Some people are slower to respond during lifestyle interventions. An adaptive “rescue” intervention may improve outcomes among slow responders. The impact of a worksite rescue intervention for early slow responders was evaluated.
Employees ≥21 years old with prediabetes were stratified to intervention using a 2.5% weight loss (%WL) threshold at week 5. Outcomes were assessed at baseline and at 4 months using mixed-effect and linear regression models.
Significant improvement occurred in mean %WL, glycemia, total cholesterol, and triglycerides in the standard compared with the adaptive (Group Lifestyle Balance Plus [GLB+]) intervention (all P≤ 0.01). However, GLB+ participants also experienced a significant reduction in %WL and glycemia (all P < 0.01). The %WL at week 5 significantly predicted %WL at 4 months (P < 0.0001). The between-group difference of 4-month %WL was not significant for someone achieving 2.5%WL at week 5.
Diabetes prevention programs should consider weight loss success following 1 month of treatment and offer a rescue intervention to early slow weight loss responders.
Introduction
The Diabetes Prevention Program (DPP) provided compelling evidence that lifestyle modification that results in 5–7% weight loss (%WL) decreases diabetes risk by 58% relative to control over 3.2 years of follow-up (1). At 6 months, 50% of DPP lifestyle participants achieved the ≥7%WL goal, indicating heterogeneity in response (2). Previously, weight loss success achieved during the first weeks of treatment was significantly associated with longer-term weight loss (3–6).
Behavioral weight loss programs generally provide the same intervention to all individuals regardless of early response. In these fixed interventions, some intervention components may have little relevance to an individual, which increases implementation cost while providing little benefit. Thus, “rescue” or adaptive interventions that are better tailored to the needs of early slow responders to improve outcomes have emerged in which treatment is varied based on participant progress. An adaptive intervention for early slow responders was implemented and evaluated in this study among employees with prediabetes at a university work site.
Research Design and Methods
Participants had a BMI ≥25 kg/m2 for non-Asians (≥23 kg/m2 for Asians) and were ≥21 years old with prediabetes based on clinical guidelines (7). An adaptive intervention design was used through weekly, group-based sessions. Body weight was obtained prior to each session. Based on prior research, failure to achieve >2.5%WL following the 1st month of intervention was predictive of failure to achieve ≥5%WL at follow-up (8). Thus, if participants achieved >2.5%WL at the beginning of session 5, they remained in the standard intervention, Group Lifestyle Balance (GLB), at session 5. If participants achieved ≤2.5%WL, they were stratified to the adaptive intervention, GLB Plus (GLB+), at session 5. Participants and health coaches were unaware of differences between the interventions. All procedures were followed in accordance with the ethical standards of the Ohio State University Institutional Review Board, and participants provided written, informed consent.
GLB is a manualized group-based intervention adapted from the DPP Outcomes Study described elsewhere (9), with a 7%WL goal by study end. All participants received GLB during the 1st month of intervention and met together as one group. Briefly, GLB+ helped participants link goals to personal values and specific plans for implementing change, develop goal-setting and problem-solving skills, identify and minimize obstacles that might arise, and develop implementation plans. Mindful decision making and agency for lifestyle change also were promoted (10,11).
Outcomes were collected in person using a calibrated digital scale with standing stadiometer. Glucose, cholesterol, and lipoprotein outcomes were assayed via fingerstick samples collected after a minimum 8-h fast using the Alere Cholestech LDX and the Alere Afinion. Blood pressure readings were obtained from seated participants with a digital Omron blood pressure monitor.
Between-group comparison of baseline outcomes and comparison of the change in outcomes at 4 months following the core intervention phase were determined using mixed-effect models where treatment group, time, and their interaction were fixed effects and subjects were random effects. Linear regression was conducted for the %WL expected from baseline to 4 months assuming 2.5%WL at week 5. The %WL from baseline to week 5 was a significant predictor of %WL at 4 months, and a balanced comparison of %WL should condition on equal weight loss at week 5. Thus, 2.5%WL was chosen, as it was the boundary of weight loss for both groups. A 90% CI was constructed for the mean %WL from baseline to 4 months for each group and for the mean difference between groups. If the 90% CI was contained in the (−2, +2) predefined indifference zone, then equivalence was established of the mean of the two groups with a level of significance of 0.05. Two one-tailed t tests also were conducted to compare the estimated mean difference with −2 and +2, respectively. If the P values were >0.05 in each case, equivalence was established.
An a priori power analysis showed a two-sample t test of equivalence with equal sample sizes of 103 each, threshold of indifference at 2%WL, and α level of 0.05, had 84% power when there was no difference in the mean weight loss of the two groups.
Results
There were no significant differences in demographic characteristics between groups, except the GLB+ group included more women (81.08% of group) than the GLB group (62.50% of group; P = 0.0050) (see Supplementary Table). Mean %WL was significantly greater for GLB compared with GLB+ at 4 months, but both groups experienced significant weight loss (both P < 0.0001) (Table 1). In GLB, 75% and 60% lost ≥5% and ≥7% of body weight, respectively; whereas in GLB+, 29.7% and 13.5% lost ≥5% and ≥7% of body weight, respectively.
Percentage weight change and clinical outcomes at baseline by group and the change in outcomes by group at the 4-month follow-up
| . | Baseline (mean ± SE) . | Change from baseline at 4 months (mean ± SE) . | ||||
|---|---|---|---|---|---|---|
| Outcome . | GLB (n = 80) . | GLB+ (n = 111) . | P value† . | GLB (n = 80) . | GLB+ (n = 111) . | P value† . |
| Weight change (%) | — | — | — | −8.13 ± 0.48*** | −3.46 ± 0.31*** | <0.0001†† |
| Body weight (kg) | 102.60 ± 2.71 | 102.46 ± 2.30 | 0.9701 | −8.36 ± 0.48*** | −3.63 ± 0.41*** | <0.0001 |
| BMI (kg/m2) | 35.68 ± 0.85 | 37.07 ± 0.72 | 0.2119 | −2.84 ± 0.17*** | −1.27 ± 0.14*** | <0.0001 |
| AIC (%) | 5.69 ± 0.03 | 5.68 ± 0.03 | 0.7324 | −0.20 ± 0.02*** | −0.12 ± 0.02*** | 0.0045 |
| A1C (mmol/mol) | 38.69 ± 0.36 | 38.53 ± 0.30 | 0.7324 | −2.21 ± 0.25*** | −1.26 ± 0.21*** | 0.0045 |
| Fasting glucose (mmol/L)‡ | 6.06 ± 0.06 | 5.97 ± 0.05 | 0.2708 | −0.39 ± 0.06*** | −0.18 ± 0.05** | 0.0112 |
| Total cholesterol (mmol/L) | 5.08 ± 0.11 | 5.12 ± 0.10 | 0.7991 | −0.30 ± 0.08*** | 0.03 ± 0.07 | 0.0020 |
| LDL cholesterol (mmol/L) | 3.0 ± 0.10 | 3.1 ± 0.09 | 0.4517 | −0.19 ± 0.07** | −0.06 ± 0.06 | 0.1733 |
| HDL cholesterol (mmol/L) | 1.32 ± 0.05 | 1.34 ± 0.04 | 0.7876 | −0.05 ± 0.03 | 0.02 ± 0.02 | 0.1023 |
| Triglycerides (mmol/L)§ | 0.38 ± 0.04 | 0.36 ± 0.04 | 0.6988 | −0.15 ± 0.04*** | 0.08 ± 0.04 | <0.0001 |
| Blood pressure | ||||||
| Systolic (mmHg) | 127.06 ± 1.64 | 123.92 ± 1.39 | 0.1457 | −4.87 ± 1.63** | −1.19 ± 1.37 | 0.0846 |
| Diastolic (mmHg) | 85.84 ± 1.08 | 84.34 ± 0.91 | 0.2899 | −3.39 ± 1.18** | −0.65 ± 1.0 | 0.0780 |
| . | Baseline (mean ± SE) . | Change from baseline at 4 months (mean ± SE) . | ||||
|---|---|---|---|---|---|---|
| Outcome . | GLB (n = 80) . | GLB+ (n = 111) . | P value† . | GLB (n = 80) . | GLB+ (n = 111) . | P value† . |
| Weight change (%) | — | — | — | −8.13 ± 0.48*** | −3.46 ± 0.31*** | <0.0001†† |
| Body weight (kg) | 102.60 ± 2.71 | 102.46 ± 2.30 | 0.9701 | −8.36 ± 0.48*** | −3.63 ± 0.41*** | <0.0001 |
| BMI (kg/m2) | 35.68 ± 0.85 | 37.07 ± 0.72 | 0.2119 | −2.84 ± 0.17*** | −1.27 ± 0.14*** | <0.0001 |
| AIC (%) | 5.69 ± 0.03 | 5.68 ± 0.03 | 0.7324 | −0.20 ± 0.02*** | −0.12 ± 0.02*** | 0.0045 |
| A1C (mmol/mol) | 38.69 ± 0.36 | 38.53 ± 0.30 | 0.7324 | −2.21 ± 0.25*** | −1.26 ± 0.21*** | 0.0045 |
| Fasting glucose (mmol/L)‡ | 6.06 ± 0.06 | 5.97 ± 0.05 | 0.2708 | −0.39 ± 0.06*** | −0.18 ± 0.05** | 0.0112 |
| Total cholesterol (mmol/L) | 5.08 ± 0.11 | 5.12 ± 0.10 | 0.7991 | −0.30 ± 0.08*** | 0.03 ± 0.07 | 0.0020 |
| LDL cholesterol (mmol/L) | 3.0 ± 0.10 | 3.1 ± 0.09 | 0.4517 | −0.19 ± 0.07** | −0.06 ± 0.06 | 0.1733 |
| HDL cholesterol (mmol/L) | 1.32 ± 0.05 | 1.34 ± 0.04 | 0.7876 | −0.05 ± 0.03 | 0.02 ± 0.02 | 0.1023 |
| Triglycerides (mmol/L)§ | 0.38 ± 0.04 | 0.36 ± 0.04 | 0.6988 | −0.15 ± 0.04*** | 0.08 ± 0.04 | <0.0001 |
| Blood pressure | ||||||
| Systolic (mmHg) | 127.06 ± 1.64 | 123.92 ± 1.39 | 0.1457 | −4.87 ± 1.63** | −1.19 ± 1.37 | 0.0846 |
| Diastolic (mmHg) | 85.84 ± 1.08 | 84.34 ± 0.91 | 0.2899 | −3.39 ± 1.18** | −0.65 ± 1.0 | 0.0780 |
Analyses are based on mixed effect models where treatment group, time, and group by time interaction are the fixed effects and subjects are random effects using JMP 15 software (SAS Institute, 2019). A P value <0.0125 was used for statistical significance to account for the Bonferroni correction of four comparisons of between- and within-group changes.
P value based on Welch t test.
Six individuals removed from analyses due to nonfasting sample.
Variable was modeled on the logarithmic scale, and P values are for transformed data. Summary statistics are the original scale for data reporting. One outlier was removed from analyses. Within-group change from baseline
P < 0.0125;
P < 0.001.
At 4 months, participants in GLB achieved significantly greater reductions in A1C, glucose, total cholesterol, and triglycerides compared with GLB+. Analysis of within-group change found that GLB achieved significant reductions in all outcomes except for HDL cholesterol. GLB+ achieved significant reductions in weight, A1C, and glucose.
Week 5%WL predicted %WL at 4 months for both groups (P < 0.0001). Between-group comparison of weight loss at 4 months should condition on equal weight loss at week 5. The 90% CI for the mean difference between GLB and GLB+, assuming 2.5%WL at week 5, was −1.39% to 1.52%. Thus, means were equivalent (within 2% difference) according to the predefined criterion, and the null hypothesis of nonequivalence was rejected. The two one-tailed t tests that compared the estimated mean difference with −2 and +2, respectively yielded P values of 0.01 and 0.0144. Since the 90% CI, and consequently 95% CI contained 0, the mean %WL at 4 months was not significantly different between groups assuming 2.5%WL at week 5.
Conclusions
This is one of few trials to evaluate the impact of an adaptive intervention for early slow weight loss responders among adults with prediabetes. Both GLB and GLB+ core interventions facilitated significant weight loss and improvement in glycemia. However, the first month of intervention is a critical time for initiating weight loss efforts.
Given the adaptive intervention design, a between-group comparison of weight loss should consider treatment stratification and balance the comparison by starting from the same reference point. There was no significant between-group difference in mean weight loss at 4 months assuming someone achieved 2.5%WL at week 5. Prior research found that slow weight loss responders lost a mean of 1.3% to 3.94%WL at 3 months (12–14). The weight loss achieved by GLB+ (3.46%) exceeded the weight loss achieved by slow responders in two of the prior studies where no additional intervention was provided and was similar to the third study where preportioned food was provided free of charge (14). No food was provided to GLB+ participants.
Results from the DPP found that diabetes risk decreased by 10% for each percentage point of weight loss achieved in the lifestyle arm, independent of change in glycemia (15). Therefore, even modest weight loss in GLB+ likely promotes diabetes risk reduction. Furthermore, GLB+ was implemented without additional contact time, as the intervention dose was held constant with the same number and duration of sessions as GLB.
In summary, clinical practice should assess weight loss following the 1st month of intervention and provide rescue treatment to early slow responders to improve weight and glycemia among adults with prediabetes. Future research is needed to identify predictors of early weight loss success.
Clinical trial reg. no. NCT03382873, clinicaltrials.gov
This article contains supplementary material online at https://doi.org/10.2337/figshare.20388144.
Article Information
Acknowledgments. The authors gratefully acknowledge the commitment and dedication of the study participants.
Funding. Research reported in this publication was supported by the National Institutes Health National Institute of Diabetes and Digestive and Kidney Diseases grant R01DK112930 and by National Center for Advancing Translational Sciences award number UL1TR002733.
The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Center for Advancing Translational Sciences or the National Institutes of Health.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. C.K.M. oversaw participant recruitment and enrollment, contributed to development of both interventions, oversaw intervention implementation, and wrote the manuscript. H.N.N. performed statistical analyses. J.C. contributed to GLB+ intervention development and reviewed and edited the manuscript. K.F. contributed to GLB+ intervention development and reviewed and edited the manuscript. S.L. contributed to implementation of the GLB+ intervention and reviewed and edited the manuscript. H.N.N. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
