Behavioral weight loss interventions for people with physical disabilities: A systematic review

The aim of this study is to examine which interventions lead to clinically significant weight loss among people with physical disabilities.


| INTRODUCTION
According to the World Health Organization, approximately 15% of people worldwide live with different types of disability. 1 Although there are over a billion people with physical disabilities, and the number is rising, 1,2 there is a lack of research investigating their needs and challenges.Individuals with disabilities face different barriers, such as health disparities, compared with the general population. 2,3For instance, adults with a disability are less likely to engage in physical activity and are more prone to obesity and cardiovascular diseases, 4,5 as they might face challenges maintaining a healthy lifestyle, particularly a healthy weight. 6Pain and disability limit the possibilities for physical activity, and physical barriers prevent people from accessing health-promoting surroundings such as gyms or parks. 6,710] In view of the unique needs of people with physical disabilities and the growing issue of overweight and obesity, which could lead to further difficulties for affected persons, it is pertinent to identify effective weight loss interventions for this target group.2][13] Considering that weight loss in people with obesity might improve physical function, it is vital to outline different weight management interventions and detect effective treatments for these populations. 5This is especially relevant as people with physical disabilities might have fewer options to, for example, exercise and be physically active. 7The current investigation aims at assessing the effect of behavioral weight-loss interventions for people with physical disabilities and at identifying interventions that lead to clinically relevant weight loss.

| Eligibility criteria
The eligibility criteria were defined by using the Population-Intervention-Comparison Outcome-Study Design (PICOS) schema as outlined in Table S1.The studies had to fulfill the following criteria to be included in the systematic review: (a) randomized controlled trials  15 Within the context of this systematic review, the term "physical disability" will be understood broadly based on the impairment definition of the ICF: "problems in body function or structure such as a significant deviation or loss." 14,16Studies that investigated people with intellectual disabilities or involved other participants as most of their study population (≥60%) were excluded.We focused on behavioral strategies that circumvent bariatric surgery and medication risks and, therefore, could be applied more broadly.Accordingly, surgical and pharmacological interventions and studies with no control group were excluded.

| Search strategy and eligibility assessment
To find eligible trials, we conducted a systematic search from inception to May of 2022 in the following electronic databases: PubMed, Scopus, and the Cochrane Central Register of Controlled Trials (CENTRAL).The detailed search strategy with the respective number of results can be found in Table S2.As disability is a broad concept that tries to grasp, for instance, cognitive and physical impairments, 17 there is no clear definition of the term. 18Accordingly, terms like "physical disability," "physical impairment," "functional limitation," or "mobility impairment" were also defined differently. 16,19Thus, the search terms included "people with physical disabilities" and a variety of synonyms, as well as different conditions associated with a physical disability such as "spina bifida" or "musculoskeletal conditions."Terms were derived from previous systematic reviews examining this target group. 7,20,21ter records were identified, duplicates were removed with the support of referencing software (EndNote X9.3.3) as recommended by Bramer et al., 22 and titles and abstracts were screened.Full texts of the remaining records were assessed for eligibility.In the second step, the reference lists of the included studies were screened to identify additional eligible studies.JH and VO conducted the eligibility and quality assessment independently and solved discrepancies by discussion.

| Data extraction and quality assessment
The following study characteristics were extracted: authors, publication year, country, area (where the intervention took place), study type/design, study duration, age, sex, intervention type, and primary outcomes of interest.
An extensive description of the study participants allows statements about the differences and the comparability 23 in this "heterogeneous population with a wide variety of individual health needs." 24erefore, data on the study population included the number of participants within each treatment group with a description of the physical disability.
The methodological quality of the papers was assessed with the Risk of Bias tool (RoB 2.0) from the Cochrane Collaboration. 25,26e risk of bias assessment involves responding to "signaling questions" in five bias domains: bias arising from the randomization process; bias due to deviations from intended interventions; bias due to missing outcome data; bias in the measurement of the outcome; and bias in the selection of the reported result.Studies are classified with high or low risk of bias or as raising some concerns regarding the bias domains.The assessment was done based on an "intention-to-treat" effect 27 and the outcome of interest (weight or BMI) of this review.

| Data synthesis
We grouped the studies by treatment types, including dietary, physical activity, education/coaching, or multi-component interventions.
Interventions were considered, based on previous research, 28 as short-term if their duration is less than 24 weeks and as long term, if they were longer than 24 weeks.
Mean weight changes in kilograms and percent with the standard deviations were derived to assess the intervention effect.Furthermore, the mean weight and the standard deviation in the pre-test and followups were extracted with confidence intervals if they were reported.If pre-and post-test data were reported without the mean difference in weight, the difference was calculated.If standard deviations were reported, Cohen's d was extracted or calculated as the standardized mean difference between and within groups.For the calculation of within-group differences, we used the following formula with the average standard deviation of two groups as the denominator 29 : : For the between-group difference, we compared the follow-up data of the intervention group(s) with the control group data, using the following formula: q : To account for baseline differences, we also calculated the standardized mean difference with pooled standard deviations from the pre-and post-test using the following formula 30 : The resulting effect size was used to approximate the number needed to treat (NNT) using the Kramer and Kupfer method. 31If the authors did not report the number of participants included in the analyses at follow-up, the number of participants at baseline was used.If studies reported the median and interquartile range, means were estimated based on previous recommendations, 32 using the following formula:

| Risk of bias in included studies
Six studies had a low risk of bias, 48,54,61,66,68,91 23 had a high risk 34,38,40,41,45,46,[50][51][52][53]55,56,58,67,70,72,74,77,80,84,90,95 (five of these were pilot studies), and the rest of the included publications raised some concerns regarding the bias domains (see Figure 2). The main reasons for the "high risk judgment were missing outcome data, lack of information regarding deviations from intended intervention, and inappropriate outcome measurement (weight was measured by self-report), which could have threatened internal validity.97 The highrisk judgment in the outcome data domain related to, for example, high attrition rates and authors not providing reasons for participants leaving trials.The absence of blinding procedures and the lack of description of an intention-to-treat analysis prevented a low-risk judgment in the deviations from the intended intervention domain.One publication had a high risk regarding the randomization, 80 indicated by baseline differences between groups and insufficient reporting of the randomization process.Concerns regarding the selection of the reported result were mostly raised because no pre-specified analysis plan was provided or mentioned in the publications.
Cohen's d of 0.15 indicates a slight difference between the two groups after the intervention.An approximated NNT of 13 indicated that for every 13 individuals receiving the intervention, one person would achieve weight reduction in comparison to the control group.

| Education-based interventions
A pilot trial for people with lower extremity amputation used information material and health coaching via phone, reducing participants' weight after 26 weeks on average by 3.3 kg. 57Over time, the coached group and a self-directed control group differed (insignificantly) by 4 kg.
An effect size of 0.13 suggested a small difference between both groups at the follow-up.Another pilot study 59 included education about exercise and dieting in groups and reported that the weight of the participants who had a stroke or multiple sclerosis declined on average by 2.41 kg after 12 weeks.This change was insignificant compared to an active control group.Ravaud et al. 60 provided individual education sessions with the physicians of patients with knee osteoarthritis.Their intervention led to an average of À1.11 ± 2.49 kg weight change from the baseline body weight, which did not differ significantly from their usual care comparison group. 60The remaining two education and counseling interventions 58,61 did not report a relevant weight change.

| Multi-component interventions
][84][85]88,94,96 The dietary component was complemented by educational or physical activity aspects.Bliddal et al. 66 utilized formula diets (between 800 and 1,200 kcal/day) and offered group meetings with educational components for participants with knee osteoarthritis.Compared to the control group, their intervention significantly reduced 7.5% (95% CI [5.2, 9.9]) of participants' initial Note: Data for each individual study can be found in Table S3.
a Other conditions include spina bifida, lumbar instability, lower extremity amputation, multiple conditions, frailty, knee pain, osteosarcopenic obesity, cerebral palsy, and using a wheelchair.
body weight after 52 weeks.Messier et al., 80 as an example of treatment without caloric restriction, combined exercise programs with nutrition education, leading to an average weight reduction of 9.3% after 24 weeks in the intervention group.This was significantly different from the 1.63% weight loss of participants in a control group receiving only the exercise program.
F I G U R E 2 Risk of bias assessment for the intentionto-treat effect.

| Clinically relevant weight loss
Fourteen studies 54,55,63,[66][67][68]70,80,81,[83][84][85][86]96 employed interventions, leading to a clinically relevant weight loss (>5% reduction of the initial average weight or reduction of one BMI unit) (Figures 3-5). Two of these trias were dietary based, 54,55 and the remaining 12 used multicomponent approaches.63,[66][67][68]70,80,81,[83][84][85][86]96 Most (n = 10) of these multi-component interventions had a dietary component combined with physical activity or educational aspects.Christensen et al., 67 for instance, used low energy formula diets with weekly nutritional instructions and behavioral therapy, which, compared to the control group, reduced an average of 6.8% (95% CI [5.5., 8.1]) of participants' initial body weight after 8 weeks.nosed patients from hospitals or general practitioners.33-36,38,40- 61,64,66-69,74,75,78-81,83,84,87-93,95,96 The included studies involved a wide variety of participants with physical disabilities.The limited findings might apply to populations (BMI > 30) with obesity and mild-to-moderate mobility impairments that are older than 50 years on average.The effects of interventions, for instance, for younger people with more severe disabilities remain to be addressed as only three of the included studies targeted children and young adults.Similarly, only seven identified trials targeted working-age adults under 40 years.35,41,45,50,51,53,65 As people with disabilities are a heterogeneous population, the heterogeneity in the studied groups is not necessarily an issue because "evidence suggests that there are important overarching barriers that people [with disabilities] experience." 24 Hoever, the lack of high-quality investigations limits generalizations for these populations from this review.
Incompleteness regarding outcome data was also an apparent issue.Twenty-one studies were mainly focused on weight-related outcomes.The remaining papers had other primary outcomes.Considering selective outcome reporting in health research, 101,102 this could be one reason weight-related data were not reported as extensive (see Table S5).In this context, it is notable that power calculations for trials are usually based on the primary outcomes. 103Hence, the studies could have been not powered enough to detect significant changes.
Due to the high heterogeneity of studies, low study quality, ambiguity, and imprecision in findings, it can be assumed that the range of applicability of the findings is questionable.

| Potential biases and limitations in the review process
Theoretically, two potentially limiting issues need to be addressed.At Centers for Disease Control, 6 these measures could "underestimate the amount of fat" in individuals.For wheelchair users, for example, there are other weight loss measurements, such as the anthropometric index. 104This challenge could not be avoided as included studies did not report such measures.Second, it is debated if a clinically significant weight loss of 5% should be set as a goal in interventions.
Ross, 105 for instance, argues that such a goal could lead to frustration among the participants and suggests that focusing on causal behaviors might be an alternative.Although this motivational aspect should not be neglected, the current review used the 5% threshold as it is linked to health benefits and provides an option to classify treatments as helpful. 106art from the theoretical concerns, it should be mentioned that the extracted means are measures, which are susceptible to outliers, and because standard deviations (or other spread measures) were only partially available, the findings should be regarded and interpreted with caution.However, means and mean differences were included primarily because they were the most frequently used study outcomes.

| Agreements and disagreements with other reviews
In general, this review supports the claim of other Another systematic review examining the effects of weight management programs for people with intellectual disabilities 13 concluded that effective treatments for their target group differ from interventions for the general population.The current review cannot support these findings or make any final statements in this regard.This might be due to the different examined populations, or because of the existing limitations in both, this and Spanos et al. 13 review.

(
RCTs) published in English; (b) including people with physical disabilities (at least 60% of the study population based on the International Classification of Functioning, Disability and Health [ICF] definition below) 14 ; (c) focusing on behavioral weight-loss interventions (e.g., dietary, exercise or counseling interventions); and (d) reporting at least one of the following outcomes: Body mass index (BMI), weight loss (in kg or percent), clinically relevant weight loss defined as at least 5% percent loss of the initial body weight, or reduction of at least one BMI unit.
incorporated calorie restrictions with meal replacements and health-related education in individual and group sessions in their first intervention group.The second intervention group received an exercise intervention with, for example, strength training in addition to the treatment of the first intervention group, which indicates the aim for an overall lifestyle change for the participants.Short-term interventions (<24 weeks) were mostly aimed at specific behavior changes while often integrating short-term rewards or motivations.An example of a short-term intervention is Thomsen et al. 61 counseling intervention, which focused on reducing sedentary behavior by utilizing motivational counseling and SMS text messages as reminders for participants.The dietary interventions and the multi-component interventions, which included dietary aspects, made use of restricted calorie intake (from 400 to 1800 kcal/day) over different periods.The physical activity and the counseling treatments were mostly short-term and integrated face-to-face sessions (in groups or as individuals) with an educational aspect (knowledge transfer).Education and knowledge transfer in the multi-component interventions were present in the form of instructions and counseling on top of behavioral treatments.

3. 5 |
Intervention effects3.5.1 | Physical activity-based interventionsOne exercise-based study for adults with spinal cord injury reported an average decrease of 0.2 ± 0.2 BMI units after 6 weeks for an intervention group participating in indoor hand-bike training sessions.35This change differed significantly from the control group, which gained on average of 0.3 ± 0.4 BMI units.A Cohen's d effect size of 0.19 indicated a small difference in BMI between the groups after the intervention.The approximated NNT of 10 suggested that F I G U R E 1 PRISMA flowchart of identified studies and the selection process.From: Page et al. 115 Double reporting: different articles reporting the same data (multiple publications for the same study).one person would achieve a weight reduction for every 10 individuals receiving the intervention compared to the control group.The other eight exercise trials did not report relevant weight decreases (Table

4 | DISCUSSION 4 . 1 | 100 4. 2 |
Main resultsThis review adds to a developing body of literature, including underrepresented target groups in health research.In distinction to intellectual disabilities, physical impairments present different challenges in accessing health-promoting environments such as gyms or parks for exercise.Finding appropriate weight loss interventions for this target group could ultimately improve physical function or slow the progression of long-term conditions and adverse health consequences.5Based on our review, there is limited evidence that extensive dietary interventions or long-term multi-component interventions combining F I G U R E 3 Weight changes in percent (%) in education-, dietary-, and exercise-based interventions (n = 20).dietary approaches with exercise and health education might lead to weight loss for individuals with mild-to-moderate mobility impairments and overweight, especially people with osteoarthritis.Dietary intervention components might be practical weight reduction options for people with disabilities having difficulties in mobility and face barriers to access to physical activity.However, it is notable that dietary F I G U R E 4 Weight changes in percent (%) in multi-component interventions (n = 30).interventions should be practical in real-life settings. 98Dietary treatment components leading to weight loss in this review worked with low (around 800 kcal) or very low energy diets (around 400 kcal) or moderate calorie restrictions (around 1,000 kcal) for specific periods.Although similar interventions already showed significant weight loss in other populations, 99 questions of real-life application and adherence to the interventions remain unanswered. 98,Overall completeness, applicability, and certainty of evidence Conceptual problems, especially issues in definitions of disabilities, limit the ability to compare the studies and assess the interventions in terms of practicality.Yet, due to the comprehensiveness of this review and the detailed inspection of multiple intervention types, we provide an inclusive overview of current behavioral weight interventions for people with physical disabilities.Indeed, this review used a general and far-reaching definition of physical disabilities to ensure a broad scope.Included studies defined study populations differently, while most (n = 49) used specific diagnostic criteria or included diag- first, it should be questioned if weight and BMI are suitable and valid measures for people with physical disabilities.According to the US F I G U R E 5 Weight changes in BMI units in multi-component, dietary-, and exercise-based interventions (n = 9).

investigations 107 -
110 that there is a lack of well-designed, high-quality trials examining interventions suitable for the individual needs of people with disabilities.Due to the complexity of the needs of people with physical disabilities, we align with previous research recommending multi-component interventions.[109][110][111][112][113][114]However, we emphasize that these should include a dietary component, as most(12 out of 14)   of the effective treatments presented included a calorie restriction intervention.Maston et al.99 reviewed 11 publications concerning low and very low-energy diets with meal replacements aiming at weight loss for populations with severe obesity.Although their findings were limited due to a lack of controlled trials, they concluded that the described calorie-restricted diets are effective in achieving clinically relevant weight loss in severely obese populations.On that account and based on the similar (but limited) findings in this study, it follows that effective dietary interventions for people with physical disabilities are similar to those for other target groups.However, the studies, which lead to clinically relevant weight loss in this review, also dealt with populations affected by overweight and obesity.

5 |
CONCLUSIONThe importance of targeting obesity in underrepresented subgroups of the population has received growing interest in recent years.Finding appropriate weight management strategies for people with physical disabilities is a preventive effort to control the potential progression of impairment.Currently, there is no definitive evidence for one effective weight loss strategy for people with physical disabilities and obesity.However, our findings demonstrate a general tendency that multi-component weight loss interventions can induce clinically relevant weight loss in people with physical disabilities.Dietary elements in these interventions are a practical option to suit this target group's complex needs.Such strategies effectively improved the participants' weight but required a more rigorous research methodology to generalize results.AUTHOR CONTRIBUTIONS Jihad Hossaini, Vanesa Osmani, and Stefanie Klug conceptualized the study and defined the methodology used in the review.Jihad Hossaini and Vanesa Osmani conducted the systematic search, screening, eligibility assessment, and quality judgment and drafted the manuscript.Jihad Hossaini carried out the data extraction, calculation, and visualization, and Vanesa Osmani verified and revised the syntheses.Jihad Hossaini, Vanesa Osmani, and Stefanie Klug reviewed and contributed to the final manuscript.provided open-access funding.Open Access funding enabled and organized by Projekt DEAL.
Summary of general characteristics of included studies (n = 60).
There is limited evidence for effective weight loss interventions targeting people with physical disabilities.Dietary and multi-component interventions with calorie restrictions and meal replacements or exercise programs might help achieve clinically relevant weight loss, specifically in older adults with obesity (age > 50, BMI > 30) with mild-to-moderate mobility impairments.However, findings in this review are based on heterogenous investigations with partially low quality and, therefore, should be treated with caution.4.6 | Implications for researchFuture research should aim to reduce disparities while addressing definition and classification issues of physical disabilities or provide specific, consistent, and comprehensive descriptions of study populations when conducting and designing controlled trials.There is a lack of interventional studies examining a variety of populations with physical disabilities, for instance, younger age groups in different settings such as schools, clinics, nursing homes, and assisted living facilities, or working-age adults with disabilities under the age of 40.