Plain English summary

Obesity leads to a decrease in health-related quality of life (HRQOL) and mood. Lifestyle strategies may improve HRQOL and mood. However, the efficacy of adding exercise to a weight-loss diet on HRQOL and mood is unclear. Determining combined effects of weight-loss diet and exercise compared with weight-loss diets alone on mood status will help physicians, researchers, and policy makers in prescribing optimal lifestyle to improve HRQOL and depression. In this study, we assess the effects of weight-loss diet plus exercise compared with weight-loss diets alone on HRQOL and depression. Our findings did not reveal any beneficial effects of adding exercise to a weight-loss diet on HRQOL and depression in overweight and obese adults. Further studies are needed to investigate the benefits of exercise plus diet intervention on mood status. Given the well-known benefits of exercise on different aspects of health, exercise should be recommended as an important component of lifestyle strategies for obesity management.

Introduction

Obesity remains a global health challenge [1] as prevalence has nearly tripled since 1975 [2]. Obesity and related co-morbidities (such as cardiovascular diseases, type 2 diabetes mellitus, fatty liver disease, certain cancers, Alzheimer, and depression) [3] are associated with a reduction in health-related quality of life (HRQOL) [4, 5]. Both direct and indirect costs of impaired HRQOL and depression (such as the costs related to diagnosis, treatment, unemployment, and loss of income) are considerable [6,7,8,9].

Given that obese individuals are at a higher risk for depression and mood disorders [10], managing obesity is critical. Adopting lifestyle interventions (such as diet, physical activity, or behavioral therapy) that are cost-effective and have few side effects are initial non-pharmacological approaches for obesity management [11]. Data from the National Weight Control Registry (NWCR) demonstrates that both changes in diet and physical activity are key strategies for obesity treatment [12]. Dietary weight-loss interventions are effective for body composition changes; however, systematic reviews and meta-analyses assessing the effects of weight-loss diets on depression [13, 14] and HRQOL [15] are inconsistent. Adopting a calorie-restricted diet might reduce some aspects of HRQOL, including social and economic benefits [15]. Systematic reviews and meta-analyses assessing the effects of exercise during weight-loss interventions on the HRQOL [16, 17] and depression [18, 19] have shown beneficial effects.

There are a large number of systematic reviews and meta-analyses directly comparing the effects of diet alone or in conjunction with exercise on the improvement of health outcomes [20,21,22,23,24]. However, to the best of our knowledge, only one systematic review compares energy-restricted diets and combined energy-restricted diet and exercise on HRQOL [25]; and no meta-analysis was performed. The authors indicated that a definitive conclusion was not feasible based on existing evidence, and a more detailed and comprehensive systematic review and meta-analysis was needed. Moreover, previous clinical trials examining the effect of adding exercise to energy-restricted diets on HRQOL [26,27,28,29,30,31,32] or depression [27, 33,34,35,36] were also inconsistent. Although some studies found a beneficial effect on depression [28, 33] and HRQOL [26, 29, 30, 34], others reported null or negative effects [5, 27, 34, 35, 37]. For instance, Thomson et al. stated that the emotional response to diet plus exercise interventions might reduce over time, negatively affecting depressive symptoms. Highly perceived time pressure induced by work and family responsibilities can be a major barrier to healthy eating and participation in physical activity [5]. In addition, Bowen et al. found that intensive exercise may harm some important aspects of HRQOL (body or joint pain, social interactions, and mood), especially in initially sedentary individuals [38]. Determining the combined effects of diet and exercise interventions compared with energy-restricted diets alone on mood status will assist physicians, researchers, and policymakers in prescribing optimal lifestyle modifications to improve HRQOL and depression [33]. Therefore, the objective of the present meta-analysis was to critically assess whether the addition of exercise to energy-restricted dietary programs in adult populations has a greater beneficial effect on HRQOL and mood status when compared with energy-restricted diets alone.

Methods

The current study followed the preferred reporting guidelines for systematic reviews and meta-analyses (PRISMA) [39], and was registered in the international prospective register of systematic reviews (PROSPERO, www.crd.york.ac.uk/PROSPERO; identifier: CRD4202173434). The detailed design and rationale have been described in detail elsewhere [40].

Literature search

PubMed/MEDLINE, EMBASE, ISI (Web of sciences), Scopus, and Google Scholar were searched up to Apr 2021 by using Medical Subject Headings (MeSH) and non-MeSH keywords outlined in Supplementary Table 1. No language restriction was applied for searching the databases. The reference lists of all relevant studies were checked to identify any additional studies.

Eligibility criteria

Titles, abstracts, and full texts were screened and crosschecked by six reviewers (Z.Y, S.S, SH.R, S.B, S.MT, and T.Z) based on the following pre-defined inclusion criteria:

  1. (i)

    Studies conducted in overweight or obese adults (age ≥ 18 years and BMI ≥ 25 kg/m2);

  2. (ii)

    Studies published as original articles using randomized or non-randomized controlled clinical trial study designs (either parallel or cross-over);

  3. (iii)

    If there were multiple published reports from the same study, we included the publication with the longest follow-up or a greater number of included participants;

  4. (iv)

    Studies that compared a weight-loss diet plus any type of exercise (i.e., aerobic or resistance) with an isocaloric weight-loss diet;

  5. (v)

    Studies reporting mean baseline, after the intervention, or change values, and their corresponding standard deviations (SDs) for depression, SF36-MCS (SF36-mental component summary), SF36-PCS (SF36-physical component summary), or total scores of disease-specific HRQOL questionnaires.

Studies were excluded from the systematic review if they:

  1. (i)

    Were conducted on animals, children, adolescents, professional athletes, and pregnant, or lactating women;

  2. (ii)

    Were published as reviews, observational studies, or study protocols;

  3. (iii)

    Were single arm in design;

  4. (iv)

    Included different types of weight-loss diets in both arms;

  5. (v)

    Included non-isocaloric diets in both arms;

  6. (vi)

    Included follow-up durations of less than two weeks because the time interval would be too short for effective weight-loss diets;

The consensus opinion of the reviewers resolved any disagreement about study selection.

Data extraction and management

As duplicating the data extraction process reduces both the risk of making mistakes and a single person’s biases [41], the data were extracted by two independent investigators (SH.R, T.Z) from eligible studies and any disagreement was resolved by discussion with the principal investigator (A.S-A).

Study and participants’ characteristics: the study design (parallel/cross-over), geographical location and duration; the number of participants in each arm, the participants’ age, sex, health condition, and being physically active or sedentary at baseline; Intervention details: The characteristics of the exercise program (including exercise duration, intensity, frequency, and being conducted under supervision or not); Comparison details: The characteristics of the weight-loss diet (including the amount of calorie-restriction, type of diet, and macronutrient distributions); Outcome measures: The type of questionnaire; assessing HRQOL and depression as primary outcomes or not; baseline, post-intervention, or change from baseline mean ± standard deviation (SD) for depression or quality of life scores, P-values for within-group and between-group comparisons.

Risk of bias assessment in individual studies

The quality of included articles was assessed using the Cochrane collaboration risk of the bias assessment tool [42] considering seven domains: (i) random sequence generation, (ii) allocation concealment, (iii) blinding of participants and personnel, (iv) blinding of outcome assessment, (v) incomplete outcome data, (vi) selective reporting, and (vii) dietary compliance. Studies were categorized as low risk of bias, high risk of bias, or unclear risk of bias for each domain. To obtain the overall quality of studies, we did not take the “blinding of participants and personnel domain” into account because of the difficulty in blinding participants and personnel in lifestyle modification trials. Considering the remaining domains (random sequence generation, allocation concealment, blinding of outcome assessment, incomplete outcome data, selective reporting, and dietary compliance) as key domains, the studies were categorized as low risk (low risk for all domains), unclear risk (unclear risk for at least one domain), and high risk (high risk for at least one domain). Any disagreement was resolved through discussion with the principal investigator (A.S.A).

The overall quality of evidence

The quality of evidence for each outcome was assessed using The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach based on the following domains: assessed risk of bias, inconsistency, indirectness, imprecision, and publication bias [43,44,45]. The quality of evidence was categorized as high, moderate, low, and very low.

Statistical analyses

The difference in mean changes and their standard deviations (SDs) between the two arms of each study was used as the effect size for SF36-MCS and SF36-PCS quantitative synthesis. Depression and HRQOL disease-specific questionnaire scores were assessed using different tools in different studies. For this reason, standardized mean differences are calculated and Hedges’g statistic is used.

Two [35, 46] out of six [5, 27, 33,34,35, 46] studies reported the mean change values for depression. Therefore, the SDs for mean change values were calculated using correlation r of studies [35, 46] that reported baseline, after intervention and change values (r = 0.57 for depression).

Because none of the studies provided data on changes in SF36 and HRQOL disease-specific scores, the SDs for change values were calculated by using a correlation coefficient of 0.5. The meta-analysis was repeated using correlation coefficients of 0.2 and 0.8 to check the sensitivity of the meta-analysis to selected correlation coefficients. The overall estimates were calculated using both the DerSimonian and Laird random-effects model [47] and the inverse variance heterogeneity (IVhet) model [48] which has more coverage likelihood compared to DerSimonian and Laird random-effects model.

Between-study heterogeneity was assessed using Cochran’s Q-test and I-squared (I2) statistic [49]. The Cochran’s Q test with a P-value of < 0.05 was considered a statistically significant heterogeneity, and I2 with values of greater than 75% was considered the between-study heterogeneity [40].

To explore the potential source of heterogeneity, several subgroup analyses were performed based on: country (Astralia/ Canada/ Iran/ USA/ Netherlands), sex of participants (male / female / both), intervention duration (≤ 12 weeks / > 12 weeks), type of questionnaire [Hospital Anxiety and Depression Scale (HADS), Geriatric Depression Scale (GDS)], outcome (primary / secondary), intervention under supervision (yes / no / partially / not mentioned), risk of bias assessment (low / unknown / high), baseline physical activity status (not mentioned / sedentary), additional weight management program (not mentioned / behavior modification / routine classes), type of exercise program (strength / aerobic / strength plus aerobic / not mentioned), type of dietary intervention (diet / education / guideline), and reporting data (complete / incomplete). In addition, meta‐regression analysis was performed to evaluate if prespecified factors can explain the variation between studies.

Publication bias for each outcome was assessed by inspecting funnel plots and conducting statistical asymmetry tests (Egger’s regression asymmetry test [50] and Begg’s adjusted rank correlation test [51]), if there were at least ten studies accessible for meta-analysis [42]. Sensitivity analysis was performed to assess the robustness of overall effects by sequentially removing individual studies from the meta‐analysis [52]. All of the analyses were performed using STATA, version 11.2 (Stata Corp, College Station, TX), and two-sided P values ≤ 0.05 were considered statistically significant.

Results

Literature search

A total of 42,189 publications were identified during the initial database literature search; of those, 469 articles were included in the full-text screening stage. Out of these, 454 articles were excluded (119 studies were repeated publications on the same study; the energy deficit of diets was not equal in the diet alone and diet plus exercise groups for 34 trials; the full texts of 17 articles were not accessible; the outcomes of 148 publications were not relevant; 76 trials did not conduct a weight-loss program; 1 study represented the data after a follow-up period; 4 studies did not report total scores for depression, SF36-PCS; SF36-MCS or HRQOL disease-specific tools and 55 articles were removed for other reasons). Sixteen arms from fifteen articles were included in the present systematic review. Of those, one study did not provide sufficient data based on each arm of the intervention [37]. Thus, this trial was not included in the quantitative synthesis for depression. Therefore, fifteen arms from fourteen studies were included in the meta-analysis (Fig. 1).

Fig. 1
figure 1

Flow diagram of study selection process

Full size image

Characteristics of the included studies

Sixteen arms from fifteen articles were included in the systematic review which were published from 1995 to 2016. All trials assessing the effects of exercise as an adjuvant to energy-restricted diets on HRQOL or depression were parallel in design. However, some cross-over studies assessed the effects of adding exercise to energy-restricted diets on the other outcomes that were not included in the current study. The majority of included studies were from the United States of America (10 publications [27, 28, 31,32,33,34, 36, 37, 46, 53]), others were from Canada [26], Netherlands [35], Australia [54], and Iran [29]. The duration of the studies ranged from 8 to 72 weeks.

All studies were conducted on overweight/obese individuals. Two studies [33, 36] were conducted in females, two studies were completed on males [27, 29], and others included participants from both sexes [26, 28, 30,31,32, 34, 35, 46, 53, 54]. The calorie restriction of diets in all studies varied from 200 to 1000 kcal/day, and The exercise programs of included studies were aerobic [26, 29, 30, 32, 33, 35], resistance [53], or both [28, 31, 34, 36, 37, 46, 54].

Two independent couples of arms from the Lalonde et al. study [26] are included in the current study. Moreover, the two Messier et al.’s studies [28, 53] are independent since they are conducted at different times and on different samples with different aims. The characteristics of included studies are provided in Table 1.

Table 1 Characteristics of included studies
Full size table

Risk of bias and quality of evidence

A total of 14 trials were included in the quality assessment. Of these, two RCTs [31, 34] RCTs [31, 34] were categorized as low risk of bias, eight studies [26,27,28, 30, 33, 35, 36, 54] were classified as unclear risk of bias; and four studies [29, 32, 46, 53] were classified as high risk. Approximately 80% of trials [26,27,28, 30,31,32,33,34, 36, 46, 54] were categorized as low risk of bias for random sequence generation, and three studies [31, 34, 54] reported the approach to conceal their randomization (allocation concealment). As blinding of the participants and personnel is impossible in lifestyle modification trials, the domain of blinding of participants and personnel was not taken into account in the overall score of quality assessment. Seven trials [27, 28, 30, 31, 33, 34, 46] were low risk of bias for blinding of outcome assessor, and eleven trials [26,27,28, 30,31,32,33,34, 36, 53, 54] were low risk of bias for incomplete outcome data. There was no evidence of selective outcome reporting in the included trials, and eight studies were categorized as low risk of bias for dietary compliance [26,27,28, 30,31,32,33,34]. The details of the risk of bias assessments are shown in Supplementary Table 2.

The GRADE approach was used to assess the quality of evidence. The quality of evidence was low for the effect of adding exercise to the energy-restricted diet on disease-specific HRQOL, and depression status; and high for the effect of adding exercise to the energy-restricted diets on wither SF36-PCS or SF36- MCS (Supplementary Table 3).

Meta-analyses

Depression

Meta-analysis of six trials (including 462 participants) [27, 33,34,35,36, 46] found that adding exercise to the energy-restricted diet interventions had no significant effect on depression scores neither by using the random effect model (hedges’g = − 0.04, 95% CI: − 0.28, 0.20, P = 0.73) (Fig. 2) nor by using the IVhet model (hedges’g = − 0.04, 95% CI: − 0.30, 0.21, P = 0.73). Moreover, the between-study heterogeneity was moderate [40] (Q statistic = 6.92, P for heterogeneity = 0.22, I2 = 27.8%) for both models. The results remained non-significant even after conducting several subgroup analyses (Supplementary Table 4).

Fig. 2
figure 2

Forest plot describing the effect of exercise plus a weight-loss diet compared with a weight-loss diet alone on depression using a random-effects model

Full size image

Physical component summary (PCS) scores of SF36 questionnaire

Meta-analysis of eight arms from seven trials (including 467 participants) [26,27,28,29,30,31,32] found that adding exercise to the energy-restricted diet interventions had no significant effect on SF36-PCS scores with no evidence of between-study heterogeneity neither by using the random effect model (weighted mean difference (WMD) = 1.51, 95% CI: − 0.16, 3.18, P = 0.08; I2 = 0.0%, P for heterogeneity = 0.78, Fig. 3) nor by using the IVhet model (WMD = 1.51, 95% CI: − 0.16, 3.18, P = 0.08; I2 = 0.0%, P for heterogeneity = 0.78). This result was marginally significant (P < 0.1) but not clinically important. The results remained non-significant even after performing several subgroup analyses (Supplementary Table 5).

Fig. 3
figure 3

Forest plot representing the effect of exercise plus a weight-loss diet compared with a weight-loss diet alone on SF36- PCS using a random-effects model

Full size image

Mental component summary (MCS) scores of SF36 questionnaire

The results of meta-analysis of seven arms from six trials (428 participants) [26,27,28,29,30,31] revealed that adding exercise to the energy-restricted diet interventions had no significant effect on SF36-MCS scores with no evidence of between-study heterogeneity neither by using the random effect model (WMD = 0.64, 95% CI: − 1.00, 2.28, P= 0.44; I2= 0.0 %, P for heterogeneity = 0.62, Figure 4) nor by using the IVhet model (WMD = 0.64, 95% CI: − 1.00, 2.28, P= 0.44; I2= 0.0 %, P for heterogeneity = 0.62). Several subgroup analyses were conducted and no significant effects were found (Supplementary Table 5).

Fig. 4
figure 4

Forest plot depicting the effect of exercise plus a weight-loss diet compared with a weight-loss diet alone on SF36- MCS using a random-effects model

Full size image

Disease-specific health-related quality of life questionnaire scores

Meta-analysis of 5 trials (266 participants) [30, 32, 34, 53, 54] found that adding exercise to the energy-restricted diet interventions had no significant effect on disease-specific questionnaire scores with no between-study heterogeneity (hedges’g = 0.16, 95% CI: − 0.09, 0.40, P = 0.20; I2 = 0.0%, P for heterogeneity = 0.91), using neither the random effect model (Fig. 5) nor the IVhet model.

Fig. 5
figure 5

Forest plot depicting the effect of exercise plus a weight-loss diet compared with a weight-loss diet alone on disease-specific questionnaire scores using a random-effects model

Full size image

Due to the lack of evidence of between-study heterogeneity and the limited number of included studies, we did not perform subgroup analysis for this outcome.

Meta‐regression

On meta-regressions, no significant associations were found between the effects of energy-restricted diet plus exercise compared with energy-restricted diets alone on depression and HRQOL and following study-level covariates: type of questionnaire, type of dietary intervention, type of exercise program, intervention duration, outcome, intervention under supervision, risk of bias assessment result, sex of participants, baseline physical activity status, additional weight management program (supplementary table 6).

Sensitivity analysis and Publication bias

With regards to sensitivity analyses, the overall effects of exercise on depression and HRQOL did not change by removing individual studies from the analyses, and visual inspection of the funnel plots revealed no evidence of publication bias (supplementary Fig. 1 [A-C]).

Discussion

The current systematic review and meta-analysis of RCTs examining the effects of energy-restricted diet plus exercise compared with energy-restricted diets alone did not significantly affect the quality of life and depression. These findings remained non-significant in several pre-defined subgroup analyses. Some studies have reported the beneficial effects of adding exercise to an energy-restricted diet [26, 28,29,30, 33, 34]. However, a recent systematic review [25] indicated that although four out of nine included studies supported the role of exercise in the improvement of HRQOL when added to energy-restricted diets, a definitive conclusion was not feasible because of the limited number of included studies. In addition, five more trials [26, 27, 29, 32, 54] which were discounted in Van den Hoek et al. systematic review [25], were included in the current study. It seems that two [26, 54] out of them did not match the inclusion criteria (not proper mean BMI of their participants) of their review [25].

In the present meta-analysis, the random effect and IVhet models were used to account for between-study heterogeneities and floor effects. However, it is essential to acknowledge that there were limitations with the design of the individual studies which might have biased the results in the direction of the non-significant effects on depression and HRQOL of exercise as an adjuvant to diet. These results, found in the present review, were also found in most previous clinical trials [27, 34,35,36,37, 54].

These findings may be due to the following reasons: The first and the most important reason may be limiting the inclusion criteria to trials with an isocaloric weight-loss diet in their arms. This may have attenuated the between-group differences in weight loss which is a key mediator in lifestyle modification-mood status relationship [55]. The second reason is that the generic HRQOL tools (such as SF36) are not specific enough to explore the subtle changes [56] and are susceptible to ceiling and floor effects [57]. Given that a methodological challenge in studies comparing the effect of diet plus exercise and diet alone is the masking of exercise effects by the magnitude of weight-loss effects [35], the questionnaires which are applied in such studies should be sensitive enough to detect the slight differences. Thus, SF36 might not be an ideal questionnaire in this regard. To address these limitations, some studies have proposed the use of the IWQOL questionnaire as a disease-specific tool in addition to the SF36 in trials since the disease-specific tools do not have such limitations [57,58,59,60].

The summary effect of adding exercise to an energy-restricted diet on disease-specific HRQOL scores was insignificant in the present meta-analysis. Two [30, 34] out of five [30, 32, 34, 53, 54] included studies used the IWQOL as an obesity-specific tool. However, these two studies [30, 34] did not indicate any significant between-group differences in weight-loss (as a key linkage between lifestyle interventions and HRQOL [55]); as such, they failed to address the effects of exercise plus diet or diet-related weight loss on the HRQOL. Nevertheless, Napoli et al. [34] reported an improvement in HRQOL in the diet plus exercise group compared with the exercise group in the 6-month follow-up period during which all participants were required to maintain their weight loss.

The studies included in the present systematic review used different tools for assessing depression; among these, BDI may be a more practical tool for weight-loss interventions. BDI has two primary factors including cognitive/affective and somatic items. Somatic items (such as body image) might help to explore the effect of weight-loss interventions on depression [61]. However, Geliebter et al. [46] utilized the BDI questionnaire to assess depression status, but failed to show any significant effect. This finding may be related to relatively short duration (8 weeks) and low sample size. Although the depression tools used in these studies were different, we used bias-corrected standardized mean difference (Hedges’ g) to reduce the effect of the between-scale heterogeneity.

To the best of our knowledge, this is the first systematic review and meta-analysis investigating the effects of adding exercise to energy-restricted diets on quality of life and depression in overweight/obese adults. The strengths of the current meta-analysis are as follow: (1) it was conducted on clinical trials which are the gold standard methods for evaluating the effectiveness of interventions; (2) a comprehensive search method was applied to identify all potential studies; (3) the exercise protocol details (such as intensity and type of exercise) were considered in the analyses; (4) in the present study, trials with isocaloric weight-loss diets as part of their research arms were included to prevent the confounding effects of the amount of calorie restriction of the diet; (5) the quality of the overall provided evidence was assessed for each outcome using the GRADE system (6) we have used two different approaches to utilize any information available to provide an estimate of the sought overall effect, and appropriate sensitivity analysis adds strength to the evidence to achieve a proper estimation.

The present meta-analysis also has some limitations which should be considered when interpreting its results: (1) most of our included studies had low or unclear risk of bias according to the Cochrane Collaboration tool, and high-quality studies are still needed; (2) most of the trials included in the present study have assessed HRQOL or depression as secondary outcomes; therefore, these studies did not have enough statistical power to address the effects of energy-restricted diets plus exercise in comparison to energy-restricted diets alone on HRQOL or depression; (3) the limited number of trials included in our review may cause a relatively low internal and therefore external validity; (4) given that this study was conducted on overweight or obese adults (age ≥ 18 years and BMI ≥ 25 kg/m2), the results of the current study might not be generalized to non-obese individuals and other age groups; (5) The outcome variables were assessed by questionnaire and their assessment are prone to a degree of measurement error; (6) As the majority of the included studies did not report the change values in the intervention and control groups, the SDs for change values were calculated by using a hypothetical correlation coefficient of 0.5. However, to check the sensitivity of the meta-analysis to selected correlation coefficients, we repeated the meta-analyses using the lower and upper limits of the confidence interval. It should be mentioned that the results were not sensitive to different values of “correlation coefficient”; (7) as blinding of participants and personnel was not possible for the interventions assessing the effect of diet or exercise, the observed effects in the included studies might have been affected.

In conclusion, our meta-analysis did not support the improving or reducing effect of adding exercise to an energy-restricted diet for improving depression and HRQOL compared to an energy-restricted diet alone in overweight or obese adults. However, the effect of adding exercise to energy-restricted diets on SF36-PCS was marginally significant. It should be noted that well-designed RCTs are still needed to investigate further the uncertainty of the benefits of exercise plus diet intervention on mood status. Until then, given the well-known benefits of exercise on different aspects of health, exercise should be recommended as an essential component of lifestyle interventions for obesity management.