Nutrition and Exercise Interventions to Improve Body Composition for Persons with Overweight or Obesity Near Retirement Age: A Systematic Review and Network Meta-Analysis of Randomized Controlled Trials

The retirement phase is an opportunity to integrate healthy (nutrition/exercise) habits into daily life. We conducted this systematic review to assess which nutrition and exercise interventions most effectively improve body composition (fat/muscle mass), body mass index (BMI), and waist circumference (WC) in persons with obesity/overweight near retirement age (ages 55–70 y). We conducted a systematic review and network meta-analysis (NMA) of randomized controlled trials, searching 4 databases from their inception up to July 12, 2022. The NMA was based on a random effects model, pooled mean differences, standardized mean differences, their 95% confidence intervals, and correlations with multi-arm studies. Subgroup and sensitivity analyses were also conducted. Ninety-two studies were included, 66 of which with 4957 participants could be used for the NMA. Identified interventions were clustered into 12 groups: no intervention, energy restriction (i.e., 500–1000 kcal), energy restriction plus high-protein intake (1.1–1.7 g/kg/body weight), intermittent fasting, mixed exercise (aerobic and resistance), resistance training, aerobic training, high protein plus resistance training, energy restriction plus high protein plus exercise, energy restriction plus resistance training, energy restriction plus aerobic training, and energy restriction plus mixed exercise. Intervention durations ranged from 8 wk to 6 mo. Body fat was reduced with energy restriction plus any exercise or plus high-protein intake. Energy restriction alone was less effective and tended to decrease muscle mass. Muscle mass was only significantly increased with mixed exercise. All other interventions including exercise effectively preserved muscle mass. A BMI and/or WC decrease was achieved with all interventions except aerobic training/resistance training alone or resistance training plus high protein. Overall, the most effective strategy for nearly all outcomes was combining energy restriction with resistance training or mixed exercise and high protein. Health care professionals involved in the management of persons with obesity need to be aware that an energy-restricted diet alone may contribute to sarcopenic obesity in persons near retirement age. This network meta-analysis is registered at https://www.crd.york.ac.uk/prospero/ as CRD42021276465.


Introduction
Overweight and obesity are serious disorders with prevalence rates among older Europeans of about 60% and 20%, respectively [1]. In the United States, the obesity prevalence is even higher at around 40% [2]. These rates have steadily increased worldwide over the last 40 y in men and women [2][3][4][5]. Its high prevalence and serious social, economic, and health consequences make it one of the major global health problems [6][7][8].
Obesity is a major risk factor for several diseases, including type 2 diabetes mellitus, coronary artery disease, cerebral vascular disease, arterial hypertension, dyslipidemia, and several types of cancer. All of these conditions contribute to a reduction in both the quality of life and life expectancy [7,8]. For example, an increase in a society's BMI by 2 points shortens the life expectancy by 0.7 to 1 y [9]. Furthermore, obesity is accompanied by burdens such as falls, disability, or care dependency, especially in older adults [7,8,10].
Obesity is characterized by excessive fat accumulation [11] that often occurs during the process of aging. This especially occurs in persons aged 45 to 70 y, with a weight peak observed at middle age, i.e., 50 to 65 y of age [8,[12][13][14]. Aging is accompanied not only by a gradual increase in body fat (BF) stores but also a decrease in muscle mass, muscle function, and water retention. Simultaneously suffering from obesity and the progression of the aging process can lead to sarcopenic obesity, a condition that combines the loss of muscle mass, strength, and function with an increase in adiposity [15,16]. This affects a remarkably large group of people, with prevalence rates of up to 33.5% observed, e.g., in the older US population [17], which suggests that many obese people simulateously suffer from sarcopenia. Because loss of muscle mass is often accompanied by an increase in fat mass, body weight may remain stable [18,19], meaning that a stable or even decreasing body weight can mask increasing adiposity [19].
The retirement age is generally between 48 and 67 y in the Organization for Economic Co-operation and Development countries [20]; this is exactly in the age range during which the previously described major changes in body composition occur. Therefore, the retirement phase represents a window of opportunity to decelerate the associated deterioration in body composition. This phase is a period of change. In most cases, this change is not gradual, but occurs abruptly one day when a person no longer needs to go to work and needs to start redesigning their everyday life [12]. A recent longitudinal study showed that 61% of people in this age group changed their lifestyle during the retirement phase [21]. People that changed their lifestyle by reducing risk factors for obesity, such as poor diet or inactivity, showed smaller physical declines over time in later life. This finding underlines the great potential for implementing healthy nutrition and exercise habits and thus increasing the disability-free life expectancy in the retirement phase [21].
Nutrition and exercise interventions are considered as firstline therapies for treating individuals with overweight and obesity [4,11,22]. These should not only be effective in reduction of BF but also in preserving muscle mass. This is even more important in older adults to prevent the occurence of disability [11]. Several systematic reviews have summarized different nutrition and exercise interventions in older persons with overweight and obesity [23][24][25][26][27][28][29][30], but we identified only one that focused on people near retirement age [23]. This review, however, examined the effectiveness of dietary interventions on healthy eating habits and not on obesity parameters such as body composition or anthropometric parameters.
Thus, prior to the current review, a comprehensive systematic review and network meta-analysis (NMA) of the effects of nutrition and exercise interventions in persons near retirement age was lacking. Such an analysis is highly beneficial for the scientific community and clinical practice, because the results can provide recommendations for effective interventions for people who are overweight or obese in this target group. The specific aim of conducting this systematic review using NMA methodology was to assess which nutrition and exercise interventions are most effective for improving the body composition (fat mass and muscle mass), BMI, and waist circumference (WC) in persons with overweight or obesity near retirement age (55 to 70 y of age).
respective age group (middle-aged and aged) and made small adaptations for each database searched (see Supplementary Table 1).

Study selection
The study selection process was conducted with the systematic review software COVIDENCE (Veritas Health Innovation). We included RCTs using a parallel or crossover design, based on our predefined Population, Intervention, Control, Outcome (PICO) question (Table 1). We excluded studies focusing on persons with specific health conditions, such as cancer, type 2 diabetes, heart failure, or pulmonary diseases, as well as studies with specific target groups, such as soccer players or truck drivers. We further excluded RCTs with pharmaceutical or behavioral interventions other than nutrition or activity interventions for obesity and studies that lacked a clear description of the intervention and weight maintenance studies. Title and abstract screening as well as full-text screening were performed based on inclusion and exclusion criteria by 2 authors independently of one another (MT, SB, DE). The numbers and reasons for the exclusion of studies are listed in the flow chart (see Figure 1). Any disagreements were resolved by a discussion involving a third person (DE).

Data extraction and quality assessment
Two reviewers independently extracted data from the final included full-text articles, and disagreements were resolved by a discussion involving a third person. We generated a standardized data extraction template, including the study characteristics, patient characteristics, intervention(s), adherence to the intervention(s), and patient outcomes. Prior to the data extraction, we piloted the template of 2 studies to identify and edit possible shortcomings in the template. The methodological quality of the RCTs was assessed by 2 independent reviewers using the Cochrane Risk of Bias tool, referring to the Cochrane Handbook for Systematic Reviews of Interventions [32].

Data synthesis: interventions
Based on the different interventions used in the identified studies, we created 12 pragmatic intervention/control categories: 1) no intervention, 2) energy restriction (i.e., caloric restriction of 500 to 1000 kcal), 3) energy restriction plus high protein intake (1.1-1.7 g/kg body weight/d), 4) intermittent fasting (5:2 diet), 5) mixed exercise (aerobic and resistance training), 6) resistance training, 7) aerobic training, 8) high-protein intake plus resistance training, 9) energy restriction plus high-protein intake plus exercise, 10) energy restriction plus resistance training, 11) energy restriction plus aerobic training, and 12) energy restriction plus mixed exercises (aerobic and resistance training combined). Studies that could not be assigned to any of these categories and studies that included study arms comparing similar interventions that would have been in the same category were described narratively.

Statistical analysis
The change in the outcome measures was used for the statistical analysis. This change was calculated as the mean difference (MD) between the baseline and follow-up for each treatment group. Most included studies provided means and standard deviations at baseline and for specific follow-up dates. We then calculated the MD and the standard deviation of change referring to the Cochrane Handbook for Systematic Reviews of Interventions and assuming a correlation of 0.85 [33]. This value was chosen because we observed a high correlation between the baseline and follow-up measures in studies where baseline, follow-up, and change measures were reported. If not directly provided, further steps were taken to calculate the corresponding values. This included the use of P values and confidence intervals; where ranges were reported, we applied the same methodology as Hozo et al. [34]. The NMAwas based on a random effects model, and correlations in multi-arm studies were considered [35]. The common heterogeneity variance τ 2 used in the random effects model was estimated with a generalized DerSimonian-Laird estimator [36]. To assess inconsistency, the between-designs Q-value was calculated based on a full design-by-treatment interaction model for random effects [37]. If studies had more than one arm belonging to the same combined treatment group, we combined the corresponding means, standard deviations, and sample sizes [38]. We used Egger's test to appraise the data for potential publication bias, i.e., to identify asymmetry in the funnel plot [39]. For models where we compared effects within the same outcome measure, we used the MD. In models where we compared more than one outcome measure, we used the standardized mean difference (SMD) to assure comparability. For one model, we prioritized data, choosing the first available outcome measures in the following order: BF in %, BF in kg, muscle mass (LBM/ FFM), WC, then BMI. Here, we considered the different direction of positive effects for the fat mass and muscle mass.
Additionally, we performed several subgroup and sensitivity analyses. First, we analyzed data stratified by sex, from studies with only females, only males, or reporting on both sexes separately. Second, we distinguished between the duration of the intervention(s) being less or equal to 14 wk or being more than 14 wk. Third, we repeated analyses excluding all studies identified as having a high risk of bias. Fourth, we grouped the interventions even further into the following 3 categories: nutrition, exercise, or nutrition plus exercise. In all subgroup analyses, we used one fat mass as an outcome measure (body fat or BF in %) and muscle mass as an outcome measure (LBM/FFM), deciding on a case-by-case basis and using the ones that were reported more frequently in order to include the highest number of studies in the analysis. A 2-sided P value of less than 0.05 was considered to be statistically significant. All analyses were performed in R (Version 4.1.3).

Results
We identified and screened 11,073 records, from which 92 RCTs met the selection criteria ( Figure 1). Of these, 6 RCTs included interventions that could not be assigned to one of the 12 intervention/control categories (i.e., different types of oil supplementation, vegan diet, and Mediterranean diet), 2 RCTs included less than 8 participants, and 18 RCTs compared interventions within the same intervention category (e.g., comparison of energy restriction with target 500 kcal and 1000 kcal). For these reasons, we included 66 studies  that involved nutrition and exercise interventions in persons with overweight/obesity near retirement age and included a total of 4957 participants in the final NMA. Figure 1 illustrates the literature review process.

Characteristics of the included RCTs
Of all 92 identified studies, we identified 21 RCTs that focused on nutrition interventions alone, 25 RCTs that focused on exercise interventions alone, and 46 RCTs that combined nutrition and exercise interventions. Overall, we identified 82 different interventions described in the studies, which we assigned to 12 categories created based on the interventions used in the primary studies.
Most studies included both male and female participants (n ¼ 51), followed by studies that only included female participants (n ¼ 36). The included studies were conducted in 19 different countries, and mostly in the United States (n ¼ 49), see Table 2. The sample sizes ranged from 11 to 543 participants. The intervention period ranged from 8 wk to 6 mo, with the most common intervention periods being 12 wk (n ¼ 33 or 36%), 26 wk (n ¼ 21 or 23%), and 16 wk (n ¼ 11 or 12%). Most of the exercise studies (n ¼ 19 or 76%) performed training sessions 3 times a week for 20 to 60 min. The most common nutrition intervention was energy restriction (with or without meal replacement, mostly with a reduction in fat content and/or carbohydrates, including low-and very low-calorie diets).

Risk of bias
Overall, 19 RCTs (21 %) were rated as having a high risk of bias for at least one domain. Thirty-four studies (37%) were judged as having a low risk of bias in the domain random sequence generation, whereas 32 studies (35%) had a low risk of bias in the domain allocation concealment. The assessment of participant and personnel blinding revealed that 18 studies (20%) had a low risk of bias, and 31 studies (34%) had a low risk of bias for blinding the outcome assessor. Seventy-two (78%) of the studies had a low risk for presenting incomplete outcome data, and 40 (43%) had a low risk for selective reporting. The Egger's test results for publication bias were not significant for all outcomes (P < 0.05). A summary of risk of bias assessment is provided in Figure 2.

NMA
We conducted the NMA model separately for the 5 outcomes (%BF, BF in kg, LBM/FFM, WC, BMI) and one model using all         outcomes. The network geometry for the outcomes BF mass in kg and LBM/FFM, as well as the network geometry using all studies according to the predefined prioritization of outcomes is presented in Figure 3. All other network geometries can be found in Supplementary Figure 1. By grouping interventions into specific treatment groups, we obtained a dense network that enabled us to make many direct comparisons. The network where we used the prioritization shows the highest number of direct comparisons. Figure 4 illustrates the results of the NMA for all outcomes separately as well as for the single model using all studies with a prioritization of outcomes.
alone. No significant effect in terms of reducing BF was observed for the 5:2 diet and resistance training combined with a highprotein diet. Aerobic training was only significant in terms of losing BF in % (P ¼ 0.02, MD: À1.41, 95% CI: À2.58, À0.25).

Inconsistency
We observed no signs of inconsistency in the networks when comparing changes in BF in kg (P ¼ 0.859), BF in % (P ¼ 0.986), LBM/FFM (P ¼ 0.232), and prioritization (P ¼ 0.461). However, we observed inconsistencies in the network when comparing changes in BMI (P < 0.001) and WC (P < 0.001). In the network comparing the change in BMI, it was necessary to remove 3 studies to obtain a P value above the level of significance [66,95,106]. In the network comparing the change in WC, we found 3 studies [52,66,101] that contributed to inconsistency; when removing all 3 of them, the network no longer showed signs of inconsistency (P ¼ 0.219), but the same overall results were obtained as for the main analysis.

Subgroup and sensitivity analyses
No considerable differences were observed when only analyzing studies with women or men. An intervention duration of more than 14 wk was associated with more pronounced weight loss, but the order of effectiveness did not change between categories of interventions. Results did not change substantially when excluding studies with a high risk of bias. The subgroup analysis results supported the hypothesis that interventions combining nutrition and exercise most effectively improve body composition and anthropometric parameters.

Summary of studies not included in the network meta-analysis
In our literature review, we identified 26 studies  that could not be included in the network meta-analysis. Eleven of these studies were with mixed nutrition and exercise interventions, 3 were with only exercise interventions, and 12 were with only nutritional interventions. All 26 studies were highly heterogeneous in terms of the interventions used; therefore, they are difficult to compare. A detailed overview about the used interventions as well as the main results of the single studies can be found in Supplementary Table 2.

Discussion
The aim of conducting this systematic review and NMA was to evaluate which nutrition and exercise interventions most effectively improve body composition (fat mass and muscle mass) and anthropometric measures (BMI and WC) in persons with overweight or obesity near or around retirement age. In the NMA models, we identified several effective nutrition and exercise interventions for this target group. A reduction in BF could be best achieved by applying the measures of energy restriction combined with any kind of exercise or with high-protein intake. Energy restriction alone also reduced BF, but to a lesser extent, and on the contrary, energy restriction alone tended to decrease muscle mass. Muscle mass could only be significantly increased with mixed exercise (resistance and aerobic) interventions, but all other interventions that included exercise effectively preserved muscle mass. A decrease in BMI and/or WC could be achieved with nearly every intervention except aerobic training alone, resistance training alone, or resistance training combined with a high-protein diet. Overall, the most effective strategy for loss of fat mass while maintaining or increasing muscle mass was the combination of energy restriction and exercise (resistance or mixed) and/or a high-protein diet.
Energy restriction to achieve a negative energy balance is still a key therapeutic weight loss strategy recommended in evidence-based guidelines [22,133]. However, according to our results, energy restriction alone does not seem to be an appropriate approach for persons in retirement age. Although it results in weight and fat loss, it tends to result in a loss of muscle mass, which may be considered as an adverse effect as it increases risk of disability, metabolic impairments, mortality, or a low quality of life in this group [134]. In addition, LBM loss is a major factor in weight regain, as LBM is the main driver of energy expenditure [135].
Because aging is associated with muscle loss, great efforts should be made to preserve or even increase muscle mass as well as muscle function and quality in aging or aged persons. Our results show that muscle preservation can be effectively achieved by combining resistance training, mixed exercise, or mixed exercise with high-protein content foods with an energy-restricted diet. This result is similar to those of other systematic reviews that concluded that resistance training added to energyrestricted diets prevents muscle loss in other target groups, e.g., in generally older individuals [25,136]. According to our results, resistance training or the combination of resistance and aerobic training yielded much better results than aerobic training alone regarding muscle mass preservation. This result also agrees with those of other studies. A recent systematic review in older adults, for example, concluded that only resistance training could effectively improve muscle strength, whereas aerobic training could not [137].
The term energy-restricted diet is referred to in several different approaches, including low-carbohydrate or low-fat diets, low-and very low-calorie diets, diets using formula products, time-restricted eating, and many more [138]. Our results do not allow us to determine which approach is best because it was not possible to examine all different approaches to reach energy restriction due to the limited number of similar studies that included people near retirement age. However, the specific approach might not be important as long as the energy intake is below the energy requirements (currently 500 to 1000 kcal). For example, a systematic review comparing intermittent and continous energy restrictions in adults did not find different effects, but both forms of energy restriction resulted in similar amounts of weight loss [139]. Whether weight loss is achieved with a moderate, low-or very low-calorie diet also seems to be of secondary importance, but guidelines address concerns that a lowor very low-calorie diet may be less likely to be nutritionally complete [133], also because very low-calorie intake may induce more intense LBM loss [140]. Another systematic review found that low-fat diets are not more successful than higher-fat, low-carbohydrate diets with regard to long-term weight change in adults [141]. The protein amount, source, and quality are also important components of an energy-restricted diet for maintaining muscle [142]. The amino acid composition (and notably the essential amino acid content) and the timing of protein intake (and especially regarding exercise training) [143] but also of other nutrients should be considered, such as vitamin D [144], as well as the fat quality, either alone or combined with other interventions [145].
This evidence implies that several potentially effective ways to restrict energy intake exist. It is likely to be much more important that persons with overweight or obesity find an approach that suits their lifestyle [133]. The time near retirement is a period that offers a unique opportunity for changing lifestyle habits; this time should be used to change diet and behavior in ways that are comfortable over longer periods of time and ensure a healthy diet over the long term to maintain weight loss. Women and men may also react differently to nutritional and exercise interventions [146], but this was not supported by our NMA results because the subgroup analyses separating men and women did not reveal different results from those of the full analysis.
Risk of bias assessment results indicate that risk of bias of the included studies was mostly unclear, i.e., the reporting was very poor in most of the included studies, and risk of reporting bias was mostly unclear or high. Furthermore, only 48% of all identified studies investigating nutritional interventions could be included in the NMA. This was due to the heterogeneity of the nutritional intervention studies (i.e., highly heterogeneous interventions) and the (frequent) lack of a control group with no intervention or usual care. These studies compared similar interventions but yielded few results in terms of observed differences between the study arms. The quality of the studies including the exercise interventions was much higher. Of these, 88% of the exercise studies and 83% of the studies including nutritional interventions combined with exercise interventions could be included in the NMA, demonstrating a much higher homogeneity. This may also be due to the fact that there are many more different food and diet options available to improve body composition than exercise options.
This systematic review and meta-analysis included a considerably high number of studies with the predefined target group and was conducted systematically based on the recommendations in the Cochrane Collaboration Handbook. This NMA allowed an indirect comparison to be included in the statistical model, enhancing the significance and coverage of the whole model [147][148][149]. However, this study also had some limitations. The method used to group treatments actually treated different treatments as the same, which might have biased our results.
The included studies of this review are heterogeneous, varying in terms of the study duration, the participants' sex, age and weight, or the country in which the studies were conducted. We found that some of our networks showed evidence of inconsistency. We were able to partly explain the cause of this inconsistency. Although we could not reject the null hypothesis of no inconsistency, this does not imply that the network is consistent. Nevertheless, this study contributes important evidence that should be considered when developing recommendations for improving body composition aimed at persons with overweight and obesity near retirement age.

Conclusion
The overall results of this NMA indicate that the most effective strategy to improve body composition, i.e., losing fat without increasing risk of sarcopenia in persons with obesity around retirement age, was combining energy restriction with resistance training or with mixed exercise (resistance combined with aerobic exercise) and/or high-protein intake. Without training, an energy-restricted diet with or without added protein helped individuals lose fat mass but also tended to result in losses of muscle mass. To lose fat while preserving muscle, interventions involving aerobic training, intermittent fasting, resistance training combined with a high-protein diet, and energy restriction alone or in combination with a high-protein diet were not suitable, because they either tend to decrease muscle mass or do not reduce BF.
The important life period near retirement provides individuals with an opportunity to start establishing new healthy nutrition and exercise habits and to incorporate evidence-based nutrition and exercise interventions into their daily routines. The main aim is to prevent dependency and disability in older age. Healthcare professionals involved in the management of persons with obesity must be aware that an energy-restricted diet alone probably contributes to the development of sarcopenic obesity in persons of retirement age. To simultaneously lose weight and maintain muscle mass, the combination of energy restriction and resistance training is probably the best way forward.

Acknowledgments
The authors' responsibilities were as follows -DE, SB: