake was significant in VDM (0.06 servings per session, p=0.046) but not in I-PDM. This aligns with time-by-treatment findings on F/V consumption, in which the total change in F/V was higher in the VDM than in the I-PDM (1.9 vs 0.5 servings, respectively).
DISCUSSION
To the authors’ knowledge, this is the first study assessing the impacts of virtual DPP delivery during the pandemic among low-income adults in the RGV. The findings confirm the robust effectiveness of DPP regardless of the delivery mode and add to the growing evidence supporting the impact of the 12-month program when delivered virtually.21 On average, all participants experienced a 6.7 lb reduction in weight regardless of session attendance or delivery mode, underscoring the promise of both delivery modes in reducing the risk of T2DM.
This study employed a comparative effectiveness framework to evaluate virtual delivery as an alternative to traditional in-person DPP delivery. The approach emphasized real-world performance of virtual delivery relative to established in-person methods, reflecting its emergence from pandemic necessity rather than formal superiority testing.
The findings support virtual delivery as a viable alternative, with both modalities achieving meaningful clinical improvements. From a public health perspective, demonstrating comparable outcomes has significant implications for program accessibility, particularly for underserved populations facing traditional barriers such as transportation challenges and work schedule conflicts. The flexibility of virtual programming, evidenced by higher evening session availability, may enhance accessibility for individuals with competing responsibilities.
Each additional session of DPP was associated with a slight, nonsignificant weight increase (0.46 lb per session in I-PDM and 0.31 lb per session in VDM), likely reflecting selection effects or unmeasured confounding in session attendance rather than a true dose–response relationship. Importantly, the overall finding is that average weight loss was achieved across both delivery modes. The I-PDM group experienced greater improvements in PA per session than the VDM group (45.2 vs 2.13 MET-min per session, respectively), although these changes were not statistically significant. However, the per-session change in F/V intake was statistically significant in the VDM group (0.06 servings per session, p=0.046) but not in I-PDM. Moreover, VDM participants averaged slightly greater weight loss and showed changes in PA and F/V consumption compared with I-PDM participants.
In addition, in this study covering DPP programs delivered between 2018 and 2021, average attendance and retention rates were higher in VDM group (70.90% and 81.06%) than in the I-PDM group (58.46% and 51.16%, respectively); however, these differences were not significant. Significant differences were seen between the groups in class delivery time. Virtual programs were flexible and often scheduled in the evenings, possibly increasing accessibility for individuals with work or childcare responsibilities. These results represent key findings given the potential benefits of virtual health promotion and weight loss interventions addressing lifestyle changes in overcoming many of the limitations faced in in‐person–based interventions, facilitating attendance, and improving health outcomes.
Previous studies have shown mixed results when comparing VDM with I-PDM. For instance, Taetzsch and colleagues22 led a nonrandomized trial study among 2 military installations that received a DPP-Group Lifestyle Balance program either in person or through a videoconference-adapted version. Unlike the results seen in this study, they found no significant differences in weight loss over 12 weeks between participants in the in-person program and those in the videoconference intervention (6.2% and 5.3%, respectively).22 However, the authors found differences in weight change trajectory between groups over time. Furthermore, Das et al.23 assessed the effect of program delivery of a new commercial weight loss behavioral intervention on participants’ outcomes and found no statistical significance that program delivery mode was associated with percentage weight loss. Inconsistent with their results, this study showed average weight reduction in the adjusted regression when stratifying by BMI and controlling for baseline characteristics. Participants in the VDM had improved average weight in the BMI ≥30 stratum compared with those in the I-PDM (−8.19 lb vs −5.19 lb, respectively).
A further major finding from the analyses is that delivering DPP virtually resulted in improved program outcomes, including a reduction in average weight, improved PA levels, and increased daily F/V consumption after completing the program. After adjusting for baseline characteristics, there was a statistically significant increase in PA (575 MET-min/week, p<0.001) in VDM only. For F/V consumption, results showed a statistically significant increase in both modes, but a greater improvement was seen in the VDM (2 servings vs 0.5 in I-PDM). Consistent with the findings, Iglesias and colleagues14 also assessed pre–post changes in health behaviors in the VDM of their community health worker–led program that focused on reducing cardiovascular disease risk among economically disadvantaged Latinos.
These mixed findings in the literature highlight the importance of comparative effectiveness research approaches that examine real-world implementation rather than studies conducted under ideal research conditions. This study contributes to this evidence base by demonstrating that virtual delivery can serve as an effective alternative to in-person delivery, particularly important for programs serving populations in underserved areas or with limited resources.
A key strength of this study is that it is a secondary analysis of real-world, evidence-based group lifestyle intervention that comprises a structured diet, PA, and behavior change focused on reducing the risk of T2DM among adults. Building from a pragmatic trial research orientation that aims to demonstrate real-world effectiveness,24 a benefit of such practice-based research is the generalizability of the outcomes. In addition, changes in outcomes were evaluated after completion of the 1-year program. The more extended study period provides adequate time to observe a potential change in health outcomes. Furthermore, given the paucity of data available from the target population, the sample addresses an important gap in the literature on the contributions of the DPP among low-income communities in the RGV, TX.
From a comparative effectiveness perspective, several factors support virtual delivery as a viable alternative. Virtual delivery demonstrated notably higher retention rates (81.06% vs 51.16% for I-PDM), suggesting that this modality may help address engagement challenges that commonly affect lifestyle interventions. The added flexibility of virtual scheduling, with more evening options, may also better accommodate the diverse needs of working adults and caregivers. Taken together, these practical advantages, coupled with comparable health outcomes, position virtual delivery as a valuable, durable addition to the DPP delivery model rather than merely a temporary pandemic adaptation.
Limitations
A major threat to internal and external validity is differential history effect. Although some hybrid cohorts overlapped time periods, most in-person delivery occurred before the pandemic, and most virtual delivery occurred during the pandemic. This temporal separation could have introduced time-related confounding because the pandemic represented a unique and unprecedented period of significant lifestyle disruptions. In VDM programs, factors influencing results could include participants having more time on their hands and wanting to focus on their health during the pandemic or other unmeasured, time-related external factors (novelty and disruption effect).25 The authors cannot infer from this study that outcomes from VDM would be as positive under normal, nonpandemic conditions.
Measurement methods also differed by delivery mode. For the VDM group, data were collected virtually: participants self-reported their weight using digital scales provided by the program, and PA and F/V intake were collected through phone interviews. On the other hand, I-PDM group had weight measured directly in person, and PA and F/V intake were self-reported in person. These differences may have introduced recall bias and measurement bias, particularly for self-reported outcomes, as well as potential social desirability bias. Bias in self-reported PA can be avoided in the future using data from activity trackers. In addition, data on participants who were offered, referred to, or dropped from the program were not collected, which may have limited the ability to assess this potential bias, including personal choice, differences related to geographic location, or other factors. Furthermore, potential bias from differential dropout during the program could not be assessed.
As described in the Methods, hybrid cohorts were classified by predominant delivery mode, which may have introduced minor misclassification. Most participants in the sample were Hispanic and female, reflecting the demographics of the RGV region; as a result, the generalizability of these findings to more diverse populations may be limited. In addition, analyses included only DPP-eligible individuals who completed both baseline and 12-month assessments; no statistical methods were used to address missing data, which may have introduced selection bias and further limited generalizability.
CONCLUSIONS
This comparative effectiveness analysis demonstrates that virtual delivery of DPP programs represents a viable alternative to traditional in-person delivery, with both approaches achieving meaningful improvements in health outcomes, including reducing average weight, increasing weekly PA minutes, and improving daily F/V consumption, among underserved communities at high risk for T2DM. The results suggest that 1-year health behavior, weight loss, and DPPs such as DPP delivered virtually may be considered for Hispanic and other underrepresented populations who are living with obesity and at risk for T2DM.
Acknowledgments
ACKNOWLEDGMENTS
The researchers and authors thank all community members, partners, lifestyle coaches, and the Diabetes Prevention Program research and evaluation team at the University of Texas Health Science Center at Houston School of Public Health in Brownsville for their support with this study. The IRB of the University of Texas Health Science Center at Houston reviewed and approved this study. During the preparation of this work, the authors used ChatGPT (OpenAI) to assist with manuscript formatting, grammar, and language revision and Claude (Anthropic) for table creation and methodologic consistency. The authors reviewed and edited the content as needed and take full responsibility for the content of the publication.
Disclaimer: Findings and conclusions are those of the researchers and authors and do not represent the official views of the Centers for Disease Control and Prevention.
Funding: None.
Declaration of interest: None.
CRediT AUTHOR STATEMENT
Ghadir Helal Salsa: Conceptualization, Methodology, Writing - original draft, Writing - reviewing & editing, Visualization. Andrew E. Springer: Conceptualization, Methodology, Writing - review & editing, Supervision. Nalini Ranjit: Formal analysis, Writing - review & editing. John Wesley McWhorter: Writing - review & editing. Belinda M. Reininger: Validation, Investigation, Writing - review & editing, Supervision.
Footnotes
Supplementary material associated with this article can be found in the online version at doi:10.1016/j.focus.2025.100457.
Appendix. Supplementary materials
REFERENCES
- 1.National Diabetes Statistics Report. Centers for Disease Control and Prevention. https://www.cdc.gov/diabetes/php/data-research/index.html Updated 2024. Accessed November 19, 2024.
- 2.Tabák A.G., Herder C., Rathmann W., Brunner E.J., Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet. 2012;379(9833):2279–2290. doi: 10.1016/S0140-6736(12)60283-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Centers for Disease Control and Prevention; 2024. Prediabetes – your chance to prevent type 2 diabetes.https://www.cdc.gov/diabetes/prevention-type-2/prediabetes-prevent-type-2.html UpdatedAccessed March 19, 2022. [Google Scholar]
- 4.Centers for Disease Control and Prevention; 2024. About the national diabetes prevention program.https://www.cdc.gov/diabetes-prevention/programs/?CDC_AAref_Val=https://www.cdc.gov/diabetes/prevention/about.htm UpdatedAccessed March 14, 2022. [Google Scholar]
- 5.Centers for Disease Control and Prevention; 2024. Preventing type 2 diabetes with the lifestyle change program.https://www.cdc.gov/diabetes-prevention/lifestyle-change-program/index.html UpdatedAccessed November 19, 2024. [Google Scholar]
- 6.Kramer M.K., Kriska A.M., Venditti E.M., et al. Translating the diabetes prevention program: a comprehensive model for prevention training and program delivery. Am J Prev Med. 2009;37(6):505–511. doi: 10.1016/j.amepre.2009.07.020. [DOI] [PubMed] [Google Scholar]
- 7.Aziz Z., Absetz P., Oldroyd J., Pronk N.P., Oldenburg B. A systematic review of real-world diabetes prevention programs: learnings from the last 15 years. Implement Sci. 2015;10:172. doi: 10.1186/s13012-015-0354-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Venditti E.M., Kramer MK. Diabetes prevention program community outreach: perspectives on lifestyle training and translation. Am J Prev Med. 2013;44(4)(suppl 4):S339–S345. doi: 10.1016/j.amepre.2012.12.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Palmer K.N.B., Garr Barry V.E., Marrero D.G., et al. Intervention delivery matters: what mothers at high risk for type 2 diabetes want in a diabetes prevention Program-results from a comparative effectiveness trial. Diabetes Ther. 2020;11(10):2411–2418. doi: 10.1007/s13300-020-00891-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Joiner K.L., Nam S., Whittemore R. Lifestyle interventions based on the diabetes prevention program delivered via eHealth: a systematic review and meta-analysis. Prev Med. 2017;100:194–207. doi: 10.1016/j.ypmed.2017.04.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ross J.A.D., Barron E., McGough B., et al. Uptake and impact of the English National Health Service digital diabetes prevention programme: observational study. BMJ Open Diabetes Res Care. 2022;10(3) doi: 10.1136/bmjdrc-2021-002736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.McLean G., Band R., Saunderson K., et al. Digital interventions to promote self-management in adults with hypertension systematic review and meta-analysis. J Hypertens. 2016;34(4):600–612. doi: 10.1097/HJH.0000000000000859. https://DOI.ORG/10.1097/HJH.0000000000000859 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Inglis S.C., Clark R.A., Dierckx R., Prieto-Merino D., Cleland J.G.F. Structured telephone support or non-invasive telemonitoring for patients with heart failure. Cochrane Database Syst Rev. 2015;2015(10) doi: 10.1002/14651858.CD007228.pub3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Iglesias A., Ambrose A., Coronel-Mockler S., et al. Impact of transitioning to virtual delivery of a cardiovascular health improvement program for Latinos during the COVID-19 pandemic. BMC Public Health. 2022;22(1):1935. doi: 10.1186/s12889-022-14291-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Gaddis J.M., Arellano E., Pulido K., et al. Burden of diabetes mellitus in the medically underserved rio grande valley. Cureus. 2024;16(9) doi: 10.7759/cureus.70088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Centers for Disease Control and Prevention; 2024. Are you eligible to join the national DPP lifestyle change program?https://www.cdc.gov/diabetes-prevention/eligible-lifestyle-change-program/index.html UpdatedAccessed November 19, 2024. [Google Scholar]
- 17.Centers for Disease Control and Prevention; 2024. New American Diabetes Association (ADA)/CDC prediabetes risk test.https://nationaldppcsc.cdc.gov/s/article/New-American-Diabetes-Association-ADA-CDC-Prediabetes-Risk-Test UpdatedAccessed November 19, 2024. [Google Scholar]
- 18.Godin G., Shephard RJ. Godin leisure-time exercise questionnaire. Med Sci Sports Exerc. 1997;29(6)(suppl):S36–S38. doi: 10.1097/00005768-199706001-00009. [DOI] [Google Scholar]
- 19.Cappuccio F.P., Rink E., Perkins-Porras L., McKay C., Hilton S., Steptoe A. Estimation of fruit and vegetable intake using a two-item dietary questionnaire: a potential tool for primary health care workers. Nutr Metab Cardiovasc Dis. 2003;13(1):12–19. doi: 10.1016/S0939-4753(03)80163-1. [DOI] [PubMed] [Google Scholar]
- 20.Albright A.L., Gregg EW. Preventing type 2 diabetes in communities across the U.S: the National Diabetes Prevention Program. Am J Prev Med. 2013;44(4)(suppl 4):S346–S351. doi: 10.1016/j.amepre.2012.12.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Welshons K.F., Johnson N.A., Gold A.L., Reicks M. Diabetes prevention program outcomes by in-person versus distance delivery mode among ethnically diverse, primarily lower-income adults. Digit Health. 2023;9 doi: 10.1177/20552076231173524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Taetzsch A., Gilhooly C.H., Bukhari A., et al. Development of a videoconference-adapted version of the community diabetes prevention program, and comparison of weight loss with in-person program delivery. Mil Med. 2019;184(11–12):647–652. doi: 10.1093/milmed/usz069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Das S.K., Brown C., Urban L.E., et al. Weight loss in videoconference and in-person iDiet weight loss programs in worksites and community groups. Obesity (Silver Spring) 2017;25(6):1033–1041. doi: 10.1002/oby.21854. [DOI] [PubMed] [Google Scholar]
- 24.Ford I., Norrie J. Pragmatic trials. N Engl J Med. 2016;375(5):454–463. doi: 10.1056/NEJMra1510059. [DOI] [PubMed] [Google Scholar]
- 25.Bracht G.H., Glass GV. The external validity of experiments. Am Educ Res J. 1968;5(4):437–474. doi: 10.3102/00028312005004437. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.