You cannot register for this webinar

Stepped wedge cluster-randomized trials (SW-CRTs) are increasingly used to evaluate interventions in healthcare settings, yet contamination due to physician collaboration remains an overlooked challenge. This study examines spillover effects in an SW-CRT assessing a video game intervention aimed at increasing Advance Care Planning (ACP) billing among hospitalists, which previously found no direct impact of the intervention (OR: 0.96, 95% CI: 0.88-1.06, p=0.42). Here, we introduce a method leveraging physician social networks constructed from shared-patient encounters to quantify intervention diffusion beyond directly treated individuals. While the intervention itself showed a borderline significant effect on ACP billing (OR 1.175, 95% CI: 0.999–1.383, p = 0.052), network-based spillover effects were substantial (OR 2.794, 95% CI: 2.484–3.142, p < 0.001), demonstrating that non-intervened physicians were strongly influenced by their treated peers. Moreover, as spillover increased, the direct intervention effect diminished (OR 0.889, 95% CI: 0.799–0.999, p = 0.032), suggesting an interplay between direct and indirect effects. These findings highlight the critical need to account for physician collaboration when designing and analyzing SW-CRTs to avoid underestimating intervention effects and to better understand the mechanisms driving behavior change in clinical practice.