Effects of Classroom Active Desks on Children and Adolescents’ Physical Activity, Sedentary Behavior, Academic Achievements and Overall Health: A Systematic Review

The purpose of this systematic review was to examine the effects of active desks in the school setting on sedentary behavior, physical activity, academic achievements and overall health among children and adolescents aged 5–17 years. A systematic literature search was conducted using five databases until October 2020. Twenty-three studies were included. Studies reported an increase of around 36% in energy expenditure for cycling desks and between 15% and 27.7% for upright active desks. Children increased inhibitory control and selective attention capacity while using cycling desks. A heterogeneous quality of design and of results were observed limiting comparisons and conclusions for each active desk. Despite the lack of strong methodology for the included studies, active desks appear to be a promising intervention in classrooms to improve health-related outcomes in children aged 5–17 years. Due to weak methodology, future studies with stronger study designs and methodology are needed to better inform policy and practice about the role of classroom active desks on health-related outcomes in children and adolescents.


Introduction
Concerns and research regarding the effects of sedentary behaviors and physical inactivity on overall health have been growing for the last decades, leading nowadays to a better identification of their independent and joint implications [1,2]. While sedentary behaviors is defined as any waking behavior characterized by an energy expenditure ≤1.5 metabolic equivalents, while in a sitting, reclining or lying posture [3,4], physical inactivity is typically defined as "the non-achievement of physical activity guidelines" [5]. Both sedentary behaviors and physical inactivity have substantially increased in our societies, with physical inactivity being identified as the main cause for about 1.6 million deaths worldwide [6] and leading to a public health cost of $53.8 billion per year [7]. Due to their important implication in the risks of all-cause mortality and cardio-metabolic morbidity as well as in some cancer occurrence [8], both sedentary behaviors [9,10] and physical inactivity [11,12] are of public health concern today. In children and adolescents, it has been found with device-based measurements that daily sitting time takes over 50% of the waking day at 7 years and 75% at 15 years [13]. This high level of sedentariness, combined with the fact that about 80% of children and adolescents are inactive (i.e., not reaching the physical activity recommendations) [14,15], led some scientists to propose the existence of what they called a "Sedentary & Inactive" profile [16]. Not only physical inactivity and sedentary behaviors have been found to be associated with early metabolic and cardiovascular risk in children and adolescents [17][18][19][20][21], they have also been found to be related to a decrease in cognitive performance and academic achievements [22][23][24].
Knowledge and behaviors developed during childhood have been shown to influence their future behaviors as adults [25]. In particular, children's physical activity and sedentary behaviors have been shown to not only determine their actual health but also their adolescent and adult behaviors and health [26]. Since children spend at least one third of their waking time in class [27], school appears as an ideal setting to promote health and induce behavioral change [28]. Targeting school time and the school place to promote healthy active behaviors necessitates however to face the highly sedentary nature of the children's class time. In that context, the literature shows a growing number of experiments trying to implement interventions aimed at breaking and reducing this sedentary time during class [29,30]. The use of active desks in the classroom (e.g., standing desks, sit-to-stand desks, cycling desks, stability balls) has been especially studied [31][32][33][34], with studies showing for instance that sit-to-stand desks seem to reduce sedentary time in the classroom [31] or increase energy expenditure with the use of bike desks [34]. These studies are providing some promising results and our aim is to conduct a systematic analysis of these works to have a better understanding of their effects.
Previous reviews have examined the effects of standing desks on children and adolescents [35,36]. Regarding, active desks, while some already systematically reviewed their effects on academic achievement [37] or questioned their use among specific groups (e.g., overweight and obese) [38], no review has specifically studied the impact of classroom active desks on cognitive, academic and overall health-related (physical, metabolic and mental health) outcomes among children and adolescents. Having a global picture on the role of classroom active desks on improving health-related outcomes of children and adolescents is needed to inform policy and practice.
Thus, the objective of the present systematic review was to analyze the existing literature on the implementation of active desk in the school environment and examine their effects on physical activity, sedentary behavior, academic achievements and overall health in children and adolescents aged 5-17 years.

Methods
This research is registered in PROSPERO as CRD42020196096. This review was completed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) to identify and collate studies [39].

Search Strategy
A literature search was conducted for studies from year 1990 to October 2020 using the following electronic bibliographic databases: PubMed, ScienceDirect, ResearchGate, Google Scholar and Medline (Cochrane Library). The search terms included the key words "desk* or workstation* or work station" AND "treadmill OR pedaling OR cycling OR bicycl* OR bik* OR active OR exercise ball* OR swiss ball* OR stability ball* OR dynamic seating OR active sitting OR standing OR stepping OR stand up OR position, standing OR standing position* OR sit-to-stand OR sit stand OR stand/sit OR stand biased OR adjustable furniture OR height adjustable" AND "school* OR class* OR child* OR student* OR academic institution". To identify articles potentially missed during the literature search, reference lists of candidate articles were reviewed. Table 1. Active desks characteristics and range of price.

Active Desk Type Description Range of Price (USD) Pictures
Upright active desk Corresponds to standing desk, sit-to-stand desk or stand-biased desk.

Data Collection
Full texts from the articles were imported from a reference manager software (Zotero software; 5.0.21, CHNM, GMU, USA). After removal of duplicates, a screening was conducted by two independent authors on titles and abstracts to assess study eligibility (CC, TG). Identical procedure was used by the same authors on full text articles (CC, TG). Any disagreement regarding eligibility for inclusion was discussed until consensus emerged as made among the research team members. Each author completed data extraction files for every paper included. The process for trial inclusion is shown in the PRISMA flow chart (Figure 1).

Risk of Bias, Study Quality Assessment and Result Consideration
Risk of bias was independently examined by two authors (CC and TG) using the Cochrane risk of bias tool [60] (Table 5). Selection bias, performance bias, detection bias, attrition bias and reporting bias were assessed. The quality of evidence for each outcome by type of study design was determined using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework [61] (Table 6). Any divergences were reported to the research team (MD, DT, LM). We did not exclude studies on the basis of risk of bias or low quality evidence. Importantly, the results of all the included studies and their directions, have been reported whether or not a statistical analysis was performed and if yes, précising whether the results reached or not the level of significance.     [34,40,42,44,56]. b One study did not detail the age and sex of participants. Additionally, the results of body composition assessment were incomplete (the quality of evidence was downgraded from "high" to "moderate"). c Includes two non-randomized studies [41,47]. d Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). e Inconsistencies have been reported in the unit used in the results (the quality of evidence was downgraded from "low" to "very low"). f Includes seven randomized controlled studies [30,31,33,34,42,44,55]. g Inconsistencies have been reported in the number of participants (the quality of evidence was downgraded from "high to "moderate"). h Several studies did not achieve statistical analyses (the quality of evidence was downgraded from "moderate" to "low"). i Includes six non-randomized controlled studies [31,43,46,51,53,57]. j Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). k One study did not achieve statistical analyses (the quality of evidence was already at "very low"). l Includes two non-randomized studies [41,47]. m Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). n One study did not achieve statistical analyses (the quality of evidence was already at "very low"). o Includes one randomized study [50]. p Includes four cross-over studies [32,49,52,54]. q Includes three randomized controlled studies [31,44,55]. r One study did not achieve statistical analyses (the quality of evidence was downgraded from "high" to "moderate"). s Includes six non-randomized controlled studies [31,43,46,51,53,57]. t Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). u One study did not achieve statistical analyses (the quality of evidence was already at "very low"). v Includes one randomized study [50]. w Includes four cross-over studies [32,49,52,54]. x Includes four randomized controlled studies [30,34,40,42]. y One study did not achieve statistical analyses (the quality of evidence was downgraded from "high" to "moderate"). z Includes one non-randomized non-controlled study [43]. aa Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). ab One study did not achieve statistical analyses (the quality of evidence was already at "very low"). ac Includes one non-randomized study [41]. ad Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). ae One study did not achieve statistical analyses (the quality of evidence was already at "very low"). af Includes four randomized controlled studies [33,34,44,59]. ag One study did not achieve statistical analyses (the quality of evidence was downgraded from "high" to "moderate"). ah Includes four non-randomized controlled studies [43,45,51,57]. ai Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). aj Includes 3 non-randomized studies [47,48,58]. ak Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). al Includes one randomized controlled study [44]. am One study did not detail the number of participants for this outcome and did not achieve statistical analyses (the quality of evidence was downgraded from "high" to "moderate"). an Includes two non-randomized non-controlled study [51,53]. ao Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). ap One study did not achieve statistical analyses (the quality of evidence was already at "very low"). aq Includes one non-randomized study [47]. ar Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). as Includes two cross-over studies [32,49]. at Includes three randomized controlled studies [30,44,55]. au Studies reported mixed findings (the quality of evidence was downgraded from "high" to "moderate"). av Inconsistencies have been reported in the unit used in the results (the quality of evidence was downgraded from "moderate" to "low"). aw Includes four non-randomized controlled studies [43,46,51,53]. ax Studies reported mixed findings (the quality of evidence was downgraded from "low" to "very low"). ay Includes two randomized controlled studies [34,44]. az One study did not achieve statistical analyses (the quality of evidence was downgraded from "high" to "moderate").

Description of Studies
The PRISMA flow diagram presented in Figure 1 summarizes the study selection process. The search strategy initially found a total of 1677 references after removing duplicates. A total of 1635 articles were excluded after screening of titles and abstracts. Full text copies were obtained for 44 articles; of which 25 articles matched the inclusion criteria and were thus included in this systematic review. The main reasons for studies exclusion among the remaining were: (1) study design did not meet inclusion criteria (n = 8); (2) intervention did not use an active desk (n = 5); (3) population was not children without health issues (n = 3); (4) full texts were not available (n = 2); and (5) active desks were already integrated in classroom (n = 1). One article included two different study designs [31].
All studies had an intervention duration from two weeks to two years. Verloigne et al. [55] suggested a rotation every half class while Clemes et al. [31] recommended to use active desks at least 30 min per day (Australian study) and one hour per day (English study). Some studies suggested also to practice active desks at least one hour per day [44,57] or for four class hours of 50 min per week [34]. Several studies did not indicate the active desks time and frequency use [47,49,51,58,61]. Some interventions enabled active desks to be free to use [30,[40][41][42][43][46][47][48]52,53] or to use it for the entire school day [32,33,45,59]. In one study, active desks were only used for the evaluations [58].

Body Composition
Six studies assessed body composition when using upright active desks [40][41][42]44,47,56] and one with cycling desks [34]. However, one study did not detail their results on this outcome [40] (Table 6). Wendel et al. [56] found a significant difference in BMI for interventional group compared to the control group after two years of intervention (−5.24 for BMI percentile) ( Table 3). Other studies did not report any change in BMI with the use of an upright active desk.
Torbeyns et al. [34] observed a significant effect of time for height, body weight, fat mass percentage and waist circumference without condition effect. However, traditional desks group reported a significantly higher BMI while cycling desks group did not find any difference.

Sedentary Behaviors
Thirteen articles using upright active desks assessed sedentary behaviors [31,32,43,44,46,[49][50][51][52][53][54][55]57], while only one used cycling desks [30]. As presented in Table 3, two studies observed that children, when using upright active desks, spent significantly less time sedentary than the control group, using objective measurements [46,54]. Other studies did not find any difference for the interventional group [49,53]. Moreover, Ee et al. [32] observed no significant difference for whole day sedentary time but reported a significant reduction in sitting time during school hours for the intervention group compared to the control group. Similar results have been reported in four other articles [31,44,52,53]. Additionally, four studies reported a reduction of sitting time between T0 and T1 for the intervention group [44,46,51,58]. Similar results have been found in another study but were not statistically significant [49,52].
Fedewa et al. [30] reported a decreased of 9.5% in sedentary time for the intervention group compared to the control group.
For light physical activity, four studies reported no significant changes for interventional group compared to the control group [32,50,52,59,61]. For MVPA, studies found contradictory results while two studies did not find any change [32,49]. Kidokoro et al. [46] observed a significant increase in MVPA for the intervention group between pre-and post-intervention. Another study [54] found that MVPA decreased for the intervention group during school years but less than the control group. Additionally, they reported that the benefit of upright active desk was greater among students initially determined as less active.
Regarding step counts, Benden et al. [41] reported an increase of this outcome without statistical analyses (Tables 3 and 6). In another study, they reported similar results at mid intervention but not at the end [42]. No significant effects were observed in two other studies [47,50]. In the article of Clemes et al. [31], the study in Australian school reported no significant effect while the British ones showed an increase for the intervention group in post intervention.
For stepping time a significant decrease was reported for the intervention group [55] or no effect [31,57]. One study [51] observed a significant increase while Clemes et al. [44] found similar results but no statistical analyses have been reported.
Torbeyns et al. [34] assessed the effect of cycling desks on physical activity with a questionnaire. Interventional group and control group decreased their physical activity time between pre-and post-intervention but no condition effect was observed. Despite the lack of statistical analyses (Tables 2 and 4), one study reported, with an objective measurement, an increase of light physical activity and MVPA for the intervention group compared to the control group [30].
One study using stability balls assessed physical activity and missed to observe any difference between the interventional group and the control group after the intervention [33]. Additionally, all groups decreased their physical activity level and their step count between pre-and post-intervention.

Energy Expenditure
Four studies assessed energy expenditure with the use of upright active desks [42,43,48,54] and two with cycling desks [30,34]. All upright active desks studies observed an increase between 15% and 25.7% in energy expenditure for interventional groups compared to control groups [42,43,48,54] (Table 2).
Cycling desk studies reported also an increase of energy expenditure. Torbeyns et al. [34] showed a significant increase in energy expenditure (36%) using cycling desks compared to traditional desks. Fedewa et al. [30] reported similar results without any statistical analyses (Tables 3 and 6).

Physical Capacities and Cardiometabolic Health
Physical capacities were only evaluated in one study that used cycling desks [34]. The authors reported an increase in the performance during the 20 m shuttle run test in their interventional group compared to the control group (+0.6 interval) (Table 3). Moreover, there was a significantly lower rate of perceived exertion (RPE) in the interventional group compared to the control group after 22 weeks. For cardiometabolic health, only Clemes et al. [44] assessed blood pressure with the use of an upright active desks. They reported an increase in systolic blood pressure in the interventional group but the authors did not perform statistical analyses (Tables 2 and 4).

Cognitive and Academic Performance
Seven studies assessed cognitive and academic performance when using upright active desks [44,48,49,51,53,55,56], two studies with cycling desks [34,58] and two with stability balls [33,59]. Concerning executive functions (working memory, inhibitor control, cognitive flexibility), visual working memory was assessed in two studies using upright active desk and two studies using cycling desks and no change was reported [34,53,56,57]. As detailed in Table 4, inhibitory control has been assessed in three studies, and the use of cycling desks shown to significantly increase the inhibitor control in the intervention group compared to the control group with an higher increase of accuracy for the intervention group (4.21%) [58]. One of the studies that used upright active desk reported an improvement in both reaction time and accuracy [57] while the other reported no significant change [48]. The reaction time for cognitive flexibility decreased after intervention in the study that used upright active desks [57] Regarding to academic engagement and attention, two studies using upright active desks reported an increase in the intervention group compared to the control group [43,45] without any change in concentration and classroom management [33,44,55]. A study using stability balls reported more interaction time with teachers but the time working with other students or independently were reduced compared to the control group after intervention. Both groups observed improvement in mathematics and literacy but they were not related specifically to the intervention [59]. Mehta et al. [48] assessed several outcomes where they primarily observed a significant increase in cognitive performance with the use of upright active desks compared to traditional ones.

Fatigue and Musculoskeletal Pain Symptoms
Six studies, all with upright active desks, assessed fatigue and musculoskeletal pain symptoms [32,44,51,52,55,59]. Three studies reported no difference on those outcomes between upright active desks and traditional desks [44,51,55]. Significant changes have been reported in two studies [32,49] with a decrease of pain symptoms in the neck and shoulder area. Nonetheless, a study observed that 51% of children have experienced pain in legs and back area with the use of upright active desks [53] (Table 4).

Process Evaluation
Acceptability and feasibility have been assessed in several studies [30,43,44,46,51,53,55]; one was cycling desks [30] and others were upright active desks. One study reported retention rates of 100% for schools and 97% for children with an overall recruitment rate at 33% [44] (Table 4). Studies have shown a good acceptability of upright active desks in children [48,50,51], with a willingness to use it in the future and a reduction of sleepiness [46]. From teachers' perspective, they have declared a positive influence of upright active desks to complete tasks and are willing to continue teaching with upright active desks [53]. One study reported that parents have felt a positive impact on their children's behavior at school [43]. However, one study [55] reported some negative effects with the use of upright active desks such as a slight deterioration of the relation with classmates. Authors also reported, a decrease of the mean duration and habit to use upright active desks over time. Most of those observations were reported in primary schools; secondary schools observed an improvement of the attitude towards the desk [55].
For cycling desks, authors [30] observed no change in attention and task completion compared to traditional desks. Students also experimented a reduction of fidgeting. Their preference to sit on cycling desks compared to traditional desks was higher despite the lack of a comfortable seat. Overall, cycling desks have been perceived by teachers and students as a positive tool to improve the environment of school class.
It was determined by the review team that a meta-analysis was not possible due to high levels of heterogeneity across studies; narrative syntheses were employed instead. The overall quality of the included studies was low due to methodological inconsistencies, in addition of the heterogeneity in terms of statistical and clinical characteristics (Tables 5 and 6).

Discussion
We are currently at a time where sedentary behaviors are a worldwide concern and classroom active desks have been proposed as a potential solution to counterbalance their adverse effects on health-related outcomes. Several reviews evaluated the effect of some specific types of active desks [35,36] on some specific outcomes such as academic achievement and cognitive outcomes [37]. The present work is the first systematic analysis of the existing literature on active desk implementation in the school environment and their effects on physical activity, sedentary behavior, academic achievements and overall health. According to our results, (i) cycling desk may be a promising active desk to increase physical activity while reducing sedentary behaviors; also, cycling desk is associated with positive cognitive performance and is well-received in the school environment; (ii) studies need to better identify and detail their active desks use; (iii) further studies have to use stronger methodologies to enable comparisons and conclusions regarding the real effects of each active desks.
Among all the included studies that assessed body composition, little or none effect was observed from the use of upright active desks or cycling desks. The only study that found positive changes in body composition was the study that lasted 2 years with upright active desks [56]. This suggests that the time of exposure to active desks can be an important parameter to consider. Additionally, the lack of observed effect on body composition in the reviewed studies can be potentially explained by the low level of energy expenditure generated by active desks. While active desks substantially increase students' energy expenditure compared to traditional desks [30,34,42,43,48,54], the magnitude of responses may not be sufficiently important to induce significant changes in body composition. However, it is important to notice that the range of increase in energy expenditure is not the same across active desks, with cycling desks generating a higher energy expenditure compared with upright active desks. According to our analysis, active desks also seem to positively influence sedentary behaviors. Indeed, by using upright active desks, students spend more time in a standing position and less time seated. Even though "standing" is not included in the definition of sedentary behavior [4], the energetic cost of this passive posture can be under 1.5 METs [62] and this long-term position can be a potential source of musculoskeletal pain [63]. From that perspective, replacing traditional desks by active desks (maybe not only standing), which increase energy expenditure, may be promising due to the replacement of a sitting time to an active behavior. Concerning cognitive and academic performance, all studies reported either no change or an improvement in students, leading to consider the non-deleterious impact of active desks on cognition. This finding is particularly relevant, as the implementation of active desks is clearly dependent on the willingness of the academic actors and parents. Beyond the cognitive aspect, active desks were well received by students and teachers in most studies, suggesting the possibility that active desks can be easily implemented in the school setting.

Methodological Concerns
The tremendous amount of sitting time spent in classrooms led scientists to examine how active desks for children and adolescents can be used to reduce sedentary behaviors. There has been a constant increase of studies focusing on this target in recent years. However, by systematically reviewing the current literature on the topic, we observed several methodological issues. The lack of strong and reliable results did not allow us to perform any meta-analysis to avoid misleading errors [64]. Not only the lack of methodological consistency between studies is concerning but also the relatively low quality of the included works (Tables 5 and 6) are certainly the main conclusions emerging from the present systematic review.
Indeed, although there is an increasing number of RCT on the topic in the scientific literature, only 10 of the 23 included studies were RCT in the present analysis. The heterogeneity of designs makes any comparison difficult regarding the potential benefits of active desks on health. Similarly, the variety of methods used to evaluate similar parameters (e.g., evaluation of physical activity and/or sedentary behaviors using accelerometers or inclinometers or questionnaires and expressed as counts, vector of magnitudes, activity or standing/sitting time for instance) prevent any strong collective evidence.
As previously suggested in the literature [65], studies should use parameters indicated in the last available recommendation to evaluate sedentary behaviors, sedentary time and physical activity. Moreover, studies assessing sedentary behaviors and physical activity should also consider the recording time for valid and reliable data. In other words, to observe and understand behavioral changes and compensatory effects, studies should not only record data during class time but rather on overall days, school and non-school day.
Importantly, while the recent years have seen a growing number of studies implementing active desks at school at a time where it was necessary to adopt new solutions and strategies to counteract the adverse effects of sitting time, these studies missed to clearly detail the exact conditions of use of their active desks. Indeed, as underlined in our systematic analysis, most of the included studies do not provide details regarding the time of use of their active desks, the instructions given to the teachers and pupils, which once more, make any practical recommendation hazardous. Characterization of the workload and details regarding the practical instructions should be a priority for investigators to understand at which frequency, intensity and duration active desks are driving benefits or adverse effects. In addition, it appears even difficult to clearly understand which kind of active desk has been used when reading some studies. Indeed, while some studies claim to use standing desks and formulate recommendations and conclusions regarding the use of the standing position at school, it appears that some stools are provided with each desk and that the exact time spent standing is not evaluated or even presented. Literally, it may be possible that users can sit or recline on the stool most of their class time. Considering this information, there is a clear risk of misunderstanding by using the term standing desk when it refers to a stand-biased desk. To avoid any misconception, the following definitions for those three active desks are proposed: (i) standing desk: desk which enables users to be in a standing position, without allowing any support to sit or recline; (ii) sit-to-stand desk: desk enabling users to switch from a sitting to a standing position at their discretion by adjusting the desk height; (iii) stand-biased desk: desk which enables users to be in a standing position while having a support such as a stool to sit or recline at their discretion.
Upright active desks represent a majority of the included study, while cycling desks and stability balls represent only five studies. Regarding results obtained in adults, cycling desk is suggested as the best compromise between all active desks [66] but not enough studies have assessed its effects in children and adolescents. While further studies are needed in the pediatric population, we also encourage future investigations to consider the effects of such active desks on physical discomfort, cognitive performance, physical capacities or physiological components that remain underexplored. Similarly, some ergonomic and process evaluations should be considered, which would benefit for a better use and implementation of these desks. Effectively, whether active desks are showing positive effects on several outcomes, one priority remains to understand if they are well accepted in school class by children, teachers and parents.
Furthermore, studies are essentially focusing on primary schools (17 of the 23 studies included). As previously said, children at 15 years spend on average 75% of their waking day in a sitting position [13]. Then, there is obviously a lack of active desks implementation in secondary school. From this perspective, scientists should also consider secondary levels. Therefore, to better understand the effect of active desks, further investigations should focus on large sample RCT follow-up in primary and secondary level (long-term follow-up), assessing multicomponent outcomes with valid, reproducible and reliable methods, while quantifying the workload. There is a need for a better description of the active desks use and condition of its use to avoid any misconception and inaccuracies. Additionally, scientists must consider the feasibility and the implementation of active desks in the school environment.

Conclusions
Active desks appear as a promising tool to reduce sedentary behaviors in school environment. In the present state of knowledge, the effects of all active desks appear not equivalent, mainly due to the difference in body activation and energy expenditure. Regarding the relatively low number of available studies and the high degree of heterogeneity in terms of quality, design and methods, comparisons and conclusions remain difficult at the moment. The present systematic analysis calls for further well-designed studies to better understand the effects of the use of active desks among children and adolescents in order to inform policy and practice.