Complementary to the “Survival Analysis for Junior Researchers Conference” (see above), an afternoon workshop on multi-state modelling was jointly organised by the RSS Young Statisticians’ Section and the Leeds/Bradford local group on 14 April 2016. Several SAfJR delegates stayed on for the workshop and were joined by local group members. The meeting was introduced and chaired by Professor Linda Sharples (Leeds Institute of Clinical Trials Research, University of Leeds).
The first speaker was Dr Andrew Titman (Department of Mathematics and Statistics, University of Lancaster) who gave a talk entitled ‘Multi-state modelling: An overview’ to provide an introduction to the methodology. The talk gave an overview of the principal methods and key assumptions used in multi-state models, covering both continuously observed processes (where much of the machinery from standard survival analysis carries across) and interval-censored or panel-observed data (where there are additional computational challenges, and analysis is usually parametric). The methods were illustrated through application to progression-free and overall survival in cancer studies, and modelling the onset of cardiac allograft vasculopathy in post-heart-transplantation patients.
To offer an example of current research in the field, Dr Aidan O’Keeffe (Department of Statistical Science, University College London) gave a presentation entitled ‘Multi-state models and causal arguments: Application to a study of clinical damage in psoriatic arthritis’. The presentation illustrated the use of multi-state models as a method for assessing a causal effect of one process on another in the context of psoriatic arthritis, and showed how multi-state models can be used to assess the causal relationship between disease activity (tenderness and swelling) and clinical joint damage.
The final speaker was Dr Howard Thom (School of Social and Community Medicine, University of Bristol) whose talk was entitled ‘Using Parameter Constraints to choose State-Structures in Cost-effectiveness Modelling’. This research addressed the question of structural uncertainty in cost-effectiveness decision models – in particular, the choice of state-structure when using a multi-state model. Key model outputs, such as treatment recommendations and identification of future research needs, may be sensitive to this choice of state-structure. Dr Thom described a new method that involves re-expressing a model with merged states as a model on the larger state space, meaning that standard statistical methods for comparing models with a common reference dataset can be used. This methodology was then applied to data for prescribing anti-depressants by depression severity.
Overall, the workshop gave a comprehensive overview of methodologies in the field, and illustrated them through two very different applications in causal assessment of the relationship between different elements of arthritis and to health economics decision modelling.
The slides from the event can be found at the local group website: