- Open Access
Systematic review of the characteristics of school-based feasibility cluster randomised trials of interventions for improving the health of pupils in the UK
Pilot and Feasibility Studies volume 8, Article number: 132 (2022)
The last 20 years have seen a marked increase in the use of cluster randomised trials (CRTs) in schools to evaluate interventions for improving pupil health outcomes. Schools have limited resources and participating in full-scale trials can be challenging and costly, given their main purpose is education. Feasibility studies can be used to identify challenges with implementing interventions and delivering trials. This systematic review summarises methodological characteristics and objectives of school-based cluster randomised feasibility studies in the United Kingdom (UK).
We systematically searched MEDLINE from inception to 31 December 2020. Eligible papers were school-based feasibility CRTs that included health outcomes measured on pupils.
Of 3285 articles identified, 24 were included. School-based feasibility CRTs have been increasingly used in the UK since the first publication in 2008. Five (21%) studies provided justification for the use of the CRT design. Three (13%) studies provided details of a formal sample size calculation, with only one of these allowing for clustering. The median (IQR; range) recruited sample size was 7.5 (4.5 to 9; 2 to 37) schools and 274 (179 to 557; 29 to 1567) pupils. The most common feasibility objectives were to estimate the potential effectiveness of the intervention (n = 17; 71%), assess acceptability of the intervention (n = 16; 67%), and estimate the recruitment/retention rates (n = 15; 63%). Only one study was used to assess whether cluster randomisation was appropriate, and none of the studies that randomised clusters before recruiting pupils assessed the possibility of recruitment bias. Besides potential effectiveness, cost-effectiveness, and the intra-cluster correlation coefficient, no studies quantified the precision of the feasibility parameter estimates.
Feasibility CRTs are increasingly used in schools prior to definitive trials of interventions for improving health in pupils. The average sample size of studies included in this review would be large enough to estimate pupil-level feasibility parameters (e.g., percentage followed up) with reasonable precision. The review highlights the need for clearer sample size justification and better reporting of the precision with which feasibility parameters are estimated. Better use could be made of feasibility CRTs to assess challenges that are specific to the cluster design.
Cluster randomised trials (CRTs) are studies in which clusters (groups) of individuals are allocated to trial arms, and outcomes are measured on the individual participants . These clusters might be geographical locations (e.g., cities), organisations (e.g., workplaces) or social units (e.g., households). Clusters may be chosen as the randomisation unit for different reasons, including logistical reasons, to prevent contamination that could otherwise occur between trial arms if individuals were randomised, or because the intervention is designed to be administered at the cluster level . CRTs are often used to investigate complex interventions. They usually require more participants and can be more complicated to design, conduct and analyse than individually randomised controlled trials (RCTs) [1,2,3,4,5,6]. Therefore, it is important to assess the feasibility of the study processes and design uncertainties before a definitive CRT of intervention effectiveness is conducted.
Prior to a definitive trial, a feasibility study can be used to determine whether the research is something that can be done, whether it should be done and how it should be done . Feasibility studies focus on areas of uncertainty in trial delivery, such as the randomisation process, recruitment and follow-up rates, acceptability to the participants of the trial processes and the intervention itself, implementation of the intervention, data collection processes, selection of outcome measures, potential harms related to the intervention and trial, knowledge of parameters that will inform the sample size calculation for the definitive trial, and potential effectiveness of the intervention. The randomised pilot trial is a type of feasibility study that involves conducting the future definitive trial or part of it on a smaller scale . For ease of understanding, this paper refers to randomised pilot trials as feasibility studies. Other types of feasibility study include non-randomised parallel group and single-arm trials, which also focus on developing trial methodology and interventions, and testing processes prior to a full-scale RCT [7, 8]. However, such designs cannot be used to test specific uncertainties such as the randomisation process and the willingness of participants to be randomised. Feasibility CRTs differ from those done in advance of individually RCTs in that they may be used to address concerns that are specific to CRTs, including evaluating the possibility for recruitment bias in studies where clusters are randomised before individual participants are recruited  and obtaining estimates of the intra-cluster correlation coefficient (ICC) of the primary outcome to support the calculation of the sample size for the definitive trial, although some authors caution that the resulting estimates will often be imprecise due to the small number of clusters typically included in such studies . Other general feasibility considerations apply at both the cluster and individual levels, such as ease of recruitment, rate of loss to follow-up and acceptability of the intervention. Methodological considerations that are unique to the conduct of feasibility CRTs include the need to take account for clustering when calculating the sample size for and reporting the precision of feasibility parameter estimates from such studies .
In recent years, CRTs have been increasingly used to evaluate interventions for improving educational outcomes in schools  and complex interventions for improving child health outcomes [12,13,14]. Schools provide a natural environment in which to recruit and deliver public health interventions to children due to the amount of time they spend there . The CRT design is suited to the natural clustered structure found in schools (pupils within classes within schools), but there are challenges to delivering trials in this setting. For example, schools and teachers often have stretched and limited resources, and implementing an intervention and participating in a trial can be challenging, given that the primary focus of schools is the education of pupils. A recent systematic review of definitive school-based CRTs found that 52% of the studies required a member of school staff to deliver components of the intervention . Obtaining a representative sample of schools is important for external validity and inclusiveness , but recruitment of schools and pupils is also a challenge. Another potential feasibility issue regards which type of cluster to randomise in the school setting for a given trial, such as entire schools, year groups, classrooms or teachers. For example, there may be a choice between randomising schools and randomising classrooms; the former would be better to minimise the chance of contamination between trial arms but the latter would have the advantage of a smaller design effect  and, therefore, greater power for a fixed total number of recruited pupils . In comparison to the primary care setting, CRTs for evaluating health interventions have only relatively recently been used in schools in the UK and, therefore, there is a smaller pool of experience available from previous studies [1, 14]. Given these uncertainties, feasibility trials have an important role to play in the design and execution of definitive school-based CRTs.
Authors have previously discussed the growing literature described as ‘feasibility’ or ‘pilot’ studies, and the associated methodological challenges . The characteristics of feasibility studies generally [10, 16, 17] and cluster randomised feasibility studies specifically [18, 19] have been summarised, but, to date, no systematic review has focussed on the characteristics of school-based feasibility CRTs for improving pupil health outcomes. The aim of this systematic review is to summarise the key design features and report the feasibility-related objectives of school-based feasibility CRTs in the United Kingdom (UK) that measure health outcomes on pupils. It follows our previous systematic review of full-scale definitive CRTs in the school setting . Through summarising the design features of these studies, the findings of this review will highlight particular areas where improvements could be made to the conduct of feasibility CRTs. The reporting of their feasibility objectives will help identify areas in which better use of such studies could be made to address uncertainties that are specific to the CRT design.
Data sources and search methods
This review has been reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement  as evidenced in the PRISMA checklist (see Additional file 1: Table S1) and was registered with PROSPERO (ID CRD: 42,020,218,993; www.crd.york.ac.uk/prospero).
Peer-reviewed school-based feasibility CRTs, indexed on MEDLINE (through Ovid), were the source of data for the review. MEDLINE was systematically searched from inception to 31 December 2020. A pragmatic decision was made to search MEDLINE only due to time constraints and available resources. The search strategy (Table 1) was developed using terms from the MEDLINE search strategy by Taljaard et al.  to identify CRTs, and this was combined with school concept terms, including the ‘Schools’ MeSH term. This was the same search strategy used in our previous systematic review to identify definitive school-based CRTs . The search was limited to English language papers.
Inclusion and exclusion criteria
The review included school-based feasibility CRTs that measured health outcomes on pupils and were conducted in the UK. It focussed on the UK to align with available resources and to summarise data from a single education system relevant to the research team. The population of included studies was pupils attending pre-school, primary or secondary school in the UK. ‘Pre-school’ was defined as an organisation offering early childhood education to children before they begin compulsory education (i.e., primary school). This included nursery schools and kindergartens. Eligible clusters could be any school-related unit (e.g., schools, classes, year groups). Studies that randomised school-related units as well as other types of clusters (e.g., towns, hospitals, households) were eligible for inclusion in the review as long as the results of the study were shown separately for the school clusters (i.e., the authors did not pool results across the different types of clusters). Any health-related intervention(s) were eligible. The primary outcome had to be measured on pupils and be health related. Studies with education-related primary outcomes were excluded. All types of CRT design were eligible, including parallel group, factorial, crossover and stepped wedge trials.
Only randomised external feasibility studies were included in the systematic review. The definition of feasibility study used to identify eligible papers was that used by Eldridge et al.  which states “A feasibility study asks whether something can be done, should we proceed with it, and if so, how.” Therefore, eligible studies had to be assessing some element of feasibility in the intervention and/or trial methodology, ahead of a definitive trial. This was determined by looking for the terms, ‘pilot’, ‘feasibility’ or ‘explanatory’ in the title and abstract and by examining the aims and objectives of each study. Internal pilot studies that are part of the actual definitive trial, where the data from the pilot phase are included in the main analysis  were excluded. Non-randomised parallel group feasibility studies and single-arm feasibility studies were excluded. Definitive CRTs were not eligible for inclusion in this review.
If there was more than one publication of the results for an eligible feasibility CRT, the paper presenting quantitative results related to the feasibility objectives was designated the key study report (index paper) and used for data extraction. Papers that did not report the results of the feasibility objectives were excluded along with protocol/design articles, cost-effectiveness/economic evaluations and process evaluations.
Sifting and validation
Titles and abstracts were downloaded into Endnote  and screened by two independent reviewers (KP & SEd/OU) for eligibility against inclusion criteria. Studies for which inclusion status was uncertain were included for full-text screening. Full-text articles were assessed against inclusion criteria by two reviewers (KP & SEd) using a pre-piloted coding method. Any uncertainties were resolved by consulting a third reviewer (OU).
The data extraction form was pre-piloted in Microsoft Excel by KP and SEd. One investigator (KP) extracted data from all included studies. A second reviewer (SEd or OU) independently extracted data for validation. If there was uncertainty regarding a particular article, the data obtained were checked by another member of the team (MN) and resolved by further discussion.
The items of information extracted are listed as follows:
Publication details: year of publication, journal name, funding source and trial registration status.
Setting characteristics: country (England, Scotland, Wales, Northern Ireland) in which the trial took place, school level, types of school recruited and participant information.
Intervention information: health area, intervention description and type of control arm.
Primary outcome information: name of primary outcome.
Study design: justification for using cluster trial design, type of cluster, method of randomisation, timing of randomisation of clusters relative to recruitment of pupils, number of trial arms, allocation ratio and length of follow-up.
Sample size information: justification for sample size, targeted number of schools, clusters and pupils; number of recruited schools, clusters and pupils.
Objectives of feasibility study: test randomisation process (yes/no), test willingness to be randomised (at cluster and/or individual levels) (yes/no), estimate recruitment rate (at cluster and/or individual levels) (yes/no), estimate retention/follow-up rate (at cluster and/or individual levels) (yes/no), test implementation of the intervention (yes/no), test compliance with the intervention (yes/no), assess acceptability of the intervention (at cluster and/or individual levels) (yes/no), assess acceptability of trial procedures (at cluster and/or individual levels) (yes/no), test the feasibility of blinding procedures (yes/no), test data collection process (yes/no), test outcome measures (yes/no), estimate standard deviation for continuous outcomes (or control arm rate for binary outcomes) (yes/no), test consent procedures (yes/no), identify potential harms (yes/no), estimate potential effectiveness of intervention (yes/no), estimate costs of delivering the intervention (yes/no), estimate the intra-cluster correlation coefficient (ICC) of the primary outcome (yes/no) and calculate the sample size required for the definitive trial (yes/no). Only formal feasibility objectives were extracted; these were obtained from the Background and Methods sections of the included articles.
Ethics and consent procedures: Was ethical approval provided? (yes/no).
Other design characteristics of methodological interest: analysis method used to estimate potential effectiveness of the intervention, baseline cluster-level characteristics, ICC estimates (and 95% confidence intervals (CIs)) and whether study concluded that a definitive trial is feasible (yes/yes (with modifications)/no).
Study characteristics were described using medians, interquartile ranges (IQRs) and ranges for continuous variables, and numbers and percentages for categorical variables, using Stata 17 software . Formal quality assessment of the papers was not performed as it was not necessary for summarising characteristics of studies. However, some of the data extracted and summarised in the review are indicative of the reporting quality of included studies based on the items in the CONSORT extension for both CRTs  and pilot studies . This includes details on the rationale for using the CRT design, the rationale for the target sample size and ethical approval procedures.
After deduplication, 3247 articles were identified through MEDLINE, 62 were eligible for full-text screening and 24 were included in the review [27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50]. Out of 38 excluded studies, 28 did not meet the inclusion criteria, and 10 met inclusion criteria but were excluded as they described the same study as a designated ‘index paper’. The PRISMA flow diagram  is shown in Fig. 1.
School-based feasibility CRTs for health interventions on pupils have been increasingly used in the UK since the first publication in 2008 (Fig. 2). Included articles were published across 11 different journals: Pilot and Feasibility Studies (n = 5), International Journal of Behavioural Nutrition and Physical Activity (n = 4), Public Health Research (n = 4), BMJ Open (n = 3), Health Technology Assessment (n = 2), Archives of Disease in Childhood (n = 1), BMC Public Health (n = 1), British Journal of Cancer (n = 1), British Journal of Psychiatry (n = 1), Prevention Science (n = 1) and Trials (n = 1). Ten articles described their study as a ‘pilot trial’, six as a ‘feasibility trial’, four as a ‘feasibility study’, two as an ‘exploratory trial’, one as a ‘pilot feasibility trial’ and one as a ‘pilot study’. Twelve (50%) studies were funded by the National Institute for Health Research. Eight (33%) studies were registered prospectively, thirteen (54%) retrospectively, and three (13%) did not state registration status.
Tables 2 and 3 summarise the characteristics of included studies.
Three quarters of studies (n = 18; 75%) took place in England. Just over half (n = 13; 54%) took place exclusively in secondary schools, 8 (33%) took place exclusively in primary schools, 2 (8%) exclusively in pre-schools and 1 (4%) study included both primary and secondary schools. Fifteen (63%) studies provided information about the types of schools included in their sample and, of these, 14 (93%) included “state” schools.
Intervention and control type
Eleven (46%) studies delivered interventions for improving physical activity, 4 (17%) in physical activity and nutrition, 2 (8%) in alcohol misuse, 2 (8%) in sexual health and 1 (4%) in each of illicit drug misuse, bullying, behavioural/social difficulties, body image, and dating and relationship violence.
The main types of intervention components included resources and materials for schools (n = 11; 46%), classroom lessons (n = 10; 42%) and physical activity lessons (n = 5; 21%). Almost all studies (n = 23, 96%) had intervention components that had to be delivered to entire clusters (‘cluster–cluster’ interventions  (pages 25 to 30))—e.g., classroom-delivered lessons  and physical activity sessions . Two (8%) had intervention components that were directed at individual pupils (‘individual-cluster’ interventions )—e.g., goal-setting [40, 50]. Eighteen (75%) had intervention components that were delivered by a professional or person internal to the cluster (‘professional-cluster’ interventions )—e.g., teachers , member of school staff  and fellow pupils/peers . Eight studies (33%) had intervention components that were delivered by someone external to the cluster (‘external-cluster’ interventions )—e.g., ‘active play practitioners’ , researchers  and dance teachers .
The most common type of control arm was usual care (n = 21; 88%). Two (8%) studies used an active control arm, and one (4%) study had two control arms (a usual care arm and an active control arm).
Justification for the use of the CRT design was provided in only 5 (21%) studies. The reasons given were that the intervention was designed to be delivered to entire clusters [30, 47, 48] and to minimise contamination between trial arms [44, 50]. Twenty-three (96%) studies randomised schools and the remaining study randomised classrooms . In the latter study , random allocation was carried out at the level of the classroom for ‘pragmatic considerations’. Thirteen (54%) studies used some form of restricted allocation to balance cluster characteristics between the trial arms.
Most studies (n = 21; 88%) had two trial arms and most allocated clusters in a 1:1 ratio (n = 17; 71%). The median (IQR; range) length of follow up was 7 (3 to 12; 2 to 24) months.
Twelve (50%) studies recruited pupils before randomisation of clusters, four (17%) randomised clusters before recruiting pupils, and in eight (33%) studies, it was unclear whether or not randomisation occurred before pupils were recruited. Only 13 (54%) studies reported baseline characteristics of the schools.
Ethical approval was obtained and reported in 22 (92%) studies. One study stated that ethical approval was sought but the local research committee said it was not required as the study did not involve patients or NHS staff. The remaining study did not state whether ethical approval was obtained.
Of the 24 studies included in this review, three (13%) provided details of a formal sample size calculation. One of these studies based their sample size on being able to estimate feasibility parameters (e.g., participation rates, questionnaire response rates) with a specified level of precision , and the other two studies based their sample size on power to detect a specified intervention effect [29, 48]. Only one (4%) study allowed for clustering in their sample size calculation . Nineteen studies provided informal justification for their sample size calculation, based on one or more reasons: seven (29%) studies based their target sample size on recommendations from previous articles, six (25%) studies stated that a formal sample size calculation was not needed, four (17%) studies said their target sample size was determined by resource and/or time constraints, three (13%) studies provided a general statement that their sample size was considered sufficient to address the objectives of the feasibility CRT, and one (4%) study aimed to recruit as many clusters and participants as possible. Two (8%) studies did not provide any justification for their choice of sample size.
The median (IQR; range) target sample size was 7.5 (5 to 8; 2 to 20) schools, 7.5 (5 to 8; 2 to 20) clusters and 320 (150 to 1200; 50 to 1852) pupils. The median (IQR; range) achieved sample size was 7.5 (4.5 to 9; 2 to 37) schools, 8 (5.5 to 9.5; 2 to 37) clusters and 274 (179 to 557; 29 to 1567) pupils. Two studies included just 2 schools, with 1 school allocated to each trial arm [34, 35]. The studies that reported both targeted and achieved recruitment numbers at the cluster (n = 18) and pupil (n = 13) levels achieved those targets in 94% and 46% of studies, respectively.
Objectives of feasibility study
Formal feasibility objectives were specified by all 24 studies (summarised in Table 3). Of the 18 objectives assessed in this review, the median (IQR; range) number addressed per study was 5 (4 to 7.5; 1 to 11). The most common objectives were to estimate the potential effectiveness of the intervention (n = 17; 71%; including two studies that sought to undertake a definitive test of effectiveness [29, 48]), assess acceptability of the intervention (n = 16; 67%), estimate the recruitment rate (n = 15; 63%), estimate the retention/follow-up rate (n = 15; 63%) and test outcome measures (n = 14; 58%). Two studies included estimation of the intra-cluster correlation coefficient of the primary outcome to be used in the planned definitive study as a formal objective of the feasibility study. No studies tested the feasibility of blinding or consent procedures. All studies reported additional feasibility outcomes beyond those formally stated as objectives.
The following feasibility objectives were stated specifically at the level of the cluster: assess acceptability of the intervention (n = 10; 42%), estimate retention/follow-up rate (n = 7; 29%), estimate recruitment rate (n = 6; 25%), assess willingness to be randomised (n = 4; 17%) and assess acceptability of the trial procedures (n = 3; 13%). One (4%) feasibility CRT had the formal objective of assessing the appropriateness of cluster randomisation . None of the feasibility studies used their research to assess the type of cluster that should be randomised. Of the 4 studies that randomised clusters before recruiting pupils, none investigated the possibility of recruitment bias.
Analyses were undertaken to investigate if the target sample size differed according to whether or not the studies addressed specific feasibility objectives. Many objectives were only formally stated in a small number of studies; therefore, it was hard to identify clear patterns in the data. The twelve studies that assessed potential effectiveness aimed to recruit a median (IQR; range) of 7 (3.5 to 8; 2 to 20) schools, similar to the targeted recruitment in the remaining studies (7.5 (6 to 8; 5 to 12)).
All studies reported estimates of feasibility parameters, but, other than for potential intervention effectiveness, cost-effectiveness and the intra-cluster correlation coefficient, no studies quantified the precision of these estimates. Five of the eight (63%) studies that reported estimates of the ICC for the provisional primary outcome of the planned definitive study provided 95% confidence intervals (95 CIs) for these. Table 4 reports the ICC estimates. As expected the 95% confidence intervals were generally wide given that the sample size is small for estimating the ICC. Notably, however, the upper bound for two ICC estimates was only 0.03, which provides useful information on plausible true values of the parameter despite those studies having only 6  and 19  clusters.
Of the 20 studies that reported intervention effect estimates, nine (45%) used an adjusted individual-level analysis method to allow for clustering, 4 (20%) used a cluster-level analysis method, four (20%) did not allow for clustering and three (15%) did not state the analytical method. Eight studies reported p values with the intervention effect estimate, contrary to published guidance for feasibility studies [25, 26].
Eleven (46%) studies concluded that the definitive trial was feasible, 11 (46%) said the definitive trial would be feasible with modifications and two (8%) said that the planned study was not feasible. Through searching the literature and personal correspondence with the authors, it was established that of the 24 feasibility CRTs included in the review, 11 are known to have progressed to definitive trials [28, 29, 31, 36, 39,40,41, 44, 46, 49, 50]. Of these, nine had concluded that the definitive trial was feasible, and two had concluded that the definitive trial would be feasible with modifications.
This is the first systematic review to summarise the characteristics and objectives of school-based feasibility CRTs of interventions to improve pupil health outcomes in the UK. The review found an increase in such studies since the earliest included paper was published in 2008. This mirrors the increase in definitive CRTs in this area reported in our parallel review  and highlights the rising popularity of health-based CRTs in the school-setting. The increase in feasibility CRTs may partly be due to the publication of the 2006 MRC guidelines for the evaluation of complex interventions  which highlights the importance of conducting feasibility studies ahead of full-scale trials. The relatively large number of feasibility CRTs with interventions for increasing physical activity indicates the growing importance of adolescent physical activity as a public health priority, and the use of schools in order to deliver these types of intervention . The review of school-based definitive CRTs also reflected the increasing use of the design to evaluate physical activity interventions . Based on what was observed in the review of definitive school-based CRTs, there were fewer than expected feasibility studies in the area of socioemotional functioning. This is despite the increased awareness of the prevalence of these health conditions and research funding in this area .
A previous review of feasibility CRTs found that, among other objectives, assessing the implementation of the intervention (n = 9, 50%) was the most common . The studies included in the current review sought to address a range of feasibility objectives; most commonly estimating potential effectiveness of the intervention, assessing acceptability of the intervention, estimating the recruitment and follow-up rates and testing the outcome measures. It was notable, however, that few studies formally stated objectives that were related to uncertainties that are unique to the cluster design. This finding is similar to another review of feasibility CRTs which also stated that few studies investigated issues specific to the complexities of the design . None of the 4 studies that randomised clusters before recruiting pupils investigated the potential for recruitment bias as a feasibility objective. In the current review, only one study assessed whether a cluster design was needed, and none used the research to decide on the type of school-based cluster (e.g., school versus classroom) that was best to randomise. It may be the case that the need for cluster randomisation and the appropriate type of cluster to allocate had a strong theoretical basis, negating the need for empirical justification, but only 5 of the 24 studies provided a rationale for the cluster design even though the CONSORT extension for CRTs  recommends reporting this.
The studies included in this review were heterogeneous in their formal feasibility objectives, and this may have influenced specific features of their design, such as sample size and length of follow-up. The designs may also have been influenced by other factors such as budget, time and practical constraints.
Only three (13%) studies in the review reported details of a formal calculation for the target sample size [29, 33, 48], and only one accounted for clustering in the sample size calculation . These results are similar to that found in a previous systematic review of feasibility CRTs which reported that only one of the 18 studies reported a formal sample size calculation based on the primary feasibility objective . A quarter of the included papers in the current review stated that a formal sample size calculation was not needed, and some authors have argued that it is not always appropriate in feasibility studies . In a recent review of current practice in feasibility studies, only 36% reported sample size calculations . Also, when surveyed, some journal editors stated they were willing to accept pilot studies for publication that did not report a sample size calculation . The precision with which parameters are estimated in feasibility CRTs should be reported, especially given the small number of clusters that are typically included in such studies. Despite this, apart from when assessing the effectiveness of the intervention, cost-effectiveness and estimating the ICC, this was not done by any papers in the current review. Correspondingly, a formal sample size calculation based on the feasibility objectives that allows for clustering  is appropriate to estimate parameters precisely and, therefore, minimise the uncertainty regarding the assumptions that are made for the subsequent definitive study [16, 57].
Our review found the median number of clusters recruited (eight) was similar to a previous review of feasibility studies . Based on results from a simulation study, it has been suggested that as many as 30 or more clusters may be required in a feasibility CRT in order to avoid downwardly biased and imprecise estimates of the number of clusters required to test the intervention effect in the subsequent definitive CRT; this is largely due to the imprecision with the ICC is estimated in the feasibility study . The current review found only one study that recruited more than 30 clusters , and it is difficult to achieve this level of recruitment due to funding and practical constraints. Smaller feasibility studies may, however, still provide informative estimates of many parameters. Two of the feasibility studies in the review, despite including only 6  and 19  clusters, were able to estimate the intra-cluster correlation coefficient with a 95% confidence interval upper bound of 0.03, which could rule out the need for unattainably large sample sizes in the definitive study. Many studies report feasibility objectives in the form of percentages (e.g., follow-up rates, intervention adherence rates). Eldridge and colleagues  provide formulae for calculating the sample size required in feasibility CRTs to estimate percentages based on individual-level characteristics (e.g., whether the pupil was followed up) with a confidence interval of specified width, whilst allowing for clustering. Assuming the ICC for the feasibility characteristic is 0.05, a study with 8 schools and 240 pupils (an average sample size based on the findings in the current review) is large enough to estimate the percentage with a margin of error no greater than 10 percentage points based on a 95% confidence interval. There will generally be little precision for estimating percentages based on cluster-level characteristics since this is determined by the, typically, small number of schools (clusters) in feasibility studies.
Another important reason to recruit sufficient clusters to feasibility CRTs is to assess how the intervention might be implemented and the trial delivered in a range of different types of cluster . Parameter estimates will only be useful to the extent that the clusters and individuals in the feasibility study are broadly representative and reflect the diversity of the population from which the sample in the definitive trial will be drawn . In the context of school-based trials, important aspects of representativeness include single sex versus co-educational schools, state versus independent schools, and deprived versus non-deprived areas. In the current review, only 54% of studies reported baseline characteristics of the schools, although this is higher than found in a previous systematic review of feasibility CRTs where only 11% of studies reported baseline cluster-level characteristics .
The current systematic review found that of the 13 studies that reported both targeted and achieved numbers of pupils recruited, those targets were only achieved in 46% of studies. Our previous systematic review of definitive school-based CRTs found that only 77% of studies achieved their target recruitment of pupils . The facilitators and barriers to the recruitment and retention of pupils to school-based CRTs have been discussed in detail in the literature [58,59,60], including the type of intervention being offered and the perceived benefits of the study (e.g., sexual education) [58, 60], lack of time , incompatibility of the intervention with the needs of pupils or parents or with the school’s ethos  and a lack of incentivisation .
Strengths and limitations
A strength of the review is that a predefined search strategy was used to identify feasibility cluster randomised trials in the school setting. The protocol was publicly available prior to conducting the review. Screening, piloting of the data extraction form and data extraction were conducted by two independent reviewers. A pragmatic decision was made to limit the review to the UK in order to align with available resources and to make it more focused.
A limitation is the decision to use only the MEDLINE database. MEDLINE was chosen as health-based studies were the focus of this review. We acknowledge that further articles may have been found by searching other databases, grey literature and through citation searching. The search strategy was translated in EMBASE, DARE, PsycINFO and ERIC databases to search for additional eligible school-based CRTs published between 2017 and 2020 and resulted in identification of only one further unique eligible article. Therefore, we feel the pragmatic approach to only use MEDLINE to perform this search did not result in omission of a significant body of relevant evidence.
The systematic review only included feasibility studies that used the cluster randomised trial design and not other types, such as non-randomised parallel group and single-arm feasibility studies. We focussed on CRTs because we were interested in studies that could be used to assess a wide range of uncertainties for definitive CRTs, but we acknowledge that the systematic review may, therefore, not include some relevant knowledge of practice in non-randomised feasibility studies. While the approach used was not comprehensive, it enabled us to efficiently identify studies of interest that were undertaken in advance of planned definitive CRTs.
A further limitation of the review is that data were not extracted on consent procedures used by the included studies. As found in our previous review of definitive school-based CRTs , this information was inconsistently reported across studies making it challenging to summarise. This highlights the need for more comprehensive reporting of the consent procedures in these studies.
Cluster randomised feasibility studies are increasingly used in the school setting to test feasibility prior to definitive trials. Although these studies usually include few schools, the average sample size of those included in this review would be large enough to estimate percentages based on pupil characteristics that are used to address feasibility objectives (e.g., the percentage followed up) with a reasonable level of precision. The review has highlighted the need for clearer justification for the target sample size of school-based feasibility CRTs and to report the precision with which feasibility parameters are estimated in these studies. The characteristics of the recruited schools in feasibility CRTs could be better described to help understand the extent to which the feasibility parameter estimates are applicable to the planned definitive trial and other future similar trials. Finally, better use could be made of feasibility CRTs in the area of school-based pupil health research to assess challenges that are specific to the cluster trial design.
Availability of data and materials
The datasets generated and/or analysed during the current study are not publicly available because they are also being used for a wider ongoing programme of research but are available from the corresponding author on reasonable request.
Body mass index
Cluster randomised trial
Database of Abstracts of Reviews of Effects
Excerpta Medica Database
Education Resources Information Center
Intra-cluster correlation coefficient
Medical Literature Analysis and Retrieval System Online
Medical Subject Headings
Moderate to vigorous physical activity
Preferred Reporting Items for Systematic Reviews and Meta-Analyses
Psychological Information Database
Eldridge S, Kerry S. A practical guide to cluster randomised trials in health services research. Chichester: Wiley; 2012.
Donner A, Klar N. Design and analysis of cluster randomization trials in health research. Chichester: Wiley; 2000.
Cornfield J. Randomization by group: a formal analysis. Am J Epidemiol. 1978;108(2):100–2.
Campbell M, Walters S. How to design, analyse and report cluster randomised trials in medicine and health related research. Chichester: Wiley; 2014.
Hayes R, Moulton L. Cluster randomised trials. Florida: CRC Press; 2009.
Murray D. Design and anaylsis of group-randomized trials. New York: Oxford University Press; 1998.
Eldridge SM, Lancaster GA, Campbell MJ, Thabane L, Hopewell S, Coleman CL, et al. Defining feasibility and pilot studies in preparation for randomised controlled trials: development of a conceptual framework. PLoS ONE. 2016;11(3): e0150205.
Lancaster GA, Thabane L. Guidelines for reporting non-randomised pilot and feasibility studies. Pilot and Feasibility Studies. 2019;5(1):114.
Eldridge S, Kerry S, Torgerson DJ. Bias in identifying and recruiting participants in cluster randomised trials: what can be done? BMJ. 2009;339: b4006.
Eldridge SM, Costelloe CE, Kahan BC, Lancaster GA, Kerry SM. How big should the pilot study for my cluster randomised trial be? Stat Methods Med Res. 2016;25(3):1039–56.
Spybrook J, Zhang Q, Kelcey B, Dong N. Learning from cluster randomized trials in education: an assessment of the capacity of studies to determine what works, for whom, and under what conditions. Educ Eval Policy Anal. 2020;42(3):354–74.
Walleser S, Hill SR, Bero LA. Characteristics and quality of reporting of cluster randomized trials in children: reporting needs improvement. J Clin Epidemiol. 2011;64(12):1331–40.
Goesling B. A practical guide to cluster randomized trials in school health research. J Sch Health. 2019;89(11):916–25.
Parker K, Nunns M, Xiao Z, Ford T, Ukoumunne OC. Characteristics and practices of school-based cluster randomised controlled trials for improving health outcomes in pupils in the United Kingdom: a methodological systematic review. BMC Med Res Methodol. 2021;21(1):152.
Fazzari MJ, Kim MY, Heo M. Sample size determination for three-level randomized clinical trials with randomization at the first or second level. J Biopharm Stat. 2014;24(3):579–99.
Billingham SA, Whitehead AL, Julious SA. An audit of sample sizes for pilot and feasibility trials being undertaken in the United Kingdom registered in the United Kingdom Clinical Research Network database. BMC Med Res Methodol. 2013;13(1):1–6.
Thabane L, Ma J, Chu R, Cheng J, Ismaila A, Rios LP, et al. A tutorial on pilot studies: the what, why and how. BMC Med Res Methodol. 2010;10(1):1–10.
Chan CL, Leyrat C, Eldridge SM. Quality of reporting of pilot and feasibility cluster randomised trials: a systematic review. BMJ Open. 2017;7(11): e016970.
Kristunas CA, Hemming K, Eborall H, Eldridge S, Gray LJ. The current use of feasibility studies in the assessment of feasibility for stepped-wedge cluster randomised trials: a systematic review. BMC Med Res Methodol. 2019;19(1):12.
Moher D, Liberati A, Tetzlaff J, Altman DG, Gro P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6(7):e1000097.
Taljaard M, McGowan J, Grimshaw JM, Brehaut JC, McRae A, Eccles MP, et al. Electronic search strategies to identify reports of cluster randomized trials in MEDLINE: low precision will improve with adherence to reporting standards. BMC Med Res Methodol. 2010;10(1):1–8.
Lancaster GA, Dodd S, Williamson PR. Design and analysis of pilot studies: recommendations for good practice. J Eval Clin Pract. 2004;10(2):307–12.
The EndNote Team. EndNote. EndNote X9 version ed. Philadelphia: Clarivate; 2013.
StataCorp. Release 17. College Station: StataCorp LLC; 2021.
Campbell MK, Piaggio G, Elbourne DR, Altman DG. Consort 2010 statement: extension to cluster randomised trials. BMJ. 2012;345: e5661.
Thabane L, Hopewell S, Lancaster GA, Bond CM, Coleman CL, Campbell MJ, et al. Methods and processes for development of a CONSORT extension for reporting pilot randomized controlled trials. Pilot Feasibility Stud. 2016;2:25.
Barber SE, Jackson C, Hewitt C, Ainsworth HR, Buckley H, Akhtar S, et al. Assessing the feasibility of evaluating and delivering a physical activity intervention for pre-school children: a pilot randomised controlled trial. Pilot Feasibility Stud. 2016;2(1):12.
Bonell C, Fletcher A, Fitzgerald-Yau N, Hale D, Allen E, Elbourne D, et al. Initiating change locally in bullying and aggression through the school environment (INCLUSIVE): a pilot randomised controlled trial. Health Technol Assess. 2015;19(53).
Carlin A, Murphy MH, Nevill A, Gallagher AM. Effects of a peer-led Walking In ScHools intervention (the WISH study) on physical activity levels of adolescent girls: a cluster randomised pilot study. Trials. 2018;19(1):31.
Clemes SA, Bingham DD, Pearson N, Chen Y-L, Edwardson CL, McEachan RRC, et al. Stand out in class: restructuring the classroom environment to reduce sitting time – findings from a pilot cluster randomised controlled trial. Int J Behav Nutr Phys Act. 2020;17(1):55.
Corder K, Brown HE, Schiff A, van Sluijs EMF. Feasibility study and pilot cluster-randomised controlled trial of the GoActive intervention aiming to promote physical activity among adolescents: outcomes and lessons learnt. BMJ Open. 2016;6(11): e012335.
Corepal R, Best P, O’Neill R, Kee F, Badham J, Dunne L, et al. A feasibility study of ‘The StepSmart Challenge’ to promote physical activity in adolescents. Pilot Feasibility Stud. 2019;5(1):132.
Forster AS, Cornelius V, Rockliffe L, Marlow LA, Bedford H, Waller J. A cluster randomised feasibility study of an adolescent incentive intervention to increase uptake of HPV vaccination. Br J Cancer. 2017;117(8):1121–7.
Gammon C, Morton K, Atkin A, Corder K, Daly-Smith A, Quarmby T, et al. Introducing physically active lessons in UK secondary schools: feasibility study and pilot cluster-randomised controlled trial. BMJ Open. 2019;9(5): e025080.
Ginja S, Arnott B, Araujo-Soares V, Namdeo A, McColl E. Feasibility of an incentive scheme to promote active travel to school: a pilot cluster randomised trial. Pilot and Feasibility Studies. 2017;3(1):57.
Jago R, Sebire SJ, Cooper AR, Haase AM, Powell J, Davis L, et al. Bristol girls dance project feasibility trial: outcome and process evaluation results. Int J Behav Nutr Phys Act. 2012;9(1):83.
Jago R, Sebire SJ, Davies B, Wood L, Edwards MJ, Banfield K, et al. Randomised feasibility trial of a teaching assistant led extracurricular physical activity intervention for 9 to 11 year olds: Action 3:30. Int J Behav Nutr Phys Act. 2014;11:114.
Johnstone A, Hughes AR, Bonnar L, Booth JN, Reilly JJ. An active play intervention to improve physical activity and fundamental movement skills in children of low socio-economic status: feasibility cluster randomised controlled trial. Pilot and Feasibility Studies. 2019;5(1):45.
Kipping RR, Payne C, Lawlor DA. Randomised controlled trial adapting US school obesity prevention to England. Arch Dis Child. 2008;93(6):469–73.
Lloyd JJ, Wyatt KM, Creanor S. Behavioural and weight status outcomes from an exploratory trial of the Healthy Lifestyles Programme (HeLP): a novel school-based obesity prevention programme. BMJ Open. 2012;2(3): e000390.
Lohan M, Aventin Á, Clarke M, Curran RM, McDowell C, Agus A, et al. Can Teenage men be targeted to prevent teenage pregnancy? A feasibility cluster randomised controlled intervention trial in schools. Prev Sci. 2018;19(8):1079–90.
McSweeney L, Araújo-Soares V, Rapley T, Adamson A. A feasibility study with process evaluation of a preschool intervention to improve child and family lifestyle behaviours. BMC Public Health. 2017;17(1):248.
Meiksin R, Crichton J, Dodd M, Morgan GS, Williams P, Willmott M, et al. A school intervention for 13- to 15-year-olds to prevent dating and relationship violence: the project respect pilot cluster RCT. Public Health Res. 2020;8(5).
Newbury-Birch D, Scott S, O’Donnell A, Coulton S, Howel D, McColl E, et al. A pilot feasibility cluster randomised controlled trial of screening and brief alcohol intervention to prevent hazardous drinking in young people aged 14–15 years in a high school setting (SIPS JR-HIGH). Public Health Res. 2014;2(6).
Sahota P, Christian M, Day R, Cocks K. The feasibility and acceptability of a primary school-based programme targeting diet and physical activity: the PhunkyFoods Programme. Pilot Feasibility Stud. 2019;5(1):152.
Sebire SJ, Jago R, Banfield K, Edwards MJ, Campbell R, Kipping R, et al. Results of a feasibility cluster randomised controlled trial of a peer-led school-based intervention to increase the physical activity of adolescent girls (PLAN-A). Int J Behav Nutr Phys Act. 2018;15(1):50.
Segrott J, Rothwell H, Hewitt G, Playle R, Huang C, Murphy S, et al. Preventing alcohol misuse in young people: an exploratory cluster randomised controlled trial of the Kids, Adults Together (KAT) programme. Public Health Res. 2015;3(15).
Sharpe H, Schober I, Treasure J, Schmidt U. Feasibility, acceptability and efficacy of a school-based prevention programme for eating disorders: cluster randomised controlled trial. Br J Psychiatry. 2013;203(6):428–35.
White J, Hawkins J, Madden K, Grant A, Er V, Angel L, et al. Adapting the ASSIST model of informal peer-led intervention delivery to the Talk to FRANK drug prevention programme in UK secondary schools (ASSIST + FRANK): intervention development, refinement and a pilot cluster randomised controlled trial. Public Health Res. 2017;5(7).
Wright B, Marshall D, Adamson J, Ainsworth H, Ali S, Allgar V, et al. Social Stories™ to alleviate challenging behaviour and social difficulties exhibited by children with autism spectrum disorder in mainstream schools: design of a manualised training toolkit and feasibility study for a cluster randomised controlled trial with nested qualitative and cost-effectiveness components. Health Technol Assess. 2016;20(6).
HM Government. Types of School. [Available from: https://www.gov.uk/types-of-school]. Accessed 01 Sept 2021.
Raab GM, Butcher I. Balance in cluster randomized trials. Stat Med. 2001;20(3):351–65.
Moulton LH. Covariate-based constrained randomization of group-randomized trials. Clin Trials. 2004;1(3):297–305.
Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, Petticrew M. Developing and evaluating complex interventions: the new Medical Research Council guidance. BMJ. 2008;337: a1655.
Guthold R, Stevens GA, Riley LM, Bull FC. Global trends in insufficient physical activity among adolescents: a pooled analysis of 298 population-based surveys with 1· 6 million participants. Lancet Child Adolesc Health. 2020;4(1):23–35.
Sadler K, Vizard T, Ford T, Marchesell F, Pearce N, Mandalia D, et al. Mental health of children and young people in England, 2017. Leeds: NHS Digital; 2018.
Arain M, Campbell MJ, Cooper CL, Lancaster GA. What is a pilot or feasibility study? A review of current practice and editorial policy. BMC Med Res Methodol. 2010;10(1):67.
Aventin Á, Lohan M, Maguire L, Clarke M. Recruiting faith- and non-faith-based schools, adolescents and parents to a cluster randomised sexual-health trial: experiences, challenges and lessons from the mixed-methods Jack Feasibility Trial. Trials. 2016;17(1):365.
Henderson M, Wight D, Nixon C, Hart G. Retaining young people in a longitudinal sexual health survey: a trial of strategies to maintain participation. BMC Med Res Methodol. 2010;10(1):9.
Pound B, Riddell M, Byrnes G, Kelly H. Perception of social value predicts participation in school-based research. Aust N Z J Public Health. 2000;24(5):543–5.
Kitty Parker and Obioha Ukoumunne were supported by the National Institute for Health Research Applied Research Collaboration South West Peninsula. Saskia Eddy received a Doctor of Philosophy (PhD) studentship from Barts Charity. SEd received training from the Medical Research Council (MRC)—National Institute of Health Research (NIHR) Trials Methodology Research Partnership (TMRP). Barts Charity, the MRC, the NIHR and the TMRP have no role in the study design, collection, management, analysis or interpretation of data, writing of the report or the decision to submit the report for publication. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.
This research was funded by the National Institute for Health Research Applied Research Collaboration South West Peninsula. Saskia Eddy received a Doctor of Philosophy (PhD) studentship from Barts Charity.
Ethics approval and consent to participate
Consent for publication
Sandra Eldridge is a member of the editorial board of the journal Pilot and Feasibility Studies.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Parker, K., Eddy, S., Nunns, M. et al. Systematic review of the characteristics of school-based feasibility cluster randomised trials of interventions for improving the health of pupils in the UK. Pilot Feasibility Stud 8, 132 (2022). https://doi.org/10.1186/s40814-022-01098-w
- Cluster randomised trials
- Feasibility study
- Pilot study
- Public health
- Randomised trials
- Research methods
- Systematic review
- Trial methodology