| Summary: | Background: There is growing interest in the use of mobile phone interventions to promote behaviour change and their efficacy has been assessed extensively using exploratory methods (e.g., pilot studies and proof of concept). However, in an era of evidence-based practice, calls have been made for digital interventions to be evaluated with the same rigour as other forms of behaviour change interventions i.e. with randomised controlled trials.
Aims: This presentation aims to overview a number of the issues which make randomised controlled trials of mobile phone interventions more complex than for traditional face-to-face behaviour change interventions and how these can be addressed in the statistical analysis.
Method(s)/results: The issues and their possible solutions will be discussed in relation to currently published randomised controlled trials of mobile phone behaviour change interventions. These issues include missing data and drop-out, randomisation failure, confounding effects of time, and the presence of clustering. Possible solutions which will be considered include imputation, multi-level modelling, regression adjustment, the use of Bayes factors and matching procedures.
Conclusions: This presentation will highlight a number of issues for consideration when evaluating mobile phone interventions in order to aid researchers active in the mHealth field.
|
|---|
