Therapeutic Exercise to Improve Motor Function Among Children with Down Syndrome Aged 0 to 3 Years: A Systematic Literature Review

Objective: To determine the effect of therapeutic exercises on the motor function of children with Down syndrome (DS) aged 0 to 3 years. Data Sources: A search was carried out on PubMed, PEDro, EMBASE, SCIELO, Lilacs, Cochrane library without publication date restrictions for the terms. Study Selection: The search yielded 1384 eligible articles, which were screened by 2 reviewers. RCTs that would have evaluated the effectiveness of therapeutic exercise were selected, and that would have reported the effectiveness in the outcomes. Data Extraction: The methodology and results of the studies were critically appraised in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes guidelines. Data Synthesis: Six studies were included. Two types of therapeutic exercises, aerobic and neuromuscular, were identied. A therapeutic aerobic exercise was performed using the treadmill, while a neuromuscular exercise was done using unstable surfaces. The exercise frequency ranged from three days to ve days a week, and the duration of each session was between six and 15 minutes. Conclusion: There is moderate to high evidence to support that therapeutic exercise promotes the occurrence of motor patterns such as gait patterns and enhances the motor skills of children with DS aged 0 to 3 years. square 95%

Type of participants: The participants of this study are children with DS aged 0 to 3 years, bearing in mind that, during this period, the interventions reported in the literature aim to enhance the occurrence of motor patterns in children with DS. After age 4, the therapeutic goal is mainly focused on enhancing or rehabilitating the motor functions 18 .

Type of Interventions
The study included all the therapeutic interventions that are duly applied and systematically planned physical exercises with speci c prescription parameters in terms of intensity, frequency, and duration, among others, with the aim of promoting, improving, or maintaining the motor function of children with DS. Subsequently, long-duration interventions, in which the work of large muscle groups were promoted, were classi ed as aerobic exercise 19 . Therapeutic neuromuscular exercise was classi ed as the exercise aimed at improving the participants' balance or exibility and as resistance training, short-duration exercises whose energy system was mainly anaerobic.

Outcomes
The reviewed literature showed publications on the key outcomes of being included in the RCTs assessing interventions in people with DS, however, these are cognitive and developmental outcomes 20 . Considering that a set of basic outcomes established within the motor function in children with DS were not identi ed in the reviewed literature, two groups of experts were consulted, one consisting of parents and/or caregivers of children with DS and another consisting of clinical experts.
The rst group consisted of eight parents of children with DS from the Corporation of Down Syndrome of the city of Bogotá, Colombia. The group of experts comprised a general practitioner, with a Master's Degree in physiology and with expertise in physical exercise; a professional in psychology with experience in psychosocial support; a pediatric neurophysiotherapist with a Master's Degree in Physiotherapy; a physiotherapist with a Master's Degree in Epidemiology and with expertise in prescribing physical and therapeutic exercise; and three physiotherapists with extensive clinical experience treating people with DS.
The group of parents/caregivers rated the outcomes through a printed self-administered questionnaire prepared to that effect, and the group of experts rated the outcomes using a virtual questionnaire. Both groups prioritized outcomes on a 1-9 scale following the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology 21 . The outcomes included in this review were those considered important but not critical (scores of 4 to 6) and critical for clinical decision-making (scores of 7 to 9) 21 .
The outcomes de ned for this age group were gait, balance, motor development, ne motor skills, and executive functions.

Search and Identi cation of Studies
The search strategy was designed based on the Population, Intervention, Comparison and Outcome (PICO) elements of the questions asked. These terms were adapted according to the languages of the different databases explored. A systematic search was conducted from January to December 2019, on databases such as PubMed, PEDro, EMBASE, SciELO, Lilacs, and the Cochrane Library. Additionally, other sources of evidence were consulted to allow the identi cation and analysis of published and unpublished literature (gray literature) that would not have been detected through a systematic search. Manual searches were conducted in the documents found in the reference lists and in journals specialized in the subject. In addition, Epistemonikos was consulted for previous systematic reviews on this topic in order to review the primary studies included in them, and an evidence-based matrix was built based on this information.
This process was developed during the months of January to December 2019.

Selection of Studies
Study selection based on titles and abstracts was performed independently by two trained reviewers (EIRG and YSV). RCTs that assessed the effectiveness of therapeutic exercise and reported the effectiveness in the prioritized outcomes-gait, balance, motor development, ne motor skills, and executive functionswere selected.
Each assessor generated BibTeX les of the selected studies. Using a bibliographic manager, duplicates were regarded as studies with agreement between the assessors, and those that were not duplicated were reviewed individually by the two assessors and their eligibility was discussed and determined. The eligibility of those studies without a discussion-based consensus was decided by a third assessor.
Studies that did not include at least one of the outcomes or applied a combination of therapeutic exercise interventions and pharmacological interventions were excluded from the study.

Extraction and Handling of Variables
Data were extracted through pre-designed data collection formats. The data from the variables were collected for the comparison of the studies and the measurement of outcomes.
For the gait outcome, the data such as time-to-event or changes in the kinematic or kinetic parameters of this variable were extracted. For the balance outcome, the data on displacement of the center of mass or time maintaining postural balance were extracted. The independent variable comprised the type, mode, frequency, intensity, duration of the interventions, place of performance of the interventions (i.e., outpatient consultation or home) and the person in charge of applying the intervention (i.e., physiotherapist, other professional, family member, or caregiver).
Further data extracted from the population were age, sex, sample size for each group, and cognitive impairment.

Assessment of Study Quality
Two independent assessors evaluated the risk of bias for each study using the Cochrane Collaboration tool 22 . The risk was assessed as low risk of bias, high risk of bias, and unclear risk of bias taking into account six domains: random sequence generation (selection bias), allocation concealment (selection bias), participant and staff blinding (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data, and selective outcome reporting (reporting bias). The rating of risk of bias was assessed using the RevMan 5.1 software 23 .

Assessment of the Certainty of Evidence
The assessment of the certainty of the evidence found was carried out using the GRADE approach 21 . The evidence found for each of the outcomes was rated considering the risk of bias, inconsistency, direct or indirect evidence and imprecision, the risk of selective outcome reporting, and the dose-response gradient. These outcomes were classi ed using a three-level ordinal scale that included very serious, serious, and not serious, except for the risk criteria for selective outcome reporting (not detected or strong suspicion), the size of the effect (no effect, large, or very large), the confounding factors (no effect, it would reduce the effect demonstrated, or suggest spurious effect), and the dose-response gradient (no or yes), in which nominal and ordinal scales with other levels were used 24 .

Synthesis of Data
The selected body of evidence was assessed by prioritized outcomes. Each outcome described the population's features; the parameters of the interventions including the exercise mode applied, frequency, intensity, and duration of the interventions applied in the said studies; and the quantitative results achieved with their level of signi cance, shown in Table 1. The data were synthesized on a Microsoft Excel base, extracting data from the population's features, randomization methods, outcome measures, duration of follow-up, and assessment methods from each study. The meta-analysis considered direct comparisons between the experimental group who did the interventions (aerobic exercise and resistance exercise) and a control group who performed educational activities, recreational activities, or continuity with activities of daily living or interventions other than those of interest for this review.
Averages and standard deviations of the data available from the selected studies were extracted from the prioritized outcomes included in the studies. When the studies reported standard errors of the mean, the standard deviations were obtained by multiplying standard errors of the mean by the square root of the sample size. Standardized mean differences (SMDs) and 95% con dence intervals (95% CI) were calculated to combine the results of the studies using different measures for the same concept or of studies presenting variability in its features.
Heterogeneity between trials was assessed using the chi-squared test, a signi cance value of p <0.05 after due consideration of the value of I2 25 . In the presence of heterogeneity of the results determined by I2 greater than 70%, the results were combined using the random effects model and the 95% CI was calculated. All of the above were carried out with the RevMan 5 software 23 .

Selection of Studies
A total of 1384 studies were found as a result of the systematic literature search. 239 studies were found in other sources that included the bibliographic references of the studies found in the systematic search and in those provided by the group of experts, amounting to a total of 1623 identi ed studies. Of these studies, 88 duplicated ones were excluded and 1178 studies were excluded considering the review of the titles and abstracts. The two assessors reviewed a total of 357 full-text studies, of which 347 were excluded because they did not meet the eligibility criteria, mainly due to the type of design, and because they did not include any of the prioritized outcomes for the systematic review. The ow chart of the studies found and included in the body of evidence is presented in Figure 1. Finally, six primary studies reported in eleven journals (thread) were included. Table 1 shows thread articles, primary studies, and reports linked to them 26 . Table 1 Primary studies and reports.
References of the studies included in this review: Reports linked to the studies included in this review: Five studies showed unclear bias 27,29,34,35 due to the fact that, although they mentioned the random allocation, the allocation methods for participants were not clear and whether the personnel in charge of maintaining the random allocation was masked was not established.

Blinding
Due to the nature of the interventions used, the assessment of the risk of bias took into account the masking of outcomes by the assessors in each study. In the studies included in this review, four studies showed a high risk of bias due to lack of masking 27,29,35 , since neither the staff nor the participants were masked. In the remaining studies 14,34 , the risk of bias was unclear ( Figure 3).

Outcomes with Incomplete Data
Two of the studies included presented unclear risk of bias, as participants who did not adhere to the treatment protocol were excluded from the analysis 14,34 .
Only one study presented a high risk of bias on this item since 5 out of 12 participants were missing in one of the comparison groups. Only the results from 7 participants were included in the analysis 27 .

Selective Reporting
One study was found to present high risk of bias 27 . The time of the independent walking event is considered an important outcome; nonetheless, the authors did not clearly report the time elapsed from the commencement of the study to the event of interest. Additionally, some data such as the analysis of video recordings collected during follow-ups were not reported.

Other Potential Sources of Bias
None of the studies included in the review clearly mentioned the training processes of the outcome assessors or the adjustment and calibration processes of the equipment used, which is the reason why all of the studies, with the exception of the Looper study, were considered to have unclear risks of bias 27 .
This information is summarized in Figures 2 and 3.

Types of Therapeutic Exercise and Modes of Application in Physiotherapy Interventions in Children Aged 0 to 3 years
In the literature included, only two types of therapeutic exercises were reported; the rst one was classi ed as aerobic exercise as it included longer-duration interventions that promoted the work of large muscle groups 19 . Five out of six studies included in this review applied this type of exercise and all coincided in the way the exercise was applied, by using the treadmill 14,27,29,35,36 .
The second type of exercise identi ed was neuromuscular, namely, the exercise that aims to improve the balance or exibility of the participants. It mainly includes unstable surface activities 19 . The study by Harris SR was the only one including this type of exercise 34 . Table 3 includes the characteristics of the interventions along with their respective application parameters.

Frequency, Intensity, and Duration of the Interventions Used in This Population
In those studies that applied aerobic therapeutic exercises using the treadmill (mode), the frequency ranged from three days 36 to ve days a week 14,27,29,34,35 .
The duration of each session varied between six 29,32,37 , eight 27,35 and fteen minutes 36 . The intensity was determined by the treadmill's speed, which ranged With regard to the person who applied the intervention, this was carried out by professionals in the case of the studies by Lowe, L. 36 and Angulo-Barroso, R 37 .
In Looper, J. 27 , Wu, J. 29 , and Ulrich, D. A. 35 studies, parents were trained to apply the intervention at home.
Harris SR et al. 34 assessed an intervention that was different from the aerobic exercise. They applied neuromuscular exercise with a frequency of 3 times a week for 9 weeks, 40 minutes a day. This intervention was carried out by parents at home after receiving previous training.

Outcomes Assessed in the Studies Included in the Review
Of the outcomes proposed for assessment, no evidence was found for the executive function, balance, and ne motor outcomes in this population. Table 2 includes the features of the studies included in this review.

Gait
Five studies reviewed the effect of therapeutic exercise on the participants' gait. Angulo-Barroso 14 , Looper, J. 27 , Wu, J. 29 , and Ulrich, D. A. 35 studies included the average time to achieve independent gait ( Figure 4). Wu, J. 29 included 30 children with an average age of 10 months. These participants were included in the study when they could remain seated for 30 seconds. The outcome they assessed was the time to achieve independent gait and kinematic parameters of gait, as in the study published by Angulo-Barroso 14 . Finally, Lowe, L. 36 included 24 participants in his study, with ages that ranged from 26 to 51 months, with the aim of assessing gait performance using the ten-minute gait test ( Figure 5).

Assessment of the Certainty of the Evidence Identi ed
The certainty of the evidence for the gait and motor development outcomes was moderate and high, respectively ( Table 3). The certainty of the evidence was considered moderate mainly due to the imprecision of the results obtained in the primary studies and the indirect evidence from one of the studies 36 .  Bearing in mind that motor function is a construct that encompasses multiple outcomes and that therapeutic exercise interventions under prescription parameters may favor one outcome over another, this review determined outcomes based on the preferences of parents and caregivers and professional experts treating this population. It intended to identify responses in the literature that could meet their interests and, in the case of the experts, to give response to their concerns when making clinical decisions about which type of intervention to use and effective prescription parameters to achieve successful outcomes of interest that will ultimately become the therapeutic goals of clinical interventions.
The evidence identi ed was scarce in terms of interventions and selected outcomes and their quality. Although they corresponded to randomized clinical experiments, they presented high risk and unclear risk of bias in aspects that jeopardize the internal validity of the study and therefore the certainty when measuring the effect, for example, in the random allocation 27,29,34,35 , in the concealment 27,29,29,35 , in the selective data reporting 27 , and in the follow-up losses, which could lead to selection bias 14,34 . Furthermore, the sample sizes were small, which may explain the width of the con dence intervals and the insigni cant differences reported by some studies 34,36 .
suggest the use of this intervention as an adjunct to the interventions carried out in rehabilitation centers. However, before recommending its use at home, budget impact and cost-effectiveness analyses would be required to determine whether the bene ts achieved would justify the cost of including these interventions 41 .
Only one study applied the intervention with the aim of enhancing the gait patterns of children aged between 26 and 51 months. In this case, no signi cant differences were found that resulted from the intervention, which may be explained by the frequency and duration parameters, as the frequency was three days a week and 15 minutes a day. Another explanation for these results could be the small size of the sample, which could result in a type 2 error. Another reason could be the selection bias since there was a difference in the number of girls and boys included and because the population included children with DS and cerebral palsy, among others, and the authors did not carry out a subgroup analysis 42 .
The other outcome reported in literature was motor development. This was the purpose of studies that included aerobic exercise using a treadmill 14,27,29,35,36 and neuromuscular exercise 34 . Signi cant differences were reported when using the parameters.
Only one study reported the application of this type of exercise to improve the motor development in children with DS. The authors did not report signi cant differences in the outcome measured using the Bayley and Peabody Scales. There is evidence of the effectiveness of this type of exercise in improving the balance in older children with DS 43,44 ; however, this outcome was not measured in the aforementioned study.
There are innumerable interventions regularly used in physical rehabilitation in institutions treating children with DS that include rehabilitation approaches such as Bobath and Vojta, among others. Hydrotherapy and hippotherapy interventions are also offered in the management of these children. Surprisingly, there is no good-quality evidence to support the use of these modalities 45 . Interventions such as hydrotherapy or aquatic therapy, which has been proven to be effective in improving clinical variables in other populations 46,47 , did not provide evidence that could support their use in the subject population of this review.
Future studies are expected to assess the effects of interventions that are currently used with robust research designs. New evidence is required that increases certainty regarding the measurement of the effects achieved by the studies herein reported. Additionally, it is important to include budget impact and costeffectiveness analyses for the interventions mentioned herein.

Limitations of the Study
One limitation of the study is the low number of studies that ful lled the eligibility criteria in terms of outcomes. Therefore, future studies may yield different results for the outcomes posed in this review.
The small number of studies was also reported by the authors themselves, which does not allow for a comparative analysis between prescription parameters and even the mode of application of the exercise.
No studies in children aged less than nine months were identi ed.

Conclusions
There is moderate to high evidence to support that therapeutic exercise promotes the occurrence of motor patterns such as gait patterns and improves the motor skills in children with DS aged 0 to 3 years.
More better-quality evidence is required to validate the mode of application and other prescription parameters discussed here.
Future research is required to support the use of effective prescription parameters of the many interventions currently employed in care settings within this population.   Risk of bias summary: review authors' judgments on each risk of bias item for each included study.  Aerobic exercise (treadmill) versus control (activities of daily living) outcome: independent gait, time(days)-to-event. Aerobic exercise (treadmill) versus control (physiotherapy plan) outcome: gait speed, measurement unit (m/s), and follow-up (4 weeks and 6 weeks).