The Cool Little Kids randomised controlled trial: Population-level early prevention for anxiety disorders

The strongest precursor of internalising problems in young children is temperamental inhibition, manifested as fearfulness and a tendency to withdraw from new situations [29, 31‐34]. Additional known risks are harsh and overprotective parenting, and parent internalising problems [9, 10, 34‐39]. Together, these account for up to 45% of the variance in early childhood internalising symptoms [9, 10, 35].

The only randomised trials testing a parenting prevention model in inhibited preschool children were conducted by Rapee. Rapee's Cool Little Kids program is the first (and, thus far, only) effective early childhood prevention program for internalising disorders [30, 40‐42]. Targeting child inhibition and overprotective parenting, this parenting program aims to help preschool children become resilient to situational fears and abstract distressing worries. It teaches parents strategies to modify their preschool child's fear and distress, as well as their own (if relevant), based on standard principles for treating internalising disorders in children and adults [11, 42].

Two successful efficacy trials of the Cool Little Kids program have been reported. Rapee's first trial [41] recruited 146 children aged 4 years with temperamental inhibition, measured by parent-report questionnaire (>85th percentile) and intensive laboratory observation. Intervention parents received a university-based prevention program from teams of two clinical psychologists offering six group sessions designed to reduce overprotective parenting in response to early fearful behaviour. By age 5 years, the intervention children had developed significantly fewer anxiety disorders than controls (50% vs. 64%). These effects were even larger by age 7 (40% vs. 69%) [42]. Rapee's second study [40] recruited 71 inhibited preschoolers whose parents themselves had internalising disorders. The intervention group received an eight-session version of the program which extended to focus on parent anxiety as well as overprotective parenting. Six months later, the intervention children had substantially fewer internalising disorders, diagnosed in only 53% of the intervention group compared to 93% of controls.

Rapee's two efficacy trials are at the cutting edge of prevention research and have major potential public health implications. Population conclusions, however, are precluded by their sample bias (university location and self-selection by advertisement) and the labour-intensive laboratory observation methods used for selection. The unaddressed challenge is to determine 'real world' effectiveness across an entire population. We report the protocol for the next step - to conduct a population-level translational randomised trial.

The aims of this trial are to (a) determine the balance of benefits and harms of systematically screening preschoolers for temperamental inhibition and of a parenting intervention program offered to those at risk, (b) examine the impacts on child internalising problems at school entry, and (c) evaluate cost-effectiveness.

We hypothesise that children whose parents enter the program will do better one and two years after baseline (the first two years of school for most) than 'usual care' control children on the outcomes: i) fewer children with internalising disorders, ii) lower mean scores on early child internalising symptoms, iii) lower mean scores on harsh and overprotective parenting, iv) lower mean scores on parent internalising problems. We anticipate that the prevention program will be acceptable and cost-effective.

Figure 1 summarises the time line of the trial. It shows graphically the features at each stage that are common to both the intervention and control groups, and unique to the intervention group, in the manner suggested by Perera et al. [43].

Cool Little Kids is prospectively registered with an international clinical trials registry (ISRCTN30996662), will be conducted in line with ISPOR guidelines for cost-effectiveness trials alongside RCTs [44], and reported in accordance with the CONSORT statement. Project approval has been obtained from the Ethics in Human Research Committee of the Royal Children's Hospital, Melbourne, Australia

Extending Cool Little Kids to a population randomised trial requires more than evidence supporting intervention effectiveness. It requires 1) a universal service system attended by all or almost all 4 year olds, and 2) an acceptable screening tool to systematically identify the children at risk. Primary health care settings were not considered to be an appropriate setting for this screening, because in many countries health practitioners lack the time and resources to screen for this type of problem. In Australia, preschool services are accessed by almost all 4 year olds (95%) [45], at a time when parents and early childhood teachers are concerned about children's impending readiness for school [46]. Brief universal screening can therefore be reliably placed in the preschool setting. The known risk factors for internalising problems are child temperamental inhibition, parents' own internalising problems and overprotective and harsh discipline parenting. The potential benefit versus harm of screening will therefore be considered. For preschool children, it is likely to be more acceptable both to parents and the educational preschool setting to screen inhibition as a child precursor of internalising problems, rather than their parents' mental health and parenting characteristics.

Preschool services from four government areas will be selected to span the sociodemographic spectrum. Written and verbal briefing (study aims, recruiting procedure) will be delivered to preschools across local government areas in the second half of 2010. Recruitment will take place in early 2011 over the first few months of the preschool year. Preschools will distribute a study package including an Information Statement with a screening and consent questionnaire to all parents of children enrolled in their year prior to school. Parents will return this screening and consent questionnaire in a confidential envelope to a sealed letter box at their child's preschool. The questionnaire includes The Short Temperament Scales [47] to screen for child temperamental inhibition (see Table 1) and demographic items.

After scoring, the study team will notify parents by letter if their child does not score highly on temperamental inhibition and is therefore ineligible for the trial. Eligible parents will be mailed the detailed trial Information Statement and Consent form with the baseline questionnaire ascertaining family risk for internalising problems and more detailed demographic characteristics (see Table 1). The study team will also telephone all eligible parents to explain trial procedures in detail and answer parents' questions.

After recruitment and consent, the statistician will coordinate a concealed web-based randomisation process. Computer-generated sequences of random numbers will be used to determine the trial arm status of each child. A block randomisation process, stratified for each preschool separately, will minimise the imbalance between the numbers of intervention and control participants within each preschool. After families are randomised, the study team will notify all parents by letter whether they are in the intervention or control arm, and arrange group session bookings for intervention parents. Participant allocation will be concealed to the clinician assessing diagnostic outcomes; however, the group facilitators and parents cannot be blinded to their group allocation.

Approximately one (2012) and two (2013) years after baseline, a clinician blind to group allocation will interview parents to ascertain child internalising diagnoses (see Table 1). These interviews will be conducted by telephone, which has demonstrated validity and is less expensive than face-to-face interviews [48]. A questionnaire mailed to parents will further measure child internalising symptoms, parenting practices, parent mental health and service use.

The randomised trial will have a sample size of around 500 inhibited children, providing 80% power at the 5% level of significance to detect the level of reduction in internalising problems found in Rapee's prior efficacy trials, namely a 14% reduction in child internalising disorders between the intervention (50%) and control (64%) groups [41]. This sample size estimation considers an achievable attrition rate for a quality trial, allowing for up to 20% loss to follow up.

Regression analyses will compare child outcomes and costs at school-entry (1 and 2 years after baseline) between the intervention and control arms using the intention-to-treat principle, where participants are analysed in the groups to which they were randomised. Logistic regression will be used for binary outcomes (presenting the percentages for each trial arm and odds ratio between them) and linear regression for continuous outcomes (presenting the mean and standard deviation for each trial arm and mean difference between them). Both unadjusted analyses and analyses adjusted for potential confounding factors determined a priori before randomisation (child gender, family socioeconomic level, parent mental health) will be implemented. Tests of interaction, specified a priori, will explore differential intervention effects for children who are higher on risk factors at baseline. In line with Rapee's second efficacy trial [40], we anticipate larger group differences might emerge for children who are higher on risk factors at baseline.

Parents and group facilitators will also be mailed questionnaires at the end of the intervention (late 2011) to evaluate the screening process and intervention, where applicable. All parents will report on perceived harms and benefits of screening. Intervention parents and group facilitators will report on intervention acceptability, extent to which components were implemented, and barriers to attendance. Group facilitators will complete a standard content checklist after each group session to record the degree to which the intervention was delivered. In addition, a research assistant will observe a random 10% of parent groups and evaluate protocol adherence using the content checklist.

Most of the study outcomes are largely clinical in nature (e.g. diagnosis of anxiety disorders, scores of the various clinical scales). This means that the majority of analyses will be cost-effectiveness analyses (CEA). While such analyses are important they have limitations in deciding whether the costs and benefits of interventions represent good value-for-money or not. Cost-utility analysis (CUA) is a form of cost-effectiveness analysis where the costs are measured in monetary terms and the consequences in a generic outcome metric capable of capturing both mortality and morbidity effects, allowing judgments regarding the 'worth' or the 'value' of interventions to be made (as well as comparisons to other interventions both within and across the different disorder/diseases). Parental utility will be measured using the Assessment of Quality of Life Scale -8 Dimension (AQoL8D) [49]. There are no existing tools to measure utility in preschool children, though as the children age measures such as the Health Utilities Index 3 (HUI) may be used [50]. Alternatively, "proxy" utilities may need to suffice as used in a recent study which modeled the efficacy credentials of the Cool Little Kids intervention [51]. It is because the economic analyses will be comprised of both CEA and CUA analyses, the appropriate study-frame is a cost-consequence analysis.

The underlying principle in identifying costs relevant to the economic evaluation is that the inclusion/exclusion criteria should mirror the study perspective. This economic evaluation intends to adopt a largely societal perspective, so all costs to the health sector, participants and their families as well as other sectors impacted upon by the intervention will be included. As there are no previous economic evaluations in this area, a comprehensive resource use questionnaire is proposed to capture a wide variety of resource use for both parents and children (including time costs of parents). Costs associated with the initial research, design and set-up of the intervention and the development of any program materials will be excluded, as these are largely one-off 'sunk' costs (unrecoverable past expenditures), and will not be incurred in on-going routine implementation of the intervention.

The technique of 'bootstrapping' will be used to obtain confidence intervals for cost effectiveness ratios, since parametric techniques are inappropriate for use on both skewed variables and ratios. The sensitivity of the results will also be tested against: different discounting scenario; variation in the utility weights for the children; and, differences in unit cost prices. Costs falling upon the health sector, patients or their families, the government and other sectors will be presented in total and disaggregated form.

The value of economic evaluations alongside effectiveness trials is that a comprehensive analysis of the costs and benefits at an individual level can be obtained. However, such evaluations are usually time limited and do not capture the longer term costs and benefits associated with such interventions. Modelling techniques are a valuable addition to trial based evaluations. Firstly, the longer term costs and consequences associated with interventions can be estimated, though such analyses unavoidably require some assumptions to be made (e.g. a sustained effect of the intervention beyond the duration of the trial). Secondly, modelling allows the costs and benefits at a population level to be estimated. Such information is important to decision makers who are set with the task of allocating national health care budgets and must choose between different interventions. Such modelling of the current intervention is feasible and would build upon a previous study which evaluated the population cost-effectiveness of the Cool Little Kids intervention [51].