Switching diets after 6-months does not result in renewed weight loss: a secondary analysis of a 12-month crossover randomized trial

0
Switching diets after 6-months does not result in renewed weight loss: a secondary analysis of a 12-month crossover randomized trial

Full details of the study design and outcomes from the parent study have been described elsewhere26. Here we briefly describe methods relevant to this analysis.

Study design

The parent study was a 2 × 2 design assessing the effectiveness of a LF versus LC diet among insulin resistant (IR) and insulin sensitive (IS) adults who were overweight or obese26. We suggest the terms “insulin resistance” and “insulin sensitivity” here be interpreted cautiously because we used a proxy measure for this, rather than a direct measure (expanded discussion is provided elsewhere26). Briefly, insulin resistance status was assessed by calculating an area under the curve of insulin concentrations from four blood samples taken during an oral glucose tolerance test (OGTT) (time: 0, 30, 60, and 120 min)26. Individuals above the median (separately for men and women) were considered more IR and individuals below the median were considered more IS. Participants were then assigned to either their first diet assignment (LF or LC) via a random number generator in Microsoft Excel.

Participants

Participants were recruited from the local Palo Alto, CA community primarily through media advertisements. Premenopausal women and men aged 18–50 years were invited to enroll if their BMI was 28–40 kg/m2, body weight was stable over the previous 2 months, and medications were stable for ≥ 3 months. Potential participants were excluded if they self-reported hypertension (except for those stable on antihypertension medications); type 1 or 2 diabetes mellitus; heart, renal, or liver disease; cancer or active neoplasms; hyperthyroidism unless treated and under control; taking any medications known to affect weight/energy expenditure or blood lipids; smoking; alcohol intake ≥ 3 drinks/day; pregnancy, lactation, no menstruation for the previous 12 months, or plans to become pregnant within the next year. All study participants provided written informed consent. Procedures for this study were followed in accordance with the ethical standards from the Declaration of Helsinki. The study was approved by the Stanford University Human Subjects Committee (Protocol ID: 23438, approved: 2013.12.17).

Intervention

The intervention consisted of two 6-month phases. Participants ate an assigned diet (healthy LF or LC) for 6 months then switched to the opposite diet for an additional 6 months. During each phase, participants received 14 1-h nutrition education classes led by a health educator. Classes were delivered in-person once a week for the first 8 weeks, every other week for the next 8 weeks, and once a month for the last 8 weeks.

The curriculum of the nutrition education classes emphasized the Limbo-Titrate-Quality approach for defining a healthy LF and LC diet. There were three components of this approach. The first component was “Limbo,” or “How low can you go?” This involved participants in the LF group reducing their total daily fat intake to 20 g/day or less and participants in the LC group reducing their total daily carbohydrate intake to 20/g a day or less for the first 8 weeks. The second component, “Titrate,” involved participants incrementally adding back 5 g of fat or carbohydrates per day to their assigned diet for 1–4 weeks (e.g., going from 20 g of total fat to 25 g of total fat for the LF group). An important part of the second component was for participants to identify the lowest level of daily fat or carbohydrates they felt that they could maintain long term. The third component was “Quality,” which emphasized diet quality. Participants were encouraged to consume nutrient dense foods, fresh vegetables and fruits, and to prepare meals at home while avoiding heavily processed foods, foods with added sugars, refined white flour products, and foods with trans fats. In summary, the Limbo-Titrate-Quality approach was designed to motivate participants to achieve the lowest possible level of fat or carbohydrate intake, that is, an approach that was equally ambitious with maximal overall nutritional quality and a dietary pattern that could be continued for a lifetime.

There were no caloric restriction requirements for this dietary intervention. Nutrition education classes also addressed mindful eating, body acceptance, sugar addiction, getting adequate sleep, and maintaining healthy levels of physical activity. Participants were encouraged to track their dietary intake using daily food journals. Participants were also encouraged to be physically active and were provided with pedometers (Omron HJ-112 Digital Pocket Pedometer) to track their activity.

Sociodemographics

Self-reported sociodemographic data on age, gender, race/ethnicity, marital status, education, and employment status were collected at baseline.

Anthropometric measures

Participants’ current body weight was measured to the nearest 0.1 kg at each time point (baseline, 3, 6, 9, and 12 months) using a calibrated scale (Scaletronix). Participants’ height was measured at baseline to the nearest millimeter using a standard wall-mounted stadiometer. Average daily energy expenditure was assessed using the Stanford 7-day physical activity recall39.

Dietary data

Dietary intake data was collected via 3 unannounced, 24-h dietary recalls within a 2-week time window at each time point (baseline, 3, 6, 9, and 12 months) using the Nutrition Data System for Research (NDS-R) software [Nutrition Coordinating Center (NCC), University of Minnesota, versions 4.05.33 (2011) and 4.06.34 (2012)]. Recalls were conducted on two weekdays and one weekend day, nonconsecutive whenever possible.

Cardiovascular risk factors

Blood samples for analysis of plasma lipids (including high-density lipoprotein (HDL-C), low-density lipoprotein (LDL-C), and triglycerides), and insulin and glucose were collected after participants fasted for ≥ 10 h. Insulin was collected at baseline, 6, and 12 months. All other outcomes were collected at baseline, 3, 6, 9 and 12 months. HDL-C was measured by liquid selective detergent followed by enzymatic determination of cholesterol40. LDL-C was calculated according to Friedewald et al. equation41. Total plasma insulin in serum was measured by radioimmunoassay42, and blood glucose was measured using a modification of the glucose oxidase/peroxidase method (Diabetes Research Center, Washington University, St Louis, MO)43,44.

Statistical analyses

Participant demographics and baseline clinical characteristics were summarized overall and by arm (diet order) as mean (standard deviation) or n (percent) for continuous and categorical variables, respectively.

For the primary analysis, we characterized percent weight change (PWC) before and after introducing the second diet in the crossover. The primary outcome is percent weight change at 3–6 months versus the percent weight change at 6–9 months. Absolute weight change is presented for visual observation. As prespecified in the statistical analysis plan, the primary analysis includes all available data. We fit a linear mixed model with fixed effects: categorical time (3, 6, 9, 12 months with 6 months as the reference since it is the end of the first diet and occurs before starting the new diet), order (e.g., study arm), insulin status (resistance vs sensitive), gender; and with a random effect to account for the correlated observations over time of each participant. We performed visual inspections for the model assumptions: normality of residuals and homogeneity of variance, using Q-Q plots and scatterplots, respectively. We provide model estimates for the difference in percent weight change at each time period (relative to the 6-month timepoint) along with 95% confidence intervals (CIs). These estimates account for the cross-over design and stratification randomization variables: insulin status and gender. Additionally, we also presented crude estimates for each time period, i.e., average percent weight change (95% CI). In a stratified analysis, we fit a similar linear mixed model by diet order (excluding study arm as a fixed effect) and present model estimates with 95% CIs. Additionally, we present mean percent weight change (95% CI) by diet order and insulin status.

For the secondary outcomes of LDL-C, HDL-C, triglycerides, fasting glucose, and fasting insulin, we fit a linear mixed model similar to the primary model and present model and crude estimates with 95% CIs, stratified by diet order. For blood lipids, a log transform was used on the outcome in the model to resolve departures from normality.

In a subgroup analysis, we fit a linear mixed model similar to the primary model, but included only those participants that were weight stable or gaining before the diet change at 6 months (i.e., those that can re-start weight loss). We considered participants to be weight stable or gaining if their percent weight change at the end of phase 1 (6 month timepoint) was greater than or equal to − 2 percent (weight loss) relative to the weight change at the 3-month timepoint. Estimates overall and by diet order are provided. Also, we presented crude estimates with 95% CI using only weight stable or gaining participants for the secondary outcomes.

Last, we characterized the percent weight change observed in this study (i.e., 12-month trial comparing low-fat to low-carbohydrate with a diet crossover at 6 months) and the percent weight change data from the DIETFITS dietary clinical trial (i.e., 12-month trial comparing low-fat to low-carbohydrate with the same diet for the entire study duration).

Data were analyzed in RStudio (Version 1.2.5042, RStudio Team, 2020, PBC, Boston, MA, USA).

link

Leave a Reply

Your email address will not be published. Required fields are marked *