Outcomes addressed in randomized controlled lifestyle intervention trials in community‐dwelling older people with (sarcopenic) obesity—An evidence map

Obesity and sarcopenic obesity (SO) are characterized by excess body fat with or without low muscle mass affecting bio‐psycho‐social health, functioning, and subsequently quality of life in older adults. We mapped outcomes addressed in randomized controlled trials (RCTs) on lifestyle interventions in community‐dwelling older people with (sarcopenic) obesity. Systematic searches in Medline, Embase, Cochrane Central, CINAHL, PsycInfo, Web of Science were conducted. Two reviewers independently performed screening and extracted data on outcomes, outcome domains, assessment methods, units, and measurement time. A bubble chart and heat maps were generated to visually display results. Fifty‐four RCTs (7 in SO) reporting 464 outcomes in the outcome domains: physical function (n = 42), body composition/anthropometry (n = 120), biomarkers (n = 190), physiological (n = 30), psychological (n = 47), quality of life (n = 14), pain (n = 4), sleep (n = 2), medications (n = 3), and risk of adverse health events (n = 5) were included. Heterogeneity in terms of outcome definition, assessment methods, measurement units, and measurement times was found. Psychological and quality of life domains were investigated in a minority of studies. There is almost no information beyond 52 weeks. This evidence map is the first step of a harmonization process to improve comparability of RCTs in older people with (sarcopenic) obesity and facilitate the derivation of evidence‐based clinical decisions.


| INTRODUCTION
Obesity and sarcopenic obesity (SO) in older adults are major public health issues, due to their increasing prevalence. 1,2 Obesity prevalence among older adults (60 years and older) ranges from 20.9% in Europe to 43.3% in the United States. [3][4][5] The prevalence of SO is difficult to establish due to the different definitions and cutoffs applied. 6 Nevertheless, a recent meta-analysis from Gao et al. (2021) estimates that the global prevalence of SO for older adults (60 years and older) is estimated to be 11%. 7 Both disorders are associated with negative health consequences, such as premature mortality, increased risk of falls, poor physical functioning, comorbidity burden, and reduced quality of life, increasing the risk of loss of independence and institutionalization. [8][9][10] The recommended first-line therapies for (sarcopenic) obesity in older adults are lifestyle interventions aiming at the loss of body weight and fat mass and consist of diet modifications, increasing physical activity or specific exercise training, and behavioural therapy. 11,12 Reviews and guidelines on obesity treatment in older adults have been published suggesting that lifestyle interventions in older adults are effective in reducing body weight and favoring combined interventions including dietary and exercise components. 9,11,[13][14][15] The optimal content and dose of interventions, are not well established, one reason being the lack of systematic reviews with meta-analyses enabling to identify optimal treatment strategies. 11,16 Clinical decisions about treatment should be based on outcomes of high-quality randomized controlled trials. 17 Therefore, the selection of outcomes is important relative to adequate power, the fit with the target population, and the avoidance of null findings. 17 The lack of pooled analyses might be due to the heterogeneity in assessed outcome domains, outcomes, units of measurement, and the time of outcome measurement. 18 For better comparability of studies and purposes of data pooling, a harmonization of at least a core outcome set (COS) of important measures should exist. 17,19 A COS helps avoid ineffective interventions and outcome-reporting bias by providing a list of the minimum outcomes to be measured in RCTs. 17 The development of a COS requires as a first step a comprehensive review of the existing literature and the extraction of the outcomes assessed as well as outcome-related methodology used in available RCTs. 17 For this purpose, an evidence map is considered appropriate, listing the evidence, identifying gaps, and providing results in a userfriendly format. 20 We created such an evidence map to provide an overview of outcomes reported in RCTs on lifestyle interventions in community-dwelling older people with (sarcopenic) obesity. Specifically, we addressed the following questions: Which outcomes from which domains have been measured in lifestyle intervention RCTs in community-dwelling older adults with (sarcopenic) obesity? Which methods were used and at which time points were the outcomes assessed? Do the identified outcomes and methods to assess these outcomes depend on whether obesity is treated or SO is treated?

| METHODS
This evidence map has been developed in the frame of the Effective SLOPE project (EffectS of Lifestyle interventions in Older PEople with obesity: a systematic review and network meta-analysis; PROSPERO: CRD42019147286). 16 The reporting of this study complies with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) for Scoping Reviews (PRISMA-ScR) and the extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews (PRISMA-S) (Tables S1 and S2). 21 24 In addition, we used the Cochrane sensitivity-precision maximizing search filter for RCTs. For the search in CENTRAL, we did not consider entries from trial registries due to resource limitations. Search strategies were peer reviewed by an external information specialist (EM, University of Freiburg, Germany). Searches were fully re-ran with the last search date in May 2022. When re-running the searches, records known from earlier searches (i.e., duplicates within a database) were removed based on their database accession numbers.
We did not set any restrictions regarding language or publication time.
The detailed search strategies are shown in supporting information  25 Results from ClinicalTrials.gov were exported as CSV files and were screened according to eligibility criteria.

| Selection process
At least two reviewers (IGE, EK, DS, and GT) independently screened titles/abstracts and full texts according to prespecified eligibility criteria using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia; available at https://www.covidence. org). If information was lacking, we contacted the corresponding author/s two times over a span of 2 weeks.
For translation of articles published in English, Spanish, or German, we involved colleagues who were fluent/native in these respective languages; for translation of articles published in other languages, we used online translators (e.g., https://www.deepl.com/home).
Conference abstracts and theses not additionally published in a peerreviewed journal were excluded.

| Design of primary studies
We included RCTs (parallel and cluster).

| Population
Studies in community-dwelling adults with a minimum age of 60 years and a mean age of ≥65 years were considered. At least one obesity criterion for participants to be included in the RCTs had to be applied: total body fat mass ≥35% (women) and ≥25% (men), waist circumference of ≥88 cm (women) and ≥102 cm (men), and BMI using the standard adult cutoff of ≥30 kg/m 2 . 26 Cutoffs validated in specific populations (e.g., Asian [≥25 kg/m 2 ]) were also accepted. 27 Studies stating "obese" without providing a clear definition or criteria with references were excluded.
For SO, we used the definition provided in the primary RCTs, as long as one of the above-mentioned obesity criteria was met.
RCTs in mixed samples of people with overweight and obesity were excluded. However, authors were contacted and asked for data on the obesity subgroup. If additional data were provided, the respective RCT was considered for the current analysis.

| Interventions
Lifestyle interventions were considered if the intervention consisted of diet modifications (e.g., calorie restriction), exercise (aerobic exercise, resistance exercise, or both), or behavioural therapy, as well as their combinations with all types of delivery and doses. Due to the time needed to respond to treatment, the minimum intervention duration was set to 12 weeks.

| Comparators
As comparators, any lifestyle intervention and control groups (e.g., usual care and health counseling/education) were considered as a relevant comparator group.

| Outcomes
All reported health outcomes were deemed relevant. Related outcomes, such as environmental factors (e.g., walkability) and behaviour changes (e.g., level of physical activity and dietary intake) were not considered.
Articles presenting the data on genetic outcomes only were excluded.

| Data extraction
Two reviewers (DS and IGE) extracted the data of included references independently (using a pre-piloted data extraction table in Microsoft Excel 2016). Disagreements were solved by discussion or with the help of a third reviewer (EK). For each RCT, the study characteristics (first author, year of publication, country, obesity phenotype, obesity criterion for inclusion, sample size, study duration, and mean age) were extracted. Further, to map relevant information on the outcomes assessed in the RCTs, a classification scheme (Table 1) was created and used to extract (in addition to the outcome itself) data on outcome domain, type, (sub) category, method, units, and time of measurement.

| Data synthesis
Results on outcomes are presented for all included RCTs and separately for obesity and SO trials. To obtain an overview of the frequency of assessed outcomes for all included outcome domains, a bubble chart with four dimensions for each outcome domain was created. The x axis presents the four time-based categories of measurement (12 to 19 weeks, 20 to 26 weeks, 27 to 52 weeks, and more than 52 weeks).
The y axis represents the number of studies reporting at least one outcome in each outcome domain. The size of the bubbles represents the number of outcomes for each domain, and the color of the bubbles represents the obesity phenotype addressed in the RCTs (obesity or SO).
Outcomes were counted and heatmaps were created for each domain based on the classification scheme in Table 1. A heatmap visualizes data in a compact form by representing numbers with corresponding colors. 31  from the patients about their health status/condition. 28 • Observer-reported outcome (ORO): measurement based on a report of observable signs, events, or behaviors to a patient's health condition by someone other than the patient or a health professional. 28 • Clinician-reported outcome (CRO): measurement based on a report that comes from a trained healthcare professional after observation of a patient's health condition. 28 • Performance outcome (PerfO): measurement based on standardized tasks actively undertaken by a patient according to a set of instructions. 28 • Biomarkers (BM): a measurement that is considered as an indicator of normal biological processes, pathogenic processes, or biological responses to an exposure or intervention. May include molecular, histologic, radiographic, or physiologic characteristics. 29 Outcome category and subcategory Specification of outcome domains in categories and subcategories (for psychological outcomes and QoL) based on the measurement aim, e.g., body composition was categorized in fat, muscle, and bone; biomarkers were categorized in glucose metabolism, lipids, hormones, and so on.

Outcome
• Health-related patient assessment used as an endpoint and providing a rating score (categorical or continuous). 30 • Outcomes with different names or slight differences in administration but addressing the same concept and measurement aim were categorized as one outcome (e.g., gait speed measured in 400 m or 10 m). • Unique outcomes (outcomes in one RCT only) were reported separately.

Unit
Scoring and reported units of each outcome were listed.
Time of measurement Based on information on baseline, intermediate/interim, post-intervention, and follow-up outcome assessment assigning of time measurement to one of the four categories: 1. 12-19 weeks 2. 20-26 weeks 3. 27-52 weeks 4. More than 52 weeks When outcomes were measured more than once, all times were extracted.
Note: This table was the basis for creating the heatmaps. This information was extracted from each included randomized controlled trial (RCT). considered for the heatmaps. The outcomes in the heatmaps were sorted based on their category. The heatmaps' shade colors represent the frequency of outcome measurement. The shade colors were chosen according to traffic light colors, where green represents the outcomes most frequently reported, yellow represents a midpoint, and red the least reported outcomes.
All figures were created with the statistical software R Version 4.1.0.
The bubble chart was generated using the ggplot2 package (v3.3.3 Wickham, New-York, USA, 2016). 32 The heatmaps were created with the gt package (v0.3.0 Iannone, Boston, USA, 2021). 33 Separate heatmaps for each obesity phenotype were created for the domains physical function, body composition and anthropometry, and biomarkers. Due to a limited amount of data, only a combined heatmap, including both obesity and SO, was created for the domains quality of life, psychological, physiological, pain, sleep, medications, and risk for adverse health event and medical conditions.

| RESULTS
Of 57,721 unique records, 109 articles of 54 studies were included in the evidence map as they provided information on outcomes assessed ( Figure 1 and Table S4).  The unique accession numbers from all database searches are available upon request.     (Table 3).

| Physical function
In total, 42 different outcomes ( Few studies administered Patient-Reported Outcomes (PROs) that were categorized into (I)ADLs, lower extremity functioning, and osteoarthritisspecific physical function ( Figure S1 and Table S5). The three most frequently reported outcomes in obesity RCTs were gait speed (n = 16), VO 2 max/peak (n = 16), and the Short Physical Performance Battery (SPPB) (n = 11) ( Figure S2). The most frequently reported outcomes in SO RCTs were grip strength (n = 6), gait speed (n = 6), and chair rise (n = 4) ( Figure S3). None of the SO RCTs measured VO 2 max/peak, and only one trial measured endurance by the 2-min step test ( Figure S4).
Outcomes were most frequently reported between 12 and 26 weeks.
Applied methods and given units differed for 16 of the 42 outcomes. For instance, the SPPB was assessed as SPPB, modified SPPB, and expanded SPPB; and different score ranges were reported: 0-12, 1-12, and 0-4.

| Body composition and anthropometry
The included RCTs reported 120 outcomes related to body composition and anthropometry ( only once (Table S6). The domain was categorized into body mass, fat, muscle, bone, cardiovascular, and hepatic outcomes ( Figure S4 and Table S6). All reported outcomes were clinician-reported outcomes Several technologies (e.g., dual-energy X-ray absorptiometry, bioelectrical impedance analysis, magnetic resonance imaging, hydrostatic weighing, air displacement plethysmography, or computed tomography scan) were used to assess body composition outcomes.

| Biomarkers
The biomarkers domain had the highest number of different outcomes (n = 190) (Table 3), with 134 (70.5%) of them assessed only once (Table S7). The outcome categories were blood lipids, glucose metabolism, inflammation, hormones, vitamins, bone metabolism, kidney, liver metabolism, plasma proteins, proteins of skeletal muscle, and muscular health ( Figure S7 and Table S7). The vast majority of the biomarkers were measured in the blood (n = 143). Few were also measured in saliva

| Psychological
The psychological domain summarizes 47 outcomes, which were all assessed in obesity RCTs. Of these outcomes, 31 (66.0%) were unique outcomes ( Figure S11).

| Quality of life
The quality of life domain consists of 14 different outcomes (Table 3)

| Pain
Pain was measured exclusively in obesity trials using questionnaires (PROs) in four trials with three (75%) of the outcomes being unique.
The Patient-Reported Outcomes Measurement Information Systemshort form (pain subscale) (n = 2) reported between 12 and 19 weeks was used twice ( Figure S13).

| Sleep
The domain sleep comprises only two outcomes which were measured between 12 and 26 weeks of intervention by single obesity trials in the form of questionnaires (PRO) ( Table S8).

| Medications
The medications domain was self-reported and was assessed in three obesity RCTs ( Figure S14). The number of medications taken by the participants and the change in the number of medications following Note: The domains and the number of outcomes reported by the included RCTs are shown. Some outcomes were reported in both phenotype groups.
lifestyle interventions were reported, three times between 12 and 19 weeks and once between 20 and 26 weeks of intervention.

| Risk for adverse health events and medical conditions
Overall six RCTs (12.2%) (four obesity and two SO RCTs) reported metabolic syndrome risk (n = 3), sarcopenia (n = 2), major mobility disability, frailty, and falls (all n = 1) ( Figure S15) The most frequently reported outcomes overall are listed in Table 4.

| DISCUSSION
To the best of our knowledge, this is the first evidence map providing an overview of the outcomes and related methodology reported in lifestyle intervention RCTs in community-dwelling older adults with obesity and SO.
We  emphasizes the necessity of including both, muscle mass and muscle strength. 100 Of the seven in this evidence map included SO studies, only one fulfilled these consensus criteria. 57 Outcomes in the domains quality of life, psychological, pain, sleep, and medications were rarely reported and only in RCTs of the obesity phenotype. They constitute mostly PROs and are recommended for use in RCTs. 103,104 As these domains are directly linked to obesity and aging, this is an important research gap that needs to be addressed in future studies. [105][106][107] Obesity could affect the patients' mental health leading, among others, to depression, stress, and low self-esteem. 107,108 Therefore, it is not surprising that obesity is associated with lower quality of life. 107,108 The body weight stigma not only affects the patient's mental health but is also impacting their social participation. 107 Social participation was not an outcome in any included RCT. Considering that advanced age is also linked to loneliness, older people with obesity are at particular risk. 109 Pain and sleep are also determinants of quality of life, and both are associated with obesity and aging. In addition, pain is a barrier to performing everyday activities and losing body weight. 107,110 The number of medications taken is high in older people with obesity. 111 In diabetic adults with overweight and obesity, it has been shown that body weight loss is associated with a reduction in antidiabetic and antihypertensive drugs. 112 Heterogeneity was not only introduced by using diverse outcomes but also by differences in applied methodology within outcomes. In addi-