Friday, May 21, 2010: 5:10 PM
Grand Ballroom E Level 5 (Philadelphia Marriott Downtown)
4:45 PM
A. Pickles
,
Health Methodology Research Group, University of Manchester, Manchester, United Kingdom
J. Green
,
Child and Adolescent Psychiatry, The University of Manchester, Manchester, United Kingdom
H. McConachie
,
Institute of Health and Society, The University of Newcastle, United Kingdom
T. Charman
,
Centre for Research in Autism and Education, Institute of Education, University of London, London, United Kingdom
C. R. Aldred
,
Psychiatry Research, University of Manchester, Manchester, United Kingdom
T. PACT Consortium
,
Psychiatry Research, University of Manchester, Manchester, United Kingdom
Background:
Properly conducted therapeutic trials should be a rich source of information on how change in interaction, behaviour and functioning can be brought about. The careful design that is required for rigorous evaluation can also provide a unique opportunity for a detailed decomposition of how change diffuses through the interactional environment and to do this in a way that may be causally meaningful. Such analyses will be key to learning not just what interactions are sensitive therapeutic targets but also where transmission of effect is blocked or dissipates. Such knowledge acquisition is essential to the process of revision, refinement and enhancement of therapies that should be the aim of our therapeutic research. However, for us to be able to use trials in this way requires that we elaborate in parallel both our designs and our analysis of findings. Objectives: We describe recent methodological concerns and advances that properly exploit randomisation even while examining relationships among post-randomisation measures and then consider some of the implications these have for measurement protocols and design.
Methods: Using the Pre-School Autism Communication Trial (PACT) as an illustrative vehicle we describe the combination of trajectory models that characterise the joint development of parent and child interaction during progress of the trial and counterfactual thinking that helps characterise the impact of treatment in a way that continues to exploit randomisation. We consider how these models can be extended to consider the transmission or disattenuation of effects to final outcome, and the appropriate ways in which we should assess the extent to which effects are transmitted along/mediated in the hypothesized manner. In so doing we highlight some of the implicit but implausible assumptions commonly made in the analysis of mediation.
Results: Trial data is complete but due to a pre-publication embargo on the study findings we are unable to summarise the empirical findings in this abstract. However all data and analysis will be available for presentation at the IMFAR meeting.
Conclusions: Treatment trials can and should be viewed and designed both as tests of proposed treatments and as research tools with which to explore and refine the therapeutic pathway.
