Effect of removing an oncologist-directed behavioral nudge on sustaining rates of serious illness conversations: An interrupted time series analysis.

Abstract

242 Background: Behavorial economic nudges are increasingly used in oncology settings to encourage evidence-based practices, such as serious illness conversations (SIC). Prior work demonstrated the effectiveness of an oncologist-directed nudge identifying patients at high risk of 180-day mortality, which led to a 3-fold increase in SIC documentation. It is unknown whether such nudges must continue indefinitely to sustain effective behavior change, or whether clinician behavior change would persist in their absence. The objective of this analysis was to examine the effect of removing an oncologist-directed nudge on overall SIC rates. Methods: From 2019 to 2022, clinicians within our institution received a nudge consisting of weekly performance feedback on SIC completion and opt-out text messages identifying patients at high-risk of 180-day mortality based on a validated machine-learning algorithm. In March 2022, at the conclusion of clinical trial NCT04867850, this program ended. For this analysis, the population included adult patients seen by a medical oncologist at one of 11 practices between Sept 2021 and Sept 2022. Patient encounters with an SIC documented in the previous year were excluded, as well as those occurring during a washout period (March-April 2022). SICs were attributed to an encounter if documented within 30 days. The primary outcome was the number of SICs documented per 1000 patient encounters. We used linear probability models and an interrupted time series analysis at the patient-encounter level to analyze the association between nudge cessation and SIC rates. All analyses were unadjusted. Results: 111,459 encounters were included in the final analysis (56.9% prior to nudge cessation). Demographic data showed mean age 64.16 (SD 13.29), 57.02% female, 75.25% non-Hispanic white, 61.88% Medicare. There were 1,517 unique encounters with an SIC documented within 30 days (1.36%), with 1,030 pre-cessation (1.62%) and 487 post-cessation (1.01%). Rates of SICs were 15.3 per 1000 encounters in February 2022 (pre-cessation) vs. 9.7 in May 2022 (post-cessation), representing a 37% decline (Table). Unadjusted interrupted time series analyses showed a significant decrease in documented SICs per 1000 encounters (odds ratio 0.64, 95% CI 0.57 - 0.71). Conclusions: Removal of a clinician nudge was associated with a significant decline in SIC rates, suggesting a lack of sustained clinician behavior change without an ongoing behavioral nudge. Because nudges may be required indefinitely to promote evidence-based practices such as SICs, their design should be minimally intrusive and easily adaptable to evolving care standards. Adjusted analyses, marginal effects analysis, and time-to-event analyses are in progress.[Table: see text]