Adherence and dropout in exercise‐based interventions in childhood obesity: A systematic review of randomized trials

Our objective was to systematically examine the characteristics of exercise interventions on adherence and dropout in children and adolescents with obesity. PubMed, Embase, PsycINFO, Lilacs, Scielo, and The Cochrane Central Register of Controlled Trials and reference lists of relevant articles were searched. We included randomized controlled trials with exercise interventions for pediatric patients with obesity presenting data on dropout and/or adherence. Two reviewers screened the records independently for eligibility with disagreements being resolved by a third reviewer. Twenty‐seven studies with 1268 participants were included. Because of high heterogeneity and poor reporting of adherence, it was not possible to perform a meta‐analysis. Dropout prevalence was calculated, and subgroup analyses comparing different types of exercise and a meta‐regression with potential moderators were performed. We found a dropout rate of 13%. Subgroup analyses did not identify significant differences. The duration of the exercise presented a moderating effect on dropout, suggesting that longer exercise sessions may lead to higher dropout in children and adolescents with obesity. Because of the poor adherence data, it is not clear which exercise characteristics may moderate adherence. To improve the quality of childhood obesity care, it is mandatory that future studies present adherence data. Systematic review registration: PROSPERO CRD42021290700.


| INTRODUCTION
Physical exercise is an effective and safe intervention to treat and prevent childhood obesity, and the positive effects go beyond body mass index (BMI) reduction and body composition improvements. 1idence suggests that engaging in regular exercise also enhances cardiorespiratory fitness, lowers triglyceride levels, and improves low-density lipoprotein and total cholesterol. 2Aligning with this perspective, the Physical Activity Guidelines for Americans recommend that children and adolescents aged 6 to 17 years engage in 60 min or more of moderate-to-vigorous intensity physical activities each day, including aerobic, muscle-strengthening, and bone-strengthening activities. 3e short-term benefits of regular exercise on health-related outcomes are well documented, 4 and alarming statistics from global trends in physical activity highlight a prevalent deficiency in adequate exercise.Globally, only 19% of students aged 11-17 years adhere to recommended activity levels that highlight the need for the implementation of established effective policies and programs to enhance physical activity, as well as increased investment to address the diverse causes perpetuating the low participation rates. 5herence is broadly defined as the degree to which a patient (or participant) follows the recommendations of a prescription or intervention. 6There is a consensus that adherence to an intervention must be reported in all clinical trials.In an exercise-based intervention, adherence is calculated as the number of sessions attended related to the sessions planned. 7,8When a patient (or participant) withdraws from an intervention for any reason, it is defined as a dropout.Along with adherence, dropout rate is an important factor to be considered when interpreting data from clinical trials and should also be clearly reported.
Low levels of adherence pose challenges by potentially undermining the benefits of exercise on health outcomes, both within the specific context of randomized controlled trials (RCTs) and within the broader population. 9Although there is evidence suggesting that adolescents with a high level of peer support are more likely to engage in sufficient physical activity, research on exercise adherence remains imperative, as no single variable can comprehensively explain or predict adherence. 10better understanding of exercise programs' characteristics, such as type, intensity, and duration, and their associations with adherence and reduced dropout, may lead to the development of more robust guidelines to support clinicians and policymakers to prescribe effective interventions for children and adolescents with obesity thereby improving long-term success rate.Therefore, the objective of the present systematic review is to describe those exercise program characteristics that may be related to improved adherence and reduced dropout in children and adolescents with obesity.

| METHODS
Reporting of this study is in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA). 11The protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO) (registration number CRD42021290700), and we incurred no deviations from the a priori review protocol; this systematic review and meta-analysis protocol has not been published elsewhere.There are no restrictions regarding publication date."Pediatric Obesity" OR "Adolescent Obesity" OR "Obesity in Adolescence" OR "Childhood Onset Obesity" OR "Obesity in Childhood" OR "Child Obesity" OR "Childhood Obesity."The following terms were excluded using the Boolean operator "NOT": "Behavior Therapy" "Life Style," and "Prevention."Filters were used to select only RCTs when available.

| Sources of information and search strategy
The search strategy was validated by testing whether it could identify relevant studies.Two additional manual searches were conducted to identify other eligible studies and to complement the electronic searches.Reference lists of eligible studies were consulted and references from systematic reviews with similar population and intervention were also analyzed; the processes were blindly performed in duplicate by JG and ES.Detailed search strategies for each database are available in the Supporting Information.

| Selection of studies and data extraction
Studies identified from all six databases were imported into EndNote.To identify multiple reports on the same study, the clinical trial register number, author names, dates, and sample characteristics were analyzed, and in cases where there was more than one register with the same sample, only the report that provided the most complete information for review was considered.
Eligible outcomes were adherence (defined as the percentage of sessions attended relative to sessions offered) and dropout (defined as the number of participants who discontinued the study during the intervention period).Among studies that did not report exercise adherence but did report the number of sessions attended and the number of sessions offered, exercise adherence was calculated (attendance/offered session*100).
For classification of exercise intensity, the author's criteria were used primarily.When the author did not rate intensity, the Physical Activity Guidelines for Americans criteria were used. 12For studies that required data confirmation and/or did not include the values required for analysis, authors were contacted by email.

| Risk of bias
Risk of bias was assessed independently by two trained reviewers (JG and RCA) using the Physiotherapy Evidence Database (PEDro) rating scale.This instrument consists of 11 items and a score that varies between 0 and 10 points.However, the maximum score on the scale for complex interventions (interventions involving exercise) is 8; therefore, among these clinical trials, it was not possible to blind the therapist and the subjects.The PEDro scale has excellent levels of reliability and validity 13,14 and assesses the quality of the study (items 2 to 9) and the presence of statistical information that allows the interpretation of the results (items 10 and 11).Any discrepancies in the evaluation were discussed until a consensus was reached.

| Data analysis
We conducted a meta-analysis with random effects using JAMOVI version 2.3. 15For the meta-analysis procedures, we used the MAJOR R 1.2.2 package 16 in the following sequence: initially, the dropout prevalence was calculated together with the confidence interval (95% CI) among all studies.Then, a subgroup analysis was performed comparing dropout rates between different types of exercise (resistance, aerobic, and combined).Subsequently, we performed a metaregression analysis with potential moderators that could influence dropout rates across studies.The percentage of girls in the studies, type of exercise performed, weekly frequency, duration, and intensity of exercise were included as moderators.The significance level adopted was p < 0.05.Heterogeneity was assessed using the I 2 statistic, where values >50% were considered as having substantial heterogeneity. 17blication bias was analyzed by visual inspection of the funnel plot and by the Begg-Mazumdar Kendall's tau 18 and Egger bias tests. 19After analyzing the funnel plot with the results of all studies, we performed the trim and fill 20 adjustment to remove extreme outliers and recalculated the pooled dropout rates.
In line with the review protocol, we synthesized the evidence narratively when meta-analyses could not be conducted due to heterogeneity.

| Study characteristics
Initial search of the databases identified 916 records, and 183 additional records were retrieved in the update search, identifying a total of 1099 studies.After removing duplicates and assessing titles, abstracts, and full texts for eligibility, 18 studies were selected for data extraction (16 retrieved in the initial search and two from the update search).After reviewing the reference list of included studies and systematic reviews with similar populations and interventions, 48 titles were found, of which nine were considered eligible, resulting in the 27 studies included in this review.For details and reasons for exclusion, see Figure 1.

| Adherence
Only nine of the 27 studies reported participant adherence, three studies 21,24,25 did not specify how they defined adherence, and all other studies defined adherence as the percentage of sessions attended of the sessions planned.The studies had a mean PEDro score of 5 (range 7-4 points).Because of high heterogeneity in adherence (I 2 = 99.2%;p < .001),meta-analyses could not be conducted.In line with the review protocol, we synthesized the data narratively.
Three studies 25,33,40 that reported adherence used HIIT as an exercise intervention, with an average workout duration of 20 min (range 11-25 min) and a total of 65 participants.The average adherence to the HIIT protocol was 86% (range 68%-100%), and there were 13 dropouts.Two studies 25,33 used protocols of bouts of 15 s at 90%-100% of maximal aerobic speed and 15 s at 50% maximal aerobic speed with 3 min of rest between bouts.In one study, 40 four 4-min exercise cycles at 90%-100% Maximum Heart Rate (HRmax) were interspersed with 3 min of active recovery at 50%-70% HRmax.
The Racil 33 and Meng 25 studies had higher adherence (100% and 89%, respectively) and lower dropout (zero and three participants, respectively) than the Dias et al. study, 40 in which adherence was 68% and 10 participants dropped out.
Two studies 30,33 used moderate-intensity interval training as an exercise intervention.Tan et al. 30 incorporated a 40-min protocol that included 4 min of exercise at 50% of heart rate (HR) (with a target HR of 148 beats per minute) followed by 1 min of rest.Racil et al. 33 followed a 24-min workout protocol with bouts of 15 s of exercise at 80% maximal aerobic speed, interspersed with 15 s of exercise at 50% maximal aerobic speed, and 3 min of rest between bouts.With a total of 37 participants, the average adherence was 99%, and there were no dropouts in either study.
Two studies 28,39 used RT as an intervention.In one of the studies, RT was used alone, 28 and in the other, RT was combined with continuous endurance exercise. 39In the RT study, 22 subjects participated in the protocol, which consisted of 57 min of muscle strength and neuromuscular training exercises focusing on the hip abductors and quadriceps muscles at low intensity.Adherence was 57%, and three participants dropped out of the intervention.In the study that combined RT with continuous endurance exercise, a total of 12 subjects participated in the protocol with four to seven resistance stations alternating with three to five aerobic stations at moderate intensity for approximately 45 min, no participant dropped out, and adherence was 98%.
Five of the nine studies in which adherence was presented used continuous endurance exercise as an exercise intervention, 21,22,24,25,40 with a total of 132 participants and an average workout duration of 37 min, adherence was 81% (range 56%-95%), and a total of 25 dropouts were registered.Three studies 22,25,40 used moderate-intensity interventions: in the McCormack et al.
There were no dropouts in the study by McCormack et al., 22 four dropouts in the study by Meng et al., 25 and eight dropouts were recorded in the study by Dias et al. 40 Two studies 21,24 used low-intensity continuous endurance exercise, and one study used high-intensity continuous endurance exercise.In the Tan et al. study, 24  20% below ventilatory threshold 1 with a group of 31 participants who exercised for approximately 38 min at a high intensity equivalent to ventilatory threshold 1.The group exercising at low intensity had 80% adherence, and six participants dropped out of the intervention, and the group exercising at high intensity showed 83% adherence, and five participants dropped out of the intervention.

| Dropout
Of the 27 studies 26 clearly reported dropout; in the study by Hagströmer et al., 45 there were a total of 16 dropouts, but it was not stated which of these subjects were in the exercise group and which were in the control group.The study was not included in the metaanalysis.The 26 studies had a mean PEDro score of 5 (range 7-4 points).
The dropout rate in studies that used aerobic exercise programs was 12% (IC 95% 8 a 16%, I 2 = 61%); in studies that used resistance exercise was 16% (IC 95% 5 a 27% I 2 = 0%); and for programs with Forest plot for meta-analysis of the 34 exercise groups reporting dropout (excluding Hagströmer et al.).
F I G U R E 3 Funnel chart of the combined analysis.
Meta-regression analysis revealed that dropout rates were not moderated by the percentage of girls in the studies, type of training, and exercise intensity and frequency.The only training variable that showed a moderating effect on the dropout rate was the workout duration (β = 0.036; SE = 0.015; Z = 2.44; p = 0.015).The details of all meta-regressions are summarized in Table 2.

| Risk of bias
Table 3 displays the scores achieved by the studies for each item and the total score on the PEDro scale.The PEDro score of the studies ranged from 4 to 7, with a median of 5 points.Only four studies (15%) performed a concealed allocation of volunteers and presented blinding of the evaluators.Fifteen studies (56%) had losses to follow-up of less than 15%, and only four (15%) studies reported performing intention-to-treat data analyses.

| DISCUSSION
The results of this systematic review indicate that the measurement and reporting of adherence rates are generally inadequate, with only 33% of the identified RCTs reporting adherence.Because of the high heterogeneity between the studies, it was not possible to identify potential moderators of adherence.Of the variables examined, only one moderator of dropout was identified, and that was workout duration.Based on the available data, we cannot determine whether the characteristics of the exercise intervention (type, workout duration, and intensity) affect adherence rates.Consequently, we cannot make a firm recommendation to clinicians on which training approach is most effective in improving adherence to exercise-based interventions in children and adolescents with obesity.
Among the studies analyzed, the average adherence was 84%, and the dropout rate was 13%, indicating that children and adolescents with obesity can sustain the exercise program over 12 weeks, with varying exercise durations, modalities, and intensities.Only a minority of the studies (23%) reported adherence less than 80%.
However, it is possible that this high level of adherence in the RCTs is Meta-analysis of dropout rates after trim and fill adjustment for the 19 exercise groups.
T A B L E 2 Meta-regression of dropout moderators in children and adolescents with obesity.Note: • adds a point on the score; adds no point on the score.The item "eligibility criteria" is not included in the final score.overestimated because the majority of studies did not report adherence.
Another consideration regarding the high adherence and low dropout rates is the potential influence of the intrinsic characteristics of the study design.Consistent with findings from other studies, RCTs typically demonstrate higher adherence levels due to their highly supervised nature, in contrast to the comparatively lower adherence often observed in unsupervised and observational studies. 48Additionally, the type of recruitment may have also played a role in the high adherence rates.For instance, studies recruiting participants directly from schools or clinics often benefit from a more controlled and supervised environment, potentially contributing to the observed high retention rates.Our study's primary focus was the evaluation of adherence and dropout within the framework of structured exercise for children and adolescents with obesity.The intentional focus on structured programs was guided by the intent to highlight the distinctive challenges and characteristics inherent in such interventions.
By acknowledging these limitations and analyzing exercise interventions, we could identify one moderator of dropout: workout duration.In other words, the longer the workout duration, the higher the dropout rate.This discovery aligns with the findings of Collado-Mateo et al., 9 who observed in a study involving patients with chronic diseases and older adults, that longer intervention durations resulted in lower adherence among individuals participating in the program.
Unfortunately, reporting of reasons for dropout was inconsistent.
Most studies reported nonspecific reasons such as personal/family reasons.It would be interesting to obtain more specificity from this information to corroborate the data found, so we recommend that researchers provide more details on the reasons reported by participants for dropping out of the exercise intervention as they may help to identify possible moderators.
Because of the short workout duration, HIIT is often described as a facilitator to adherence.According to previous studies among adolescents with normal weight, HIIT is more enjoyable compared with exercise at a moderate intensity continuously. 49,50In the present study, HIIT showed 86% adherence and continuous moderate-intensity exercise 76%, but different types of protocols elicited different levels of adherence.HIIT protocols with bouts of 15 s at 90%-100% maximal aerobic speed with 15 s at 50% maximal aerobic speed showed higher adherence rates and lower dropouts than bouts of 4 min at 90%-100% HRmax with 3 min of active recovery at 50%-70% HRmax.The same occurred during continuous moderate-intensity exercise: 20 min of aerobic exercise at 60%-80% of HR reserve or 30 min at 60%-65% of maximal aerobic speed resulted in better adherence rates and lower dropout rates than 44 min of exercise at 60%-70% HRmax.
In the present study, two protocols 30,33  Little is known in the literature about adherence to RT programs.
In the present review, only one RCT 28 presented adherence rates to RT.The study showed an adherence below 60%, but when compared with another RCT, 39 we see that adherence to RT programs increased and dropout rates decreased when RT was combined with continuous endurance exercise.Even though the comparisons between groups did not identify significant differences within the dropout rate, in studies that used aerobic exercise programs, dropout rates were 12%, and in studies that used resistance training programs, it was 16%.In contrast, among those studies that used combined programs, dropout rates were 5%, which may suggest that combining RT with continuous endurance exercise may be a better option for children and adoles- Inclusion criteria were as follows: (a) study design: RCTs; (b) population: pediatric patients (aged 5 to 19 years) with obesity (according to the author's criteria); (c) intervention: structured exercise interventions according to a predetermined frequency and/or intensity; (d) control: Participants maintain their current level of physical activity.(e) Outcome: data on dropout and/or adherence; (f) language: English, Spanish, and Portuguese.Studies offering a financial incentive for attending sessions and those with prescribed diet were excluded.
Two independent investigators (JG and ES), with experience and appropriate training, searched six electronic databases: PubMed, Embase, PsycINFO, Lilacs, Scielo, and The Cochrane Central Register of Controlled Trials.An initial search occurred in January 2022 followed by an update in January 2023.Searches within Medical Subject Headings and Descriptors in Health Science were performed to identify appropriate search terms.Additionally, reference articles were reviewed to identify common keywords based on titles and abstracts.A pilot search strategy was conducted, and additional keywords were identified and incorporated into the strategy.A search strategy was created combining variations between two groups of keywords: (1) Exercise: "Exercise" OR "Physical exercise" OR "Physical training" OR "Aerobic exercise" OR "Endurance training" OR "Circuit-based exercise" OR "Resistance Training" OR "Strength Training"; and (2) Children and adolescents with obesity: Duplicates were independently excluded by JG and ES, first using the software tool and then visually by comparison.Eligibility was verified in three steps: titles, abstracts, and full text.JG and ES independently assessed all titles and abstracts and excluded those not meeting the eligibility criteria; disagreements were resolved by a third reviewer (JB), senior researcher.The full text of all eligible studies was retrieved and independently reviewed, again, for eligibility by JG and ES.An Excel spreadsheet was used to extract the following data from the included articles: (a) author, (b) year of publication, (c) title, (d) clinical trial registration, (e) obesity definition, (f) age, (g) sex and, when the study had a mixed sample of boys and girls, percentage of girls in the group, (h) details of the exercise intervention: duration (weeks), frequency (times per week), type, intensity, and duration, and (i) dropout and adherence data: initial and final sample of the exercise group, percentage of adherence to exercise, and reasons for dropout.Data extraction was also performed independently by two trained reviewers (JG and ES) and then compared with any discrepancies being resolved through discussion.
cents with obesity than RT alone.This is a comprehensive review of adherence and dropout among exercise interventions conducted in line with PRISMA guidance.However, several issues need further consideration.The findings should be interpreted with some caution because of the low quantity of studies reporting of adherence.In addition, adherence may have been further inflated because the interventions were relatively brief, and there is a possibility of bias because studies not published in English were not included.5| CONCLUSIONOnly 33% of existing exercise RCTs performed among children and adolescents with obesity report adherence and program dropout data during the intervention period.To appropriately identify effective interventions and correctly interpret health outcome results from an exercise study, adherence should be reported.Researchers are urged to include clear definitions and presentation of adherence data in future studies so that clinicians can make a balanced cost-benefit decision regarding the implementation and maintenance of physical exercise.
Characteristics of included studies.
and high 21 intensity as exercise interventions.Two studies did not provide clear information on training intensity: the study by Meyer et al. 29 stated that the intensity F I G U R E 1 RISMA flow diagram.T A B L E 1 Physiotherapy evidence database (PEDro) score of the included randomized controlled trials.