Encounter-based randomization did not result in contamination in a shared decision-making trial: a secondary analysis

Abstract

ObjectivesTo estimate the level of contamination in an encounter-randomized trial evaluating a shared decision-making (SDM) tool. Study Design and SettingWe assessed contamination at three levels: (1) tool contamination (whether the tool was physically present in the usual care encounter), (2) functional contamination (whether components of the SDM tool were recreated in the usual care encounters without directly accessing the tool), and (3) learned contamination (whether clinicians “got better at SDM” in the usual care encounters as assessed by the OPTION-12 score). For functional and learned contamination, the interaction with the number of exposures to the tool was assessed. ResultsWe recorded and analyzed 830 of 922 randomized encounters. Of the 411 recorded encounters randomized to usual care, the SDM tool was used in nine (2.2%) encounters. Clinicians discussed at least one patient-important issue in 377 usual care encounters (92%) and the risk of stroke in 214 encounters (52%). We found no significant interaction between number of times the SDM tool was used and subsequent functional or learned contamination. ConclusionDespite randomly assigning clinicians to use an SDM tool in some and not other encounters, we found no evidence of contamination in usual care encounters.