Predictors of Effective Scaling: A Meta-Epidemiological Study of Bias in Early-Stage Studies to Prevent Chronic Disease

PROJECT SUMMARY/ABSTRACT Developing behavioral interventions that are both effective/efficacious and that can be scaled is a major challenge in the field of behavioral science. Many behavioral interventions demonstrate early promise in their testing, only to fail to be effective during a larger, more well-powered evaluation. This common pattern (i.e. initial promise then fail) wastes valuable resources and time. The objective of this proposal is to investigate the role of preliminary, early-stage studies (e.g. pilot, feasibility studies) in the development of public health behavioral interventions with attention to internal and external biases introduced during early stage testing. This study will capitalize on developed search strategies and established resources to conduct a series of meta-epidemiological reviews of behavioral interventions targeting adult physical activity and dietary behaviors. In these reviews, behavioral interventions with a published early-stage study and a published, more well-powered trial of the same/similar intervention will be systematically coded for the presence of external and internal validity biases. Meta-regression models will be run to understand how the presence/absence of each bias in early-stage studies, influences the effect size and significance of larger iterations of the intervention. This project’s goal is to inform future physical activity and diet/nutrition behavioral intervention studies, increasing their efficiency and ultimately, the effectiveness of behavioral intervention research to address chronic, non-communicable disease. Aim 1 is to identify the prevalence and impact of external validity biases in a sample of adult physical activity and diet interventions. Aim 2 is to conduct qualitative interviews with lead/senior authors from Aim 1 to better understand the reasons for identified differences/similarities between early-stage and more well-powered studies. Through the execution of this project the following will be gained: extensive skills in systematic-review, meta-analytic techniques, meta-epidemiological methods, qualitative research processes, and scientific communication including peer-reviewed publication, scientific presentation, mentor peer-reviewed, and a drafted post-doctoral grant application. The proposed F31 study is significant because findings from the proposed study will identify characteristics that lead to successful and unsuccessful scaling of well-powered trials, potentially increasing the speed of behavioral intervention development. This study is innovative because it will be among the first to establish the prevalence and impact of external and internal biases in adult physical activity and dietary behavioral interventions.