Background
Evidence from clinical trials shows that behavioural interventions aimed at helping individuals to acquire the knowledge and skills to change dietary intake, physical activity, and other weight-related behaviours can lead to successful weight loss [
1]. Behavioural weight loss interventions typically encourage changes in dietary intake that result in reduced energy intake, as well as strategies to change behaviours such as self-monitoring, goal setting and implementation plans [
2,
3]. However, relatively few behavioural weight loss interventions have detailed the changes that occur in dietary intake or their associations with weight change [
4‐
8]. This information may help to identify the effective characteristics of interventions [
9].
Behavioural interventions delivered in commercial group-based programmes usually result in greater weight losses than those led by primary care staff, which might be explained by greater reductions in energy intake and/or specific changes in diet quality [
10,
11]. Here we report changes in dietary intake and test their relationships with weight loss using data from a 12 month, multicenter, randomised controlled trial (RCT) that compared weight loss after referral to free access to a commercial weight loss programme (CP) or weight management advice provided as standard care (SC) in a primary care setting [
10]. In this secondary analysis of RCT data, we investigated changes in self-reported dietary energy density, macronutrient (total fat, saturated fat, protein, carbohydrate, sugars) and fibre intakes as indicators of diet quality or because of their associations (probable or confirmed) with body weight [
12‐
20]. We have previously shown that participants randomised to the CP lost more weight than those receiving SC [
10]. Accordingly, we hypothesised that participants who followed the CP would have achieved greater reductions in dietary energy density, total energy, total fat, saturated fat, total carbohydrate and sugar intake, and greater increases in dietary fibre and protein intake than the SC group at 6 and 12 months after baseline. We also examined which of these dietary changes were associated with the greatest weight loss at 6 and 12 months.
Discussion
This longitudinal analysis of a behavioural weight loss intervention has revealed that participants in both intervention groups achieved significant improvements in diet quality. These improvements persisted to 12 months from baseline and included reductions in total energy intake, dietary energy density, total fat (g/d and g/MJ), saturated fat (g/d and g/MJ), carbohydrates (g/d) and sugars (g/d), along with increases in dietary fibre density and protein density (g/MJ). Participants who followed a group-based commercial programme achieved greater improvements in dietary fibre density and protein density (in addition to greater reductions in total energy intake, total fat, and saturated fat) than those in standard care. Regardless of intervention group, only increases in dietary fibre density and protein density were associated with weight loss.
An intention to treat analysis of data from all study centres in this RCT previously showed that those randomised to the CP lost significantly more weight (adjusted difference: 2.77 kg, 95% CI: 2.03, 3.50 kg; based on last observation carried forward) at 12 months after baseline [
10]. This association was robust when different methods were used to treat missing weight data, as well as completers-only analyses [
10]. The present explanatory analysis restricted to participants from the Australian and UK centres, showed similar differences in weight loss compared to the full trial. Taken together, this suggests that greater increases in dietary fibre density and protein density among participants in the CP group may have contributed to their greater weight loss, compared to SC.
Evidence from observational and experimental studies suggest that increased fibre intake may prevent weight gain, through reductions in appetite and energy intake [
19]. A high fibre intake is associated with diets of lower energy density and total energy intake, as fibre increases the weight of food consumed (leading to greater satiety) without providing additional energy [
33]. Mechanisms linking dietary fibre with enhanced appetite control include increased food chewing and gut transit times, enhanced gut hormone secretion and fibre fermentation in the colon (leading to greater short chain fatty acids), which may enhance satiety [
19]. However, the Scientific Advisory Committee on Nutrition review of Carbohydrates and Health failed to find a consistent association between dietary fibre and body weight in RCTs or cohort studies [
20]. The effect on body weight may depend on fibre type, but no clear dose-response relationships for individual fibre types have been identified [
19].
Meta-analyses of short and long term (≥12 months) trials indicate that higher protein diets can lead to greater weight loss over the short term and better weight loss maintenance [
18]. Dietary protein is proposed to influence key metabolic targets that may enhance weight loss, including sustaining satiety during negative energy balance, maintaining basal energy expenditure despite weight loss, and the sparing of fat free mass [
34]. It has been suggested that optimum protein intake for weight loss can be achieved by maintaining absolute protein intake while reducing carbohydrate and fat intake in an energy restricted diet [
34]. Consistent with this, we observed no significant changes in absolute protein intake (g/d) and accordingly, reductions in absolute total fat and carbohydrate intakes are likely to have led to the observed increases in dietary protein density (g/MJ), which were linked to weight loss.
Not all of the dietary changes investigated in this study were associated with weight loss, but the observed improvements in diet quality may confer other health benefits. Replacing saturated fats with polyunsaturated fats reduces the risk of coronary heart disease [
15,
35,
36] and may help in the prevention and management of Type 2 diabetes [
15,
37]. Evidence from RCTs suggests that a reduction in sugar intake may lead to improvements in blood pressure and blood lipids [
38], and lower intakes of sugar-sweetened beverages are associated with a decreased risk of Type 2 diabetes [
20].
This study adds to the small body of evidence of improvements in energy, macronutrient, and fibre intakes reported by other comparable behavioural weight loss interventions. A 12-week commercial behavioural weight loss intervention for men (the “SHED IT” trial) [
4] reported that after 3 and 6 months, men who received self-help resources plus online support for weight loss significantly reduced their intakes of total fat, saturated fat, carbohydrate and sugars, and increased their energy from protein, but made no significant changes in fibre intake [
4]. In another Australian web-based commercial weight loss intervention, no significant changes in macronutrient or fibre intake were observed after 12 weeks [
5]. The precision and statistical power of these studies may have been reduced however, by using a FFQ to detect dietary changes over a short time period, and by focussing on a relatively small proportion of people who lost ≥5% of their body weight.
The PREMIER trial (
n = 745) in the US tested two 18-month behavioural lifestyle interventions: one including established recommendations (EST); one including EST plus support to adopt the Dietary Approaches to Stop Hypertension (DASH) diet (EST + DASH); or an advice only group [
8]. All groups significantly decreased their mean total energy intake at 6 and 18 months [
6]. In addition, both intervention groups significantly decreased mean proportions of energy from total fat, saturated fat, other fats, and carbohydrate, with the greatest changes in the EST + DASH group. The EST + DASH group also significantly increased mean fibre intake and energy from protein [
6].
Following on from the PREMIER trial, the US Weight Loss Maintenance Study included an initial 6 month intensive behavioural weight loss intervention (Phase I) in which all participants (
n = 1685) were encouraged to follow the DASH diet [
7]. Dietary changes were reported for 828 participants who had lost at least 4 kg body weight at the end of Phase 1. These changes included significant reductions in total energy intake and energy from fat, and significant increases in fibre intake and energy from carbohydrate and protein [
7].
Not all studies report the relationship between changes in dietary intake and weight loss. However in the SHED IT trial, those who lost at least 5% of their baseline weight (
n = 49, 18%) reported significantly greater reductions in total energy, energy from sugars, and greater increases in energy from protein (but not fat, carbohydrate, or fibre) compared with unsuccessful completers. In the PREMIER trial, a secondary analysis including 501 participants who were overweight or obese at baseline (68% obese) found that a 1% decrease in energy from total fat between baseline and 6 months was associated with a 0.06% decrease in weight (
p < 0.05) [
8]. Unlike the present study, no associations were observed between weight change and energy from protein, and dietary fibre was not examined (no associations were observed between weight loss and total energy or total carbohydrate intake) [
8]. In the initial 6 month intensive behavioural weight loss intervention (Phase 1) of the Weight Loss Maintenance study, it was found that substituting dietary fat with protein or carbohydrate, or substituting carbohydrate with protein, was associated with greater weight loss at 6 months. Unlike our present study, fibre intake was not associated with greater weight loss [
7]. This study benefited from a large sample size, however, by restricting the analysis to only those participants who lost at least 4 kg weight and who completed a FFQ at four time points, the likelihood of selection bias was increased.
Collectively, these studies indicate that structured dietary programmes can achieve specific improvements in dietary intake which may enhance weight loss. An increase in energy from protein was frequently linked with weight loss but the effects of total fat, carbohydrates and fibre are less clear.
Unlike earlier comparable studies that used an FFQ to assess dietary changes [
4,
5,
7] the present study included 4-day food diaries that were linked to country-specific food composition data to provide detailed information on food intake at three time points over 12 months. The food diary or record is an ideal dietary assessment as it collects rich data on food choice, portion size and cooking methods at the meal-level, which are highly relevant to understanding eating behaviours conducive to weight loss. Furthermore, food diaries can capture dietary intake over a short time period and are well suited as repeated dietary assessments over the duration of a trial. Whereas, a FFQ does not collect detailed data at the meal level and by design, is better suited for summarising usual dietary intake over an extended time period, typically 1 year. However, it is recognised that dietary under-reporting is an issue with all dietary assessments that are based on self-report [
25]. We did not attempt to adjust for dietary under-reporting because all respondents were expected to be purposefully reducing their energy intake in this weight loss intervention, and current methods to identify dietary under-reporting assume energy balance. We did not use a baseline indicator of under-reporting and assume this remained constant over the course of the trial, as this would be likely to incorporate additional misclassification error. However, the use of nutrient densities in this analysis would have mitigated the impact of dietary under-reporting to some degree [
25]. Furthermore, our analysis examined individual-level changes in dietary intake over time (using longitudinal models), rather than comparing group-level mean intakes, as is commonly reported. Including data on dietary intake and weight up until 12 months after baseline is another advantage, as many published studies of this nature are brief interventions averaging 12 weeks in duration or report associations up to 6 months [
5,
7,
8].
High participant dropout rates are very common in weight loss trials [
39] and a high dropout rate (non-completion of dietary assessments) was evident in this secondary analysis. As changes in weight and/or dietary intake may have differed in completers and non-completers, a biased study population may have resulted from participant dropout, reducing the generalisability of our findings. Gender bias is also not uncommon in weight loss interventions. Our study population was predominantly female (88%) and therefore, the results are less generalisable to men. However, the use of multivariate, longitudinal models to examine dietary changes and their relationships with weight loss at three time points is a major study strength. Mixed effects models utilised all available follow up data rather than limiting the analyses to those who completed all follow ups only. The multivariate models analysed changes in dietary intake taking account of baseline values and individual level changes while adjusting for relevant confounders. Furthermore, in our analysis we separated the cross-sectional from the longitudinal associations between changes in dietary intake and weight loss.