Documente Academic
Documente Profesional
Documente Cultură
more nuanced understanding of what individual and Third, we need to accept that many developmental
contextual factors contribute to treatment success or concerns (such a language disorder) are incredibly
failure is more clinically relevant than a simple ‘does stable. In these cases, change requires sustained input,
this work?’ question. Blanket approaches in such especially as child
volatile settings are unlikely to fully address individual behaviours and the envi- There is increasing
needs. ronments children expe- focus on implementa-
Finally, Burgoyne et al. (in this issue) investigate rience are often changing.
tion, or turning
an adaptation of a tried and tested approach to What if our research ques-
developing children’s oral language skills that uses tion shifted from ‘does this
therapeutic research
parents, as opposed to clinicians or educators, as the work’ to ‘what would it discoveries into practi-
agents of change. The treatment was much more take to achieve the desired cal benefit for human
intensive, designed to be delivered several times a outcome?’ With that in health and well-being.
week over a 30-week period, and included a follow-up mind, intervention trials In practice, this means
assessment at 6 months that tested both core out- could usefully compare testing interventions
comes, and cascading impacts on early literacy. different implementation that can readily be
Parents received a modest amount of training methods, varying dosage implemented and
(1.5 hr) to deliver the programme and thus, this has and agents of delivery. ‘scalable’ in real-world
potential to be extremely cost-effective given the Technology presents an settings. This forces
intended intensity. important tool for meeting
researchers to consider
The statistical approach to analysis employed demand when resources
latent variables, which have the advantage of explor- are stretched, but there
the practical and
ing change in the underlying construct, unfettered by may be limited contexts in financial challenges of
measurement error. This large study provided evi- which this is applicable. intervention delivery
dence of a moderate (d = ~.35) treatment effect on Finally, in many fields, and raises important
language that was maintained at follow-up and we see a move to provi- questions about
accompanied by a similar degree of difference on sion of ‘universal’ ser- dosage and implemen-
early literacy measures. These differences were vices, designed to be tation.
harder to detect on individual tests, where measure- delivered to all children
ment error may obscure real change. with a view to reducing the numbers of children
A challenge for this type of programme is that referred to tertiary services. While these efforts are
parents, on average, completed only 50% of the laudable, there will always be a need for specialist
intended sessions. Issues of parent language and interventions targeted at individuals who will not
literacy (as the programme involved shared book benefit from a universal approach. We therefore need
reading) may limit the extent to which some families to maintain flexibility in the system to deal with
may access this intervention. Once again, participants varying and ongoing clinical need. One size does not
were not clinically referred, with only 20% falling below fit all, and while this presents challenges, it is also
commonly accepted cut-offs for language disorder. It is what drives innovation and our understanding of
encouraging that these children make as much pro- developmental psychopathology.
gress during the intervention period as more verbally
able peers, but replication in clinical cohorts is needed.
Each of these studies represents an important Courtenay Frazier Norbury
advance in their respective fields and should influence C.N. is professor of developmental language and
choices about effective approaches to supporting chil- communication disorders at the University College
dren and young people faced with developmental chal- London. She is a joint editor for JCPP. She has
lenges. Collectively, they also inform us about some declared that she has no competing or potential
common issues that require wider dialogue and con- conflicts of interest in relation to this editorial.
sideration by those of us undertaking intervention
research. First, testing developmental cascade models
Courtenay Frazier Norbury
requires longitudinal data, and very few intervention
studies include follow-up periods of sufficient duration
References
to really test cascading effects. Second, statistical power
Bornstein, M.H., Hahn, C.-S., & Putnick, D.L. (2016). Stability
is a critical issue that should concern us all. We of core language skill across the first decade of life in
know that underpowered studies can reveal positive children at biological and social risk. Journal of Child
findings that are likely to be statistical artefacts, rather Psychology and Psychiatry, 57, 1434–1443.
than true effects (Button et al. 2013). There is a growing Button, K.S., Ioannidis, J.P.A., Mokrysz, C., Nosek, B.A., Flint,
consensus that trials of psychological processes should J., Robinson, E.S.J., & Munafò, M.R. (2013). Power failure:
why small sample size undermines the reliability of neu-
be powered to detect medium sized effects, are practical
roscience. Nature Reviews Neuroscience, 14, 365–376.
in terms of the sample sizes required, and can provide Green, J. (2016). Editorial: Ingenious designs and causal
meaningful results. In addition, even seemingly inference in child psychology and psychiatry. Journal of
small changes can have large cascading impacts Child Psychology and Psychiatry, 57, 549–551.
(Bornstein, Hahn, & Putnick, 2016), but we need to Masten, A.S., & Cicchetti, D. (2010). Developmental cascades.
explicitly model these using longitudinal trial designs. Development and Psychopathology, 22, 491–495.