Evaluating interventions to promote routine preventive screenings: a comparison of analytical outcomes

Often in public health, we are interested in promoting routine preventive screenings (e.g., blood glucose monitoring, hypertension screening, or mammography). Evaluating novel interventions to encourage frequent screenings using randomized controlled trials can help inform evidence-based health promotion programs. When the desired behavior change is a recurrent event, specifying the most meaningful study outcomes may prove challenging.

To understand the efficiency of multiple approaches for evaluating an intervention seeking to increase regular health screenings we (a) simulated several replications of a trial with a positive intervention effect under various censoring scenarios, (b) formulated three different analytical outcome definitions (screening a certain number of times during the entire study period versus not, screening at least once within a clinically meaningful time period versus not, "hazard" or instantaneous rate of screening), and (c) compared them with regard to interpreting results and estimating power at different sample sizes.

Approaches which better utilize detailed prospective data, while also accounting for within-participant correlations, are less likely to miss the actual underlying benefits conferred by a new prevention strategy compared to relying on a dichotomous measure derived from aggregating events over the study duration. Such approaches are also more powerful in realistic scenarios wherein some participants are lost to follow-up over time.

Researchers should carefully consider the choice of analytical outcomes and strive to employ more efficient approaches that model comprehensive event-specific information, rather than summarizing repeated measures into less-informative dichotomous responses, while designing and conducting trials with recurrent preventive screenings.