Farrin, AJ; Nixon, J; Stevens, J; Roberts, K; Foy, R

14 Pages·2017·1.67 MB·English

Bryantetal.Trials (2017) 18:40
DOI10.1186/s13063-016-1732-3
STUDY PROTOCOL Open Access
Effectiveness of an implementation
optimisation intervention aimed at
increasing parent engagement in HENRY, a
childhood obesity prevention programme -
the Optimising Family Engagement in
HENRY (OFTEN) trial: study protocol for a
randomised controlled trial
Maria Bryant1* , Wendy Burton1, Bonnie Cundill1, Amanda J. Farrin1, Jane Nixon1, June Stevens2,3, Kim Roberts4,
Robbie Foy5, Harry Rutter6, Suzanne Hartley1, Sandy Tubeuf7, Michelle Collinson1 and Julia Brown1
Abstract
Background: Family-based interventions to prevent childhood obesity dependupon parents’taking action to
improve diet and other lifestyle behaviours intheirfamilies. Programmes that attract and retain high numbers of
parents provide an enhanced opportunity to improve public health and are also likely to be more cost-effective
than thosethat do not. We have developed a theory-informed optimisation interventionto promoteparent
engagement within an existing childhood obesity prevention group programme,HENRY (Health Exercise Nutrition
for theReally Young).Here, we describe a proposal to evaluate the effectiveness of this optimisation intervention in
regard to theengagement ofparents and cost-effectiveness.
Methods/design: The OptimisingFamily Engagementin HENRY(OFTEN) trial is a clusterrandomised controlled
trial being conducted across 24 local authorities (approximately 144 children’s centres) which currently deliver
HENRY programmes. The primary outcome will be parental enrolment and attendance at the HENRY programme,
assessed using routinelycollected process data. Cost-effectivenesswill be presented interms of primary outcomes
usingacceptabilitycurvesandthroughelicitingthewillingnesstopayfortheoptimisationfromHENRYcommissioners.
Secondaryoutcomesincludethelongitudinalimpactoftheoptimisation,parent-reportedinfantintakeoffruitsand
vegetables(asaproxytocompliance)andotherparent-reportedfamilyhabitsandlifestyle.
(Continuedonnextpage)
*Correspondence:[email protected]
1LeedsInstituteofClinicalTrialsResearch,UniversityofLeeds,LeedsLS29JT,
UK
Fulllistofauthorinformationisavailableattheendofthearticle
©TheAuthor(s).2016OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.0
InternationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinkto
theCreativeCommonslicense,andindicateifchangesweremade.TheCreativeCommonsPublicDomainDedicationwaiver
(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Bryantetal.Trials (2017) 18:40 Page2of13
(Continuedfrompreviouspage)
Discussion:Thisinnovativetrialwillprovideevidenceontheimplementationofatheory-informedoptimisation
interventiontopromoteparentengagementinHENRY,acommunity-basedchildhoodobesitypreventionprogramme.
Thefindingswillbegeneralisabletootherinterventionsdeliveredtoparentsinothercommunity-basedenvironments.
Thisresearchmeetstheexpressedneedsofcommissioners,children’scentresandparentstooptimisethepotential
impactthatHENRYhasonobesityprevention.Asubsequentclusterrandomisedcontrolledpilottrialisplannedto
determinethepracticalityofundertakingadefinitivetrialtorobustlyevaluatetheeffectivenessandcost-effectiveness
oftheoptimisedinterventiononchildhoodobesityprevention.
Trialregistration:ClinicalTrials.govidentifier:NCT02675699.Registeredon4February2016.
Keywords:Engagement,Trial,Implementation,Childhoodobesity,Parent,Recruitment
Background all parenting interventions delivered in community set-
Although the most recent data suggest a levelling off of tings to optimise their impact and so that the degree to
the prevalence of childhood obesity in the United which they are effective can be assessed with confidence.
Kingdom,levels remainhighandthere are clear inequal- In contrast to conventional pathways, in which imple-
ities, with rates continuing to rise in more deprived and mentation research is conducted following trials for clin-
ethnic minority groups [1]. Childhood obesity impacts ical effectiveness [23], it is argued that there is a need
physiological and psychological health, which tracks into for comprehensive early-phase (evidentiary) evaluation
adulthood [2, 3], increasing the risk of morbidity and and enhancement of complex interventions such as
mortality [4, 5]. It imposes significant costs on the U.K. those designed to prevent childhood obesity [24], prior
economy, with an expected sevenfold increase in related to the conduct of large randomised controlled trials
NHS costs by 2020 and a forecasted £2 billion annual (RCTs). This novel approach ensures that factors which
spend by 2030 [6]. Tackling obesity is a national public limit trial outcomes, such as low adherence, are mini-
health priority in terms of both treatment and preven- mised prior to dedicating the resources required to con-
tion. Establishing healthy behaviours in early childhood duct a large trial which may identify no evidence of
iscriticalforoptimumgrowthanddevelopment[7].Fur- effectiveness,perhapsasaresult ofpoorcompliance.
ther, poor eating patterns developed early can persist The present trial involves the evaluation of an imple-
and are associated with chronic diseases in adulthood mentation enhancement ‘optimisation intervention’ of
(e.g., cardiovascular disease, type 2 diabetes mellitus [8]). an existing preschool obesity prevention group
Once established, obesity is difficult to reverse [9], programme, HENRY (Health Exercise Nutrition for the
strengthening the case for primary prevention [10, 11]. Really Young), to promote parent enrolment and attend-
To reverse trends in obesity, there is a clear need to en- ance prior to establishing its clinical effectiveness and
gage parents in shaping healthy weight-related behav- cost-effectiveness. HENRY is an 8-week group-delivered
iours intheirchildren. programme provided to parents of preschool children. It
Design, implementation and evaluation of interven- wasdeveloped in2006with jointfunds from theDepart-
tions to prevent or treat obesity may be expensive and ment of Health and the Department for Education (then
time-consuming, and researchers often report a lack of called the Department of Children, Schools and Families)
effectiveness [12–15]. Within the literature on childhood and is currently commissioned and delivered across the
obesity, low parent enrolment and attendance at group- United Kingdom by 32 local authorities providing more
delivered programmes, along with a lack of reported be- than 150 programmes each year. It is delivered in com-
haviour change, are commonly described to have had a munity settings, often in children’s centres by children’s
substantial impact on group dynamics, effectiveness and centre staff [25]. HENRY uses a responsive approach to
cost-effectiveness[16–18]. provide practical guidance and improve parenting skills
There is a need for pragmatic,‘real-world’ evaluations aimed at enhancing family lifestyle and children’s centre
of interventions to understand the generalisability of the environments. Preliminary data indicate that HENRY
interventions across everyday practice [19, 20]. Further, may be effective at preventingchildhoodobesityandim-
in order for interventions to be accurately implemented proving family health [25], although there is not yet evi-
and interpreted, the underpinning behaviour change dence from anRCT.
techniques need to be clearly defined [21]. Implementa- Despite some indications of success of HENRY from
tion optimisation strategies can then be adopted to tar- audit [25] and qualitative data [26–28], process evalu-
get a specific intervention component [22]. It is ation indicates implementation targets are often not
imperative that parent engagement be improved across met. Lack of involvement by some parents may be a
Bryantetal.Trials (2017) 18:40 Page3of13
threat to the viability of this programme and other, simi- developedtooptimiseparentengagementinobesitypre-
lar community-based interventions. The effectiveness of vention programmes, and which have been evaluated
interventions in real-world conditions may not match usingan RCTdesign. Tomeet thisneed,we developeda
the efficacy found in research studies [29]. This may be parent engagement optimisation intervention following
particularlyrelevanttopopulation-basedpreventionpro- Medical Research Council (MRC) guidance for develop-
grammes,inwhichparentsmaybelesslikelytobemoti- ment of complex interventions [38]. The Consolidated
vated to attend because their children show no clinical Framework for Implementation Research was used to
symptoms [30, 31]. Thus, it is imperative to create tai- understand contextual factors associated with the imple-
lored methods to maximise parent participation and en- mentation of HENRY and their impact on parent en-
hance successful implementation in childhood obesity gagement [43]. The Behaviour Change Wheel [44]
preventionprogrammes. provided guidance for the development of the interven-
Failure to attend interventions is complicated byissues tion, which features use of the ‘COM-B’ model (capabil-
of health inequalities; socio-demographic attributes; and, ity, opportunity and motivation) to dissect behaviours
in some cases, challenges related to safeguarding [30, 32, into their individual and interacting components and
33]. Literature on non-attendance in other areas (clinical consider which of these should be targeted in an inter-
appointments) indicates that attendance is lower in cer- vention to bring about the desired behaviour change. A
tainpopulations andmay beanindicator of vulnerability rapid ethnography was conducted in five children’s cen-
[34]. This is pertinent to obesity because prevalence tres to understand the environmental, social, psycho-
rates continue to rise in children in lower socio- social, political and economic factors associated with
economic or ethnic minority groups. In addition to the parent engagement in HENRY. This study gathered data
anticipated public health benefits, the economic benefits from approximately 190 h of observations, 22 interviews
of implementation optimisation in programmes such as with children’s centre managers and staff, HENRYcom-
HENRYare substantial, with one parenting intervention missioners,andHENRYcoordinatorsandfacilitators.Six
calculating an extra £800 cost per child in programmes focus groups were also conducted with parents who had
that run with 8 parents compared with those running at attended the programme. We convened an intervention
theintendedcapacityof12 parents[35]. development team consisting of experts in intervention
Unlike the development of other robust behaviour development, obesity, applied health and behaviour
change interventions, initiatives to improve parental at- change,alocalauthorityrepresentative,aHENRYparent
tendance at clinical appointments have been confined champion, a parent who has attended the programme,
largely to simple approaches such as text reminders [34, and a representative from the HENRY team. The inter-
36, 37]. To promote behaviour change, enhance the vention team used evidence from the ethnography, from
transparency of interventions and guide their generalis- the literature, and from their own experiences and ex-
ability, it is recognised that a systematic approach is pertise to develop a multi-component optimisation
used during intervention development, which is under- intervention which attempts to promote parent engage-
pinned by theories of behaviour change and guided by ment and positive behaviour change in families by sup-
an intervention-planning framework such as the Behav- portinglocal authorities,children’scentres,HENRYstaff,
iour Change Wheel [24, 38]. Lessons can be learnt from and parents using candidate implementation and deliv-
child mental health research, in which many interven- ery strategies (Career Development Fellowship CDF-
tions have been developed and tested using robust 2014-07-052). Here we report the protocol ofour cluster
theory-based approaches which carefully consider parent randomised controlled trial (cRCT) (using routine
engagement [33]. Research inthis fieldhasexplored per- process data) to evaluate the optimisation intervention
ceived barriers to participation with suggested strategies for its ability to engage parents in HENRY prior to con-
to engage parents, including promotion of self-efficacy ducting afurtherevaluation ofitsclinical effectiveness.
and treatment motivation. Such strategies perceive par-
ents as the central agents of change and aim to modify Methods/design
the pre-treatment experience [33, 39]. RCTs evaluating Design
parent engagement in mental health services [40] and This study is a two-arm, multi-centre cRCT across 24
drug misuse [41] indicate that parental engagement can local authorities (local governments) (supporting ap-
be increased with a dedicated optimisation component, proximately 144 children’s centres) in the United
although a further trial focused on cultural relevance for Kingdom to determine the effectiveness and cost-
child behaviour problems to optimise engagement [42] effectiveness of an optimisation intervention to promote
didnotfindanyincrease. parent engagement in the HENRY programme com-
To our knowledge, there are no theory-based, multi- pared with standard HENRY practice (Fig. 1). Although
component optimisation interventions that have been the children’s centres deliver the HENRY programme,
Bryantetal.Trials (2017) 18:40 Page4of13
Fig.1Trialdesignflowdiagram.HENRYHealthExerciseNutritionfortheReallyYoung,ITTIntentiontotreat
owing to the multi-component and interactive nature of 1. Theeffectofthe optimisationinterventionon
the optimisation intervention, local authorities will be achievingcombinedparent enrolment,attrition and
the units of randomisation (i.e., clusters) to prevent con- compliancetargets
tamination between the randomised groups. Outcomes 2. Theeffectofthe optimisationinterventiononparent
will be assessed through routinely collected data avail- compliancewith HENRY programmecontent
able foreach HENRYprogrammeat the parentand chil- 3. Theeffectofthe optimisationinterventionon
dren’s centre levels. Information on commissioners’ parent-reportedfamily habitsand lifestyle
willingness to pay (within local authorities) related to 4. Thepotential longitudinalimpact(sustainability) of
the optimisation intervention will be elicited. A process theoptimisationinterventiononenrolmentand
evaluation will be conducted alongside the cRCT in ac- attritionincentresthatprovide datafrom more
cordance with the MRC guidance on evaluating process thanoneprogramme
incomplex interventions[19].
Setting
The study will be conducted in local authorities with
Objectives
children’s centres where staff are currently trained to de-
Primary objectives of the trial are to determine the
liver the HENRY programme. There are currently 32
following:
local authorities (approximately 144 centres) in the
United Kingdom running the programme, all of which
1. Theeffectofthe optimisationinterventionapplied
provide process data to the central HENRY office for
toHENRYcomparedwith standardHENRYin
monitoring/quality assurance (QA) purposes. These data
regardtoincreasing parentenrolmentinHENRY
willbeanonymisedandtransferred foranalysis.
programmesorreducingparent attrition within
HENRYprogrammes
2. Thewillingnessto payofcommissionersofthe Inclusioncriteria
HENRYprogrammeperadditionallevelofparent Local authoritiescomprisingthe following:
engagementrelatedtotheoptimisationintervention
(Costswillbepresentedasanincrementalcost per (cid:1) Localauthoritiesthatcommission HENRYand
increaseofparentengagement,sothatcommissioners consent fortheircentrestobeinvolved inthe
areabletoidentifythepointatwhichadditionalcosts research
areacceptablyoffsetbythebenefitsofengaging (cid:1) CommissioningHENRYprogrammesdelivered by
effectivelywithparents.) trainedstaff whohave beencertifiedbyHENRY
Secondaryobjectivesofthestudy aretodetermine: Children’scentrescomprisingthefollowing:
Bryantetal.Trials (2017) 18:40 Page5of13
(cid:1) Centres providingroutinelycollecteddataforthe
mostrecentHENRY programmethatthey ranat
thepointofrandomisation
Exclusioncriteria
Local authorities:
(cid:1) Localauthoritieswhich plantodecommission the
HENRY interventionduringtheperiodofthetrial
orwhich arenotplanningtorunanyHENRY
programmesduringthetrialperiod
Children’scentres:
(cid:1) Centres which wererecruitedinthe development
phaseoftheoptimisationintervention(participating
intherapidethnography)
(cid:1) Centres which arenotplanningtorunanyHENRY
programmesduringthetrialperiod
There will be no exclusions based on the demograph-
ics of children’s centres, but location will be monitored
to ensure inclusion of those with diverse social and
environmental characteristics. Randomisation will also
stratify by area-level deprivation to ensure balance be-
tweentrial arms.
Recruitmentandconsent
Local authorities and their centres across the United Fig.2Recruitment/opt-outprocess.HENRYHealthExerciseNutrition
fortheReallyYoung,LALocalauthorities
Kingdom will be identified from an existing database of
HENRY delivery sites (see Fig. 2). Those meeting the eli-
gibilitycriteriawillbecontactedbyinvitationletter(pos-
tal and electronic) issued jointly by the HENRY central conducted in public health. The Leeds Institute of Clin-
office and the University of Leeds. HENRY is currently ical Trials Research (LICTR) has experience in using an
commissioned across the whole of the United Kingdom, opt-out approach and has found that it (1) is more
with most sitessituatedwithin England. likely than opt-in methods to provide a more represen-
Explicit consent will not be sought. Instead, opt-out tative or ‘typical’ participating sites; (2) appears more
consent will be sought at the cluster level (local author- effective and efficient in promoting research participa-
ities) and from the centres within each local authority. tion by sites; and (3) is acceptable, provided that suffi-
It will be possible for centres to decline participation in cient sensitivity and safeguards are applied [46]. In this
the randomisation, even if they are based within a con- instance, individual-/family-level consent is not re-
senting local authority. However, if a local authority de- quired, because all patient-level data will be unlinked
clines to take part in the trial, none of its centres will and anonymous and is routinely collected by centres
be eligible to take part. Exclusion criteria related to fur- and submitted to the central HENRY office. Centres
ther commissioning of HENRY (local authorities) and will not be asked to conduct any additional data collec-
plans to run HENRY programmes (centres) will be self- tion from parents for the purposes of this trial. Our
reported by commissioners and centres, respectively. parent advisory panel (patient and public involvement
This opt-out method has been chosen because this is a group) and trial steering committee (TSC) have both
low-risk trial in which anonymised data will be ex- approved this process.
tracted remotely. It has been advocated to promote a Local authorities and children’s centres will be given
simple and efficient trial design, and literature suggests 30 days from the date on which the initial information
that it is acceptable to participants [45]. It also repli- letter is sent to decline participation (confirmed by re-
cates, as far as possible, the ‘real-life’ conditions under corded delivery). Centres that opt out after randomisa-
which such quality improvement initiatives are usually tion willbetreatedas‘withdrawn’.
Bryantetal.Trials (2017) 18:40 Page6of13
Unitofrandomisation for the duration of the trial, the HENRY QA team have
Local authorities will be randomised in a 1:1 allocation agreedtoreallocatedareasthattheyareresponsiblefor,so
ratio (HENRY+optimisation intervention, HENRY as thatonlyonepersonwillhavecontactwithareasallocated
standard) by a statistician at LICTR, using an algorithm to the optimisation intervention. Other QA staff will be
for covariate-constrained randomisation [47] to achieve aware that they have standard QA responsibility (e.g., an-
a balanced allocation between the trial arms on (1) local sweringquestions,providingencouragement,checkingon
authority baseline level of parental engagement with progress) for areas not randomised to the optimisation,
HENRY (proportion of centres recruiting a minimum of but will not know the details or content of the optimisa-
eight parents per programme, proportion of centres tionintervention.Tofurtheravoidcontamination,thede-
retaining at least 75% of parents for a minimum of five tailsoftheoptimisationinterventionwillnotbepublished
of eight sessions), (2) proportion of centres running at untilafterthefinalanalysis.Unplannedmasking(unblind-
least one HENRY programme in 2015, (3) size of local ing)willbemonitoredandreportedtotheTSC.
authorities (number of children’s centres participating),
and (4) area deprivation (proportion of centres in the Intervention
least/most deprived quintiles as ranked by the 2015 HENRY
Index of Multiple Deprivation at the Lower Layer Super HENRY is an 8-week programme delivered in children’s
Output Area) [48]. Randomisation will be performed for centres with the aim of providing parents with skills,
all local authorities at a single time point after baseline knowledge and confidence to support healthy lifestyles
data are transferred, prior to the implementation of the among their preschool children and their families, in-
intervention. All possible enumerations of allocation to cluding parenting skills, emotional well-being and activ-
the trial arms will be generated, and an imbalance statis- ity. The programme was set up in 2006 with the aim of
ticwillbe calculatedforeachallocation. A setofoptimal reversing rising trends in school entry age obesity.
allocations which minimise the imbalance will be pro- HENRY is currently delivered within 32 local authorities
vided, and, in accordance with the principles of Inter- across England and Wales by trained health and com-
national Conference on Harmonisation of Technical munity practitioners who undergo two stages of training
Requirements for Registration of Pharmaceuticals for [26]. Stage 1, centre-level training, is a 2-day workshop
Human Use guidance on randomness, the final alloca- designed to equip centre staff with knowledge and skills
tion will be randomly selected from among this optimal to promote and provide healthy nutrition within early
set [49]. Results of the randomisation procedure will be years settings and support parents to provide healthy
sent by the LICTR statistician to the HENRYcentral of- family lifestyles and nutrition for their families. The the-
fice, which will be responsible for informing local au- oretical underpinning combines proven models of be-
thoritiesoftheirtreatmentallocation. haviour change, including the Family Partnership Model,
motivational interviewing and solution-focused support.
Blinding Stage 2, practitioner-level training to deliver the HENRY
Owing to the nature of the intervention, it will not be programme to families, is provided to approximately
possible to blind allocation within intervention sites or four practitioners per centre after completing Stage 1 to
of those involved in the optimisation intervention train- deliver the 8-week HENRY programme. The aim of this
ing. Families attending the HENRY programme have stage is to build parents’skills, knowledge and confidence
been made aware that HENRY uses data anonymously to change old habits, provide healthier nutrition for their
for research, but they have not been told explicitly about youngchildren,andencouragehealthierlifestyles[25,27].
the Optimising Family Engagement in HENRY (OFTEN) Programme contentincludessessionsonlifestyleandeat-
trial or whether their local authority has been assigned inghabits(e.g.,familymeals),balancinghealthymealsand
to the optimisation. HENRY is coordinated by a central snacks,child-appropriateportionsizes,parenting,physical
team that is responsible for coordinating training and activityandemotionalwell-being.
for providing support and QA to areas that commission Current HENRY QA practice involves the review of
HENRY. Because the HENRYcentral office is integral to processdata bydedicatedindividualsintheHENRYcen-
the delivery of the optimisation intervention, it will not tral office with provision of written and oral feedback.
be possible to blind the HENRYcentral office (as it will This QA will continue in both trial arms and will be
communicate the randomisation allocation to all those monitored. Staff delivering current QA activities will be
responsible for its implementation); however, staff who blindedtotreatmentallocation.
are responsible for collating data and transferring it to
LICTR will remain blinded to allocation. The chief ex- HENRY+parentengagementoptimisationintervention
ecutive officer of HENRY will be blinded to treatment We summarise only the key features of the optimisation
allocationand willnot attendTSCmeetings.In addition, intervention in the trial protocol to minimise
Bryantetal.Trials (2017) 18:40 Page7of13
contamination (i.e., uptake of initiatives of the interven- – Parentingself-efficacy(assessedviafourparenting
tion by local authorities randomised to the control arm) confidence itemswith 5-point Likertscales,
and maintain blinding. The optimisation intervention modifiedfromthevalidatedParentingSelf-Agency
consists of interacting components delivered to local Measure[50]).Thismeasureassessestheparents’
authorities, children’s centres and HENRY facilitators. estimateoftheirabilitytoinfluencetheirchildand
It includes additional support to stakeholders, modifi- environmenttoleadtopositivedevelopment,
cation of HENRY marketing to clarify perceptions of whichisakeygoaloftheHENRYapproach.In
the programme, and methods to modify the pre- orderforthequestionnairetobefeasibleto
programme experience for parents. Strategies have administeraspartoftheservice,HENRYchoseto
been developed to increase parent motivation to enrol reducethenumberofitemstoasinglescaleof
in HENRY and promote parent self-efficacy to con- parentingconfidence.Internalconsistencyofthese
tinue to attend. Implementation of the optimisation itemsinapreviousHENRYevaluationwashigh,
intervention will begin immediately following random- andinternalconsistencyinthissamplewashigh
isation. All components of the optimisation interven- (Cronbach’salpha=0.82atbaseline,0.74at
tion will be implemented within 6 months following completion)[51].
randomisation. – Eating behaviours(basedonthe GolanFamily
EatingandActivity Habits Questionnaire[52]).
HENRYasstandard This validated measureissensitive tochangeand
Local authorities randomised to the control arm will demonstrates highlevelsofreliabilityandvalidity.
continue to deliver HENRY programmes as per standard HENRYfacilitatorsroutinelyadminister sixitems
practice. relatedtofamily behaviours, includedparent
report offamilysitting togetherformeals,
Outcomes watchingtelevisionduringmealtimes,consuming
Primaryoutcome take-outmeals, consumptionofhome-cooked
The effectiveness of the optimisation intervention will foods,stoppingeatingwhenfulland choosing
be determined by comparing parent engagement in healthymeals. Individualevaluationofthese
HENRY+optimisation intervention vs HENRY alone. itemswillbeundertakenbecause reliabilityfor
The co-primary outcomes are (1) the proportion of cen- thescalewaspoorwhenpreviouslytestedusing
tres enrolling at least eight parents per programme and HENRYfamilies(Cronbach’salpha=0.52at
(2)the proportion ofcentres withat least75%ofparents baseline,0.56atcompletion).
attending fiveofeight sessionsperprogramme.Bothwill – Family activity (viaabriefHENRYquestionnaire
be evaluated 12 months post-randomisation. The opti- asking parents toreportthe frequency that
misation intervention will be deemed effective if either parentsand childrenengageinexercise).This
theenrolmentorattrition goalsaremet. questionnairehasbeen developedbespoketouse
within theHENRYevaluationsand includes
Secondaryoutcomes parentalreportoftheir ownactivities thatresult
Secondary outcomes, measured 12 months post- inbreathlessness, withresponsecategories
randomisation (unless stated otherwise), include some rangingfromnone to<1h,2h,3hor>3h.
further assessment of parent engagement to explore the Children’sactivityisassessed usingparental
degree to which the intervention was effective. Data report asenergetic play,with responsecategories
from additional questionnaires which are routinely ad- rangingfromnone to5–15minutes, 20–30
ministered by HENRY facilitators at the beginning and minutes,30minutesto1h,or>1h.
end of each 8-week course will also been analysed to ex- – Child screentime(viaaparentreportedusingthe
plore the potential impact of the optimisation on behav- bespokeHENRYquestionnaire).Thisassessesuse
iouraloutcomes,outlinedbelow: oftelevisions,DVDs,computers,smartphones,
and soforth withresponsecategoriesranging
(cid:1) Proportionofcentresachievingalltargets for from noneto<1h,1–2h,2–3h,or>3h.
enrolment,attrition andparentcompliance – Intakeofkeyindicatorfoodsperday(assessedvia
(cid:1) Parent adherencetoHENRY programmecontent, 14itemsfromamodified validatedFood
definedastheproportionofparentsreportingan FrequencyQuestionnaire[53]completed in
increaseof0.5inthedailyfrequencyofconsumption relation tothe parent andchild).Parents are
offruitsandvegetablesbychildrenperprogramme askedtoestimatehowoften (never,onceper
(cid:1) ImpactofHENRYonparentingandfamilyhealth, month,once perfortnight,1–7daysperweek)
assessed byparent-report,including thefollowing: they consumedeach ofthe14itemsorgroupsof
Bryantetal.Trials (2017) 18:40 Page8of13
foods (e.g.,freshfruit,sweets,chocolate, water), optimisation intervention is cost-effective for a range of
with spacetoreport thenumberoftimes maximum monetary values that a decision-maker might
consumed.Ithaspreviouslydemonstrated bewillingtopayforaparticularunitchangeinoutcome.
sensitivitytochange with parents attending Cost-effectiveness analysis leaves it to decision-makers
HENRYsessions [25]. to form their own view of the relative importance of
(cid:1) Longitudinalimpact ofHENRY+optimisation these results [54]; therefore, in a second part of the
interventionvsHENRY alone(parentengagement work, we will conduct a meeting with commissioners to
[enrolmentandattrition]within centresduringthe determine their willingness to pay per additional unit of
12-month follow-up inlocalauthoritieswhich effectiveness (parent engagement) before concluding
provide dataformorethan oneprogrammeper whether the optimisation intervention can be considered
centreatfollow-up) (Thisevaluationwillexplore cost-effective. We will use contingent valuation tech-
whetherthe optimisationinterventionneedstime to niques and design an experiment to estimate the point
‘bed-in’and/orwhetheranyearlyeffects are at which commissioners consider that the costs of the
maintainedovertime.) programme are acceptably offset by the benefits of en-
gagingeffectivelywithparents.
Processevaluation
A process evaluation will be conducted by an LICTR re- Samplesize
searcher in accordance with MRC guidance on evaluat- We assumed that 25% of the 32 local authorities cur-
ing process in complex interventions [19]. In brief, the rently running HENRY will not be eligible or will opt
aims ofthis evaluation will beto (1) examinethe uptake, out of the study, leaving 24 local authorities (12 per
delivery and acceptability of the optimisation interven- arm). Power calculations for this fixed sample size (with
tion across local authorities/children’s centres (reach); alleligible centres intheUnitedKingdom invitedtotake
(2) explore the effects of individual intervention compo- part) were therefore conducted to examine anticipated
nents at increasing parent engagement, along with po- power for various intervention effects, at the 5% signifi-
tential unintended consequences; and (3) consider the cance level, in each of the primary outcomes and for the
utility of theories underpinning the intervention and ap- composite endpoint (enrol at least eight parents per
proach used to develop the intervention design. Process programme and retain ≥75% of parents attending five of
measures will include quantitative measurement of eight sessions) (see Table 1 for scenarios). On the basis
centre uptake and fidelity of delivered trial components; of data from previous HENRY programmes (all terms in
qualitative and quantitative analysis of stakeholder 2014), we assumed an average of 6 children’s centres per
perceptions of intervention components considering local authority providing a total of 144 children’s centres
individual, organisational and contextual factors; quanti- (72 per arm), an intra-cluster correlation coefficient
tative measurement of enrolment methods used to re- (ICC) between 0.05 and 0.1, a coefficient of variation in
cruit HENRY programme participants; and qualitative cluster size of 0.54, and the following estimates of the
and/or quantitative measurement of predicted behaviour outcomes in the control sites (HENRY alone): 55% of
change outcomes among stakeholders targeted in the centres will enrol at least eight parents per programme;
implementation optimisation. Quantitative data required 50% of centres will retain ≥75% of parents attending five
for process evaluation will be gathered and reported on of eight sessions; and 25% of centres will enrol at least
paper case report forms by a researcher within the eight parentsper programme andretain ≥75%of parents
LICTR, who will also collect qualitative data from a ran- attending five of eight sessions. Thus, with the antici-
dom selection of sites in person, including interviews pated number of centres (24 local authorities, 144 chil-
andobservations. dren’scentres), we will haveat least80% power to detect
meaningful improvements in differences of between 25%
Economicevaluation and 30% in either of the primary endpoints or the com-
Economicevaluationwillmodelthebenefitsofoptimisa- posite endpoint at the 5% significance level, assuming
tion intervention over the costs of implementation and theICCisnogreater than 0.05.
will seek commissioners’ willingness to pay for this. It
willincludethe costsrequiredtodelivertheintervention Datacollectionandtransfer
(gathered and transferred from HENRY) and the rou- In this trial, we will use routinely collected data for all
tinely collected outcomedata. primary and secondary outcomes, except for the process
The cost-effectiveness analysis in the present trial will evaluation data, which will be gathered by an independ-
include an array of trial endpoints (including parent en- ent LICTR researcher. HENRY has designed a centra-
rolment and attrition rates) and costs. Cost-effectiveness lised process for evaluating programmes, developed for
acceptability curves will describe the probability that the QA purposes and reporting to commissioners of the
Bryantetal.Trials (2017) 18:40 Page9of13
Table1Powercalculationsforenrolmentandattritionendpoints information on participant satisfaction, parent-reported
forvariousestimatesoftheinterventioneffectandafixedsample compliance,parentingandfamilylifestyle(Table2),which
sizeof144children’scentresin24localauthorities aregatheredinquestionnairesonthefirstandlastdaysof
Outcomein Percentagepointincrease Powerfor Powerfor each HENRY programme. This information is submitted
thecontrol inintervention ICC=0.1 ICC=0.05 to the central HENRY office (without any personal iden-
Enrolment(≥8parentsperprogramme) tifiersforthefamilies),whichcheckthedataforcomplete-
55% 5% 6% 7% nessandcollatetheinformationina.csvfile.
10% 15% 18% For the trial period, centre-level baseline data from the
most recent programme delivered prior to randomisa-
15% 29% 35%
tion (winter term 2015) will be used. Follow-up data will
20% 49% 58%
be gathered 12 months after randomisation. It will in-
25% 70% 79%
clude routine data collected from the autumn and spring
30% 87% 93% terms (2016–17), allowing 6 months for intervention de-
Attrition(≥75%ofparentsattendingfiveofeightsessions) livery at all sites followed by the delivery of two terms of
50% 5% 6% 7% HENRY programmes. In accordance with the planned
intention-to-treat (ITT) analysis, data will be collected
10% 15% 17%
from centres that are not compliant with the interven-
15% 28% 34%
tion. For those withdrawing from the trial, only data col-
20% 47% 55%
lected uptothepoint ofwithdrawal willbeused.
25% 67% 76% Alldata transferred from HENRYtothe LICTRwillbe
30% 84% 91% inthe form ofunlinked,anonymised datasetsandwill be
Enrolmentandattrition(atleasteightparentsperprogrammeand75% transferred via a secure, encrypted system. Only data for
ofparentsattendingfiveofeightsessions) centres participating in the trial will be transferred. Data
25% 5% 7% 8% transfer agreements (including details on the sender, re-
cipient, content of transfer/general purpose of transfer
10% 17% 19%
and any data-processing limitations) will be set up be-
15% 31% 37%
tween HENRY and the LICTR prior to the transfer of
20% 49% 58%
any data. Queries pertaining to outliers or missing data
25% 67% 76% willbesent toHENRYinaccordance withLICTR stand-
30% 82% 89% ard operating procedures. Any new data that are identi-
ICCIntra-clustercorrelationcoefficient fied following queries will also be transferred via the
secure,encryptedsystem.Missingdata,exceptindividual
service. For the purpose of this trial, baseline data at the data items collected via parent questionnaires, will be
centre level include routine data on enrolment and at- chased until it is received, confirmed as not available, or
tendance at programmes that have been run before the trial is at analysis. For missing item-level data within
randomisation. These data also include anonymous questionnaires (item non-response), no attempts will be
Table2Datacollectionsummary
Data Screening Randomisation Baseline On-going Follow-up
Localauthoritylevel
Clustereligibility X
Stratificationfactors X
Programmeenrolment X X
Programmeattrition X X
Implementationdata(e.g.,fidelity) X
Parentlevel(anonymisedroutine)
Infantdiet(fruitandvegetableintake) X X
Familyhabitsandlifestyle X X
Implementation/processdata X
Interventionlevel
Costsforinterventiondeliveryforeconomicevaluation X X X X X
Xindicateswhendataaregathered
Bryantetal.Trials (2017) 18:40 Page10of13
made to retrieve missing data; only missing question- ofthedata structure,andthereliabilityoftwo-level(par-
naireswillbe chased. ents nested within centres or local authorities) and
All data provided will be stored, handled and processed three-level (parents nested within centres within local
inaccordancewiththeprinciplesofthe1998DataProtec- authorities) random-effects models will also be investi-
tion Act, the operation of the agreement and the study gated.Theproportionofcentresachievingcombineden-
publication policy. The rights for this data belong to the rolment, attrition and parent adherence will be analysed
study sponsor, and no processing, including further data using the same methods described for the primary out-
transfer in whole or in part to a third party, is permitted comes. The longitudinal impact of the intervention on
otherthanasstatedinthedatatransferagreements. parental engagement will be analysed among those local
authoritiesprovidingdataformorethanoneprogramme
Analysis per centre at follow-up, using logistic regression with
Statistical analysis of the quantitative elements of the random effects adjusting for the time point of the
trial is the responsibility of the LICTR statisticians. The programme.
analysis plan outlined in this section will be reviewed, Appropriate scoring manuals will be followed, and
and a detailed statistical analysis plan will be written and missing items within individual outcome measures will
approved, before any formal analyses are undertaken. be treated according to instructions for that particular
Statistical analyses will be carried out by the ITT measure. For all other outcomes (without instructions
principle, and statistical significance will be assessed at for dealing with missing data), half rule will be used,
the two-sided 5% significance level. The ITT population substitution of the mean of the answered questions for
is defined as analysis according to the randomisation that specific subscale for the missing responses as long
and regardless of compliance with the protocol or with- as at least half the questions are answered [57]. If more
drawal from thetrial. than 50% of the items are missing, the outcome will be
assignedasmissing.
Primaryanalysis
Theprimaryoutcomesassociatedwithenrolmentorattri- Datamonitoring
tion will be compared between the intervention and con- Trial supervision includes a core project team, a study
trolarmsthroughacluster-levelanalysis,usingaweighted management group, and a TSC. The core project team
t test of cluster-level proportions and a two-stage process willcomprisethe chiefinvestigatorandaresearchassist-
to adjust for covariates [55]. Intervention effects and ant, with oversight from all other monitoring groups.
corresponding95%confidenceintervalswillbepresented. Thestudymanagementgroupwillcomprisethe chiefin-
As part of a sensitivity analysis, the reliability of random vestigator, clinical trial research unit team (research as-
effectslogisticregressionwillalsobeinvestigated. sistant [WB], statisticians [BC and MC], data
In the primary analyses, missing data will be assumed management, trial management [SH] and health econo-
tobemissingcompletely atrandom;thatis,analyseswill mist [ST]). TheTSC will comprise an independent chair
be performed using complete cases (including those (a professor of chronic disease and public health) plus
where sufficient item-level data exists). The extent of independent expertise in statistics, qualitative and
unit non-response (data missing from a whole course) mixed-methods research, behaviour change and a parent
will be assessed, and the reason for missingness and the (a volunteer from our parent advisory group). Because
missingdata mechanismwillbeinvestigated. thetrialisusingroutine data,aseparatedatamonitoring
A sensitivity analysis accounting for all participants in and ethics committee was not convened. Rather, the in-
the ITT population assuming data missing at random, dependent TSC will adopt a safety monitoring role, and
usingmultipleimputation,willbeperformed,andpoten- will establish a subcommittee to review safety issues
tial predictors of missingness will be investigated. The should this become necessary. Serious adverse events
number of imputations will be determined using Bodner are not anticipated, and, as such, a separate protocol for
rule of thumb, where number of imputations is similar unmasking has not be produced. This will, however, be
tothepercentageofcasesthatare complete[56]. reviewed by the TSC as a regular item on the meeting
agenda.
Secondaryanalysis TheTSC operates in line with the LICTR terms of ref-
Adherence to HENRY programme content and the im- erence as amended and agreed by TSC members at their
pact of the HENRY optimisation on parenting and fam- first meeting. Data provided to the LICTR will be moni-
ily health will be analysed using a cluster-level analysis tored for quality and completeness by the LICTR, using
of either proportions or means, depending on the out- established verification, validation and checking pro-
come, as described for the primary analysis. As part of cesses. Clinical governance issues pertaining to all as-
sensitivity analyses, the degree of clustering at each level pects of routine management will be brought to the